GROWING VOICE COMPETITION SPOTLIGHTS URGENCY OF IP TRANSITION By Patrick Brogan, Vice President of Industry Analysis

RESEARCH BRIEF NOVEMBER 22, 2013 GROWING VOICE COMPETITION SPOTLIGHTS URGENCY OF IP TRANSITION By Patrick Brogan, Vice President of Industry Analysis An updated USTelecom analysis of residential voice competition data reaffirms previous research showing that an overwhelming majority of consumers are dropping traditional switched telephone landlines from incumbent local exchange carriers (ILECs) in favor of modern mobile and Internet Protocol (IP) options. As utilization of the switched network declines and the transition to IP networks approaches critical mass, the economics of maintaining dual networks becomes more challenging and the need to address IP transition issues becomes increasingly urgent. This analysis provides historical estimates and straight line projections through year-end 2013 for both household-level voice service and total voice connections. As previously projected, by the end of 2012 U.S. households using wireless-only phone service had surpassed households using traditional ILEC switched landlines. By the end of 2013, wireless-only households will be almost 45 percent while households using traditional landlines will have dropped to 26 percent. In addition, for the first time, a greater portion of landline households will be using non-switched services than will be using the traditional switched network (see Chart 1). From 2000 to 2013, ILECs lost an estimated 62 percent of switched access lines and 70 percent of retail residential lines (see Chart 2) due to intensive facilities-based competition from wireless and cable companies. Chart 1: ILEC Switched vs. Wireless-Only and Interconnected VoIP 1 U.S. Household Primary Line Service Penetration and Straight-Line Projections (Percent of Telephone ) 100% 90% 93% 89% 84% 80% 70% 60% 50% 40% 30% 20% 10% 0% 76% 68% 60% 54% 31% 36% 33% Landline Other 26% 29% Than ILEC Switched 22% 17% 27% 26% 26% ILEC Switched 13% 23% 10% 20% 7% 18% 5% 15% 10% 6% 3% 4% 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012P 2013P 46% 39% 40% 45% Wireless Only Sources: FCC, CDC, Census, USTelecom Analysis (2008-13P); and FCC, CDC, NCTA, Financial Reports, USTelecom Analysis (2003-7) 1 In this chart, the ILEC Switched category includes all non-cable, non-ilec switched services, on the assumption that they are resold ILEC switched services. ILEC VoIP lines are included in the Landline Other Than ILEC Switched category. 1

These projections provide further support for USTelecom s December 2012 petition seeking a declaratory ruling from the Federal Communications Commission (FCC) that ILECs should no longer be subject to dominant carrier regulation. In particular, as detailed in USTelecom s December 2012 petition to the FCC, the mounting evidence of non-dominance undermines the core rationale for much legacy regulation of ILEC switched residential voice services. The estimates and projections below for households and connections are based on an analysis of the most current landline data from the FCC, wireless substitution data from the Centers for Disease Control (CDC), and household data from Census. This updates and revises previous analyses published in a January 2012 research brief, a November 2012 research paper, and an April 2013 research brief based on earlier data releases. Analysis As of year-end 2012, the share of U.S. telephone households for which a traditional ILEC-provided voice service (ILEC switched) was almost 33 percent: just less than 31.5 percent ILEC retail switched service and one percent resold ILEC switched wholesale service. Of the remaining telephone households at year-end 2012, 40 percent of households had cut the cord, relying entirely on wireless telephone service. Almost 20.5 percent were using cable telephony, mostly interconnected voice over Internet Protocol (VoIP); 2 percent were using independent VoIP providers; and 5 percent were using an ILEC VoIP service. 2 Compared to year-end 2011, ILEC switched share was down at least 6 percentage points from 39 percent (nearly 37.5 percent switched retail and nearly 1.5 percent resold switched wholesale). Cord-cutting households were up approximately 4.5 percentage points from less than 36 percent to just over 40 percent; cable telephony was up 1 percentage point from almost 19.5 percent to almost 20.5 percent; independent VoIP was down almost 1 percentage point; and ILEC VoIP grew almost 2 percentage points from more than 3 percent. The declining trends in switched voice service and the migration to IP networks continue unabated. In the December 2012 non-dominance petition and the November 2012 research paper, USTelecom for the first time included historical estimates for household voice telephony shares and projections, based on actual data for 2003 through mid-year 2011 and straight-line projections through year-end 2013. Previously, USTelecom had provided only point-in-time estimates of household voice shares. This analysis includes updated historical data and 2013 projections based on the latest available data from FCC (mid-2012), CDC (year-end 2012), and Census (2012 and 2011 revised), plus methodological revisions. 3 The revised projection for ILEC switched service is 26 2 Numbers do not add to 100 percent due to rounding. 3 In addition to new data, the methodology for the new straight-line projections is different. The previous projections were based on straight-line trends in actual households converted to percentages. The new projections are based on straight-line trends in the percentages. The change is discussed in the methodology appendix below. 2

percent of households for year-end 2013. The revised wireless-only projection is nearly 45 percent of households by year-end 2013. The revised projection for landlines other than ILEC switched 29 percent in 2013, consisting of about 23 percent cable and non-ilec VoIP and 6 percent ILEC VoIP. 4 Detailed year-end households results for 2008-13 are shown in the table in Appendix A below, with the yellow highlighted row corresponding to the ILEC switched share in Chart 1. Access Lines and Connections Analysis Switched access line data emphasize the degree to which utilization of ILEC switched networks has declined since line growth peaked around the year 2000. Unlike the householdslevel analysis above, the access line data below include both residential and business lines and they do not adjust for multiple lines per household or business establishment. By the end of 2012 total access lines using the ILEC switched network had fallen from 186 million to 80 million, and are projected to fall to 71 million by end of 2013, or a decline of 62 percent from the year 2000. For retail residential service, lines fell an estimated 70 percent over the same period. See Chart 2. Chart 2: Declining Utilization of ILEC Switched Telephone Network ILEC Switched Access Lines 2000 2013 (millions) 200 175 150 125 100 75 50 25 186 Est. 120-125 182 177 170 163 156 94 148 89 137 82 124 73 111 64 99 56 89 48 80 42 71 36 Total Retail and Wholesale 62% Loss Residential Retail 70% Loss - 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012P 2013P Source: FCC and USTelecom analysis. Includes primary and non-primary lines. Excludes ILEC VoIP and UNE-L unbundled loops. Projections are 6-month straight-line run rates from mid-2012. Pre 2005, carriers with <10,000 lines did not report and FCC did not report residential lines. An analysis of all U.S. voice connections, including wireless connections unadjusted for multiple subscriptions per household, underscores that ILECs are no longer dominant providers of voice service. By the end of this year, switched ILEC connections including business and 4 Overall, the revised 2013 household projections are largely consistent with previous projections, within a percentage point or two, plus or minus, of the two most recent projections. 3

residential, retail and wholesale switched lines will represent approximately 15 percent of U.S. voice connections, 17 percent if ILEC VoIP connections are included (see Chart 3). Chart 3: ILEC Switched Lines Are an Exceedingly Small Portion of Voice Connections Change in Share of Voice Connections, 2000 to 2013 100% 90% 80% 34.4% 70% 60% 5.3% 69.9% Wireless 50% 40% 30% 20% 10% 0% 60.4% 13.1% 15.1% 2000 2013P Non-ILEC ILEC VoIP 1.9% ILEC Switched Source: FCC, USTelecom Analysis. Incudes residential and business. Straight-line projections from mid-2012 with 6-month run rate. Summary The declining share and utilization of ILEC switched networks provides evidence that ILECs are no longer dominant voice providers. It underscores the urgency among providers and regulators to address IP transition issues. At the national level, even when including ILEC VoIP and wholesale, ILEC share of primary-line voice service was no more than 37 percent of households at the end of 2012. ILEC switched service was less than 33 percent in 2012, falling to 26 percent in 2013. When taking into account wireless mostly households that still had landlines, but predominantly used their wireless phones, ILECs were the provider of first choice for no more than 28 percent of U.S. households by the end of 2012, declining to an estimated 23 percent in 2013. However one slices the data, the portion of households getting voice service over ILEC-owned end-user facilities, switched or VoIP, is significantly lower than the roughly 60 percent market share threshold at which AT&T long distance was found to be non-dominant in the mid-1990s. This was true as of year-end 2012 and is even more so today as competitive and technological trends continue. Of course, traditional voice access is only one of the many options for communications. Internet voice and video communication, email, wireless text messaging, instant messaging, and communication through social networking sites offer non-traditional alternatives and these services are growing in popularity. For the most current data and analysis of these nontraditional alternatives, please see USTelecom s January 2013 research brief and the associated November 2012 research paper referenced in the introduction above. 4

Appendix A U.S. Household Voice Telephony Choices 2008 to 2013 Projected 2008 Share of Phone 2009 Share of Phone 2010 Share of Phone 2011 Share of Phone 2012 Projected Share of Phone 2013 Projected Share of Phone Total 117.1 n/a 117.4 n/a 119.3 n/a 120.8 n/a 122.0 n/a 123.1 n/a Total Phone 114.9 100.0% 115.1 100.0% 116.9 100.0% 118.2 100.0% 119.4 100.0% 120.6 100.0% Total Wired 89.7 78.1% 85.1 73.9% 80.4 68.7% 76.0 64.3% 71.5 59.9% 66.6 55.3% ILEC Wired 69.7 60.7% 62.7 54.5% 55.7 47.6% 49.7 42.1% 44.6 37.3% 39.1 32.4% ILEC Switched (retail and wholesale) 69.5 60.5% 61.8 53.7% 53.4 45.6% 45.9 38.8% 38.8 32.5% 31.4 26.1% ILEC Retail 66.7 58.1% 59.4 51.6% 51.3 43.9% 44.2 37.4% 37.5 31.4% 30.8 25.5% Non-ILEC, Non-Cable Switched (assumed resold ILEC) 2.7 2.4% 2.4 2.1% 2.1 1.8% 1.7 1.4% 1.3 1.1% 0.7 0.5% ILEC VoIP 0.2 0.2% 0.9 0.8% 2.3 2.0% 3.9 3.3% 5.8 4.8% 7.7 6.4% ILEC Est. "Wireline Mostly" 56.9 49.6% 49.9 43.3% 42.9 36.7% 37.5 31.8% 32.9 27.6% 27.9 23.2% ILEC Est. "Wireless Mostly" 12.8 11.1% 12.8 11.1% 12.8 10.9% 12.2 10.3% 11.7 9.8% 11.2 9.3% Non-ILEC Wired 20.0 17.5% 22.4 19.4% 24.7 21.1% 26.3 22.3% 26.9 22.6% 27.5 22.8% Non-ILEC Switched (excludes ILEC wholesale) 2.5 2.2% 2.3 2.0% 1.9 1.6% 1.7 1.5% 1.4 1.1% 1.0 0.8% Non-ILEC VoIP 17.6 15.3% 20.1 17.5% 22.8 19.5% 24.6 20.8% 25.6 21.4% 26.6 22.0% Non-ILEC Est. "Wired Mostly" 15.7 13.7% 17.1 14.9% 18.3 15.6% 19.2 16.3% 19.3 16.2% 19.3 16.0% Non-ILEC Est. "Wireless Mostly" 4.3 3.8% 5.3 4.6% 6.4 5.5% 7.1 6.0% 7.6 6.4% 8.2 6.8% Cable Operator 17.0 14.8% 19.8 17.2% 21.6 18.5% 22.9 19.4% 24.5 20.5% 26.0 21.6% Other than Cable Operator 3.0 2.6% 2.6 2.3% 3.0 2.6% 3.4 2.9% 2.5 2.1% 1.5 1.3% Cord Cutters 25.1 21.9% 30.0 26.1% 36.6 31.3% 42.2 35.7% 47.9 40.1% 53.9 44.7% No-Phone 2.2 n/a 2.3 n/a 2.4 n/a 2.6 n/a 2.6 n/a 2.6 n/a Sources: FCC, CDC, Census, USTelecom. Numbers may not add due to rounding. Projections in gray shaded areas are straight-line estimates based on most recent six-month trends in available data. A-1

Methodology and Assumptions Appendix B: Technical Discussion of Analysis and Results When estimating shares of wireless and wireline residential telephony adoption, it is useful to normalize the analysis and make other simplifying assumptions. These issues are discussed in turn below. Also discussed below are issues associated with combining different data sources, time series analysis, and projections. Finally, there is a discussion of methodological revisions. Normalizing by household This analysis takes a conservative approach and normalizes shares by household in order to understand the degree to which residential purchasers have completely replaced ILEC primary line service with an alternative technology (e.g., wireless, cable, over-the-top VoIP). In other words, the analysis attempts to answer two simple questions: does the household choose to purchase landline service or not; if it chooses landline service, does it choose an ILEC or an alternative? Residential landline service is usually purchased at the household level. When households purchase landline subscriptions, they buy a single line for joint use by each member of the household, even if there are multiple outlets and telephones in the home. Simultaneous use of the line for separate voice calls is not possible. A small portion of households today buy more than one line, known as non-primary lines. The analysis attempts to back those out to focus on the first basic question: landline or not? On the other hand, when people buy wireless service they generally buy a subscription or family plan, and within a household each individual user has a separate wireless device capable of simultaneous use. At a minimum, the ratio of wireless phones per wireless household is much greater than the ratio of landlines per landline household. Under our estimates and assumptions, at a national level, the ratios would be a little over one landline connection per landline household and approximately three wireless connections per wireless household. But to be consistent, the analysis focuses on the basic choice: landline or not? Therefore, by using household-level wireless substitution data from CDC, it effectively backs out multiple wireless subscriptions in a household, much like non-primary landlines. So, normalizing the analysis at the household level is both methodologically simple and conservative. It is also a practical choice. The CDC wireless substitution report offers household level results, providing a ready means of normalizing wireless adoption by household and comparing to household-level landline adoption. On the other hand, it would be difficult to normalize landlines by users to compare to wireless users because there is no easy way to distribute the purchase of a single landline among multiple household users. Thus, methodological problems are likely to arise using other approaches to normalizing shares. For example, subscriptions or connections while providing a useful gauge of total B-1

revenue generating units commingle wireless users engaging in complete wireless substitution ( cord cutting ) with those who purchase complementary wireless and wireline services, not to mention the potential for commingling with business-only wireless subscriptions. One might perform a useful analysis if it were possible to isolate the number of wireless phones per cord-cutting household. Such an analysis would also have to include non-primary landlines. But absent hard data or a good proxy, the result is heavily driven by assumptions about wireless subscriptions per cord-cutting household, and the range of feasible outcomes is wide. Additional simplifying assumptions The analysis assumes a household is either a landline household or not. So long as a household buys a wired line, it is a landline household, even if the household also purchases wireless services. The analysis then backs out a reasonable estimate of non-primary lines, as discussed below. Conversely, a wireless-only household is treated as if there is only one primary voice connection per household, even if there are multiple wireless devices in the household. This assumption furthers the primary objective of isolating complete replacement: has the household cut the cord or not? The CDC data identify the portion of combined landline-wireless households that use wireless mostly, but this analysis reports that separately as a supplemental point. Another simplifying and largely safe assumption is that households don t mix and match landline providers; i.e., they don t purchase a primary line from an ILEC and a non-primary line from a cable operator, or vice versa. Data combination and time series issues To construct a holistic analysis that accounts for all households while eliminating double counting, it is necessary to use data from multiple sources. Sources include surveys, such as Census and CDC, and industry-wide data collections, such as the FCC local telephone competition data. Surveys by nature contain small margins of error. To complicate matters further, the data sources do not reflect the exact same point in time. The FCC local telephone competition data reflect mid-year and year-end. So, in order to standardize data at a point in time, in this case the end of a half or a year, USTelecom makes straight-line adjustments to the CDC and Census data. USTelecom also made some assumptions or extrapolations where data are not readily available, such as non-primary lines and the portion of cable voice that is business rather than residential. Thus, while such a holistic approach entails a small degree of imprecision, the alternative is to rely on more precise but narrower, less integrated, and therefore less useful analyses. In terms of time series, the historical analysis is limited to 2003 and later years. Census household data are available back to 1940, but CDC household telephone status data were not available before 2003. CDC data provide the only practical starting point, with data on wireless-only, landline, and no-phone households. It is then possible to estimate the distribution of landline households among ILECs and non-ilecs based on FCC local telephone competition data. The FCC data are available in their current format from 2008 to present, allowing a reasonably granular level of precision. A different estimation methodology is necessary for 2003 to 2007. NCTA data were used to estimate cable B-2

telephone households; financial reports were used to estimate independent VoIP households; and remainder was assumed to be ILEC switched households, including resold ILEC switched lines. A number of additional simplifying assumptions were necessary for 2003 to 2007. Since the FCC did not report VoIP prior to year-end 2008, and ILEC VoIP customers were very few in 2008, they were assumed to be effectively zero before 2008. All or most cable telephony customers were assumed to be residential and 95 percent primary line. Vonage, the only independent VoIP provider for which public data were readily available, was assumed to represent approximately three-quarters of the independent VoIP industry; subscribers were assumed to be approximately 95 percent U.S., 95 percent primary line, and nearly all residential. This methodology trades off a small amount of precision to generate a holistic analysis that is roughly consistent and accurate over time. Moreover, there was no reason to believe that with the several sources of imprecision identified such as survey error and assumptions about business or primary lines that the estimates are biased in any particular direction. Rather, these estimates represent a very reasonable ballpark estimate of the portion of U.S. households using ILEC Switched, Wireless-Only, and Landline Other Than ILEC Switched options for voice telephony. The small amount of imprecision in no way puts into question the core conclusion: that the portion of households using ILEC Switched telephony is well below traditional thresholds for dominant carrier classification. Straight-line Projections Year-end 2012 and 2013 projections are straight-line projections based on six-month run rates from year-end 2011 to mid-year 2012. This is the third time this analysis has included projections. The initial projections in our the December 2012 non-dominance petition and the November 2012 research paper were based on six-month run rates from year-end 2010 to mid-year 2011. An April 2013 USTelecom research brief made projection based on the six-month run-rate from mid-year 2011 to year-end 2011, with a minor methodological adjustment. The initial projections were based on six-month run rates in households for each category, and shares were calculated as a percentage of the sum of households for all categories. The April 2013 analysis, and this analysis, estimate total household growth and project the changes in share based on the six-month run rates in the percentage share for each category. This constraint ensures top-line household growth is in line with historical trends while having no significant effect on proportionate household shares for the various categories. U.S. USTelecom starts the analysis by estimating the number of U.S. households, based on U.S. Census data from table HH-1, by Type, 1940 to Present. The HH-1 data are based on the Current Population Survey conducted in March each year. For a rough approximation of year-end households, assuming March represents roughly the first quarter end, it is possible to make linear estimates of quarterly growth from the end of the first quarter in a given year to the end of the first quarter in the next year. With the release of the March 2012 household data, Census has revised its March 2011 households up from 118.682 million to 119.927 million. B-3

Telephone Market share estimates should be based only on households that actually have telephone service. According to the CDC wireless substitution data, at the end of 2012, just over 2 percent of households had no telephone. Landline The most recent residential landline data are available from the FCC mid-year 2012 local telephone competition report. The starting point is retail landlines in other words, services sold to the end-user. From there, adjustments are made for cases where a non-ilec retail provider resells ILEC switched lines. The analysis assumes all non-cable, non-ilec switched services fall under this category and count them as ILEC switched lines. Non-Primary Lines As discussed above, to normalize shares by household, it is necessary to back out nonprimary lines so that, after adding landline households to wireless cord-cutting households and no-phone households, the number of households equals the number of total households estimated from Census data. Under this assumption, there were approximately six and a half million non-primary landlines as of mid-year 2012, which are allocated among ILECs and non-ilecs in proportion to their overall shares of landlines. This is a slight methodological change. In prior versions of this analysis, 75 percent were allocated ILECs and 25 percent to non-ilecs. The new methodology shifts a small portion of household share from non-ilecs to ILECs. The result under our allocation assumption is that non-primary lines account for approximately 8 percent of ILEC and non-ilec lines as of mid-year 2012. ILEC Landline The FCC telephone competition data indicate that ILECs had 50.2 million total residential lines at mid-year, including 44.9 million switched lines and 5.2 million interconnected VoIP lines. The analysis introduces another methodological change here: elimination of residential lines in U.S. territories because those areas are not included in the CDC household share estimates. These line figures are available from the FCC telephone competition reports. After eliminating 515,000 ILEC switched lines for territories, there were 44.4 million switched ILEC lines. Based on assumptions discussed above about the allocation of non-primary lines, ILECs were providing approximately 4.0 million non-primary lines at year-end 2011, including almost 3.6 million switched non-primary lines and more than 400,000 interconnected VoIP non-primary lines. As a result, ILECs were providing retail service to a total of 45.7 million households or 38 percent of 118.9 million telephone households. This figure consists of 40.9 million switched households and 4.8 million VoIP households, or 34 percent and 4 percent, respectively. Further, as discussed in greater detail below, resold ILEC lines may account for approximately 1 percent of telephone households. So, it appears that ILECs provided either retail or wholesale telephone service to a maximum of 40 percent of U.S. households at mid-year 2012. B-4

Cable and Other Landline The FCC telephone competition data indicate there were 27.4 million non-ilec residential interconnected VoIP connections and 3.4 million non-ilec switched connections. Like ILEC landlines, a methodology adjustment was made to back out residential lines in U.S. territories: 13,000 switched and 87,000 VoIP. The resulting figures are 27.3 million for VoIP and 3.4 million for switched lines. These are allocated among three categories: cable telephony, over-the-top interconnected VoIP, and other. The analysis makes a simplifying assumption for territories that all VoIP lines are cable and all switched lines are non-cable. Cable telephony accounted for approximately 20 percent of telephone households at midyear 2012. For non-ilec services sold using coaxial cable technology in broadband bundles a good proxy for overall cable telephony there were 25.2 million interconnected VoIP connections. Backing out 87,000 for territories leaves 25.1 million plus 1.7 million switched connections, for a total of 26.8 million connections, including residential and non-residential. To cross check this number, NCTA data indicate there were 25.3 million cable telephony subscribers as of year-end 2011 and 26.7 million as of year-end 2012, with a midpoint of 26.0 million. 5 Thus, the FCC and the NCTA data provide similar starting points. Two additional adjustments are needed to derive the share of telephone households. First approximately 96 percent of cable telephone subscribers were residential in mid-2012, so there were an estimated 25.8 million cable residential lines. Second, non-primary lines are allocated in proportion to landline shares and 2.1 million are backed out for cable. As a result, there were an estimated 23.7 million cable residential primary line telephone households, or approximately 20 percent of the 118.9 million telephone households in the U.S. Over-the-top interconnected VoIP providers accounted for nearly 3 percent of telephone households. The FCC reports the number of retail residential interconnected VoIP lines. The portion that was not cable was likely predominantly over-the-top interconnected VoIP provided by a third party that was not the underlying broadband provider. As of mid-year 2012, the FCC reported 27.3 million non-ilec residential interconnected VoIP lines, of which approximately 23.1 million were cable, i.e., sold in broadband bundles over coaxial cable, leaving 3.2 million non-ilec non-cable VoIP connections. Accounting for the possibility that some were non-primary lines, there were an estimated 3.0 million over-the-top VoIP primary lines, which was just under 2.5 percent of 118.9 million telephone households. Other landline providers accounted for slightly less than 1.5 percent of telephone households. Switched retail residential lines provided by non-ilecs not provided via cable telephony were presumably resold ILEC service, where the ILEC was the wholesaler and a competitive provider was the retailer. As of mid-year 2012, the FCC reported 3.4 million non-ilec switched residential lines, of which 1.7 million are presumed to be cable telephony, leaving another 1.7 million non-cable non-ilec 5 See http://www.ncta.com/stats/cablephonesubscribers.aspx (visited March 7, 2013). B-5

switched connections. Eliminating 13,000 lines in territories and accounting for the possibility that some were non-primary lines, there were an estimated 1.6 million noncable non-ilec switched primary lines, which is less than 1.5 percent of telephone households. Wireless-Only The CDC semi-annual wireless substitution surveys take place between January and June for the first half and between July and December for the second half of each year. The analysis assumes that the result reflects the midpoint of the survey period, or the end of the first quarter for the first half survey and the end of the third quarter for the second half survey. It also assumes linear growth between each semi-annual survey to estimate the mid-year and year-end percentages. According to the CDC wireless substitution data, 38.2 percent of U.S. households surveyed from July 2012 through December 2012 cut the cord and abandoned wireline phone service altogether. Similarly, CDC estimated that 35.6 percent of U.S. households surveyed from January 2012 to June 2012 had cut the cord. The midpoint of 37.0 percent approximates the mid-year 2012 share of total U.S. households. Since 2.1 percent of U.S. households are not telephone households, this represented 37.8 percent of U.S. telephone households, or approximately 44.9 million U.S. households that had cut the cord as of mid-year 2012. Looking forward, based on an average six-month change rate of 2.3% percentage points over the last three years, the first half 2013 figure is projected to be 40.5 percent, and the mid-point of 39.3 percent is a reasonable estimate for the percentage of total U.S. households cutting the cord as of yearend 2012. This represents 40.1 percent of telephone households. It is possible to adjust wireless substitution percentages for the probability that the underlying wireline carrier is affiliated with the cord-cutter s wireless provider. Such an analysis would provide only a rough estimate based on national averages for wireless subscriber and wireline footprint shares. The result would not affect the broad conclusion, and national shares would remain well below the 60 percent threshold at which AT&T long distance was found to be nondominant. Such an analysis is not provided here, however, because the driver of this analysis is the dominance rationale for regulation of traditional wireline telephony. Wireless Mostly According to the CDC data, during the second half of 2012 an additional 15.9 percent of total households had both landline and wireless telephone service, but used wireless predominantly. This was the same as the first half of 2012, meaning it provides a reasonable midpoint estimate for the year. Allocating among ILEC and non-ilec landline providers in proportion to landline shares indicates that ILECs likely are the first provider of choice to 10 percent fewer telephone households, or only about 30 percent of households mid-year 2012 (34 percent if ILEC VoIP is included in the calculation). Projections for year-end 2012 are 28 percent ILEC landline-mostly (and between 32 and 33 percent if ILEC VoIP is included.) B-6

Key Methodology Changes Compared to April 2013 Analysis There are two methodology changes in this analysis compared to previous versions, and these revisions have been applied to historical data back to year-end 2008. First, non-primary lines are allocated among ILECs and non-ilecs in direct proportion to the share of landlines. Previously, 75 percent of second lines were allocated to ILECs and 25 percent to non-ilecs. Within each category, VoIP and switched shares are allocated in proportion to shares, as they were before. Second, in order to better synchronize the FCC data with CDC and Census data, which do not include U.S. territories, territory lines were backed out of FCC totals. The analysis relies on a simplifying assumption that all non-ilec VoIP lines in territories are cable and all non-ilec switched lines in territories are non-cable. This change makes little difference in total primary lines compared to previous analyses, which effectively backed those out on the assumption that they were non-primary lines. This new approach results in better assumptions about non-primary lines and, in combination with new allocation of non-primary lines discussed above, should result in more finely tuned allocations among ILEC, non-ilec, VoIP, and switched lines overall. B-7