DATA SHOW VOICE COMPETITION GAINING IN ALL STATES By Patrick Brogan, Vice President of Industry Analysis

RESEARCH BRIEF DECEMBER 17, 2013 DATA SHOW VOICE COMPETITION GAINING IN ALL STATES By Patrick Brogan, Vice President of Industry Analysis USTelecom analysis of state-by-state data show competition for voice services is significant and gaining ground across all of the states. This research updates previous USTelecom state voice competition analyses 1 and quantifies household shares for wireless and landline alternatives to voice service from traditional providers using the legacy switched landline phone network. It complements a recent USTelecom analysis of national-level voice competition with more granular state detail. Both analyses estimate the portion of households that chose either to disconnect landline service altogether and go wireless-only, or to use alternative landline services, especially cable telephony and interconnected Voice over Internet Protocol (VoIP). 2 The state analysis provides more evidence that it is time for federal and state governments to address Internet Protocol (IP) transition issues, including USTelecom s petition to the FCC to find that traditional switched voice providers, known as incumbent local exchange carriers (s), are no longer dominant providers of voice communications. This state-level analysis includes new historical time series estimates from year-end 2008 through year-end 2011 based on the most current available wireless data from the Centers for Disease Control (CDC) and corresponding wireline competition data from the Federal Communications Commission (FCC). 3 Nationwide, USTelecom estimated that 60 percent of U.S. telephone households received their primary voice connection via switched service at year-end 2008, falling to 39 percent at year-end 2011. USTelecom further projects that this figure will fall to 26 percent by year-end 2013. While the most current state data are available only through 2011, it is very likely that state-level trends are, in varying degree, in line with the national trend of declining switched shares. 4 Table 1 shows competitive voice shares as a percent of U.S. households by state from 2008 to 2011. The data underscore that a large portion of households in all states had already shifted to alternatives to switched services by the end of 2011, with competition likely gaining share in most or all states in recent years. 1 USTelecom previously published an analysis of state-level data in 2011 and May 2013. Those analyses are not directly comparable to this new state analysis due to methodological upgrades made to maximize the consistency between the current state and national analyses. The differences are discussed in detail in Appendix B - A. 2 Interconnected VoIP is a term defined by the FCC to include VoIP services that can send and receive calls to and from the public switched telephone network. It includes Internet phone services such as Vonage and most cable telephony, but excludes computer to computer IP telephony services such as Skype or FaceTime. Throughout the remainder of this Research Brief, VoIP means interconnected VoIP. 3 CDC provides data through mid-year 2011. Estimates to year-end 2011 are straight-line. The FCC has provided actual data for year-end 2011. 4 This state-level analysis does not contain state-by-state share projections through 2013. It is very likely that all or most states are following the national trend of declining shares. See Appendix B - Technical Notes. 1

For 2008 and 2011, Table 1 shows the estimated portion of telephone households in three categories based on how they receive their primary telephone service: switched landlines; landlines other than switched; and wireless-only. The three categories sum to 100 percent of telephone households. 5 The last set of columns shows the change in percentage points for each of these categories during the three year period from 2008 to 2011. The states are ranked by switched share in 2011, from lowest to highest. The range statistics at the bottom of the table are shown from low to high for switched, and high to low for the other categories, so that within each column, the top row shows the state with the largest shift away from traditional landline voice services, and vice versa. Detailed data by state are available in Appendix A - State Details. Among the states, 6 the switched portion ranged from 28 percent at the low end to 51 percent at the high end at the close of 2011. The figures for 2011 reported here have likely declined even further in the last two years, in line with national trends. As noted above, switched share at the national level has been projected to fall from 39 percent at the end of 2011 to 26 percent by the end 2013. 7 The median switched portion was 37 percent. Only one of the 46 states for which data were available, Hawaii, had switched share greater than 50 percent; and even its 51 percent share for 2011 was lower than AT&T s long distance share when the FCC found in the mid-1990s that it was no longer a dominant provider of long distance. Four-fifths of states had switched household shares of 41 percent or less in 2011. Contrast these 2011 figures with 2008, just three years earlier, when switched services were being used by 44 percent to 71 percent of telephone households; the median switched share was 58 percent; and 38 of the 46 states examined had switched shares greater than 50 percent. 8 From 2008 to 2011, the median state saw the share of households using switched service decline by 22 percentage points, an average of more than seven percentage points per year. Four-fifths of states saw a decline of at least 17 percentage points. With continuing losses in recent years, it is clear that in the switched service has gone from a majority to a small and shrinking minority of households across the country. 5 These categories correspond to USTelecom s November 22, 2013 national voice competition analysis. Percentages are given as a share of telephone households. At the national level approximately 98 percent of households are telephone households and 2 percent have no telephone. 6 The analysis includes 46 states technically, 45 states and the District of Columbia. It excludes five states because data were not available from either the FCC (Alaska missing in both 2008 and 2011) or CDC (Iowa, Montana, South Dakota, and Wyoming missing in 2011). See Appendix B - Technical Notes for additional details. 7 The state switched figures exclude VoIP. This is consistent with the high-level national shares cited, which are for switched services only. The general observations would not change if VoIP were included with switched service; however, VoIP does play a greater role in some states relative to others. See Appendix B - Technical Notes for a detailed discussion of VoIP. 8 The four states missing 2011 CDC data did have 2008 CDC data. Iowa, Montana, South Dakota and Wyoming had estimated switched shares greater than 50 percent in 2008. CDC does report less statistically significant data, i.e. estimates with greater margins of error, for these four states in mid-2011. Projections to year-end are not possible. Montana, South Dakota and Wyoming had less than 50 percent switched share in mid-2011, and Iowa had about 51 percent, likely trending to 50 percent or less by year-end. So, taking these figures into consideration, 42 of 50 states had shares greater than 50 percent as of year-end 2008 and only one, maybe two, of 50 states had switched shares greater than 50 percent as of year-end 2011. 2

Table 1: Estimated State Competitive Voice Shares (Percent of U.S. Telephone Households, Year-End 2008 Year-End 2011) 2008 2011 Change Other Than Other Than Other Than States Ranked by Share Only Only Only MI 48% 26% 26% 28% 31% 41% -20% 5% 15% AZ 44% 29% 27% 29% 27% 43% -15% -2% 17% DC 58% 17% 25% 30% 19% 51% -28% 1% 27% MA 54% 32% 15% 32% 44% 25% -22% 12% 10% NH 56% 29% 14% 32% 40% 28% -25% 11% 14% FL 56% 18% 25% 32% 29% 39% -24% 10% 14% RI 44% 45% 10% 33% 50% 18% -12% 4% 7% KS 49% 24% 27% 33% 25% 43% -16% 0% 16% NE 46% 23% 31% 33% 23% 44% -13% 0% 13% NJ 57% 32% 11% 33% 48% 19% -24% 15% 8% NV 57% 22% 20% 33% 26% 41% -24% 3% 21% NY 50% 34% 15% 34% 43% 23% -17% 9% 8% AR 57% 9% 34% 34% 16% 50% -23% 7% 16% OK 49% 23% 28% 34% 27% 40% -15% 4% 11% OR 56% 17% 27% 34% 22% 44% -23% 6% 17% DE 60% 25% 15% 34% 35% 31% -26% 10% 17% WA 56% 21% 23% 34% 28% 38% -22% 7% 15% CO 53% 17% 30% 34% 22% 44% -19% 5% 13% UT 60% 18% 22% 35% 22% 43% -26% 4% 21% TX 56% 13% 30% 35% 19% 45% -21% 6% 15% IL 58% 19% 23% 36% 27% 38% -23% 8% 15% CT 60% 29% 11% 36% 42% 21% -24% 13% 11% TN 59% 14% 27% 37% 22% 41% -22% 8% 14% GA 59% 17% 24% 37% 23% 40% -21% 6% 16% MS 57% 9% 34% 37% 14% 48% -19% 5% 15% WI 58% 20% 22% 38% 24% 38% -20% 3% 17% ID 61% 9% 30% 38% 12% 50% -23% 3% 20% MD 63% 21% 16% 38% 31% 31% -25% 10% 15% OH 56% 21% 23% 38% 23% 39% -18% 2% 16% ND 47% 22% 30% 39% 15% 46% -8% -7% 15% SC 61% 15% 24% 39% 19% 42% -22% 4% 18% NC 59% 17% 24% 39% 22% 38% -20% 6% 14% LA 60% 17% 23% 39% 23% 38% -21% 6% 15% MN 58% 17% 25% 40% 24% 37% -19% 6% 12% IN 62% 14% 24% 40% 23% 37% -23% 9% 14% AL 64% 13% 23% 41% 20% 39% -23% 6% 16% VA 57% 21% 22% 41% 28% 31% -15% 6% 9% KY 50% 19% 32% 42% 21% 37% -8% 2% 6% CA 66% 18% 16% 42% 25% 33% -24% 8% 16% ME 60% 18% 23% 42% 21% 36% -18% 4% 14% PA 66% 18% 15% 44% 29% 28% -23% 10% 12% MO 69% 12% 20% 44% 18% 38% -24% 6% 19% VT 68% 13% 19% 44% 21% 35% -24% 8% 16% NM 66% 7% 27% 46% 12% 43% -21% 5% 16% WV 71% 11% 18% 49% 21% 30% -22% 10% 12% HI 66% 16% 17% 51% 20% 29% -15% 3% 12% AK n/a n/a n/a n/a n/a n/a n/a n/a n/a IA 28% 72% 57% n/a n/a n/a n/a n/a n/a MT 19% 81% 64% n/a n/a n/a n/a n/a n/a SD 14% 86% 53% n/a n/a n/a n/a n/a n/a WY 22% 78% 57% n/a n/a n/a n/a n/a n/a Ranges and Averages Low (High) 44% 45% 34% 28% 50% 51% -28% 15% 27% 20th (80th) Percentile 53% 24% 27% 33% 29% 43% -24% 9% 17% Median 58% 18% 23% 37% 23% 38% -22% 6% 15% 80th (20th) Percentile 62% 14% 16% 41% 20% 31% -17% 3% 12% High (Low) 71% 7% 10% 51% 12% 18% -8% -7% 6% Sources: CDC, FCC, Census, USTelecom analysis. AK, IA, MT, SD, WY excluded after 2011 Ranges are lowto-high for switched and high-to-low for wireless-only and landlines other than switched. 3

The data emphasize that one must assess switched share losses holistically, looking at the combined impact of wireline and wireless alternatives. It is insufficient to examine either in isolation. As of 2011, among the states, anywhere from 12 percent (New Mexico and Idaho) to 51 percent (Rhode Island) of telephone households had chosen a landline alternative to switched service, with a median of 23 percent. -only households ranged from 18 percent (Rhode Island) to 51 percent (District of Columbia). Yet, in no state was the combined share of wireless and landline alternatives to switched service less than 49 percent of households. This is because the dynamic and the relative significance of wireline and wireless competition vary across the states. The real competitive impact on s is measured by the combined losses to both wireline and wireless competitors. Thus, there are a number of states with a high ranking on landline alternatives, a low ranking on wireless-only and a high ranking for alternatives to switched service overall. Examples include many states ranked highly for total competitive share, such as Massachusetts, New Hampshire, Rhode Island, New Jersey and New York. On the other hand, some states are ranked low in wireline alternatives, high in wireless-only households and highly ranked for alternatives to switched service overall, such as the District of Columbia, Arkansas, Kansas, Nebraska and Oregon. In fact, some of the states ranked lowest for landline alternatives are ranked among the highest for wireless-only households, such as Texas, Mississippi, Idaho and North Dakota. There are also several states ranked highly for alternatives overall, with moderately strong rankings in both wireline and wireless alternatives. Examples include Michigan, Arizona, Florida and Nevada. Again, the figures are for year-end 2011 and there may have been shifts in the last couple years as the level of competition has risen overall. Regardless, relative rankings are less important in assessing competitiveness since even the lowest ranked states appear to have significant levels of competitive share on an absolute basis. Moreover, this household share analysis is a conservative method of assessing competitiveness for several reasons. First, with the focus on wireless-only households, the analysis treats any household with a landline as being fully in the landline category, ignoring dual wireline and wireless usage, especially wireless-mostly usage. 9 Second, this voice share analysis ignores communications alternatives other than interconnected voice, such as email, text messaging, computer to computer IP telephony, video chat and social networking. USTelecom has documented the extent of adoption of these alternatives in 2011; however, measuring and analyzing the competitive impacts of non-voice alternatives is very difficult at the national, let alone state level. Therefore, this analysis is limited to more easily-measured landline and wireless voice calling options. 9 -mostly households are those CDC identifies as having both landlines and wireless telephones but receiving all or most calls via wireless. For example, at the national level USTelecom projects that by year-end 2013 about 45 percent of telephone households will be wireless-only, another 55 percent will have landlines, of which between 45 percent and 50 percent will also have wireless phones, and more than 15 percent of those will be wireless-mostly. Allocating wireless-mostly households in proportion to share of landlines, switched but wireless-mostly would represent about 7 percent of households (47 percent of landlines times 15 percent) and an estimated 19 percent households in 2013 would be switched using wireline mostly. The corresponding national figure for year-end 2011 would be about 30 percent. Among the states, in 2011, this figure ranged from 21 percent to 41 percent with a median of 27 percent. See Appendix A - State Details, Table A1. 4

To summarize, at the end of 2011 the maximum household share for switched service in any state was 51 percent, the median was 37 percent and four-fifths of all states had an switched share of no more than 41 percent. Since 2011, s have continued to lose household share at a rapid pace. Since 2008, households using switched telephone service have gone from a slim majority of homes to a small and shrinking minority. These figures reflect a conservative approach since they describe only complete substitution of wireless for landlines. When taking into consideration the additional households that had both switched service and wireless phones but mostly used wireless phones, the share of households that used switched service exclusively or mostly ranged from 21 percent to 41 percent, with a median of 27 percent. Considering this state-level voice competition data, as well as the growing prevalence and popularity of non-voice communications options, it is increasingly clear that s are no longer dominant in the provision of switched voice communications services. 5

Appendix A - State Details This Appendix contains three tables showing state-by-state voice household shares. Data are included for year-end 2011 (Table A1), mid-year 2011 (Table A2), and year-end 2008 (Table A3). Mid-year 2011 data are included because CDC provides actual estimates that reflect midyear 2011 and the year-end 2011 estimates for wireless-only households are based on straightline projections. Actual FCC data were available for year-end 2011, but since the estimates are derived from both FCC and CDC data sets, the mid-2011 data reflect actual data from both without any projections. So, while the data in Table A2 are older by six months, there is less estimation involved. Each table shows eight columns of data for each state. All data are given as a percentage of telephone households. At the national level, approximately 98 percent of households have telephones and 2 percent do not. At the state level, no-telephone households range from 1 percent to 4 percent with a median of 2 percent. In each table below, the first two columns show wireless-only and landline households, which sum to 100 percent. The next two columns show landline households broken down into switched and landlines other than switched. The next two columns break landlines other than switched into two further subcategories: non- landlines, which include both non- VoIP lines and non- switched lines; and VoIP landlines. 10 The next column contains wireless-mostly households. 11 The last column shows an estimate of switched households that switched service either exclusively or mostly. 12 At the bottom of each table are statistics showing ranges and averages for each column: the high, low, median, 20 th percentile and 80 th percentile. The median is a type of average, technically the figure here percentage of households at which half of states are above and half below. High and low scores are self-explanatory. Percentiles are interpreted as follows: within each column the 80 th percentile means four-fifths of states have percentage household shares at or below that percentage, down to the lowest percentage; the 20 th percentile means onefifth of states have percentage household shares at or below that percentage, down to the lowest percentage. 10 Most non- lines are VoIP, largely cable telephone customers. There are some states where switched landlines represent a large portion landline competition. These include states like Arizona, Arkansas, Nebraska, North Dakota and Rhode Island. Some cable companies, such as Cox, still use switches deployed before VoIP became widely available. 11 -mostly is a CDC term defined as households that have both wireline and wireless phones but receive most or all calls on their wireless phones The A household share reported by CDC is adjusted to reflect share of telephone households. 12 -mostly households include both switched and other than switched households. Therefore, when adding wireless-mostly households to other measures of competitive share, it is necessary to add only the portion attributable to the. Otherwise, it would double count the portion that is attributable to alternatives, which are already included in the competitive share measure. The best approach is to allocate wireless-mostly households in proportion to switched share of landline households in each state. Note: it is possible to do this calculation to include VoIP, but such calculations are not included here. A-1

The range statistics are ordered differently so that the top rows reflect the greatest shift away from traditional landline voice service. Thus, the first, fourth, fifth, sixth, and seventh columns showing gains by alternatives to switched service are ordered from high-tolow. The second, third and eighth columns showing landline and households are presented from low-to-high. For the wireless-only and other than switched columns, the 20 th percentile can be especially insightful if it represents a significant share of households. This would indicate that most states the four-fifths above that level have seen significant shifts to competitive alternatives. For example, in Table A1 the 20 th percentile for wireless-only is 31 percent, which indicates that the bottom one-fifth of households had 31 percent or fewer wireless-only homes, down to the low of 18 percent. But, the other four-fifths of households had 31 percent or more households with wireless-only, up to the high of 51 percent. Similarly, for the switched columns, if the 80 th percentile is a low percentage, it would indicate that four-fifths of states have lower switched shares. Thus, in 2011, fourth-fifths of states had switched household shares of 41 percent or less, down to the low of 28 percent. The analysis covers only the years 2008 to 2011 due to certain data limitations. In particular, CDC provides a time series of state-level data reflecting mid-year 2007 to mid-year 2011 for wireless-only households. 13 The FCC has published actual data through year-end 2012. The FCC provides state-level local telephone competition data going back to the late 1990s through year-end 2012; however, before year-end 2008, the FCC data do not contain the breakdowns necessary for this analysis. Therefore it is necessary to limit the analysis to 2008 on one end and 2011 on the other. 13 For other figures, such as wireless-mostly and no-telephone households, data are available only for the most recent period covered in the CDC releases. As a result, data reflecting wireless-mostly and no-telephone households for mid-year 2011 and year-end 2009 were used as proxies for year-end 2011 and year-end 2008, respectively. There is not much volatility in wireless-mostly and no-telephone households over time, so the results will not be skewed. A-2

Table A1: Estimated State Competitive Voice Shares (Percent of U.S. Telephone Households, Year-End 2011) State Only Other Than Non- VoIP but Mostly (Mid 2011) Less Allocated Mostly AL 39% 61% 41% 20% 18% 2% 13% 32% AK 36% 64% n/a n/a n/a n/a n/a n/a AZ 43% 57% 29% 27% 27% 0% 15% 22% AR 50% 50% 34% 16% 14% 2% 15% 23% CA 33% 67% 42% 25% 21% 4% 18% 30% CO 44% 56% 34% 22% 22% 0% 17% 24% CT 21% 79% 36% 42% 36% 6% 15% 29% DE 31% 69% 34% 35% 28% 7% 17% 25% DC 51% 49% 30% 19% 17% 2% 15% 21% FL 39% 61% 32% 29% 24% 4% 16% 24% GA 40% 60% 37% 23% 20% 3% 20% 25% HI 29% 71% 51% 20% 20% 0% 14% 41% ID 50% 50% 38% 12% 12% 0% 13% 28% IL 38% 62% 36% 27% 23% 4% 17% 26% IN 37% 63% 40% 23% 20% 3% 15% 30% IA n/a n/a n/a n/a n/a n/a n/a n/a KS 43% 57% 33% 25% 22% 2% 14% 25% KY 37% 63% 42% 21% 20% 1% 16% 31% LA 38% 62% 39% 23% 21% 2% 15% 30% ME 36% 64% 42% 21% 21% 0% 11% 35% MD 31% 69% 38% 31% 22% 9% 18% 28% MA 25% 75% 32% 44% 38% 6% 17% 24% MI 41% 59% 28% 31% 27% 4% 14% 21% MN 37% 63% 40% 24% 24% 0% 17% 29% MS 48% 52% 37% 14% 13% 1% 12% 29% MO 38% 62% 44% 18% 13% 4% 16% 33% MT n/a n/a n/a n/a n/a n/a n/a n/a NE 44% 56% 33% 23% 23% 0% 14% 25% NV 41% 59% 33% 26% 25% 1% 19% 23% NH 28% 72% 32% 40% 40% 0% 14% 26% NJ 19% 81% 33% 48% 41% 7% 23% 24% NM 43% 57% 46% 12% 12% 0% 14% 35% NY 23% 77% 34% 43% 39% 4% 17% 26% NC 38% 62% 39% 22% 21% 1% 12% 32% ND 46% 54% 39% 15% 15% 0% 9% 33% OH 39% 61% 38% 23% 21% 2% 17% 28% OK 40% 60% 34% 27% 25% 2% 20% 22% OR 44% 56% 34% 22% 22% 0% 13% 26% PA 28% 72% 44% 29% 25% 4% 16% 34% RI 18% 82% 33% 50% 41% 9% 19% 25% SC 42% 58% 39% 19% 17% 2% 15% 29% SD n/a n/a n/a n/a n/a n/a n/a n/a TN 41% 59% 37% 22% 20% 2% 16% 27% TX 45% 55% 35% 19% 14% 5% 18% 24% UT 43% 57% 35% 22% 22% 0% 14% 26% VT 35% 65% 44% 21% 21% 0% 10% 38% VA 31% 69% 41% 28% 22% 6% 21% 29% WA 38% 62% 34% 28% 28% 0% 19% 24% WV 30% 70% 49% 21% 21% 0% 11% 41% WI 38% 62% 38% 24% 21% 3% 11% 31% WY n/a n/a n/a n/a n/a n/a n/a n/a Ranges and Averages High (Low) 51% 82% 28% 50% 41% 9% 23% 21% 20th (80th) Percentile 43% 69% 33% 29% 27% 4% 18% 24% Median 38% 62% 37% 23% 22% 2% 15% 27% 80th (20th) Percentile 31% 57% 41% 20% 18% 0% 13% 32% Low (High) 18% 49% 51% 12% 12% 0% 9% 41% Source: Federal Communications Commission, Centers for Disease Control, Census and USTelecom Analysis. Percentages rounded. Year-end 2011 is straight-line projection from mid-year 2011. Ranges are high-to-low except for second, third and eighth columns, which are low-to-high. A-3

Table A2: Estimated State Competitive Voice Shares (Percent of U.S. Telephone Households, Mid-Year 2011) State Only Other Than Non- VoIP but Mostly Less Allocated Mostly AL 37% 63% 44% 19% 18% 2% 13% 35% AK 32% 68% n/a n/a n/a n/a n/a n/a AZ 41% 59% 31% 28% 28% 0% 17% 22% AR 48% 52% 40% 12% 11% 2% 17% 27% CA 30% 70% 46% 24% 21% 3% 20% 33% CO 41% 59% 38% 21% 21% 0% 19% 26% CT 20% 80% 40% 41% 35% 5% 17% 31% DE 29% 71% 39% 33% 27% 5% 19% 28% DC 48% 52% 34% 18% 17% 1% 16% 24% FL 37% 63% 36% 27% 24% 3% 18% 26% GA 37% 63% 41% 22% 19% 2% 22% 27% HI 28% 72% 53% 19% 19% 0% 16% 42% ID 48% 52% 40% 12% 12% 0% 14% 29% IL 35% 65% 39% 25% 22% 4% 18% 28% IN 35% 65% 38% 27% 23% 4% 16% 28% IA n/a n/a n/a n/a n/a n/a n/a n/a KS 41% 59% 35% 25% 22% 2% 16% 26% KY 36% 64% 45% 19% 19% 1% 17% 32% LA 36% 64% 41% 23% 22% 2% 17% 30% ME 34% 66% 45% 21% 21% 0% 12% 37% MD 28% 72% 44% 27% 21% 6% 20% 32% MA 23% 77% 36% 41% 37% 4% 19% 27% MI 38% 62% 31% 31% 27% 4% 16% 23% MN 34% 66% 45% 20% 20% 0% 19% 32% MS 46% 54% 41% 14% 13% 1% 14% 31% MO 35% 65% 48% 17% 13% 4% 18% 35% MT n/a n/a n/a n/a n/a n/a n/a n/a NE 41% 59% 35% 24% 24% 0% 16% 26% NV 37% 63% 36% 26% 25% 1% 21% 24% NH 25% 75% 35% 40% 40% 0% 15% 28% NJ 18% 82% 38% 44% 39% 5% 25% 26% NM 39% 61% 50% 11% 11% 0% 15% 37% NY 21% 79% 37% 42% 39% 3% 18% 28% NC 35% 65% 43% 22% 21% 1% 13% 34% ND 44% 56% 40% 15% 15% 0% 10% 33% OH 36% 64% 41% 23% 21% 2% 19% 29% OK 37% 63% 36% 27% 26% 1% 23% 23% OR 41% 59% 37% 22% 22% 0% 15% 28% PA 25% 75% 48% 27% 24% 3% 17% 37% RI 16% 84% 36% 48% 43% 5% 21% 27% SC 39% 61% 43% 19% 17% 1% 16% 31% SD n/a n/a n/a n/a n/a n/a n/a n/a TN 38% 62% 41% 21% 19% 2% 18% 29% TX 43% 57% 39% 18% 14% 4% 20% 26% UT 40% 60% 40% 21% 21% 0% 15% 30% VT 31% 69% 49% 21% 21% 0% 11% 41% VA 29% 71% 46% 25% 22% 3% 23% 31% WA 36% 64% 37% 27% 27% 0% 21% 25% WV 28% 72% 52% 20% 20% 0% 13% 43% WI 36% 64% 40% 24% 21% 3% 13% 32% WY n/a n/a n/a n/a n/a n/a n/a n/a Ranges and Averages High (Low) 48% 84% 31% 48% 43% 6% 25% 22% 20th (80th) Percentile 41% 72% 36% 27% 27% 4% 20% 26% Median 36% 64% 40% 23% 21% 2% 17% 29% 80th (20th) Percentile 28% 59% 45% 19% 18% 0% 15% 33% Low (High) 16% 52% 53% 11% 11% 0% 10% 43% Source: Federal Communications Commission, Centers for Disease Control, Census and USTelecom Analysis. Percentages rounded. Ranges are high-to-low except for second, third and eighth columns, which are low-to-high. A-4

Table A3: Estimated State Competitive Voice Shares (Percent of U.S. Telephone Households, Year-End 2008) State Only Other Than Non- VoIP but Mostly Less Allocated Mostly AL 23% 77% 64% 13% 13% 0% 14% 52% AK 20% 80% n/a n/a n/a n/a n/a n/a AZ 27% 73% 44% 29% 29% 0% 19% 32% AR 34% 66% 57% 9% 9% 0% 16% 43% CA 16% 84% 66% 18% 17% 0% 21% 49% CO 30% 70% 53% 17% 17% 0% 16% 41% CT 11% 89% 60% 29% 28% 2% 15% 50% DE 15% 85% 60% 25% 25% 0% 18% 47% DC 25% 75% 58% 17% 17% 0% 17% 45% FL 25% 75% 56% 18% 18% 0% 17% 44% GA 24% 76% 59% 17% 17% 0% 20% 44% HI 17% 83% 66% 16% 16% 0% 18% 52% ID 30% 70% 61% 9% 9% 0% 15% 47% IL 23% 77% 58% 19% 19% 1% 18% 45% IN 24% 76% 62% 14% 13% 1% 15% 50% IA 28% 72% 57% 14% 14% 0% 17% 44% KS 27% 73% 49% 24% 24% 1% 13% 40% KY 32% 68% 50% 19% 19% 0% 14% 39% LA 23% 77% 60% 17% 17% 0% 17% 47% ME 23% 77% 60% 18% 17% 0% 12% 51% MD 16% 84% 63% 21% 21% 0% 22% 47% MA 15% 85% 54% 32% 32% 0% 15% 44% MI 26% 74% 48% 26% 25% 1% 16% 38% MN 25% 75% 58% 17% 17% 0% 16% 46% MS 34% 66% 57% 9% 9% 0% 15% 44% MO 20% 80% 69% 12% 12% 0% 16% 55% MT 19% 81% 64% 17% 17% 0% 14% 53% NE 31% 69% 46% 23% 23% 0% 17% 35% NV 20% 80% 57% 22% 22% 0% 15% 47% NH 14% 86% 56% 29% 29% 0% 14% 48% NJ 11% 89% 57% 32% 32% 0% 22% 43% NM 27% 73% 66% 7% 7% 0% 12% 55% NY 15% 85% 50% 34% 34% 0% 14% 42% NC 24% 76% 59% 17% 17% 0% 16% 46% ND 30% 70% 47% 22% 22% 0% 10% 41% OH 23% 77% 56% 21% 20% 0% 18% 44% OK 28% 72% 49% 23% 22% 0% 17% 37% OR 27% 73% 56% 17% 17% 0% 15% 45% PA 15% 85% 66% 18% 18% 0% 17% 53% RI 10% 90% 44% 45% 45% 0% 17% 36% SC 24% 76% 61% 15% 15% 0% 19% 46% SD 14% 86% 53% 33% 33% 0% 9% 47% TN 27% 73% 59% 14% 14% 0% 17% 45% TX 30% 70% 56% 13% 13% 0% 21% 40% UT 22% 78% 60% 18% 18% 0% 14% 49% VT 19% 81% 68% 13% 13% 0% 16% 55% VA 22% 78% 57% 21% 21% 0% 18% 44% WA 23% 77% 56% 21% 21% 0% 17% 43% WV 18% 82% 71% 11% 11% 0% 14% 59% WI 22% 78% 58% 20% 20% 1% 11% 50% WY 22% 78% 57% 21% 21% 0% 13% 47% Ranges and Averages High (Low) 34% 90% 44% 45% 45% 2% 22% 32% 20th (80th) Percentile 27% 84% 53% 24% 24% 0% 18% 42% Median 23% 77% 58% 18% 18% 0% 16% 45% 80th (20th) Percentile 16% 73% 63% 14% 14% 0% 14% 50% Low (High) 10% 66% 71% 7% 7% 0% 9% 59% Source: Federal Communications Commission, Centers for Disease Control, Census and USTelecom Analysis. Percentages rounded. Ranges are high-to-low except for second, third and eighth columns, which are low-to-high. A-5

Appendix B - Technical Notes Comparison and Consistency with Previous Analyses USTelecom has issued a series of national voice competition analyses, the most recent was released in a November 22, 2013 Research Brief, which updated an April 3, 2013 Research Brief. USTelecom also issued a state voice competition analysis in a May 30, 2013 Research Brief. This updated state analysis is intended to achieve maximum methodological consistency with the most recent national analysis and to minimize skews arising from the use of several data sources. It is also updated to include historical data for 2008 to show the trend through 2011. See the November 22, 2013 Research Brief for a full discussion of the national methodology. This analysis makes methodological adjustments from previous state analyses in order to fine tune the results and minimize distortions arising from the use of several data sources. In particular, the analysis relies upon household data from Census, share of households from CDC for no-phone, wireless-only, and landline households, and line count data from FCC for allocating landlines among s and Non-s, and switched, and VoIP service. All of these data sets are subject to certain margins of error, which make inconsistencies inevitable. The May 30, 2013 state analysis started by taking the share of households that were either wireless-only or used a landline from CDC. It then added actual FCC line counts for landlines other than switched, including non- switched, non- VoIP, and VoIP, and after backing out second lines, divided by the number of Census households to calculate the percentage of households. The remaining percentage of households was then assumed to reflect switched primary line households. This approach, referred to herein as the residual method, lent itself to minor distortions in a handful of cases in which using the literal reported FCC lines adjusted for second lines would result in telephone households greater than or less than 100 percent of telephone households reported by Census, given the share allocated to wireless-only by CDC. The methodology effectively addressed the inconsistency by assuming that switched lines were the residual. This meant that when actual FCC line counts implied greater than 100 percent of households, the switched share was truncated to bring the total down to 100 percent; and when actual FCC line counts implied less than 100 percent of households, switched share was effectively augmented to bring the total up to 100 percent. In the former case, switched share is skewed down; in the latter case, switched share is skewed up relative to its proportionate state share within the FCC data. In both cases, landlines other than switched were taken as given by the FCC and the skew in terms of share was the inverse of the skew for switched. In order to correct for the potential distortions described above, the new analysis employs what will be referred to as the proportionate method. The key adjustments are described below. The starting point remains the CDC for wireless-only and landline households; however, the new analysis allocates landline households among s and non-s in proportion to the FCC data for each state. In this way, to the extent actual FCC line B-1

counts do not align perfectly with Census and CDC household data, the difference is more evenly spread among s and non-s, minimizing the distortion of household share. Table B1 describes the results, state-by-state, using the residual and proportionate methods and the difference. The data in the table make two points. First, the change in methodology did not make a big difference for most states: 40 of 46 states for which data were available shifted less than three percentage points, plus or minus. Second, the results under the new proportionate methodology are on balance more conservative with respect to shares than the old methodology. Twenty-four of the 46 states saw an increase in switched share under the new proportionate methodology; 22 saw a lower switched share. Of the states that saw large shifts, greater than three percentage points, five showed more conservative results under the new methodology, i.e. an increase in the switched share percentage points (Delaware +4.6, Maine +3.3 percent, New Hampshire +7.2, New Jersey +4.3, and Vermont +6.9). Only one state showed a decline in switched share percentage points (Rhode Island -7.0). In addition, the previous state analysis backed out second lines using the same assumptions used in the previous (April 3, 2013) national analysis: approximately 10 percent for s and 6 percent non-s. In the most recent national analysis (November 22, 2013), USTelecom used a different approach, allocating second lines in proportion to and non- landline shares. This state analysis effectively does the same by allocating remaining landline households in proportion to and non- state shares. This approach effectively allows for more variation among states in second line adoption, rather than applying a rigid national assumption. One minor difference between this state analysis and the latest national analysis is the state analysis cannot distinguish between non- switched telephony provided by cable and non-cable providers due to data limitations. In the national analysis, non-, non-cable switched telephony providers, are assumed to resell wholesale services and are included in the switched category. Unfortunately, this skews estimates of competitive share up slightly. Such providers, however, account for only 1.5 percent of national telephone households at the end of 2011, and this percentage is declining. Therefore, the impact on the results is likely very small. Of course, resellers are competitors, just not facilities-based competitors, which has been the focus of the USTelecom voice competition analyses. On the other hand, the CDC state data are reported somewhat differently than the CDC national data, and may understate competitive share. The CDC state data report the percentage of adults living in wireless-only households, which is slightly different that the percent of households that are wireless-only. In the CDC national data, they report both: wireless-only households during the second half of 2011 were 34 percent and the percent of adults living in those households was 32.3 percent. Thus, there is a fair chance that this state analysis understates wireless-only share of telephone households since the only available state data from CDC are based on percentage of adults, not percentage of households. At the national level, the difference is a little over 1.5 percent. This understatement is roughly on par with the overstatement for noncable non- switched share of telephone households, though there is no guarantee that the two factors would balance in all states, if data were available. B-2

Table B1 Impact of Proportionate vs. Residula Method on Shares (Percenage of households) Old Residual Method New Proportionate Method Difference (New Minus Old) State Other Than Non- VoIP Other Than Non- VoIP Other Than Non- AL 41% 20% 18% 2% 41% 20% 18% 2% 0% 0% 0% 0% AK n/a n/a n/a 0% n/a n/a n/a n/a n/a n/a n/a n/a AZ 29% 27% 27% 0% 29% 27% 27% 0% 0% 0% 0% 0% AR 32% 18% 16% 2% 34% 16% 14% 2% 2% -2% -2% 0% CA 41% 26% 22% 4% 42% 25% 21% 4% 1% -1% -1% 0% CO 32% 24% 24% 0% 34% 22% 22% 0% 2% -2% -2% 0% CT 33% 45% 39% 6% 36% 42% 36% 6% 3% -3% -3% 0% DE 29% 39% 32% 8% 34% 35% 28% 7% 5% -5% -4% -1% DC 27% 22% 20% 2% 30% 19% 17% 2% 3% -3% -3% 0% FL 31% 30% 25% 4% 32% 29% 24% 4% 1% -1% -1% 0% GA 38% 22% 19% 3% 37% 23% 20% 3% -1% 1% 1% 0% HI 53% 18% 18% 0% 51% 20% 20% 0% -2% 2% 2% 0% ID 37% 13% 13% 0% 38% 12% 12% 0% 1% -1% -1% 0% IL 36% 26% 22% 4% 36% 27% 23% 4% -1% 1% 0% 0% IN 41% 22% 19% 3% 40% 23% 20% 3% -1% 1% 1% 0% IA n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a KS 35% 23% 21% 2% 33% 25% 22% 2% -2% 2% 2% 0% KY 43% 19% 19% 1% 42% 21% 20% 1% -1% 1% 1% 0% LA 41% 22% 20% 2% 39% 23% 21% 2% -1% 1% 1% 0% ME 39% 25% 25% 0% 42% 21% 21% 0% 3% -3% -3% 0% MD 36% 33% 23% 10% 38% 31% 22% 9% 2% -2% -2% -1% MA 29% 46% 40% 6% 32% 44% 38% 6% 2% -2% -2% 0% MI 26% 33% 29% 4% 28% 31% 27% 4% 2% -2% -2% 0% MN 38% 25% 25% 0% 40% 24% 24% 0% 2% -2% -2% 0% MS 38% 14% 12% 1% 37% 14% 13% 1% -1% 1% 1% 0% MO 46% 16% 12% 4% 44% 18% 13% 4% -2% 2% 1% 0% MT n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a NE 32% 24% 24% 0% 33% 23% 23% 0% 1% -1% -1% 0% NV 32% 28% 27% 1% 33% 26% 25% 1% 2% -2% -2% 0% NH 24% 47% 47% 0% 32% 40% 40% 0% 7% -7% -7% 0% NJ 29% 52% 44% 8% 33% 48% 41% 7% 4% -4% -4% -1% NM 46% 11% 11% 0% 46% 12% 12% 0% 0% 0% 0% 0% NY 36% 41% 37% 4% 34% 43% 39% 4% -2% 2% 2% 0% NC 39% 23% 21% 1% 39% 22% 21% 1% 0% 0% 0% 0% ND 38% 17% 17% 0% 39% 15% 15% 0% 2% -2% -2% 0% OH 38% 23% 20% 2% 38% 23% 21% 2% 0% 0% 0% 0% OK 34% 27% 25% 2% 34% 27% 25% 2% 0% 0% 0% 0% OR 32% 24% 24% 0% 34% 22% 22% 0% 2% -2% -2% 0% PA 44% 28% 25% 4% 44% 29% 25% 4% 0% 0% 0% 0% RI 40% 43% 35% 8% 33% 50% 41% 9% -7% 7% 6% 1% SC 39% 20% 18% 2% 39% 19% 17% 2% 1% -1% -1% 0% SD n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a TN 37% 22% 20% 2% 37% 22% 20% 2% 0% 0% 0% 0% TX 36% 18% 14% 4% 35% 19% 14% 5% -1% 1% 1% 0% UT 35% 22% 22% 0% 35% 22% 22% 0% 0% 0% 0% 0% VT 37% 28% 28% 0% 44% 21% 21% 0% 7% -7% -7% 0% VA 42% 28% 22% 6% 41% 28% 22% 6% 0% 0% 0% 0% WA 32% 30% 30% 0% 34% 28% 28% 0% 2% -2% -2% 0% WV 50% 20% 20% 0% 49% 21% 21% 0% -1% 1% 1% 0% WI 38% 24% 21% 3% 38% 24% 21% 3% 0% 0% 0% 0% WY n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a Source: Federal Communications Commission, Centers for Disease Control, Census and USTelecom Analysis. Percentages rounded. VoIP B-3

State Estimates for Year-End 2011; No State-by-State Projections to 2013 The state analysis attempts to normalize all data to year-end 2008 and year-end 2011. Actual FCC data are available for year-end 2008 and year-end 2011. The Census data are based on a March survey; year-end 2011 data are derived by straight-line quarterly estimates from March 2011 to March 2012. CDC state-level data reflect overlapping 12-month periods, from January to December, and July to June. July 2008 to June 2009 data are assumed to reflect yearend 2008. The most recent data for January to December 2011 are assumed to reflect mid-year 2011; year-end figures for 2011 are derived using straight-line growth from the prior period. This method generated year-end estimates that are consistent with CDC s national data, which are released semi-annually. The CDC semi-annual national data are assumed to reflect a mid-point for each half of the year and therefore require adjustments to year-end. USTelecom s national household voice share analysis includes projections, based on straight-line methods, for year-end 2012 and year-end 2013. This state-level analysis does not contain state-by-state share projections, due to the complexity of the exercise and the greater potential for error in making more granular projections. Nonetheless, given the available state data, which show increasing levels of competition, it is very likely that all or most states are following the national trend of declining shares. For example, for the three periods reported in the latest CDC data, there was an increase in the share of wireless-only households in every state for which data are available. Using FCC data from year-end 2011 to year-end 2012, s lost lines in all states. This is true even if VoIP gains are netted against switched line losses. Mid-Year 2011 Data for Mostly and No-Phone Households As noted above, the CDC data are based on overlapping twelve-month periods, which presumably reflect the midpoint of each period. For wireless-only households, it is possible to estimate year-end figures using straight-line methods with the existing data. There is no time series data reported for wireless-mostly households and no-phone households. Therefore, the data for those two categories is based on mid-year 2011. Applying these shares in the year-end 2011 analysis is a crude estimation, but not likely problematic. Unlike wireless-only households, at the national level the share of households that are wireless-mostly has not shown consistent upward or downward trends recently. Impact of VoIP Categorization The analysis above states that inclusion of VoIP as an alternative to switched service does not affect the broad observation that s are no longer dominant in the provision of voice communications. At the national level, VoIP accounted for approximately three percent of telephone households at the end of 2011, although its share relative to switched service is expected to grow over time and will account for an increasing share. At the state level, VoIP shares fall in a larger range, from zero to almost ten percentage points at the end of 2011. Therefore, it is appropriate to determine whether inclusion of VoIP affects the conclusion in states where it commands the largest shares. B-4

An analysis of Table A1 with year-end 2011 estimates indicates that the categorization of VoIP does not have a material impact on the overall conclusion. First, the 80 percentile for VoIP share of households was 4.3 percent, meaning 80 percent of states with data available showed VoIP share less than 4.3 percent. Second, for the 10 states Connecticut, Delaware, Maryland, Massachusetts, Montana, New Jersey, New York, Rhode Island, Texas and Virginia where VoIP share was greater than 4.3 percent, other competitive alternatives to s also had very high shares, such that total share including VoIP remained less than 50 percent ranging from 37 percent to 48 percent. Only three states had combined switched and VoIP share greater than 42 percent: Virginia and Maryland with 47 percent and Montana with 48 percent. Again, this is an estimate for year-end 2011 and shares have declined since then. Not surprisingly, high--voip states were states where non- landline competition was strong, with all but two of the 10 states (Texas and Montana) also being ranked better than average (median) in non- switched and VoIP competition. Somewhat more surprising, most of these states were not strong in wireless-only shares, with only the same two (Texas and Montana) greater than average for wireless-only households. But these states that were ranked lower in wireless-only households are dense Eastern states where, as the CDC has noted, wireless cord-cutting has not been historically as great as the rest of the country. Meanwhile wireless-mostly rankings in these states are typically greater than average. States Excluded Due to Data Issues This new state analysis excludes five states because data were not available from either the FCC (Alaska) or CDC (Iowa, Montana, South Dakota and Wyoming). These states collectively represent only 2.1 percent of U.S. households. Therefore excluding them from the analysis is not likely to significantly skew the results. However, since they likely represent a larger share of rural households, it is worth exploring how these states measure up to the rest of the states. For the states missing in the CDC data, CDC does provide estimates, however, the standard errors are large, so it is inappropriate to include in the analysis with the other states. Table B1 below presents the CDC estimates for these states, which are available only for the period January to December 2011, presumably reflecting mid-year 2011. Table B2: Estimated Voice Shares for CDC Missing States (Percent of Telephone Households, Mid-Year 2011) State Other Than Non- ( and VoIP) VoIP - Only - Mostly - Only Standard Error - Mostly Standard Error IA 32% 68% 51% 17% 17% ~0% 16% 5.9% 3.8% MT 33% 67% 49% 19% 19% ~0% 18% 6.2% 4.1% SD 33% 67% 24% 43% 43% ~0% 17% 6.1% 3.9% WY 34% 66% 45% 21% 20% 0.4% 18% 6.3% 4.1% Source: Federal Communications Commission, Centers for Disease Control, Census and USTelecom Analysis. Percentages rounded. B-5

The data and estimates in Table B2 reflect mid-year 2011. Therefore comparisons with year-end data in Table 1 and Table A1 above are likely to be complicated since the dates are out of sync. Table A2 above provides a better comparison since it shows the same analysis as Table 1, but for mid-year 2011. Assuming the CDC estimates for these states are accurate, the table shows that these states are within the range of the other states analyzed. With the exception of South Dakota, which ranks very highly, they would be near the lower end of the range. B-6