Federal Communications Commission DA 18-99 Before the Federal Communications Commission Washington, D.C. 20554 In the Matter of International Comparison Requirements Pursuant to the Broadband Data Improvement Act International Broadband Data Report ) ) ) ) ) ) GN Docket No. 17-199 SIXTH REPORT Adopted: February 2, 2018 Released: February 2, 2018 By the Chief, International Bureau: TABLE OF CONTENTS Para. I. INTRODUCTION .................................................................................................................................. 1 II. BACKGROUND .................................................................................................................................... 2 III. DISCUSSION ........................................................................................................................................ 5 IV. HIGHLIGHTS OF THE FINDINGS ..................................................................................................... 7 A. Broadband Speed Comparison ......................................................................................................... 8 B. Broadband Price Comparison ........................................................................................................ 12 C. High-Speed Broadband Deployment Comparison with Europe .................................................... 16 D. Demographics Dataset ................................................................................................................... 22 E. Market and Regulatory Developments .......................................................................................... 23 F. Other Relevant Information ........................................................................................................... 32 1. Broadband Subscription (OECD Countries) ........................................................................... 32 2. Efforts to Improve International Broadband Data. .................................................................. 35 V. CONCLUSION .................................................................................................................................... 36 VI. PROCEDURAL MATTERS ................................................................................................................ 37 APPENDIX A – Country List APPENDIX B – Broadband Speed Comparison APPENDIX C – Broadband Price Comparison APPENDIX D – High-Speed Broadband Deployment Comparison with Europe APPENDIX E – Demographics Dataset APPENDIX F – Market and Regulatory Developments I. INTRODUCTION 1. As required by Section 103(b) of the Broadband Data Improvement Act (BDIA), we issue this sixth International Broadband Data Report (Report), which provides comparative international information on broadband services and, where possible, a year-to-year measure of the extent of broadband service capability in the United States and select communities and countries abroad.1 In this Report, we compare fixed and, for the first time, mobile broadband (LTE) speeds in the United States with the selected countries, to the extent data are available. We improve upon our pricing comparison from 1 47 U.S.C. § 1303(b). The Broadband Data Improvement Act, Pub. L. No. 110-385, 122 Stat. 4096 (2008), is codified in Title 47, Chapter 12 of the United States Code. 47 U.S.C. § 1301 et seq. Federal Communications Commission DA 18-99 2 previous reports by providing a more comprehensive assessment of the competitiveness of broadband in each country and the value that broadband providers are delivering to consumers. We include a comparison of high-speed fixed and mobile broadband deployment in the United States and in Europe. Finally, we present demographic, market, and other regulatory information relating to broadband service capability. We include the highlights of our findings in this Report, and present the detailed data sources and additional discussion in the relevant appendices. II. BACKGROUND 2. The BDIA requires the Commission to include as part of its assessment in the annual broadband deployment report “information comparing the extent of broadband service capability (including data transmission speeds and price for broadband service capability) in a total of 75 communities in at least 25 countries abroad for each of the data rate benchmarks for broadband service utilized by the Commission to reflect different speed tiers.”2 The BDIA directs the Commission to choose international communities comparable to various communities in the United States with respect to population size, population density, topography, and demographic profile.3 The Commission is required to include “a geographically diverse selection of countries” and “communities including the capital cities of such countries.”4 The Commission must “identify relevant similarities and differences in each community, including their market structures, the number of competitors, the number of facilities-based providers, the types of technologies deployed by such providers, the applications and services those technologies enable, the regulatory model under which broadband service capability is provided, the types of applications and services used, business and residential use of such services, and other media available to consumers.”5 3. Thirteenth Broadband Deployment Notice of Inquiry and Comments. In the Thirteenth Broadband Deployment Notice of Inquiry, the Commission sought comment “on how to best interpret the statutory obligations” of the BDIA and how “to improve our next international assessment.”6 The Commission asked whether it should continue using the same overall approach to the international assessment as in past years7 and whether it should select “at least 25 countries that have developed broadband markets, which have readily available data.”8 The Commission asked whether it should continue to present, for example, actual broadband speeds in different countries by using the publicly 2 47 U.S.C. § 1303(b)(1). Several terms that we use in this Report, such as “broadband,” “advanced telecommunications capability,” and “availability” may have specialized meanings in other contexts, and nothing in this Report should be read to suggest that our use of terminology here is intended to affect the meanings of other specialized terms in the context of the 2018 Broadband Deployment Report or in other proceedings. See, e.g., Inquiry Concerning the Deployment of Advanced Telecommunications Capability to All Americans in a Reasonable and Timely Fashion, GN Docket No. 17-199, Broadband Deployment Report, FCC 18-10, (Feb. 2, 2018) (2018 Broadband Deployment Report). The 2018 Broadband Deployment Report incorporates by reference this Report to fulfill the requirements of Section 103(b) of the BDIA. 47 U.S.C. § 1303(b). 3 47 U.S.C. § 1303(b)(2). 4 Id. 5 Id. § 1303(b)(3). 6 Inquiry Concerning the Deployment of Advanced Telecommunications Capability to All Americans in a Reasonable and Timely Fashion, GN Docket No. 17-199, Thirteenth Section 706 Report Notice of Inquiry, 32 FCC Rcd 7029, 7044, para. 51 (2017) (Thirteenth Broadband Deployment Notice of Inquiry). 7 Id. at 7044-45, para. 52. Past international assessments have included comparisons of actual broadband download speeds, prices for fixed and mobile service plans, rural and non-rural broadband deployment (comparing the United States and Europe), demographic data (population, income, and education), and a summary of market and regulatory aspects of the countries included for comparison. 8 Id. at 7045, para. 53. (continued….) Federal Communications Commission DA 18-99 3 available speed test data provided by Ookla, proprietor of speedtest.net.9 The Commission also sought comment on “how best to compare fixed and mobile broadband pricing in the United States with the selected countries,” including whether to use a hedonic approach, and if so, whether previously captured price variables are adequate.10 The Commission also sought comment on whether there are other sources of international broadband pricing data that could better improve the quality and usefulness of the comparisons.11 The Commission also invited commenters to provide any relevant qualitative and quantitative data or suggest data sources that could improve our analysis.12 4. We received very few comments regarding how to best interpret the statutory obligations of the BDIA or ways to improve the international assessment. One commenter recognized the inherent difficulty in comparing the United States to other countries, but stated “it can still be instructive to compare similar locales in various countries.”13 Some commenters stated that international comparisons are of limited use, indicating, for example, that it is “doubtful that 25 comparable countries exist” due to the United States’ large size, number of sparsely populated areas, and low density cities as compared to large foreign cities.14 No commenter specifically offered or suggested alternate sources of broadband speed or price data.15 III. DISCUSSION 5. Selection of Countries for Comparison. The BDIA directs the Commission to report information comparing the extent of broadband service capability “in a total of 75 communities in at least 25 countries abroad.”16 To implement the statutory directive, we selected 28 foreign countries for comparison with the United States. To guide our selection of countries with comparable communities for this Report, we developed several criteria to meet the statutory directive of developing a geographically diverse and detailed set of data on international broadband service capability.17 First, we attempted to select countries with comparable communities18 by selecting countries that have more developed broadband markets. We focused on those Organization for Economic Cooperation and Development (OECD) Member countries that have the highest levels of broadband adoption under the assumption that 9 Id. at 7045, para. 54. 10 Id. at 7045-46, para. 55. A hedonic regression provides an empirical summary of how prices vary with the characteristics of a good. Our hedonic regression builds on the price index by allowing adjustment of prices for cost and demographic differences across countries and then predicting broadband prices for each country at the average U.S. values of these variables. See infra Appx. C at para. 2, note 5. 11 Thirteenth Broadband Deployment Notice of Inquiry, 32 FCC Rcd at 7045-46, para. 55. 12 Id. at 7045, para. 52. 13 Judd Rodgers Comments at 1. 14 Richard Bennett Comments at 8. 15 Some commenters referenced various international broadband reports suggesting that the United States lags behind other countries for various broadband metrics, such as speed. See, e.g., Thaddeus Ballantine Comments at 1 (citing Akamai data to state that “we [the United States] are not in the front in terms of bandwidth in fact we are 14th based on average download speed and 16th in terms of average peak speed”); David G. Mackay Comments at 1 (citing Speedtest.net/global-index to note that “the US ranks at number 9 for fixed broadband, and 46th for mobile”). 16 47 U.S.C. § 1303(b)(1). 17 Id. § 1303(b)(2). 18 Id. (“The Commission shall choose communities for the comparison under this subsection in a manner that will offer, to the extent possible, communities of a population size, population density, topography, and demographic profile that are comparable to the population size, population density, topography, and demographic profile of various communities within the United States.”). (continued….) Federal Communications Commission DA 18-99 4 countries with higher levels of broadband adoption have more developed broadband markets.19 Second, we included only countries for which substantial and relevant information is available.20 The OECD also regularly compiles demographic and broadband data on all Member countries, and other broadband data such as speeds, prices, and deployment for these countries are readily available from other public sources.21 6. We combined the OECD fixed and mobile broadband subscription data for each country to assess which OECD countries have the highest broadband subscriptions.22 Based on these data, the United States is included in the top 27 ranked OECD countries. For comparison with the United States, we selected the following top 26 OECD foreign countries (in alphabetical order): Australia, Austria, Belgium, Canada, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Iceland, Ireland, Italy, Japan, Latvia, Luxembourg, Netherlands, New Zealand, Norway, Portugal, South Korea, Spain, Sweden, Switzerland, and United Kingdom. To increase geographic diversity and representation of the Americas region, we also included OECD Members Chile and Mexico although these countries are not among the top 27 foreign countries with the highest levels of broadband adoption. By including two additional countries, we compare up to 28 foreign countries to the United States (for a total of up to 29 countries), which exceed the statutory minimum of 25 comparison countries.23 Our international comparisons reflect that the sources, definitions, and/or time periods of available data often differ by country and by dataset.24 19 We note that the OECD’s subscription metrics define transmission speeds of at least 256 kbps in one direction as “broadband service” for both fixed and mobile Internet access, which is slower than the 25 Mbps download/3 Mbps upload speeds which the Commission considers as “advanced telecommunications capability.” 2018 Broadband Deployment Report at para. 15. However, it is our view that the OECD’s subscription metric, when applied across all countries, including the United States, provides a sufficiently reliable metric for selecting the countries to be discussed in this Report. In particular, the OECD Broadband Portal provides data on the number of broadband subscriptions (fixed and mobile) per 100 inhabitants. OECD, Broadband Portal, Fixed and Wireless Broadband Subscriptions per 100 Inhabitants, Table 1.2 (Dec. 2016) (OECD Broadband Subscriptions Table 1.2), http://www.oecd.org/sti/broadband/1.2-OECD-WiredWirelessBB-2016-12.xls. The OECD broadband subscription data include fixed broadband subscriptions with downloads speeds that are at least 256 kbps (DSL, satellite, terrestrial fixed wireless) or greater than 256 kbps (cable, fiber, broadband over power lines, leased lines), and mobile subscriptions which advertise data speeds of 256 kbps or greater. OECD, Broadband Methodology: OECD Broadband Subscriptions Criteria (2015) (OECD Broadband Methodology), http://www.oecd.org/sti/broadband/broadband-methodology.htm. 20 International Comparison Requirements Pursuant to the Broadband Data Improvement Act; International Broadband Data Report, GN Docket No. 15-191, 31 FCC Rcd 2667, 2668-69, para. 3 (2016) (Fifth International Broadband Data Report). 21 See OECD, Members and Partners, http://www.oecd.org/about/membersandpartners/ (last visited Jan. 16, 2018) (identifying OECD Member countries). 22 OECD Broadband Subscriptions Table 1.2. 23 47 U.S.C. § 1303(b)(1); see infra Appx. A. 24 The data relied upon in this Report come from a variety of sources, including contractual arrangements with TeleGeography, S&P Global (formerly SNL Kagan), and Ookla, staff research, and publicly available records. Federal Communications Commission DA 18-99 5 IV. HIGHLIGHTS OF THE FINDINGS 7. In this section, we present highlights of our findings. The detailed presentation of the analysis and data is in the relevant appendices. A. Broadband Speed Comparison 8. In Appendix B, we present information on “data transmission speeds” for broadband service capability as directed by the BDIA.25 We present data on actual fixed and mobile broadband speeds based on data gathered by Ookla for the United States and 27 comparison countries for a ranking of fastest actual speed (1st) to slowest (28th).26 Broadband speeds are often presented as either advertised speed or the actual speed.27 We obtained actual speed data from Ookla28 for our fixed and mobile international speed comparisons, which are collected primarily from software-based tests on an end user’s device using speedtest.net.29 The data are aggregated at the city level and include observations in 2014, 2015, and 2016 for both U.S. and international cities. 9. Fixed Broadband Speed Results. The United States ranked 10th out of 28 countries in 2016 in terms of actual download speeds (55.07 Mbps) weighted by the number of tests in each city—an improvement from a ranking of 11th in 2015 (40.38 Mbps) and 15th in 2014 (28.09 Mbps).30 In 2016, the median weighted download speed for the United States increased to 55.44 Mbps from 40.17 Mbps in 2015 and 28.29 Mbps in 2014, and the U.S. ranking improved from 14th in 2014 and 13th in 2015 to 10th out of the 28 comparison countries.31 25 47 U.S.C. § 1303(b)(1) (“As part of the assessment and report required by section 1302 of this title, the Federal Communications Commission shall include information comparing the extent of broadband service capability (including data transmission speeds and price for broadband service capability) in a total of 75 communities in at least 25 countries abroad for each of the data rate benchmarks for broadband service utilized by the Commission to reflect different speed tiers.”). 26 Due to data availability, the broadband speed comparison does not include Latvia. See infra Appx. B at para. 11. 27 Actual speed is measured primarily by two methods: (i) by installing special hardware on an end user’s device that enables the hardware to measure actual download and upload speeds and (ii) software-based tests. Steve Bauer, David Clark, William Lehr, Massachusetts Institute of Technology, Understanding Broadband Speed Measurements at 11-17 (2010) (MIT Report), https://www.measurementlab.net/publications/understanding-broadband-speed- measurements.pdf. In this Report, “actual speed” refers to mean actual speed unless otherwise specified. See infra Appx. B, para. 2, note 4. 28 We obtained speed data through a contractual arrangement with Ookla. Ookla, Ookla Speedtest (Ookla Speedtest), http://www.speedtest.net (last visited Jan. 16, 2018). We note that generally, crowd-sourced data (such as Ookla’s) can provide the benefit of generating a large volume of data at a very low cost and of measuring actual consumer experience on a network in a wide variety of locations, indoor and outdoor. Crowd-sourced data, however, are often not collected pursuant to statistical sampling techniques, and may require adjustments to construct a representative sample from the raw data. For instance, crowd-sourced mobile data come from a self- selected group of users, and there often is little control for most tests regarding such parameters as when people implement the test, whether the test is performed indoors or outdoors, the geographic location of the tester, and the vintage of the consumer’s device. See, e.g., Implementation of Section 6002(b) of the Omnibus Budget Reconciliation Act of 1993; Annual Report and Analysis of Competitive Market Conditions With Respect to Mobile Wireless, Including Commercial Mobile Services, WT Docket No. 13-135, Seventeenth Report, 29 FCC Rcd 15311, 15405-06, para. 191 (2014) (Seventeenth Mobile Wireless Competition Report). 29 Ookla’s mobile speed measurements are derived from customer tests run by Ookla Speedtest mobile apps that measure the performance of mobile connections. Ookla, Ookla Speedtest Mobile Apps (Ookla Speedtest Mobile), http://www.speedtest.net/mobile/ (last visited Jan. 16, 2018). 30 See infra Appx. B, para. 14, Tbl. 2 and Fig. 1. 31 Id. at para. 15, Tbl. 3. Our calculations are based on the median of the city-level averages reported by Ookla. Because the data are aggregated at the city level and we do not have access to individual speed test records, we (continued….) Federal Communications Commission DA 18-99 6 10. Mobile Broadband Speed Results. With regard to actual mobile broadband speeds, the United States ranked 24th out of 28 countries in both 2016 (19.98 Mbps) and 2015 (15.58 Mbps), and 18th in 2014 (12.62 Mbps).32 Actual mobile download speeds in the United States increased by approximately 58 percent from 2014 to 2016.33 In 2016, the median weighted download speed for the United States increased to 19.36 Mbps from 12.62 Mbps in 2014, and its ranking decreased from 17th to 24th out of the 28 comparison countries.34 11. Historical Overview of U.S. Fixed Broadband Speed. Based on data from past International Broadband Data Reports, we present U.S. fixed download speeds and rankings from 2012 to 2016 to show some indication of how speeds and U.S. rankings have evolved over time. We note that due to differences in the Ookla data from 2012-2013 and the data from 2014-2016, the earlier data are not directly comparable to the later data.35 Nevertheless, the data indicate that for the United States, both speeds and international rank have been on a rising trend since 2012.36 Based on mean speed measurement, the United States ranked 25th fastest of 40 countries in 2012 (14.50 Mbps), and 26th fastest of 40 countries in 2013 (18.67 Mbps).37 As noted above, the mean U.S. speed rank has since risen to 10th fastest of 28 countries in 2016 (55.07 Mbps).38 B. Broadband Price Comparison 12. In Appendix C, we present in detail our fixed and mobile broadband price analysis as directed by the BDIA.39 We examine advertised broadband prices for both fixed and mobile service plans in the United States and up to 28 comparison countries depending on data availability (for a total of up to 29 countries). Between June and August of 2017, we collected a stratified random sample of the prices and terms for almost 3,000 fixed and mobile broadband plans from the websites of broadband providers in the United States and the selected countries.40 cannot compute a true median. Here, the median refers to the median of the aggregated (average) annual city speed tests weighted by sample size, and average refers to the averages at the city level as provided by Ookla. Id. at para. 15, note 14. 32 Id. at para. 20, Tbl. 7 and Fig. 2. 33 Id. at para. 20. 34 Id. at para. 21, Tbl. 8. 35 The Fourth International Broadband Data Report and the Fifth International Broadband Data Report relied on Ookla speed data for 2012 to 2014 that consisted of daily speed test results for all cities. See International Comparison Requirements Pursuant to the Broadband Data Improvement Act; International Broadband Data Report, GN Docket 14-126, Fourth Report, 30 FCC Rcd 14994, 15225, 15248, Appx. F (2015) (Fourth International Broadband Data Report); Fifth International Broadband Data Report, 31 FCC Rcd at 2821,Appx. F (2016) (Fifth International Broadband Data Report). Ookla speed data for 2014 to 2016 used in this report consists of city speed test results averaged up to the yearly level, amounting to far fewer observations than the prior methodology. See infra Appx. B, paras. 10, 24-25. 36 See infra Appx. B, para. 25. 37 Id. at para. 25, Tbl. 11. 38 Id. at para. 25, Tbl. 12. 39 47 U.S.C. § 1303(b)(1) (“As part of the assessment and report required by section 1302 of this title, the Federal Communications Commission shall include information comparing the extent of broadband service capability (including data transmission speeds and price for broadband service capability) in a total of 75 communities in at least 25 countries abroad for each of the data rate benchmarks for broadband service utilized by the Commission to reflect different speed tiers.”). 40 Fixed broadband price data include prices for both standalone broadband and bundles consisting of broadband and video service. Mobile broadband price data include primarily postpaid smartphone plans (both single and shared line) that allowed both unlimited voice calling and texting. Additionally, postpaid plans refer to plans that are paid (continued….) Federal Communications Commission DA 18-99 7 13. We rank the countries by fixed and mobile price from the least expensive (1st) to most expensive (e.g., 29th) according to three different methodologies. As in previous reports, the first method compares countries according to unweighted average prices for standalone broadband plans within certain download speed ranges and mobile plans within bands of data usage allowances. To more closely match the characteristics of the comparison communities and their broadband offerings with those in the United States, we present country rankings by two additional methodologies: a broadband price index41 and a hedonic price index.42 Our additional assessments seek to better assess how the U.S. market is performing relative to other markets after accounting for quality differences as well as market-level cost and demographic differences that are known to affect pricing, such as population density, income, and education levels. The hedonic price index also allows an adjustment for observable differences in broadband quality across countries (e.g., speed and usage limits) and generates prices for a set of standardized broadband plans in every country to produce a price index that accounts for all of these factors and is comparable across countries.43 For example, factors affecting rankings include income, where the United States has the 6th highest income of the 29 countries, and urban population density, where the United States is the 8th lowest.44 As discussed below, the broadband and hedonic price index analysis demonstrate that accounting for country differences in cost, demographic, and quality factors give different assessments of the state of the U.S. broadband economy relative to other countries. 14. Fixed Broadband Pricing. For fixed broadband prices, under the first method comparing unweighted average prices, the United States ranks 18th out of 23 countries that offer fixed standalone broadband plans with download speeds of at least 25 Mbps and less than 100 Mbps, and 26th out of 28 countries that have fixed standalone plans with download speeds of 100 Mbps or greater.45 When taking into account fixed broadband bundled with video service, the United States ranks 10th out of 20 countries with download speeds of at least 25 Mbps and less than 100 Mbps.46 For the highest speed bundle plans with download speeds of 100 Mbps or greater, fixed broadband in the United States ranks 23rd out of 25 countries that offer such plans.47 Using the second approach, the fixed broadband price index analysis, the United States ranks 21st out of 29 countries aggregating both standalone and bundled broadband products.48 However, using the third approach, the fixed hedonic price index analysis that adjusts for after usage (i.e., not prepaid or “pay-as-you-go” plans), and smartphone plans refer only to plans that have a data component. See infra Appx. C at paras. 14-15 and 40, note 84. 41 The price index measures the dollar amount that U.S. broadband subscribers would need to have added or subtracted from their incomes to purchase the same basket of broadband services under the pricing structures in other countries. Quantity weights for our price index are the share of broadband subscribers in the United States that subscribe to each of the four broadband speed tiers we have chosen for analysis. See id. at paras. 23-28. 42 A hedonic regression provides an empirical summary of how prices vary with the characteristics of a good. Our hedonic regression builds on the price index by allowing adjustment of prices for cost and demographic differences across countries and then predicting broadband prices for each country at the average U.S. values of these variables. Id. at paras. 29-32. 43 The pricing analysis is designed to account for: (1) the different costs of deploying and operating broadband networks; (2) demographic differences that affect demand for broadband service; (3) multi-product bundling in broadband pricing; (4) different product offerings in each country; and (5) the availability and quality of complementary content and applications. Id. at para. 7. 44 See id. at paras. 30, 70, Tbl. C1. 45 Id. at para. 22, Tbl. 1b. 46 Id. at para. 70, Tbl. C6. 47 Id. 48 Id. at paras. 4, 28, Tbl. 3. (continued….) Federal Communications Commission DA 18-99 8 cost, demographic, and quality differences across the countries, shows that the United States ranks 7th out of the 29 countries.49 15. Mobile Broadband Pricing. For mobile broadband prices, under the first method, the United States ranks 18th out of the 22 sampled countries based on unweighted average prices of individual plans with data usage allowances of 2 GB or less.50 For the highest usage individual plans with data usage allowances greater than 10 GB, the United States ranks 21st out of the 28 countries that offer such plans.51 According to the second method, the mobile broadband price index, the United States ranks 25th out of the 29 countries in individual plan pricing, and 18th out of the 29 countries in shared data plan pricing (i.e., “family plans” with multiple lines).52 Combining individual and shared data plan pricing, the overall rank of the United States is 20th out of the 29 countries in the mobile broadband price index.53 Relying on the third approach, the mobile hedonic price index that adjusts for country-level cost, demographic, and quality differences, the United States ranks 10th out of the 29 countries.54 C. High-Speed Broadband Deployment Comparison with Europe 16. In Appendix D, as directed by the BDIA, we compare “the extent of broadband service capability” by examining international fixed high-speed broadband deployment55 in the United States and 21 European countries (EU21).56 We rely on data gathered in June 2015 and June 2016 by the FCC and the European Commission (EC). For the first time, we also compare mobile high-speed broadband deployment in the EU21 and the United States by focusing exclusively on LTE, which is the baseline industry standard for the marketing of mobile broadband service.57 We examine U.S. deployment of fixed 49 Id. at paras. 4, 32, Tbl. 4. 50 Id. at paras. 5, 47, Tbl. 5 51 Id. 52 Id. at paras. 5, 52, Tbl. 7. 53 Id. 54 Id. at paras. 5, 54, Tbl. 8. 55 47 U.S.C. § 1303(b)(1) (“As part of the assessment and report required by section 1302 of this title, the Federal Communications Commission shall include information comparing the extent of broadband service capability (including data transmission speeds and price for broadband service capability) in a total of 75 communities in at least 25 countries abroad for each of the data rate benchmarks for broadband service utilized by the Commission to reflect different speed tiers.”). 56 We refer to the set of countries that we compare here as the EU21 countries, as we selected 21 of the 31 European countries addressed in the EC Broadband Report as comparison countries for purposes of this Report. See European Commission, Broadband Coverage in Europe 2016: Mapping Progress Towards the Coverage Objectives of the Digital Agenda at 5 (2017) (EC Broadband Report), https://ec.europa.eu/digital-single-market/en/news/study- broadband-coverage-europe-2016. The EC Broadband Report discusses the 28 member countries of the EU, as well as Iceland, Norway, and Switzerland. Our deployment comparison in this Report assesses the 21 countries that overlap with the European countries selected for the comparison overall. The 21 countries we include in our analysis are: Austria (AT), Belgium (BE), Czech Republic (CZ), Denmark (DK), Estonia (EE), Finland (FI), France (FR), Germany (DE), Greece (EL), Ireland (IE), Italy (IT), Latvia (LV), Luxembourg (LU), Netherlands (NL), Portugal (PT), Spain (ES), Sweden (SE), United Kingdom (UK), Iceland (IS), Norway (NO), and Switzerland (CH). 57 Implementation of Section 6002(b) of the Omnibus Budget Reconciliation Act of 1993; Annual Report and Analysis of Competitive Market Conditions with Respect to Mobile Wireless, Including Commercial Mobile Services, WT Docket No. 17-69, Twentieth Report, 32 FCC Rcd 8968, 9018, para. 73 (2017) (Twentieth Mobile Wireless Competition Report). We note that the 2018 Broadband Deployment analyzes mobile LTE coverage data associated with 5 Mbps/1 Mbps and higher minimum advertised speeds in the United States and supplements that data with actual on the ground 10 Mbps/3 Mbps and higher median speed data measurements. In Appendix D, we (continued….) Federal Communications Commission DA 18-99 9 broadband with download speeds 30 Mbps or higher to match the European definition of high-speed broadband.58 To match the fixed technologies used in the EC Broadband Report, we do not include satellite technology in the comparison of the United States and the EU21. 17. Fixed High-Speed Broadband Deployment Results. In the United States, fixed high- speed broadband reached 90 percent of all households and 62 percent of rural households by June 2016,59 which is up from 89 percent of all households and 58 percent of rural households as of June 2015.60 In the EU21, fixed high-speed broadband reached 76 percent of all households and 41 percent of rural households by June 2016,61 which is up from 72 percent of all households and 30 percent of rural households as of June 2015.62 18. The U.S. deployment gap between non-rural and rural areas decreased from 39 percentage points as of June 2015 to 35 percentage points as of June 2016.63 In the EU21, the gap between non-rural and rural areas decreased from 48 percentage points as of June 2015 to 42 percentage points as of June 2016.64 19. The United States ranked 11th out of the 22 countries in 2015 and 10th out of the 22 countries in 2016 with respect to fixed high-speed broadband deployment.65 The United States ranked 10th out of the 22 countries in 2015 and 9th out of the 22 countries in 2016 in terms of its rural coverage of high-speed broadband.66 20. Mobile High-Speed Broadband Deployment Results. In the United States, mobile LTE coverage was widely deployed by the end of June 2016, reaching over 99 percent of all households and 98 percent of rural households.67 In the EU21, during the same period, mobile LTE coverage reached 97 percent of all households and 83 percent of rural households.68 21. Historical Overview. Our historical overview for 2012 to 2016 shows that the United States had higher deployment rates than the EU21 countries as a whole during the period both generally, analyze mobile LTE coverage regardless of minimum advertised speeds or actual speeds to match the EC Broadband Report. 58 EC Broadband Report at 11. This standard differs from that used for the fixed broadband deployment data for the United States in the 2018 Broadband Deployment Report. That Report shows deployment of fixed broadband with 25 Mbps download and 3 Mbps upload speed. In addition, we assess deployment as of June 2016 and June 2015 to match the European data, while the 2018 Broadband Deployment Report measures deployment as of December 2016 and December 2015. 59 See infra Appx. D, paras. 12, 14, Figs. 2, 4. 60 Id. at paras. 11, 13, Figs. 1, 3. 61 Id. at, paras. 12, 14, Figs. 2, 4. 62 Id. at paras. 11, 13, Figs. 1, 3. 63 See id. at para. 16, Figs. 5, 6. 64 Id. 65 See id. at para. 17, Figs. 7, 8. 66 Id. at para. 18, Figs. 9, 10. 67 See id. at para. 21, Figs. 17, 18. 68 Id. (continued….) Federal Communications Commission DA 18-99 10 and separately in rural and non-rural areas.69 Deployment increased during the period, with the EU21 countries having a somewhat higher growth rate.70 D. Demographics Dataset 22. The BDIA directs the Commission to compare broadband development in international communities comparable to U.S. communities in terms of population size, density, topography, and demographic profile.71 The dataset in Appendix E contains information for, in the aggregate, almost 300 province/county communities. We present updated demographics data for the United States and the 28 comparison countries on a sub-national basis, including the latest figures for such indicators as population size, population density, gross domestic product (GDP), and educational attainment.72 Consistent with our approach in previous reports, we provide the most recent publicly available data for each variable in the community dataset in Appendix E. We present data available from the OECD for all countries in our sample, except for the United States and Canada, based on available data ranging from 2012 to 2016.73 Data for the United States and Canada were obtained from other sources.74 We also present topography information for the United States and the comparison countries based on information from the Central Intelligence Agency’s The World Factbook.75 E. Market and Regulatory Developments 23. In Appendix F, as required by the BDIA, we identify the relevant similarities and differences between the United States and the 28 comparison countries with respect to multiple criteria.76 First, we discuss the regulatory models for fixed broadband deployment. Second, we provide a list of regulators and, where relevant, the ministries responsible for regulating broadband. Third, we provide information concerning the major fixed and mobile broadband competitors and the types of technologies used to provide broadband. Finally, we present data on the types of activities that consumers in the United States and the comparison countries engage in while accessing the Internet. We rely on several 69 See id. at para. 22, Figs. 19, 20. 70 Id. 71 47 U.S.C. § 1303(b)(2). 72 Using this sub-national data, one can draw comparisons across both international and domestic cities, states, and regions. As is the case in the United States, intra-country variations are greater than the inter-country differences. In particular, differences in population density, dispersion, and income may create significant variations. For example, the lower population density and greater size of the United States present unique challenges for broadband deployment. 73 OECD, OECD.Stats: Regions and Cities (OECD Regions and Cities), http://stats.oecd.org/ (last visited Jan. 16, 2018). 74 For the United States, we present 2016 data from the U.S. Census Bureau. For Canada, we present 2016 data from the Canadian Radio-television and Communications Commission. U.S. Census Bureau, Percent Of Households With A Broadband Internet Subscription (U.S. Census Households With Broadband Subscription), https://factfinder.census.gov/faces/tableservices/jsf/pages/productview.xhtml?pid=ACS_16_1YR_GCT2801.US01P R&prodType=table (last visited Jan 16, 2018); Canadian Radio-television and Communications Commission, Communications Monitoring Report at 279 (2017) (Canada Communications Monitoring Report), https://crtc.gc.ca/eng/publications/reports/PolicyMonitoring/2017/cmr2017.pdf. 75 Central Intelligence Agency, The World Factbook (last updated Jan. 3, 2018) (CIA World Factbook), https://www.cia.gov/library/publications/resources/the-world-factbook/. 76 47 U.S.C. § 1303(b)(3) (“The Commission shall identify relevant similarities and differences in each community, including their market structures, the number of competitors, the number of facilities-based providers, the types of technologies deployed by such providers, the applications and services those technologies enable, the regulatory model under which broadband service capability is provided, the types of applications and services used, business and residential use of such services, and other media available to consumers.”). (continued….) Federal Communications Commission DA 18-99 11 sources, such as the TeleGeography GlobalComms Database77 and the EC’s Digital Economy and Society Index for this information.78 24. Fixed Broadband Regulatory Models. Based on our analysis, there are two basic fixed broadband regulatory models—a facilities-based competition model and an open access regulatory model—with some blending of the two models in most countries. There is also significant variation in regulatory approaches for implementing the same model. The facilities-based competition model relies on competition between the incumbent telecommunications operator and other operators (e.g., cable). The open access regulatory model features the use of mandated wholesale access to the incumbent’s fixed network to create service-based competition by encouraging market entry at the retail level. Under either model, investment in fixed broadband networks may be primarily market- or state-aid-driven, depending on the extent to which governments subsidize network deployment. 25. National Broadband Developments. Many countries around the world recognize the fundamental role that broadband plays in helping to achieve social and economic development and continue to take steps to expand broadband access and use. According to the International Telecommunication Union (ITU), as of September 2017, 156 countries have introduced national broadband plans.79 Many countries are developing comprehensive digital agendas80 that go beyond conventional supply-side policies designed to expand broadband coverage and also include demand-side policies to promote affordability, Information and Communication Technology skills, and e-government, among other cross-cutting issues.81 26. Although approaches to national digital strategies vary significantly across countries, many governments continue to establish measurable targets to monitor broadband progress.82 For example, the EU’s Digital Single Market strategy calls on EU Member States to provide download speeds of 30 Mbps for all citizens and ensure that 50 percent of households have subscriptions of greater than 100 Mbps by 2020.83 Moreover, by 2025, all schools, transport hubs, and main providers of public services should have access to Internet connections with download speeds of 1 Gbps, and all European households should have access to networks offering download speeds of at least 100 Mbps.84 Finland, for its part, recently began the process of updating its national broadband strategy to redefine new speed and availability targets for 2025 and 2030.85 77 TeleGeography GlobalComms Database (TeleGeography GlobalComms Database), www.telegeography.com (last visited Jan. 16, 2018). 78 European Commission, Digital Single Market: Digital Economy & Society Index (DESI) 2017 (2017), https://ec.europa.eu/digital-single-market/en/news/digital-economy-and-society-index-desi-2017. 79 Broadband Commission for Sustainable Development, The State of Broadband: Broadband Catalyzing Sustainable Development at 42 (2017) (State of Broadband Report 2017), http://www.broadbandcommission.org/publications/Pages/SOB-2017.aspx. 80 Id. at 44; see also OECD, OECD Digital Economy Outlook 2017 at 34 (2017) (noting that “[national digital strategies] have become the norm across OECD countries”) (OECD Report 2017), http://www.oecd.org/internet/oecd-digital-economy-outlook-2017-9789264276284-en.htm. 81 State of Broadband Report 2017 at 44. 82 OECD Report 2017 at 37. 83 European Commission, Digital Single Market: Broadband Strategy & Policy, https://ec.europa.eu/digital-single- market/en/broadband-strategy-policy (last visited Jan. 16, 2018). 84 Id. 85 TeleGeography CommsUpdate, Finnish Communications Ministry Begins Preparing New National Broadband Strategy (July 13, 2017), https://www.telegeography.com/products/commsupdate/articles/2017/07/13/finnish- communications-ministry-begins-preparing-new-national-broadband-strategy/. (continued….) Federal Communications Commission DA 18-99 12 27. 5G Development. Approaches to 5G development and the status of development efforts vary across countries. The United States has taken steps to make available low-, mid-, and high-band spectrum for next-generation wireless networks. For instance, in 2017, the Commission completed the auction for new 600 MHz licenses86 and granted licenses for the 600 MHz Band (low-band).87 On August 3, 2017, the Commission adopted a Notice of Inquiry seeking comment on ways to expand opportunities for next generation services—particularly wireless broadband services—in spectrum bands between 3.7 and 24 GHz (i.e., mid-band spectrum).88 On November 16, 2017, the Commission adopted a Report and Order that, among other things, makes available 1,700 megahertz of millimeter wave (mmW) spectrum for terrestrial next generation wireless use, including 5G (high-band spectrum),89 which was in addition to the nearly 11 gigahertz of spectrum opened up in July 2016 for wireless broadband operations in frequencies above 24 GHz (high-band spectrum).90 Finally, we note that, in April 2016, the Commission created a new Citizens Broadband Radio Service, establishing an innovative three-tier framework to enable shared wireless broadband use of the 3550-3700 MHz band (3.5 GHz band);91 the Commission is currently considering several rule changes that would further facilitate the implementation of 5G networks in this band.92 86 Incentive Auction Closing and Channel Reassignment Public Notice; The Broadcast Television Incentive Auction Closes; Reverse Auction and Forward Auction Results Announced; Final Television Band Channel Assignments Announced; Post-Auction Deadlines Announced, AU Docket No. 14-252 et al, Public Notice, 32 FCC Rcd 2786 (2017). The Commission released a Public Notice on April 13, 2017, marking the completion of the incentive auction and the start of a 39-month post-auction transition period. Id. 87 Incentive Auction Task Force and Wireless Telecommunications Bureau Grant 600 MHz Licenses; Auction No. 1002, Auction No. 1002, Public Notice, 32 FCC Rcd 4832 (2017). 88 Expanding Flexible Use in Mid-Band Spectrum Between 3.7 and 24 GHz, GN Docket No. 17-183, Notice of Inquiry, 32 FCC Rcd 6373 (2017). 89 Use of Spectrum Bands Above 24 GHz For Mobile Radio Services, GN Docket No. 14-177, et al., Second Report and Order, Second Further Notice of Proposed Rulemaking, Order on Reconsideration, and Memorandum Opinion and Order, FCC 17-152 (rel. Nov. 22, 2017). 90 Use of Spectrum Bands Above 24 GHz For Mobile Radio Services et al., GN Docket No. 14-177 et al., Report and Order and Further Notice of Proposed Rulemaking, 31 FCC Rcd 8014 (2016); Press Release, FCC, FCC Takes Steps to Facilitate Mobile Broadband and Next Generation Wireless Technologies in Spectrum Above 24 GHz (July 14, 2016), https://apps.fcc.gov/edocs_public/attachmatch/DOC-340301A1.pdf. 91 Amendment of the Commission’ s Rules with Regard to Commercial Operations in the 3550- 3650 MHz Band, Report and Order and Second Further Notice of Proposed Rulemaking, GN Docket No. 12-354, 30 FCC Rcd 3959 (2015); Press Release, FCC, FCC Puts Final Rules in Place for New Citizens Broadband Radio Service (Apr. 28, 2016), https://apps.fcc.gov/edocs_public/attachmatch/DOC-339104A1.pdf. 92 Promoting Investment in the 3550-3700 MHz Band; Petitions for Rulemaking Regarding the Citizens Broadband Radio Service, GN Docket No. 17-258, Notice of Proposed Rulemaking and Order Terminating Petitions, FCC 17- 134 (rel. Oct. 24, 2017); Press Release, FCC, FCC Seeks Comment on Promoting Investment in the 3.5 Ghz Band (Oct. 24, 2017), https://apps.fcc.gov/edocs_public/attachmatch/DOC-347378A1.pdf. (continued….) Federal Communications Commission DA 18-99 13 28. Other countries, including Austria,93 Canada,94 the Czech Republic,95 and France,96 among others, are in the midst of allocating additional spectrum across various bands to support 5G services. In Japan, with guidance from the Ministry of Internal Affairs & Communications (MIC), Japan’s three wireless operators—NTT DOCOMO, KDDI, and SoftBank Mobile—are committed to launching commercial 5G services by the Tokyo Summer Olympics in 2020.97 In partnership with industry, the South Korean government aims to introduce trial 5G services by the 2018 Winter Olympics in PyeongChang and to launch full commercial 5G services by December 2020. In January 2017, South Korea’s Ministry of Science, ICT and Future Planning (MSIP) announced plans to allocate an additional 1,300 megahertz of spectrum for 5G services by 2018.98 29. In September 2016, the EC adopted a 5G Action Plan, which encouraged EU Member States to develop national 5G roadmaps by the end of 2017, and to identify at least one major city to be “5G-enabled” by the end of 2018, with the goal of expanding 5G coverage to all urban areas and major transportation hubs by 2025.99 The plan also sought to identify “pioneer spectrum bands” to harmonize the initial launch of 5G services across Europe.100 30. Other countries are also in the process of developing policy and regulatory frameworks to facilitate the development of next-generation wireless technologies. For example, in March 2017, the United Kingdom (UK) issued a national 5G strategy focused on three main outcomes: accelerating the deployment of 5G networks; maximizing the productivity and efficiency benefits to the UK from 5G; and creating new opportunities for UK businesses at home and abroad, and encouraging inward investment.101 93 See Austrian Regulatory Authority for Broadband and Telecommunications (RTR), Award Procedure 3.4-3.8 GHz (2018), https://www.rtr.at/en/tk/FRQ5G_2018 (last visited Jan 16, 2018). 94 See Innovation, Science and Economic Development Canada, Consultation on Releasing Millimetre Wave Spectrum to Support 5G (June 2017), http://www.ic.gc.ca/eic/site/smt-gst.nsf/eng/sf11298.html (last visited Jan. 16, 2018). 95 See Czech Telecommunication Office (CTU), End of Auction for Radio Frequencies in 3,7 GHz Band and Allocation of Frequency Ranges (Aug. 30, 2017), https://www.ctu.eu/end-auction-radio-frequencies-37-ghz-band- and-allocation-frequency-ranges; see also, e.g., Sarah McBride, Czech Republic Auction Sees a Surge in Demand for 5G Spectrum (Aug. 1, 2017), https://ovum.informa.com/resources/product-content/te0007-001171. 96 See Arcep, Consultation Publique du 13 juillet 2017 au 7 septembre 2017: Attribution de fréquences de la bande 3410 - 3460 MHz pour le très haut débit radio en France métropolitaine (July 13, 2017), https://www.arcep.fr/uploads/tx_gspublication/consult-attribution-THD_radio-juil2017.pdf. 97 See, e.g., Yuki Nakamura, Ministry of Internal Affairs & Commc’ns, Japan’ s 5G Policy Perspectives (Nov. 9, 2016), https://5g-ppp.eu/wp-content/uploads/2016/11/Opening-1_Yuji-Nakamura.pdf. 98 TeleGeography CommsUpdate, Korean Ministry Looks to Boost 5G Bandwidth (Jan. 19, 2017), https://www.telegeography.com/products/commsupdate/articles/2017/01/19/korean-ministry-looks-to-boost-5g- bandwidth/. 99 European Commission, 5G for Europe: An Action Plan at 4 (2016), https://ec.europa.eu/digital-single- market/en/news/communication-5g-europe-action-plan-and-accompanying-staff-working-document. 100 Id. at 5. Subsequently, in November 2016, the Radio Spectrum Policy Group (RSPG) considered various spectrum bands and identified the 3400-3800 MHz band as the primary band suitable for the introduction of 5G services in Europe. See Radio Spectrum Policy Group, Strategic Roadmap Towards 5G for Europe: Opinion of Spectrum-Related Aspects for Next-Generation Wireless Systems (5G) (2016), http://rspg-spectrum.eu/wp- content/uploads/2013/05/RPSG16-032-Opinion_5G.pdf. 101 See Department for Culture, Media & Sport, Next Generation Mobile Technologies: A 5G Strategy for the UK (2017), https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/597421/07.03.17_ 5G_strategy_-_for_publication.pdf. (continued….) Federal Communications Commission DA 18-99 14 In July 2017, Spain launched a public consultation on the development of a national 5G plan.102 In October 2017, Australia introduced a 5G strategy, outlining government actions to support the timely rollout of 5G services.103 31. At the international level, the ITU Radiocommunication Sector (ITU-R) and Regional Groups continue to focus on the harmonization of spectrum bands and their conditions of use in order to promote the timely implementation of technological changes and the benefits of economies of scale. The ITU-R has created an agenda item for the 2019 World Radiocommunication Conference (WRC-19) to consider spectrum for IMT-2020 (5G) in several frequency ranges above 24 GHz.104 In addition, administrations and industry are participating in regional bodies committed to the advancement of 5G and building consensus towards standardization, spectrum, research and development, and cooperation with other strategic industry sectors. Some examples include the 5G Infrastructure Public Private Partnership (5G PPP),105 the 5G Forum,106 and 5G Americas,107 which host meetings and summits to harmonize global efforts. F. Other Relevant Information 1. Broadband Subscription (OECD Countries) 32. We report fixed and mobile broadband subscription in the United States and in OECD countries below. The OECD’s subscription metrics define transmission speeds of at least 256 kbps in one direction as “broadband service” for both fixed and mobile Internet access.108 This is considerably slower than the 25 Mbps download/3 Mbps upload speeds which the Commission considers as “advanced telecommunications capability.”109 In this section, we use the OECD’s broadband definition to present comparable subscription statistics from OECD countries, including the United States. As the OECD notes, subscription is measured using different indicators and different reference dates across various countries and where a particular country falls in these rankings may be influenced by population density and dispersion, income, and other factors.110 We provide an update on international efforts to improve data on broadband below. 33. As the most populous member of the OECD, the United States ranked first in 2016 in the sheer number of fixed broadband subscriptions with approximately 106,327,000 subscriptions.111 In 102 Ministry of Energy, Tourism & the Digital Agenda (MINETAD), Digital Agenda Opens a Public Consultation on the National Plan of 5G (July 6, 2017), http://www.minetad.gob.es/en- US/GabinetePrensa/NotasPrensa/2017/Paginas/consulta-publica-5g20170706.aspx. 103 See Australian Government Department of Communications and the Arts, 5G—Enabling the Future Economy (Oct. 2017), https://www.communications.gov.au/documents/5g-enabling-future-economy. 104 See State of Broadband Report 2017 at 21, 23. 105 See generally 5G Infrastructure Public Private Partnership, About the 5GPPP, https://5g-ppp.eu/ (last visited Jan. 16, 2018). 106 See generally 5G Forum, 5G Forum, https://www.5gforum.org/english (last visited Jan. 16, 2018). 107 See generally 5G Americas, 5G Americas, http://www.5gamericas.org/en/ (last visited Jan. 16, 2018). 108 OECD Broadband Methodology; see also supra note 19. 109 2018 Broadband Deployment Report at para. 15. 110 The United States has about one-quarter the population density of Europe, one-tenth that of Japan, and one- fifteenth that of South Korea. 111 OECD, Broadband Portal, Total Fixed and Wireless Broadband Subscriptions by Country, Table 1.1 (Dec. 2016) (OECD Broadband Subscriptions by Country Table 1.1), http://www.oecd.org/sti/broadband/1.1-TotalBBSubs-bars- 2016-12.xls. The OECD fixed broadband subscription data include fixed broadband subscriptions with downloads (continued….) Federal Communications Commission DA 18-99 15 2015, the number of fixed broadband subscriptions in the United States was approximately 100,865,000.112 By comparison, Japan ranked second in 2016 with 38,743,212 fixed subscriptions.113 In 2015, the number of fixed broadband subscriptions in Japan was 37,610,780.114 With respect to penetration, measured by the number of subscriptions per 100 inhabitants, the United States ranked 16th out of 35115 countries in 2016 for percentage of population with overall fixed broadband subscriptions, with 32.8 broadband subscriptions per 100 inhabitants.116 Switzerland ranked first in 2016 in fixed broadband subscription penetration with 50.1 subscriptions per 100 inhabitants.117 In 2015, the United States ranked 16th of 35 countries with 31.77 fixed broadband subscriptions per 100 inhabitants.118 34. In terms of sheer number of mobile broadband subscriptions,119 the United States ranked first out of OECD’s 35 countries with approximately 409,173,000 subscriptions with data plans as of December 2016120 (compared to approximately 375,504,000 subscriptions as of December 2015 for the first place rank).121 By comparison, Japan ranked second with 193,237,268 wireless broadband subscriptions in 2016 (also second with 176,010,204 in 2015).122 The OECD’s 2016 subscription data also rank countries according to the number of subscriptions per 100 inhabitants.123 By this metric, the United States ranked fourth out of the 35 OECD countries in 2016 in mobile broadband subscriptions, with 126.3 mobile broadband subscriptions per 100 inhabitants124 (by comparison, Japan ranked first in speeds that are at least 256 kbps (DSL, satellite, terrestrial fixed wireless) or greater than 256 kbps (cable, fiber, broadband over power lines, leased lines); OECD Broadband Methodology. 112 The 2015 data are no longer available on the OECD Broadband Portal. OECD, Broadband Portal, Table 1.1.1 (Dec. 2015). 113 OECD Broadband Subscriptions by Country Table 1.1. 114 The 2015 data are no longer available on the OECD Broadband Portal. OECD, Broadband Portal, Table 1.1.1 (Dec. 2015). 115 Since the release of the Fifth International Broadband Data Report, the OECD added Latvia as a Member country, increasing the number of OECD countries to 35. OECD, Latvia’s Accession to the OECD, https://www.oecd.org/latvia/latvia-accession-to-the-oecd.htm (last visited Jan. 16, 2018). 116 OECD Broadband Subscriptions Table 1.2. 117 Id. 118 See OECD, Broadband Portal, Historical Time Series: Fixed and Wireless Broadband Penetration, Table 1.5 (Dec. 2016) (OECD Historical Broadband Penetration Table 1.5), http://www.oecd.org/sti/broadband/1.5- BBPenetrationHistorical-Data-2016-12.xls. 119 The OECD mobile broadband subscription data include: (1) “data and voice subscriptions,” including “mobile subscriptions which advertise data speeds of 256 kbit/s or greater and which have been used to make an Internet data connection via IP in the previous 3 months,” and (2) “data-only subscriptions” which are defined as “subscriptions to dedicated data services over a mobile network which are purchased separately from voice services either as a stand-alone service (modem/dongle) or as an add-on data package to voice services which requires an additional subscription.” OECD Broadband Methodology. 120 OECD Broadband Subscriptions by Country Table 1.1. 121 The 2015 data are no longer available on the OECD Broadband Portal. OECD, Broadband Portal, Table 1.1.2 (Dec. 2015). 122 Id. 123 OECD Broadband Subscriptions Table 1.2. 124 Id. (continued….) Federal Communications Commission DA 18-99 16 2016 with 152.4 mobile broadband subscriptions per 100 inhabitants).125 In 2015, the United States ranked sixth out of 35 countries with 116.7 mobile broadband subscriptions per 100 inhabitants.126 2. Efforts to Improve International Broadband Data 35. As we indicated in the previous reports, available data on international broadband are incomplete and generally challenging to compare because of significant gaps and variations in data collection methodologies across countries. In previous reports, we detailed the steps that the OECD and U.S. Government have taken to standardize broadband metrics.127 Since then, the OECD and U.S. Government have continued to work to standardize broadband metrics, particularly in the area of hedonic price broadband pricing analysis.128 OECD delegates published a final paper in July 2016 entitled, “Hedonic Prices for Fixed Broadband Services: Estimation Across OECD Countries,” which refined the hedonic methodology using a more robust dataset.129 The OECD also created a “beta-test” set of eight OECD members that will collect and apply the new methodology using pricing data for broadband bundles (including video services). The eight members are: Australia, Hungary, Israel, Mexico, Portugal, Spain, South Korea, and the United States. In November 2017, the initial results were previewed for OECD delegates; an initial draft paper, the finding of which OECD delegates will discuss and debate, is anticipated in May 2018. The OECD, with the support of the U.S. Government and other delegations, will continue to devote resources to the ongoing review and reform of its core broadband data collection efforts, among other initiatives, including refurbishment of the OECD’s current “Broadband Portal”130 in 2018.131 V. CONCLUSION 36. In conjunction with the Commission’s adoption of the 2018 Broadband Deployment Report, the release of this Report fulfills the obligation imposed by Section 103(b) of the BDIA.132 125 Id. 126 See OECD Historical Broadband Penetration Table 1.5. 127 Fifth International Broadband Data Report, 31 FCC Rcd at 2681, para. 45. 128 The OECD has been using a “price basket” methodology for comparing communication prices across countries for over 20 years. As concluded by the OECD project (Towards a New OECD Metrics Checklist), hedonic price analysis could provide a new perspective. Hedonic models are based on the idea that products or services can be viewed as a bundle of characteristics that are valued by both buyers and sellers. Price represents the value of characteristics of the products or services. See, e.g., Jack E. Triplett, Economic Interpretation of Hedonic Methods, Survey of Current Business, Bureau of Economic Analysis, Department of Commerce at 36-40 (Jan. 1986); see also OECD, Glossary of Statistical Terms (last updated Jul. 8, 2005), http://stats.oecd.org/glossary/detail.asp?ID=1225. 129 In June 2015, the OECD, via a contract with Dr. Carol Corrado of Georgetown University in Washington, agreed to further refine the hedonic methodology using a more robust dataset. The initial findings were presented in a draft paper in June 2015 for initial review by OECD delegates, and a follow-up expert peer review of this work was conducted at the FCC in September 2015, with over 30 attendees from the FCC, U.S. Government, World Bank, and Canadian and Mexican regulatory officials collaborating on a set of final suggestions and observations for the authors to consider as they finalized the OECD analysis. The OECD published this paper in July 2016. Carol Corrado and Olga Ukhaneva, Hedonic Prices for Fixed Broadband Services: Estimation Across OECD Countries (2016), http://www.oecd-ilibrary.org/science-and-technology/hedonic-prices-for-fixed-broadband- services_5jlpl4sgc9hj-en?crawler=true. 130 OECD, Broadband Portal, http://www.oecd.org/sti/broadband/oecdbroadbandportal.htm (last visited Jan. 16, 2018). 131 OECD, Going Digital: Making the Transformation Work for Growth and Well-Being, http://www.oecd.org/going-digital/project (last visited Jan. 16, 2018). 132 47 U.S.C. § 1303(b). Federal Communications Commission DA 18-99 17 VI. PROCEDURAL MATTERS 37. IT IS ORDERED that, pursuant to Section 103(b) of the Broadband Data Improvement Act, 47 U.S.C. § 1303(b), and pursuant to authority delegated to the International Bureau in Sections 0.51 and 0.261 of the Commission’s rules, 47 CFR §§ 0.51, 0.261, this Report, with its associated Appendices A-F, is ADOPTED. FEDERAL COMMUNICATIONS COMMISSION Thomas Sullivan Chief, International Bureau Federal Communications Commission DA 18-99 18 APPENDICES APPENDIX A: Country List APPENDIX B: Broadband Speed Comparison APPENDIX C: Broadband Price Comparison APPENDIX D: High-Speed Broadband Deployment Comparison with Europe APPENDIX E: Demographics Dataset APPENDIX F: Market and Regulatory Developments Federal Communications Commission DA 18-99 19 APPENDIX A Country List 1. In the Table below, we list the United States and the 28 foreign countries selected for purposes of this Report and identify the countries that are excluded in an Appendix with an “X”. Countries Appendix B Broadband Speed Comparison Appendix C Broadband Price Comparison Appendix D High-Speed Broadband Deployment Comparison with Europe Appendix E Demographics Dataset Appendix F Market and Regulatory Developments Australia X Austria Belgium Canada X Chile X Czech Republic Denmark Estonia Finland France Germany Greece Iceland Ireland Italy Japan X Latvia X Luxembourg Mexico X Netherlands New Zealand X Norway Portugal South Korea X Spain Sweden Switzerland United Kingdom United States X Federal Communications Commission DA 18-99 20 APPENDIX B Broadband Speed Comparison 1. As required by the BDIA, we present information on “data transmission speeds” for broadband service capability1 for both fixed and mobile broadband. This information is based on actual speed data collected by Ookla.2 The data include observations in 2014, 2015, and 2016 and include both U.S. and international cities.3 As a historical overview, we also present available data on U.S. fixed download speeds and rankings from 2012 to 2016, which show how actual speeds have evolved over time. I. DATA HIGHLIGHTS 2. Fixed Broadband Speed Results. With respect to fixed broadband speeds, the United States ranked 10th out of 28 countries in 2016 (55.07 Mbps) in terms of actual download speeds4 weighted by the number of tests in each city—an improvement over 11th in 2015 (40.38 Mbps) and 15th in 2014 (28.09 Mbps). Actual fixed download speeds in the United States increased by approximately 96 percent from 2014 to 2016. 3. In all three years, Luxembourg had the highest mean fixed broadband speed, and Greece had the lowest. Luxembourg’s mean fixed broadband speeds were 375.78 Mbps in 2016, 344.40 Mbps in 2015, and 222.13 Mbps in 2014. By contrast, Greece’s mean fixed broadband speeds were 11.83 Mbps in 2016, 9.52 Mbps in 2015, and 8.90 Mbps in 2014. However, Luxembourg is an outlier—the mean fixed broadband speed in Luxembourg was more than three times the mean speed in the second-ranked country in 2016 (Japan at 102.34 Mbps), and more than four times the mean speed in the second-ranked countries in 2015 (Japan at 81.50 Mbps) and 2014 (South Korea at 53.15 Mbps). 4. In our historical overview, we find, based on the available data, that fixed speeds for the United States have been on a rising trend since 2012, and that its rank among the selected countries has been on a rising trend since 2012. 5. Mobile Broadband Speed Results. With regard to mobile broadband speeds, the United States ranked 24th out of 28 countries in both 2016 (19.98 Mbps) and 2015 (15.58 Mbps), and 18th in 2014 (12.62 Mbps), based on a comparison of mean download speeds. Actual mobile download speeds in the United States increased by approximately 58 percent from 2014 to 2016. 6. In 2016, mean mobile download speeds ranged from a high of 39.19 Mbps in South Korea to a low of 15.24 Mbps in Chile. The highest-ranked country in the previous two years was New 1 47 U.S.C. § 1303(b)(1) (“As part of the assessment and report required by section 1302 of this title, the Federal Communications Commission shall include information comparing the extent of broadband service capability (including data transmission speeds and price for broadband service capability) in a total of 75 communities in at least 25 countries abroad for each of the data rate benchmarks for broadband service utilized by the Commission to reflect different speed tiers). 2 We obtained speed data through a contractual arrangement with Ookla. Ookla Speedtest. 3 In this Appendix, due to data limitations, we are unable to replicate some of the analysis in the Fifth International Broadband Data Report, including jitter, packet loss comparisons, or compare the percent difference between advertised and actual speeds, as this year’s Ookla dataset did not include the data. In addition, in contrast to past reports, we are unable to present hardware-based speed test measurements in the United States and the European Union (EU), as the European Commission (EC) has not released a Quality of Broadband Services in the EU covering either 2015 or 2016. See Fifth International Broadband Data Report, 31 FCC Rcd at 2801-48, Appx. F. 4 In this Appendix, “actual speed” refers to mean actual speed unless otherwise specified. We chose mean speeds over median speeds due to the data limitations. The means are weighted by the number of tests performed in each city. (continued….) Federal Communications Commission DA 18-99 21 Zealand, with a mean mobile download speed of 27.85 Mbps in 2015 and 28.09 Mbps in 2014. Thus, the mean mobile download speed of the highest-ranked country rose significantly in 2016 after remaining roughly flat in the previous two years. II. DATA SOURCES 7. Broadband speeds are often illustrated by either advertised speed or the actual speed.5 For purposes of this Report, we obtained from Ookla actual speed data for our fixed and mobile international speed comparisons. The Ookla speed data are collected primarily from software-based tests on an end user’s device using speedtest.net. 8. Fixed Speed Testing. To collect fixed speed data, Ookla measures maximum sustainable throughput between the user’s computer and the nearest server by selecting a file size based on a bit test estimate of connection speed. This method measures the speed of the broadband connection when multiple computers or programs are using it.6 Essentially, more data are used to test the faster connections than slower ones, ensuring the speed data reflect the actual speed experienced by the typical consumer.7 9. Mobile Speed Testing. Ookla’s mobile speed measurements are derived from customer tests run over Ookla’s Speedtest mobile app that measures the performance of mobile connections.8 This app is available free of charge to smartphone users.9 Once the app is downloaded, users can periodically measure the speed of their wireless connection, and these speed test observations are then used to produce Ookla’s Net Index dataset.10 Because the speed tests rely on the smartphone’s connection to the server, factors such as congestion, location of the server, proximity and access to a cell tower, and phone quality can affect the result. Therefore, the Ookla data show significant variation in different geographies, as well as among service providers. 10. Methodology and Time Period. We obtained from Ookla a test speed for every city in this Report for the years 2014, 2015, and 2016. Prior to 2015, Ookla provided a test speed for every city on a daily basis.11 Due to this change, we constructed a dataset, for every city in each country selected 5 Actual speed is measured primarily by two methods: (i) by installing special hardware on an end user’s device that enables the hardware to measure actual download and upload speeds and (ii) software-based tests. See MIT Report at 11-17; see also Seventeenth Mobile Wireless Competition Report, 29 FCC Rcd at 15405-06, para. 191 (2014) (discussing the characteristics of crowd-sourced data). 6 This is done by using multiple threads (simultaneous transfers of data) and carefully sizing the transferred payload. Ookla Speedtest, How Does the Test Itself Work? How is the Result Calculated? (Jan. 13, 2012), https://support.speedtest.net/hc/en-us/articles/203845400-How-does-the-test-iitself-work-How-is-the-result- calculated. 7 MIT Report at 3 (“[T]he Ookla/Speedtest approach – which typically results in greater measured data rates than the other approaches reviewed – was the best of the currently available data sources for assessing the speed of ISP’s broadband access service. One of the key differences that accounts for this is that the Ookla/Speedtest tools utilize multiple TCP connections to collect the measurement data which is key to avoiding the receive window limitation. These tests are also much more likely to be connected to a server that is relatively close to the client running the test.”). 8 Ookla Speedtest Mobile; Twentieth Mobile Wireless Competition Report, 32 FCC Rcd at 9034-35, para. 90 (presenting Ookla speed data). 9 Ookla gathers crowdsourced mobile speed data through the use of its Speedtest mobile app. The mobile data include tests over the carrier network and WiFi tests depending on the smartphone’s connection. 10 See Twentieth Mobile Wireless Competition Report, 32 FCC Rcd at 9034-35, para. 90. 11 In the Fifth International Broadband Data Report, we used data for 2014 which included 6.3 million observations for 17,917 cities in 40 countries from January 1, 2014 to December 15, 2014. Mean download and upload speeds on a city-by-day level were aggregated to annual measures and weighted by the number of tests to draw comparisons (continued….) Federal Communications Commission DA 18-99 22 that appear in the data for all three years—2014, 2015, and 2016. Because the cities selected may change over time, not all the same cities appear every year. We then separately compared the changes in broadband speeds for all of the cities over time. 11. Because of the change in the fixed speed data, we rely on the newly acquired 2014, 2015, and 2016 Ookla data rather than the 2014 data relied upon in the Fifth International Broadband Data Report, which were based on daily city-level observations. We also note that the Ookla data omits Latvia in its fixed and mobile speed data, and therefore we compare the United States to 27 foreign countries for a total of 28 countries in our analysis. For fixed and mobile speed results, we present the download speed Mbps rankings for the United States and the selected 27 comparison countries based on weighted mean speed. For weighted mean speed, we take the mean across cities using the sample size in each city as weights.12 In our historical overview, for the period 2012-2014 we also rely on Ookla speed data, but collected using a different methodology. 12. Data Caveats. The Ookla data can be useful in providing a high-level international comparison, but certain caveats should be noted. For instance, the physical distance of the end user to the server may influence the results of user-initiated, software-based speed measurement tests. Additionally, the actual speeds that are observed in each country reflect a combination of availability and usage. For example, a low mean download speed for a country could reflect either more people subscribing to low- speed broadband or poor performance and availability of high-speed broadband. Despite these shortcomings, the Ookla speed dataset helps in constructing international comparisons because of its large geographic scope and vast number of speed tests.13 III. INTERNATIONAL BROADBAND SPEED COMPARISON A. Fixed Broadband Speed 13. Below, we present our fixed broadband speed test results. Table 1 shows our fixed broadband summary statistics for 2014 to 2016, which contained upload and download speeds for 61,920 cities in 2014; 94,894 cities in 2015; and 128,845 cities in 2016. between capital cities and countries in previous IBDRs by using a stratified sampling approach. This approach divides the sample of cities into different non-overlapping bins according to their population level and then draws a sample from each bin. This allowed us to control for larger cities having inherently different broadband characteristics than smaller and more sparsely populated citied. Fifth International Broadband Data Report, 31 FCC Rcd at 2675, 2801-48, paras 28-29, Appx. F. 12 For both fixed and mobile, we conducted an additional estimation check of the mean and median broadband speeds by creating a subsample of the full dataset called a panel, for the cities that appeared in the data for 2014, 2015, and 2016. We then compared the changes in broadband speeds for the same cities over time, holding fixed the sample universe. The mean and median broadband speeds for the subset sample of cities did not differ greatly from the full dataset we present in the Report. 13 Since January 2008, Ookla has collected data on over 10.3 billion speed tests. Ookla, Ookla SpeedTest Intelligence, https://www.ookla.com/speedtest-intelligence (last visited Jan. 16, 2018). Federal Communications Commission DA 18-99 23 Table 1 Fixed Broadband Summary Statistics (2014-2016) All Available Data Total Dataset 2014 2015 2016 Number of Countries 28 28 28 28 Number of Cities 186,127 61,920 94,894 128,845 Mean Tests Per City 2504.66 3251.49 2267.91 2320.11 Median Tests Per City 141 47 110 246 Download (Mbps) Minimum 0.26 0.26 0.26 0.27 Maximum 924.20 917.38 923.81 924.20 Mean 37.50 27.26 38.01 44.03 Median 33.36 26.61 37.65 42.49 Upload (Mbps) Minimum 0.00 0.01 0.00 0.01 Maximum 939.82 939.82 898.70 931.10 Mean 13.28 8.91 13.18 16.29 Median 8.71 7.16 9.34 10.26 Source: Ookla SPEEDTEST intelligence data, © 2016 Ookla, LLC. All rights reserved. Published with permission of Ookla. Note: The cities that make up the complete set of observations varies from year to year in the Ookla dataset, as does the number of tests for each city. 14. As shown in Table 2 and Figure 1 below, the United States ranked 10th out of 28 countries in 2016 in terms of mean download speeds (55.07 Mbps) weighted by sample size—an improvement over 11th in 2015 (40.38 Mbps) and 15th in 2014 (28.09 Mbps). Luxembourg, which as noted earlier is an outlier, is ranked first in 2014, 2015, and 2016, with mean download speeds of 375.78 Mbps, 344.40 Mbps, and 222.13 Mbps, respectively, in the full sample. In 2016, the bottom five countries remained the same, with Greece remaining last with a mean download speed of 11.83 Mbps (compared to 9.52 Mbps in 2015 and 8.90 Mbps in 2014). Federal Communications Commission DA 18-99 24 Table 2 Mean (Weighted) Fixed Download Speed by Country (2014-2016) Country 2014 2015 2016 Rank Mbps Rank Mbps Rank Mbps Luxembourg 1 222.13 1 344.40 1 375.78 Japan 7 35.20 2 81.50 2 102.34 Iceland 6 37.32 8 45.96 3 90.36 South Korea 2 53.15 3 66.77 4 86.98 Switzerland 3 50.32 4 65.86 5 79.58 Sweden 4 47.77 5 58.14 6 73.81 Netherlands 5 45.20 6 56.54 7 67.54 Denmark 14 30.50 13 39.56 8 61.49 Spain 11 31.83 7 46.58 9 57.86 United States 15 28.09 11 40.38 10 55.07 France 21 22.02 16 33.59 11 54.80 Norway 9 33.85 10 42.12 12 54.71 Belgium 13 30.54 17 32.79 13 48.50 Finland 10 31.95 9 43.05 14 47.89 Portugal 12 30.97 14 36.70 15 46.14 New Zealand 18 26.40 19 32.07 16 44.99 Canada 20 22.85 20 31.88 17 44.29 United Kingdom 16 27.68 15 34.43 18 42.17 Germany 17 26.56 18 32.57 19 41.94 Ireland 8 34.16 12 40.09 20 40.28 Czech Republic 22 20.48 23 24.89 21 37.13 Estonia 19 24.86 21 28.71 22 35.05 Austria 24 16.63 22 28.48 23 32.58 Chile 26 14.50 26 15.16 24 24.38 Australia 25 16.13 25 18.49 25 20.12 Mexico 23 17.56 24 22.58 26 18.87 Italy 27 9.90 27 13.50 27 17.23 Greece 28 8.90 28 9.52 28 11.83 Source: Ookla SPEEDTEST intelligence data, © 2016 Ookla, LLC. All rights reserved. Published with permission of Ookla. Note: City-year observations are collapsed to the country-year level and are weighted by the number of tests. Federal Communications Commission DA 18-99 25 Figure 1 Mean (Weighted) Fixed Download Speed by Country (2014-2016) 15. In Table 3 below, we present the median (weighted) fixed download speed by country for 2014-2016 for all available data. In 2016, the median weighted download speed for the United States increased to 55.44 Mbps from 40.17 Mbps in 2015 and 28.29 Mbps in 2014, and its ranking improved from 14th in 2014 and 13th in 2015 to 10th out of the 28 comparison countries.14 Similar to the United States, most countries have mean and median speeds that are fairly constant from 2014 to 2016.15 Luxembourg is ranked first in 2014 (224.77 Mbps), 2015 (295.68), and 2016 (358.66 Mbps) in median download speed. 16. In 2014, 2015, and 2016, the median download speeds increased for the majority of countries in the data, but fell in Ireland and Mexico. In Ireland, the median download speed dropped from 41.65 Mbps in 2014 to 36.98 Mbps in 2016. In Mexico, the median download speed dropped from 18.69 Mbps in 2014 to 15.99 Mbps in 2016. 14 Because the data are aggregated at the city level and do not have individual speed test records, we cannot compute a true median. Here, the median refers to the median of the aggregated (mean) annual city speed tests weighted by sample size. By mean, we mean that the averages at the city level were provided by Ookla. Therefore, we took the median of the city level averages reported by Ookla. 15 See infra paras. 16-17, Tbls. 3 and 4. L u xe m b o u rg Ja pa n Ic e la n d S o u th K o re a S w itz e rl a n d S w e d e n N e th e rl a n d s D e nm a rk S p a in F ra n ce N o rw ay B e lg iu m F in la n d P o rt u g a l N e w Z e a la n d C a n a da U n ite d K in g d o m G er m an y Ir e la n d C ze ch R e p u b lic E st o n ia A u st ri a C h ile A us tr a lia M e xi co It a ly G re e ceU n ite d S ta te s 0 1 0 0 2 0 0 3 0 0 4 0 0 A v e ra g e ( W e ig h te d ) D o w n lo a d S p e e d ( M b p s ) 2016 2015 2014 Federal Communications Commission DA 18-99 26 Table 3 Median (Weighted) Fixed Download Speed by Country (2014-2016) Country 2014 2015 2016 Rank Mbps Rank Mbps Rank Mbps Luxembourg 1 224.77 1 295.68 1 358.66 Iceland 6 37.92 7 50.78 2 96.92 Japan 8 35.62 2 74.67 3 95.58 South Korea 2 53.38 4 66.80 4 87.97 Switzerland 3 50.66 3 67.16 5 77.27 Sweden 7 36.41 5 61.53 6 74.67 Netherlands 4 45.15 6 54.01 7 65.03 Spain 9 34.75 9 46.62 8 58.78 Denmark 12 33.39 14 39.60 9 58.47 United States 14 28.29 13 40.17 10 55.44 Norway 10 34.71 11 41.43 11 55.35 Portugal 13 32.66 12 40.35 12 50.91 France 21 20.63 15 33.18 13 47.38 Belgium 16 26.65 20 27.13 14 46.95 New Zealand 17 26.64 17 31.17 15 44.63 Finland 11 34.53 10 45.01 16 43.06 Canada 20 23.73 16 31.29 17 42.93 Germany 18 26.45 18 30.47 18 40.88 United Kingdom 19 25.03 19 30.15 19 39.97 Estonia 15 27.37 23 21.50 20 37.63 Ireland 5 41.65 8 49.31 21 36.98 Czech Republic 23 15.69 24 19.93 22 35.73 Austria 26 13.79 22 23.79 23 35.50 Chile 24 15.67 26 16.37 24 23.07 Australia 25 14.70 25 17.68 25 19.08 Mexico 22 18.69 21 24.57 26 15.99 Italy 27 10.79 27 14.71 27 15.30 Greece 28 8.80 28 9.54 28 11.83 Source: Ookla SPEEDTEST intelligence data, © 2016 Ookla, LLC. All rights reserved. Published with permission of Ookla. Note: City-year observations are collapsed to the country-year level and are weighted by the number of tests. 17. In Table 4 below, we present the mean download speeds by U.S. states and foreign countries for 2014, 2015, and 2016. Given the large population density and area of several U.S. states we compare U.S. states to foreign countries. In 2016, the highest ranked state is Kansas, which ranked 5th out of 77 states and countries with a mean speed of 80.49 Mbps. Kansas was also the highest ranked state in 2015 at 7th out of 77 U.S. states and countries, with a mean speed of 51.46 Mbps. In 2014, Delaware ranked the highest at 6th with a speed of 38.59 Mbps. Federal Communications Commission DA 18-99 27 Table 4 Mean (Weighted) Fixed Download Speeds by U.S. States and Countries (2014-2016) Country/U.S. State 2014 2015 2016 Rank Mbps Rank Mbps Rank Mbps Luxembourg 1 222.13 1 344.40 1 375.78 Japan 10 35.20 2 81.50 2 102.34 Iceland 7 37.32 15 45.96 3 90.36 South Korea 2 53.15 3 66.77 4 86.98 Kansas 41 26.01 7 51.46 5 80.49 Switzerland 3 50.32 4 65.86 6 79.58 Hawaii 61 20.20 13 46.48 7 76.08 Sweden 4 47.77 5 58.14 8 73.81 Texas 33 28.41 10 47.36 9 69.02 Netherlands 5 45.20 6 56.54 10 67.54 Tennessee 24 30.97 33 39.60 11 64.82 Utah 29 29.60 14 46.00 12 64.28 California 21 31.15 12 46.57 13 62.46 North Carolina 63 19.20 37 38.28 14 62.41 Arizona 30 29.54 28 40.38 15 62.22 Missouri 22 31.10 8 50.16 16 62.20 Denmark 26 30.50 34 39.56 17 61.49 Nevada 27 30.35 29 40.36 18 60.46 Washington 19 31.85 9 49.09 19 59.70 Georgia 38 26.58 16 45.63 20 59.26 Delaware 6 38.59 20 44.11 21 58.22 Spain 20 31.83 11 46.58 22 57.86 New Jersey 8 37.12 17 45.23 23 57.73 Colorado 44 24.89 25 41.79 24 57.29 Massachusetts 11 34.91 18 45.08 25 56.89 Maryland 9 36.43 19 44.61 26 56.70 Alaska 66 17.56 48 34.04 27 56.16 France 55 22.02 49 33.59 28 54.80 Norway 14 33.85 24 42.12 29 54.71 Louisiana 51 23.16 39 37.63 30 53.44 Source: Ookla SPEEDTEST intelligence data, © 2016 Ookla, LLC. All rights reserved. Published with permission of Ookla. Note: City-year observations are collapsed to the country-year level and are weighted by the number of tests. Federal Communications Commission DA 18-99 28 Table 4 (continued) Country/U.S. State 2014 2015 2016 Rank Mbps Rank Mbps Rank Mbps North Dakota 48 23.79 35 39.53 31 53.32 New York 12 34.50 21 43.50 32 53.16 Florida 35 28.27 38 37.84 33 51.82 New Hampshire 28 30.01 32 39.85 34 51.51 Rhode Island 17 32.54 23 42.22 35 51.38 Oklahoma 47 24.15 45 34.91 36 51.23 Oregon 43 25.29 36 39.25 37 50.77 Connecticut 15 33.07 27 40.73 38 49.94 Virginia 16 32.64 31 40.08 39 49.12 Vermont 34 28.40 26 41.35 40 49.10 Belgium 25 30.54 51 32.79 41 48.50 Finland 18 31.95 22 43.05 42 47.89 Illinois 37 27.39 42 36.22 43 47.48 South Dakota 31 29.06 41 36.67 44 46.97 Portugal 23 30.97 40 36.70 45 46.14 Pennsylvania 32 28.59 43 34.97 46 46.07 New Mexico 57 20.92 55 31.21 47 45.79 West Virginia 64 19.08 47 34.24 48 45.47 Minnesota 46 24.44 44 34.94 49 45.16 New Zealand 40 26.40 53 32.07 50 44.99 Canada 53 22.85 54 31.88 51 44.29 Indiana 52 22.87 56 31.02 52 42.88 Kentucky 73 15.64 72 21.20 53 42.84 United Kingdom 36 27.68 46 34.43 54 42.17 Germany 39 26.56 52 32.57 55 41.94 Michigan 42 25.34 50 33.05 56 41.16 Mississippi 58 20.62 61 27.99 57 40.59 Ireland 13 34.16 30 40.09 58 40.28 Alabama 54 22.29 58 29.30 59 39.56 Arkansas 62 19.64 64 27.37 60 39.28 Nebraska 49 23.67 62 27.95 61 38.92 Idaho 67 17.17 67 25.35 62 38.09 Iowa 50 23.66 57 30.10 63 37.24 Source: Ookla SPEEDTEST intelligence data, © 2016 Ookla, LLC. All rights reserved. Published with permission of Ookla. Note: City-year observations are collapsed to the country-year level and are weighted by the number of tests. Federal Communications Commission DA 18-99 29 Table 4 (continued) Country/U.S. State 2014 2015 2016 Rank Mbps Rank Mbps Rank Mbps Czech Republic 59 20.48 69 24.89 64 37.13 South Carolina 56 21.69 63 27.53 65 36.49 Estonia 45 24.86 59 28.71 66 35.05 Wyoming 71 16.43 65 26.78 67 33.30 Austria 70 16.63 60 28.48 68 32.58 Montana 68 16.88 66 26.01 69 30.62 Ohio 69 16.82 70 23.33 70 29.43 Wisconsin 60 20.34 68 25.14 71 28.69 Chile 75 14.50 75 15.16 72 24.38 Maine 74 14.97 73 18.91 73 21.35 Australia 72 16.13 74 18.49 74 20.12 Mexico 65 17.56 71 22.58 75 18.87 Italy 76 9.90 76 13.50 76 17.23 Greece 77 8.90 77 9.52 77 11.83 Source: Ookla SPEEDTEST intelligence data, © 2016 Ookla, LLC. All rights reserved. Published with permission of Ookla. Note: City-year observations are collapsed to the country/state-year level and are weighted by the number of tests. 18. In Table 5, we compared mean fixed download speeds in the capital cities of U.S. states and the selected comparison countries. We present a comparison of U.S. state capitals with the capitals of the comparison countries, as directed by the BDIA that “[t]he Commission shall include in the comparison under this subsection . . . communities including the capital cities of such countries.”16 Austin, Texas is the highest ranked U.S. state capital with a rank of 2nd of 78 capitals in 2016 (110.96 Mbps). 16 47 U.S.C. § 1303(b)(2). Federal Communications Commission DA 18-99 30 Table 5 Mean (Weighted) Fixed Download Speed by Country Capital and U.S. State Capital Cities (2014-2016) City, Country 2014 2015 2016 Rank Mbps # of Tests Rank Mbps # of Tests Rank Mbps # of Tests Luxembourg, Luxembourg 1 170.59 56,183 1 238.68 6,341 1 300.13 25,173 Austin, TX, U.S. 4 49.85 1,392,853 5 79.66 755,280 2 110.96 542,944 Reykjavik, Iceland 13 37.92 47,571 15 50.78 14,983 3 96.92 73,977 Paris, France 2 83.79 413 3 93.68 310 4 96.77 931,706 Stockholm, Sweden 7 42.26 34 4 88.35 57 5 89.34 145,213 Seoul, South Korea 3 53.38 486,932 8 66.80 424,781 6 87.97 360,565 Phoenix, AZ, U.S. 27 29.79 831,543 34 39.44 409,368 7 84.69 150,087 Jackson, MS, U.S. 52 23.93 36,284 42 34.54 28,654 8 81.03 9,728 Tokyo, Japan 17 35.62 1,051,715 7 71.21 722,467 9 80.24 838,054 Oslo, Norway 9 40.30 596,388 12 52.01 304,067 10 77.68 250,025 Salt Lake City, UT, U.S. 36 27.45 548,660 19 44.13 348,882 11 72.45 228,718 Nashville, TN, U.S. 28 29.71 213,631 36 37.74 139,669 12 69.67 99,940 Oklahoma City, OK, U.S. 29 29.63 302,669 24 41.64 157,788 13 67.35 75,466 Honolulu, HI, U.S. 59 20.47 278,913 23 41.79 254,978 14 65.14 162,823 Madrid, Spain 40 26.36 292 6 71.31 221 15 64.95 843,286 Salem, OR, U.S. 30 29.05 77,764 13 51.28 63,294 16 64.77 44,847 Atlanta, GA, U.S. 25 30.45 685,856 2 127.47 546,298 17 64.49 263,102 Olympia, WA, U.S. 24 30.54 80,179 14 51.05 56,739 18 63.87 34,271 Dover, DE, U.S. 6 44.16 29,250 11 52.32 22,005 19 63.46 16,963 Bern, Switzerland 26 30.10 5 30 40.62 6 20 61.17 38,788 Lansing, MI, U.S. 31 28.82 87,624 9 59.95 69,863 21 60.34 45,758 Amsterdam, Netherlands 60 20.39 213 73 18.14 536 22 59.60 275,193 Raleigh, NC, U.S. 62 19.82 319,470 30 40.62 326,968 23 58.42 138,758 Denver, CO, U.S. 50 24.18 999,544 33 40.13 657,396 24 57.75 406,555 Concord, NH, U.S. 21 32.64 24,537 29 40.69 21,694 25 57.61 13,741 Copenhagen, Denmark 15 36.05 163 27 40.96 60 26 57.60 137,735 Annapolis, MD, U.S. 8 40.35 35,926 17 46.51 30,999 27 56.94 18,602 Trenton, NJ, U.S. 12 38.90 77,050 18 44.36 57,904 28 56.65 39,850 Washington, DC, U.S. 22 32.52 374,696 20 43.14 319,088 29 55.97 225,414 Source: Ookla SPEEDTEST intelligence data, © 2016 Ookla, LLC. All rights reserved. Published with permission of Ookla. Note: City-year observations are collapsed to the country/state-year level and are weighted by the number of tests. Federal Communications Commission DA 18-99 31 Table 5 (continued) City, Country 2014 2015 2016 Rank Mbps # of Tests Rank Mbps # of Tests Rank Mbps # of Tests Wellington, New Zealand 38 26.64 190,525 58 29.74 103,951 30 55.50 55,734 Boston, MA, U.S. 16 35.87 238,628 28 40.91 198,723 31 55.25 125,587 Bismarck, ND, U.S. 19 33.77 22,482 16 50.68 18,991 32 53.41 6,590 Boise, ID, U.S. 66 18.53 208,995 63 25.56 150,117 33 52.07 55,313 Lisbon, Portugal 14 36.10 710,256 21 42.89 404,664 34 51.96 321,952 Sacramento, CA, U.S. 35 27.51 456,600 26 41.07 336,064 35 51.69 226,513 Des Moines, IA, U.S. 55 23.18 165,510 54 31.31 92,922 36 51.36 43,534 Harrisburg, PA, U.S. 20 32.90 53,234 41 35.67 46,885 37 51.14 28,359 Indianapolis, IN, U.S. 49 24.34 400,721 48 32.93 207,688 38 50.73 144,284 Tallinn, Estonia 10 39.49 88 70 22.52 13 39 50.37 155,547 Providence, RI, U.S. 23 30.55 59,039 25 41.20 52,422 40 49.71 41,257 Saint Paul, MN, U.S. 47 24.75 345,834 55 31.29 197,739 41 48.73 53,777 Prague, Czech Republic 61 20.26 158 60 28.33 318 42 48.63 326,808 Santa Fe, NM, U.S. 56 22.73 48,269 39 36.39 43,234 43 48.16 35,323 Richmond, VA, U.S. 18 34.08 213,371 32 40.22 135,960 44 47.44 58,126 Springfield, IL, U.S. 46 25.00 60,565 35 38.00 43,286 45 47.24 30,066 Hartford, CT, U.S. 37 27.31 26,263 43 34.48 27,040 46 46.98 19,279 Baton Rouge, LA, U.S. 42 26.01 145,260 38 37.36 135,154 47 46.78 98,397 Little Rock, AK, U.S. 51 24.04 104,140 49 32.87 76,665 48 45.40 28,628 Montgomery, AL, U.S. 33 27.87 71,190 45 34.43 57,707 49 45.29 10,428 Dublin, Ireland 11 39.19 533 66 24.41 423 50 43.56 136,915 Ottawa, Canada 43 25.78 653,967 40 35.67 298,477 51 43.08 180,493 Helsinki, Finland 41 26.06 1,061 10 57.67 1,007 52 43.06 348,992 Columbus, OH, U.S. 69 17.54 461,117 59 29.58 323,834 53 41.75 231,898 Source: Ookla SPEEDTEST intelligence data, © 2016 Ookla, LLC. All rights reserved. Published with permission of Ookla. Note: City-year observations are collapsed to the country/state-year level and are weighted by the number of tests. Federal Communications Commission DA 18-99 32 Table 5 (continued) City, Country 2014 2015 2016 Rank Mbps # of Tests Rank Mbps # of Tests Rank Mbps # of Tests Vienna, Austria 5 44.76 461 22 42.23 186 54 41.44 472,649 Tallahassee, FL, U.S. 53 23.60 144,559 52 32.26 131,740 55 41.23 78,334 Cheyenne, WY, U.S. 63 19.25 47,024 47 34.02 31,616 56 40.87 25,077 Madison, WI, U.S. 34 27.80 215,057 50 32.84 127,063 57 40.74 93,532 Charleston, WV, U.S. 72 15.49 76,161 37 37.65 53,811 58 39.59 7,196 Helena, MT, U.S. 68 18.16 35,778 56 30.41 25,671 59 39.10 17,819 Brussels, Belgium 44 25.76 224 71 19.33 88 60 39.04 172,244 Berlin, Germany 39 26.45 1,097,561 61 28.13 539,123 61 36.11 296,017 London, United Kingdom 45 25.03 4,236,152 57 30.15 1,414,159 62 34.98 674,685 Montpelier, VT, U.S. 71 15.57 12,315 67 24.37 8,243 63 34.95 5,057 Carson City, NV, U.S. 54 23.20 20,547 65 24.57 14,468 64 34.54 11,339 Lincoln, NE, U.S. 75 12.93 191,240 72 18.35 159,150 65 34.44 118,833 Jefferson City, MO, U.S. 64 19.12 45,576 53 31.70 40,953 66 33.92 25,982 Pierre, SD, U.S. 32 28.41 4,329 46 34.29 1,917 67 33.76 1,358 Topeka, KS, U.S. 57 22.14 79,929 51 32.38 52,739 68 32.19 13,069 Albany, NY, U.S. 58 21.31 82,087 62 26.86 52,239 69 31.38 37,918 Canberra, Australia 70 16.61 241,423 68 24.28 56,172 70 29.63 3,769 Juneau, AK, U.S. 48 24.36 6,296 44 34.47 5,056 71 28.50 1,033 Columbia, SC, U.S. 67 18.51 135,502 69 23.03 80,076 72 28.20 47,885 Santiago, Chile 77 9.41 2,546 78 8.69 1,514 73 22.83 913,652 Rome, Italy 76 10.79 2,503,159 76 14.71 1,025,307 74 20.83 601,526 Augusta, ME, U.S. 74 13.61 23,152 74 15.64 12,095 75 18.73 4,634 Frankfort, KY, U.S. 73 13.72 9,690 75 14.89 15,780 76 14.13 12,060 Athens, Greece 78 8.80 1,341 77 9.54 349 77 11.94 809,196 Mexico City, Mexico 65 18.69 4,770,066 64 24.57 1,907,816 78 8.79 129 Source: Ookla SPEEDTEST intelligence data, © 2016 Ookla, LLC. All rights reserved. Published with permission of Ookla. We note that we cannot draw statistical conclusions from cities with less than 300 tests per year. Note: The yearly observation for Mexico, Federal District, Mexico and for Mexico City, Federal District, Mexico are averaged (weighted by total tests) for 2014 is 8.00 Mbps (88,347) and 18.89 Mbps (4,681,719). For 2015, it is 9.80 Mbps (29,043) and 24.80 Mbps (1,878,773). Federal Communications Commission DA 18-99 33 B. Mobile Broadband Speeds 19. Table 6 below presents our mobile broadband summary statistics for all available data from 2014 to 2016. The data contain upload and download speeds for 96,350 cities in 2014; 98,538 cities in 2015; and 120,159 cities in 2016. Table 6 Mobile Broadband Summary Statistics (2014-2016) All Available Data Total Dataset 2014 2015 2016 Number of Countries 28 27 28 28 Number of Cities 185,640 96,350 98,538 120,159 Mean Tests Per City 488.79 485.00 443.87 528.67 Median Tests Per City 30 25 20 47 Download (Mbps) Minimum 0.26 0.26 0.26 0.26 Maximum 190.41 96.91 129.20 190.41 Mean 18.02 12.85 16.68 22.73 Median 16.38 12.55 15.87 21.87 Upload (Mbps) Minimum 0 0 0 0 Maximum 72.88 45.19 72.88 72.54 Mean 7.37 5.31 7.04 9.10 Median 7.05 5.37 7.09 9.10 Source: Ookla SPEEDTEST intelligence data, © 2016 Ookla, LLC. All rights reserved. Published with permission of Ookla. Note: The cities that make up the complete set of observations varies from year to year in the Ookla dataset, as does the number of tests for each city. The data did not include any mobile speed observations for Latvia. 20. Below, Table 7 and Figure 2 show that the United States dropped from 18th (12.62 Mbps) in 2014 to 24th (15.58 Mbps) in 2015, and remained at 24th out of the 28 comparison countries (19.98 Mbps) in 2016. During this period, however, mobile download speeds in the United States increased by approximately 58 percent. Federal Communications Commission DA 18-99 34 Table 7 Mean (Weighted) Mobile Download Speed by Country (2014-2016) Country 2014 2015 2016 Rank Mbps Rank Mbps Rank Mbps South Korea 3 20.76 5 25.35 1 39.19 Netherlands 10 15.00 7 24.15 2 39.08 Norway 9 15.08 8 23.90 3 38.03 Australia 5 19.05 2 27.24 4 36.57 Denmark 6 18.49 6 24.22 5 33.12 Luxembourg 4 19.73 4 26.12 6 32.47 Austria 20 11.70 14 19.45 7 31.09 Iceland 2 25.43 3 26.41 8 30.93 New Zealand 1 28.09 1 27.85 9 30.36 Greece 11 21.15 10 29.34 Switzerland 13 14.12 9 22.42 11 28.07 Belgium 8 15.59 16 17.99 12 27.22 France 11 14.97 13 19.62 13 26.87 Finland 15 13.37 12 20.86 14 26.63 Sweden 12 14.24 18 17.22 15 26.15 Canada 7 16.92 10 21.26 16 26.02 Estonia 25 10.37 23 16.21 17 24.27 Spain 23 11.05 22 16.61 18 24.14 United Kingdom 21 11.11 20 16.92 19 24.00 Czech Republic 14 13.58 17 17.71 20 23.14 Germany 17 12.66 21 16.75 21 22.85 Italy 22 11.10 15 18.11 22 22.03 Portugal 16 12.72 19 17.05 23 20.31 United States 18 12.62 24 15.58 24 19.98 Ireland 19 12.05 26 12.92 25 16.34 Japan 24 10.80 25 14.28 26 15.95 Chile 27 6.82 27 9.13 27 15.61 Mexico 26 8.34 28 9.06 28 15.24 Source: Ookla SPEEDTEST intelligence data, © 2016 Ookla, LLC. All rights reserved. Published with permission of Ookla. Note: City-year observations are collapsed to the country-year level and are weighted by the number of tests. Federal Communications Commission DA 18-99 35 Figure 2 Mean (Weighted) Mobile Download Speed by Country (2014-2016) 21. In Table 8, we present the median mobile weighted download speed by country for 2015 to 2016. In 2016, the median weighted download speed for the United States increased to 19.36 Mbps from 12.62 Mbps in 2014, and its ranking decreased from 17th to 24th of the 28 comparison countries. Similar to the United States, most countries have mean and median speeds that are fairly constant in the 2014, 2015, and 2016 data.17 In terms of speed measurements, the United States has improved from 2014 to 2016 when comparing both the median and mean download speeds weighted by the sample size. 17 See supra para. 20, Tbl. 7; see infra para. 21, Tbl. 8. Because the data are aggregated at the city level and do not have individual speed test records, we cannot compute a true median. Here, median refers to the median of the aggregated (mean) daily city speed tests weighted by sample size. As a summary statistic, medians are not affected by outliers in the data while means are. S ou th K or ea N et he rl an ds N o rw ay A us tr al ia D en m ar k L ux em bo ur g A u st ria Ic el an d N ew Z ea la nd G re ec e S w itz er la nd B el gi um F ra nc e F in la nd S w ed e n C an a da E st on ia S pa in U ni te d K in gd om C ze ch R ep ub lic G e rm an y Ita ly P or tu ga l Ir el an d Ja pa n C hi le M e xi coU ni te d S ta te s 0 1 0 20 30 40 A ve ra ge (W e ig ht e d) D o w nl oa d S p ee d (M b ps ) 2016 2015 2014 Federal Communications Commission DA 18-99 36 Table 8 Median (Weighted) Mobile Download Speed by Country (2014-2016) Country 2014 2015 2016 Rank Mbps Rank Mbps Rank Mbps Netherlands 12 14.61 8 23.29 1 41.12 South Korea 3 20.79 7 23.58 2 38.91 Norway 8 16.68 5 25.64 3 36.59 Australia 5 20.32 2 27.91 4 35.88 New Zealand 1 31.87 1 29.99 5 32.94 Denmark 6 19.18 6 24.63 6 32.60 Austria 20 12.49 13 20.26 7 32.44 Luxembourg 4 20.66 4 25.88 8 32.43 Iceland 2 25.49 3 27.30 9 32.01 France 11 14.80 14 19.98 10 29.44 Switzerland 13 14.61 9 22.84 11 28.71 Finland 14 13.88 11 21.80 12 27.99 Greece 12 21.15 13 27.66 Estonia 25 10.66 23 16.29 14 27.31 Sweden 10 14.97 19 17.61 15 27.29 Belgium 9 15.88 16 18.65 16 26.77 Canada 7 17.14 10 21.82 17 26.30 Spain 21 11.99 20 17.54 18 25.53 United Kingdom 23 11.36 22 16.35 19 23.48 Germany 19 12.59 21 17.29 20 23.43 Czech Republic 18 12.59 18 17.89 21 23.37 Italy 22 11.71 17 18.43 22 23.01 Portugal 16 12.69 15 18.92 23 21.52 United States 17 12.62 24 15.26 24 19.36 Ireland 15 12.76 26 13.46 25 16.51 Japan 24 11.13 25 14.95 26 15.97 Mexico 26 8.29 28 8.97 27 15.39 Chile 27 6.53 27 9.61 28 15.28 Source: Ookla SPEEDTEST intelligence data, © 2016 Ookla, LLC. All rights reserved. Published with permission of Ookla. Note: City-year observations are collapsed to the country-year level and are weighted by the number of tests. 22. In Table 9 below, we present the mean download speeds by U.S. states and foreign countries for 2014, 2015, and 2016. Given the large population density and area of several U.S. states we compare U.S. states to foreign countries. The highest ranked state for all three years is Washington, which ranked 20th out of 77 states and countries in 2016 with a mean speed of 23.70 Mbps, 13th in 2015 with a speed of 20.80 Mbps, and 8th in 2014 with a speed of 16.17 Mbps. Additionally, four U.S. states – Washington, Minnesota, Georgia, and Michigan – ranked in the top 25 for download speeds when compared to other U.S. states and countries in our study. Federal Communications Commission DA 18-99 37 Table 9 Mean (Weighted) Mobile Download Speeds by U.S. States and Countries (2014-2016) Country/U.S. State 2014 2015 2016 Rank Mbps Rank Mbps Rank Mbps South Korea 3 20.76 5 25.35 1 39.04 Netherlands 12 15.00 7 24.15 2 38.21 Norway 10 15.08 8 23.90 3 36.46 Australia 5 19.05 2 27.24 4 35.77 Denmark 6 18.49 6 24.22 5 33.12 Luxembourg 4 19.73 4 26.12 6 32.25 Iceland 2 25.43 3 26.41 7 30.93 Austria 47 11.70 16 19.45 8 30.82 New Zealand 1 28.09 1 27.85 9 30.26 Greece 11 21.15 10 29.34 Switzerland 19 14.12 9 22.42 11 27.86 Belgium 9 15.59 20 17.99 12 27.17 France 13 14.97 15 19.62 13 26.70 Finland 25 13.37 12 20.86 14 26.61 Sweden 17 14.24 24 17.22 15 26.15 Canada 7 16.92 10 21.26 16 25.56 Estonia 65 10.37 34 16.21 17 24.04 Spain 55 11.05 29 16.61 18 23.88 United Kingdom 53 11.11 26 16.92 19 23.87 Washington 8 16.17 13 20.80 20 23.70 Minnesota 14 14.88 19 18.01 21 23.57 Georgia 11 15.00 14 20.31 22 23.51 Czech Republic 23 13.58 21 17.71 23 23.14 Michigan 22 13.75 17 18.92 24 22.79 Germany 37 12.66 27 16.75 25 22.77 Italy 54 11.10 18 18.11 26 22.03 Rhode Island 15 14.75 22 17.51 27 21.16 Ohio 30 12.97 38 15.79 28 21.10 Alabama 21 13.75 23 17.48 29 21.09 Oregon 28 13.04 30 16.60 30 21.08 Source: Ookla SPEEDTEST intelligence data, © 2016 Ookla, LLC. All rights reserved. Published with permission of Ookla. Note: City-year observations are collapsed to the country-year level and are weighted by the number of tests. Federal Communications Commission DA 18-99 38 Table 9 (continued) Country/U.S. State 2014 2015 2016 Rank Mbps Rank Mbps Rank Mbps Illinois 29 13.03 33 16.23 31 20.91 Connecticut 18 14.18 28 16.64 32 20.80 New Jersey 24 13.55 32 16.37 33 20.80 New York 35 12.84 37 15.83 34 20.74 South Dakota 16 14.28 46 14.54 35 20.65 Indiana 45 11.97 42 15.04 36 20.63 Florida 39 12.22 31 16.58 37 20.55 Kansas 58 10.99 47 14.40 38 20.44 Wisconsin 34 12.92 39 15.56 39 20.30 Portugal 36 12.72 25 17.05 40 20.25 California 32 12.94 35 16.11 41 20.21 North Dakota 64 10.45 60 12.98 42 20.12 Pennsylvania 26 13.22 41 15.50 43 19.90 Delaware 33 12.93 40 15.50 44 19.81 Massachusetts 20 13.92 36 16.04 45 19.64 Missouri 49 11.46 53 13.78 46 18.93 Iowa 50 11.27 64 12.73 47 18.78 Kentucky 59 10.91 65 12.67 48 18.43 Maryland 46 11.85 51 14.02 49 18.25 Arkansas 27 13.18 43 14.96 50 18.10 Texas 40 12.13 45 14.57 51 17.93 New Hampshire 38 12.55 44 14.73 52 17.92 Tennessee 31 12.96 50 14.16 53 17.91 Virginia 48 11.65 52 13.79 54 17.70 Hawaii 63 10.53 48 14.38 55 17.67 Louisiana 52 11.14 54 13.76 56 17.52 Nebraska 57 10.99 59 13.22 57 17.03 North Carolina 42 12.04 58 13.50 58 16.98 South Carolina 43 12.01 56 13.70 59 16.89 Utah 68 10.03 62 12.79 60 16.82 Nevada 51 11.27 55 13.73 61 16.53 Ireland 41 12.05 61 12.92 62 16.34 Arizona 62 10.54 69 11.59 63 16.31 Source: Ookla SPEEDTEST intelligence data, © 2016 Ookla, LLC. All rights reserved. Published with permission of Ookla. Note: City-year observations are collapsed to the country-year level and are weighted by the number of tests. Federal Communications Commission DA 18-99 39 Table 9 (continued) Country/U.S. State 2014 2015 2016 Rank Mbps Rank Mbps Rank Mbps Oklahoma 44 11.98 57 13.54 64 16.19 Japan 60 10.80 49 14.28 65 15.84 Chile 76 6.82 75 9.13 66 15.49 Mississippi 66 10.30 66 12.04 67 15.28 Idaho 69 9.85 63 12.78 68 14.98 Mexico 74 8.34 76 9.06 69 14.96 Montana 75 8.30 73 10.28 70 14.26 Colorado 61 10.55 67 11.68 71 14.22 New Mexico 70 9.20 68 11.64 72 14.07 West Virginia 71 9.07 70 11.14 73 13.56 Maine 67 10.04 71 10.84 74 12.71 Alaska 56 11.01 74 9.80 75 12.56 Vermont 72 8.49 72 10.30 76 12.17 Wyoming 73 8.37 77 8.99 77 9.61 Source: Ookla SPEEDTEST intelligence data, © 2016 Ookla, LLC. All rights reserved. Published with permission of Ookla. Note: City-year observations are collapsed to the country/state-year level and are weighted by the number of tests. 23. In Table 10 below, we compare mean mobile download speeds in the capital cities of U.S. states and the selected comparison countries. We present a comparison of U.S. state capitals with the capitals of the comparison countries, as directed by the BDIA that “[t]he Commission shall include in the comparison under this subsection . . . communities including the capital cities of such countries.”18 Lansing, Michigan ranks as the highest U.S. state capital with a 20th ranking out of 78 capitals in 2016 (26.16 Mbps). 18 47 U.S.C. § 1303(b)(2) Federal Communications Commission DA 18-99 40 Table 10 Mean (Weighted) Mobile Download Speed by Country Capital and U.S. State Capital Cities (2014-2016) City, Country 2014 2015 2016 Rank Mbps # of Tests Rank Mbps # of Tests Rank Mbps # of Tests Amsterdam, Netherlands 1 42.07 61,000 Oslo, Norway 6 16.74 68,099 3 27.37 118,402 2 40.68 119,217 Seoul, South Korea 2 20.79 7,412 1 31.11 35,576 3 38.91 92,812 Vienna, Austria 10 20.73 169 4 34.77 450,644 Helsinki, Finland 5 32.49 347,676 Luxembourg, Luxembourg 3 20.75 16,698 4 24.88 12,525 6 32.43 17,018 Reykjavik, Iceland 7 32.01 13,160 Stockholm, Sweden 26 16.85 707 8 31.42 24,710 Wellington, New Zealand 1 31.87 5,246 2 30.85 5,256 9 30.53 9,077 Copenhagen, Denmark 11 20.72 1,126 10 30.21 100,778 Paris, France 11 29.65 366,806 Bern, Switzerland 14 20.19 25 12 28.90 13,588 Madrid, Spain 5 24.87 34 13 28.36 200,331 Lisbon, Portugal 11 15.14 28,762 19 18.92 62,737 14 28.26 50,622 Prague, Czech Republic 15 27.75 79,886 Athens, Greece 8 21.15 1,678 16 27.66 138,488 Tallinn, Estonia 17 27.31 123,540 Canberra, Australia 57 8.67 160 7 21.93 5,168 18 26.42 14,528 Brussels, Belgium 20 18.46 430 19 26.31 28,546 Lansing, MI, U.S. 8 16.49 17,279 15 20.03 16,681 20 26.16 12,964 Salem, OR, U.S. 9 16.02 16,661 17 19.14 15,605 21 25.61 11,700 Saint Paul, MN, U.S. 16 13.98 64,825 25 17.05 71,902 22 25.14 28,216 Berlin, Germany 35 12.23 50,613 16 19.88 97,767 23 24.76 172,721 Montgomery, AL, U.S. 27 12.65 12,431 21 18.25 12,322 24 24.54 11,003 Source: Ookla SPEEDTEST intelligence data, © 2016 Ookla, LLC. All rights reserved. Published with permission of Ookla. Note: City-year observations are collapsed to the country/state-year level and are weighted by the number of tests. Federal Communications Commission DA 18-99 41 Table 10 (continued) City, Country 2014 2015 2016 Rank Mbps # of Tests Rank Mbps # of Tests Rank Mbps # of Tests Rome, Italy 13 14.23 47,589 9 20.87 181,500 25 24.11 537,626 London, United Kingdom 42 11.36 875,046 30 16.35 960,512 26 23.48 794,560 Little Rock, AR, U.S. 10 15.35 19,742 23 17.85 15,900 27 23.32 13,844 Springfield, IL, U.S. 30 12.44 13,956 24 17.82 10,485 28 23.10 8,443 Tallahassee, FL, U.S. 12 14.98 19,444 13 20.43 16,238 29 23.08 12,244 Ottawa, Canada 4 17.30 24,570 12 20.64 28,881 30 22.49 46,039 Bismarck, ND, U.S. 51 10.01 3,772 53 13.17 2,191 31 22.37 1,926 Dover, DE, U.S. 5 17.05 4,357 6 23.40 3,280 32 22.31 2,742 Atlanta, GA, U.S. 20 13.65 206,342 18 19.03 182,204 33 22.28 170,471 Indianapolis, IN, U.S. 39 11.71 41,801 34 15.88 51,603 34 22.05 72,218 Columbus, OH U.S. 15 14.20 62,558 29 16.37 68,965 35 22.02 88,165 Harrisburg, PA, U.S. 7 16.67 9,571 22 18.03 8,157 36 21.41 7,399 Pierre, SD, U.S. 29 12.58 60 65 10.77 221 37 20.51 317 Des Moines, IA, U.S. 56 9.01 16,415 56 12.44 22,343 38 20.36 20,494 Dublin, Ireland 62 6.87 1 71 2.19 1 39 20.26 170,265 Annapolis, MD, U.S. 34 12.34 5,516 43 14.52 5,938 40 20.08 4,620 Lincoln, NE, U.S. 46 11.11 22,513 46 14.19 16,605 41 19.99 16,143 Washington, DC, U.S. 40 11.51 40,351 41 14.87 93,123 42 19.67 109,894 Austin, TX, U.S. 23 13.25 104,918 42 14.76 105,773 43 19.62 111,128 Providence, RI, U.S. 21 13.63 10,925 27 16.83 14,433 44 19.54 16,818 Sacramento, CA, U.S. 25 13.09 102,978 33 15.92 106,214 45 19.21 111,809 Juneau, AK, U.S. 63 6.67 21 47 14.11 948 46 19.09 662 Trenton, NJ, U.S. 19 13.65 9,610 31 16.14 7,823 47 19.07 6,216 Hartford, CT, U.S. 18 13.70 19,404 32 16.02 18,169 48 19.04 16,463 Raleigh, NC, U.S. 38 11.78 32,461 48 13.97 32,809 49 19.00 40,699 Albany, NY, U.S. 26 13.07 12,377 40 14.89 10,936 50 18.92 10,275 Olympia, WA, U.S. 17 13.71 6,611 38 15.13 6,055 51 18.58 4,688 Boston, MA, U.S. 32 12.41 31,974 37 15.14 64,185 52 18.25 88,039 Source: Ookla SPEEDTEST intelligence data, © 2016 Ookla, LLC. All rights reserved. Published with permission of Ookla. Note: City-year observations are collapsed to the country/state-year level and are weighted by the number of tests. Federal Communications Commission DA 18-99 42 Table 10 (continued) City, Country 2014 2015 2016 Rank Mbps # of Tests Rank Mbps # of Tests Rank Mbps # of Tests Salt Lake City, UT, U.S. 47 10.95 85,200 50 13.76 75,767 53 18.23 88,701 Frankfort, KY, U.S. 43 11.31 3,489 58 12.25 2,683 54 18.10 2,797 Nashville, TN, U.S. 33 12.37 61,272 51 13.44 64,362 55 17.75 80,705 Baton Rouge, LA, U.S. 31 12.41 45,621 44 14.48 29,842 56 17.70 20,941 Richmond, VA, U.S. 28 12.64 29,968 55 12.89 24,777 57 17.63 32,270 Honolulu, HI, U.S. 45 11.16 62,847 39 14.91 81,720 58 17.25 118,987 Carson City, NV, U.S. 14 14.22 795 28 16.82 1,877 59 17.14 2,945 Helena, MT, U.S. 55 9.05 1,182 66 10.69 1,592 60 17.13 1,529 Phoenix, AZ, U.S. 52 9.87 145,810 59 11.48 188,833 61 17.02 183,819 Madison, WI, U.S. 36 12.23 21,065 54 13.05 17,106 62 16.71 16,139 Jefferson City, MO, U.S. 37 11.99 7,055 49 13.94 5,013 63 16.70 3,351 Jackson, MS, U.S. 50 10.11 8,816 52 13.29 5,863 64 16.66 5,511 Mexico City, Mexico 59 8.29 115,018 69 8.97 284,585 65 16.15 576,975 Columbia, SC, U.S. 41 11.40 9,870 57 12.26 10,756 66 16.07 13,674 Boise, ID, U.S. 48 10.81 18,249 35 15.85 18,279 67 16.02 14,845 Tokyo, Japan 44 11.24 371,219 36 15.51 498,306 68 15.97 664,877 Oklahoma City, OK, U.S. 22 13.41 26,887 45 14.45 42,083 69 15.93 71,115 Topeka, KS, U.S. 61 7.76 11,441 63 11.05 7,620 70 15.56 7,240 Santiago, Chile 71 15.28 488,563 Denver, CO, U.S. 49 10.52 126,113 60 11.41 141,413 72 14.16 140,760 Augusta, ME, U.S. 24 13.16 401 61 11.35 572 73 13.66 968 Montpelier, VT, U.S. 53 9.65 478 62 11.10 407 74 13.39 307 Cheyenne, WY, U.S. 58 8.37 1,548 68 9.47 2,705 75 12.08 3,574 Charleston, WV, U.S. 60 7.91 3,984 64 10.86 4,945 76 11.93 4,657 Concord, NH, U.S. 54 9.21 926 67 10.25 1,215 77 11.33 1,443 Santa Fe, NM, U.S. 64 6.58 10,132 70 7.97 9,630 78 11.28 6,725 Source: Ookla SPEEDTEST intelligence data, © 2016 Ookla, LLC. All rights reserved. Published with permission of Ookla. We note that we cannot draw statistical conclusions from cities with less than 300 tests per year. Note: City-year observations are collapsed to the country/state-year level and are weighted by the number of tests. C. Historical Overview of Fixed Speeds from 2012 to 2016 24. Below, we present U.S. fixed download speeds and rankings from 2012 to 2016 to demonstrate how actual speeds have evolved over time.19 We rely on data presented in prior International Broadband Data Reports to make this comparison. We note that the Fourth International Broadband Data Report and the Fifth International Broadband Data Report relied on Ookla speed data for 2012 to 2014 that consisted of daily speed test results for all cities.20 As described above in our discussion of speed data for 2014 to 2016, we there rely on Ookla speed data that consists of city speed test results averaged up to the yearly level, which has far fewer observations than the prior methodology.21 19 We do not have Ookla mobile speed data prior to 2014 to conduct a similar analysis. 20 See Fourth International Broadband Data Report, 30 FCC Rcd at 15225, 15248, Appx. F; Fifth International Broadband Data Report, 31 FCC Rcd 2667, 2801, 2821, Appx. F. 21 See supra para. 10. Federal Communications Commission DA 18-99 43 25. Below, we present: (1) speed data for 2012 to 2013 under the previous methodology; (2) speed data for 2015 to 2016 under the new methodology; and (3) speed data for 2014 under both methodologies. Because of the different methodologies, we present separately the U.S. data from 2012 to 2013 (Table 11) and 2014 to 2016 (Table 12). We also note that, under the previous methodology for 2012, 2013, and 2014, the speed rankings were based on a 40-country sample. By comparison, this Report’s analysis of 2014, 2015, and 2016 was based on a 28-country sample. By comparing the 2014 data under both methodologies, we find a number of differences. For the United States, the differences are relatively minor, while the differences for many of the other 27 countries are significantly greater. For instance, the speeds of only 9 of the 28 countries are within five percent in the two datasets, while the speeds of 13 of the 28 countries show more than a 10 percent differential. Based on the data, for the United States, both speeds and rank have been on a rising trend since 2012. Table 11 Historical U.S. Fixed Broadband Speeds Previous Method (2012-2014) Year USA Rank # Countries Ranked Relative USA Ranking Mean Speed (Mbps) Data Methodology 2012 25 40 0.63 14.50 Previous 2013 26 40 0.65 18.67 Previous 2014 26 40 0.65 26.68 Previous Table 12 Historical U.S. Fixed Broadband Speeds New Method (2014-2016) Year USA Rank # Countries Ranked Relative USA Ranking Mean Speed (Mbps) Data Methodology 2014 15 28 0.54 28.09 New 2015 11 28 0.39 40.38 New 2016 10 28 0.36 55.07 New Federal Communications Commission DA 18-99 44 APPENDIX C Broadband Price Comparison 1. In this Appendix, as directed by the BDIA, we compare the pricing of fixed and mobile broadband plans across the United States with the selected 28 comparison countries.1 Between June and August of 2017, we collected a stratified random sample of advertised prices and terms for almost 3,000 fixed and mobile broadband plans from broadband providers’ websites in the United States and the selected foreign countries.2 2. The BDIA directs the Commission to compare broadband pricing in “communities of a population size, population density, topography, and demographic profile that are comparable to the population size, population density, topography, and demographic profile of various communities within the United States.”3 In this Appendix, we have ranked the countries by fixed and mobile prices from the least expensive (1st) to most expensive (e.g., 29th) according to three different methodologies. As in previous reports, we continue to produce rankings based on unweighted average prices for standalone fixed broadband plans within certain download speed ranges and mobile plans within bands of data usage allowances. To more closely match the characteristics of the comparison communities and their broadband offerings with those in the United States, we present country rankings by two additional methodologies: a broadband price index4 and a hedonic price index.5 The additional assessments seek to better assess how the U.S. market is performing relative to other markets after accounting for quality differences as well as market-level cost and demographic differences that are known to affect pricing, such as population density, income, and education levels. The hedonic price index also allows an adjustment for observable differences in broadband quality across countries (e.g., speed and usage limits) and generates prices for a set of standardized broadband plans in every country to produce a price index that accounts for all of these factors and is comparable across countries.6 The fixed and mobile analyses 1 47 U.S.C. § 1303(b)(1) (“As part of the assessment and report required by section 1302 of this title, the Federal Communications Commission shall include information comparing the extent of broadband service capability (including data transmission speeds and price for broadband service capability) in a total of 75 communities in at least 25 countries abroad for each of the data rate benchmarks for broadband service utilized by the Commission to reflect different speed tiers.”). Our comparison of broadband pricing in communities includes comparison of broadband pricing in cities, as directed by the BDIA to include “a geographically diverse selection of countries” and “communities including the capital cities of such countries.” 47 U.S.C. § 1303(b)(2). 2 The fixed and mobile broadband price datasets are available on the FCC website. FCC, International Broadband Data Report (Sixth), https://www.fcc.gov/reports-research/reports/international-broadband-data- reports/international-broadband-data-report-4. 3 47 U.S.C. § 1303(b)(2). 4 The price index measures the dollar amount that U.S. broadband subscribers would need to have added or subtracted from their incomes to purchase the same basket of broadband services under the pricing structures in other countries. Quantity weights for the price index are the share of broadband subscribers in the United States that take each of the four broadband speed tiers chosen for analysis. See infra paras. 23-28. 5 A hedonic regression provides an empirical summary of how prices vary with the characteristics of a good. In this Report, the hedonic regression builds on the price index by allowing adjustment of prices for cost and demographic differences across countries and then predicting broadband prices for each country at the average U.S. values of these variables. See id. at paras. 29-32. 6 The pricing analysis in this Report is designed to account for: (1) the different costs of deploying and operating broadband networks; (2) demographic differences that affect demand for broadband service; (3) multi-product bundling in broadband pricing; (4) different product offerings in each country; and (5) the availability and quality of complementary content and applications. Id. at para. 7. (continued….) Federal Communications Commission DA 18-99 45 demonstrate that accounting for country differences in cost, demographic, and quality factors give different assessments of the state of the U.S. broadband economy relative to other countries. I. OVERVIEW AND DATA HIGHLIGHTS 3. This Report seeks to improve upon our pricing analysis to better compare the broadband plans in the United States with the 28 comparison countries. The complexity of residential broadband offerings makes comparisons across countries difficult. Based on our data collection, we find that the features and quality of broadband service vary widely across countries and providers. Broadband service is also frequently purchased as part of a bundle of services, which makes it difficult to identify the price of the broadband service. The plans differ with respect to: (1) download and upload speeds; (2) type of technology used to deliver broadband services; (3) limitations on use, including limits on upload and download volumes; (4) contractual conditions; (5) additional services included; and (6) consequences of exceeding usage limits with some plans reducing speeds, imposing surcharges, or shutting off service. 4. Fixed Broadband Pricing Results. In this Report, we extend the analysis used in past reports to provide a more comprehensive assessment of why prices differ across countries. In contrast to recent reports, we include both standalone broadband and broadband plans that are bundled with video in our analysis to more accurately reflect what consumers in each country are paying for their broadband services. This is a particularly important issue for the United States where it is estimated that 75 percent of consumers purchase broadband in bundles at large discounts.7 The results are summarized below. x Fixed Unweighted Average Prices. o Based on unweighted average prices for standalone plans with download speeds less than 10 Mbps, we find that the United States ranks 8th out of 13 countries with such plans.8 o Based on unweighted average prices for standalone plans with download speeds of at least 10 Mbps and less than 25 Mbps, we find that the United States ranks 18th out of 25 countries with such plans.9 o Based on unweighted average prices for standalone plans with download speeds of at least 25 Mbps and less than 100 Mbps, we find that the United States ranks 18th out of 23 countries with such plans.10 x Fixed Broadband Price Index. Rather than allowing for comparisons only within a particular broadband speed tier, we combine the different offerings in each tier into a single broadband price index so that countries can be ranked and compared by a single measure of price. This weighted average Purchasing Power Parity (PPP) adjusted broadband price index estimates what U.S. consumers would expect to pay in each country for the broadband service that they consume today.11 o For standalone broadband across all speed tiers, the United States ranks 21st out of the 29 7 Kagan, a media research group within S&P Global Market Intelligence, estimates that 75% of U.S. broadband subscribers from the top 5 publicly reported MSO’s subscribe to double or triple-play bundles. 8 See id. at para. 22, Tbl. 1b. 9 Id. 10 Id. 11 PPPs are currency conversion rates that convert to a common currency and equalize the purchasing power of different currencies. In other words, PPPs eliminate the differences in price levels between countries in the process of conversion. See OECD, Purchasing Power Parities— Frequently Asked Questions (FAQs), http://www.oecd.org/std/prices-ppp/purchasingpowerparities-frequentlyaskedquestionsfaqs.htm (last visited Jan. 16, 2018). (continued….) Federal Communications Commission DA 18-99 46 countries.12 o For broadband bundled with multichannel video service, the United States ranks 19th out of the 29 countries.13 o Based on the fixed broadband price index for both standalone and bundled offers in each country, the United States ranks 21st out of the 29 countries.14 x Fixed Hedonic Price Index. We adjust broadband prices for differences across countries in demographic and cost profiles using a hedonic regression. The hedonic regression also allows us to adjust for observable differences in broadband quality across countries (e.g., the speed and usage limits of each plan) and generate prices for a set of standardized broadband plans in every country to facilitate comparisons. We then calculate a hedonic price index to compare prices across countries. This index estimates what the average U.S. consumer would expect to pay for their service in each country if that country had demographic, cost, and quality profiles similar to the United States.15 o After adjusting for differences across countries in the cost and demographic factors outlined in the BDIA, as well as differences in broadband plan characteristics, our hedonic price index estimates that, for U.S. broadband service levels, the United States ranks 14th out of the 29 countries.16 o If in our hedonic analysis we then further adjust prices for the quality of broadband content, we find that the United States ranks 7th among the 29 countries.17 5. Mobile Broadband Pricing Results. Our mobile price comparison methodology is the same that we use for fixed broadband with two exceptions. First, we classify mobile broadband products by data usage allowances rather than by download speeds. Second, we account for bundling in this sector by analyzing multi-line shared data plans (i.e., family plans) rather than the video and broadband bundling that occurs in the fixed broadband sector.18 The results of this analysis are summarized below. x Mobile Unweighted Average Prices. o For individual plans with usage allowances of 2 GB or less, the United States ranks 18th out of the 22 countries that offer such plans.19 o For the highest usage individual plans with data usage allowances greater than 10 GB, we find that the United States ranks 21st out of the 28 countries that offer plans with such high usage limits.20 12 See infra para. 28, Tbl. 3. 13 Id. 14 Id. 15 The country rankings would not change if we predicted prices at the values of these variables for any other country or the average of these variables across all countries. The only difference in our results would be in the levels of the predicted prices. 16 See id. at para. 32, Tbl. 4, Model 2. 17 See id. at para. 32, Tbl. 4, Model 4. 18 In other words, a “bundled” mobile offering consists of a multi-line package rather than a combination of broadband and video. 19 See id. at para. 47, Tbl. 5. 20 Id. (continued….) Federal Communications Commission DA 18-99 47 x Mobile Broadband Price Index. We again combine the different product offerings at each provider within each country into a single broadband price index that measures what U.S. consumers would expect to pay in each country for their mobile broadband services. o For individual plan pricing, the United States ranks 25th out of the 29 countries at $76.87. However, similar to our findings for fixed pricing, the United States ranks significantly better in bundled (i.e., shared data) plan pricing at $51.00 for 18th place.21 o Combining individual and shared data plan pricing, the overall rank of the United States is 20th out of the 29 countries.22 x Mobile Hedonic Price Index. As with our fixed analysis, we calculate a hedonic index that estimates what the average U.S. consumer would expect to pay for their level of service in each country if that country had demographic, cost, and broadband quality profiles similar to the United States. o After adjusting for differences across countries in the cost and demographic factors, as well as differences in broadband quality, our mobile hedonic price index estimates that, the United States ranks 20th out of the 29 countries.23 o If we further adjust mobile prices for content quality differences, our mobile hedonic price index finds that the United States ranks 10th across the 29 countries at an average per-line monthly price of $60.63.24 6. Combining Fixed and Mobile Hedonic Price Index Rankings. Typical consumers in the United States subscribe to both fixed and mobile services, so we also measure overall broadband affordability by calculating the average monthly cost U.S. consumers would pay to subscribe to both services in each country. We find that the United States ranks 8th overall by this measure at $123.62 per month for a mobile and fixed broadband connection.25 II. FIXED BROADBAND PRICING ANALYSIS 7. The great challenge in conducting international price comparisons is that the supply and demand factors that generate different broadband prices and offerings vary widely from one country to the next. An analysis that seeks to make “apples to apples” comparisons of broadband prices across countries would, at a minimum, need to account for: (1) the different costs of deploying and operating broadband networks; (2) demographic differences that affect demand for broadband service; (3) multi- product bundling in broadband pricing; (4) different product offerings in each country; and (5) the availability and quality of complementary content and applications. We examine each of these factors below, describing how each factor may affect international price comparisons and how we account for it in our fixed pricing analysis. 8. Cost and Demographic Differences. Each country is a separate broadband market with different supply and demand conditions that give rise to the observed market structure and pricing. One primary factor in determining the costs of deploying a broadband network is population density. Countries with lower population densities (e.g., the United States) will have much higher per-household 21 See id. at para. 52, Tbl. 7. 22 Id. 23 See id. at para. 54, Tbl. 8, Model 2. 24 See id. at para. 54, Tbl. 8, Model 4. 25 See id. at para. 70, Tbl. C18. (continued….) Federal Communications Commission DA 18-99 48 deployment costs than countries with much higher densities (e.g., South Korea).26 Standard economic models of entry would predict relatively higher prices in the United States since for a given level of demand, the markup over marginal cost required to cover the fixed costs of deployment would need to be higher even if the market is competitive.27 As a result, countries with high population densities would be expected, all else equal, to have lower markups over cost and lower broadband pricing.28 On the demand side, we would expect that demographic characteristics such as higher income and education levels would lead to higher broadband demand, all else equal.29 Therefore, in our hedonic price index, we adjust for the types of cost and demographic differences across countries outlined in the statute by including controls for population density, education, and income in the hedonic regression.30 We then predict broadband prices for each country at the average United States values of these variables. 9. Broadband and Video Bundling. Consumer preferences for subscription video services, the quality of these services, and the extent of consumer bundling of video with broadband also vary widely across countries. The United States generally has both higher prices and subscription rates for multichannel video service than other countries. For example, while only 20 percent of German households subscribe to multi-channel video services, over 80 percent of U.S. households subscribe to such services despite much higher monthly fees.31 The explanation for these differences may be in large part due to differences in the quality of video services across countries.32 Table 1a shows the total and per capita investment costs of programming networks by country.33 U.S. programming networks spend 26 Fiber deployment costs in South Korea have been estimated to be as low as $110-$170 per location passed, whereas such costs are estimated at $1000-$1300 in Australia. See OECD, The Development of Fixed Broadband Networks at 8 (2014), http://www.oecd-ilibrary.org/science-and-technology/the-development-of-fixed-broadband- networks_5jz2m5mlb1q2-en. 27 See Timothy F. Bresnahan and Peter C. Reiss, Entry and Competition in Concentrated Markets, 99 J. Pol. Econ. 977, 981 (1991) (Equation 3), http://pages.stern.nyu.edu/~acollard/bresnahan-reiss.pdf. 28 This discussion assumes that fixed deployment costs are not subsidized by the government. To the extent they are, this may lower the markup required to cover fixed costs and therefore result in lower prices if the market is sufficiently competitive. In Appendix F, we indicate that many countries subsidize broadband deployment. However, we did not account for this in our pricing analysis. If we had, the U.S. ranking likely would have improved given the greater governmental broadband subsidization in the comparison countries relative to the United States. See infra Appx. F. 29 Although generally we would expect higher demand to increase prices all else equal, the effect of higher demand on prices is indeterminate as it depends on how higher demand affects marginal costs, competition, and the elasticity of broadband demand. 30 Income levels will also affect broadband input costs in a country since labor and materials will be more expensive in countries with higher income levels. We account for the effects of income levels on prices using two methods. The first uses the U.S. dollar (USD) broadband price as the dependent variable in the hedonic regression and includes country-level income as an independent variable. The second uses the PPP price as the dependent variable and does not include income as an independent variable in the regression. We find that the U.S. fixed and mobile broadband rankings are the same under either approach. 31 Jorn Krieger, German Pay-TV Market On the Rise (Jul. 15, 2015), http://www.broadbandtvnews.com/2015/07/15/german-pay-tv-market-on-the-rise/; New Street Research, 4Q15 Cable Trends Review (2016), http://www.newstreetresearch.com/download/NSR-Cable_16-03- 09_%284Q15_Cable_Trends%29.pdf. 32 Another contributing factor may be the substantially larger public broadcasting sector in Germany and other European countries. 33 The total investment costs of programming networks were obtained from Kagan, a media research group within S&P Global Market Intelligence. Kagan, a media research group within S&P Global Market Intelligence, TV Network Summary, https://platform.mi.spglobal.com/web/client?auth=inherit#industry/tv_NetworksSummary (subscription only) (last visited Jan. 16, 2018). FCC staff calculated the investment per capita costs by dividing the (continued….) Federal Communications Commission DA 18-99 49 nearly ten times more than networks in the United Kingdom, the second largest investors in programming, and nearly twice as much per capita. Economics would again predict that video bundles in the United States would need to cost significantly more than other countries in order to recoup their large fixed investment costs in programming quality. Country Investment ($000)* Investment Per Capita Denmark 531,335 93.10 Finland 223,222 40.68 France 2,304,892 34.53 Germany 2,148,024 26.14 Italy 1,674,811 27.61 Netherlands 1,181,428 69.58 Norway 528,743 101.47 Spain 1,038,761 22.37 Sweden 874,125 88.73 United Kingdom 6,600,854 100.96 United States 61,270,385 189.62 Table 1a 2016 Programming Investment (USD) *Source : Kagan, a media research group within S&P Global Market Intelligence 10. Differences in consumer preferences for video also lead to different firm pricing strategies across countries. While providers in the United States use mixed bundle pricing with steep discounts, pricing models with additive or minimal bundle discounts prevail in many other countries. These different pricing strategies are evident in our data. Table C2 in the Appendix of Supplementary Tables calculates the average percentage discount off broadband services when they are bundled with video for each country.34 We find that consumers in the United States receive a 20 percent discount on average compared to only 11 percent across all other countries.35 This impacts our broadband pricing comparisons for two reasons. First, broadband customers in the United States receive substantial discounts when they bundle video and broadband, and 75 percent of customers benefit from these discounts by purchasing bundled services.36 Second, standalone broadband prices are likely higher in the United States than other countries due to a pricing structure that incentivizes customers to purchase bundles.37 As a result, the data collection and analysis need to include both standalone and bundled offerings in each country to accurately reflect the prices that consumers actually pay for their broadband total investment costs by the country population totals. We obtained the population size of each country from the OECD. See infra Appx. E, para. 1 and Tbl. 1. 34 See infra para. 70, Tbl. C2. 35 Id. 36 See supra para. 4, note 7. 37 When pursuing a mixed bundling strategy, firms will generally raise the price of the standalone goods in order to draw consumers into purchasing the bundled product. See Yongmin Chen and Michael H. Riordan, Profitability of Product Bundling, 54-1 International Economic Review 35-57 (2013), https://www.jstor.org/stable/23352318?seq=1#page_scan_tab_contents. Also, while consumers that purchase both goods are generally better off, those that purchase a single product are generally worse off. See Daniel R. Vincent, Insights From Recent Economic Analysis of Bundling (2016) https://transition.fcc.gov/bureaus/mb/docs/policy/video_marketplace/position_statement_Vincent_2016.pdf. A full analysis of the welfare consequences of different bundling strategies by country is beyond the scope of our work. Federal Communications Commission DA 18-99 50 services. We account for bundled offerings in both our broadband price index and our hedonic price index by estimating the prices of various broadband speed tiers in each country when bundled with video services and including these prices in all of our indices. 11. Different Product Offerings. Different underlying consumer preferences for speed and other characteristics results in very different product offerings across countries. Speeds up to a gigabit are routinely available in the United States while these types of products are often not available in other countries. In our data, broadband offerings exhibit tremendous variation in their characteristics, including: download and upload speeds; type of technology; limitations on usage; and consequences of exceeding usage limits (e.g., access speed reductions, surcharges, service cut-off). As such, broadband quality differences need to be accounted for to not disfavor countries with demand for higher quality service. We account for quality differences in our hedonic index by using our hedonic regression to predict provider-specific prices for a set of standardized broadband products (e.g., plans with no long- term contract, unlimited usage, etc.) so that our index measures what the price would be for the exact same broadband plans in each country. 12. Content Quality and Diversity. Internet content quality and variety also vary widely across countries. Access to a broad range of valuable applications and content over both fixed and mobile connections increases the value that each user derives from broadband service. In the language of economics, Internet content would be considered a complement to broadband. Access to better content in one country, or even if the same content is more highly valued in one country compared to another, will increase the demand for broadband services. Access to high-quality content can affect both the demand for and costs of providing broadband, since: (1) consumers receive greater benefits from subscribing to broadband service; and (2) more or better content increases data usage and results in greater per subscriber variable costs due to increased bandwidth charges (i.e., transit payments) and fixed capital costs from increased capacity investment in the network. Since our hedonic price index seeks to calculate how much more or less consumers in the United States would pay for broadband holding the quality and benefits of the broadband service in each country fixed, we need to account for unobserved differences in content quality. We do so by including a proxy measure for content quality in our hedonic regression and then predict prices in each country at the U.S. value of this measure.38 13. Evidence that there are large differences in fixed broadband usage across countries is provided in Table C4 in the Appendix of Supplementary Tables.39 It shows that the average monthly data 38 Table C3 contains three measures of content quality by country. See infra para. 70, Tbl. C3. This table provides estimates for the number of websites in each country’s top level domain(s) (e.g., “.fr” is the top level domain in France), the number of web pages in the top level domain(s) and the percent of the top ten million web sites that are in each country’s dominant language. These measures are discussed in more detail in the Technical Appendix. Due to the large difference in English content relative to all other languages, we simply enter a dummy variable for whether the country’s dominant language is English into the model to control for content availability. The coefficient on this dummy variable is approximately the percentage difference in broadband prices between English and non-English speaking countries. See infoplease, Languages Spoken in Each Country of the World (infoplease, Languages Spoken), https://www.infoplease.com/languages-spoken-each-country-world (last visited Jan. 16, 1018); W3Techs, Historical Trends in the Usage of Content Languages for Websites (W3Techs, Usage of Content Languages), https://w3techs.com/technologies/history_overview/content_language (last visited Jan. 16, 2018); DomainTools, Domain Count Statistics for TLDs (DomainTools, TLD Count Statistics), http://research.domaintools.com/statistics/tld-counts/ (last visited Jan. 16, 2018); google.com (using Google Search Engine). 39 See infra para. 70, Tbl. C4; see also OECD Broadband Subscriptions by Country Table 1.1; Cisco, Advanced Editor (Cisco, Advanced Editor), https://www.cisco.com/c/dam/m/en_us/solutions/service-provider/vni-forecast- widget/forecast-widget/advanced.html (last visited Jan. 16, 2018); Cisco, VNI Forecast Highlights Tool (Cisco, VNI Forecast), https://www.cisco.com/c/m/en_us/solutions/service-provider/vni-forecast-highlights.html# (last visited Jan. 16, 2018); tefficient, Is High Mobile Data Usage Cannibalizing Fixed? (Aug. 22, 2017) (Tefficient Report), http://tefficient.com/is-high-mobile-data-usage-cannibalising-fixed/#more-3480; Stats NZ, Internet Service Provider (continued….) Federal Communications Commission DA 18-99 51 consumption per fixed Internet household varies widely across our sample of countries and that U.S. households consume considerably more data than those in other countries. Among the countries for whom we have individual country data, the United States is estimated to have the highest per Internet household usage at 161.5 GB while Germany and France have the lowest at 39.5 GB and 44.2 GB, respectively.40 The high usage in the United States indicates that these households are likely deriving more benefits from the content and applications provided through their Internet connections than subscribers in other countries. A. Data Collection 14. Sampling Methodology. To determine which providers to sample in each comparison country, we used the TeleGeography GlobalComms Database to select providers with broadband market shares of at least ten percent.41 This threshold was chosen to balance data collection costs with maximizing the representativeness of our broadband pricing sample.42 We then chose cities in each country by first including the country’s capital, as directed by the statute,43 and then adding major metropolitan areas from the OECD Metropolitan Database.44 We chose major cities on the basis of ensuring geographic diversity within the country.45 We selected between one and four cities in each of the selected countries with the exception of the United States. For the United States, we collected data from ten major cities due to the greater overall population and variation in provider offerings.46 This resulted in a sample of 83 cities including all 29 capital cities. We then randomly sampled a set of ten postal codes in each city from which we randomly selected ten addresses contained within each postal Survey: 2016 (Stats NZ, ISP Survey), http://archive.stats.govt.nz/browse_for_stats/industry_sectors/information_technology_and_communications/ISPSur vey_HOTP2016/Commentary.aspx (last visited Jan. 16, 2018); Australian Bureau of Statistics, Type of Access Connection (Australian Bureau of Statistics, Access Connection), http://www.abs.gov.au/ausstats/abs@.nsf/mf/8153.0 (last visited Jan. 16, 2018). 40 Some countries are not separately reported in the Cisco data. For some of these countries, we used data compiled by the telecom consulting firm Tefficient. See Tefficient Report. Otherwise, we used the Cisco estimate of average household consumption reported for the group of countries in which a country is included (e.g., the rest of Western Europe) or an average of this measure and a Tefficient report of usage for a single provider in a country. By combining both Cisco and Tefficient data, we have some individual country level fixed usage data for 18 out of the 29 countries in our sample. For mobile, we have some individual data for every country. 41 TeleGeography GlobalComms. We obtained these data on the TeleGeography GlobalComms Database as of June 2017. There were a few exceptions to the 10% rule. For example, Verizon is estimated to have a national broadband market share below 10% in the United States, but it was sampled due to being the largest FTTP provider as well as the second largest ILEC. Due to missing fixed market shares for Japan, we used estimates from Japan’s Ministry of Communications and Telecommunications. See Ministry of Internal Affairs and Communications, Subscriptions Shares First Quarter 2017 Data Public Release, Appx. Section 2 Fixed Communications, (1) Data Communications 1. Fixed Broadband at 8 (2017) (first quarter 2017 results), http://www.soumu.go.jp/main_content/000494106.pdf; Ministry of Communications, General Communications Infrastructure Bureau, NTT East and West Fiber Wholesale Service Provisioning Conditions, No. 34-2 (2015), http://www.soumu.go.jp/main_content/000390866.pdf. 42 On average, our sample covers 83.5 percent of all broadband subscribers over all countries. 43 47 U.S.C. § 1303(b)(2). 44 OECD, Metropolitan Areas, https://stats.oecd.org/Index.aspx?DataSetCode=CITIES (last visited Jan. 16, 2018). 45 47 U.S.C. § 1303(b)(2). 46 In all cases, we tried to collect at least three cities in each country. However, in some cases the OECD data were only available for a single city. (continued….) Federal Communications Commission DA 18-99 52 code.47 These addresses were then entered into providers’ websites to determine the product offerings at every address.48 While many providers’ websites displayed “promotional splash page” plans offered generally, entering an address allowed us to capture the variation in product availability within a city and more accurate pricing information. Where we could not collect address-level plan data because providers requested to call or e-mail customers back, we collected “promotional splash page” plans and assumed these plans were available at every address. 15. Collection of Broadband Prices and Timeframe. For our analysis, we collected fixed broadband plan prices and terms from providers’ websites between June and August of 2017. Both standalone broadband plans as well as “double play” packages of broadband bundled with multichannel video services were sampled. With some exceptions, we did not collect information on “triple play” bundles of phone, Internet, and video since the extent of the bundle discount received did not tend to increase with the addition of phone service and doing so would have greatly increased the data collection burden (i.e., adding all triple play and standalone phone plans).49 In cases where a provider did not offer Internet service without a customer also subscribing to voice services, we collected the phone plan that would result in the lowest price and indicated that the plan included phone service in the data.50 In such cases, we also collected triple play bundles from the provider that included that particular phone plan to isolate the broadband price when bundled using the methodology described below. Finally, if the provider did not offer video service, there would be no bundled plans in the data for this provider. 16. Given the large number of countries, providers, and possible product offerings, we limited the scope of the collection along two dimensions. First, we assumed customers were new to the provider and did not receive any special discounts that were not available to all customers (e.g., student discounts). Second, we only recorded information for the combination of features that resulted in the lowest price for a given plan. For example, we did not select add-on features (e.g., HBO); always chose the lowest priced equipment required for the plan; and assumed consumers were willing to sign up for a two-year contract if this offered the lowest price.51 17. We collected three types of data for each plan: general information, pricing data, and non-pricing data. General information captures the name of the plan, date of collection, and currency. For pricing data, we collected all pricing information available on the provider’s website including promotions, equipment fees, installation fees, and rebates to calculate the total cost of the broadband service plan. Non-pricing data capture information such as download and upload speeds, data usage allowances, number of channels, and contract duration. 18. Data Review Process. Upon completion of the data collection, we reviewed the data for accuracy and completeness. When certain variables essential for the analysis were unavailable, we made 47 For most countries, we used the GeoNames Postal Code Database to identify postal codes in each city and then sampled addresses from each postal code to ensure that we collected plan data from different areas within every city. See GeoNames, ReadMe for GeoNames Postal Code Files (last updated Jan. 5, 2018), http://download.geonames.org/export/zip/. Where the GeoNames Postal Code Database did not have a particular city’s postal codes, we used either alternative sources of postal codes or other sub-city geographies (e.g., for South Korea, we used Administrative Districts). 48 Actual speeds are particularly important for DSL plans because providers will commonly advertise higher speeds than most customers will receive. For example, a provider may offer a DSL plan with up to 50 Mbps download speed, but on average, most customers receive a much lower speed. Sampling actual speeds at addresses within a city allows us to estimate the true speeds that the average customer of the DSL plan is receiving. 49 When providers did not offer standalone broadband plans, we collected double plays with Internet and phone as well as triple plays with phone, Internet, and video. 50 We did not collect fixed broadband plans bundled with mobile services. 51 We only collected information for contracts exceeding two-year durations when a particular plan was only available beyond a two-year contract duration. Federal Communications Commission DA 18-99 53 additional assumptions to complete the analysis such as the following: x Plans that did not indicate a usage allowance were assumed to allow unlimited usage since the vast majority of fixed data plans did not have usage limits. x If a plan advertised a promotional price but had no duration, we assumed it lasted 12 months since this was the modal promotional length for other plans. x If there was no information on equipment rental costs (modems and set-top boxes (STB)), we assumed equipment was included in the price. 19. Broadband Price Calculation. After the data was reviewed, we then calculated the total cost of the plan over the first 24 months. A 24-month price was selected to produce a comparable pricing measure across plans that accounts for all promotional and regular pricing and amortizes one-time fees over a sufficiently long-term horizon. This price was calculated as follows. ???????????? ? ???????????? ? ??????????????? ? ???????????? ? ??????????????? ? ??? ? ?????????????? ? ??????????????? ? ????????????? ? ?? ? ??????????? ? ???????? ? ???????????????? ? ?????????????? ? ????????????? ? ??????????????? ? ????????? ? ?????????????? All countries except Canada, Japan, and the United States included taxes in the prices listed on their websites.52 To make the prices in these three countries match the post-tax prices in other countries, we added taxes for plans in these countries using OECD estimates.53 All prices were then converted to United States and PPP adjusted dollars for our analysis.54 20. We then matched all bundle plans with their corresponding standalone Internet and video component plans to calculate the following bundle discount percentage: ?? ? ?? ? ? ?? ? ??? ? ?? ? ?? ? ??? ? where ?? is the standalone Internet price, ?? is the standalone video price and, ?? is the bundle price. For most bundles, we were able to collect the exact corresponding Internet and video component plans. However, in cases where providers did not offer one or both components on a standalone basis, we 52 Another form of taxes in many European and Asian countries are media licensing fees that are used to subsidize public television and radio networks. A recent paper estimates that “two-thirds of European countries and half of Asian countries, households pay a media licensing fee on top of subscription fees.” These taxes are levied on all television users and on broadband subscribers in at least one country (e.g., Denmark). Given our methodology, our calculation of the implied broadband price when bundled with video is unaffected by the inclusion or exclusion of these fees. However, to the extent that broadband subscribers in a country are subject to these fees, our analysis understates the pricing for these countries. See Roslyn Layton and Michael Horney, Innovation, Investment, and Competition in Broadband and the Impact on America’s Digital Economy (2014), https://www.mercatus.org/system/files/Layton-Competitionin-Broadband.pdf. 53 See OECD, Triple and Quadruple Play Bundles of Communication Services (2015), http://dx.doi.org/10.1787/5js04dp2q1jc-en. For Japan, we added 5 percent to all prices according to this source. 54 For converting from local currency units to U.S. dollars, we used the World Bank’s Atlas Method which reduces the impact of exchange rate fluctuations in cross-country comparisons. See The World Bank, GNI, Atlas method (current US$), https://data.worldbank.org/indicator/NY.GNP.ATLS.CD (last visited Dec. 14, 2017); see The World Bank, GNI (current US$), https://data.worldbank.org/indicator/NY.GNP.MKTP.CN (last visited Dec. 12, 2017). OECD, Purchasing Power Parities (PPP), https://data.oecd.org/conversion/purchasing-power-parities-ppp.htm (last visited on Jan. 16, 2018) (measuring PPP in terms of national currency per U.S. dollar). (continued….) Federal Communications Commission DA 18-99 54 assumed a bundle discount of zero for that particular bundled offering. 21. After calculating the discount amount from the standalone prices for each bundled plan, we applied the bundle discount percentage equally to the standalone broadband and video component plan prices to arrive at the implied price of broadband when purchased in a bundle.55 To illustrate, suppose the standalone prices for a particular video and broadband plan are $100 and $50, respectively, but the two can be purchased in a bundle for $120. Then the bundle discount is 20 percent and the implied price of the video plan when purchased in a bundle is $80 while the implied price of broadband when bundled is $40. This implied broadband price when bundled and the associated broadband characteristics would then be included as a plan in the dataset. In this manner, our analysis does not compare video and broadband bundles across countries, but rather isolates an implied price of broadband when bundled to avoid the video product comparability issues discussed above.56 B. Unweighted Average Prices 22. Table 1b below provides results comparable to previous Reports by calculating unweighted average prices for standalone broadband within certain speed tiers in each country. It also provides the number of plans that fall into each category on which the calculations are based. Using this methodology, we find that the United States ranks 18th out of 23 countries that offer plans with download speeds of at least 25 Mbps and less than 100 Mbps. For the highest speed plans with download speeds of 100 Mbps or greater, we find that the United States ranks 26th out of 28 countries in this category. Table C6 in the Appendix of Supplementary Tables provides unweighted average monthly prices for bundled products (in particular, the estimated average cost of the broadband component video-broadband bundles after accounting for the bundle discount).57 We find that the United States ranks 10th out of 20 countries that offer bundle plans with download speeds of at least 25 Mbps and less than 100 Mbps.58 For the highest speed bundle plans with download speeds at or above 100 Mbps, we find that the United States ranks 23rd out of 25 countries.59 55 Allocating the bundle discount percentage equally to each of the standalone components is equivalent to allocating the bundle discount amount in proportion to the standalone component prices. In a few rare cases where we calculated a slightly negative bundle discount, we assumed that the consumer would purchase the two separate standalone services and therefore set the bundle discount to zero. 56 The resulting price indices and rankings are robust to alternative methods of calculating implied broadband prices. We also produced analyses that included video plan dummy variables in the hedonic regression to remove the video component price from the bundle and found similar results to those reported here. 57 See infra para. 70, Tbl. C6. 58 Id. 59 Id. Federal Communications Commission DA 18-99 55 Mean Rank # Plans Mean Rank # Plans Mean Rank # Plans Mean Rank # Plans Australia 63.49 11 11 46.29 16 65 65.22 19 69 71.50 19 35 Austria 34.20 2 10 44.44 7 10 97.32 25 28 Belgium 37.35 4 4 68.18 16 6 Canada 39.89 6 4 58.48 22 7 69.65 21 20 95.43 24 16 Chile 70.33 12 6 98.07 25 6 54.26 15 12 80.17 21 12 Czech Republic 39.90 10 3 51.00 14 5 54.07 10 7 Denmark 32.56 3 2 34.93 3 5 36.59 3 10 61.98 13 12 Estonia 28.19 1 4 39.96 11 8 43.43 6 8 55.13 11 12 Finland 34.50 4 3 32.23 1 14 31.85 1 19 39.63 2 16 France 37.04 7 6 44.18 5 12 Germany 38.75 8 15 41.06 5 15 45.73 7 22 Greece 41.07 7 2 36.56 6 8 44.71 8 10 Iceland 39.83 9 4 44.66 6 15 Ireland 58.09 21 4 65.19 15 14 Italy 31.66 2 3 35.93 5 12 38.44 1 12 Japan 38.29 5 8 41.47 13 8 35.21 2 10 49.96 8 17 Latvia 35.56 4 1 47.76 11 1 43.44 4 4 Luxembourg 55.52 20 2 92.20 23 6 Mexico 49.70 17 7 75.39 23 4 104.90 27 3 Netherlands 49.72 13 6 58.64 12 15 New Zealand 59.90 23 12 59.61 17 20 68.39 17 39 Norway 48.79 9 2 52.72 19 4 70.18 22 12 113.30 28 13 Portugal 42.32 14 7 47.85 12 2 53.29 9 9 South Korea 43.19 3 30 Spain 62.76 10 6 76.15 24 3 69.65 20 21 87.50 22 15 Sweden 43.56 15 9 45.83 9 10 69.29 18 31 Switzerland 77.57 13 4 47.15 10 12 74.14 20 24 United Kingdom 40.03 12 16 55.17 16 36 63.20 14 12 United States 47.08 8 12 52.29 18 7 61.78 18 30 104.00 26 42 Average 47.40 67 47.18 243 51.52 346 67.40 479 Note : Unweighted mean prices are simple averages of all plans in the country and speed tier. Table 1b Fixed Unweighted Monthly Prices for Standalone Products (PPP) Country 0.2 ? Mbps < 10 10 ? Mbps < 25 25 ? Mbps < 100 100 ? Mbps C. Fixed Broadband Price Index 23. For purposes of comparing broadband pricing across countries, we need an estimate of “the price” of broadband in each country that accounts for all of the factors discussed above. Our approach is to follow well-established practices in the price index literature. We calculate a broadband price index for the same fixed set of broadband services to facilitate comparisons across countries. In general, a price index calculates the change in prices for a set of products or services by comparing the prices in the “base period” to those in the “comparison period.” One such index is the Consumer Price Index calculated by the Bureau of Labor Statistics of the U.S. Department of Labor.60 While the classic setting involves measuring price changes across time, our application to price changes across countries is analogous with the two periods now corresponding to two different countries. Our goal is to calculate the 60 See U.S. Department of Labor, Bureau of Labor Statistics, Frequently Asked Questions (FAQs), https://www.bls.gov/cpi/questions-and-answers.htm (last visited Jan. 16, 2018). (continued….) Federal Communications Commission DA 18-99 56 following Laspeyres broadband price index.61 24. In the formula above, pjt is the price of product j in comparison country t, pjo is the price of product j in the base country and qj0 is the share of product j in the base country. The index is therefore the ratio of the weighted average price of all of the j broadband products sold in the comparison country to the weighted average price of these same products in the base country where the weights are the percentage of broadband consumers that choose each product in the base country. 25. Ideally, we would calculate the price index over every broadband plan offered in every country. However, there are at least two difficulties in doing so. First, we would need to know the number of households that subscribe to each base country plan, and we do not have these data. Second, the broadband products available in each country are not the same, thus even if we had such weights, they would not be applicable in the comparison countries. Therefore, we need to aggregate all the available broadband plans in each country into a smaller more uniform set of products for which we have information on purchase quantities.62 We do this by aggregating all the broadband products offered in each country into j = 8 products. We define four standalone products classified by the following download speed tiers: at least 200 kbps but less than 10 Mbps; at least 10 Mbps but less than 25 Mbps; at least 25 Mbps but less than 100 Mbps; at 100 Mbps or above. We also define four additional products when these speed tiers are purchased in a video bundle.63 The resulting broadband products and their estimated U.S. market shares are shown in Table 2 below. 61 The Laspeyres price index is an upper bound for the average compensating variation from a price change. Compensating variation measures the dollar amount by which a given consumer would need to have their income adjusted to obtain the same level of utility under the comparison prices and product choice set. 62 Aggregating products in this manner is common in the differentiated products demand model literature. See Steven Berry, James Levinsohn, and Ariel Pakes, Automobile Prices in Market Equilibrium, 63-4 Econometrica 841 (1995), http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.554.3931&rep=rep1&type=pdf; Aviv Nevo, Measuring Market Power in the Ready-to-Eat Cereal Industry, 69-2 Econometrica 307 (2001), http://www.agecon.purdue.edu/academic/agec619/PP/IO_mats/Nevo%202001.pdf; Austan Goolsbee and Amil Petrin, The Consumer Gains from Direct Broadcast Satellites and the Competition with Cable TV, 72-2 Econometrica 351 (2004), http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.476.3034&rep=rep1&type=pdf. 63 The speed tier cutoffs were chosen to match quantity data available in the FCC’s 477 broadband subscription data collection. Federal Communications Commission DA 18-99 57 Product Name U.S. Product Share Standalone: 0.2 ? Mbps < 10 4.98% Standalone: 10 ? Mbps < 25 5.79% Standalone: 25 ? Mbps < 100 9.81% Standalone: 100 ? Mbps 4.42% Bundled: 0.2 ? Mbps < 10 14.93% Bundled: 10 ? Mbps < 25 17.38% Bundled: 25 ? Mbps < 100 29.42% Bundled: 100 ? Mbps 13.27% Fixed U.S. Shares by Broadband Product Table 2 Note : Product Shares were calculated by dividing each speed tier's connections by the total connections, and then multiplying by 25% for Standalone and 75% for Bundled Products. Source : FCC 2017 Internet Access Report. 26. The price index we calculate uses the United States as the base country to which other countries’ prices are compared. While the price index and relative country rankings will in general depend on which country is chosen as the base due to the use of different quantity weights, we chose the United States for several reasons. First, the focus of this Report is to evaluate whether the prices of broadband products purchased in the United States are comparable to other countries. Second, we have better estimates of the quantity weights for the United States than for any other country. Finally, the Laspeyres index ensures that U.S. broadband consumers would be at least as well-off as in higher ranked countries by measuring the dollar amount that U.S. broadband subscribers would need to have added or subtracted from their incomes to purchase the same basket of broadband services under the pricing structures in other countries. 27. The quantity weights for our price index were calculated using the FCC’s Form 477 data and are the share of broadband subscribers in the United States that subscribe to each of the four broadband speed tiers we have chosen for analysis.64 However, Form 477 does not provide an estimate of the percentage of U.S. subscribers that purchase broadband service bundled with video. For this we rely on the estimate from Kagan that 75 percent of U.S. consumers purchase broadband in a bundle.65 Calculating meaningful prices for each of our eight broadband products is more difficult. We again follow the price-index literature in implementing two common approaches: market basket and hedonic analyses. The market basket approach, discussed in Section II.C.1, calculates a simple weighted average price in each country for our eight products using the United States quantities as weights. The hedonic index discussed in Section II.D then extends the analysis by better accounting for missing product prices, quality differences within product groupings, and differences in the broadband cost and demand structures in each country. 1. Fixed Broadband Price Index Results 28. In Table 3 below, we present country rankings based on the fixed broadband price index, as well as this index divided by the average data consumption per user discussed above to calculate a 64 FCC, Internet Access Services: Status as of June 30, 2016 at 3, Fig. 2a (WCB 2017) (FCC 2017 Internet Access Report), https://apps.fcc.gov/edocs_public/attachmatch/DOC-344499A1.pdf. 65 See supra para. 4, note 7. (continued….) Federal Communications Commission DA 18-99 58 $/GB price.66 The United States ranks 21st out of 29 countries in standalone pricing, but higher for broadband bundled with video service at 19th due to large bundle discounts.67 Combining standalone and bundled pricing, the overall rank of the United States is 21st. On a dollar per GB of data consumed basis, the United States ranks 2nd out of 29 countries behind only South Korea. However, it may not be appropriate to divide the price by average data consumption. The problem is that data consumption not only affects broadband pricing as described above, but broadband pricing also likely affects data consumption. One could argue that for fixed broadband, the monthly subscription price should not affect usage since once this price is paid, most plans have no usage allowances or allowances that far exceed expected usage for most households. The flaw in this reasoning is that consumers choose whether or not to adopt broadband based on their expected monthly data usage and how much they value that usage.68 If prices are high in a country, then we would expect that consumers with lower expected data usage would not subscribe to broadband. Conversely, in countries with low prices, we would expect more low usage consumers to subscribe. As a result, given the same content, we would expect average fixed data usage to be higher for countries with high prices and lower for countries with low prices. Since higher prices in a country may lead to higher average data usage per household, it is problematic to divide price by usage that is uncorrected for this issue since this unfairly advantages countries with higher subscription prices and disadvantages those with lower prices. To account for content quality and the resulting data usage differences across countries, we enter a proxy measure of content quality that does not suffer from this issue directly into our hedonic regression. This isolates the effect of content quality on prices and allows us to predict prices from the hedonic regression holding content quality fixed. 66 The table presents the weighted average prices in each country for the indicated products. The Laspeyres index for each country would be calculated by dividing the given country’s price with the U.S. price. 67 We note that the bundle and standalone pricing measures are not strictly comparable in Table 3 because the plans that are included in each calculation may be different. For this reason, the bundle price in a country may be higher than the standalone price. 68 This is known as selection bias in the econometrics literature. Federal Communications Commission DA 18-99 59 Mean Rank Mean Rank Mean Rank Mean Rank Australia 60.78 20 70.26 27 67.89 27 0.70 12 Austria 50.69 16 39.31 6 42.15 8 0.65 8 Belgium 42.53 7 57.79 21 53.98 20 0.54 5 Canada 63.54 24 59.51 25 60.52 24 0.70 13 Chile 80.71 29 79.98 29 80.16 29 0.98 23 Czech Republic 48.29 14 45.40 16 46.12 16 1.30 28 Denmark 39.80 3 39.80 7 39.80 6 0.80 15 Estonia 44.47 9 42.78 15 43.20 12 1.21 26 Finland 35.11 1 29.89 2 31.19 2 0.70 11 France 40.41 4 26.09 1 29.67 1 0.67 9 Germany 44.81 11 40.87 9 41.86 7 0.84 18 Greece 44.55 10 56.76 20 53.71 19 0.83 17 Iceland 49.35 15 40.25 8 42.53 11 0.86 19 Ireland 63.17 23 58.54 23 59.70 23 0.57 6 Italy 35.98 2 39.05 5 38.28 5 0.90 21 Japan 40.46 5 42.74 14 42.17 9 0.73 14 Latvia 41.18 6 32.56 3 34.71 3 0.98 22 Luxembourg 71.30 27 61.72 26 64.12 26 1.00 24 Mexico 69.61 26 46.47 17 52.25 18 1.03 25 Netherlands 51.64 18 41.91 11 44.34 14 0.89 20 New Zealand 61.84 22 58.28 22 59.17 22 0.67 10 Norway 68.73 25 59.06 24 61.48 25 1.24 27 Portugal 46.52 12 41.00 10 42.38 10 0.65 7 South Korea 43.55 8 33.85 4 36.28 4 0.27 1 Spain 73.70 28 73.70 28 73.70 28 1.47 29 Sweden 47.14 13 42.23 12 43.46 13 0.43 4 Switzerland 57.24 19 50.08 18 51.87 17 0.81 16 United Kingdom 50.98 17 42.53 13 44.64 15 0.42 3 United States 61.65 21 52.62 19 54.88 21 0.34 2 Average 52.75 48.45 49.52 0.80 Note : The Standalone Index is calculated by averaging each country's weighted mean price by speed tier (Table C7) using US Product Shares as weights (Table 2). The Bundled Index is calculated by averaging each country's weighted mean price by speed tier (Table C8) using the US Product Shares as weights (Table 2). When a country did not have plans within certain speed tiers, we first assumed, if the highest speed tier was missing, the price of the highest speed tier is equal to the next highest available speed tier, and then we assumed that any other missing prices were equal to the next lowest available speed tier price. The Overall Index is the weighted average of the Standalone Index and Bundled Index using US Bundling Share (75%). Table 3 Fixed Broadband Price Indices (PPP) Standalone Bundled Overall $/GB Country Federal Communications Commission DA 18-99 60 D. Fixed Hedonic Price Index 29. A hedonic regression provides an empirical summary of how prices vary with the characteristics of a good and is a standard technique used to estimate and compare quality-adjusted prices and has been used for years in price index applications.69 To account for the remaining issues with our price index that we identified above, we develop a hedonic price index that accounts for differences between countries in: (a) cost and demographic factors; (b) quality of broadband offerings; and (c) content quality. 30. To do so, we estimate four hedonic regressions and then construct hedonic price indices from each model. Our hedonic regression is a multilevel model that allows the coefficients on each characteristic to vary by broadband provider to estimate provider-specific prices for each of our eight standardized broadband plans. While the details of the hedonic modeling are contained in the Technical Appendix, we summarize the basic approach here. The first model regresses the logarithm of each broadband plan’s price on the characteristics of that plan to account for how differences in plan characteristics (e.g., download speeds) explain differences in plan prices across countries. The second model builds upon the first by adding average income, population density, and a dummy variable for whether the country is highly educated into the model to capture how country-level differences in these demographic and cost factors influence pricing.70 The third model adds data usage per fixed Internet household to provide a comparison to the final model where we instead add a proxy variable for content quality into the regression due to the issues with using data usage discussed above.71 This final specification accounts for all of the price comparison issues we have identified. 69 U.S. Department of Labor, Bureau of Labor Statistics, Consumer Price Index, Quality Adjustment in the CPI (Nov. 20, 2017), https://www.bls.gov/cpi/quality-adjustment/home.htm. 70 See The World Bank, Urban Population (World Bank, Urban Population), https://data.worldbank.org/indicator/SP.URB.TOTL?name_desc=false (last visited Jan. 16, 2018); The World Bank, Rural Population (World Bank, Rural Population), https://data.worldbank.org/indicator/SP.RUR.TOTL (last visited Jan. 16, 2018); The World Bank, Population Total (World Bank, Population Total), https://data.worldbank.org/indicator/SP.POP.TOTL (last visited Jan. 16, 2018); The World Bank, Urban land area (sq. km) (World Bank, Urban Land Area), https://data.worldbank.org/indicator/AG.LND.TOTL.UR.K2 (last visited Jan. 16, 2018); The World Bank, Rural land area (sq. km) (World Bank, Rural Land Area), https://data.worldbank.org/indicator/AG.LND.TOTL.RU.K2 (last visited Jan. 16, 2018); The World Bank, Agricultural land (sq. km) (World Bank, Agricultural Land Area), https://data.worldbank.org/indicator/AG.LND.AGRI.K2 (last visited Jan. 16, 2018); The World Bank, Land area (sq. km) (World Bank, Land Area), https://data.worldbank.org/indicator/AG.LND.TOTL.K2 (last visited Jan. 16, 2018); The World Bank, GNI per capita, PPP (current international $) (World Bank, GNI per capita (PPP)), https://data.worldbank.org/indicator/NY.GNP.PCAP.PP.CDv (last visited Jan. 16, 2018); The World Bank, GNI Per capita, Atlas method (current US$) (World Bank, GNI per capita (Atlas)), https://data.worldbank.org/indicator/NY.GNP.PCAP.CD (last visited Jan. 16, 2018). See also OECD, OECD.Stat: Education and Training (2016) (OECD Tertiary Education), http://stats.oecd.org/ (last visited Jan. 16, 2018) (To access the data, click the left-hand column titled “Data by Theme” and click “Education and Training,” then “Education at a Glance,” then “Educational Attainment and Outcomes,” then “Educational Attainment and Labour- Force Status,” then “Educational Attainment of 25-64 Year-Olds,” then “Tertiary Education”). A country is classified as highly educated if the percentage of its population with a tertiary education exceeds the average in our samples. This classification is based on OECD data. OECD Tertiary Education. 71 As discussed in detail in the Technical Appendix, the content quality measure we use is whether the dominant language in the country is English. The results do not change substantially if we use the other measures of content quality reported in Table C3 in the specification instead of the dummy variable. Robustness checks not reported here also show that the results do not change significantly if we drop the United States from the estimation sample and then predict prices. (continued….) Federal Communications Commission DA 18-99 61 1. Hedonic Price Index Results 31. To calculate our cross-country comparison measures based on categories of download speeds, we predict firm-specific prices from the hedonic regression in each country for each of the eight standardized broadband products. For these predictions, we set the income, population density, education, and content quality variables to the U.S. values while the plan characteristics are standardized across all countries.72 This procedure effectively estimates what the prices of each plan would be in each country if income, population density, education, and content quality were at U.S. levels. We then aggregate these firm-specific price predictions for each of the eight products by using the same steps described in the Technical Appendix for the fixed broadband price index discussed above to arrive at the price that U.S. consumers would pay in each country for their broadband services if those countries had U.S. cost and demographic profiles. 32. The resulting country rankings under each model are shown in Table 4 below. These are the overall rankings over the standalone and bundled products in each country. In the unadjusted Model 1, we find that the United States ranks 23rd out of the 29 countries in our sample with an average broadband price of $58.00. Countries with lower average incomes like Latvia, the Czech Republic, and Estonia rank near the top before we correct the price levels for income. In Model 2, after we correct price levels for differences in income, education and population density, we find that the United States ranks 14th due to being a country with high income and education levels and low population density. We observe that countries with either high population densities such as South Korea or low income levels like Latvia exhibit the greatest price changes between Models 1 and 2. Model 3 includes the data usage variable but does not account for the issues with including this variable. Countries with higher prices and higher usage improve dramatically in rank if we enter this variable into the hedonic regression without correcting for prices causing higher data usage. For example, Sweden increases from 7th to 3rd and the United States jumps to 6th. Model 4 enters our content quality proxy variable of whether the country is English speaking into the hedonic regression, and finds that the United States ranks 7th least expensive out of the 29 countries. 72 We predict prices from the hedonic regression for broadband plans at the following download speeds for both standalone and bundled plans: 10 Mbps, 25 Mbps, 100 Mbps, and 1 Gbps. The plans are standardized to include only no contract plans with no phone service, a modem rental, and unlimited data usage allowances. Federal Communications Commission DA 18-99 62 Price Rank Price Rank Price Rank Price Rank Australia 78.30 28 82.81 27 102.63 26 84.45 23 Austria 48.04 17 60.59 15 73.17 11 74.02 17 Belgium 46.82 16 66.62 21 75.29 13 81.09 22 Canada 69.66 27 74.99 25 92.73 24 76.57 19 Chile 33.42 8 73.60 23 83.81 20 88.97 25 Czech Republic 26.83 3 49.18 6 69.91 9 60.49 6 Denmark 43.46 14 52.27 8 69.37 8 63.85 8 Estonia 30.65 6 56.91 12 81.68 19 69.06 12 Finland 35.00 9 37.95 1 57.49 2 51.61 1 France 30.12 5 44.04 4 61.96 4 54.25 3 Germany 36.00 12 53.62 10 75.09 12 66.06 11 Greece 35.38 10 64.51 19 80.72 17 78.66 21 Iceland 65.78 25 73.96 24 94.85 25 90.39 26 Ireland 56.79 22 62.37 16 76.46 14 64.83 9 Italy 29.62 4 48.00 5 68.80 7 59.00 5 Japan 40.12 13 53.58 9 81.47 18 72.12 15 Latvia 20.29 1 42.78 3 63.05 5 52.20 2 Luxembourg 56.32 21 54.32 11 76.83 15 72.51 16 Mexico 35.58 11 91.29 29 120.40 29 109.64 29 Netherlands 44.39 15 63.89 18 89.51 21 77.88 20 New Zealand 59.51 24 81.42 26 90.55 22 76.25 18 Norway 88.41 29 71.77 22 103.98 27 96.95 27 Portugal 30.82 7 58.27 13 72.83 10 71.15 14 South Korea 25.45 2 42.07 2 52.01 1 56.28 4 Spain 54.95 20 87.69 28 115.51 28 106.53 28 Sweden 52.48 19 52.16 7 61.08 3 70.41 13 Switzerland 66.88 26 65.01 20 91.15 23 84.46 24 United Kingdom 50.77 18 63.75 17 79.88 16 65.44 10 United States 58.00 23 59.84 14 64.75 6 62.94 7 Average 46.55 61.70 80.24 73.73 Table 4 Model 1: Unadjusted for demographics and content quality Model 2: Adjusted for demographics but not content quality Model 3: Adjusted for demographics and data usage Model 4: Adjusted for demographics and content quality Fixed Hedonic Broadband Price Indices Country Model 1 Model 2 Model 3 Model 4 III. MOBILE OVERVIEW AND DATA HIGHLIGHTS 33. The issues with comparing mobile broadband pricing across countries are very similar to those encountered in our fixed broadband pricing analysis. Mobile products also vary considerably with respect to data allowances, technology of service, speeds, contract length requirements, and other attributes that are important to consumers. Given similar issues involved in fixed and mobile price Federal Communications Commission DA 18-99 63 comparisons, we use a very similar approach in the mobile broadband pricing analysis, though certain changes were necessary. The two major differences are described below, and we explain in greater detail in the discussion that follows. x We again define a set of standardized products over which we calculate a weighted average price index in each country. However, mobile plans are generally sold by data usage allowances instead of download speed, thus we define the products by usage. In order to calculate the shares for each mobile broadband product in the weighted average, we need to estimate the percentage of U.S. consumers that would subscribe to each plan based on data usage profiles. x Mobile bundling involves the number of lines on a given plan rather than bundling other products such as video. In the United States, Cisco estimates that 75 percent of subscribers obtain their mobile service through shared data plans (i.e., “family plans”).73 Similar to fixed bundled broadband, these bundled plans are offered at greatly discounted rates and therefore need to be properly accounted for in order for our price index to reflect actual consumer purchases. 34. We again find that before adjusting for cost and demographic differences across countries, the United States ranks 20th out of 29 countries overall (combining individual and shared line plans) in our mobile broadband price index at $57.46 per month (PPP).74 However, after accounting for cross country differences in population density, income, education and broadband quality using our hedonic regression, we find that the United States now ranks 10th overall in our mobile hedonic index.75 IV. MOBILE BROADBAND PRICING ANALYSIS 35. While the methodology used in our mobile analysis is quite similar to our fixed methodology, there are some differences that we highlight below. 36. Product Quality Differences. Mobile broadband service offerings also vary substantially across countries, which makes price comparisons difficult. Data usage allowances, activation fees, roaming charges, contract lengths, and consequences of exceeding usage allowances are some of the plan characteristics that may differ both within and between mobile broadband providers. We use the same techniques for addressing these comparability issues in mobile broadband as we did with fixed broadband. We first define a set of five mobile broadband products based on the data usage allowances offered by the plan and whether the plan is a shared data plan. Our mobile broadband products consist of a three line shared data plan and four single line products based on the following usage allowances: 2 GB or less; 5 GB or less but greater than 2 GB; 10 GB or less but greater than 5 GB; and greater than 10 GB.76 37. In our hedonic price index analysis, we further standardize these products by predicting prices out of the hedonic regression for a set of standardized plans across all countries with exactly the same characteristics and with data usage allowances set at the upper threshold of the previously given 73 See Cisco, Cisco Visual Networking Index: Global Mobile Data Traffic Forecast Update, 2016– 2021 White Paper at 30, Fig. 36 (2017) (Cisco White Paper), https://www.cisco.com/c/en/us/solutions/collateral/service- provider/visual-networking-index-vni/mobile-white-paper-c11-520862.html. 74 See id. at para. 52, Tbl. 7. 75 See id. at para. 54, Tbl. 8, Model 4. 76 The shared plan product is assumed to have three lines because this is Ericsson’s estimate for the average number of lines on shared data plans in the United States. See Ericsson, North American Ericsson Mobility Report (2016) (Ericsson 2016), https://www.ericsson.com/assets/local/mobility-report/documents/2016/north-america-ericsson- mobility-report-june-2016.pdf. (continued….) Federal Communications Commission DA 18-99 64 ranges. We also control for 4G network availability in each country as an additional measure of mobile broadband product quality and predict prices at U.S. levels of coverage.77 38. Shared Plans. Mobile broadband consumers often bundle a number of lines on the same plan to receive a discount off what the service would cost if purchased as individual line plans. In the United States, these types of shared or family plans are the norm while they are much less prevalent in other countries. Cisco estimates that 75 percent of all mobile broadband subscribers in the U.S. purchase their services through shared data plans.78 This is not surprising given that our data show that U.S. consumers achieve significant savings by bundling multiple lines on a single plan. Table 5 estimates the average monthly per line cost for each of our five products at the country level. We estimate that a single line plan with a data usage allowance over 10 GB costs $93.48 on average in the United States while a shared plan that would allow similar per line data usage would cost only $49.52 per line.79 Thus, U.S. consumers save nearly $44 per line when they choose shared data plans over individual plans. In most other countries, shared plans either do not exist or do not offer as great of a savings compared to individual line plans. 39. Content Quality and Diversity. We again find that there are large differences in data usage per subscriber across countries. Table C5 in the Appendix of Supplementary Tables shows the average monthly data consumption per mobile broadband subscriber for each of the countries.80 Finland is estimated to have the highest data usage at 9.6 GB per month.81 The United States ranks 10th at 2.9 GB per subscriber per month.82 These findings are noteworthy because higher usage increases the costs of service both through higher fixed and operating costs and indicates that consumers are benefiting more from their broadband service. A. Mobile Data Collection 40. Collection of Broadband Prices and Timeframe. For our analysis, we collected mobile broadband plan prices and terms from providers’ websites between June and August of 2017.83 Given the wide scope of offerings by mobile providers, we attempted to only collect information for postpaid smartphone plans that allowed both unlimited voice calling and texting.84 However, where providers did not offer plans with unlimited minutes or unlimited text messages, we collected plans with the highest number of minutes and text messages available. We collected plan information in three broad categories: (a) general information including country, provider, plan name, and date of collection; (b) pricing information including all types of recurring and non-recurring costs of the plan such as promotional 77 See Open Signal, The State of LTE (November 2017), https://opensignal.com/reports/2017/11/state-of-lte (last visited Jan. 16, 2018). 78 See Cisco White Paper at 30, Fig. 36. 79 See infra para. 47, Tbl. 5. 80 Id.; see also Cisco, VNI Forecast; tefficient, Industry Analysis #1 2017, Mobile data – full year 2016 at 4, Fig. 2 (July 11, 2017) (tefficient, Industry Analysis), http://media.tefficient.com/2017/07/tefficient-industry-analysis-1- 2017-mobile-data-usage-and-revenue-FY-2016-per-country-11-July.pdf; OECD, Broadband Portal, Mobile Data Usage per Mobile Broadband Subscription, 2016, Table 1.14 (Dec. 2016) (OECD, Mobile Data Usage), http://dx.doi.org/10.1787/888933585343. 81 See infra para. 70, Tbl. C5. We use the average of the reported data usage estimates by country across the Cisco Mobile VNI, Tefficient and OECD data sources. 82 Id. 83 As with the fixed price data collection, we used the TeleGeography GlobalComms Database as of June 2017 to select providers with broadband market shares of at least ten percent. 84 By postpaid plans, we refer to plans that paid after usage (i.e., not prepaid or “pay-as-you-go” plans). By smartphone plan, we refer to only plans that have a data component. (continued….) Federal Communications Commission DA 18-99 65 prices, activation fees, and rebates; and (c) non-price information such as technology, usage allowance, number of minutes and text messages (when not unlimited), and above-cap data usage policy. We collected only plans available online and to new customers without any special discounts (e.g., youth or senior discounts). 41. We attempted to collect pricing information excluding the cost of handsets due to both the complexity that handsets introduce in measuring price and because most providers allow customers to bring their own devices. However, when pricing information was unavailable without selecting a handset (less than four percent of all plans), we selected the latest version of a widely available high-end handset (e.g., Apple iPhone, Samsung Galaxy) without insurance and noted whether a handset was included in the prices in a data field. 42. One of the most important price factors for mobile broadband service is the data usage allowance. We recorded the monthly allowance for each plan and any promotional data and the duration of the promotional data.85 In general, providers set a “soft” data limit per month before the provider imposes a consequence for exceeding these usage allowances.86 If a consumer exceeds the allowance, the provider may decrease mobile broadband speeds for the remainder of the month, charge overage fees (i.e., a consumer can have extra data at an extra cost), or stop the service entirely (i.e., a “hard” data limit). The structure of the data allowance policies varies by provider and can be quite complex. For example, some providers have several data allowance thresholds with different consequences for surpassing each one while other providers limit the amount of extra data a customer can buy. Some providers allow customers to choose from various data allowances. To simplify this issue for our data collection, we recorded overage fees whenever available, but recorded only the first consequence for surpassing the original usage allowance in the dataset. For overage fees, we recorded the amount of additional data and its cost (e.g., $15 for 1 GB), and if the provider offered multiple levels of additional data, we recorded the lowest priced extra data option. 43. We encountered a few issues unique to a small number of providers that required making assumptions about customer preferences.87 For providers that offered a plan with a set number of units to split between talk, text messages, and data, we split these equally across the services and recorded the exchange rate among the services (e.g., 1 unit = 1 minute = 1 text = 1 MB). Another issue was that Telia in Scandinavia offered a choice of included add-on options, including additional monthly data. If a customer could increase data on the plan without increasing the cost, we assumed the consumer would choose the additional data over the other services.88 44. Data Review Process. When the data collection was completed, we began reviewing the data for issues. When certain essential variables were missing, we made the following assumptions to complete the analysis: x If the website did not list usage allowances or consequences of exceeding usage, we assumed the plan offers unlimited data usage. x If a plan advertised a promotional price or data without specifying a duration, we again assumed the promotion lasts 12 months since this is the most frequent promotional length. 85 One complication is that some providers zero-rate data usage for certain services (e.g., music or video streaming) or include additional restricted-use data (e.g., extra data for outside the country). For the data allowance, we record only the generally available data provided with the plan. We also restricted the sample to plans with a monthly usage allowance greater than 200 MB. 86 In our regressions, “unlimited” is reserved for those plans that do not have a soft data usage allowances. For all plans, the data usage allowance is coded as the point at which such consequences begin for a user of the plan even though it may be possible to exceed this usage allowance in the month. 87 These issues affected only four providers and not all of the plans they offered. 88 For example, Telia in Denmark offered a 15 GB plan with the choice of optional services such as digital newspaper subscription, HBO Nordic, and 10 GB of additional monthly data. Federal Communications Commission DA 18-99 66 x If activation fees, access fees, other recurring and non-recurring fees, and rebates were not listed clearly on the providers’ website, we assumed that these fees were included or did not apply to the plan. 45. Broadband Price Calculation. After the data was reviewed, we then calculated the total cost of the plan over the first 24 months. A 24-month price was selected to produce a comparable pricing measure across plans that accounts for all promotional and regular pricing and amortizes one-time fees over a sufficiently long-term horizon. This price for individual plans was calculated using the formula below. For shared data plans, we also included the extra monthly cost of two additional lines. ???????????? ? ???????????? ? ??????????????? ? ???????????? ? ??????????????? ? ??? ? ?????????????? ? ??????????????? ? ????????????? ? ?? ? ?????? ? ????????? ? ???????????????? ? ?????????????? ? ????? ? ????????????? ? ????????? ? ?????? 46. Finally, we determined that all countries except some plans in Canada, Greece, and the United States included taxes in the prices listed on their websites. To allow the prices in these three countries to match the post-tax prices in other countries, we added the same percentage tax to the plan price as we used for our fixed pricing analysis.89 B. Unweighted Average Prices 47. Table 5 below presents the plan counts and unweighted average per line prices for all mobile broadband plans within the shared plan and data usage allowance categories that define our five products. These results are comparable to findings in previous Reports because we calculate simple averages within each data allowance usage. We find that the United States ranks relatively low in these price comparisons. For individual plans with usage limits of 2 GB or less, the United States ranks 18th out of 22 countries that offer such plans. For the highest usage individual plans with usage allowances greater than 10 GB, we find that the United States ranks 21st out of the 28 countries that offer plans with such high usage limits. As expected, we also observe that the per line monthly price in the United States for subscribers on shared data plans is significantly lower than comparable individual plans. A shared data plan with total usage allowances generally exceeding 20 GB costs only $49.52 per line. It is also worth noting that only 12 out of 29 countries in our study have any providers that offer shared data plans.90 89 See supra para. 19. 90 Shared plans may not have comparable usage limits to the individual plans in some countries. As a result, it is difficult to make direct price comparisons between shared and individual plans based on Table 5. We account for these usage differences in our hedonic analysis of mobile prices. See infra paras. 53-54. Federal Communications Commission DA 18-99 67 Mean Rank # Plans Mean Rank # Plans Mean Rank # Plans Mean Rank # Plans Mean Rank # Plans Australia 19.44 4 2 25.01 8 3 31.19 7 7 46.35 6 13 0.00 0 0 Austria 12.54 1 1 18.76 3 2 30.50 5 2 59.28 11 4 0.00 0 0 Belgium 0.00 0 0 41.20 16 1 54.31 15 4 56.18 8 1 0.00 0 0 Canada 74.69 22 6 85.24 26 6 111.53 23 6 166.91 28 6 58.67 11 24 Chile 0.00 0 0 67.02 23 2 112.90 24 2 84.30 19 7 0.00 0 0 Czech Republic 58.24 19 1 61.27 22 2 75.81 18 3 145.28 27 4 0.00 0 0 Denmark 0.00 0 0 21.02 4 3 26.05 2 4 40.34 4 9 24.82 2 9 Estonia 19.57 5 5 27.40 9 3 34.17 8 2 48.03 7 7 28.31 6 6 Finland 0.00 0 0 13.43 1 1 25.86 1 1 35.38 1 6 0.00 0 0 France 28.07 9 5 0.00 0 0 0.00 0 0 60.39 12 20 0.00 0 0 Germany 43.82 16 4 49.99 19 6 85.71 20 8 139.69 25 5 25.04 3 5 Greece 48.44 17 1 133.10 27 4 274.62 25 2 0.00 0 0 65.99 12 1 Iceland 16.98 2 3 23.37 6 2 30.93 6 3 56.30 9 14 27.49 5 12 Ireland 0.00 0 0 37.10 13 1 0.00 0 0 63.08 15 5 0.00 0 0 Italy 0.00 0 0 37.12 14 2 48.63 12 5 41.63 5 1 0.00 0 0 Japan 63.29 20 4 80.47 25 5 108.39 22 3 145.10 26 10 56.00 10 10 Latvia 18.07 3 5 22.85 5 2 36.06 9 2 38.26 2 3 0.00 0 0 Luxembourg 28.47 10 1 17.08 2 1 28.47 3 4 62.38 14 9 0.00 0 0 Mexico 31.39 11 3 48.19 17 5 83.51 19 12 132.13 24 3 0.00 0 0 Netherlands 63.60 21 3 49.72 18 1 68.36 16 2 107.67 23 6 0.00 0 0 New Zealand 0.00 0 0 39.06 15 3 51.70 14 4 75.85 17 3 26.04 4 3 Norway 26.49 8 3 34.79 12 3 42.27 11 2 62.21 13 6 0.00 0 0 Portugal 36.83 13 7 50.07 20 2 0.00 0 0 84.46 20 1 31.33 8 3 South Korea 40.53 14 11 57.34 21 8 75.22 17 6 99.64 22 13 0.00 0 0 Spain 34.75 12 1 0.00 0 0 50.70 13 3 66.79 16 3 0.00 0 0 Sweden 24.67 6 4 30.68 10 3 38.12 10 3 57.62 10 14 29.34 7 6 Switzerland 42.14 15 1 33.24 11 1 0.00 0 0 78.81 18 10 0.00 0 0 United Kingdom 25.38 7 5 24.99 7 6 29.75 4 10 39.61 3 19 18.94 1 3 United States 55.53 18 3 72.99 24 2 102.36 21 3 93.48 21 7 49.52 9 6 Average 28.03 79 41.47 80 57.14 103 75.42 209 15.22 88 Note : Unweighted mean prices are simple averages of all plans in the country and data allowance tier. Country Table 5 Mobile Unweighted Monthly Prices by Usage Allowance Tiers (PPP) Shared Data Plans10 < Data (GB)0 < Data (GB) ? 2 2 < Data (GB) ? 5 5 < Data (GB) ? 10 C. Mobile Broadband Price Index 48. In this section, we present a comparison of mobile broadband pricing across countries by calculating a mobile broadband price index using the same Laspeyres formula we used for fixed broadband.91 For each provider, we first calculate the average plan price for all of their offerings that fall into each of our five product categories. We then average these provider level prices for each product to the country level weighting by the market share of each provider. Finally, we calculate a weighted average of these product prices at the country level to form a single price index by using the estimated percentage of consumers in the United States that subscribe to each product to weight the product prices. 49. Product Market Shares. To perform the last step described above requires an estimate of the percentage of U.S. consumers that subscribe to each one of our five products defined by the data usage allowance. To estimate these product shares, we use Cisco Mobile VNI data coupled with an assumption on the shape of the usage distribution.92 Based on the Cisco data, we know that in the United States, 35 percent of mobile subscribers use less than 2 GB of data per month, 28 percent of mobile subscribers use between 2 GB and 5 GB, 17 percent of mobile subscribers use between 5 GB and 10 GB 91 We again calculate a Laspeyres price index that estimates how much consumers in the United States would pay for their mobile broadband plans in each of the comparison countries. The formula is identical to that used for fixed broadband. See supra para. 23-24. 92 See Cisco White Paper. Federal Communications Commission DA 18-99 68 and 10 percent of mobile subscribers use more than 10 GB. Cisco also finds that 75 percent of users subscribe to shared plans with an average usage of 4.1 GB, and 25 percent of users subscribe to individual plans with average usage of 5.5 GB of data per month. However, we do not have an estimate of the percentage of individual plan customers that fall into each of our data usage allowance categories. We only know that on average these customers have higher usage than shared plan customers and therefore need a way to estimate these percentages from the available Cisco data. 50. Consumer usage over nearly every communications network, including broadband, has been shown to be well approximated by the log-normal distribution.93 This makes estimating the distribution of data usage in a population particularly simple since a log-normal distribution is entirely determined by only two parameters; a location parameter that pins down the mean and a scale parameter that determines the shape of the usage distribution.94 Another important property of the distribution is that the percentiles are preserved if the mean of the distribution is shifted up or down.95 Using these properties of the distribution and the Cisco data, we are able to estimate the percentage of subscribers in the United States that have usage between the data usage allowances of each of our standardized mobile broadband products. The results of this approach are summarized in Table 6 below. The column with the heading “Cisco” presents Cisco’s estimate of the percentage of all U.S. mobile broadband consumers that have usage between the specified ranges of data consumption. The next column provides our estimates using a log-normal distribution calibrated to the Cisco data. We find that our estimates are a close match and that the log-normal assumption fits this data well. The next two columns provide our estimates for the percent of individual and shared plan subscribers that fall into each usage bucket. These serve as the product shares in our price index and would likely closely match the Cisco data if Cisco reported the data separately for individual and shared plans. Cisco Overall Usage Overall Usage Single Plan Usage Shared Plan Usage 0 < Usage (GB) ? 2 35.0% 35.7% 27.8% 39.0% 2 < Usage (GB) ? 5 38.0% 36.8% 36.8% 36.3% 5 < Usage (GB) ? 10 17.0% 18.3% 21.9% 16.8% 10 < Usage (GB) 10.0% 9.2% 13.5% 7.9% Plan Type Overall Individual Shared Source : Cisco White Paper . Table 6 Mobile Product Share Calculations Product 0.960 0.95 1.253 0.95 0.95 Mean Standard Deviation Estimated Distribution Parameters 1.042 51. Results. In Table 7 below, we present the country rankings based on the price index calculations described above. We present an index for individual plans, another for shared plans and an 93 I. Antoniou, V. Ivanov, Valery Ivanov & P.V Zrelov, On the Log-Normal Distribution of Network Traffic, Physica D: Nonlinear Phenomena, Volume 167, Issues 1–2, 1 (2002) at 72–85. 94 See George S. Ford, Approximating the Distribution of Broadband Usage from Publicly-Available Data at 7, n.5 (2012), http://www.phoenix-center.org/perspectives/Perspective12-03Final.pdf. A random variable is log-normally distributed if the logarithm of the variable is normally distributed. 95 Id. (continued….) Federal Communications Commission DA 18-99 69 overall index that is a weighted average of the individual and shared plan indices.96 For the overall mobile broadband price index, we allocate 75 percent of the weight to the family plan product since this fraction of customers in the United States subscribes to this product. If a country does not offer shared data plans, then the shared product price is the same as the individual product price index since customers would be required to purchase these plans instead of bundling lines at discounted rates. For the four individual plan products, we multiply the percentage of subscribers with usage that falls into each individual plan data usage category shown in the third column of Table 6 by the 25 percent of customers that subscribe to individual plans. If a country does not offer one of the four individual plan products, we follow the same procedure used in the fixed pricing analysis whereby we use the price of a plan for that provider with the next highest usage allowance. This assures that U.S. consumers are at least as well off with the plan provided as they would have been with the plan available in the United States. For example, T-Mobile USA only offers a single $70 unlimited plan, and we use this price for all four individual products for this provider. 52. Similar to our findings for fixed pricing, the United States ranks relatively lower in standalone (individual) plan pricing at 25th out of the 29 countries at $76.87 PPP, but significantly better in bundled (shared) pricing at $51.00 for 18th place.97 In individual plan pricing, Finland ranks first at $19.08 PPP per line per month while Latvia is the top country in shared data pricing at only $21.87 per line per month. Combining individual and shared data plan pricing, the overall rank of the United States is 20th. 96 The product prices by country that were used in the mobile price index calculations are presented in Table C13 of the Appendix of Supplementary Tables. See infra para. 70, Tbl. C13. 97 See infra para. 52, Tbl. 7. Federal Communications Commission DA 18-99 70 Mean Rank Mean Rank Mean Rank Mean Rank Australia 26.61 5 40.90 13 37.32 11 21.48 14 Austria 25.09 4 30.28 7 28.98 7 5.87 3 Belgium 45.85 13 48.54 17 47.87 16 70.24 25 Canada 99.96 28 60.48 23 70.35 25 51.38 22 Chile 79.07 26 92.60 28 89.21 28 74.35 26 Czech Republic 74.64 24 75.05 27 74.95 27 101.08 28 Denmark 24.94 3 22.83 2 23.36 2 5.93 4 Estonia 29.33 7 24.57 3 25.76 3 7.03 5 Finland 19.08 1 30.69 8 27.79 5 2.89 1 France 50.73 17 33.85 9 38.07 12 30.79 17 Germany 62.78 21 46.88 16 50.86 18 55.31 23 Greece 162.30 29 121.52 29 131.71 29 183.44 29 Iceland 27.49 6 26.47 4 26.72 4 7.72 6 Ireland 48.94 16 55.01 20 53.49 19 19.10 11 Italy 37.82 12 42.93 14 41.65 14 31.67 18 Japan 92.38 27 62.03 24 69.61 24 29.52 16 Latvia 24.82 2 21.87 1 22.61 1 3.42 2 Luxembourg 29.66 8 37.12 12 35.25 9 12.11 8 Mexico 65.21 22 56.33 21 58.55 22 50.43 21 Netherlands 68.67 23 74.90 26 73.34 26 90.33 27 New Zealand 47.38 14 43.84 15 44.72 15 42.31 20 Norway 37.63 11 34.77 10 35.49 10 16.15 9 Portugal 57.93 19 35.29 11 40.95 13 38.62 19 South Korea 62.16 20 57.00 22 58.29 21 16.38 10 Spain 48.29 15 51.62 19 50.79 17 58.34 24 Sweden 33.99 10 29.87 6 30.90 8 8.50 7 Switzerland 52.13 18 68.41 25 64.34 23 23.73 15 United Kingdom 31.04 9 27.22 5 28.17 6 20.48 13 United States 76.87 25 51.00 18 57.46 20 20.02 12 Average 53.20 48.41 49.61 37.88 Note : The Single Index is calculated by averaging each country's weighted mean price by data allowance tier (Table C13) using the Single Plan Usage Shares (Table 6). The Shared Index is calculated by averaging each country's weighted mean price of shared plans. For countries without shared data plans, we used a weighted average of the four single-line data allowance tiers using the Shared Plan Usage Shares as weights (see supra Table 6). $/GB Table 7 Mobile Broadband Price Indices (PPP) Country Single Shared Overall D. Mobile Hedonic Price Index 53. The mobile broadband price index in Table 7 does not account for several factors that likely affect the observed price levels in each country. We estimate a hedonic regression model to adjust prices for country-level differences in cost and demographic factors, differences in mobile broadband Federal Communications Commission DA 18-99 71 product quality (e.g., plan usage limits), and content quality. We then predict prices out of this hedonic model for a standardized set of mobile broadband products at the U.S. averages of the other control variables in order to estimate the mobile broadband prices that would be observed in each country if that country resembled the United States in cost and demographic characteristics.98 These predicted prices are then weighted together in the same manner that we used to calculate the previous price index to arrive at our hedonic price index measure for comparing mobile broadband prices across countries. 54. Our hedonic price indices for individual plans, shared plans, and overall are provided in Table 8. The differences between the four models presented are the same as our fixed pricing analysis. We again find that adjusting for cost and demographic factors is important when ranking countries by broadband pricing. Before adjusting for these income, education, and population density factors, the United States ranks 25th across the 29 countries in mobile broadband pricing (Model 1). Correcting for these factors in Model 2 changes the U.S. ranking to 20th. The U.S. ranking improves to 17th when we add actual data usage in Model 3. If we instead add our proxy measure for content quality in Model 4, we find that the United States ranks 10th overall in mobile broadband pricing. 98 For the greater than 10 GB plan, we predict the price for a 20 GB plan and for the family plan we predict at 5 GB of data usage per line. The other plan characteristics for the predictions are set to have: no contract, long distance included, unlimited text and voice, a limited data usage allowance and the handset not included in the price. Federal Communications Commission DA 18-99 72 Price Rank Price Rank Price Rank Price Rank Australia 32.16 12 36.49 7 39.54 4 45.65 5 Austria 22.94 3 33.47 6 65.68 21 47.40 6 Belgium 54.13 22 77.56 25 59.13 15 101.89 26 Canada 75.77 29 79.84 26 48.04 7 91.62 23 Chile 44.09 20 76.63 24 82.10 25 123.58 27 Czech Republic 27.31 6 41.12 9 37.86 2 55.94 8 Denmark 23.52 4 32.02 4 55.37 11 43.35 4 Estonia 16.41 2 24.21 2 41.01 5 34.58 2 Finland 24.24 5 27.27 3 53.05 10 49.40 7 France 28.12 7 50.28 14 62.37 19 82.72 19 Germany 33.23 14 50.32 15 50.66 8 70.22 12 Greece 70.83 27 105.96 29 85.07 28 146.34 29 Iceland 30.68 9 44.47 12 59.34 16 76.27 15 Ireland 30.98 10 47.54 13 72.37 23 64.16 11 Italy 36.89 17 61.55 21 77.75 24 91.18 22 Japan 56.12 23 59.14 19 58.06 14 82.09 18 Latvia 13.52 1 20.87 1 51.62 9 30.21 1 Luxembourg 33.74 15 43.01 10 63.54 20 71.99 14 Mexico 32.61 13 57.71 17 66.82 22 87.62 21 Netherlands 51.62 21 66.18 23 61.51 18 81.09 17 New Zealand 61.47 26 97.32 28 83.07 26 92.02 24 Norway 40.76 18 43.84 11 47.43 6 71.55 13 Portugal 34.52 16 55.42 16 56.67 13 79.68 16 South Korea 58.98 24 58.79 18 83.90 27 85.29 20 Spain 42.80 19 65.07 22 55.73 12 92.89 25 Sweden 32.14 11 32.15 5 35.71 1 57.46 9 Switzerland 73.78 28 80.64 27 90.61 29 134.40 28 United Kingdom 28.95 8 38.20 8 39.51 3 41.75 3 United States 61.01 25 60.89 20 60.99 17 60.68 10 Average 40.46 54.07 60.16 75.62 Model 1: Unadjusted for demographics and content quality Model 2: Adjusted for demographics but not content quality Model 3: Adjusted for demographics and data usage Model 4: Adjusted for demographics and content quality Table 8 Mobile Hedonic Broadband Price Indices Country Model 1 Model 2 Model 3 Model 4 V. TECHNICAL APPENDIX A. Fixed and Mobile Broadband Price Index Calculations 55. The methodology used to calculate the fixed broadband price indices in Table 3 and the mobile broadband price indices in Table 7 are provided below. The supplementary tables of broadband prices by individual product type referenced here are available in the appendix of supplementary tables. Federal Communications Commission DA 18-99 73 56. Step 1. To calculate our fixed broadband price index, we need estimates for each country’s price of each of the eight products in our Laspeyres price index formula. We first calculate a weighted average price within each of the products for each broadband provider where the weights are the product of city population and product availability. We then calculate a weighted average price across providers within each country and product type. The weight is now the estimated market share of each provider in each country. This weighted average is therefore more representative of the prices consumers in each country are actually paying for their broadband services. The resulting prices for each standalone product are shown in Table C7 and for each bundle product are shown in Table C8. For mobile, we assume that all plans are available on a nationwide basis in each country. We first calculate the average price for each product within each provider and then calculate a weighted average price for each product using estimated market shares for each provider as the weight. The resulting prices for each of the five products are shown in Table C13. 57. Step 2. While most of the eight products are offered in the United States, a number of countries do not have some products available. We therefore need a method to fill in these missing prices to establish a complete price index. For example, Belgian fixed providers do not have offerings below 25 Mbps, and providers in Spain and Denmark do not offer bundles. For missing standalone prices for fixed and mobile, we substitute the price of the next highest speed or usage tier that is available in the country.99 This method was chosen to ensure that our calculated price index is “proper,” meaning an upper bound to the compensation required to make the average consumer no worse off from the different prices and product choices in the comparison country. The reason that it is still an upper bound is that the primary product characteristics of broadband (e.g., speed and usage limits) are what economists call “vertical,” meaning that consumers always prefer greater amounts of them.100 Therefore, giving consumers more of certain broadband characteristics like download speed and data usage limits guarantees that consumers prefer the product that is actually available in the comparison country to the product that is only available in the base country.101 For missing fixed bundled prices, we use the standalone price that corresponds to the given bundle product’s speed tier since these products are not offered in a bundle and must be purchased on a standalone basis. For missing mobile shared data plan prices, we use the weighted average of the standalone prices in each country where the weights are the product shares calculated in Table 6. 58. Step 3. To calculate our price index for each country, we must put these eight product prices together into an overall weighted average price for broadband in each country. A Laspeyres price index would then be calculated as the ratio of this price in each comparison country to the United States average price.102 To calculate this overall average price, we first calculate the standalone and bundle country broadband rankings shown in Table 3 for fixed and Table 7 for mobile. For fixed, we calculate a weighted average price of the four products within each product category (i.e., standalone or bundled) using the percentage of U.S. consumers that subscribe to each speed tier as determined by data collected 99 For the highest speed tier, it was sometimes necessary to choose the price of the next lowest tier to fill in missing prices. 100 See Ariel Pakes, A Reconsideration of Hedonic Price Indices with an Application to PC’s (2002), http://www.nber.org/papers/w8715.pdf. For a discussion of these issues and the recommendation to use the approach outlined here. 101 The hedonic method improves upon this approach by estimating the missing prices directly from the hedonic regression. Hedonic price estimates are frequently employed in the analogous problem that arises in consumer price index calculations when a product is no longer found on store shelves. 102 See supra paras. 23-24 for Laspeyres price index formula. The difference between the comparison country price and the United States price is an estimate of the compensating variation. This measures the dollar amount that U.S. consumers would need to have their incomes adjusted by to ensure they are at least as well-off in the other country. (continued….) Federal Communications Commission DA 18-99 74 in Form 477 as the weights.103 To calculate the overall country-level price that accounts for both standalone and bundled broadband pricing, we combine the speed tier shares with the Kagan estimate that 75 percent of U.S. consumers bundle broadband with video to estimate the product shares shown in Table 2.104 For mobile, we combined the four single-line products into an overall single-line product using the product shares in Table 6, and then we combined the overall single index and shared index to produce an overall mobile broadband price index with a 25 percent weight on the single and a 75 percent weight on the shared index. 59. Step 4. To produce the Table 3 ($/GB) rankings, in the final step we simply divide the overall broadband price calculated in step 3 by the estimated average per Internet household data consumption in each country for fixed or the average per user data consumption for mobile. B. Hedonic Price Index Details 60. While the classic hedonic framework involves adjusting for changing product quality over time, accounting for product differences across firms and countries is analogous. Griliches (1961) proposes a linear hedonic regression of price on product characteristics of the form. ln(pijk) = ak + x'ib + ei In this equation, pijk is the price of plan i at provider j in country k, xi is a vector of plan characteristics and ei is a scalar error term. Under this approach, the country specific intercepts ak estimate the difference in the average price levels across countries accounting for differences in product characteristics xi. This is called the “time dummy hedonic price index” and has been widely used in comparing prices across time or space.105 However, this approach is not ideal for cross-country broadband pricing comparisons because it assumes that coefficients on product characteristics (the slope parameters b) are the same for each country. While it is plausible that the supply and demand conditions that generate the b coefficients are similar in adjacent time periods or even possibly cities within the same country, it is implausible that these conditions are similar from country to country. If broadband cost structures, determinants of demand (e.g. demographics), product offerings, ownership structure, regulatory conditions, subsidies or other conditions that impact prices vary across countries then we would expect the slope parameters to be different as well. 61. We estimate a more flexible model where the slope coefficients for certain characteristics are allowed to differ across countries and firms. However, due to sample size limitations in our pricing data, we do not estimate all of the j possible slope parameters for each product characteristic at the provider level, but rather use multilevel modeling techniques similar to those recently proposed in broadband price hedonic work at the OECD.106 The multilevel model recognizes that plans are nested within providers which are nested within countries and that prices are likely correlated within these nests. Rather than estimating separate parameters for each provider and product characteristic, the model assumes normally distributed mean-zero random effects on some product characteristics at the provider level and then estimates the variance of each random effect. The model is therefore more parsimonious since it estimates a single unknown variance parameter for each product characteristic rather than a separate slope parameter for each provider by product characteristic combination. Our base multilevel 103 This assumes that broadband bundle subscribers have the same probability of choosing each speed tier as standalone subscribers. 104 See supra para. 4, note 7. 105 See W. Erin Diewert, Saeed Heravi and Mick Silver, Hedonic Imputation versus Time Dummy Hedonic Indexes 161 (2009), http://www.nber.org/chapters/c5073.pdf. 106 See Carol Corrado and Olga Ukhaneva, Hedonic Prices for Fixed Broadband Services: Estimation Across OECD Countries (2016), http://www.oecd-ilibrary.org/docserver/download/5jlpl4sgc9hj- en.pdf?expires=1513350160&id=id&accname=guest&checksum=C9D52D13FAEA93DFFA6B2ECBD6888F4D. These models are also called “random coefficients models,” “hierarchical linear models,” and “mixed models.” (continued….) Federal Communications Commission DA 18-99 75 hedonic pricing equation (Model 1 in table 5 above) is as follows. ln(pijk) = aj + ak + x'ib + b1jx1 + uj + uk + ei 62. In this formulation, aj and ak are fixed provider and country level intercepts, while uj and uk are random intercepts for both providers and countries that measure country and provider differences from the sample means. The xi vector is the same set of product characteristics as before but now certain product characteristics, the first characteristic x1 in the example above, are assumed to have random slopes b1j that allow providers to have different costs and markups on this product characteristic.107 This random effect measures how each provider’s pricing of characteristic one differs from the pricing of the average provider in the sample as measured by the coefficient b1. In our fixed broadband hedonic models, the product characteristics with random coefficients are four download speed splines, the bundling dummy variable, the contract dummy variable, the dummy for whether phone service is included and the unlimited data usage dummy.108 In our mobile broadband hedonic models, there are random coefficients on three data usage allowance splines and the dummy indicatory variables for whether it is a shared data plan, has a contract, and whether long distance included.109 63. It is important to understand the correct interpretation of the hedonic coefficients. Under perfect competition, the b vector estimates both the marginal consumer value and marginal production costs for each product characteristic.110 However, in markets with substantial fixed costs like broadband, the coefficient also includes the markup over cost for that characteristic and these markups are complex functions of the characteristics of competing products, firm costs, consumer preferences, and market structure.111 Since we do not observe the product markup term, this is absorbed into the error term and will be correlated with the product characteristics. As such, in imperfectly competitive markets, hedonic coefficients should only be considered a reduced-form description of how prices (costs and markups) vary with changes in product characteristics and should not be given any interpretation beyond this.112 The focus should not be on the particular value or precision of any one coefficient, but rather on how 107 For the mobile hedonic model, there were an insufficient number of plans at each provider to estimate provider specific random effects on product characteristics. Instead, we estimate a model with country specific random effects on product characteristics and a provider random intercept parameter. 108 We control for download speed using a linear spline in the logarithm of download speed with knot points at the top-end of our speed categories used to define the eight broadband products (i.e., 10, 25, 100 and above 100 Mbps). 109 We control for data allowance using a linear spline in the logarithm of the data allowance with knot points at the top-end of our data allowance categories used to define mobile broadband products with the three highest data allowances (i.e., 2, 5, 10 and above 10 GB). There is no random coefficient on the 2 GB and less category because the fixed effect explained all of the variation in the data. 110 See Sherwin Rosen, Hedonic Prices and Implicit Markets: Product Differentiation in Pure Competition (1974), http://www.stern.nyu.edu/networks/phdcourse/Rosen_Hedonic_prices.pdf. 111 See Ariel Pakes, A Reconsideration of Hedonic Price Indexes with an Application to PC’s (2003) (2003 Reconsideration of Hedonic Price Indexes Paper), https://scholar.harvard.edu/files/pakes/files/hedonics_8-03.pdf; Robert C. Feenstra and Gordon H. Hanson, Foreign Investment, Outsourcing and Relative Wages (1995), http://www.nber.org/papers/w5121.pdf; Diane Bruce Anstine, How Much Will Consumers Pay? A Hedonic Analysis of the Cable Television Industry (2001), https://www.jstor.org/stable/41799034?seq=1#page_scan_tab_contents. Even if the broadband market is competitive in a country, pricing will still need to be above marginal cost in order for firms to recover their fixed deployment costs. 112 In fact, under plausible conditions the coefficients may have the opposite sign as would be expected, implying that goods with greater quantities of desirable characteristics have lower prices. See 2003 Reconsideration of Hedonic Price Indexes Paper; Ariel Pakes, Hedonics and the Consumer Price Index (2005), https://scholar.harvard.edu/files/pakes/files/hedonics-cpi_5-11-05.pdf; Ericsson 2016 (explicit derivations and Monte Carlo simulations). (continued….) Federal Communications Commission DA 18-99 76 predictive the hedonic pricing function is of provider prices in each country.113 We follow the standard hedonic approach as well but differ in that we also correct price levels for exogenous country-level factors that we expect may be correlated with costs and markups. These proxy variables include: (a) the logarithm of urban population density; (b) an indicator variable for whether the country is above the sample average in tertiary educational attainment; (c) the logarithm of income in Atlas adjusted U.S. dollars; (d) our dummy variable for whether it is a predominantly English-speaking country. The coefficients on these variables are again reduced form estimates of how prices are correlated with country-level factors and should not be given a causal interpretation for how we would expect price to change if, for example, we increased the income level of a country. 64. In Table C3, we report various proxy measures for content quality as well as each country’s primary language. The number of websites in top-level domains (TLDs) shows the count of all domains in each country’s main TLD (e.g., Germany uses .de) according to DomainTools.com. For the United States, we aggregate over several major domains: .com, .net, .org, .us, .gov, and .edu. Similarly, we used the same TLDs to report the number of web pages in the TLDs by searching Google’s search engine (“site:de”) and recording the number of search results. Another proxy measure is the percent of the top 10 million websites in each country’s primary language.114 From this data, we find that English- based website represent over 50 percent of the top 10 million websites. Although these statistics are not perfect measurements of content quality, they demonstrate that English language content is the dominant form of content available to broadband subscribers. Since the number of English language web sites is an outlier, we were concerned that adding this variable would result in estimating the effect based on the assumed functional form rather than the data. Therefore, we simply use an English dummy variable in the hedonic regressions to control for content quality differences between English and non-English speaking countries.115 65. The hedonic estimation equation we use for our final country rankings for fixed broadband is as follows. ln(pijk) = aj + ak + x'ib + zj'd + b1jx1 + uj + uk + eijk All terms are as before except now we include a vector of exogenous country level variables (income, education, population density, and the English language dummy) and the parameter vector d measures their effects on broadband prices.116 For our mobile broadband specification, we also add the availability of 4G LTE networks in each country to capture these quality differences.117 The last issue that we have not accounted for in the hedonic regression is product bundling. As noted above, most U.S. consumers purchase broadband and video service in a bundle at steeply discounted rates while significant bundle discounts and purchasing are much less common in other countries.118 It is very difficult to compare multichannel video products across countries. The product offerings in terms of channels included are 113 See 2003 Reconsideration of Hedonic Price Indexes Paper. 114 See W3Techs, Usage of Content Language for Websites, https://w3techs.com/technologies/overview/content_language/all (last visited Jan. 16, 2018). 115 We have found our results to be robust to using different measures of content quality as well as dropping the United States from the sample and then running the estimation. The United States ranks higher when we use the number of web pages or websites in English rather than the dummy variable. 116 The effective sample size for estimating the country level parameters is only 29. While this is not large, it is likely sufficient according to Monte Carlo simulations with multilevel models. See Mark L. Bryan Stephen P. Jenkins, Multilevel Modelling of Country Effects: A Cautionary Tale (2016), http://eprints.lse.ac.uk/61357/2/Jenkins_Multilevel%20modelling.pdf. 117 For our availability measure, we use the proportion of times that users in each country are able to connect to an LTE network from OpenSignal. See OpenSignal, The State of LTE (June 2017), https://opensignal.com/reports/2017/06/state-of-lte. 118 See supra paras. 4, 9-10, note 7. Federal Communications Commission DA 18-99 77 completely different across countries and even the same content may be highly watched in some countries (e.g., soccer in Europe) but uninteresting to most viewers in another country (e.g., soccer in the United States). Therefore, unlike broadband where a download speed of 25 Mbps is a more uniform product characteristic, there is no standardized video product that would be comparable across countries that would hold consumer utility fixed. Also, given the large differences in network investments in programming quality by country shown in Table 1a, it is unlikely that standard measures used to correct for video plan quality differences (e.g., number of channels) would be sufficient to account for these quality differences. 66. Since we did not believe the observable measures captured quality differences across countries, we explored alternative methods to account for broadband bundling with video services. The first is to directly control for video plan fixed effects in the hedonic regression. Including a video plan fixed effect essentially subtracts the average standalone price of the video service from the bundle so that we recover the implied price of broadband service when bundled. The method we chose to use instead, primarily due to greater simplicity, is to calculate a bundle discount and allocate this across the standalone component pricing as described above. We estimated the hedonic model using both methods and found that the estimates and resulting country rankings were similar. However, the first method was substantially more computationally demanding due to the extremely large number of video plan fixed effects in the model, so we instead chose the simpler approach. 67. Fixed Hedonic Results. Table C9 contains the coefficient estimates and robust standard errors clustered at the country level for the fixed broadband hedonic models.119 Table C10 contains the likelihood ratio tests and measures of model fit (R-squared). Each column provides the estimates for one of our four models. Model 1 regresses the logarithm of the U.S. dollar (USD) plan price on just the product characteristics. Model 2 adds the country-level zj measures to adjust for exogenous country level factors that would be expected to affect prices. Model 3 adds the country-level data usage measure to the previous model but does not correct for the endogeneity issue when including this measure. Model 4 then adds the content quality proxy in place of the data usage variable. 68. The estimated coefficients on each product characteristic generally conform to expectations. Across all models, we find that prices increase in download speeds, data usage limits, and when phone is included; they decline when consumers sign long term contracts or bundle broadband with video.120 For example, the estimated coefficient on bundling broadband with video in Model 1 indicates that the model estimates that consumers across all countries receive a 12.5 percent discount on average when they bundle broadband with video. Importantly, we also estimate that there is a large degree of heterogeneity in the hedonic function across countries and providers within countries as demonstrated by the random effects variance estimates in Table C10. 69. Adding the country-level regressors that adjust for income and population density in Model 2 shows that these variables are economically important. For example, Model 2 estimates that raising a country’s average income by 10 percent is expected to raise fixed broadband prices by 4.9 percent. Similarly, increasing the population density by ten percentage points would lower fixed broadband prices by 1 percent. Given that the United States is one of the wealthiest and least densely populated countries in the sample, adjusting for these factors has a significant impact on the estimated U.S. rankings. Evidence of this can be seen in how the estimated country random effect for the United States changes across model specification in Table C11. The estimated random intercept for the United States is .344 in Model 1 but this drops to only .059 in Model 2, which is just above the sample average. Including the content variable in Model 4 further lowers the estimated U.S. intercept below the sample average. 119 The mobile hedonic estimates can be found in Table C14. See infra para. 70, Tbl. C14. 120 Nearly 24%, or 571 out of 2,424 plans had phone service included with broadband. These are mostly DSL plans since DSL service is less costly to provide when a household also subscribes to phone service. Federal Communications Commission DA 18-99 78 VI. APPENDIX OF SUPPLEMENTARY TABLES 70. This appendix provides the supplementary tables referenced in the text. Australia 3 477 1 54,420 45,970 43.7 Austria 106 254 82 45,230 49,990 31.4 Belgium 353 721 102 41,860 46,010 37.5 Canada 4 231 1 43,660 43,420 56.3 Chile 24 669 13 13,530 23,270 22.5 Czech Republic 137 213 121 17,570 32,710 23.0 Denmark 132 407 54 56,730 51,040 38.2 Estonia 31 359 10 17,750 28,920 38.8 Finland 17 173 7 44,730 43,400 43.6 France 116 533 37 38,950 42,380 34.6 Germany 232 929 83 43,660 49,530 28.3 Greece 87 462 25 18,960 26,900 30.2 Iceland 4 226 1 56,990 52,490 40.5 Ireland 66 484 28 52,560 56,870 42.8 Italy 205 620 68 31,590 38,230 17.7 Japan 343 962 84 38,000 42,870 50.5 Latvia 35 498 10 14,630 26,090 33.4 Luxembourg 200 494 67 76,660 75,750 42.9 Mexico 59 695 24 9,040 17,740 16.8 Netherlands 487 1054 148 46,310 50,320 36.0 New Zealand 17 432 4 39,070 37,860 36.3 Norway 16 165 5 82,330 62,510 43.0 Portugal 117 579 42 19,850 29,990 23.8 South Korea 488 1903 86 27,600 35,790 46.9 Spain 93 473 31 27,520 36,340 35.7 Sweden 22 194 9 54,630 50,000 41.1 Switzerland 202 703 71 81,240 63,660 41.2 United Kingdom 258 908 50 42,390 42,100 46.0 United States 33 312 7 56,180 58,030 45.7 Note : In the Hedonic Regressions, the Higher Education Dummy Variable was defined as one if percent of population with tertiary education is greater than the sample mean. Population Density, Gross National Income Per Capita, and Tertiary Education Table C1 Country Overall Population Density Urban Population Density Rural Population Density Gross National Income Per Capita (Atlas Method, USD) Gross National Income Per Capita (PPP) % of Population with Tertiary Education Sources : World Bank, Urban Population ; World Bank, Rural Population ; World Bank, Population Total ; World Bank, Urban Land Area ; World Bank, Rural Land Area ; World Bank, Agricultural Land ; World Bank, Land Area ; World Bank, GNI per capita (PPP) ; World Bank, GNI per capita (Atlas) ; OECD Tertiary Education . Federal Communications Commission DA 18-99 79 Country Mean Rank Median Minimum Maximum # of Plans Australia 4.9% 21 7.3% 0.0% 8.1% 121 Austria 11.1% 14 9.0% 9.0% 17.6% 16 Belgium 13.2% 10 16.4% 6.8% 16.4% 6 Canada 11.2% 13 11.8% 1.7% 25.2% 67 Chile 9.8% 16 8.5% 6.6% 13.1% 24 Czech Republic 6.3% 18 3.4% 3.1% 29.7% 15 Estonia 4.1% 22 0.0% 0.0% 21.9% 57 Finland 16.3% 7 18.0% 13.0% 22.5% 18 France 14.6% 8 14.0% 10.5% 20.3% 38 Germany 11.8% 12 2.4% 0.6% 34.0% 62 Greece 0.4% 26 0.0% 0.0% 1.4% 8 Iceland 3.9% 23 3.9% 3.9% 3.9% 2 Ireland 5.1% 20 5.5% 0.0% 11.1% 16 Italy 0.1% 27 0.0% 0.0% 0.6% 9 Japan 7.9% 17 8.2% 3.1% 18.8% 25 Latvia 20.7% 4 19.8% 19.8% 23.7% 10 Luxembourg 10.2% 15 4.4% 1.9% 23.9% 20 Mexico 28.2% 1 27.6% 26.3% 30.7% 4 Netherlands 19.4% 6 33.9% 0.0% 36.0% 33 New Zealand 2.8% 25 2.9% 2.3% 3.2% 24 Norway 13.0% 11 13.0% 12.6% 13.5% 6 Portugal 23.4% 2 23.4% 22.4% 24.4% 6 South Korea 23.2% 3 32.2% 5.5% 42.0% 198 Sweden 5.8% 19 5.5% 0.0% 20.0% 74 Switzerland 3.1% 24 0.6% 0.1% 11.7% 47 United Kingdom 14.2% 9 11.1% 11.1% 24.0% 80 United States 19.8% 5 19.2% 7.4% 39.5% 369 Total 11.3% 8.5% 0.0% 42.0% 1355 Total w/o U.S. 11.0% 8.3% 0.0% 42.0% 986 Fixed Weighted Broadband Bundled Discount Rate Table C2 Note : Mean discount rates are the average discount rates weighted by city population and provider market share. Federal Communications Commission DA 18-99 80 Australia 2,805,837 820 51.2 English Austria 1,247,824 158 5.6 German Belgium 1,451,700 658 1.4 Dutch Canada 2,514,057 1,200 51.2 English Chile 409,448 16.8 5.1 Spanish Czech Republic 1,195,334 185 0.9 Czech Denmark 1,296,288 107 0.3 Danish Estonia 109,189 79.8 0.1 Estonian Finland 401,210 119 0.3 Finnish France 3,094,932 1,390 4.1 French Germany 14,400,000 1,450 5.6 German Greece 384,040 94.3 0.5 Greek Iceland 55,755 215 0.01 Icelandic Ireland 224,570 390 51.2 English Italy 2,714,357 1,420 2.4 Italian Japan 1,355,492 2,030 5.6 Japanese Latvia 107,224 85.4 0.1 Latvian Luxembourg 76,479 158 5.6 German Mexico 720,071 113 5.1 Spanish Netherlands 4,997,405 529 1.4 Dutch New Zealand 648,009 302 51.2 English Norway 678,893 280 0.1 Norwegian Portugal 248,152 102 2.6 Portuguese South Korea 960,106 80.2 0.9 Korean Spain 1,857,971 669 5.1 Spanish Sweden 1,661,108 251 0.5 Swedish Switzerland 1,775,969 632 5.6 German United Kingdom 10,400,000 1,710 51.2 English United States 159,000,000 35,167 51.2 English Note : For Number of Web Pages in Top-Level Domains (TLDs), we used Google’s search engine (“site:de”) and recorded the approximate number of search results. For the United States, we aggregate over several major domains: .com, .net, .org, .us, .gov, and .edu. Table C3 Various Measures of Content Quality Country Number of Websites in Top- Level Domains Number of Web Pages in Top-Level Domains (mm) Percent of Top 10 Million Websites in Country's Language Primary Language Sources : infoplease, Languages Spoken ; W3 Techs, Usage of Content Languages ; DomainTools, TLD Count Statistics ; google.com (using Google Search Engine) . Federal Communications Commission DA 18-99 81 Country Subscriptions Cisco Cisco Note Other Other Note Average Australia 7,374,000 86.8 Individual Country 106.1 Australian Bureau of Statistics 96.5 Austria 2,510,500 64.4 Country Group: All Western Europe 64.4 Belgium 4,265,026 49.7 Country Group: Rest of Western Europe 150.0 Tefficient: Provider - Telenet 99.9 Canada 13,347,882 85.9 Individual Country 85.9 Chile 2,904,580 53.0 Country Group: All Latin America 110.0 Tefficient: Provider - Movistar 81.5 Czech Republic 3,038,394 35.6 Country Group: Rest of Central & Eastern Europe 35.6 Denmark 2,430,002 49.7 Country Group: Rest of Western Europe 49.7 Estonia 384,787 35.6 Country Group: Rest of Central & Eastern Europe 35.6 Finland 1,712,000 64.4 Country Group: All Western Europe 25.0 Tefficient: Provider - DNA 44.7 France 27,683,000 44.2 Individual Country 44.2 Germany 31,867,148 39.5 Individual Country 60.0 Tefficient: Individual Country 49.8 Greece 3,616,705 64.4 Country Group: All Western Europe 64.4 Iceland 128,023 49.7 Country Group: Rest of Western Europe 49.7 Ireland 1,360,309 105.0 Tefficient: Individual Country 105.0 Italy 15,563,279 42.6 Individual Country 42.6 Japan 38,743,212 57.9 Individual Country 57.9 Latvia 519,154 35.6 Country Group: Rest of Central & Eastern Europe 35.6 Luxembourg 203,100 64.4 Country Group: All Western Europe 64.4 Mexico 16,277,627 50.9 Individual Country 50.9 Netherlands 7,135,000 49.7 Country Group: Rest of Western Europe 49.7 New Zealand 1,554,206 88.0 Stats New Zealand 88.0 Norway 2,120,360 49.7 Country Group: Rest of Western Europe 49.7 Portugal 3,372,571 65.0 Tefficient: Individual Country 65.0 South Korea 20,555,683 136.4 Individual Country 136.4 Spain 14,163,442 50.0 Individual Country 50.0 Sweden 3,679,768 101.2 Individual Country 101.2 Switzerland 4,198,150 64.4 Country Group: All Western Europe 64.4 United Kingdom 25,250,011 105.8 Individual Country 105.0 Tefficient: Individual Country 105.4 United States 106,327,000 161.5 Individual Country 161.5 Note : For Cisco Individual Countries, we calculated average monthly internet usage per household by using Cisco's VNI Forecast Widget to collect each available country's 2016 Consumer Fixed Internet (All Applications) average monthly usage and dividing by the OECD's number of fixed subscriptions. When available, we used individual country average monthly usage per household, but when country-level data was unavailable from Cisco or other sources, we averaged aggregate country groups and country-specific provider usage data Table C4 Fixed Subscriptions and Average Monthly Internet Usage per Household (2016) Sources : OECD Broadband Subscriptions by Country Table 1.1 ; Cisco, Advanced Editor ; Cisco, VNI Forecast ; Tefficient, Is High Mobile Data Usage Cannibalising Fixed? (Aug. 22, 2017); Stats NZ, ISP Survey ; Australian Bureau of Statistics, Access Connection . Federal Communications Commission DA 18-99 82 Country Subscriptions Cisco Tefficient OECD Average Australia 31,544,000 1.8 1.9 1.5 1.74 Austria 7,585,400 3.6 6.3 4.94 Belgium 7,480,397 0.5 0.9 0.68 Canada 24,973,809 1.3 1.5 1.37 Chile 12,914,417 1.2 1.20 Czech Republic 8,534,191 0.5 1.0 0.74 Denmark 7,058,216 3.5 4.4 3.94 Estonia 1,607,838 3.2 4.1 3.66 Finland 8,070,000 8.3 10.9 9.62 France 53,361,000 1.1 1.0 1.6 1.24 Germany 63,094,865 0.9 0.6 1.2 0.92 Greece 5,709,261 0.7 0.72 Iceland 353,903 3.0 3.9 3.46 Ireland 4,697,555 2.5 3.1 2.80 Italy 53,076,750 1.4 0.9 1.7 1.32 Japan 193,237,268 2.5 2.5 2.1 2.36 Latvia 1,555,566 5.0 8.2 6.61 Luxembourg 482,000 2.9 2.91 Mexico 74,512,528 1.6 0.7 1.16 Netherlands 15,017,000 0.6 1.0 0.81 New Zealand 4,916,375 1.1 1.06 Norway 5,038,159 1.8 2.6 2.20 Portugal 6,477,160 0.6 1.5 1.06 South Korea 55,713,362 3.2 3.6 3.8 3.56 Spain 41,471,985 0.8 0.9 0.87 Sweden 12,140,358 2.9 3.6 4.4 3.64 Switzerland 8,221,700 2.7 2.71 United Kingdom 58,706,343 1.1 1.2 1.8 1.38 United States 409,173,000 2.9 3.0 2.7 2.87 Note : For Cisco, we calculated average monthly internet usage per household by using Cisco's VNI Forecast Widget to collect each available country's Consumer Mobile Internet (All Applications) average monthly usage and dividing by the OECD's number of mobile subscriptions. Table C5 Mobile Subscriptions and Average Monthly Internet Usage per Household (2016) Sources : Cisco, VNI Forecast ; tefficient, Industry Analysis ; OECD Broadband Subscriptions by Country Table 1.1 . Federal Communications Commission DA 18-99 83 Mean Rank # Plans Mean Rank # Plans Mean Rank # Plans Mean Rank # Plans Australia 74.40 5 8 49.56 14 41 73.62 20 51 84.37 25 21 Austria 35.71 7 2 34.99 4 8 44.05 9 6 Belgium 60.12 17 6 Canada 52.55 16 11 60.04 18 39 79.04 24 17 Chile 62.92 4 6 149.33 18 3 47.27 13 9 69.51 19 6 Czech Republic 36.30 8 3 47.14 12 5 47.26 12 7 Denmark Estonia 28.19 1 4 31.42 4 14 36.92 5 14 49.06 13 25 Finland 28.91 2 2 30.12 3 7 27.64 2 8 32.61 2 1 France 23.37 1 12 17.74 1 3 34.52 4 23 Germany 42.07 12 12 39.41 6 21 39.99 7 29 Greece 43.76 13 6 66.28 19 2 Iceland 40.25 8 2 Ireland 57.80 16 16 Italy 39.84 7 3 38.21 6 6 Japan 37.68 10 3 45.96 11 22 Latvia 28.00 2 1 34.81 3 1 34.63 5 8 Luxembourg 51.46 15 5 75.20 22 15 Mexico 35.05 5 3 55.11 15 1 Netherlands 39.87 8 12 51.48 14 21 New Zealand 57.30 17 4 55.94 16 8 65.93 18 12 Norway 56.77 17 4 69.73 20 2 Portugal 35.22 6 2 45.38 10 4 South Korea 33.34 3 198 Spain Sweden 39.02 11 8 40.87 9 14 55.88 15 52 Switzerland 44.01 11 24 71.10 21 23 United Kingdom 37.48 9 20 50.60 14 56 20.73 1 4 United States 35.32 3 60 43.57 10 176 77.96 23 133 Average 45.95 80 45.30 157 45.62 459 52.96 659 Note : Unweighted mean prices are simple averages of all plans in the country and speed tier. Table C6 Fixed Unweighted Monthly Prices for Bundled Products (PPP) 0.2 ? Mbps < 10 10 ? Mbps < 25 25 ? Mbps < 100 100 ? Mbps Country Federal Communications Commission DA 18-99 84 Mean Rank Mean Rank Mean Rank Mean Rank Australia 64.08 11 47.46 16 61.67 18 72.53 20 Austria 36.92 6 46.05 7 94.49 26 Belgium 37.36 2 66.60 17 Canada 42.26 7 57.26 21 67.82 20 86.22 23 Chile 81.19 13 124.00 25 54.86 16 80.77 21 Czech Republic 42.98 13 51.55 14 54.00 8 Denmark 32.56 2 34.69 2 37.72 3 59.27 12 Estonia 34.56 3 39.24 8 46.33 8 58.32 11 Finland 35.63 5 33.26 1 32.13 1 43.54 3 France 36.20 4 43.60 5 Germany 43.49 15 44.50 5 48.73 6 Greece 41.07 6 39.02 7 48.01 12 Iceland 39.83 9 56.56 10 Ireland 61.70 23 64.29 16 Italy 31.66 1 36.30 5 37.35 1 Japan 34.75 4 39.88 10 39.64 4 49.45 7 Latvia 35.56 3 47.76 10 40.28 2 Luxembourg 52.00 18 85.91 22 Mexico 49.79 17 75.46 23 104.90 27 Netherlands 49.84 13 60.00 13 New Zealand 60.18 22 60.09 17 69.77 18 Norway 48.79 9 52.13 19 68.99 21 112.33 28 Portugal 41.99 11 47.85 11 54.58 9 South Korea 43.55 4 Spain 62.76 10 76.15 24 71.79 22 87.02 24 Sweden 43.31 14 44.90 6 61.44 14 Switzerland 77.57 12 46.59 9 71.92 19 United Kingdom 42.81 12 54.48 15 63.12 15 United States 45.32 8 52.36 20 63.65 19 87.78 25 Average 48.63 48.74 52.13 66.37 Fixed Weighted Monthly Prices for Standalone Products (PPP) Note : First, plan prices are averaged to a provider-speed-tier level weighting by the product of availability and city population (an approximate measure of the number of people in the city who have access to the plan). Then, these provider-speed-tier prices are weighted by provider market shares to produce the country- speed-tier mean prices. Table C7 Country 0.2 ? Mbps < 10 10 ? Mbps < 25 25 ? Mbps < 100 100 ? Mbps Federal Communications Commission DA 18-99 85 Mean Rank Mean Rank Mean Rank Mean Rank Australia 74.43 5 52.78 16 72.67 20 83.12 25 Austria 35.71 6 41.13 8 44.05 9 Belgium 57.79 16 Canada 51.66 15 60.93 18 75.47 24 Chile 74.00 4 149.33 18 46.78 11 69.51 20 Czech Republic 41.30 11 49.13 12 47.12 12 Denmark Estonia 34.56 2 37.36 7 44.11 9 56.21 15 Finland 28.76 1 29.93 2 29.20 2 32.61 2 France 30.10 3 17.74 1 34.86 4 Germany 40.04 10 40.46 6 43.80 8 Greece 44.18 13 66.28 19 Iceland 40.25 7 Ireland 58.54 17 Italy 39.84 5 35.37 5 Japan 37.92 8 46.39 11 Latvia 28.00 1 34.81 3 38.67 6 Luxembourg 48.49 14 71.74 22 Mexico 35.05 4 55.11 15 Netherlands 39.72 4 52.09 13 New Zealand 57.30 17 55.89 16 65.99 18 Norway 56.77 17 69.73 21 Portugal 35.22 5 45.38 10 South Korea 33.85 3 Spain Sweden 37.96 9 41.01 7 55.33 14 Switzerland 45.37 10 72.03 23 United Kingdom 41.90 12 53.05 13 20.73 1 United States 35.41 3 53.52 14 68.80 19 Average 49.43 46.35 47.17 52.78 Note : First, plan prices are averaged to a provider-speed-tier level weighting by the product of availability and city population (an approximate measure of the number of people in the city who have access to the plan). Then, these provider-speed-tier prices are weighted by provider market shares to produce the country- speed-tier mean prices. Country Table C8 Fixed Weighted Monthly Prices for Bundled Products (PPP) 0.2 ? Mbps < 10 10 ? Mbps < 25 25 ? Mbps < 100 100 ? Mbps Federal Communications Commission DA 18-99 86 Model 1 Model 2 Model 3 Model 4 -0.028 -0.028 -0.030 -0.029 (.084) (.084) (.084) (.085) 0.122 0.123 0.123 0.124 (.032) (.032) (.032) (.032) 0.114 0.113 0.112 0.113 (.026) (.026) (.026) (.026) 0.219 0.215 0.215 0.216 (.026) (.026) (.026) (.026) -0.134 -0.134 -0.134 -0.134 (.018) (.018) (.018) (.018) 0.024 0.025 0.026 0.023 (.018) (.018) (.018) (.018) -0.142 -0.158 -0.173 -0.165 (.039) (.039) (.039) (.038) 0.050 0.042 0.046 0.047 (.029) (.029) (.029) (.029) 0.106 0.107 0.107 0.107 (.026) (.026) (.026) (.026) -0.009 -0.005 -0.011 -0.010 (.093) (.087) (.084) (.085) 0.400 0.387 0.395 (.097) (.085) (.086) 0.160 0.016 0.054 (.106) (.107) (.100) -0.100 -0.161 -0.099 (.076) (.071) (.067) 0.349 (.122) 0.306 (.105) 3.832 3.838 2.449 3.816 (.078) (.076) (.491) (.069) Observations 2,478 2,478 2,478 2,478 Log Likelihood 1399.32 1412.96 1416.51 1416.66 P-Value 0.000 0.000 0.000 0.000 Table C9 Fixed Hedonic Estimation Results (USD) Spline: 100 ? Mbps Coefficient (Standard Error) Log of Urban Population Density Log(Price (USD)) Contract Dummy Spline: 0.2 ? Mbps < 10 Spline: 10 ? Mbps < 25 Spline: 25 ? Mbps < 100 Bundle Dummy Phone Dummy Unlimited Data Dummy Bring-Your-Own-Device Dummy Higher Education Dummy Note : All continuous independent variables are recentered at grand means. Model 3: Adjusted for demographics and data usage Model 4: Adjusted for demographics and content quality Log of Data Cap Constant Log of Income (USD) Model 1: Unadjusted for demographics and content quality Model 2: Adjusted for demographics but not content quality Log of Data Usage English Language Dummy Likelihood Ratio Test vs. Linear Model Federal Communications Commission DA 18-99 87 Model 1 Model 2 Model 3 Model 4 0.126 0.044 0.032 0.032 (.036) (.014) (.011) (.011) 0.093 0.092 0.093 0.095 (.057) (.056) (.056) (.057) 0.021 0.021 0.021 0.021 (.009) (.009) (.010) (.009) 0.043 0.043 0.042 0.042 (.009) (.009) (.009) (.009) 0.036 0.036 0.036 0.036 (.008) (.008) (.008) (.008) 0.019 0.019 0.019 0.019 (.004) (.004) (.004) (.004) 0.016 0.019 0.020 0.018 (.008) (.008) (.009) (.008) 0.004 0.004 0.004 0.004 (.004) (.005) (.004) (.004) 0.009 0.009 0.009 0.009 (.004) (.004) (.004) (.004) 0.101 0.088 0.081 0.084 (.051) (.046) (.044) (.044) 0.009 0.007 0.007 0.008 (.007) (.007) (.007) (.007) 0.012 0.012 0.012 0.012 (.000) (.000) (.000) (.000) Likelihood Ratio Test 1 vs. 2 2 vs. 3 2 vs. 4 P-Value 0.000 0.008 0.007 FE Only 0.061 0.402 0.388 0.537 FE and RE 0.946 0.946 0.946 0.946 FE and RE (no country RE) 0.460 0.800 0.810 0.876 FE and RE (no country and provider RE) 0.401 0.779 0.798 0.864 Model 2: Adjusted for demographics but not content quality Model 3: Adjusted for demographics and data usage Model 4: Adjusted for demographics and content quality R-Squared Values Model 1: Unadjusted for demographics and content quality Provider: Variance(Unlimited Data Dummy) Provider: Variance(Log of Data Cap) Provider: Variance(Constant) Variance(Residual) Table C10 Fixed Estimated Variances of Random Effects (USD) Provider: Variance(0.2 ? Mbps < 10) Provider: Variance(10 ? Mbps < 25) Provider: Variance(25 ? Mbps < 100) Provider: Variance(100 ? Mbps) Provider: Variance(Bundle Dummy) Provider: Variance(Contract Dummy) Coefficient (Standard Error) Country: Variance(Constant) Random Effect Parameters Provider: Variance(Bring-Your-Own-Device Dummy) Federal Communications Commission DA 18-99 88 Model 1 Model 2 Model 3 Model 4 Australia 0.531 0.263 0.211 0.107 Austria 0.145 0.062 -0.005 0.073 Belgium 0.083 0.105 -0.021 0.111 Canada 0.491 0.241 0.187 0.086 Chile -0.208 0.232 0.103 0.229 Czech Republic -0.472 -0.183 -0.102 -0.155 Denmark -0.057 -0.173 -0.152 -0.149 Estonia -0.338 -0.059 0.015 -0.047 Finland -0.179 -0.390 -0.246 -0.264 France -0.390 -0.311 -0.235 -0.279 Germany -0.173 -0.094 -0.028 -0.068 Greece -0.296 -0.023 -0.062 -0.009 Iceland 0.284 0.086 0.068 0.099 Ireland 0.254 0.034 -0.024 -0.097 Italy -0.451 -0.279 -0.185 -0.246 Japan -0.145 -0.165 -0.022 -0.057 Latvia -0.665 -0.242 -0.131 -0.214 Luxembourg 0.311 -0.024 0.044 0.065 Mexico -0.254 0.298 0.289 0.279 Netherlands 0.030 0.063 0.117 0.072 New Zealand 0.262 0.244 0.092 0.012 Norway 0.650 0.130 0.219 0.232 Portugal -0.320 -0.013 -0.050 0.003 South Korea -0.402 -0.214 -0.255 -0.110 Spain 0.157 0.289 0.288 0.289 Sweden 0.196 -0.110 -0.200 -0.002 Switzerland 0.421 0.079 0.142 0.149 United Kingdom 0.190 0.096 0.057 -0.047 United States 0.344 0.059 -0.113 -0.061 Total 0.000 0.000 0.000 0.000 Model 2: Adjusted for demographics but not content quality Model 3: Adjusted for demographics and data usage Model 4: Adjusted for demographics and content quality Table C11 Fixed Country Random Effect Estimates (USD) Country Country Random Effect Model 1: Unadjusted for demographics and content quality Federal Communications Commission DA 18-99 89 Price Rank Price Rank Price Rank Price Rank Australia 70.65 25 73.77 27 87.77 26 74.94 21 Austria 54.74 17 61.36 17 70.84 12 72.90 20 Belgium 53.55 16 64.87 21 70.86 13 76.86 24 Canada 72.11 26 71.01 26 84.47 25 72.25 19 Chile 56.17 19 68.23 24 76.23 19 80.85 26 Czech Republic 50.23 10 55.86 10 73.92 17 66.56 13 Denmark 41.06 6 47.98 6 59.84 5 56.97 6 Estonia 50.81 12 58.49 12 78.50 21 69.23 16 Finland 36.19 3 35.20 1 49.20 2 46.03 1 France 34.94 2 41.76 3 54.66 3 49.89 2 Germany 43.26 7 53.73 9 69.87 11 64.04 10 Greece 51.92 14 60.64 16 72.44 15 72.10 18 Iceland 52.90 15 58.83 13 71.76 14 70.27 17 Ireland 63.61 24 66.70 22 78.04 20 68.41 15 Italy 38.19 5 46.29 5 61.58 7 55.20 5 Japan 43.47 8 48.22 7 67.72 9 62.57 8 Latvia 37.55 4 44.54 4 61.07 6 52.92 4 Luxembourg 58.39 23 61.38 18 80.50 23 78.64 25 Mexico 78.16 29 94.92 29 119.93 29 111.94 29 Netherlands 50.68 11 63.96 20 83.35 24 75.68 22 New Zealand 57.82 21 67.40 23 73.49 16 63.72 9 Norway 74.02 27 70.57 25 95.03 27 91.82 27 Portugal 47.89 9 57.39 11 68.48 10 68.22 14 South Korea 33.33 1 39.15 2 46.20 1 50.48 3 Spain 74.99 28 88.07 28 110.18 28 104.47 28 Sweden 51.91 13 50.59 8 57.20 4 65.74 12 Switzerland 55.62 18 60.32 15 79.32 22 76.09 23 United Kingdom 57.35 20 63.00 19 75.43 18 64.14 11 United States 57.89 22 59.24 14 62.55 8 61.55 7 Average 53.43 59.77 73.81 69.81 Note : The dependent variable is log of prices (PPP) and income is not controlled for in Models 2, 3, and 4. Model 4: Adjusted for demographics and content quality Country Table C12 Fixed Hedonic Broadband Price Indices (PPP) Model 1 Model 2 Model 3 Model 4 Model 1: Unadjusted for demographics and content quality Model 2: Adjusted for demographics but not content quality Model 3: Adjusted for demographics and data usage Federal Communications Commission DA 18-99 90 Mean Rank Mean Rank Mean Rank Mean Rank Mean Rank Australia 19.44 5 22.96 6 28.17 3 48.78 7 40.90 13 Austria 12.54 1 18.77 3 32.81 7 55.66 9 30.28 7 Belgium 41.20 16 53.22 14 56.18 10 48.54 17 Canada 75.54 22 87.39 26 114.06 23 161.67 28 60.48 23 Chile 67.02 23 113.22 22 81.36 19 92.60 28 Czech Republic 58.24 19 61.49 22 75.25 19 143.31 27 75.05 27 Denmark 21.30 5 26.37 2 40.04 4 22.83 2 Estonia 19.14 4 27.61 8 34.34 8 46.86 6 24.57 3 Finland 13.43 1 25.86 1 35.10 1 30.69 8 France 29.67 10 58.84 12 33.85 9 Germany 40.40 14 50.88 20 65.48 16 136.96 24 46.88 16 Greece 48.44 17 130.90 27 284.35 25 121.52 29 Iceland 16.75 3 23.35 7 30.86 6 55.39 8 26.47 4 Ireland 37.10 14 70.53 16 55.01 20 Italy 34.74 13 44.54 12 41.63 5 42.93 14 Japan 63.69 20 82.19 25 116.70 24 139.81 25 62.03 24 Latvia 15.51 2 21.03 4 35.73 9 36.64 2 21.87 1 Luxembourg 28.47 9 17.08 2 30.78 5 64.56 14 37.12 12 Mexico 32.60 11 49.58 18 86.24 20 140.86 26 56.33 21 Netherlands 67.63 21 49.72 19 73.76 17 114.21 23 74.90 26 New Zealand 39.72 15 53.52 15 74.03 17 43.84 15 Norway 26.32 7 34.29 12 42.77 11 61.71 13 34.77 10 Portugal 35.97 13 49.00 17 84.46 20 35.29 11 South Korea 40.70 15 56.86 21 74.74 18 100.37 22 57.00 22 Spain 34.75 12 50.46 13 66.72 15 51.62 19 Sweden 24.17 6 30.25 9 38.41 10 57.26 11 29.87 6 Switzerland 42.14 16 33.24 11 79.62 18 68.41 25 United Kingdom 27.23 8 32.58 10 29.69 4 36.89 3 27.22 5 United States 56.84 18 73.16 24 103.32 21 85.32 21 51.00 18 Average 37.10 44.70 66.59 77.67 48.41 Note : Plan prices are averaged to a country-data-allowance-tier level weighting by provider market shares. For countries without shared data plans, we used a weighted average of the four single-line tiers using the Shared Plan Usage product shares as weights ( see supra Table 6). Table C13 Mobile Weighted Monthly Prices by Data Allowance Tiers (PPP) Country 0 < Data (GB) ? 2 2 < Data (GB) ? 5 5 < Data (GB) ? 10 10 < Data (GB) Shared Data Plans Federal Communications Commission DA 18-99 91 Model 1 Model 2 Model 3 Model 4 0.136 0.136 0.135 0.135 (.029) (.029) (.029) (.029) 0.230 0.230 0.231 0.231 (.051) (.051) (.051) (.051) 0.304 0.305 0.302 0.304 (.057) (.057) (.057) (.057) 0.344 0.342 0.343 0.342 (.044) (.043) (.043) (.043) -0.322 -0.322 -0.032 -0.323 (.045) (.045) (.072) (.045) 0.063 0.057 0.069 0.061 (.068) (.068) (.068) (.068) 0.111 0.090 0.011 0.115 (.125) (.129) (.125) (.128) 0.442 0.454 0.386 0.451 (.131) (.131) (.129) (.130) 0.256 0.255 0.259 0.254 (.059) (.059) (.057) (.060) 0.674 0.735 0.773 0.760 (.790) (.790) (.785) (.792) 0.112 0.116 0.122 0.115 (.101) (.101) (.101) (.102) -0.209 -0.248 -0.301 -0.280 (.586) (.586) (.581) (.587) -0.033 -0.032 -0.033 -0.031 (.072) (.072) (.072) (.072) 0.657 1.323 1.173 (.559) (.398) (.563) -0.011 -0.093 0.033 (.064) (.044) (.063) 0.112 0.261 0.127 (.198) (.136) (.182) 0.207 0.381 -0.062 (.236) (.158) (.249) -0.556 (.098) 0.564 (.257) 3.062 2.988 3.364 2.985 (.564) (.574) (.567) (.571) Observations 556 556 556 556 Log Likelihood 21.18 23.57 33.74 25.73 P-Value 0.000 0.000 0.000 0.000 Model 4: Adjusted for demographics and content quality Long Distance Dummy Unlimited Minutes Dummy Log of Minutes Unlimited Texts Dummy Log of Texts Constant Model 1: Unadjusted for demographics and content quality Model 2: Adjusted for demographics but not content quality Model 3: Adjusted for demographics and data usage Log of 4G Availability Log of Population Density Log Income (USD) Higher Education Dummy Note : All continuous independent variables are recentered at grand means. Likelihood Ratio Test vs. Linear Model Log of Data Usage Handset Dummy Table C14 Mobile Hedonic Estimation Results (USD) Log(Price (USD)) Coefficient (Standard Error) Spline: 0 < Data (GB) ? 2 Spline: 2 < Data (GB) ? 5 Spline: 5 < Data (GB) ? 10 Spline: 10 < Data (GB) Shared Dummy Unlimited Data Dummy Contract Dummy English Language Dummy Federal Communications Commission DA 18-99 92 Model 1 Model 2 Model 3 Model 4 0.024 0.024 0.023 0.024 (.014) (.014) (.014) (.014) 0.032 0.031 0.031 0.031 (.020) (.019) (.019) (.019) 0.030 0.030 0.030 0.030 (.013) (.013) (.013) (.013) 0.013 0.013 0.014 0.013 (.009) (.009) (.009) (.009) 0.102 0.114 0.143 0.119 (.058) (.067) (.084) (.070) 0.034 0.035 0.032 0.021 (.019) (.020) (.018) (.037) 0.230 0.180 0.056 0.142 (.072) (.062) (.029) (.054) 0.028 0.028 0.029 0.028 (.007) (.007) (.007) (.007) 0.028 0.028 0.028 0.028 (.002) (.002) (.002) (.002) Likelihood Ratio Test 1 vs. 2 2 vs. 3 2 vs. 4 P-Value 0.310 0.000 0.038 FE Only 0.215 0.343 0.547 0.369 FE and RE 0.929 0.929 0.929 0.929 FE and RE (no country RE) 0.903 0.903 0.904 0.903 FE and RE (no country and provider RE) 0.857 0.857 0.854 0.856 Country: Variance(5 < Data (GB) ? 10) Table C15 Mobile Estimated Variances of Random Effects (USD) Random Effect Parameters Coefficient (Standard Error) Country: Variance(2 < Data (GB) ? 5) Country: Variance(10 < Data (GB)) Country: Variance(Shared Dummy) Country: Variance(Contract Dummy) Model 3: Adjusted for demographics and data usage Model 4: Adjusted for demographics and content quality Country: Variance(Long Distance Dummy) Country: Variance(Constant) R-Squared Values Model 1: Unadjusted for demographics and content quality Model 2: Adjusted for demographics but not content quality Provider: Variance(Constant) Variance(Residual) Federal Communications Commission DA 18-99 93 Model 1 Model 2 Model 3 Model 4 Australia -0.033 -0.197 -0.228 -0.302 Austria -0.506 -0.432 0.026 -0.414 Belgium 0.257 0.305 -0.071 0.233 Canada 1.042 0.798 0.167 0.599 Chile 0.062 0.261 0.139 0.345 Czech Republic -0.216 -0.114 -0.256 -0.141 Denmark -0.346 -0.324 -0.003 -0.334 Estonia -0.694 -0.591 -0.206 -0.543 Finland -0.376 -0.546 -0.053 -0.300 France -0.683 -0.410 -0.314 -0.261 Germany 0.148 0.246 0.089 0.231 Greece 0.460 0.537 0.171 0.497 Iceland 0.083 0.150 0.247 0.333 Ireland -0.222 -0.108 0.075 -0.134 Italy -0.253 -0.067 0.017 -0.030 Japan 0.562 0.321 0.143 0.302 Latvia -0.915 -0.764 -0.114 -0.696 Luxembourg -0.340 -0.388 -0.127 -0.230 Mexico -0.151 0.071 0.047 0.122 Netherlands 0.210 0.159 -0.017 0.043 New Zealand 0.501 0.645 0.312 0.265 Norway 0.029 -0.152 -0.145 -0.034 Portugal -0.032 0.120 0.007 0.132 South Korea 0.488 0.201 0.328 0.225 Spain 0.098 0.209 -0.073 0.219 Sweden -0.102 -0.341 -0.274 -0.139 Switzerland 0.501 0.292 0.226 0.440 United Kingdom -0.138 -0.158 -0.233 -0.387 United States 0.566 0.277 0.121 -0.040 Total 0.000 0.000 0.000 0.000 Model 3: Adjusted for demographics and data usage Model 4: Adjusted for demographics and content quality Table C16 Mobile Country Random Effect Estimates (USD) Country Country Random Effect Model 1: Unadjusted for demographics and content quality Model 2: Adjusted for demographics but not content quality Federal Communications Commission DA 18-99 94 Price Rank Price Rank Price Rank Price Rank Australia 32.86 9 36.68 8 43.38 3 46.43 6 Austria 26.96 5 31.20 5 69.01 20 45.12 5 Belgium 62.26 24 67.15 23 56.78 13 89.60 20 Canada 79.43 28 88.45 27 49.51 7 101.68 25 Chile 72.13 25 93.55 28 91.70 29 150.78 28 Czech Republic 49.71 16 53.56 14 46.25 4 72.40 13 Denmark 22.20 1 24.18 1 49.29 6 33.37 1 Estonia 27.76 6 30.01 4 49.07 5 42.97 3 Finland 25.84 4 28.00 3 54.91 11 51.26 7 France 32.14 8 46.10 12 65.76 19 77.53 18 Germany 39.34 14 48.16 13 50.66 10 68.13 11 Greece 95.62 29 113.85 29 84.50 26 156.29 29 Iceland 24.52 2 32.55 7 49.51 8 57.33 8 Ireland 36.06 11 55.16 15 86.52 27 75.02 14 Italy 47.85 15 61.15 19 82.40 25 92.05 23 Japan 61.22 22 58.10 16 56.12 12 80.97 19 Latvia 25.13 3 27.76 2 65.55 18 40.17 2 Luxembourg 38.67 12 45.03 11 71.21 21 76.36 16 Mexico 73.00 26 87.43 26 87.80 28 131.89 27 Netherlands 60.00 19 60.08 17 62.37 16 75.12 15 New Zealand 60.34 20 80.17 25 74.21 22 76.69 17 Norway 39.17 13 41.68 10 50.26 9 69.36 12 Portugal 53.85 17 62.38 20 62.49 17 90.08 21 South Korea 75.65 27 62.56 21 82.09 23 90.22 22 Spain 59.94 18 68.58 24 60.41 14 99.03 24 Sweden 31.58 7 31.88 6 35.37 1 57.67 9 Switzerland 62.12 23 66.33 22 82.39 24 112.88 26 United Kingdom 32.99 10 40.67 9 42.24 2 44.42 4 United States 61.04 21 61.02 18 61.40 15 60.83 10 Average 48.60 55.29 62.87 78.13 Table C17 Mobile Hedonic Broadband Price Indices (PPP) Country Model 1 Model 2 Model 3 Model 4 Model 1: Unadjusted for demographics and content quality Model 2: Adjusted for demographics but not content quality Model 3: Adjusted for demographics and data usage Model 4: Adjusted for demographics and content quality Note : The dependent variable is log of prices (PPP) and income is not controlled for in Models 2, 3, and 4. Federal Communications Commission DA 18-99 95 Price Rank Price Rank Price Rank Australia 84.45 23 45.65 5 130.10 11 Austria 74.02 17 47.40 6 121.42 7 Belgium 81.09 22 101.89 26 182.98 24 Canada 76.57 19 91.62 23 168.19 21 Chile 88.97 25 123.58 27 212.55 27 Czech Republic 60.49 6 55.94 8 116.42 6 Denmark 63.85 8 43.35 4 107.21 5 Estonia 69.06 12 34.58 2 103.63 3 Finland 51.61 1 49.40 7 101.01 2 France 54.25 3 82.72 19 136.97 13 Germany 66.06 11 70.22 12 136.29 12 Greece 78.66 21 146.34 29 225.00 29 Iceland 90.39 26 76.27 15 166.66 20 Ireland 64.83 9 64.16 11 128.99 10 Italy 59.00 5 91.18 22 150.18 16 Japan 72.12 15 82.09 18 154.21 18 Latvia 52.20 2 30.21 1 82.41 1 Luxembourg 72.51 16 71.99 14 144.50 15 Mexico 109.64 29 87.62 21 197.26 25 Netherlands 77.88 20 81.09 17 158.96 19 New Zealand 76.25 18 92.02 24 168.28 22 Norway 96.95 27 71.55 13 168.50 23 Portugal 71.15 14 79.68 16 150.83 17 South Korea 56.28 4 85.29 20 141.58 14 Spain 106.53 28 92.89 25 199.43 26 Sweden 70.41 13 57.46 9 127.88 9 Switzerland 84.46 24 134.40 28 218.85 28 United Kingdom 65.44 10 41.75 3 107.20 4 United States 62.94 7 60.68 10 123.62 8 Average 73.73 75.62 149.35 Note : These hedonic price indices are based on our hedonic regressions that adjust for cost and demographic differences across countries. The reported price levels for each country are what an average U.S. consumer would pay for their broadband service in each country if that country had the same demand and cost profile as the United States. Table C18 Fixed & Mobile Hedonic Broadband Price Indices (USD) Country Fixed Broadband Mobile Broadband Fixed + Mobile Federal Communications Commission DA 18-99 96 APPENDIX D High-Speed Broadband Deployment Comparison with Europe 1. In this Appendix, as directed by the BDIA, we compare “the extent of broadband service capability” by examining fixed high-speed broadband deployment1 in the United States and 21 European countries (EU21).2 To match the European Commission’s (EC) definition of fixed high-speed broadband and its choice of technologies in our comparison, we examine U.S. deployment of fixed broadband with download speeds of 30 Mbps or higher,3 and do not include satellite technology. For the first time, we also compare mobile high-speed broadband deployment in the United States and EU21 by focusing exclusively on LTE, which is the baseline industry standard for the marketing of mobile broadband service.4 For our primary fixed and mobile analysis, we rely on data gathered in June 2015 and June 2016 by the FCC and the EC.5 We also present a historical overview of fixed deployment in the United States and the EU21 countries by presenting summaries of data from 2012 to 2016. Finally, we present maps that show fixed high-speed broadband deployment in the United States and Europe. I. DATA HIGHLIGHTS 2. Similar to previous years, the data show that the United States has better fixed broadband coverage than Europe, including in rural areas. The data further show that this is also true for mobile LTE broadband deployment. 3. Fixed High-Speed Broadband. In the United States, fixed high-speed broadband reached 90 percent of all households and 62 percent of rural households as of June 2016, which is up from 89 percent of all households and 58 percent of rural households as of June 2015. In the EU21, fixed high- speed broadband reached 76 percent of all households and 41 percent of rural households as of June 2016, which is up from 72 percent of all households and 30 percent of rural households as of June 2015. Our historical overview for 2012 to 2016 shows that the United States had higher deployment rates than the EU21 countries as a whole during the period both generally and separately in rural and non-rural areas. Deployment increased during the period, with the EU21countries having a somewhat higher growth rate. 1 47 U.S.C. § 1303(b)(1) (“As part of the assessment and report required by section 1302 of this title, the Federal Communications Commission shall include information comparing the extent of broadband service capability (including data transmission speeds and price for broadband service capability) in a total of 75 communities in at least 25 countries abroad for each of the data rate benchmarks for broadband service utilized by the Commission to reflect different speed tiers.”). 2 We refer to the set of countries that we compare here as the EU21, as we selected only 21 of the 31 European countries addressed in the EC Broadband Report as comparison countries for purposes of this Report. The EC Broadband Report discusses the 28 member countries of the European Union (EU), as well as Iceland, Norway, and Switzerland. See EC Broadband Report at 5. The 21 countries we include in our analysis are: Austria (AT), Belgium (BE), Czech Republic (CZ), Denmark (DK), Estonia (EE), Finland (FI), France (FR), Germany (DE), Greece (EL), Ireland (IE), Italy (IT), Latvia (LV), Luxembourg (LU), Netherlands (NL), Portugal (PT), Spain (ES), Sweden (SE), United Kingdom (UK), Iceland (IS), Norway (NO), and Switzerland (CH). 3 EC Broadband Report at 11. 4 Twentieth Mobile Wireless Competition Report, 32 FCC Rcd at 9018, para. 73. We note that the 2018 Broadband Deployment Report analyzes mobile LTE coverage data associated with 5 Mbps/1 Mbps and higher minimum advertised speeds in the United States and supplements that data with actual on the ground 10 Mbps/3 Mbps and higher median speed data measurements. In this Appendix, we analyze mobile LTE coverage regardless of minimum advertised speeds or actual speeds to match the EC Broadband Report. 5 We assess deployment as of June 2016 and June 2015 to match the European data. This speed standard differs from that used for the fixed broadband deployment data for the United States contained in the 2018 Broadband Deployment Report, which shows deployment of fixed broadband with 25 Mbps download and 3 Mbps upload speed as of December 2012 to December 2016. See 2018 Broadband Deployment Report at para. 15. (continued….) Federal Communications Commission DA 18-99 97 4. Mobile LTE Broadband. In the United States, mobile LTE coverage was already nearly ubiquitous by the end of June 2016, reaching almost 100 percent of all households and 98 percent of rural households. In the EU21, during the same period, mobile LTE coverage reached 97 percent of all households and 83 percent of rural households. This is a significant increase from the 88 percent of all households and 41 percent of rural households that mobile LTE reached as of June 2015.6 II. DATA SOURCES 5. Deployment Data. Both the FCC and the EC monitor broadband deployment. For comparison purposes, we rely on the FCC’s Form 477 fixed and mobile LTE deployment data to estimate U.S. broadband deployment.7 The Form 477 data reflect data gathered by the Commission as of the end of June in 2015 and 2016.8 We rely on data from the State Broadband Initiative (SBI) as of December 2012, 2013, and 2014, which the Commission relied on prior to the revision of the Form 477 data collection.9 We employ the centroid methodology to evaluate the deployment of LTE.10 We consider a census block to be covered by LTE if there is at least one service provider that reports coverage based on their Form 477 submission. For the EU21, we rely on fixed and mobile LTE deployment data provided in the EC Broadband Report.11 To match the fixed technologies used by the EC, we do not include satellite technology in the comparison of U.S. and EU21 deployment.12 The data gathered by the EC reflect data 6 We note that mobile LTE can be provided at various speeds in different countries. The mobile LTE deployment comparison in this Appendix is only a technology comparison. 7 FCC, Fixed Broadband Deployment Data from FCC Form 477, https://www.fcc.gov/general/broadband- deployment-data-fcc-form-477; FCC, Mobile Broadband Deployment Data from FCC Form 477, https://www.fcc.gov/mobile-deployment-form-477-data. References to the United States in this Appendix refer only to data collected from the 50 states and the District of Columbia. 2018 Broadband Deployment Report at para. 44. 8 FCC, FCC Form 477: Local Telephone Competition and Broadband Reporting—Instructions at 32 (2016), https://transition.fcc.gov/form477/477inst.pdf. 9 See Inquiry Concerning the Deployment of Advanced Telecommunications Capability to All Americans in a Reasonable and Timely Fashion, and Possible Steps to Accelerate Such Deployment Pursuant to Section 706 of the Telecommunications Act of 1996, as Amended by the Broadband Data Improvement Act, GN Docket No. 11-121, Eighth Broadband Progress Report, 27 FCC Rcd 10342, 10364-65, para. 28 (2012) (Eighth Broadband Progress Report). The SBI data were collected semi-annually through state-led efforts and maintained by the National Telecommunications and Information Administration for the National Broadband Map, in collaboration with the Commission. Id. at 10365, para. 28. 10 Inquiry Concerning the Deployment of Advanced Telecommunications Capability to All Americans in a Reasonable and Timely Fashion, and Possible Steps to Accelerate Such Deployment Pursuant to Section 706 of the Telecommunications Act of 1996, as Amended by the Broadband Data Improvement Act, 2016 Broadband Progress Report, 31 FCC Rcd 699, 730, para. 75, n.234 (2016 Broadband Progress Report) (explaining that the Commission evaluated the ability of mobile wireless providers to provide services throughout a census block by evaluating whether the provider’s shapefile overlaps the centroid of the census block); id. at 734-35, para. 82, Tbls. 4 and5 (reporting proportion of population with access to LTE technology). In the Twentieth Mobile Wireless Competition Report, the Commission presented coverage analysis based on both the centroid methodology and the actual area coverage methodology (which calculates the exact area of the block covered by each service provider by technology). See Twentieth Mobile Wireless Competition Report, 32 FCC Rcd at 9016-18, para. 71-72. At the aggregate national level, the results in terms of population covered will be similar whether the centroid methodology or the actual area coverage methodology is utilized and therefore, at that aggregate level, the centroid approach is a reasonable approach to take when comparing population based coverage. Id. at 9017-18, para. 72. 11 Previously, the EC reported deployment data as of the end of December of the targeted year. EC Broadband Report at 5. 12 Id. at 11. (continued….) Federal Communications Commission DA 18-99 98 as of the end of June 2015 and June 2016.13 We rely on data as of December 2012, 2013, and 2014.14 The EC Broadband Report provides a measure of progress towards Europe’s broadband coverage objectives.15 The EC Broadband Report is “designed to monitor the progress of EU Member States toward their specific broadband coverage objectives – namely: Universal Broadband Coverage with speeds at least 30 Mbps by 2020 and Broadband Coverage of 50 percent of households with speeds at least 100 Mbps by 2020.”16 6. Speeds. To facilitate an “apples to apples” comparison, we limit the scope of our regional comparison to fixed high-speed broadband service with at least 30 Mbps download speed, the same minimum speed used in the EC Broadband Report.17 We compare the U.S. fixed and mobile LTE deployment data to the EU21 fixed and mobile LTE deployment data as of June 2015 and June 2016. The EC Broadband Report also presents speed tier data at 2 Mbps, 30 Mbps, and 100 Mbps, which we also present below and which include any fixed technology capable of meeting those speeds.18 These data were calculated differently from the data used in the rest of this Appendix.19 7. Rural Definition. The EC Broadband Report includes data at the sub-national level— corresponding to counties, departments, or provinces—and are also broken down into rural and non-rural areas. The EC Broadband Report classifies European households as rural if they are within any square kilometer with a population of less than 100 people.20 The EC Broadband Report focuses on sub-national geographic areas with populations ranging from 150,000 to 800,000 (a geographical category known as the nomenclature of units for territorial statistics (NUTS-3)).21 NUTS-3 is a political-bureaucratic 13 Id. at 5. 14 See Eighth Broadband Progress Report, 27 FCC Rcd at 10364-65, para. 28. The SBI data were collected semi- annually through state-led efforts and maintained by the National Telecommunications and Information Administration for the National Broadband Map, in collaboration with the Commission. Id. at 10365, para. 28. 15 See generally EC Broadband Report. The EC tracks progress in 28 member states: Austria (AT), Belgium (BE), Bulgaria (BG), Croatia (HR), Cyprus (CY), Czech Republic (CZ), Denmark (DK), Estonia (EE), Finland (FI), France (FR), Germany (DE), Greece (EL), Hungary (HU), Ireland (IE), Italy (IT), Latvia (LV), Lithuania (LT), Luxembourg (LU), Malta (MT), Netherlands (NL), Poland (PL), Portugal (PT), Romania (RO), Slovakia (SK), Slovenia (SI), Spain (ES), Sweden (SE), and the United Kingdom (UK) and three additional countries: Iceland (IS), Norway (NO), and Switzerland (CH). 16 Id. at 5. 17 In the Fifth International Broadband Data Report, we used 25 Mbps given the Form 477 collected by speed tiers and was the closest speed to the EC’s 30 Mbps data for fixed high-speed broadband in the United States. See Fifth International Broadband Data Report, 31 FCC Rcd at 2671-72, para. 15. 18 EC Broadband Report at 11. The EC does not include satellite in the speed tiers. 19 Id. at 18. The speed tiers data are coverage by broadband networks capable of “realistically achieving” at least 2 Mbps, 30 Mbps, and 100 Mbps download speeds. These speed metrics were obtained at the country level only, not at the rural/regional level of the technology metrics. Id. at 19. 20 Id. at 15-16. In the EC Broadband Report, rural areas are defined by “using the Corrine land cover database and creating a database of population and land type in every square kilometre across Europe.” Id. at 16. The EC Broadband Report adds that households in square kilometers with population less than 100 were classified as rural. Id. The EC Broadband Report obtained from Point Topic updated estimates of rural population for 2015, which show that roughly 14 percent of households in the 31 EU study countries are rural. Id. According to U.S. Census block data, the U.S. rural share of households is slightly higher at 20 percent. 21 There are 1,357 NUTS-3 regions in the 31 EC study countries and 3,143 counties and county equivalents in the United States. EC Broadband Report at 11. Only 350 counties fall within the NUTS-3 population range of 150,000 to 800,000. Over 115 million Americans live in the 74 counties with populations above the NUTS-3 range, while 112 million and 87 million Americans live in counties within and below the NUTS-3 range respectively. The four least populous U.S. states (plus the District of Columbia) fall within the NUTS-3 population range. (continued….) Federal Communications Commission DA 18-99 99 jurisdiction that is a subdivision of NUTS-2. In this Report, we consider U.S. states as comparable to NUTS-2, and U.S. counties as comparable to NUTS-3 areas. There are 3,143 counties, parishes or boroughs (counties) in the United States, which would be equivalent to NUTS-3.22 8. The United States also relies on households as the unit of measurement. We rely on the U.S. Census Bureau’s method for identifying a U.S. census block23 as rural or non-rural,24 and then determine rural populations with or without broadband deployment in each county. Each county consists of multiple census blocks. In the maps below, we aggregate census block data to the county level to more closely match the level of aggregation in the EC Broadband Report.25 9. Technologies. The EC Broadband Report reports the following technologies that are capable of at least 30 Mbps download speed: VDSL, Fiber, and DOCSIS 3.0.26 The U.S. Form 477 deployment data include the following fixed technologies that are capable of at least 30 Mbps download speed: asymmetric xDSL, ADSL2, symmetric xDSL, VDSL, Cable Modem—DOCSIS 1, 1.1 and 2, Cable Modem—DOCSIS 3.0, Optical Carrier/Fiber to the End User, Copper Wireline, and Fixed Wireless. III. FIXED HIGH-SPEED BROADBAND COMPARISON 10. The figures and maps below compare fixed high-speed deployment in the United States and EU21 as of June 2015 and June 2016. A. Total and Rural Household Fixed High-Speed Broadband Deployment 11. As shown below in Figure 1, as of June 2015, fixed high-speed broadband was deployed to 89 percent of total U.S. households, while total coverage in the EU21 was 72 percent. 22 The population of U.S. counties varies widely outside of the NUTS-3 range, with the smallest having a population under 100 and the largest having a population over 10 million. 23 U.S. Census Bureau, Geographic Terms and Concepts—Block, https://www.census.gov/geo/reference/gtc/gtc_block.html (last viewed Jan. 16, 2018). 24 The U.S. Census Bureau defines rural areas as all territory, population, and housing units located outside urbanized areas and urban clusters. The U.S. Census Bureau defines an urbanized area as “consist[ing] of densely developed territory that contains 50,000 or more people,” and defines an urban cluster as “consist[ing] of densely developed territory that has at least 2,500 people but fewer than 50,000 people.” U.S. Census Bureau, Geographic Terms and Concepts—Urban and Rural, http://www.census.gov/geo/reference/gtc/gtc_urbanrural.html (last visited Jan. 16, 2018). 25 The Commission’s website provides online broadband maps. FCC, Maps, https://www.fcc.gov/reports- research/maps/ (last visited Jan. 16, 2018). 26 EC Broadband Report at 11. The EC includes all homes which are covered by at least one high-speed fixed technology capable of delivering 30 Mbps. The EC does not include satellite in the high-speed category. Federal Communications Commission DA 18-99 100 Figure 1 Fixed High-Speed Broadband Deployment All Households (June 2015) 12. Figure 2 presents fixed high-speed broadband deployment in the United States as of June 2016. We see that deployment rose from 89 percent as of June 2015 to 90 percent as of June 2016. In the EU21, deployment increased from 72 percent as of June 2015 to 76 percent as of June 2016. Figure 2 Fixed High-Speed Broadband Deployment All Households (June 2016) 13. Figure 3 below presents the fixed high-speed broadband deployment for all rural households in both the United States and EU21 as of June 2015. By June 2015, the United States had a greater percentage of rural household coverage at 58 percent as compared to the EU21’s rural coverage at 30 percent. Federal Communications Commission DA 18-99 101 Figure 3 Fixed High-Speed Broadband Deployment All Rural Households (June 2015) 14. Figure 4 compares rural household fixed high-speed broadband deployment for the United States with the EU21 rural household coverage as of June 2016. By June 2016, the United States had a greater percentage of rural household coverage of 62 percent as compared to the EU21’s rural coverage at 41 percent. Figure 4 Fixed High-Speed Broadband Deployment All Rural Households (June 2016) Federal Communications Commission DA 18-99 102 B. High-Speed Rural and Non-Rural Household Broadband Deployment 15. If we compare deployment in rural areas and non-rural areas, we observe deployment in rural areas lagging behind deployment in urban areas.27 In Figures 5 and 6 below, we compare the United States and EU21 rural and non-rural fixed high-speed broadband deployment as of June 2015 and June 2016. 16. Figures 5 and 6 below show that the U.S. deployment gap between non-rural and rural areas decreased from 39 percentage points in June 2015 to 35 percentage points as of June 2016. In the EU21, the gap between non-rural and rural areas decreased from 48 percentage points as of June 2015 to 42 percentage points as of June 2016. Figure 5 United States and EU21 Rural vs. Non-Rural Fixed High-Speed Broadband Deployment (June 2015) 27 We calculate the non-rural household coverage for both the United States and the EU21. We derive the non-rural households by subtracting the absolute number of rural households from the absolute number of total households. Federal Communications Commission DA 18-99 103 Figure 6 United States and EU21 Rural vs. Non-Rural Fixed High-Speed Broadband Deployment (June 2016) C. Total High-Speed Broadband Deployment by Country 17. Figures 7 and 8 below present the status of fixed high-speed broadband deployment for the United States and the EU21. In 2015, the United States ranked 11th among the 22 countries. In 2016, the United States ranked 10th among the 22 countries. Figure 7 Fixed High-Speed Broadband Deployment by Country for All Households (June 2015) Federal Communications Commission DA 18-99 104 Figure 8 Fixed High-Speed Broadband Deployment by Country for All Households (June 2016) D. Rural High-Speed Broadband Deployment by Country 18. Figures 9 and 10 below show the status of rural fixed high-speed broadband deployment across the United States and the EU21. The United States ranked 10th out of 22 countries in 2015 and 9th out of 22 countries in 2016 in terms of its rural coverage of high-speed broadband. Figure 9 Fixed High-Speed Broadband Deployment by Country for All Rural Households (June 2015) Federal Communications Commission DA 18-99 105 Figure 10 Fixed High-Speed Broadband Deployment by Country for All Rural Households (June 2016) E. High-Speed Fixed Broadband Deployment by Technology and Technology Combination 19. With regard to fixed high-speed broadband, cable is deployed to more U.S. households than any other technology. Similarly, DOCSIS 3.0 cable is deployed to more EU21 households than any other technology, though to a lesser extent than in the United States. Figure 11 Fixed High-Speed Broadband Deployment by Technology (June 2015) Federal Communications Commission DA 18-99 106 Figure 12 Fixed High-Speed Broadband Deployment by Technology (June 2016) F. Comparison of 2 Mbps, 30 Mbps, and 100 Mbps Fixed Broadband Deployment in the United States and the EU21 20. The charts below compare broadband at these speeds in the United States and the EU21 for 2015 and 2016. In both 2015 and 2016, the United States led at the highest speeds of broadband. Figure 13 Fixed High-Speed Broadband Deployment for All Households by Speed (June 2015) Federal Communications Commission DA 18-99 107 Figure 14 Fixed High-Speed Broadband Deployment for All Households by Speed (June 2016) IV. MOBILE HIGH-SPEED BROADBAND COMPARISON 21. We also compare mobile LTE broadband deployment in the United States and the EU21. The Figures below show mobile LTE broadband coverage for households overall and rural households in the United States and the EU21. In the United States, by June 2016, LTE mobile coverage reached nearly 100 percent of all households and 98 percent of rural households. In the EU21, by June 2016, mobile LTE coverage reached 97 percent of all households and 83 percent of rural households. Figure 15 Mobile LTE Broadband Deployment for All Households (June 2015) Federal Communications Commission DA 18-99 108 Figure 16 Mobile LTE Broadband Deployment for All Rural Households (June 2015) Figure 17 Mobile LTE Broadband Deployment for All Households (June 2016) Federal Communications Commission DA 18-99 109 Figure 18 Mobile LTE Broadband Deployment for All Rural Households (June 2016) Federal Communications Commission DA 18-99 110 V. HISTORICAL OVERVIEW OF FIXED HIGH-SPEED DEPLOYMENT, 2012-2016 22. Below, we present deployment of fixed high-speed broadband from 2012 to 2016 in the United States and EU21 at either 25 Mbps or 30 Mbps or above depending on data availability.28 We separately provide data for total, rural, and non-rural deployment. As of December 2012, the United States had a higher deployment than the EU21 and both continued to increase deployment in total, rural, and non-rural areas, with the EU21 countries having a somewhat higher growth rate. As shown in Figure 19, the total fixed high-speed deployment in the United States as of December 2012 was 80 percent, increasing to 90 percent by June 2016. By comparison, as of December 2012, the EU21 total deployment for fixed high-speed broadband was at 55 percent and increased to 76 percent by June 2016. 23. Below, Figure 20 shows that rural fixed high-speed broadband deployment in the United States as of December 2012 was 45 percent and increased to 63 percent by June 2016. As of December 2012, rural fixed high-speed broadband deployment in the EU21 was 15 percent, increasing to 41 percent by the end of June 2016. 28 For purposes of our fixed high-speed broadband deployment comparison for 2012 to 2014, we used the SBI data for 25 Mbps, which most closely matches the 30 Mbps threshold in the EC study. Federal Communications Commission DA 18-99 111 24. Figure 21 shows that non-rural fixed high-speed broadband deployment in the United States increased from 89 percent as of December 2012 to 97 percent as of June 2016. By comparison, non-rural fixed high-speed broadband deployment in the EU21 increased from 61 percent as of December 2012 to 83 percent as of June 2016. Federal Communications Commission DA 18-99 112 VI. FIXED HIGH-SPEED BROADBAND COVERAGE MAPS FOR THE UNITED STATES AND EUROPE 25. Below are maps of fixed high-speed fixed broadband coverage in the United States and Europe as of June 2016. Given that the EC Broadband Report already provides a map of their data, we reproduce that map below. Federal Communications Commission DA 18-99 113 Map 1 United States Fixed High-Speed Broadband Coverage Map (30 Mbps) June 2016 Federal Communications Commission DA 18-99 114 Map 2 Europe Fixed High-Speed Broadband Coverage (30 Mbps) June 201629 29 EC Broadband Report at 47. Federal Communications Commission DA 18-99 115 APPENDIX E Demographics Dataset 1. The BDIA directs the Commission to compare broadband development in communities comparable to U.S. communities in terms of population size, population density, topography, and demographic profile.1 In this Appendix, we present data on the population size, population density, and other indicators such as gross domestic product (GDP) and educational attainment for the United States and the comparison countries and, in the aggregate, almost 300 province/county communities. For the selected countries excluding the United States and Canada, we present the Organization for Economic Cooperation and Development’s (OECD’s) most recent published data ranging from 2012 to 2016.2 For the United States, we present 2016 data from the U.S. Census Bureau.3 For Canada, we present 2016 data from the Canadian Radio-television and Communications Commission.4 The topography information for the United States and the comparison countries is based on information from the Central Intelligence Agency’s The World Factbook.5 Table 1 Demographics Dataset Community Households with Broadband (%) Population Total Population Density (Persons per Square km) GDP Total (US$mm), PPP (Constant Real Prices 2010) GDP Per Capita, (US$) PPP (Constant Real Prices 2010) Education (% of Labor Force with Tertiary Education) Australia (ALA0) 86* 24,127,200 3 1,077,540 45,294 45 New South Wales (AU1) 85 7,725,880 10 350,643 46,030 48 Victoria (AU2) 86 6,068,040 27 243,807 41,062 50 Queensland (AU3) 86 4,844,470 3 205,893 43,084 39 South Australia (AU4) 82 1,708,180 2 65,329 38,459 40 Western Australia (AU5) 88 2,617,170 1 156,080 60,257 41 Tasmania (AU6) 81 519,128 8 17,034 32,975 34 1 47 U.S.C. § 1303(b)(2). 2 OECD Regions and Cities (To access the data on households with broadband (%), population size, population density, GDP total, GDP per capita, and educational attainment, click the left-hand column titled “Data by Theme,” and then click on “Regions and Cities,” then “Regional Statistics,” and then the following sub-categories: “Regional Social and Environmental Indicators” (select “Internet Broadband Access”); “Regional Demography” (select “Population by Age and Gender” and “Population Density and Regional Area”); “Regional Economy” (select “Regional Gross Domestic Product (GDP)” and “Regional GDP per Capita”); “Regional Innovation” (select “Educational Attainments of the Labour Force” and then “Customize”—“Selection”—“Indicator”—“Share of Labour Force with Tertiary Education (in % of labour force”). In Table 1a below, we identify the sources. The term PPP refers to Purchasing Power Parity. 3 U.S. Census Households With Broadband Subscription. In Table 1, the data for the percentage of households with broadband in all of the communities except Canada represent fixed and mobile broadband. 4 Canada Communications Monitoring Report. In Table 1, the data for the percentage of households with broadband in Canada represent fixed broadband subscription. 5 CIA World Factbook (accessing topography information of each country via the drop-down list that displays the text, “Please select a country to view,” and selecting a country-specific webpage). See also infra para. 1, Tbl. 2. Federal Communications Commission DA 18-99 116 Community Households with Broadband (%) Population Total Population Density (Persons per Square km) GDP Total (US$mm), PPP (Constant Real Prices 2010) GDP Per Capita, (US$) PPP (Constant Real Prices 2010) Education (% of Labor Force with Tertiary Education) Northern Territory (AU7) 89 244,880 0.2 15,002 61,407 44 Australian Capital Territory (AU8) 94 396,141 169 23,747 60,779 65 Austria (AT0) 85 8,690,080 105 369,329 42,780 32 Burgenland (AT0) 83 290,608 79 8,651 29,896 28 Lower Austria (AT12) 83 1,652,320 87 58,033 35,300 31 Vienna (AT13) 88 1,837,440 4,652 94,031 51,776 42 Carinthia (AT21) 84 559,846 60 20,222 36,210 31 Styria (AT22) 82 1,230,760 76 47,078 38,411 28 Upper Austria (AT31) 86 1,451,920 124 63,172 43,759 27 Salzburg (AT32) 86 545,074 77 27,103 50,043 31 Tyrol (AT33) 84 738,455 59 33,426 45,573 27 Vorarlberg (AT34) 88 383,657 151 17,511 45,969 26 Belgium (BE0) 82 11,311,100 373 459,555 40,762 42 Brussels Capital Region (BE1) 86 1,201,290 7,461 83,544 70,047 49 Flemish Region (BE2) 84 6,492,000 486 269,481 41,635 42 Wallonia BE3 79 3,617,830 215 106,227 29,434 39 Canada 83 36,286,400 4 1,515,440 42,273 27 Newfoundland and Labrador (CA10) 84 530,128 1 22,966 43,440 20 Prince Edward Island (CA11) 83 148,649 26 4,720 32,165 21 Nova Scotia (CA12) 79 949,501 18 30,691 32,533 25 New Brunswick (CA13) 86 756,780 11 25,218 33,432 20 Quebec (CA24) 80 8,326,090 6 290,676 35,193 24 Ontario (CA35) 84 13,983,000 15 582,369 42,210 30 Manitoba (CA46) 79 1,318,130 2 50,252 38,775 22 Saskatchewan (CA47) 76 1,150,630 2 60,593 53,515 22 Alberta (CA48) 87 4,252,880 7 249,064 59,589 23 British Columbia (CA59) 88 4,751,610 5 190,732 40,642 28 Yukon (CA60) -- 37,492 0.8 2,068 55,296 -- Northwest Territories (CA61) -- 44,469 0.04 3,684 83,259 -- Nunavut (CA62) -- 37,082 0.02 1,667 51,107 -- Chile 53* 18,191,900 25 375,261 20,840 33 Tarapacá (CL01) 56 344,760 8 8,768 26,036 30 Antofagasta (CL02) 73 631,875 5 35,891 57,644 38 Atacama (CL03) 57 316,692 4 7,350 23,522 29 Federal Communications Commission DA 18-99 117 Community Households with Broadband (%) Population Total Population Density (Persons per Square km) GDP Total (US$mm), PPP (Constant Real Prices 2010) GDP Per Capita, (US$) PPP (Constant Real Prices 2010) Education (% of Labor Force with Tertiary Education) Coquimbo (CL04) 48 782,801 19 10,954 14,206 26 Valparaíso (CL05) 56 1,842,880 112 32,705 17,913 38 O'Higgins (CL06) 47 926,828 57 17,243 18,768 25 Maule (CL07) 38 1,050,320 35 12,170 11,669 20 Bio-Bio (CL08) 49 2,127,900 57 28,384 13,425 29 Araucanía (CL09) 39 995,974 31 9,048 9,141 24 Los Lagos (CL10) 46 847,495 17 11,886 14,131 28 Aysén (CL11) 53 109,317 1 2,179 20,113 27 Magallanes y Antártica (CL12) 67 165,547 1 3,256 19,775 38 Santiago Metropolitan (CL13) 62 7,399,040 480 187,924 25,693 40 Los Ríos (CL14) 42 407,300 22 4,993 12,345 24 Arica Y Parinacota (CL15) 57 243,149 14 2,510 10,494 27 Czech Republic 80 10,553,800 137 313,253 29,703 22 Prague (CZ01) 91 1,267,450 2613 76,527 60,579 41 Central Bohemian Region (CZ02) 84 1,326,880 123 36,304 27,480 22 Southwest (CZ03) 82 1,214,450 71 31,876 26,269 18 Northwest (CZ04) 75 1,120,650 132 24,917 22,209 15 Northeast (CZ05) 79 1,507,210 123 36,865 24,462 18 Southeast (CZ06) 78 1,684,500 123 46,690 27,732 24 Central Moravia (CZ07) 74 1,219,390 134 30,169 24,725 19 Moravia-Silesia (CZ08) 79 1,213,310 228 29,902 24,601 18 Denmark (DK0) 92 5,707,250 133 253,126 44,537 33 Capital (DK) (DK01) 93 1,789,170 699 102,037 57,368 44 Zealand (DK02) 91 827,499 115 25,268 30,666 27 Southern Denmark (DK03) 89 1,211,770 99 48,869 40,429 28 Central Jutland (DK04) 92 1,293,310 99 51,411 39,915 30 North Jutland (DK05) 92 585,499 74 21,559 36,911 26 Estonia 85 1,315,940 30 34,182 25,986 39 Finland (FI0) 91 5,487,310 18 208,113 37,980 41 Western Finland (FI19) 88 1,379,120 24 46,887 34,020 38 Helsinki-Uusimaa (FI1B) 95 1,620,260 178 80,418 49,893 48 Southern Finland (FI1C) 93 1,160,490 37 38,985 33,576 37 Eastern and Northern Finland (FI1D) 89 1,298,460 6 40,423 31,108 36 Åland (FI20) -- 28,983 19 1,337 46,184 29 France 79 66,760,000 105 2,455,870 36,862 36 Federal Communications Commission DA 18-99 118 Community Households with Broadband (%) Population Total Population Density (Persons per Square km) GDP Total (US$mm), PPP (Constant Real Prices 2010) GDP Per Capita, (US$) PPP (Constant Real Prices 2010) Education (% of Labor Force with Tertiary Education) Île de France (FR10) 85 12,142,800 1011 742,928 61,319 46 Champagne-Ardenne (FR21) 75 1,339,570 52 41,557 31,016 27 Picardy (FR22) 73 1,935,560 100 53,331 27,567 25 Upper Normandy (FR23) 81 1,864,110 151 59,244 31,810 28 Centre-Val de Loire (FR24) 79 2,587,000 66 78,532 30,376 30 Lower Normandy (FR25) 80 1,479,130 84 43,429 29,353 28 Burgundy (FR26) 77 1,640,690 52 50,911 31,013 31 Nord-Pas-de-Calais (FR30) 76 4,094,750 330 122,863 30,031 35 Lorraine (FR41) 82 2,333,590 99 65,661 28,104 32 Alsace (FR42) 80 1,885,150 228 64,369 34,192 33 Franche-Comté (FR43) 74 1,179,470 73 33,554 28,451 29 Pays de la Loire (FR51) 77 3,743,980 117 123,061 32,977 33 Brittany (FR52) 74 3,310,340 122 103,182 31,241 36 Poitou-Charentes (FR53) 79 1,808,710 70 54,808 30,339 30 Aquitaine (FR61) 83 3,399,090 82 108,561 32,064 35 Midi-Pyrénées (FR62) 82 3,027,280 67 99,418 32,965 44 Limousin (F 63) 71 735,295 43 20,481 27,818 31 Rhône-Alpes (FR71) 81 6,574,710 150 240,516 36,734 40 Auvergne (FR72) 78 1,365,940 53 41,081 30,097 30 Languedoc-Roussillon (FR 81) 76 2,802,890 102 76,737 27,499 34 Provence-Alpes-Côte d'Azur (FR82) 81 5,024,190 160 172,020 34,301 37 Corsica (FR83) 62 330,354 38 9,715 29,542 26 Germany (DE0) 90 82,175,700 230 3,473,470 42,522 27 Baden-Württemberg (DE1) 89 10,879,600 304 528,828 48,974 29 Bavaria (DE2) 89 12,843,500 182 630,422 49,377 29 Berlin (DE3) 91 3,520,030 3946 142,526 40,781 37 Brandenburg (DE4) 84 2,484,830 84 74,951 30,328 28 Bremen (DE5) 91 671,489 1,603 36,262 54,392 27 Hamburg (DE6) 94 1,787,410 2367 125,433 70,662 34 Hesse (DE7) 91 6,176,170 293 302,410 49,292 29 Mecklenburg-Vorpommern (DE8) 89 1,612,360 69 45,766 28,501 24 Lower Saxony (DE9) 92 7,926,600 166 296,770 37,677 23 North Rhine Westphalia (DEA) 90 17,865,500 524 741,081 41,747 25 Rheinland-Palatinate (DEB) 91 4,052,800 204 151,468 37,565 25 Saarland (DEC) 88 995,597 388 40,210 40,521 21 Federal Communications Commission DA 18-99 119 Community Households with Broadband (%) Population Total Population Density (Persons per Square km) GDP Total (US$mm), PPP (Constant Real Prices 2010) GDP Per Capita, (US$) PPP (Constant Real Prices 2010) Education (% of Labor Force with Tertiary Education) Saxony (DED) 88 4,084,850 222 129,322 31,774 29 Saxony-Anhalt (DEE) 88 2,245,470 110 65,533 28,803 24 Schleswig-Holstein (DEF) 90 2,858,710 181 98,273 34,545 23 Thüringen (DEG) 88 2,170,710 134 65,214 30,140 28 Greece (GR0) 68 10,783,700 82 255,977 23,656 31 Attica (EL30) -- 3,781,270 -- 122,918 32,329 -- North Aegean (EL41) -- 196,654 -- 3,575 18,133 -- South Aegean (EL42) -- 334,791 -- 8,846 26,421 -- Crete (EL43) -- 631,812 -- 12,805 20,272 -- Eastern Macedonia, Thrace (EL51) -- 604,504 -- 10,046 16,590 -- Central Macedonia (EL52) -- 1,883,340 -- 34,436 18,233 -- Western Macedonia (EL53) -- 273,843 -- 6,033 21,928 -- Epirus (EL54) -- 336,834 -- 5,663 16,755 -- Thessaly (EL61) -- 729,442 -- 12,968 17,727 -- Ionian Islands EL62 -- 206,141 -- 4,551 22,030 -- Western Greece (EL63) -- 668,258 -- 11,667 17,394 -- Central Greece (EL64) -- 555,830 -- 11,210 20,132 -- Peloponnese (EL65) -- 581,026 -- 11,259 19,338 -- Iceland 93 332,529 3 13,095 40,446 33 Capital Region (IS01) -- 213,619 217 -- -- -- Other Regions (IS02) -- 118,910 1 -- -- -- Ireland 86 4,724,720 69 269,794 58,284 42 Border - Midlands and Western (IE01) 82 1,250,090 39 -- -- 36 Southern and Eastern (IE02) 87 3,474,630 96 -- -- 44 Italy (IT0) 77 60,665,600 206 2,015,050 33,180 19 Piedmont (ITC1) 78 4,404,250 177 155,975 35,333 18 Aosta Valley (ITC2) 75 127,329 39 5,357 41,910 16 Liguria (ITC3) 76 1,571,050 295 58,369 37,009 21 Lombardy (ITC4) 82 10,008,300 439 437,436 43,720 20 Abruzzo (ITF1) 78 1,326,510 125 39,913 30,031 18 Molise (ITF2) 73 312,027 71 7,399 23,663 19 Campania ITF3 70 5,850,850 437 123,129 21,025 18 Apulia (ITF4) 70 4,077,170 212 88,338 21,632 17 Basilicata (ITF5) 70 573,694 59 14,021 24,377 18 Calabria (ITF6) 68 1,970,520 134 40,162 20,350 19 Federal Communications Commission DA 18-99 120 Community Households with Broadband (%) Population Total Population Density (Persons per Square km) GDP Total (US$mm), PPP (Constant Real Prices 2010) GDP Per Capita, (US$) PPP (Constant Real Prices 2010) Education (% of Labor Force with Tertiary Education) Sicily (ITG1) 69 5,074,260 200 107,012 21,052 17 Sardinia (ITG2) 79 1,658,140 69 39,777 23,952 16 Province of Bolzano-Bozen (ITH1) 76 520,891 71 26,315 50,635 16 Province of Trento (ITH2) 82 538,223 88 22,788 42,371 20 Veneto (ITH3) 80 4,915,120 280 185,695 37,732 17 Friuli-Venezia Giulia (ITH4) 80 1,221,220 162 43,681 35,682 19 Emilia-Romagna (ITH5) 81 4,448,150 207 183,112 41,155 20 Tuscany (ITI1) 79 3,744,400 165 135,115 36,045 19 Umbria (ITI2) 79 891,181 108 26,254 29,400 21 Marche (ITI3) 78 1,543,750 162 49,711 32,128 19 Lazio (ITI4) 80 5,888,470 348 223,668 37,971 26 Japan (JP0) 62* 126,969,000 340 4,685,010 36,804 43 Hokkaido (JPA) 51 -- -- 168,270 30,983 35 Tohoku (JPB) 57 -- -- 296,832 32,637 29 Northern-Kanto, Koshin (JPC) 64 -- -- 354,158 35,882 36 Southern-Kanto (JPD) 76 -- -- 1,509,060 42,161 54 Hokoriku (JPE) 66 -- -- 192,156 35,850 35 Toukai (JPF) 67 -- -- 609,130 40,474 40 Kansai Region (JPG) 71 -- -- 731,748 35,177 46 Chugoku (JPH) 58 -- -- 258,027 34,542 40 Shikoku (JPI) 54 -- -- 125,482 32,134 38 Kyushu, Okinawa (JPJ) 53 -- -- 440,154 30,312 36 Latvia (LV00) 75 1,968,960 32 43,949 22,224 32 Luxembourg (LU00) 97 576,249 223 49,670 87,202 48 Mexico 38* 122,273,000 62 1,873,410 15,482 21 Aguascalientes (ME01) 40 1,304,740 232 23,820 18,498 23 Baja California Norte (ME02) 56 3,534,690 49 56,553 16,231 20 Baja California Sur (ME03) 58 786,864 11 14,587 19,094 22 Campeche (ME04) 40 921,517 16 48,646 53,582 22 Coahuila (ME05) 40 2,995,370 20 66,534 22,473 25 Colima (ME06) 47 735,724 131 11,364 15,708 22 Chiapas (ME07) 13 5,317,960 72 32,137 6,118 14 Chihuahua (ME08) 41 3,746,280 15 56,682 15,278 19 Distrito Federal (ME09) 63 8,833,420 5952 313,525 35,408 33 Durango (ME10) 28 1,782,210 14 23,375 13,246 19 Federal Communications Commission DA 18-99 121 Community Households with Broadband (%) Population Total Population Density (Persons per Square km) GDP Total (US$mm), PPP (Constant Real Prices 2010) GDP Per Capita, (US$) PPP (Constant Real Prices 2010) Education (% of Labor Force with Tertiary Education) Guanajuato (ME11) 36 5,864,020 192 83,430 14,341 15 Guerrero (ME12) 21 3,588,260 56 28,458 7,976 16 Hidalgo (ME13) 32 2,913,150 140 32,945 11,446 16 Jalisco (ME14) 47 8,022,180 102 127,866 16,122 21 Mexico (ME15) 40 17,118,500 767 177,443 10,518 20 Michoacán (ME16) 25 4,627,900 79 45,038 9,798 15 Morelos (ME17) 40 1,943,040 397 22,028 11,471 20 Nayarit (ME18) 34 1,246,200 45 13,096 10,701 21 Nuevo Leon (ME19) 59 5,157,780 80 141,131 27,750 28 Oaxaca (ME20) 18 4,037,360 43 30,022 7,483 14 Puebla (ME21) 25 6,254,600 183 60,611 9,786 18 Queretaro (ME22) 43 2,034,030 174 44,005 21,953 23 Quintana Roo (ME23) 49 1,619,760 38 31,010 19,691 20 San Luis Potosi (ME24) 40 2,778,000 45 37,871 13,754 20 Sinaloa (ME25) 38 3,009,950 53 41,181 13,798 26 Sonora (ME26) 57 2,972,580 17 55,465 18,912 23 Tabasco (ME27) 21 2,407,860 97 43,518 18,255 21 Tamaulipas (ME28) 44 3,583,300 45 57,317 16,176 23 Tlaxcala (ME29) 23 1,295,780 324 10,742 8,403 19 Veracruz (ME30) 26 8,106,140 113 93,422 11,610 18 Yucatán (ME31) 47 2,145,880 54 29,458 13,903 19 Zacatecas (ME32) 27 1,588,420 21 20,133 12,774 17 Netherlands (NL0) 95 16,979,100 504 769,032 45,398 34 Groningen (NL11) 98 583,721 251 29,741 50,942 -- Friesland (NL12) 95 646,040 194 20,664 31,979 -- Drenthe (NL13) 95 488,629 185 15,696 32,124 -- Overijssel (NL21) 95 1,144,280 344 42,301 37,026 -- Gelderland (NL22) 96 2,035,350 410 77,163 37,994 -- Flevoland (NL23) 100 404,068 286 14,014 34,779 -- Utrecht (NL31) 96 1,273,610 922 67,415 53,142 -- NL32 North Holland (NL32) 94 2,784,850 1,045 161,170 58,113 -- South Holland (NL33) 95 3,622,300 1,291 164,114 45,446 -- Zeeland (NL34) 92 381,252 214 13,452 35,308 -- North Brabant (NL41) 96 2,498,750 509 116,121 46,565 -- Limburg (NL) (NL42) 93 1,116,260 519 42,801 38,315 -- New Zealand (NZ0) 75 4,692,700 18 149,885 33,236 38 Federal Communications Commission DA 18-99 122 Community Households with Broadband (%) Population Total Population Density (Persons per Square km) GDP Total (US$mm), PPP (Constant Real Prices 2010) GDP Per Capita, (US$) PPP (Constant Real Prices 2010) Education (% of Labor Force with Tertiary Education) Northland (NZ11) Region -- 171,400 -- 3,647 21,972 -- Auckland Region (NZ12) -- 1,614,400 -- 54,871 35,936 -- Waikato Region (NZ13) -- 449,200 -- 12,211 28,345 -- Bay of Plenty Region (NZ14) -- 293,500 -- 7,639 27,059 -- Gisborne Region (NZ15) -- 47,800 -- 5,145 24,976 -- Hawke's Bay Region (NZ16) -- 161,500 -- -- Taranaki Region (NZ17) -- 116,700 -- 5,441 47,399 -- Manawatu-Wanganui Region (NZ18) -- 236,900 -- 5,715 24,583 -- Wellington Region (NZ19) -- 504,800 -- 20,270 41,249 -- Tasman-Nelson-Marlborough (NZ21) -- 146,300 -- 5,171 29,381 -- West Coast Region (NZ22) -- 32,500 -- -- Canterbury Region (NZ23) -- 599,900 -- 20,435 35,582 -- Otago Region (NZ24) -- 219,200 -- 6,322 29,877 -- Southland Region (NZ25) -- 98,000 -- 3,018 31,279 -- Norway (NO0) 96 5,210,720 17 307,691 59,301 42 Oslo and Akershus (NO01) 96 1,251,690 250 77,208 -- 53 Hedmark and Oppland (NO02) 93 384,221 8 13,128 -- 33 South-Eastern Norway(NO03) 97 984,764 29 34,413 -- 37 Agder and Rogaland (NO04) 98 768,179 33 37,151 -- 37 Western Norway (NO05) 99 890,719 19 42,614 -- 41 Trøndelag (NO06) 99 449,457 12 18,655 -- 40 Northern Norway (NO07) 91 481,694 5 19,419 -- 38 Portugal (PT0) 73 10,341,300 112 276,281 26,673 23 North (PT) PT11 70 3,603,780 169 81,444 22,543 20 Algarve (PT15) 71 441,929 88 12,089 27,369 19 Central Portugal (PT16) 68 2,256,360 80 52,262 23,123 20 Lisbon (PT17) 82 2,812,680 933 100,553 35,772 33 Alentejo (PT18) 62 724,391 23 17,643 24,205 18 Azores (PT) (PT20) 79 245,766 106 5,824 23,671 15 Madeira (PT) (PT30) 78 256,424 320 6,400 24,849 18 South Korea (KR0) 99 50,801,400 511 1,696,970 33,654 43 Capital Region (KR01) 99 25,268,200 2159 831,313 33,213 47 Gyeongnam Region (KR02) 98 7,832,300 635 276,130 35,296 40 Federal Communications Commission DA 18-99 123 Community Households with Broadband (%) Population Total Population Density (Persons per Square km) GDP Total (US$mm), PPP (Constant Real Prices 2010) GDP Per Capita, (US$) PPP (Constant Real Prices 2010) Education (% of Labor Force with Tertiary Education) Gyeonbuk Region (KR03) 98 5,092,300 256 159,039 31,182 39 Jeolla Region (KR04) 98 5,072,170 247 157,638 31,084 38 Chungcheong Region (KR05) 97 5,432,010 328 214,296 40,302 39 Gangwon Region (KR06) 97 1,511,020 91 42,462 28,288 36 Jeju (KR07) 98 593,455 321 16,092 27,694 40 Spain (ES0) 81 46,445,800 92 1,472,320 31,698 38 Galicia (ES11) 78 2,720,540 93 76,407 28,012 37 Asturias (ES12) 79 1,041,030 99 29,126 27,860 45 Cantabria (ES13) 77 582,548 111 16,735 28,658 41 Basque Country (ES21) 82 2,164,140 301 91,097 42,084 51 Navarra (ES22) 82 637,540 62 25,302 39,722 46 La Rioja (ES23) 79 312,815 62 10,800 34,483 39 Aragón (ES24) 80 1,318,740 28 46,113 34,866 38 Comunidad de Madrid (ES30) 88 6,424,840 806 277,815 43,374 49 Castile and León (ES41) 77 2,454,860 26 73,316 29,725 37 Castile-la Mancha (ES42) 78 2,049,150 26 50,588 24,605 30 Extremadura (ES43) 78 1,085,120 27 23,643 21,723 30 Catalonia (ES51) 82 7,408,850 232 279,490 37,754 39 Valencia (ES52) 80 4,933,050 213 138,752 28,108 34 Balearic Island (ES53) 81 1,135,630 228 37,271 32,974 29 Andalusia (ES61) 80 8,405,300 97 196,894 23,433 30 Murcia (ES62) 81 1,466,510 130 37,680 25,718 29 Ceuta (ES63) 83 84,663 4456 2,179 25,733 -- Melilla (ES64) 85 84,777 6521 1,977 23,344 -- Canary Islands (ES70) 81 2,135,720 287 55,956 26,259 29 Sweden (SE0) 89 9,851,020 24 432,516 44,138 37 Stockholm (SE11) 90 2,231,440 342 137,714 62,181 45 East Middle Sweden (SE12) 89 1,638,830 43 61,611 37,793 35 Småland with Is (SE21) 83 834,276 25 30,785 37,079 29 South Sweden (SE22) 93 1,459,880 105 54,100 37,273 37 West Sweden (SE23) 89 1,963,470 67 84,447 43,238 37 North Middle Sweden (SE31) 87 838,747 13 29,409 35,171 27 Central Norrland (SE32) 82 371,273 5 14,014 37,821 29 Upper Norrland (SE33) 85 513,111 3 20,339 39,667 35 Switzerland (CH0) 86 8,327,130 208 442,360 54,021 38 Lake Geneva Region (CH01) 85 1,593,840 192 80,446 51,627 39 Federal Communications Commission DA 18-99 124 Community Households with Broadband (%) Population Total Population Density (Persons per Square km) GDP Total (US$mm), PPP (Constant Real Prices 2010) GDP Per Capita, (US$) PPP (Constant Real Prices 2010) Education (% of Labor Force with Tertiary Education) Espace Mitteland (CH02) 81 1,842,250 188 89,987 49,515 35 Northwestern Switzerland (CH03) 87 1,128,720 579 61,825 55,660 40 Zurich (CH04) 93 1,466,420 883 95,126 66,246 44 Eastern Switzerland (CH05) 87 1,153,490 102 54,315 47,658 32 Central Switzerland (CH06) 87 790,458 185 40,924 52,584 35 Ticino (CH07) 77 351,946 128 19,738 56,646 35 United Kingdom (UK0) 92 65,382,600 270 2,476,520 38,025 40 North East (UKC) 91 2,632,290 307 73,807 28,074 31 North West (UKD) 92 7,187,230 510 233,071 32,498 36 Yorkshire and The Humber (UKE) 91 5,406,550 351 162,992 30,200 34 East Midlands (UKF) 89 4,693,620 300 145,435 31,120 34 West Midlands (UKG) 89 5,772,080 444 177,945 30,935 33 Eastern (UKH) 92 6,105,480 319 216,400 35,620 37 London (UKI) 93 8,759,410 5572 562,240 64,756 55 South East (UKJ) 95 8,992,570 472 370,207 41,352 43 South West (UKK) 93 5,492,330 230 187,214 34,240 39 Wales (UKL) 89 3,105,860 150 82,887 26,669 37 Scotland (UKM) 92 5,376,610 69 189,075 35,232 46 Northern Ireland (UKN) 88 1,858,540 137 51,125 27,593 36 United States (US0) 81 323,128,000 35 16,489,800 51,303 40 Alabama (US01) 75 4,863,300 37 183,725 37,811 35 Alaska (US02) 86 741,894 1 48,538 65,732 34 Arizona (US04) 83 6,931,070 24 267,691 39,205 37 Arkansas (US05) 71 2,988,250 22 109,419 36,740 29 California (US06) 85 39,250,000 97 2,283,350 58,331 40 Colorado (US08) 87 5,540,550 21 288,713 52,911 46 Connecticut (US09) 84 3,576,450 285 232,748 64,816 46 Delaware (US10) 83 952,065 188 63,240 66,855 41 District of Columbia (US11) 80 681,170 4284 112,400 167,205 64 Florida (US12) 81 20,612,400 148 817,224 40,314 39 Georgia (US13) 81 10,310,400 69 458,211 44,857 38 Hawaii (US15) 83 1,428,560 86 73,963 51,644 44 Idaho (US16) 79 1,683,140 8 60,319 36,448 36 Illinois (US17) 82 12,801,500 89 714,892 55,590 43 Indiana (US18) 79 6,633,050 71 309,238 46,715 35 Iowa (US19) 80 3,134,690 22 160,144 51,264 40 Federal Communications Commission DA 18-99 125 Community Households with Broadband (%) Population Total Population Density (Persons per Square km) GDP Total (US$mm), PPP (Constant Real Prices 2010) GDP Per Capita, (US$) PPP (Constant Real Prices 2010) Education (% of Labor Force with Tertiary Education) Kansas (US20) 80 2,907,290 14 137,701 47,293 41 Kentucky (US21) 77 4,436,970 43 177,852 40,192 34 Louisiana (US22) 74 4,681,670 42 220,210 47,147 31 Maine (US23) 81 1,331,480 17 52,725 39,663 40 Maryland (US24) 86 6,016,450 238 336,203 55,974 46 Massachusetts (US25) 86 6,811,780 335 446,248 65,679 50 Michigan (US26) 81 9,928,300 67 430,964 43,433 39 Minnesota (US27) 84 5,519,950 27 302,141 55,039 45 Mississippi (US28) 71 2,988,730 25 97,375 32,542 32 Missouri (US29) 79 6,093,000 34 270,993 44,544 38 Montana (US30) 79 1,042,520 3 41,627 40,300 37 Nebraska (US31) 82 1,907,120 10 104,243 54,975 42 Nevada (US32) 81 2,940,060 10 128,575 44,477 30 New Hampshire (US33) 86 1,334,800 57 67,973 51,084 45 New Jersey (US34) 84 8,944,470 466 522,436 58,321 46 New Mexico (US35) 74 2,081,020 7 85,891 41,193 35 New York (US36) 82 19,745,300 161 1,319,140 66,638 47 North Carolina (US37) 79 10,146,800 80 455,872 45,393 39 North Dakota (US38) 81 757,952 4 51,403 67,910 44 Ohio (US39) 81 11,614,400 110 562,180 48,408 37 Oklahoma (US40) 77 3,923,560 22 171,141 43,755 33 Oregon (US41) 85 4,093,470 16 200,264 49,706 39 Pennsylvania (US42) 81 12,784,200 110 653,128 51,016 41 Rhode Island (US44) 83 1,056,430 390 51,579 48,830 43 South Carolina (US45) 77 4,961,120 64 184,966 37,778 36 South Dakota (US46) 80 865,454 4 43,474 50,642 39 Tennessee (US47) 77 6,651,190 62 290,654 44,036 34 Texas (US48) 81 27,862,600 41 1,500,010 54,607 35 Utah (US49) 85 3,051,220 14 135,733 45,306 38 Vermont (US50) 81 624,594 26 27,641 44,152 43 Virginia (US51) 83 8,411,810 82 442,697 52,809 45 Washington (US53) 87 7,288,000 42 409,872 57,162 44 West Virginia (US54) 74 1,831,100 29 68,391 37,086 31 Wisconsin (US55) 81 5,778,710 41 277,972 48,164 40 Wyoming (US 56) 83 585,501 2 36,683 62,588 38 Figures marked with an asterisk (*) were calculated by FCC staff using simple averages of OECD data. Federal Communications Commission DA 18-99 126 Table 1a Sources for Demographics Dataset Country Households with Broadband (%) Population Total Population Density (Persons per Square km) GDP Total (US$mm), PPP1 (Constant Real Prices 2010) GDP Per Capita, (US$) PPP (Constant Real Prices 2010) Education (% of Labor Force with Tertiary Education) Australia 2015, OECD 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2014, OECD Austria 2016, OECD 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2014, OECD Belgium 2016, OECD 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2014, OECD Canada 2016, CRTC 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2013, OECD Chile 2013, OECD 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2014, OECD Czech Republic 2016, OECD 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2014, OECD Denmark 2016, OECD 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2014, OECD Estonia 2016, OECD 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2014, OECD Finland 2016, OECD 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2014, OECD France 2016, OECD 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2014, OECD Germany 2016, OECD 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2014, OECD Greece 2016, OECD 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2014, OECD Iceland 2014, OECD 2016, OECD 2016, OECD 2013, OECD 2013, OECD 2014, OECD Ireland 2016, OECD 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2014, OECD Italy 2016, OECD 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2014, OECD Japan 2015, OECD 2016, OECD 2016, OECD 2013, OECD 2013, OECD 2010, OECD Latvia 2016, OECD 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2014, OECD Luxembourg 2016, OECD 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2014, OECD Federal Communications Commission DA 18-99 127 Country Households with Broadband (%) Population Total Population Density (Persons per Square km) GDP Total (US$mm), PPP1 (Constant Real Prices 2010) GDP Per Capita, (US$) PPP (Constant Real Prices 2010) Education (% of Labor Force with Tertiary Education) Mexico 2015, OECD 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2010, OECD Netherlands 2016, OECD 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2014, OECD New Zealand 2012, OECD 2016, OECD 2016, OECD 2014, OECD 2014, OECD 2012, OECD Norway 2016, OECD 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2014, OECD Portugal 2016, OECD 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2014, OECD Spain 2016, OECD 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2014, OECD South Korea 2014, OECD 2016, OECD 2016, OECD 2014, OECD 2014, OECD 2014, OECD Sweden 2016, OECD 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2014, OECD Switzerland 2014, OECD 2016, OECD 2016, OECD 2014, OECD 2014, OECD 2014, OECD United Kingdom 2016, OECD 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2014, OECD United States 2016, Census Bureau 2016, OECD 2016, OECD 2015, OECD 2015, OECD 2013, OECD Federal Communications Commission DA 18-99 128 Table 2 Country Topography Country Topography Australia Located between the Indian Ocean and the Pacific Ocean, the country is the world’s smallest continent but the sixth largest country. It is comprised of mostly low plateau with deserts and a fertile plain in the southeast. Austria A landlocked Central European country about the size of South Carolina and slightly more than two-thirds the size of Pennsylvania. The terrain is a mostly mountain to the west and south while mostly flats or gently sloping land along the eastern and northern margins. Belgium Bordering the North Sea, the country, about the size of Maryland, has flat coastal plains in the northwest, central rolling hills, and the rugged mountains of the Ardennes Forest in the southeast. Canada Slightly larger than the United states, it is the world’s largest country that borders only one country. The third largest country in the world, the terrain is mostly plains with mountains in the west, and lowlands in the southeast. Chile Slightly smaller than twice the size of Montana, the South American country’s terrain incudes low coastal mountains, a fertile central valley, and the rugged Andes in the east. Czech Republic About two-thirds the size of Pennsylvania and slightly smaller than South Carolina, the terrain consists of rolling plains, hills, and plateaus surrounded by low mountains to the west and very hilly areas to the east. Denmark Bordering the Baltic and North Sea, the area is slightly less than twice the size of Massachusetts, with the terrain composed of low and flat to gently rolling plains. The area also includes several major islands. Estonia About twice the size of New Jersey, the Eastern European country’s terrain includes marshy lowlands, mostly flat in the north but hilly in the south. Finland Slightly smaller than Montana, the Northern European country’s terrain is mostly low, with flat to rolling plains interspersed with lakes and low hills. France Slightly less than the size of Texas, the country has five overseas regions with varying terrain. In the French metropole in Western Europe, the terrain is mostly comprised of flat plains or gently rolling hills in the north and west. The remainder is mountainous. Germany Slightly smaller than Montana, the terrain consists of lowlands in the north, uplands in the center, and the Bavarian Alps in the south. Greece Slightly smaller than Alabama, the Southern European country borders the Aegean, Ionian, and the Mediterranean Seas. It has a mountainous terrain with ranges extending into the sea as peninsulas or chains of islands. Iceland Occupies an area about the same size as Kentucky. The terrain is mostly comprised of plateaus interspersed with mountain peaks and ice fields. The coast is deeply indented by bays and fiords. Ireland Occupies an area larger than West Virginia, The terrain is mostly flat to rolling interior plains which are surrounded by rugged hills and low mountains. The west coast has sea cliffs. Federal Communications Commission DA 18-99 129 Country Topography Italy Slightly larger than Arizona, the Southern European country borders the Mediterranean Sea. The terrain is mostly rugged and mountainous. Some plains and coastal lowlands also make up its terrain. Japan The island chain nation occupies an area slightly smaller than California. The terrain is mostly rugged and mountainous. Latvia The Eastern European country is slightly smaller than West Virginia. Low plains mark the terrain as most of the country is composed of fertile low-lying plains with some hills in the east. Luxembourg Occupies an area slightly smaller than Rhode Island. The terrain is mostly comprised of gently rolling uplands with broad, shallow valleys. In the north, there are uplands to slightly mountainous terrain and a steep slope down to Moselle flood plain in the southeast. Mexico Slightly less than three times the size of Texas, the terrain is comprised of high rugged mountains, low coastal plains, high plateaus, and desert. Netherlands Bordering the North Sea, it is less than twice the size of New Jersey. The terrain includes mostly coastal lowland and reclaimed land and some hills to the southeast. New Zealand An island archipelago in Oceania, it is almost twice the size of North Carolina and about the size of Colorado. The terrain is mostly mountainous with large coastal plains. Norway The Northern European country is slightly larger than New Mexico. It has a glaciated terrain, with mostly high plateaus and rugged mountains broken by fertile valleys. Small-scattered plains are also part of the terrain, as well as a coastline deeply indented by fjords, with arctic tundra in the north. Portugal Occupies an area slightly smaller than Virginia. The west-flowing Tagus River divides the country’s terrain. The north is mountainous toward the interior, while the south is characterized by rolling plains. South Korea Occupies an area slightly smaller than Pennsylvania, while slightly larger than Indiana. The terrain is mostly hills and mountains with wide coastal plains in the west and south. Spain Slightly more than twice the size of Oregon, the terrain is large and flat with dissected plateaus surrounded by rugged hills. Mountains are in the north. Sweden Occupies an area slightly larger than California. The terrain is mostly flat or gently rolling lowlands, and mountains to the west. Switzerland The landlocked Central European country is slightly less than twice the size of New Jersey. The terrain is comprised of mostly mountains with a central plateau of rolling hills, plains, and large lakes. United Kingdom The Atlantic archipelago occupies an area slightly smaller than Oregon. The terrain is mostly rugged hills and low mountains, with level to rolling plains in the east and the southeast. United States Slightly larger than China and more than twice the size of the European Union, the terrain is comprised of a vast central plain, mountains in the west, hills and low mountains in the east, rugged mountains and broad river valleys in Alaska, and rugged, volcanic topography in Hawaii. Source: CIA World Factbook Federal Communications Commission DA 18-99 130 APPENDIX F Market and Regulatory Developments 1. In this Appendix, as required by the BDIA, we identify the relevant similarities and differences between the United States and the 28 comparison countries with respect to multiple criteria.1 First, we discuss the regulatory models for fixed broadband deployment.2 Second, we provide a list of regulators and, where relevant, the ministries responsible for regulating broadband. Third, we provide information concerning the major fixed and mobile broadband competitors and the types of technologies used to provide broadband. Finally, we present data on the types of activities that consumers in the United States and the comparison countries engage in while using the Internet. I. BROADBAND REGULATORY MODELS A. Summary of Fixed Broadband Regulatory Models 2. Based on our analysis, there are two basic fixed broadband regulatory models – a facilities-based competition model, and an open access regulatory model – with some blending of the two models in most countries. There is also significant variation in regulatory approaches for implementing the same model. The facilities-based competition model relies on competition between the incumbent telecommunications operator and cable operators. The open access regulatory model features the use of mandated wholesale access to the incumbent’s fixed network to create service-based competition by encouraging market entry at the retail level. Under either model, investment in fixed broadband networks may be primarily market- or state-aid-driven, depending on the extent to which governments subsidize network deployment. 3. Below, in Table 1, we identify the primary regulatory models for the United States and the comparison countries for the past 10-15 years. 1 47 U.S.C. § 1303(b)(3) (“The Commission shall identify relevant similarities and differences in each community, including their market structures, the number of competitors, the number of facilities-based providers, the types of technologies deployed by such providers, the applications and services those technologies enable, the regulatory model under which broadband service capability is provided, the types of applications and services used, business and residential use of such services, and other media available to consumers.”). 2 47 U.S.C. § 1303(b)(3) (“The Commission shall identify . . . the regulatory model under which broadband service capability is provided.”). Federal Communications Commission DA 18-99 131 Table 1 Regulatory Models Country3 Facilities-Based Competition Open Access Regulations Australia X Austria X Belgium X Canada X Czech Republic X Denmark X Estonia X Finland X France X Germany X Greece X Iceland X Ireland X Italy X Japan X Latvia X Luxembourg X Netherlands X New Zealand X Norway X Portugal X South Korea X Spain X Sweden X Switzerland X United Kingdom X United States X 4. Regulatory Model Blending. There is some blending of the two models within each group of countries. In the United States, the Commission initially imposed an unbundling obligation on traditional wireline services and facilities (e.g., digital subscriber line (DSL)),4 but eliminated these requirements for wireline broadband Internet access service providers in 20055 and also declined to impose unbundling requirements on certain incumbent local exchange carrier (LEC) next-generation 3 This chart includes the 28 comparison countries discussed in the Report except for Chile and Mexico because the regulatory model in these countries is unclear. 4 Appropriate Framework for Broadband Access to the Internet over Wireline Facilities et al., CC Docket Nos. 02- 33 et al., Report and Order and Notice of Proposed Rulemaking, 20 FCC Rcd 14853, 14872, para. 31 (2005) (Wireline Broadband Order), aff'd Time Warner Telecom, Inc. v. FCC, 507 F.3d 205 (3d Cir. 2007). 5 Id. at 14876, para. 41 (“[S]ubject to a one-year transition period for existing wireline broadband transmission services, all wireline broadband Internet access service providers are no longer subject to the Computer II requirement to separate out the underlying transmission from wireline broadband Internet access service and offer it on a common carrier basis.”). (continued….) Federal Communications Commission DA 18-99 132 networks.6 Canada likewise introduced unbundling requirements for DSL, but facilities-based competition between incumbent telecommunications and cable operators has been the dominant characteristic.7 5. South Korea initially relied on facilities-based competition between the incumbent and two entrants to achieve a high rate of broadband penetration.8 During this initial phase, DSL was classified as a relatively unregulated service, and the two entrants provided broadband service through cable modem using cable facilities leased from the Korea Electric Power Corporation (KEPCO), a state- owned utility.9 After the incumbent had surpassed the entrants to become the dominant market leader, South Korea then shifted to a regulatory regime that imposed open access and related requirements on the incumbent’s legacy network.10 6. Similarly, the extent of facilities-based competition varies across countries with an open access regulatory model. In Europe, the cable network footprint ranges from no presence in several countries to nearly ubiquitous (over 95 percent) coverage of households in a few others,11 but cable deployment, where it exists, “is generally limited to densely populated urban areas and, to a lesser extent, some suburban areas.”12 Facilities-based competition from cable operators is recognized as one of the 6 Review of the Section 251 Unbundling Obligations of Incumbent Local Exchange Carriers et al., CC Docket Nos. 01-338 et al., Report and Order and Order on Remand and Further Notice of Proposed Rulemaking, 18 FCC Rcd 16978, 17141, para. 272 (2003) (“[W]e decline to attach unbundling requirements to the next-generation network capabilities of fiber-based local loops, i.e., those loops that make use of fiber optic cables and electronic or optical equipment capable of supporting truly broadband transmission capabilities . . . .”); Richard Bennett, American Enterprise Institute G7 Broadband Dynamics: How Policy Affects Broadband Quality in Powerhouse Nations at 65- 66 (2014) (AEI Broadband Report), http://www.aei.org/wp-content/uploads/2014/11/G7-Broadband-Dynamics- Final.pdf; Wolfgang Briglauer and Klaus Gugler, The Deployment and Penetration of High-Speed Fiber Networks and Services: Why are EU Member States Lagging Behind? at 820 (2013) (EU Member States Paper), https://www.wu.ac.at/fileadmin/wu/d/i/iqv/Gugler/Artikel/bg_tp.pdf; OECD, Broadband Networks and Open Access at 12 (2013) (OECD Broadband Report), http://www.oecd-ilibrary.org/docserver/download/5k49qgz7crmr- en.pdf?expires=1514991235&id=id&accname=guest&checksum=F273F202DDDA94BC2907BD4B211FB749. 7 OECD Broadband Report at 11; AEI Broadband Report at v (identifying facilities-based competition as Canada’s model). 8 See Sujin Choi, Facilities to Service Based Competition, Not Service to Facilities Based, for Broadband Penetration and Investment: A Comparative Study between the United States and South Korea at 27-29 (2011) (South Korea Comparative Study Paper), https://ssrn.com/abstract=1989168; Kenji Kushida and Seung-Youn Oh, The Political Economies of Broadband Development in Korea and Japan, 47 Asian Survey, 481, 493 (2007) (Korea and Japan Broadband Development Paper), http://repository.brynmawr.edu/cgi/viewcontent.cgi?article=1029&context=polisci_pubs. 9 One of the entrants provided broadband service through both cable modem and DSL. 10 See South Korea Comparative Study Paper at 29; Korea and Japan Broadband Development Paper at 496-97; OECD Broadband Report at 12-14. The OECD Broadband Report finds that mandatory LLU is not available in South Korea, but the South Korea Comparative Study Paper indicates otherwise. The OECD Broadband Report also characterizes the availability of KEPCO’s cable infrastructure as a type of open access arrangement, but the South Korea Comparative Study Paper cites studies concluding that government subsidies, rather than open access to cable facilities, were the main factor creating facilities-based competition. 11 Body of European Regulators for Electronic Communications, Challenges and Drivers of NGA Rollout and Infrastructure Competition at 11 (2016) (BEREC Report) (identifying four countries—Belgium, Switzerland, Malta and the Netherlands—as having ubiquitous or nearly ubiquitous cable coverage and three countries—Italy, Greece and Iceland—as having no coverage), http://berec.europa.eu/eng/document_register/subject_matter/berec/reports/6488-berec-report-challenges-and- drivers-of-nga-rollout-and-infrastructure-competition. 12 Id. at 12. (continued….) Federal Communications Commission DA 18-99 133 main drivers of Next Generation Access (NGA) network rollout in a number of European countries, such as Belgium and the Netherlands.13 National regulators in some European Union (EU) member states have decided to transition from uniform nationwide access pricing to geographically differentiated access regimes in which access prices and wholesale obligations vary according to the degree of facilities-based inter-platform competition prevailing in local geographic markets.14 7. Regional and National Variation in Open Access Regimes. The open access regulatory model varies depending on the access price and other non-price access terms and conditions such as the minimum number of lines competitors can lease. For example, the lower the access price, the greater the incentive for new providers to enter the retail market. Conversely, a high access price likely discourages retail entry. As the access price increases, an open access regime approaches and, at the limit, might effectively converge with a facilities-based model.15 8. There is significant regional and national variation in the scope and details of access regulations.16 The EU regulatory framework extended open access regulations to fiber-based NGA networks in EU member states.17 According to some European analysts, the EU regulatory framework imposes the most comprehensive and strict access obligations on NGA networks in comparison with regulatory policies in the leading East-Asian fiber nations as well as the United States.18 The EU framework does not apply to all European countries, however. The regulatory model used in Switzerland, which is not a member of the EU, promotes infrastructure-sharing and co-investment arrangements for fiber-based NGA networks based on partnership agreements between the incumbent operator and local utilities, instead of mandatory fiber wholesale access obligations.19 9. In addition, although EU member states are bound by a common regulatory framework and other directives, implementation at the level of the individual EU member states has given rise to different national approaches to access pricing and other wholesale obligations for both legacy networks and NGA networks.20 National regulators have some discretion to adapt the common EU legal framework for open access regulations to local market conditions and the national policy goals of their respective countries. In Germany and the United Kingdom, this resulted in what one analyst characterizes as a “contingent model” that differentiates the incumbent’s wholesale access obligations based on the type 13 Id. at 10-12, 29. 14 See Roberto Balmer, Geographic Regulation and Cooperative Investment in Next Generation Broadband Networks: A Review of Recent Literature and Practical Cases (2013), https://ssrn.com/abstract=2369049; Marc Bourreau, Carlo Cambini and Steffen Hoernig, Geographic Access Markets and Investments at 3-5 (2015) (Geographic Access Markets Paper), http://docentes.fe.unl.pt/~shoernig/papers/geo_2013Dec28.pdf. 15 AEI Broadband Report at 66. 16 See id. at vi, 66-76. 17 Wolfgang Briglauer, Carlo Cambini, and Michal Grajek, Centre for European Economic Research, Speeding Up the Internet: Regulation and Investment in European Fiber Optic Infrastructure at 3 (2017) (Internet Regulation and Investment Paper), https://ssrn.com/abstract=2962532. 18 Id. at 1; Wolfgang Briglauer, Stefan Frubing, and Ingo Vogelsang, Centre for European Economic Research, The Impact of Alternative Public Policies on the Deployment of New Communications Infrastructure—A Survey at 4 (2015) (Communications Infrastructure Deployment Public Policies Paper), http://ftp.zew.de/pub/zew- docs/dp/dp15003.pdf. 19 See BEREC Report at 129 (stating that, under Switzerland’s regulatory approach, “[r]egulation occurs only upon request by an alternative operator (ex post regulation); ULL regulation is by law limited to copper local loops such that fibre local loops are not regulated . . . .”); Communications Infrastructure Deployment Public Policies Paper at 2, 16. 20 AEI Broadband Report at 69-74; BEREC Report at 31-38, 44-129; Internet Regulation and Investment Paper at 19-20. (continued….) Federal Communications Commission DA 18-99 134 of network technology, with easy access to legacy networks and more restricted access to advanced networks.21 10. A similarly differentiated approach to the open access regulatory model was developed in Japan. Like the EU, the Japanese regulator extended wholesale access obligations to the incumbent’s fiber infrastructure, but the unbundling requirements initially established for fiber allowed the incumbent to offer access terms and conditions that were viewed as less effective in promoting service-based competition than those for copper local loops.22 11. Finally, Australia and New Zealand adopted open access regimes for fiber NGA networks that differ from those in both European nations and Japan. Both countries opted to condition the deployment of publicly-funded national fiber access networks on a strict form of open access that involves structural separation of network ownership and wholesale operations from the provision of retail services.23 12. Public Funding. National regulatory models differ with respect to the use of other types of policy instruments, including public funding of broadband networks. The U.S. regulatory model reserves the use of subsidies primarily to improve network coverage and quality in rural areas, where network deployment may be unprofitable, while relying primarily on private investment to finance advanced network deployment in urban areas.24 Canada has a similar approach, with broadband largely deregulated in urban markets but heavily subsidized in rural areas. 25 The main difference is Canadian geography and its vast, sparsely populated northern areas. 13. In Europe, national state aid projects for NGA fiber network deployment appear to be heavily targeted at rural areas and so-called “white areas,” which are defined as locations that currently do not have broadband coverage by the same type of infrastructure (either basic broadband or NGA)26 and where no deployment is likely to take place on a commercial basis within the next three years.27 Examples include the national public funding programs for NGA network deployment of Finland, France, 21 AEI Broadband Report at 71. 22 Id. at 67-69 (explaining that competitors could lease a single line for ADSL but had to lease a bundle of eight strands for fiber, and that, as of 2007, the incumbent had installed about 79% of local fibers and controlled nearly 71% of the retail market for fiber, but controlled only about 37% of the retail market for ADSL despite having installed nearly 100% of copper lines); EU Member States Paper at 828. 23 See OECD Broadband Report at 13, 18-19; Bronwyn Howell, Mythological Musings From the Antipodes (Dec. 18, 2013) (Howell 2013), http://www.aei.org/publication/mythological-musings-antipodes/; Jeffrey Eisenach, Australia’s Failed Expe riment in Government-Owned Broadband (Mar. 6, 2014) (Eisenach 2014), http://www.aei.org/publication/australias-failed-experiment-in-government-owned-broadband/. 24 AEI Broadband Report at 9. 25 Id. at 9, 67. 26 BEREC Report at 27, n.47 (citing Commission Regulation (EU) No 651/2014 of 17 June 2014 declaring certain categories of aid compatible with internal market in application of Articles 107 and 108 of the Treaty, 26.6.2014 Official Journal of the European Union L187/1 at L187/29 (2014) (“‘Basic broadband networks’ means networks with basic functionalities which are based on technology platforms such as asymmetric digital subscriber lines (up to ADSL2+ networks), non-enhanced cable (e.g. DOCSIS 2.0), mobile networks of third generation (UMTS) and satellite systems”), https://publications.europa.eu/en/publication-detail/-/publication/1291bb4c-fcfe-11e3-831f- 01aa75ed71a1/language-en. 27 BEREC Report at 27. See European Commission, EU Guidelines for the Application of State Aid Rules in Relation to the Rapid Deployment of Broadband Networks at 14, paras. 63, 66 (2013) (defining “white areas” as those in which there is no broadband infrastructure and it is unlikely to be developed in a period of 3 years), http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:C:2013:025:0001:0026:EN:PDF. (continued….) Federal Communications Commission DA 18-99 135 Germany, Portugal, and the United Kingdom.28 Funding levels and the project design vary across countries, and not all European countries have established public funding programs. 14. Countries that rely on relatively more comprehensive public subsidy programs to drive broadband investment appear to be limited in number, but they include some national leaders in broadband performance. In Europe, for example, the northern European Scandinavian nations (Denmark, Iceland, Norway, and Sweden) have “a long-lasting history” of state aid programs that are credited with helping to make these countries “European forerunners” in broadband coverage and penetration.”29 In the case of Sweden, a package of measures adopted in 2000 to promote broadband development included funding for a national backbone network.30 In addition, municipalities and community-owned local utilities have become the most important alternative operators in the Nordic countries and the Netherlands by directly investing in the deployment of fiber (FTTH/B) networks.31 Sweden uses municipal networks to drive investment in advanced broadband networks. Municipal networks were deployed in over 200 of the 290 Swedish municipalities as of 2015, and accounted for 23 percent of fixed broadband investments in 2014.32 15. Japan and South Korea use public subsidies to drive investment in broadband networks.33 Both countries have a history of offering direct and indirect forms of financial aid (tax incentives, low- or no-interest loans) to support the rollout of broadband networks by service providers.34 The government of South Korea fostered the development of facilities-based competition by subsidizing network deployment for cable and DSL entrants as well as the incumbent telecom operator.35 In both countries, the state-aid- driven approach continued with the subsequent establishment of major national public funding programs for the deployment of NGA fiber networks.36 16. The experience of other countries shows that government-subsidized broadband may not always be successful. For instance, the Australian government budgeted $43 billion AUD (about $40 billion USD) for its government-owned and operated National Broadband Network (NBN). The project was eventually scaled back after repeated delays and cost overruns that caused the budget to grow to $72.6 billion AUD in four years.37 17. Relationship Between Regulatory Models and Public Funding. In countries with the open access model, public funding for broadband deployment tends to be closely linked to open access requirements.38 The structural separation arrangements for both Australia’s and New Zealand’s publicly- 28 BEREC Report at 66-73, 106-10, and 130-33. 29 EU Member States Paper at 826, 829. 30 Bengt G Molleryd, Development of High-Speed Networks and the Role of Municipal Networks at 49 (2015) (OECD High-Speed Networks Report), http://www.oecd-ilibrary.org/docserver/download/5jrqdl7rvns3- en.pdf?expires=1511963783&id=id&accname=guest&checksum=254E92D62FAAF806F6F53E3E5D6724ED. 31 See EU Member States Paper at 826, 829; OECD High-Speed Networks Report at 15; BEREC Report at 23-25, 27 (singling out Switzerland in addition to Denmark and Sweden as leading examples of investment by local municipalities and community-owned utilities in NGA rollout). 32 OECD High-Speed Networks Report at 50. 33 See EU Member States Paper at 826-29. 34 Id.; Korea and Japan Broadband Development Paper at 494-502. 35 South Korea Comparative Study Paper at 23. 36 EU Member States Paper at 827-29. 37 See Eisenach 2014; Howell 2013; Bronwyn Howell, Government-Subsidized Fiber: Careful What You Wish For (Mar. 26, 2014) (Howell 2014), http://www.aei.org/publication/government-subsidized-fiber-careful-wish/. 38 See OECD Broadband Report at 19-23. (continued….) Federal Communications Commission DA 18-99 136 funded national fiber access networks illustrate this link. In addition, under EU guidelines on national state aid projects for broadband deployment, network operators are generally obligated to provide wholesale access to service providers for a period of at least seven years.39 In Europe and elsewhere, many publicly-funded municipal networks also operate based on an open access model.40 18. In the case of the facilities-based model, South Korea’s subsidization of broadband deployment differentiates the Korean approach from the U.S. and Canadian variants. The United States is widely regarded as having adopted a deregulatory and largely market-driven broadband strategy, and Canada’s model is regarded as having a similar approach.41 In South Korea, by contrast, the facilities- based model was combined with a state-aid-driven approach to encourage broadband investment.42 South Korea’s broadband strategy is often compared with that of Japan. Despite the differences in their regulatory models, the governments of both countries are viewed as taking a relatively interventionist approach to broadband development.43 II. REGULATORS FOR THE UNITED STATES AND THE COMPARISON COUNTRIES 19. Table 2 identifies the regulator and, where relevant, the ministry responsible for regulating broadband, in the United States and the respective comparison countries. Table 2 Regulator and/or Ministry by Country (2017) Country Regulator/Ministry Australia Australian Communications and Media Authority Department of Communications and the Arts Australian Competition and Consumer Commission Austria Austrian Regulatory Authority for Broadcasting and Telecommunications Telecom Control Commission Austrian Communications Authority Belgium Belgian Institute for Postal Services and Telecommunications Minister of Development Cooperation, Digital Agenda, Telecom and Postal Services Canada Canadian Radio-television and Telecommunications Commission (CRTC) Innovation, Science and Economic Development Canada (ISED) Chile Subsecretaria de Telecomunicaciones Ministry of Transport and Telecommunications Czech Republic Czech Telecommunication Office Czech Association of Electronic Communications Denmark Danish Business Authority Danish Competition and Consumer Authority Danish Energy Agency Ministry of Energy, Utilities, and Climate Estonia Ministry of Economic Affairs and Communications Technical Regulatory Authority Finland Finnish Communications Regulatory Authority Ministry of Transport and Communications France Autorite de Regulation des Communications Electroniques et des Postes Ministry of the Economy, Finance and Industry National Frequency Agency 39 Id. at 20; BEREC Report at 27. 40 See OECD Broadband Report at 21-23; OECD High-Speed Networks Report at 16. 41 See EU Member States Paper at 819; AEI Broadband Report at xii, 9, 67. 42 South Korea Comparative Study Paper at 7-8. 43 See EU Member States Paper at 828; Korea and Japan Broadband Development Paper at 482. Federal Communications Commission DA 18-99 137 Country Regulator/Ministry Germany Bundesnetzagentur Federal Ministry of Transport and Digital Infrastructure Greece Hellenic Telecommunications and Post Commission Ministry of Infrastructure, Transport, & Networks Iceland Post and Telecom Administration in Iceland Ministry of Transport and Local Government Ireland Commission for Communications Regulation Department of Communications, Climate Action, & Environment Italy Communications Regulatory Authority Ministry of Economic Development Japan Ministry of Internal Affairs and Communications Latvia Public Utilities Commission Ministry of Transport Electronic Communications Office Luxembourg Institut Luxembourgeois de Regulation Mexico Instituto Federal de Telecomunicaciones Secretaria de Comunicaciones y Transportes Netherlands Netherlands Authority for Consumers and Markets Ministry of Economic Affairs & Climate Policy Radiocommunications Agency New Zealand Commerce Commission Ministry of Business, Innovation, and Employment Norway Norwegian Communications Authority Ministry of Transport and Communications Portugal Autoridade Nacional de Comunicações Autoridade da Concorrencia South Korea Korea Communications Commission Ministry of Science and ICT Korea Communications Agency Spain Comision Nacional de los Mercados y la Competencia Ministerio de Energia, Turismo y Agenda Digital Sweden Swedish Post and Telecom Authority Ministry of Enterprise and Innovation Switzerland Federal Communications Commission (ComCom) Federal Office of Communications (OFCOM) United Kingdom Office of Communications Department for Digital, Media, Culture, and Sport Competition and Markets Authority United States Federal Communications Commission (FCC) Source: TeleGeography GlobalComms Database. III. INTERNATIONAL BROADBAND COMPETITION AND USAGE 20. Below, we provide information concerning the major fixed and mobile broadband competitors and the types of technologies used to provide broadband in the United States and in the 28 comparison countries. Generally, major competitors are those competitors with a market share of at least 10 percent. A. Fixed Broadband Competition 21. In Table 3 below, we provide information concerning the major fixed broadband competitors in the United States and in the 28 comparison countries. Federal Communications Commission DA 18-99 138 Table 3 Major Fixed Broadband Competitors and Types of Technology by Country (2017) Country Major Fixed Broadband Competitors Types of Technology Australia iiNet Cable HFC DOCSIS 3.0, DSL ADSL, DSL ADSL2+, DSL SHDSL, DSL VDSL2, LAN/FTTx Singtel Optus Cable HFC, DSL ADSL, DSL ADSL2+, Satellite Telstra Cable HFC, Cable HFC DOCSIS 3.0, DSL ADSL, DSL ADSL2+, LAN/FTTx FTTH, Satellite TPG Telecom DSL ADSL, DSL ADSL2+, LAN/FTTx FTTB Austria A1 Telekom Austria DSL ADSL, DSL ADSL2+, DSL G.fast, DSL VDSL2, VDSL2 Vectoring, LAN/FTTx FTTx UPC Cable Ethernet over Cable TV, Cable HFC DOCSIS 3.0, DSL ADSL, DSL ADSL2+ Belgium Proximus DSL ADSL, DSL ADSL2+, DSL SDSL, DSL VDSL, DSL VDSL2, VDSL2 Vectoring, LAN/FTTx FTTH Telenet Cable HFC DOCSIS 3.0 Canada Bell Canada Enterprises DSL ADSL, DSL ADSL2+, DSL VDSL, DSL VDSL2, LAN/FTTx FTTB, LAN/FTTx FTTH, LAN/FTTx FTTN Rogers Communications Cable HFC, Cable HFC DOCSIS 3.0, Cable HFC DOCSIS 3.1, DSL ADSL, LAN/FTTx FTTB, LAN/FTTx FTTH Shaw Communications Cable HFC, Cable HFC DOCSIS 2.0, Cable HFC DOCSIS 3.0, LAN/FTTx FTTH Telus Corporation Cable HFC, DSL ADSL, DSL ADSL2+, DSL VDSL2, LAN/FTTx FTTB/FTTH, LAN/FTTx FTTN Videotron Cable HFC, Cable HFC DOCSIS 3.0, Cable HFC DOCSIS 3.1 Chile Claro Chile Cable HFC, DSL ADSL Movistar Chile DSL ADSL, DSL VDSL, LAN/FTTx FTTB/FTTH VTR Cable HFC DOCSIS 3.0 Czech Republic O2 Czech Republic DSL ADSL, DSL ADSL2+, DSL VDSL, LAN/FTTx FTTx RIO Media Cable HFC, LAN/FTTx FTTC/FTTH UPC Ceska Republika Cable HFC DOCSIS 2.0, Cable HFC DOCSIS 3.0 Denmark Stofa Cable Ethernet over Cable TV, Cable HFC, Cable HFC DOCSIS 3.0, LAN/FTTx FTTB/FTTH TDC Cable HFC, Cable HFC DOCSIS 3.0, Cable HFC DOCSIS 3.1, DSL ADSL, DSL ADSL2+, DSL VDSL2, DSL VDSL2 Vectoring, DSL VDSL2- Vplus, LAN/FTTx FTTx Estonia Telia Eesti DSL ADSL, DSL ADSL2+, DSL VDSL, LAN/FTTx Starman Cable HFC, Cable HFC DOCSIS 3.0 STV Cable Cable, LAN/FTTx FTTB/FTTH Finland DNA Finland Cable HFC, Cable HFC DOCSIS 3.0, Cable HFC DOCSIS 3.1, DSL ADSL, DSL ADSL2+, DSL VDSL2 Elisa Corporation Cable HFC, Cable HFC DOCSIS 3.0, DSL ADSL, DSL ADSL2+, DSL SDSL, DSL VDSL2, LAN/FTTx FTTB/FTTH Federal Communications Commission DA 18-99 139 Country Major Fixed Broadband Competitors Types of Technology Telia Finland Cable HFC, DSL ADSL, DSL ADSL2+, DSL VDSL2, LAN/FTTx FTTH France Bouygues Telecom DSL ADSL, DSL ADSL2+, DSL VDSL2, LAN/FTTx FTTH Iliad DSL ADSL, DSL ADSL2+, DSL VDSL2, LAN/FTTx FTTH Orange France DSL ADSL, DSL ADSL2+, DSL VDSL2, LAN/FTTx FTTH, LAN/FTTx FTTN SFR Group Cable HFC, Cable HFC DOCSIS 3.0, Cable HFC DOCSIS 3.1, DSL ADSL, DSL ADSL2+, DSL VDSL, LAN/FTTx FTTH, LAN/FTTx FTTN Germany Telekom Deutschland DSL ADSL, DSL ADSL2+, DSL VDSL, DSL VDSL2 Vectoring, LAN/FTTx FTTH United Internet DSL ADSL, DSL ADSL2+, DSL SHDSL Unitymedia Cable HFC, Cable HFC DOCSIS 3.0, LAN/FTTx FTTB/FTTH Vodafone Germany Cable Ethernet over Cable TV, Cable HFC DOCSIS 3.0, DSL ADSL, DSL SHDSL, DSL VDSL, DSL VDSL2 Vectoring, LAN/FTTx FTTH Greece Cosmote DSL ADSL, DSL ADSL2+, DSL VDSL2, LAN/FTTx FTTC, Satellite Forthnet DSL ADSL, DSL ADSL2+, DSL SHDSL, DSL VDSL2, LAN/FTTx FTTB, Satellite VSAT Vodafone Greece DSL ADSL, DSL ADSL2+, DSL VDSL2, LAN/FTTx FTTB, Satellite Wind Hellas DSL ADSL, DSL ADSL2+, DSL SDSL, DSL VDSL2, LAN/FTTx FTTC Iceland 365 Media DSL ADSL, DSL VDSL, LAN/FTTx FTTH Siminn DSL ADSL, DSL ADSL2+, DSL SDSL, DSL VDSL, DSL VDSL2 Vectoring, LAN/FTTx FTTC, LAN/FTTx FTTH Vodafone Iceland DSL ADSL, DSL ADSL2+, DSL VDSL, DSL VDSL2 Vectoring, LAN/FTTx FTTH Ireland eir DSL ADSL, DSL ADSL2+, DSL VDSL2, DSL VDSL2 Vectoring, LAN/FTTx FTTC, LAN/FTTx FTTH Sky Ireland DSL ADSL2+, LAN/FTTx FTTC Virgin Media Ireland Cable HFC DOCSIS 2.0, Cable HFC DOCSIS 3.0 Vodafone Ireland DSL VDSL2, LAN/FTTx FTTB Italy Fastweb DSL ADSL, DSL ADSL2+, DSL VDSL2, LAN/FTTx FTTC, LAN/FTTx FTTH Telecom Italia DSL ADSL, DSL ADSL2+, DSL VDSL2, LAN/FTTx FTTC, Satellite VSAT Vodafone Italy DSL ADSL2+, DSL VDSL, LAN/FTTx FTTC, LAN/FTTx FTTH Wind Tre DSL ADSL, DSL ADSL2+, DSL SHDSL, LAN/FTTx FTTH Japan J:COM Cable HFC, Cable HFC (DOCSIS 3.0) KDDI Cable HFC (DOCSIS 3.0), LAN/FTTx FTTH NTT (NTT East & NTT West) Cable Ethernet over Fibre, DSL ADSL, DSL ADSL2+, LAN/FTTx FTTH Softbank DSL ADSL, DSL ADSL2+, LAN/FTTx FTTH Latvia Baltcom Cable HFC DOCSIS 3.0, LAN/FTTx FTTB/FTTH Lattelecom Group DSL ADSL, DSL ADSL2+, DSL VDSL, Federal Communications Commission DA 18-99 140 Country Major Fixed Broadband Competitors Types of Technology LAN/FTTx Luxembourg Post Luxembourg DSL ADSL, DSL ADSL2+, DSL VDSL, LAN/FTTx FTTB/FTTH Tango DSL ADSL, DSL ADSL2+, DSL VDSL, LAN/FTTx FTTx Mexico Megacable Cable HFC, Cable HFC DOCSIS 3.0 Telefonos de Mexico DSL ADSL, DSL ADSL2+, LAN/FTTx FTTH Netherlands KPN DSL ADSL, DSL ADSL 2+, DSL VDSL2, DSL VDSL2 Vectoring, LAN/FTTx FTTH VodafoneZiggo Cable Ethernet over Cable TV, Cable HFC DOCSIS 2.0, Cable HFC DOCSIS 3.0 New Zealand CallPlus DSL ADSL2+, DSL VDSL, LAN/FTTx FTTC/FTTH Spark DSL ADSL, DSL ADSL2+, DSL VDSL, LAN/FTTx FTTN Vodafone New Zealand Cable HFC DOCSIS 3.0, Cable HFC DOCSIS 3.1, DSL ADSL, DSL ADSL2+, DSL VDSL2, LAN/FTTx FTTH, Satellite VSAT Norway Altibox LAN/FTTx FTTH Telenor Norge Cable Ethernet over Cable TV, Cable HFC DOCSIS 3.0, DSL ADSL, DSL ADSL2+, DSL SHDSL, DSL VDSL2, LAN/FTTx FTTH/FTTN TDC Nordic Cable HFC, Cable HFC DOCSIS 3.0, DSL ADSL, DSL SHDSL, DSL VDSL Portugal Nos Cable HFC DOCSIS 3.0 PT Portugal DSL ADSL, DSL ADSL2+, LAN/FTTx FTTB/FTTH Vodafone Portugal DSL ADSL, DSL ADSL2+, LAN/FTTx FTTx South Korea KT Corp Cable Ethernet/LAN, DSL ADSL, DSL VDSL, LAN/FTTx FTTH, Satellite LG Uplus Cable Ethernet/LAN, Cable HFC, Cable HFC DOCSIS 3.0, Cable HFC DOCSIS 3.1, LAN/FTTx FTTC, Powerline (PLC or BPL) SK Broadband Cable HFC, Cable HFC DOCSIS 3.0, DSL ADSL, DSL VDSL, LAN/FTTx Ethernet/LAN, LAN/FTTx FTTB/FTTH Spain Orange Espana DSL ADSL, DSL ADSL2+, LAN/FTTx FTTH Telefonica Espana DSL ADSL, DSL ADSL2+, DSL VDSL2, LAN/FTTx FTTH Vodafone Spain Cable HFC, Cable HFC DOCSIS 2.0, Cable HFC DOCSIS 3.0, DSL ADSL, DSL ADSL2+, DSL VDSL2, LAN/FTTx FTTH Sweden Com Hem Cable HFC, Cable HFC DOCSIS 3.0, LAN/FTTx FTTB/FTTH Telenor Sweden Cable HFC, DSL ADSL, DSL ADSL2+, DSL VDSL2, LAN/FTTx, LAN FTTx FTTB/FTTH Telia Sweden DSL ADSL, DSL ADSL2+, DSL VDSL2, LAN/FTTx FTTB/FTTH Switzerland Swisscom DSL ADSL, DSL G.fast, DSL SDSL, DSL VDSL2, DSL VDSL2 Vectoring, LAN/FTTx FTTB/FTTH Sunrise Communications DSL ADSL, DSL ADSL2+, DSL VDSL, LAN/FTTx FTTH, LAN/FTTx FTTx UPC Switzerland Cable HFC DOCSIS 3.0 Federal Communications Commission DA 18-99 141 Country Major Fixed Broadband Competitors Types of Technology United Kingdom BT Group DSL ADSL, DSL ADSL2+, DSL G.fast, DSL SDSL, DSL VDSL2, LAN/FTTx FTTC, LAN/FTTx FTTH Sky DSL ADSL, DSL ADSL2+, LAN/FTTx FTTH TalkTalk DSL ADSL, DSL ADSL2+, LAN/FTTx FTTH Virgin Media Cable Ethernet over Fibre (EFM-F), Cable HFC, Cable HFC DOCSIS 3.0 United States AT&T DSL ADSL, DSL G.fast, LAN/FTTx FTTH, LAN FTTx FTTN Charter Communications Cable Ethernet over Fibre (EFM-F), Cable HFC, Cable HFC DOCSIS 3.0 Comcast Corp. Cable Ethernet over Cable TV, Cable HFC DOCSIS 3.0, Cable HFC DOCSIS 3.1 Verizon Communications DSL ADSL, LAN/FTTx FTTB/FTTH Sources: TeleGeography GlobalComms Database, as of January 2018; Ministry of Internal Affairs and Communications, Subscriptions Shares First Quarter 2017 Data Public Release, Appx. Section 2 Fixed Communications, (1) Data Communications 1. Fixed Broadband at 8 (2017) (first quarter 2017 results), http://www.soumu.go.jp/main_content/000494106.pdf; Ministry of Communications, General Communications Infrastructure Bureau, NTT East and West Fiber Wholesale Service Provisioning Conditions, No. 34-2 (2015), http://www.soumu.go.jp/main_content/000390866.pdf. B. Mobile Broadband Competition 22. Table 4 provides information concerning the major mobile broadband competitors in the United States and in each of the 28 comparison countries. Federal Communications Commission DA 18-99 142 Table 4 Major Mobile Broadband Competitors and Types of Technology by Country (2017) Country Major Mobile Broadband Competitors Types of Technology Australia Optus Mobile 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G TD-LTE, 4G LTE- Advanced Telstra 3.5G W-CDMA, 4G LTE, 4G LTE- Advanced Vodafone Hutchison Australia 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Austria A1 Telekom Austria 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Hutchison Drei Austria 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced T-Mobile Austria 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Belgium BASE Company 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE Orange Belgium 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Proximus 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Canada Bell Canada Enterprises 2.5G CDMA2000, 3G CDMA2000, 3G W- CDMA, 3.5G CDMA2000, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Rogers Communications 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Telus Corporation 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Chile Claro Chile 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Entel Chile 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Movistar Chile 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE Czech Republic O2 Czech Republic 2G GSM, 2.5G GSM, 3G CDMA2000, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced T-Mobile Czech Republic 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Vodafone Czech Republic 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Denmark Hi3G Access Denmark 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced TDC 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Telenor Denmark 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Telia Denmark 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Estonia Elisa 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Tele2 Eesti 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G Federal Communications Commission DA 18-99 143 Country Major Mobile Broadband Competitors Types of Technology W-CDMA, 4G LTE, 4G LTE-Advanced Telia Eesti 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Finland DNA Finland 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Elisa Corporation 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Telia Finland 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced France Bouygues Telecom 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Free Mobile (Iliad) 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Orange France 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced SFR Group 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Germany Telefonica Deutschland Holding 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Telekom Deutschland 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Vodafone Germany 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Greece Cosmote 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Vodafone Greece 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Wind Hellas 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE Iceland Nova 3G W-CDMA, 3.5G W-CDMA, 4G LTE Siminn 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Vodafone Iceland 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE Ireland Eir Group Mobile 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE Hutchison 3G Ireland 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Vodafone Ireland 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Italy Telecom Italia 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Vodafone Italy 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Wind Tre 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE Japan KDDI (au) 3G CDMA2000, 3.5 CDMA2000, 4G LTE, 4G LTE-Advanced NTT DOCOMO 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Softbank Mobile 2G PHS, 3G W-CDMA, 3.5G W-CDMA, 4G PHS, 4G LTE Latvia Bite Latvia 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Federal Communications Commission DA 18-99 144 Country Major Mobile Broadband Competitors Types of Technology Latvijas Mobilais Telefons 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Tele2 Latvia 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Luxembourg Orange Luxembourg 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Post Luxembourg 2G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE Tango 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Mexico AT&T Mexico 2.5G iDEN, 3G W-CDMA, 3.5G W-CDMA, 4G LTE Movistar Mexico 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE Telcel 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE Netherlands KPN 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced T-Mobile Netherlands 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced, 4G TD-LTE LTE-Advanced VodafoneZiggo 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced New Zealand Spark 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Two Degrees Mobile 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE Vodafone New Zealand 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE Norway Telenor Norge 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Telia Norge 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Portugal Nos 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced PT Portugal 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Vodafone Portugal 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced South Korea KT Corp 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced LG Uplus 2.5G CDMA, 2.5G CDMA2000, 3.5G CDMA2000, 4G LTE, 4G LTE-Advanced SK Telecom 2.5G CDMA, 2.5G CDMA2000, 3G CDMA2000, 3G W-CDMA, 3.5G W- CDMA, 4G LTE, 4G LTE-Advanced Spain Orange Espana 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Telefonica Espana 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Vodafone Spain 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Sweden Hi3G Access Sweden 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G TD-LTE, 4G LTE-Advanced, 4G TD-LTE Federal Communications Commission DA 18-99 145 Country Major Mobile Broadband Competitors Types of Technology LTE-Advanced Tele2 Sweden 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Telenor Sweden 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Telia Sweden 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Switzerland Salt 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Sunrise Communications 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Swisscom 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced United Kingdom BT Group 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Hutchison 3G UK 3G W-CDMA, 3.5G W-CDMA, 4G LTE O2 UK 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE Vodafone UK 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced United States AT&T Mobility 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Sprint Corporation 2.5G CDMA 2000, 3G CDMA2000, 3.5G CDMA2000, 4G LTE, 4G LTE-Advanced, 4G TD-LTE T-Mobile US 2G GSM, 2.5G GSM, 3G W-CDMA, 3.5G W-CDMA, 4G LTE, 4G LTE-Advanced Verizon Wireless 2.5G CDMA2000, 3G CDMA2000, 3.5G CDMA2000, 4G LTE, 4G LTE-Advanced, 4.5G LTE-Advanced Pro Source: TeleGeography GlobalComms Database, as of January 2018. Federal Communications Commission DA 18-99 146 IV. TYPES OF APPLICATIONS AND SERVICES USED44 23. Table 5 presents data on the types of activities that consumers in the United States and the 28 comparison countries engage in while accessing the Internet. We provide these data as required by BDIA to identify “the types of applications and services used” and “business and residential use of such services . . . .”45 Table 5 Types of Applications and Services Used by Country (2014-2017) Country Online News Play or Download Games, Music, Videos Watch Video on Demand Make a Telephone or Video Call Participate in Social Networks Online Banking Online Shopping Australia46 N/A 24% N/A N/A 40% N/A 18% Austria47 66% 79% 14% 32% 58% 63% 68% Belgium 65% 72% 12% 44% 80% 75% 65% Canada48 55% 25% (gaming), 30% (music), 36% (video) N/A 20% 59% 68% 46% Chile49 51% N/A N/A N/A 92% 41% 35% Czech Republic 82% 72% 4% 40% 55% 63% 57% Denmark 72% 90% 49% 60% 77% 91% 84% Estonia 89% 84% 24% 47% 66% 90% 64% Finland 85% 91% 37% 34% 66% 92% 72% France 56% 75% 12% 34% 47% 69% 75% Germany 72% 78% 23% 31% 56% 59% 82% 44 The unit for the types of applications and services used in each country reflects the percentage of population. Country-specific data that were obtained from the European Commission’s International Digital Economy and Society Index (DESI) 2017 reflect the percentage of individuals who used Internet in the last three months as of the publication of DESI 2017. See European Commission, Digital Economy and Society Index (DESI) 2017 (2017) (DESI 2017), https://ec.europa.eu/digital-single-market/en/news/digital-economy-and-society-index-desi-2017. 45 47 U.S.C. § 1303(b)(3). 46 See European Commission, International Digital Economy and Society Index (I-DESI) at 81-86 (2016) (2016 I- DESI), https://ec.europa.eu/digital-single-market/en/news/2016-i-desi-report. The numbers reflected in the chart represent approximates based on the 2015 data displayed in the I-DESI. 47 DESI 2017. Data for the types of applications and services in the following countries were obtained from the Digital Economy and Society Index (DESI) 2017: Austria, Belgium, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Ireland, Italy, Latvia, Luxembourg, Netherlands, Portugal, Spain, Sweden, and the United Kingdom. 48 Canadian Internet Registration Authority (CIRA), CIRA Internet Factbook 2016: Internet Use in Canada (2016), https://cira.ca/factbook/domain-industry-data-and-canadian-Internet-trends/internet-use-canada. 49 Pew Research Center, Internet Seen as Positive Influence on Education but Negative on Morality in Emerging and Developing Nations: Internet Usage More Common Among the Young, Well-Educated and English Speakers at 23 (2015), http://assets.pewresearch.org/wp-content/uploads/sites/2/2015/03/Pew-Research-Center-Technology-Report- FINAL-March-19-20151.pdf. (continued….) Federal Communications Commission DA 18-99 147 Country Online News Play or Download Games, Music, Videos Watch Video on Demand Make a Telephone or Video Call Participate in Social Networks Online Banking Online Shopping Greece 85% 77% 12% 46% 68% 28% 45% Iceland50 95% 54% N/A 51% 84% 93% 68% Ireland 49% 73% 24% 42% 70% 64% 71% Italy 60% 79% 15% 34% 60% 42% 41% Japan51 55.9% 28.2% (gaming), 51% (video) 20.2% 49.2% 51.0% 14.7% 45.6% Latvia 84% 77% 15% 51% 71% 78% 55% Luxembourg 89% 89% 29% 54% 69% 73% 80% Mexico52 N/A 23% 6% N/A 29% N/A 9% Netherlands 75% 88% 39% 39% 66% 91% 79% New Zealand53 43% 28% 33% N/A 42% N/A 16% Norway54 94% 89% 54% 48% 78% 94% 79% Portugal 78% 83% 9% 39% 74% 41% 43% South Korea55 76.9% 77.0% N/A 39.9% 57.9% 50.8% 50.7% Spain 78% 83% 27% 31% 67% 54% 54% Sweden 87% 91% 49% 51% 75% 89% 80% Switzerland56 76% 23% 19% N/A 29% 60% 10% United Kingdom 68% 80% 34% 49% 73% 68% 87% United States 43%57 49% 28%61 N/A 69%62 43% 79% 50 European Commission, Digital Economy and Society Index (DESI) 2015: Country Profile – Iceland, https://ec.europa.eu/digital-single-market/en/scoreboard/iceland (last visited Jan. 16, 2018). 51 Data for the types of applications and services in Japan were provided by the Embassy of Japan. The data include specifically the percentage of individuals that subscribe to video on demand (20.2%). 52 See 2016 I-DESI at 81-86. The numbers reflected in the chart represent approximates based on the 2015 data displayed in the I-DESI. 53 See id. 54 European Commission, Digital Economy and Society Index (DESI) 2017: Norway, https://ec.europa.eu/digital- single-market/en/scoreboard/norway (last visited Jan. 16, 2018). 55 Data for the types of applications and services in South Korea were provided by the Embassy of South Korea. The data reflect information from a 2016 survey by the Ministry of Science and ICT and the Korea Internet and Security Agency. The data include information specifically pertaining to the percentage of subscribers to video on demand (“N/A”). 56 See 2016 I-DESI at 81-86. The numbers reflected in the chart represent approximates based on the 2015 data displayed in the I-DESI. 57 Pew Research Center, Americans’ Online News Use is Closing in on TV News Use (Sept. 7, 2017), http://www.pewresearch.org/fact-tank/2017/09/07/americans-online-news-use-vs-tv-news-use/. 61 Pew Research Center, About 6 in 10 Young Adults in U.S. Primarily Use Online Streaming to Watch TV (Sept. 13, 2017), http://www.pewresearch.org/fact-tank/2017/09/13/about-6-in-10-young-adults-in-u-s-primarily-use- online-streaming-to-watch-tv/. 62 Pew Research Center, Social Media Fact Sheet (Jan. 12, 2017), http://www.pewinternet.org/fact-sheet/social- media/. (continued….) Federal Communications Commission DA 18-99 148 Country Online News Play or Download Games, Music, Videos Watch Video on Demand Make a Telephone or Video Call Participate in Social Networks Online Banking Online Shopping (gaming),58 41% (audio),59 78% (video)60 (mobile),63 62% (online)64 (ever),65 15% (weekly)66 58 See Pew Research Center, Gaming and Gamers at 2 (2015), http://assets.pewresearch.org/wp- content/uploads/sites/14/2015/12/PI_2015-12-15_gaming-and-gamers_FINAL.pdf. 59 IFPI, Music Consumer Insight Report at 6 (2016), http://www.ifpi.org/downloads/Music-Consumer-Insight- Report-2016.pdf. 60 Statista, Percentage of Internet Users Who Watch Online Video Content on Any Device as of January 2017, by Country (2017), https://www.statista.com/statistics/272835/share-of-internet-users-who-watch-online-videos/. 63 Board of Governors of the Federal Reserve System, Consumers and Mobile Financial Services at 1 (2016), https://www.federalreserve.gov/econresdata/consumers-and-mobile-financial-services-report-201603.pdf. 64 Bank of America, Trends in Consumer Mobility Report at 5 (2016), http://newsroom.bankofamerica.com/files/press_kit/additional/2016_BAC_Trends_in_Consumer_Mobility_Report. pdf. 65 Pew Research Center, Online Shopping and E-Commerce at 2 (2016), http://www.pewinternet.org/2016/12/19/online-shopping-and-e-commerce/. 66 Id.