This is the accessible text file for GAO report number GAO-06-426 
entitled 'Telecommunications: Broadband Deployment Is Extensive 
throughout the United States, but It Is Difficult to Assess the Extent 
of Deployment Gaps in Rural Areas' which was released on May 5, 2006.

This text file was formatted by the U.S. Government Accountability 
Office (GAO) to be accessible to users with visual impairments, as part 
of a longer term project to improve GAO products' accessibility. Every 
attempt has been made to maintain the structural and data integrity of 
the original printed product. Accessibility features, such as text 
descriptions of tables, consecutively numbered footnotes placed at the 
end of the file, and the text of agency comment letters, are provided 
but may not exactly duplicate the presentation or format of the printed 
version. The portable document format (PDF) file is an exact electronic 
replica of the printed version. We welcome your feedback. Please E-mail 
your comments regarding the contents or accessibility features of this 
document to Webmaster@gao.gov.

This is a work of the U.S. government and is not subject to copyright 
protection in the United States. It may be reproduced and distributed 
in its entirety without further permission from GAO. Because this work 
may contain copyrighted images or other material, permission from the 
copyright holder may be necessary if you wish to reproduce this 
material separately. 

United States Government Accountability Office: 

GAO:

Report to Congressional Committees:

May 2006:

Telecommunications:

Broadband Deployment Is Extensive throughout the United States, but It 
Is Difficult to Assess the Extent of Deployment Gaps in Rural Areas:

GAO-06-426:

GAO Highlights: 

Highlights of GAO-06-426, a report to congressional committees.

Why GAO Did This Study: 

Both Congress and the President have indicated that access to broadband 
for all Americans is critically important.  Broadband is seen as a 
critical economic engine, a vehicle for enhanced learning and medicine, 
and a central component of 21st century news and entertainment.  As 
part of our response to a mandate included in the Internet Tax 
Nondiscrimination Act of 2004, this report examines the factors that 
affect the deployment and the adoption of broadband services.  In 
particular, this report provides information on (1) the current status 
of broadband deployment and adoption; (2) the factors that influence 
the deployment of broadband networks; (3) the factors that influence 
the adoption, or purchase, of broadband service by households; and (4) 
the options that have been suggested to spur greater broadband 
deployment and adoption.

What GAO Found: 

About 30 million American households have adopted broadband service, 
but the Federal Communications Commission’s (FCC) data indicating the 
availability of broadband networks has some weaknesses.  FCC conducts 
an extensive data collection effort using its Form 477 to assess the 
status of advanced telecommunications service in the United States.  
For its zip-code level data, FCC collects data based on where 
subscribers are served, not where providers have deployed broadband 
infrastructure.  Although it is clear that the deployment of broadband 
networks is extensive, the data may not provide a highly accurate 
depiction of local deployment of broadband infrastructures for 
residential service, especially in rural areas.

A variety of market and technical factors, government efforts, and 
access to resources at the local level have influenced the deployment 
of broadband infrastructure.  Areas with low population density and 
rugged terrain, as well as areas removed from cities, are generally 
more costly to serve than are densely populated areas and areas with 
flat terrain.  As such, deployment tends to be less developed in more 
rural parts of the country.  Technical factors can also affect 
deployment. GAO also found that a variety of federal and state efforts, 
and access to resources at the local level, have influenced the 
deployment of broadband infrastructure.

A variety of characteristics related to households and services 
influence whether consumers adopt broadband service.  GAO found that 
consumers with high incomes and college degrees are significantly more 
likely to adopt broadband.  The price of broadband service remains a 
barrier to adoption for some consumers, although prices have been 
declining recently.  The availability of applications and services that 
function much more effectively with broadband, such as computer gaming 
and file sharing, also influences whether consumers purchase broadband 
service.  

Stakeholders identified several options to address the lack of 
broadband in certain areas.  Although the deployment of broadband is 
widespread, some areas are not served, and it can be costly to serve 
highly rural areas.  Targeted assistance might help facilitate 
broadband deployment in these areas.  GAO found that stakeholders have 
some concerns about the structure of the Rural Utilities Service’s 
broadband loan program.  GAO was also told that modifications to 
spectrum management might address the lack of broadband infrastructure 
in rural areas.  Also, because the cost of building land-based 
infrastructure is so high in some rural areas, satellite industry 
stakeholders noted that satellite broadband technology may be the best 
for addressing a lack of broadband in those regions.  While several 
options such as these were suggested to GAO, each has some challenges 
to implementation.  Also, a key difficulty for analyzing and targeting 
federal aid for broadband is a lack of reliable data on the deployment 
of networks.  

What GAO Recommends:

GAO recommends that FCC develop information regarding the cost and 
burden that would be associated with various options for improving the 
information available on broadband deployment and report this 
information to the relevant Senate and House committees to help them 
determine what actions, if any, are necessary.  FCC provided technical 
comments on this report, but did not comment on this recommendation.  

[Hyperlink, http://www.gao.gov/cgi-bin/getrpt?GAO-06-426].

To view the full product, including the scope and methodology, click on 
the link above. For more information, contact JayEtta Z. Hecker at 
(202) 512-2834 or heckerj@gao.gov. 

[End of section]

Contents:

Letter:

Results in Brief:

Background:

About 30 Million American Households Purchase Broadband Service; 
Despite Evidence of Substantial Broadband Deployment throughout the 
United States, It Is Difficult to Assess Deployment Gaps in Some Areas:

A Variety of Market and Technical Factors, in Addition to Government 
Involvement and Access to Resources at the Local Level, Have Influenced 
the Deployment of Broadband:

A Variety of Household and Service Characteristics Influence the 
Adoption of Broadband:

Stakeholders Identified Several Options to Address the Lack of 
Broadband in Certain Areas, but Challenges Exist with Implementation:

Conclusion:

Recommendation for Executive Action:

Agency Comments and Our Evaluation:

Appendixes:

Appendix I: Scope and Methodology:

Appendix II: Data Reliability:

Appendix III: Broadband Deployment and Adoption Models:

Design of Our Broadband Deployment and Adoption Models:

Data Sources:

Assessing Broadband Deployment:

Estimation Methodology and Results:

Appendix IV: Additional Communications Technologies:

Appendix V: Comments from Industry Participants:

Appendix VI: GAO Contact and Staff Acknowledgments:

Tables:

Table 1: Deployment Model: Definitions and Sources of Variables:

Table 2: Adoption Model: Definitions and Sources of Variables:

Table 3: Deployment Model: Descriptive Statistics:

Table 4: Deployment Model: Estimation Results:

Table 5: Adoption Model: Descriptive Statistics:

Table 6: Adoption Model: Estimation Results:

Figures:

Figure 1: Status of Household Computer Ownership and Internet 
Connection:

Figure 2: Household Online Connection:

Figure 3: Percentage of Households Subscribing to Broadband, by Type of 
Location:

Figure 4: Factors Influencing Subscription to Broadband:

Abbreviations:

3G: third generation:

ADSL: asymmetric digital subscriber line:

BPL: broadband over power lines:

CTIA: Cellular Telecommunications and Internet Association:

DBS: direct broadcast satellite:

DSL: digital subscriber line:

FCC: Federal Communications Commission:

FTTH: fiber to the home:

HFC: hybrid-fiber coaxial:

IEEE: Institute of Electrical and Electronics Engineers:

IG: inspector general:

ILEC: incumbent local exchange carrier:

Kbps: kilobits per second:

Mbps: megabits per second:

MSA: metropolitan statistical area:

NARUC: National Association of Regulatory Utility Commissioners:

NATOA: National Association of Telecommunications Officers and Advisors:

NCTA: National Cable and Telecommunications Association:

NTCA: National Telecommunications Cooperative Association:

NTIA: National Telecommunications and Information Administration:

RUS: Rural Utilities Service:

SIA: Satellite Industry Association:

UNE: unbundled network element:

USF: Universal Service Fund:

USIIA: US Internet Industry Association:

USTA: United States Telecom Association:

VoIP: voice over Internet protocol:

Wi-Fi: wireless fidelity:

WiMAX: Worldwide Interoperability for Microwave Access:

WISP: wireless Internet service provider:

WISPA: wireless Internet Service Providers Association:

United States Government Accountability Office:
Washington, DC 20548:

May 5, 2006:

The Honorable Ted Stevens: 
Chairman: 
The Honorable Daniel K. Inouye: 
Co- Chairman: 
Committee on Commerce, Science, and Transportation: 
United States Senate:

The Honorable Joe L. Barton: 
Chairman: 
The Honorable John D. Dingell: 
Ranking Minority Member: 
Committee on Energy and Commerce: 
House of Representatives:

The universal availability of high speed Internet access over broadband 
technologies--commonly referred to as broadband Internet access--has 
become a national goal.[Footnote 1] The Telecommunications Act of 1996 
directed the Federal Communications Commission (FCC) and state 
commissions to encourage the deployment of advanced telecommunications 
capability. Similarly, in 2004, the President stated that there should 
be a national goal for universal, affordable access to broadband 
technology by 2007. The importance placed on access to broadband 
correlates to its many benefits for individuals and society. Broadband 
is seen as a critical economic engine, a vehicle for enhanced learning 
and medicine, and a central component of 21st century news and 
entertainment.

As part of our response to a mandate included in the Internet Tax 
Nondiscrimination Act of 2004, this report examines the factors that 
affect the deployment--that is, the building of infrastructure over 
which broadband services can be provided--and the adoption of broadband 
services. We focus particularly on the deployment and adoption of 
broadband to households, as opposed to businesses or institutions. In 
particular, this report provides information on (1) the current status 
of broadband deployment and adoption; (2) the factors that influence 
the deployment of broadband networks; (3) the factors that influence 
the adoption, or purchase, of broadband service by households; and (4) 
the options that have been suggested to spur greater broadband 
deployment and adoption. In January 2006, we released a report that 
examined the impact of the Internet tax moratorium on state and local 
tax revenues, as also mandated by the law.[Footnote 2]

To respond to the objectives of this report, we selected eight states 
and conducted case studies on the status of broadband deployment and 
adoption. For each of the states--Alaska, California, Kentucky, 
Massachusetts, North Dakota, Ohio, Texas, and Virginia--we interviewed 
state and local officials, including local franchising authorities, 
state public utility regulators, and representatives from governors' 
offices; state industry and government associations; private cable and 
telephone providers; wireless Internet service providers; and municipal 
and cooperative telecommunications providers. We also spoke with a 
variety of individuals and organizations knowledgeable about broadband 
services, such as national industry associations and experts. We spoke 
with representatives from FCC, the National Telecommunications and 
Information Administration of the Department of Commerce, and the Rural 
Utilities Service (RUS) of the Department of Agriculture. To assess the 
status of broadband deployment and to understand the factors affecting 
the deployment and adoption of broadband, we used survey data from 
Knowledge Networks/SRI's The Home Technology MonitorTM: Spring 2005 
Ownership and Trend Report. Knowledge Networks/SRI interviewed 
approximately 1,500 randomly sampled households, asking questions about 
each household's purchase of Internet services and the availability of 
cable television service. Using these data, we estimated two 
econometric models: One model examined the factors affecting broadband 
deployment and the second examined the factors affecting households' 
adoption of broadband services. We combined the household survey data 
with information from FCC's Form 477 filings, which contain information 
on companies' provision of broadband services by zip codes. This 
enabled us to develop information about what options for broadband a 
particular household would have. To assess the impact of Internet taxes 
on broadband deployment and adoption, we contacted officials in 48 
states and the District of Columbia to determine whether the state, or 
local governments in the state, imposed taxes on Internet access in 
2005; we did not evaluate the level of taxation. We concluded that 
information from Knowledge Networks/SRI and FCC (with modifications 
discussed later in this report) was sufficiently reliable for the 
purpose of this report. All percentage estimates from the Knowledge 
Networks/SRI survey have margins of error of plus or minus 7 percentage 
points or less, unless otherwise noted. See appendix I for a more 
detailed discussion of the overall scope and methodology for this 
report, including a discussion of how we selected the case-study 
states; appendix II for an assessment of the data reliability of the 
Knowledge Networks/SRI survey; and appendix III for a more detailed 
explanation of, and results from, our deployment and adoption models.

We conducted our work from April 2005 through February 2006 in 
accordance with generally accepted government auditing standards.

Results in Brief:

About 30 million American households purchase, or have adopted, 
broadband service, but it is difficult to assess the extent of gaps in 
the availability of broadband in local markets. Using a survey of 
American households, we found that 28 percent--or about 30 million-- 
subscribed to broadband service in 2005. In addition, 30 percent of 
surveyed households subscribed to a dial-up Internet service, and 41 
percent did not access the Internet from their home. Among households 
subscribing to broadband service, we found roughly an equal share 
taking cable modem and digital subscriber line (DSL) service, the two 
primary broadband services at this time. Households in rural areas were 
less likely to subscribe to broadband service, compared with households 
in urban and suburban areas. On a semiannual basis, FCC conducts an 
extensive data collection effort using its Form 477 to assess the 
availability of advanced telecommunications service in the United 
States. As of July 2005, FCC has found that 99 percent of Americans 
live in the 95 percent of zip codes that have at least one broadband 
provider reporting to be serving at least one subscriber. These data 
clearly indicate that deployment of broadband networks has been 
extensive. However, for its zip-code level data, FCC collects data 
based on where subscribers are served, not where providers have 
deployed broadband infrastructure. Based on our analysis is appears 
that these data may not provide a highly accurate depiction of 
deployment of broadband infrastructures for residential service in some 
areas.[Footnote 3]

A variety of market and technical factors, as well as federal and state 
government efforts and access to resources at the local level have 
influenced the deployment of broadband infrastructure. Most 
importantly, companies contemplating the deployment of broadband 
infrastructure consider both the cost to deploy and operate a broadband 
network and the expected demand for broadband service. We found it is 
more costly to serve areas with low population density and rugged 
terrain with terrestrial facilities than it is to serve areas that are 
densely populated and have flat terrain. It also may be more costly to 
serve locations that are a significant distance from a major city. As 
such, these important factors have caused deployment to be less 
developed in more rural parts of the country. Firms also consider the 
extent of existing competition in the broadband market when making 
deployment decisions: New entrants are more likely to enter markets 
with no competitors, but at the same time, we found that incumbent 
cable and telephone companies may respond to entry by new companies by 
rolling out broadband in markets where they had not yet provided 
service. Even when cost and demand factors are favorable, technical 
factors can limit the deployment of broadband service in certain 
contexts. For example, DSL--the primary broadband service provided by 
telephone companies--can generally extend only 3 miles[Footnote 4] from 
the central office with copper plant, which precludes many households 
from obtaining DSL service.[Footnote 5] Finally, we found that a 
variety of federal and state government efforts as well as access to 
resources at the local level have influenced the deployment of 
broadband infrastructure. At the federal level, one of the programs of 
the Universal Service Fund (USF)--known as the High Cost Fund--has 
indirectly facilitated broadband service in more rural areas. 
Similarly, the Department of Agriculture's Rural Utilities Service 
(RUS) provides grants and loans to promote broadband service in rural 
areas. At the local level, access to rights-of-way, pole attachments, 
wireless-tower sites as well as the video franchising process can 
influence the pace of deployment. We also found that strong leadership 
within a community can help promote broadband deployment by, for 
example, enhancing the likely market success of companies' entry into 
rural markets. Finally, using our econometric model, we found that the 
imposition of taxes was not a statistically significant factor 
influencing the deployment of broadband.

A variety of characteristics related to households and services 
influence whether consumers purchase (or adopt) broadband service. 
Based on our econometric model, we found that several characteristics 
of households influence the adoption decision. Our model showed that 
households with high incomes were 39 percentage points more likely to 
adopt broadband than lower-income households, and those with a college- 
educated head of household were 12 percentage points more likely to 
purchase broadband than households headed by someone who did not 
graduate from college. While rural households are less likely to adopt 
broadband, our findings indicate that this difference may be related in 
part to the lower availability of broadband in rural areas. In 
addition, based on discussions with stakeholders, we identified several 
characteristics of broadband service that influence whether a consumer 
purchases the service. The price of broadband service remains a barrier 
to adoption of broadband service for some consumers, although prices 
have been declining recently. The availability of applications and 
services that either require or function much more effectively with 
broadband--such as computer gaming and file sharing--also influences 
whether a particular consumer purchases broadband service. Using our 
model, we found that the imposition of the tax was not a statistically 
significant factor influencing the adoption of broadband service at the 
5 percent level. It was statistically significant at the 10 percent 
level, perhaps suggesting that it is a weakly significant factor. 
However, giving the nature of our model, it is unclear whether this 
finding is related to the tax or other characteristics of the states in 
which the households resided.

Targeted government assistance might help facilitate the deployment of 
broadband service, and stakeholders we spoke with identified several 
options to spur greater deployment of broadband service in rural 
America. However, each of the policy options that stakeholders 
discussed with us had challenges to their implementation. For example, 
a few of the stakeholders we spoke with expressed concerns about the 
structure of the Rural Utilities Service's broadband loan program. 
Also, several of the stakeholders suggested that modifications to 
spectrum management might address the lack of broadband infrastructure 
in rural areas. Finally, because the cost of building land-based 
infrastructure is so high in some rural areas, satellite industry 
stakeholders noted that satellite broadband technology may be the best 
option for addressing a lack of broadband in those regions. Ultimately, 
we found that a key difficulty for analyzing and targeting any federal 
aid for broadband is a lack of reliable data on the deployment of 
networks.

We provided a draft of this report to the Department of Agriculture, 
the Department of Commerce, and FCC for their review and comment. The 
Department of Agriculture had no comments on the draft. The Department 
of Commerce and FCC provided technical comments that we incorporated, 
as appropriate.

In the draft, GAO recommended that FCC identify and evaluate strategies 
for improving the 477 data such that the data provide a more accurate 
depiction of residential broadband deployment throughout the country. 
In oral comments regarding this recommendation, FCC staff noted that 
the commission had recently determined that it would be costly and 
could impose large burdens on filers--particularly small entities--to 
require any more detailed filings on broadband deployment. As such, we 
recommend that FCC develop information regarding the degree of cost and 
burden that would be associated with various options for improving the 
information available on broadband deployment and should provide that 
information to the Senate Committee on Commerce, Science, and 
Transportation and the House Energy and Commerce Committee in order to 
help them determine what actions, if any, are necessary going forward. 
FCC did not comment on our final recommendation.

We also provided a draft of this report to several associations 
representing industry trade groups and state and local government 
entities for their review and comment. Specifically, the following 
associations came to GAO headquarters to review the draft: Cellular 
Telecommunications and Internet Association (CTIA), National 
Association of Regulatory Utility Commissioners (NARUC), National 
Association of Telecommunications Officers and Advisors (NATOA), 
National Cable and Telecommunications Association (NCTA), National 
Telecommunications Cooperative Association (NTCA), Satellite Industry 
Association (SIA), US Internet Industry Association (USIIA), United 
States Telecom Association (USTA), and Wireless Internet Service 
Providers Association (WISPA). Officials from CTIA, NARUC, and NTCA did 
not provide comments. Officials from NATOA, NCTA, SIA, and USIIA 
provided technical comments that were incorporated, as appropriate. 
USTA and WISPA provided comments that are discussed in appendix V.

Background:

Internet access became widely available to residential users by the mid 
1990s. For a few years, the primary mechanism to access the Internet 
was a dial-up connection, in which a standard telephone line is used to 
make an Internet connection. A dial-up connection offers data 
transmission speeds up to 56 kilobits per second (Kbps). Broadband, or 
high-speed, Internet access became available by the late 1990s. 
Broadband differs from a dial-up connection in certain important ways. 
First, broadband connections offer a higher-speed Internet connection 
than dial-up--for example, some broadband connections offer speeds 
exceeding 1 million bits per second (Mbps) both upstream (data 
transferred from the consumer to the Internet service provider) and 
downstream (data transferred from the Internet service provider to the 
consumer).[Footnote 6] These higher speeds enable consumers to receive 
information much faster and thus enable certain applications to be used 
and content to be accessed that might not be possible with a dial-up 
connection. Second, broadband provides an "always on" connection to the 
Internet, so users do not need to establish a connection to the 
Internet service provider each time they want to go online.

Consumers can receive a broadband connection to the Internet through a 
variety of technologies. These technologies include, but are not 
limited to, the following:

* Cable modem. Cable television companies first began providing 
broadband service in the late 1990s over their hybrid-fiber coaxial 
networks. When provided by a cable company, broadband service is 
referred to as cable modem service. Cable providers were upgrading 
their infrastructure at that time to increase their capacity to provide 
video channels in response to competition from direct broadcast 
satellite (DBS) providers such as DirecTV® and Dish Network. By also 
redesigning their networks to provide for two-way data transmission, 
cable providers were able to use their systems to provide cable modem 
service. Cable modem service is primarily available in residential 
areas, and although the speed of service varies with many factors, 
download speeds of up to 6 Mbps are typical. Cable providers are 
developing even higher speed services.

* DSL. Local telephone companies provide digital subscriber line (DSL) 
service, another form of broadband service, over their telephone 
networks on capacity unused by traditional voice service. Local 
telephone companies began to deploy DSL service in the late 1990s--some 
believe, in part, as a response to the rollout of cable modem service. 
To provide DSL service, telephone companies must install equipment in 
their facilities and remove devices on phone lines that may cause 
interference. While most residential customers receive asymmetric DSL 
(ADSL) service with download speeds of 1.5 to 3 Mbps, ADSL technology 
can achieve speeds of up to 8 Mbps over short distances. Newer DSL 
technologies can support services with much higher download speeds.

* Satellite. Currently, three providers of satellite service can offer 
nearly ubiquitous broadband service in the United States. These 
providers use geosynchronous satellites that orbit in a fixed position 
above the equator and transmit and receive data directly to and from 
subscribers. Signals from satellites providing broadband service can be 
accessed as long as the user's reception dish has a clear view of the 
southern sky. Therefore, while the footprint of the providers' 
transmission covers most of the country, a person living in an 
apartment with windows only facing north, or a person living in house 
in a heavily wooded area might not be able to receive Internet access 
via satellite. Earlier Internet services via satellite could only 
receive Internet traffic downstream--that is, from the satellite to the 
subscriber--and upstream Internet traffic was transmitted through a 
standard telephone line connection. Currently, however, satellite 
companies provide both upstream and downstream connections via 
satellite, eliminating the need for a telephone line connection and 
speeding the overall rate of service. Transmission of data via 
satellite typically adds one-half to three-fourths of a second, causing 
a slight lag in transmission and rendering this service less well- 
suited for certain applications over the Internet. While satellite 
broadband service may be available throughout the country, the price 
for this service is generally higher than most other broadband modes; 
both the equipment necessary for service and recurring monthly fees are 
generally higher for satellite broadband service, compared with most 
other broadband transmission modes.

* Wireless. Land-based, or terrestrial, wireless networks can offer a 
broadband connection to the Internet from a wide variety of locations 
and in a variety of ways. Some services are provided over unlicensed 
spectrum and others over spectrum that has been licensed to particular 
companies.[Footnote 7] In licensed bands, some companies are offering 
fixed wireless broadband throughout cities. Also, mobile telephone 
carriers--such as the large companies that provide traditional cell 
phone service--have begun offering broadband mobile wireless Internet 
service over licensed spectrum--a service that allows subscribers to 
access the Internet with their mobile phones or laptops as they travel 
across cities where their provider supports the service. Such services 
are becoming widely deployed and are increasingly able to offer high- 
speed services. A variety of broadband access technologies and services 
are also provided on unlicensed spectrum--that is, spectrum that is not 
specifically under license for a particular provider's network. For 
example, wireless Internet service providers generally offer broadband 
access in particular areas by placing a network of antennae that relay 
signals throughout the network. Subscribers place necessary reception 
equipment outside their homes that will transmit and receive signals 
from the nearest antenna. Also, wireless fidelity (Wi-Fi) networks-- 
which provide broadband service in so-called "hot spots," or areas up 
to 300 feet--can be found in cafes, hotels, airports, and offices. Some 
technologies, such as Worldwide Interoperability for Microwave Access 
(WiMAX), can operate on either licensed or unlicensed bands, and can 
provide broadband service up to approximately 30 miles in a line-of- 
sight environment.

Under section 706 of the Telecommunications Act of 1996, Congress 
directs FCC to encourage deployment of advanced telecommunications 
capability, which includes broadband, to all Americans. In implementing 
the act, FCC has treated the two most widely available broadband 
services--cable modem and DSL service--as information services having a 
telecommunications component. FCC's approach of not treating such 
services as telecommunications services has important legal 
implications because a service defined as a telecommunications service 
could be subject to regulation under Title II of the Communications 
Act, which imposes substantial common carrier regulations unless the 
commission choose to forebear from their enforcement. As part of its 
responsibilities, FCC periodically issues a report to Congress on the 
status of advanced telecommunication capability in the United States. 
To prepare this report, FCC developed a periodic reporting requirement 
using Form 477. In November 2004, FCC modified its rules regarding the 
filing of the 477 form, which went into effect for the companies' 
second filing in 2005. Specifically, FCC removed existing reporting 
thresholds,[Footnote 8] and companies are now required to report their 
total state subscribership by technology.[Footnote 9]

About 30 Million American Households Purchase Broadband Service; 
Despite Evidence of Substantial Broadband Deployment throughout the 
United States, It Is Difficult to Assess Deployment Gaps in Some Areas:

We found that in 2005, about 30 million American households--or 28 
percent--subscribed to broadband, although households in rural areas 
were less likely to subscribe to broadband service than were households 
in urban and suburban areas. Households appear to subscribe to cable 
modem and DSL services--the two primary broadband services--in 
approximately equal numbers. FCC requires providers of broadband 
service to report on the geographic areas in which they serve 
subscribers, but these data are sometimes used to infer the status of 
deployment of companies' Internet infrastructure. Some stakeholders 
find FCC data collection efforts useful for comparison of adoption of 
broadband across states, but we found that the data may not be as 
useful for understanding the status of broadband deployment across the 
country.

About 30 Million American Households Purchase Broadband Service:

Based on survey data from 2005,[Footnote 10] we found that 28 percent 
of American households subscribe to broadband service. Figure 1 
illustrates how American households access the Internet, by various 
technologies, and also shows the percentage of households that do not 
own a computer. As shown, 30 percent of American households subscribe 
to dial-up access, and about 41 percent of American households do not 
have an Internet connection from home. Of those households that do not 
access the Internet, more than 75 percent do not have a computer in the 
home, while the remaining households own a computer but do not have 
online access.

Figure 1: Status of Household Computer Ownership and Internet 
Connection:

[See PDF for image] 

Source: GAO analysis of Knowledge Networks/SRI's The Home Technology 
Monitor. Spring 2005 Ownership and Trend Report.

[End of figure] 

Among online households, we found 50 percent subscribe to dial-up 
service, and 48 percent subscribe to a broadband service.[Footnote 11] 
Additionally, we found that of those households subscribing to a 
broadband service, roughly half purchase DSL service and half purchase 
cable modem service. (See fig. 2 for the types of connections purchased 
by online households.)

Figure 2: Household Online Connection:

[See PDF for image] 

Source: GAO analysis of Knowledge Networks/SRI's The Home Technology 
Monitor. Spring 2005 Ownership and Trend Report.

[End of figure] 

Finally, we found that households residing in rural areas were less 
likely to subscribe to broadband service than were households residing 
in suburban and urban areas.[Footnote 12] Seventeen percent of rural 
households subscribe to broadband service, while 28 percent of suburban 
and 29 percent of urban households subscribe to broadband service. (See 
fig. 3 for the percentage of urban, suburban, and rural households 
purchasing broadband service.)

We also found that rural households were slightly less likely to 
connect to the Internet, compared with their counterparts in suburban 
areas.[Footnote 13]

Figure 3: Percentage of Households Subscribing to Broadband, by Type of 
Location:

[See PDF for image] 

Source: GAO analysis of Knowledge Networks/SRI's The Home Technology 
Monitor. Spring 2005 Ownership and Trend Report.

[End of figure] 

Deployment of Broadband Appears to Be Extensive, but FCC's Form 477 
Data May Not Provide an Accurate Depiction of Gaps in Broadband 
Deployment:

In order to fulfill its responsibility under section 706 of the 
Telecommunications Act, FCC collects data on companies' broadband 
operations. In early 2004, FCC initiated a proceeding to examine 
whether it should collect more detailed information for its broadband 
data gathering program than had previously been collected. 
Specifically, FCC asked whether it should do several things to enhance 
the broadband data including (1) requiring providers to report the 
speeds of their broadband services, (2) eliminating the reporting 
threshold such that all providers of broadband--no matter how small-- 
must report to FCC on its services, and (3) requiring that providers 
report the number of connections by zip code. In late 2004, FCC 
released an order in which it decided to require all providers--no 
matter how small--of broadband to report in the 477 data collection 
effort on broadband and also required providers to report information 
about their services within speed tier categories. The commission 
decided not to require providers to report the number of connections 
(or subscribers) that they serve within each zip code or the number of 
connections in speed tiers or by technology within each zip code, 
finding that finding that such a requirement would impose a large 
burden on filers (particularly smaller entities), and would require 
significant time to implement. In particular, several providers 
commented in the 2004 proceeding that it would be costly and burdensome 
to develop the software and systems to generate the detailed zip code- 
level data and that the cost and burden of further reporting 
requirements would likely outweigh the benefits of more substantial 
information on broadband deployment in the United States. On the other 
hand, 3 state utility commissions asked FCC to gather more information 
within zip codes or by some other Census boundary because such 
information is, in their view, important for tracking broadband 
availability.

Based on the modifications to the filing requirements FCC implemented, 
FCC collects, through its Form 477 filings, information on several 
aspects of each company's provision of broadband at the state level, 
such as the total number of subscribers served, the breakdown of how 
those subscribers are served by technology, and estimates within each 
technology of the percentage of subscribers that are residential. For 
each technology identified in the state reporting, providers also 
submit a list of the zip codes in which they serve at least one 
customer. As discussed above, companies do not report the number of 
subscribers served or whether subscribers are business or residential 
within each zip code; they also do not report information on the 
locations within the zip code that they can serve.

In July 2005, FCC found that 99 percent of the country's population 
lives in the 95 percent of zip codes where at least one provider 
reported to FCC that it serves at least one high-speed subscriber as of 
December 31, 2004. In 83 percent of the nation's zip codes, FCC noted 
that subscribers are served by more than one provider, and the 
commission noted that for roughly 40 percent of zip codes in the United 
States, there are five or more providers reporting high-speed lines in 
service. Although these data indicate that broadband availability is 
extensive, we found that FCC's 477 data may not be useful for assessing 
broadband deployment at the local level.[Footnote 14] While FCC, in 
general, notes that the 477 zip-code data are not meant to measure 
deployment of broadband, in its July 2005 report,[Footnote 15] the 
commission states that in order to be able to evaluate deployment, the 
commission "instituted a formal data collection program to gather 
standardized information about subscribership to high speed services. . 
. ." (Emphasis added. ) Based on our analysis, we found that collecting 
data about where companies have subscribers may not provide a clear 
depiction of their deployment, particularly in the context of 
understanding the availability of broadband for residential 
users.[Footnote 16]

One quandary in analyzing broadband deployment is how to consider the 
availability of satellite broadband services. Even though broadband 
over satellite may not be seen by some as highly substitutable for 
other broadband technologies because of certain technical 
characteristics or because of its higher cost, satellite broadband 
service is deployed: Three companies have infrastructure in place to 
provide service to most of the country.[Footnote 17] The actual 
purchase of satellite broadband is scattered throughout the country and 
shows up in FCC's 477 zip-code data only where someone actually 
purchases the service. It is not clear how satellite service should be 
judged in terms of deployment. Since it is available throughout the 
entire country, one view could be that broadband is near fully 
deployed. Alternatively, it could be viewed that satellite broadband-- 
while available in most areas--does not reflect localized deployment of 
broadband infrastructure and should therefore not be counted as a 
deployed broadband option at all. In either case, FCC's zip-code data 
on satellite broadband--which is based on the pattern of scattered 
subscribership to this service--does not seem to be an appropriate 
indicator of its availability.

Aside from the question of how to view satellite deployment, other 
issues arise in using subscribership indicators for wire or wireless 
land-based providers at the zip-code level as an indicator of 
deployment. These issues include the following:

* Because a company will report service in a zip code if it serves just 
one person or one institution in that zip code, stakeholders told us 
that this method may overstate deployment in the sense that it can be 
taken to imply that there is deployment throughout the zip code even if 
deployment is very localized. We were told this issue might 
particularly occur in rural areas where zip codes generally cover a 
large geographic area. Based on our own analysis, we found, for 
example, that in some zip codes more than one of the large established 
cable companies reported service. Because such providers rarely have 
overlapping service territories, this likely indicates that their 
deployment was not zip-code-wide and that the number of providers 
reported in the zip code overstates the level of competition to 
individual households. We also found that a nontrivial percentage of 
households lie beyond the 3-mile radius of their telephone central 
office, indicating that DSL service was unlikely to be available to 
these homes.

* Companies report service in a zip code even if they only serve 
businesses. One academic expert we interviewed expressed a concern 
about this issue. Based on our own analysis, we found that many of the 
companies filing 477 data indicating service in particular zip codes 
only served business customers. As such, the number of providers 
reported as serving many zip codes is likely overstated in terms of the 
availability of broadband to residences.

* FCC requires that companies providing broadband using unbundled 
network elements (UNE)[Footnote 18] report their broadband service in 
the zip code data. When a provider serves customers using UNEs, they 
purchase or lease underlying telecommunications facilities from other 
providers--usually incumbent telephone companies--to serve their 
customers. Having these providers report their subscribers at the state 
level is important to ensure that correct numbers on the total 
subscribers of broadband service is obtained. However, while UNE 
providers may make investments in infrastructure, such as in 
collocation equipment, they do not generally own or provide last-mile 
connectivity for Internet access. Thus, counting these providers in the 
zip-code-level data may overstate the extent of local infrastructure 
deployment in the sense that several reporting providers could be 
relying on the same infrastructure, owned by the incumbent telephone 
company, to provide broadband access. 

Image: ConnectKentucky:

[See PDF for Image]

Source: ConnectKentucky. 

[End of image]

Based on our analysis, we believe that the use of subscriber indicators 
at the zip-code level to imply availability, or deployment, may 
overstate terrestrially based deployment. We were able to check these 
findings for one state--Kentucky--where ConnectKentucky, a state 
alliance on broadband, had done an extensive analysis of its broadband 
deployment. ConnectKentucky officials shared data with us indicating 
that approximately 77 percent of households in the state had broadband 
access available as of mid-2005. In contrast, we used population data 
within all zip codes in Kentucky, along with FCC's 477 zip-code data 
for that state, and determined that, according to FCC's data, 96 
percent of households in Kentucky live in zip codes with broadband 
service at the end of 2004. Thus, based on the experience in Kentucky, 
it appears that FCC's data may overstate the availability and 
competitive deployment of nonsatellite broadband.

Additionally, to prepare our econometric models, we relied on FCC's 477 
data to assess the number of providers serving the households 
responding to Knowledge Networks/SRI's survey. Based on FCC's data, we 
found that the median number of providers reporting that they serve zip 
codes where the households were located was 8; in 30 percent of these 
zip codes, 10 or more providers report that they provide service. Only 
1 percent of respondents lived in zip codes for which no broadband 
providers reported serving at least one subscriber, according to FCC's 
data. To better reflect the actual number of providers that each of the 
survey respondents had available at their residence, we made a number 
of adjustments to FCC's provider count based on our analysis of the 
providers, certain geographic considerations, and information provided 
by the survey respondents.[Footnote 19] After making these adjustments, 
the median number of providers for the respondents fell to just 2, and 
we found that 9 percent of respondents likely had no providers of 
broadband at all.

Despite these concerns about FCC's 477 data, several stakeholders, 
including a state regulatory office and a state industry association, 
said they found FCC's data useful. An official at a state governor's 
office also noted that analysis of FCC data allowed them to make 
conclusions about the extent of deployment in their state. Similarly, 
an official in another governor's office said that they use FCC's data 
to benchmark the accessibility of broadband in their state because it 
is the only data available. A few academic experts also told us that 
they use FCC's data.

A Variety of Market and Technical Factors, in Addition to Government 
Involvement and Access to Resources at the Local Level, Have Influenced 
the Deployment of Broadband:

Several market characteristics appear to influence providers' broadband 
deployment decisions. In particular, factors related to the cost of 
deploying and providing broadband services, as well as factors related 
to consumer demand, were critical to companies' decisions about whether 
to deploy broadband infrastructure. At the same time, certain technical 
factors related to specific modes of providing broadband service 
influence how and where this service can be provided. Finally, a 
variety of federal and state government activities, as well as access 
to resources at the local level, have influenced the deployment of 
broadband infrastructure.

Several Key Market Factors Related to the Cost of Service and Demand 
Influence Deployment Decisions:

As companies weigh investment decisions, they consider the likely 
profitability of their investments. In particular, because broadband 
deployment requires substantial investment, potential providers 
evaluate the cost to build and operate the infrastructure, as well as 
the likely demand--that is, the expected number of customers who will 
purchase broadband service at a given price--for their service. Based 
on our interviews, we found that several cost and demand factors 
influence providers' deployment decisions.

Cost Factors:

The most frequently cited cost factor affecting broadband deployment 
was the population density of a market. Many stakeholders, including 
broadband providers, state regulators, and state legislators, said 
population density--which is the population per square mile--was a 
critical determinant of companies' deployment decisions. In particular, 
we were told that the cost of building a broadband infrastructure in 
areas where people live farther apart is much higher than building 
infrastructure to serve the same number of people in a more urban 
setting. As such, some stakeholders noted that highly rural areas-- 
which generally have low population density--can be costly to serve. 
Results from our econometric model confirm the views of these 
stakeholders. We found that densely populated and more urbanized 
locations were more likely to receive broadband service than were less 
densely populated and rural locations. For example, we found that urban 
areas were 9 percentage points more likely to have broadband service 
available than were rural areas.

Terrain was also frequently cited as a factor affecting broadband 
deployment decisions. In particular, we were told that infrastructure 
build-out can be difficult in mountainous and forested areas because 
these areas may be difficult to reach or difficult on which to deploy 
the required equipment. Conversely, we were told that flat terrain 
constitutes good geography for telecommunications deployment. For 
wireless providers, we were told that terrain concerns can present 
particular challenges because some wireless technologies require "line- 
of-sight," meaning that radio signals transmitted from towers or 
antennas need an unobstructed pathway--with no mountains, trees, or 
buildings--from the transmission site to the reception devices at 
users' premises. Terrain can also affect satellite broadband 
availability in rural areas that have rolling hills or many trees that 
can obstruct a satellite's signal.

Some stakeholders also said costs for what is known as "backhaul" are 
higher for rural areas and can affect the deployment of broadband 
networks in these areas. Backhaul refers to the transmission of 
information--or data--from any of a company's aggregation points to an 
Internet backbone provider that will then transmit that data to any 
point on the Internet. This is also sometimes referred to as the 
"middle mile." Internet traffic originating from rural areas may need 
to travel a long distance to a larger city to connect to a major 
Internet backbone provider. Because the cost of transmitting over this 
distance--that is, the backhaul--can be high, one stakeholder noted 
that backhaul costs are another barrier to deployment in rural areas. 
However, using our econometric model, we did not find that the distance 
to a metropolitan area with a population of 250,000 or more--our proxy 
for backhaul--was associated with a lower likelihood of broadband 
deployment. 

Image: Backhaul: 

[See PDF for Image] 

Source: GAO. 

[End of image]

Demand Factors:

Based on our interviews with stakeholders, we found that certain demand 
factors influence providers' deployment decisions. In particular, 
because stakeholders noted that potential providers seek to deploy in 
markets where demand for their service will be sufficient to yield 
substantial revenues, certain elements of markets were identified as 
affecting the demand for broadband:

* Ability to aggregate demand. For rural locations, officials we spoke 
with stressed the importance of aggregating sufficient demand. For 
example, officials in one state told us that to justify the cost of 
deployment in rural areas where population density is low, 
telecommunications providers need to be able to aggregate all of the 
possible demand to make a business case. We were also told that 
aggregation is furthered by ensuring that a large "anchor tenant" will 
subscribe to the service. Possible anchor-tenant customers described to 
us included large businesses, government agencies, health-care 
facilities, and schools. Because the revenues from such customers will 
be significant, two stakeholders noted that the anchor tenant alone 
will help to cover a significant portion of the providers' expenses.

* Degree of competition. We found that the degree of existing broadband 
competition in a local market can inhibit or encourage deployment, 
depending on the circumstances. Some new entrants--companies not 
already providing a telecommunications service in an area--reported 
that they avoid entering markets with several existing providers and 
seek out markets where incumbent telephone and cable companies do not 
provide broadband service. The lack of existing service enables the 
entrant company to have the potential to capture many customers. At the 
same time, stakeholders told us that deployment by a new entrant often 
spurred incumbent telephone or cable providers to upgrade their 
infrastructures so as to compete with the entrant in the broadband 
market.

* Technological expertise. A few stakeholders noted that demand will be 
greater in areas where potential customers are familiar with computers 
and broadband, as these individuals are more likely to purchase 
broadband service.

Stakeholders we spoke with rarely mentioned the per-capita income of a 
service area as a factor determining deployment. In fact, a few 
stakeholders credited cable franchising requirements with ensuring 
deployment to low-income areas; in some cases, cable franchise 
agreements require cable providers to build out to all parts of the 
service territory. However, a 2004 study did find that areas with 
higher median incomes were more likely to have competitive broadband 
systems.[Footnote 20] Similarly, results from our econometric analysis 
indicate that areas with higher per-capita income are more likely to 
receive broadband service than are areas with lower per-capita income.

Taxation of Internet Access:

Using our econometric model, we did not find that taxation of Internet 
access by state governments influenced the deployment of broadband 
service. Taxes can raise consumer prices and reduce revenues and impose 
costs on providers, and thereby possibly reduce the incentive for 
companies to deliver a product or service. To assess the impact of 
Internet taxes on broadband deployment, we contacted officials in 48 
states and the District of Columbia[Footnote 21] to determine whether 
the state, or local governments in the state, imposed taxes on Internet 
access. To incorporate this analysis into our model, we used a binary 
variable to indicate the presence of the tax; that is, each state was 
placed into one of two groups, states with a tax and states without a 
tax. As such, this binary variable could potentially capture the 
influence of other characteristics of the states, in addition to the 
influence of the tax. While the parameter estimate in our model had the 
expected sign--indicating that the imposition of taxes may reduce the 
likelihood of broadband deployment--it was not statistically 
significant.

Certain Critical Technical Factors Affect Broadband Deployment:

Many stakeholders we spoke with commented on issues related to 
technical characteristics of networks that provide broadband. In 
particular, many noted that certain technical characteristics of 
methods used to deliver broadband influence the extent of its 
availability. In terms of issues discussed for established modes of 
broadband delivery, we were told the following:

* DSL service can generally be provided over telephone companies' 
copper plant to residences and businesses that are within approximately 
3 miles from the telephone company's facility, known as a central 
office. However, if the quality of the telephone line is not good, the 
distance limit can be reduced--that is, it may only be possible to 
provide DSL for locations within some lesser distance--perhaps 2 miles-
-from a central office. We were told, for example, that in some rural 
areas, DSL is more limited by distance because the telephone lines may 
be older. While the distance limit of DSL can be addressed by deploying 
certain additional equipment that extends fiber further into the 
network, this process can be expensive and time consuming.

* Across spectrum bands used to provide terrestrial wireless broadband 
service, technical characteristics vary: In some cases, signals may 
travel only a short distance, and in other cases, they may travel 
across an entire city; in some cases there may be a need for line-of- 
sight from the transmission tower to the user, but in other cases, the 
signals may be able to travel through walls and trees. Some 
stakeholders mentioned that wireless methods hold great promise for 
supporting broadband service.

* Satellite technology can provide a high-speed Internet service 
throughout most of the United States. However, the most economical 
package of satellite broadband service generally offers, at this time, 
upstream speeds of less than 200 kilobits per second, and therefore 
this service does not necessarily meet FCC's definition of advanced 
telecommunications services, while it does meet FCC's definition of 
high-speed service. Despite the near universal coverage of satellite 
service, consumers need a clear view of the southern sky to be able to 
receive transmissions from the satellites. Additionally, transmission 
via satellite introduces a slight delay, which causes certain 
applications, such as VoIP (i.e., telephone service over the Internet), 
and certain computer gaming to be ill-suited for use over satellite 
broadband.

Some emerging or expanding broadband technologies do not currently have 
significant subscribership, but have the potential to be important 
means of broadband service in the coming years. These technologies 
include deep fiber deployment (e.g., fiber to the home), WiMAX, 
broadband over power lines (BPL), and third-generation (3G) cellular. 
Each of these technologies has technical considerations that are 
influencing the nature of deployment. See appendix IV for a discussion 
of these technologies.

Federal and State Government Efforts, and Access to Resources at the 
Local Level, Have Impacted the Deployment of Broadband:

We found that government involvement in several venues, and access to 
resources at the local level, have affected the deployment of broadband 
networks throughout the nation. In particular, we found that (1) 
certain federal programs have provided funding for broadband networks; 
(2) some state programs have assisted deployment; (3) state and local 
government policies, as well as access to resources at the local level, 
can influence broadband deployment; and (4) broadband deployment-- 
particularly in more rural settings--is often influenced by the extent 
of involvement and leadership exercised by local government and 
community officials.

Federal Programs Have Funded Broadband Infrastructure:

We found that several federal programs have provided significant 
financial assistance for broadband infrastructure.

* The Universal Service Fund (USF) has programs to support improved 
telecommunications services. The high-cost program of the USF provides 
eligible local telephone companies with funds to serve customers in 
remote or rural areas where the cost of providing service is higher 
than the cost of service in more urbanized areas. The high-cost funds 
are distributed to providers according to formulas based on several 
factors, such as the cost of providing service, with funds distributed 
to small rural incumbent local exchange carriers (ILEC) and larger 
ILECs serving rural areas based on different formulas. Competitive 
local exchange carriers can also qualify to receive high-cost funds. 
While high-cost funds are not specifically targeted to support the 
deployment of broadband infrastructure, these funds do support 
telecommunications infrastructure that is also used to provide 
broadband services. We were told by some stakeholders in certain states 
that high-cost support has been very important for the upgrade of 
telecommunications networks and the provision of broadband services. In 
particular, some stakeholders in Alaska, Ohio, and North Dakota told us 
that high-cost support has been critical to small telephone companies' 
ability to upgrade networks and provide broadband services. 
Additionally, the e-rate program of the USF has provided billions of 
dollars in support of Internet connectivity for schools and libraries. 
Another USF program, the Rural Health Care Program, provides assistance 
for rural health facilities' telecommunications services.

* Some programs of the U.S. Department of Agriculture's Rural Utilities 
Service (RUS) provide grants to improve rural infrastructures providing 
broadband service. The Community Connect Program provides grants to 
deploy transmission infrastructures to provide broadband service in 
communities where no broadband services exist, and requires grantees to 
wire specific community facilities and provide free access to broadband 
services in those facilities for at least 2 years. Grants can be 
awarded to entities that want to serve a rural area of fewer than 
20,000 residents. Approximately $9 million was appropriated in 2004 as 
well as in 2005 for this purpose.

* RUS's Rural Broadband Access Loan and Loan Guarantee program provides 
loans[Footnote 22] to eligible applicants to deploy infrastructures 
that provide broadband service in rural communities that meet the 
program's eligibility requirements. A wide variety of entities are 
eligible to obtain loans to serve small rural communities. To obtain a 
4 percent loan, the applicant must plan on serving a community with no 
previously available broadband service, but loans at the Treasury 
interest rate do not have such a requirement.

* The Appalachian Regional Commission's Information Age Appalachia 
program focuses on assisting in the development and use of 
telecommunications infrastructure. The program also provides funding to 
assist in education and training, e-commerce readiness, and technology- 
sector job creation. We were told that in Kentucky, funding from the 
commission assisted the development and operations of ConnectKentucky, 
a state alliance that focuses on broadband deployment and adoption. The 
Appalachian Regional Commission also provided some funding for projects 
in Ohio and Virginia.

Various State Programs Assist the Deployment of Broadband Services:

A number of states we visited have had programs to assist the 
deployment of broadband services, including the following:

* The Texas Telecommunications Infrastructure Fund began in 1996 and 
according to an official of the Texas Public Utility Commission 
committed to spend $1 billion on telecommunications infrastructure in 
Texas. Public libraries, schools, nonprofit medical facilities, and 
higher education institutions received grants for infrastructure and 
connectivity to advanced communications technology. The program is no 
longer operational.

* ConnectKentucky is an alliance of technology-focused businesses, 
government entities, and universities that work together to accelerate 
broadband deployment in the state. ConnectKentucky focuses on three 
goals: (1) raising public awareness of broadband services, (2) creating 
market-driven strategies to increase demand--particularly in rural 
areas, and (3) initiating policy to reduce regulatory barriers to 
broadband deployment.

* The Virginia Tobacco Indemnification and Community Revitalization 
Commission partially funded Virginia's Regional Backbone Initiative. 
The backbone initiative is designed to stimulate economic development 
opportunities by encouraging the creation of new technology-based 
business and industry in southern Virginia, which has historically 
relied heavily on tobacco production.

Local Issues and Access to Resources Impact the Deployment of Broadband 
Services:

The ability of a company to access local rights-of-way, telephone and 
electric poles, and wireless-tower sites can influence the deployment 
of broadband service. In particular, a few stakeholders we spoke with 
said difficulty in gaining access to these resources can serve as a 
barrier to the rapid deployment of broadband service because accessing 
these resources was a time-consuming and expensive process. Companies 
often require access to rights-of-way--such as areas along public 
roads--in order to install infrastructure for broadband service. In 
some instances, companies can face challenges gaining access to rights- 
of-way, which can hinder broadband deployment. For example, we were 
told that in one California community, providers had difficulty 
bringing wires across a highway, which limited their ability to provide 
service in all areas of the community. Some companies also require 
access to telephone and electric poles to install their broadband 
infrastructure. Depending on the entity owning the pole, we were told 
that acquiring access to poles could be costly and time consuming. For 
example, one BPL provider told us that it encountered difficulty 
accessing poles owned by the telephone company. Finally, wireless 
companies need access to towers or sites on which they can install 
facilities for their broadband infrastructure. A few stakeholders we 
spoke with told us that gaining this access can be a difficult process. 
For example, one company said providers are often challenged by the 
need to learn each town's tower-siting rules. While some stakeholders 
identified problems gaining access to these resources, other 
stakeholders did not identify access to rights-of-way, poles, and other 
resources as issues in deploying broadband services.

We found that the video-franchising process can also influence the 
deployment of broadband service because companies may be building 
infrastructure to simultaneously provide both video and broadband 
services. To provide video service, such as cable television, companies 
usually must obtain a franchise agreement from a local government. Some 
stakeholders assert that the video-franchising process can delay the 
deployment of broadband service because providers must negotiate with a 
large number of local jurisdictions. Further, these negotiations can be 
time consuming and costly. As a result, these stakeholders believe that 
local franchising can hinder their ability to deploy broadband 
infrastructures. Alternatively, some stakeholders believe that the 
video-franchising process is important because it helps promote 
deployment of broadband service to all areas of a community. For 
example, some jurisdictions have ubiquity requirements mandating 
deployment to all areas of a community, including those that are less 
affluent. These stakeholders argue that without the local ubiquity 
requirement, service providers could "cherry pick" and exclusively 
provide broadband services to more economically desirable areas.

In some instances, municipal governments provide broadband 
infrastructure and service. For example, we spoke with officials in 
five municipal governments that provide wire-based broadband service, 
often in conjunction with the government's electric utility. We also 
spoke with one municipal government that provided wireless broadband 
service. A few of these municipal government officials told us that 
their municipality had undertaken this deployment because they believe 
that their communities either do not have, or do not have adequate, 
private broadband service. A significant number of stakeholders we 
interviewed support a municipality's right to provide broadband 
services and believe that broadband service is a public utility, such 
as water and sewer. Some support municipal deployment of broadband, 
regardless of whether other providers are available in that area, while 
other stakeholders support a municipality's right to deploy broadband 
service only if there are no other broadband providers serving the 
area. However, other stakeholders we spoke with oppose municipal 
government deployment of broadband service. These stakeholders believe 
that municipal governments are not prepared to be in the business of 
providing broadband and that municipal deployment can hinder private- 
sector deployment.

Community Leadership Encourages the Deployment of Broadband Services:

We found that the cost of serving rural areas presents a challenge to 
the nationwide goal of universal access to broadband. One of the ways 
that some communities have addressed the lack of market entry into 
rural areas has been through initiatives wherein community leaders have 
worked to enhance the likely market success of private providers' entry 
into rural broadband markets. For example, some community leaders have 
worked to aggregate demand--that is, to coordinate the Internet needs 
of various users so that a potential entrant would be able to support a 
business plan. We were told that this leadership--sometimes by key 
government officials, sometimes through partnerships--was seen as 
critical in helping to spur the market in some unserved areas.[Footnote 
23] The following examples illustrate this point: 

Image: Berkshire Connect Attracts Deployment to Western Massachusetts: 

[See PDF for Image] 

Source: Berkshire Connect, Inc. 

[End of image]

* In Massachusetts, several regional coalitions that have been called 
"connect" projects focus on demand aggregation as a tool to encourage 
further deployment of telecommunications backbone and broadband 
networks in more rural parts of the state that were not well served by 
broadband providers. In particular, three such regional groups said 
their demand aggregation model is designed to maximize the purchase of 
broadband services in their region by working with local hospitals, 
schools, home businesses, small business, and residents to demonstrate 
the full extent of the demand for broadband and thus encourage private 
investment in infrastructure. For the one project that was the most 
developed, a few stakeholders told us that the group had been critical 
in helping to spur infrastructure development in the area, and that 
leadership by State government was important to the development of the 
initiative.

* ConnectKentucky, as discussed earlier, is an example of a state 
coalition taking a leadership role to develop information on state 
deployment levels, educate citizens about the benefits of broadband 
service, and advocate broadband-friendly policies with the state 
legislature. Throughout our meetings in Kentucky, the work of 
ConnectKentucky was stated to have been instrumental in the development 
of a common understanding of the state of broadband deployment and 
adoption as well as in instigating new initiatives to advance the 
market. The key element of ConnectKentucky that was cited as crucial to 
its success was leadership from state government, in particular from 
the governor's office.

* In Alaska, we found that in one remote area--Kotzebue, a community 26 
miles above the Arctic Circle--strong local leadership was important to 
the development of a public-private partnership that provides improved 
medical care to the region. The local leadership from the health 
cooperative brought together parties in the community and worked with 
them to develop a plan to provide enhanced health service throughout 
the community's villages. The Maniilaq Health Center uses a wireless 
"telecart" with a video camera that can send high-quality, real-time 
sound and video between the center and Anchorage. The center's 
physicians are able to perform procedures under the guidance of experts 
in Anchorage who can "remotely" look over the physicians' shoulders. In 
addition, there are village clinics staffed by trained village health 
aides. These village clinics are connected to the main health center 
via a broadband link that allows them to share records and diagnoses 
via the telecart.

A Variety of Household and Service Characteristics Influence the 
Adoption of Broadband:

We developed an econometric model to assess the many factors that might 
influence whether a household purchases broadband service. The model 
examined two types of factors: the tax status of states in which 
respondents live, and the characteristics of households. We also 
discussed these issues, as well as the influence of characteristics and 
uses of broadband service, with stakeholders.

Based on our model and interviews with stakeholders, we identified 
several characteristics of households that influence broadband 
adoption. First, our model indicated that high-income households are 39 
percentage points more likely to purchase broadband service than are 
low-income households.[Footnote 24]Similarly, some stakeholders we 
spoke with stated that adoption of broadband service is more widespread 
in communities with high income levels. A key underlying factor may be 
that computer ownership is substantially higher among higher-income 
households, according to a survey conducted by the Census Bureau. 
Second, our model results showed that households with a college 
graduate are 12 percentage points more likely to subscribe to broadband 
services compared with households without a college graduate. In fact, 
when discussing the effects of education on the demand for broadband, 
we were told that some college graduates see broadband as a necessity 
and would be less likely to choose to live in a rural area that did not 
have adequate broadband facilities. Third, we found that households 
headed by young adults are more likely to purchase broadband than are 
households headed by a person 50 or older.[Footnote 25] Similarly, a 
few stakeholders we spoke with said that older adults are less likely 
to purchase broadband. This may be the case because older Americans 
generally have lower levels of computer ownership and computer 
familiarity. We also were told that households with children in school 
are more likely to have broadband service. Figure 4 provides some 
descriptive statistics to illustrate the relationship between several 
demographic characteristics and the adoption of broadband.

Figure 4: Factors Influencing Subscription to Broadband:

[See PDF for image] 

Source: GAO analysis of Knowledge Networks/SRI's The Home Technology 
Monitor: Spring 2005 Ownership and Trend Report. 

[End of figure] 

We also examined whether households residing in rural areas were less 
likely to purchase broadband service than those living in urban areas. 
As noted earlier, we found that only 17 percent of rural households 
subscribe to broadband service. Our model indicated, however, that when 
the availability of broadband to households, as well as demographic 
characteristics, are taken into account, rural households no longer 
appear less likely than urban households to subscribe to broadband. 
That is, the difference in the subscribership to broadband among urban 
and rural households appears to be related to the difference in 
availability of the service across these areas, and not to a lower 
disposition of rural households to purchase the service.

In addition to household characteristics, we also found that 
characteristics and uses of broadband service available to consumers 
can also influence the extent to which households purchase broadband 
service.

* Some stakeholders we spoke with mentioned that the price of broadband 
service is an important factor affecting a household's decision to 
purchase this service. Some stakeholders mentioned, for example, that 
one of the key reasons for the recent surge in DSL subscribership is 
due to recent price declines for the service: Some providers are now 
offering DSL for less than $15 per month. Conversely, because satellite 
broadband service is expensive and also requires the upfront purchase 
of expensive equipment needed to receive the satellite signal, several 
of those we spoke with said that the expense of satellite broadband 
deters its purchase. In fact, a recent study suggests that areas served 
by multiple providers, where prices may tend to be lower, may have 
higher rates of broadband adoption.[Footnote 26] However, because we 
lacked data on the price of broadband service, we were unable to 
include this variable in our econometric model.[Footnote 27] We did not 
find that the number of companies providing broadband service affected 
the likelihood that a household would purchase broadband service.

* Some stakeholders also told us that the availability of applications 
and content not easily accessible through dial-up, as well as the 
degree to which consumers are aware of and value this availability, 
contribute to a household's decision to adopt broadband. For example, 
some functions, applications, and content--such as gaming, VoIP, and 
music and video downloads--either need or function much more 
effectively with broadband service than with dial-up service and, as 
such, make broadband a major attraction for households that value these 
types of services and content. Alternatively, some applications, such 
as e-mail, function adequately with dial-up service, and for households 
that primarily use the Internet for e-mail, there may be little need to 
upgrade to broadband service. Several of those we spoke with noted that 
a "killer application"--one that nearly everyone would view as 
essential and might entice more American households to adopt broadband-
-has not yet emerged.

* We also examined whether the tax status of the state in which each 
survey respondent lived influenced their likelihood to adopt broadband 
service. As mentioned earlier, we used a binary variable to represent 
the presence of Internet taxation. As such, the variable may capture 
the influence of other characteristics of the states in which the 
households resided, in addition to the influence of the tax. Further, 
lacking a variable for the price of broadband service, we cannot assess 
how the imposition of the tax influenced the price of the service. 
Using our model, we found that the parameter estimate had the expected 
sign--indicating that the imposition of the tax may have reduced the 
likelihood that a household would purchase broadband service. While the 
estimate was not statistically significant at the 5 percent level, it 
was statistically significant at the 10 percent level, perhaps 
suggesting that it is a weakly significant factor. However, given the 
nature of our model, it is unclear whether this finding is related to 
the tax or other characteristics of the states in which households 
resided.

Stakeholders Identified Several Options to Address the Lack of 
Broadband in Certain Areas, but Challenges Exist with Implementation:

Stakeholders we spoke with identified several options to facilitate 
greater broadband service in unserved areas; however, each option poses 
special challenges. RUS broadband programs provide a possible means for 
targeted assistance to unserved areas, but stakeholders raised concerns 
about the effectiveness of the loan program and its eligibility 
criteria. USF programs have indirectly facilitated broadband deployment 
in rural areas, but it is unclear whether the program should be 
expanded to directly support broadband service. Finally, wireless 
technologies could help overcome some of the cost and technological 
limitations to providing service in remote locations, but congestion 
and the management of the spectrum remain possible barriers.

RUS Broadband Programs Could Provide a Source of Targeted Assistance, 
but Stakeholders Identified Several Concerns with the Programs:

As mentioned earlier, RUS provides support through grants and loans to 
improve rural infrastructures providing broadband service. The 
Community Connect Broadband grant program provides funding for 
communities where no broadband service currently exists. One loan 
program, which provides loans at 4 percent, also requires that no 
existing broadband providers be present in a community, but loans at 
the Treasury interest rate are available to entities that plan to serve 
communities with existing broadband service. Several stakeholders with 
whom we spoke, as well as the findings of a recent report by the 
Inspector General (IG) of the Department of Agriculture, raised 
concerns about these programs:[Footnote 28]

* Effectiveness of loans. It is not clear whether a loan program--such 
as the RUS loan program--is effective for helping rural areas gain 
access to broadband services. RUS requires applicants to submit an 
economically viable business plan--that is, applicants must show that 
their business will be sufficiently successful such that the applicant 
will be capable of repaying the loan. But developing a viable broadband 
business plan can be difficult in rural areas, which have a limited 
number of potential subscribers. As a result, RUS has rejected many 
applications because the applicant could not show that the business 
plan demonstrated a commercially viable and sustainable business. In 
fact, the agency has been unable to spend all of its loan program 
funds. Since the inception of the program in 2002, the agency has 
fallen far short of obligating the available funding in this program. 
For example, RUS officials told us that in 2004, they estimated that 
the appropriations for the broadband loan program could support 
approximately $2.1 billion in loans, but only 28 percent of this 
amount--or $603 million--was awarded for broadband projects. RUS 
officials also told us that its 2005 appropriations could support just 
over $2 billion in loans, but only 5 percent--or $112 million--was 
awarded to broadband projects. One stakeholder we spoke with suggested 
that a greater portion of RUS funds should be shifted from loans to 
grants in order to provide a more significant level of assistance for 
rural broadband deployment. RUS officials noted that they are currently 
evaluating the program and recognize that the program criteria limit 
the ability of the agency to utilize their full loan funding.

* Competitive environment requirements. During our interviews, some 
stakeholders expressed concerns about how the presence of existing 
broadband deployment was considered in evaluating RUS grant and loan 
applications. In the case of the grant program, RUS approves 
applications only for communities that have no existing broadband 
service. Some local government officials and a company we spoke with 
noted that this "unserved" requirement for RUS grants can disqualify 
certain rural communities that have very limited Internet access-- 
perhaps in only one small part of a community.[Footnote 29] 
Alternatively, regarding the Treasury rate loan program, a few 
providers and the IG's report criticized the program for supporting the 
building of new infrastructure where infrastructure already existed. In 
particular, we learned that loans were being let for deployment in 
areas that already had at least one provider and in some cases had 
several providers. As such, it is not clear whether these funds are 
being provided to communities most in need. RUS officials noted, 
however, that the statute specifically allows such loans. Additionally, 
the issue of how the status of existing service is gauged was a concern 
for one provider we spoke with. RUS obtains information about existing 
providers from applicants, and agency officials told us that agency 
field representatives review the veracity of information provided by 
applicants during field visits. However, RUS officials told us that FCC 
zip-code data is not granular enough for their needs in evaluating the 
extent of broadband deployment in rural areas.

* Community eligibility. A few local officials we spoke with criticized 
the community size and income eligibility requirements for the grant 
and loan programs. In Massachusetts, one stakeholder said that most 
small towns in part of that state exceed RUS's population requirements 
and thus do not qualify for grants or loans. The grant and loan 
programs also have per-capita personal income requirements. One service 
provider in Alaska said that the grant program income eligibility 
requirements can exclude Alaskan communities, while failing to take 
into account the high cost of living in rural Alaska.

* Technological neutrality. Satellite companies we spoke with said 
RUS's broadband loan program requirements are not readily compatible 
with their business model or technology. Once a company launches a 
satellite, the equipment that individual consumers must purchase is the 
remaining infrastructure expense. Because the agency requires 
collateral for loans, the program is more suited for situations where 
the providers, rather than individual consumers, own the equipment 
being purchased through the loan. Yet, when consumers purchase 
satellite broadband, it is common for them to purchase the equipment 
needed to receive the satellite signal, such as the reception dish. 
Additionally, broadband service must be provided at a speed of at least 
200 kilobits in both directions--which is not necessarily the case for 
satellite broadband--for it to qualify for RUS loans. Moreover, RUS 
officials noted that for satellite broadband providers to be able to 
access RUS loans, they would have to demonstrate that each customer 
lives in a community that meets the community size eligibility 
requirement. As such, this program may not be easily utilized by 
satellite broadband providers. Yet for some places, satellite could be 
a cost-effective mechanism to provide broadband infrastructure into 
rural areas. For example, in 2005, the RUS Community Connect program 
provided grants to 19 communities that average 554 residents and 194 
households. The total cost of these grants was roughly $9 million. 
Thus, RUS spent an average of $2,443 per covered household,[Footnote 
30] but the cost per household that adopted broadband would be even 
higher since only a subset of these households would choose to 
subscribe to broadband service. By contrast, two satellite providers we 
spoke with estimate that their consumer equipment and installation 
costs are roughly $600 per subscribing household. These figures might 
not fully represent the full nature of the services provided through 
the grant program and those available via satellite; for example, 
grantees of the RUS program are required to provide free Internet 
service to community centers.

USF Programs Indirectly Support Broadband Service, but Several 
Stakeholders Expressed Concerns:

While the USF program does not directly fund broadband service, the 
funding provided to support telecommunications networks indirectly 
supports the development of infrastructure that can provide many 
communications services, including broadband. USF's high-cost program 
helps maintain and upgrade telecommunications networks in rural areas. 
Three stakeholders we spoke with in Alaska, Ohio, and North Dakota 
attributed the relative success of broadband deployment in rural areas 
to the USF program. Additionally, the Schools and Libraries Program and 
the Rural Health Care Program help facilitate broadband service to 
specific locations; according to two providers in Alaska, these 
programs have been very beneficial in bringing some form of broadband 
service to rural Alaskan villages that might have received no service 
without these government programs.

However, stakeholders we spoke with identified several concerns about 
the USF program:

* Large ILECs serving rural areas and rural ILECs receive high-cost 
fund support under different formulas. The two types of ILECs have 
different eligibility criteria under which they can qualify to receive 
high-cost support and more support is provided to rural companies than 
to nonrural companies serving rural areas.[Footnote 31] Two 
stakeholders we spoke with suggested that the eligibility criteria 
should be modified, such that the criteria better reflect the cost to 
provide service in particular areas, rather than the type of company 
providing the service. Alternatively, two stakeholders we spoke with 
favor the current eligibility criteria and funding mechanism.

* Two stakeholders we spoke with expressed concerns about a lack of 
coordination across USF funding sources, which could lead to 
inefficient use of funds and inadequate leveraging of funds. For 
example, in Alaska, two stakeholders noted that governments and 
providers receive "silos" of funding for schools, libraries, and rural 
health centers. Because the programs are narrowly defined, multiple 
entities might be the recipient of funding for broadband service, which 
could lead to multiple broadband connections in relatively small rural 
communities. One stakeholder noted that since each entity might use 
only a fraction of its available broadband capacity, there can be 
capacity for Internet traffic available for other uses or users, but 
funding recipients are sometimes not allowed to share this capacity, 
either with other entities or with residents in the community. Thus, 
communities may be unable to leverage the available funding for other 
uses.

* While two stakeholders we spoke with suggested expanding the USF 
program to include broadband service, we found little support for this 
overall. Some stakeholders we spoke with expressed concern about 
funding the USF program at current levels of support. These 
stakeholders fear that expanding the USF program to include broadband 
service, which would increase program expenditures and thus require 
additional funding, could undermine support for the entire USF program.

Resolving Spectrum Congestion and Management Concerns Could Facilitate 
Greater Wireless Broadband Service:

As mentioned previously, certain wireless technologies hold the 
potential for supporting broadband service in difficult-to-serve rural 
areas. In less densely populated areas, installing wire-based 
facilities for cable modem and DSL service represents a significant 
cost factor. Therefore, certain wireless technologies may be a lower- 
cost way to serve rural areas than wireline technologies.

While wireless technologies hold the promise of expanding the 
availability of broadband, some stakeholders we spoke with expressed 
concern about the degree of congestion in certain bands as well as the 
management of spectrum. For example, in some geographic areas, we heard 
that congestion in certain unlicensed spectrum bands makes providing 
wireless broadband Internet access more difficult, and a few 
stakeholders said that with more unlicensed spectrum, wireless 
providers could support greater broadband deployment. Additionally, 
wireless providers we spoke with also expressed concern about the 
management of spectrum, particularly the quality of certain bands and 
quantity of spectrum available for wireless broadband service. Two 
stakeholders mentioned that spectrum allocated to wireless broadband 
service is susceptible to having communications obstructed by 
interference from trees and buildings. In a 2005 report, we noted that 
experts agreed that the government should evaluate its allocation of 
spectrum between licensed and unlicensed uses.[Footnote 32] But we also 
noted that these experts failed to agree on whether FCC should dedicate 
more or less spectrum to unlicensed uses. In June 2006, FCC will 
conduct an auction of spectrum dedicated to advanced wireless services, 
which will make available 90 MHz of spectrum for wireless broadband 
services. FCC staff also noted that the commission has other efforts 
underway to increase available spectrum for wireless broadband services.

Conclusion:

In the past several years, the importance of broadband for Americans 
and for the American economy has been articulated by interested 
stakeholders, as well as by the President, Congress, and the last 
several FCC chairmen. Universal availability of broadband has been set 
forth as a policy goal for the near term--2007. And progress toward 
this goal has been substantial. The availability of broadband to 
residential consumers has grown from its nascent beginnings in the 
latter part of the 1990s to broad coverage throughout the country. In 
the last 10 years, providers in traditional communications industry 
segments--telephone and cable--have upgraded and redesigned miles of 
their networks in order to offer broadband services. The provision of 
broadband through various wireless means, as well as over the existing 
electricity infrastructure, have also been developed, and for many, if 
not most Americans, the burgeoning broadband marketplace is 
characterized by competitive choice in broadband access and creative 
and ever-expanding applications and content. Many would consider the 
rollout of broadband infrastructure as a success story of 
entrepreneurial initiative.

But not all places or people have experienced the full benefits of this 
rapid rollout of broadband services. As with many other technologies, 
the costs of bringing broadband infrastructure to rural America can be 
high. For private providers who must weigh the costs and returns of 
their investments, the feasibility of serving the most rural parts of 
our country may not work within a reasonable business model. While 
there are federal support mechanisms for rural broadband, it is not 
clear how much impact these programs are having or whether their design 
suggests a broad consideration of the most effective means of 
addressing the problem. And one of the difficulties of assessing the 
gaps in deployment and where to target any federal support is that it 
is hard to know exactly where broadband infrastructure has not been 
deployed. FCC does collect data on the geographic extent of providers' 
service, but these data are not structured in a way that accurately 
illustrates the extent of deployment to residential users. Without 
accurate, reliable data to aid in analysis of the existing deployment 
gaps, it will be difficult to develop policy responses toward gaps in 
broadband availability. This could hinder our country's attainment of 
universally available broadband. And as the industry moves quickly to 
even higher bandwidth broadband technologies, we risk leaving some of 
the most rural places in America behind.

Recommendation for Executive Action:

In a draft of this report provided to FCC for review and comment, GAO 
recommended that FCC identify and evaluate strategies for improving the 
477 data such that the data provide a more accurate depiction of 
residential broadband deployment throughout the country. In oral 
comments regarding this recommendation, FCC staff acknowledged that the 
477 data have some limitations in detailing broadband deployment, but 
also noted that there had recently been a proceeding examining its 
broadband data collection efforts and that some changes to the data 
collection had been implemented. In that proceeding, the commission 
also determined that it would be costly and could impose large burdens 
on filers--particularly small entities--to require any more detailed 
filings on broadband deployment. Although FCC staff told us that 
analysis of potential costs had been conducted, exact estimates of 
these costs and burdens have not yet been determined. Moreover, many 
have expressed concern about ensuring that all Americans--especially 
those in rural areas--have access to broadband technologies. 
Policymakers concerned about full deployment of broadband throughout 
the country will have difficulty targeting any assistance to that end 
without accurate and reliable data on localized deployment. As such, we 
recommend that FCC develop information regarding the degree of cost and 
burden that would be associated with various options for improving the 
information available on broadband deployment and should provide that 
information to the Senate Committee on Commerce, Science, and 
Transportation and the House Energy and Commerce Committee in order to 
help them determine what actions, if any, are necessary to employ going 
forward.

Agency Comments and Our Evaluation:

We provided a draft of this report to the Department of Agriculture, 
the Department of Commerce, and the Federal Communications Commission 
for their review and comment. The Department of Agriculture provided no 
comments. The Department of Commerce and FCC provided technical 
comments that we incorporated, as appropriate. FCC did not comment on 
the final recommendation contained in this report.

We also provided a draft of this report to several associations 
representing industry trade groups and state and local government 
entities for their review and comment. Specifically, the following 
associations came to GAO headquarters to review the draft: Cellular 
Telecommunications and Internet Association (CTIA), National 
Association of Regulatory Utility Commissioners (NARUC), National 
Association of Telecommunications Officers and Advisors (NATOA), 
National Cable and Telecommunications Association (NCTA), National 
Telecommunications Cooperative Association (NTCA), Satellite Industry 
Association (SIA), US Internet Industry Association (USIIA), United 
States Telecom Association (USTA), and Wireless Internet Service 
Providers Association (WISPA). Officials from CTIA, NARUC, and NTCA did 
not provide comments. Officials from NATOA, NCTA, SIA, and USIIA 
provided technical comments that were incorporated, as appropriate. 
USTA and WISPA provided comments that are discussed in appendix V.

We are sending copies of this report to the appropriate congressional 
committees and to the Secretary of Agriculture, the Secretary of 
Commerce, and the Chairman of the Federal Communications Commission. We 
will also make copies available to others upon request. In addition, 
the report will be available at no charge on the GAO Web site at  
[Hyperlink, http://www.gao.gov]. 

If you have any questions about this report, please contact me at (202) 
512-2834 or heckerj@gao.gov. Contact points for our Offices of 
Congressional Relations and Public Affairs may be found on the last 
page of this report. Contact information and major contributors to this 
report are listed in appendix VI.

Signed By:

JayEtta Z. Hecker: 
Director: 
Physical Infrastructure Issues:

[End of section]

Appendix I: Scope and Methodology:

The objectives of the report were to provide information on (1) the 
current status of broadband deployment and adoption, (2) the factors 
that influence the deployment of broadband networks, (3) the factors 
that influence the adoption of broadband service by households, and (4) 
the options that have been suggested to spur greater broadband 
deployment and adoption. To respond to the four objectives, we used a 
variety of approaches.

To gather opinions for all four objectives, we employed a case-study 
approach. This approach allowed us to identify issues at the state and 
local level that would not be apparent in nationwide data. We selected 
eight states for our case studies: Alaska, California, Kentucky, 
Massachusetts, North Dakota, Ohio, Texas, and Virginia. We selected 
these states based on Census Bureau data on statewide income, 
urbanization, population density, and percentage of households using 
the Internet. We also considered whether each state taxed Internet 
access. We sought to include states in diverse categories of each of 
our selection criteria. In each state, we interviewed state and local 
officials, including local franchising authorities, state public 
utility regulators, representatives from state governor's offices; 
associations; private cable and telephone providers; wireless Internet 
service providers; and municipal and cooperative telecommunications 
providers.

We also spoke with a variety of individuals and organizations 
knowledgeable about broadband services. In particular, we spoke with 
industry providers, trade associations, and academic experts. We also 
spoke with representatives from the Federal Communications Commission 
(FCC), the National Telecommunications and Information Administration 
of the Department of Commerce, and the Rural Utilities Service of the 
Department of Agriculture.

To assess the factors influencing the deployment and adoption of 
broadband, we used survey data from Knowledge Networks/SRI's The Home 
Technology MonitorTM: Spring 2005 Ownership and Trend Report. Knowledge 
Networks/SRI is a survey research firm that conducted a survey on 
household ownership and use of consumer electronics and media. 
Knowledge Networks/SRI interviewed approximately 1,500 randomly sampled 
telephone households, asking questions about the household's purchase 
of computers and Internet access. All percentage estimates from the 
Knowledge Networks/SRI survey have margins of error of plus or minus 7 
percentage points or less, unless otherwise noted. See appendix II for 
a discussion of the steps we took to evaluate the reliability of 
Knowledge Networks/SRI's data. Using the data from Knowledge Networks/ 
SRI, we estimated two econometric models. One model examined the 
factors affecting broadband deployment. We also developed a model to 
examine the factors affecting a household's adoption of broadband 
services. See appendix III for a more detailed explanation of, and 
results from, our deployment and adoption models.

To assess the status of broadband deployment, we used FCC's Form 477 
data that identified companies providing broadband service by zip code. 
We used FCC's data to identify the companies reporting to provide 
broadband service in the zip codes where respondents to Knowledge 
Networks/SRI's survey resided. To assess the reliability of FCC's Form 
477 data, we reviewed documentation, interviewed knowledgeable 
officials, and performed electronic testing of the data elements used 
in our analyses. We made several adjustments to these data, such as 
excluding satellite companies and companies only providing service to 
businesses. See appendix III for more on our methodology concerning 
adjustment to FCC's 477 data. With these adjustments to the data, we 
determined that they were sufficiently reliable for the purposes of 
this report.

We conducted our work from April 2005 through February 2006 in 
accordance with generally accepted government auditing standards.

[End of section]

Appendix II: Data Reliability:

To obtain information on the types of Internet access purchased, or 
adopted, by U.S. households, we purchased existing survey data from 
Knowledge Networks Statistical Research (Knowledge Networks/SRI). Their 
survey was completed with 1,501 of the estimated 3,127 eligible sampled 
households for a response rate of 48 percent. The survey was conducted 
between February 22 and April 15, 2005.

The study procedures yielded a sample of members of telephone 
households in the continental United States using a national random- 
digit dialing method. Survey Sampling Inc. (SSI) provided the sample of 
telephone numbers, which included both listed and unlisted numbers and 
excluded blocks of telephone numbers determined to be nonworking or 
business-only. At least five calls were made to each telephone number 
in the sample to attempt to interview a responsible person in the 
household. Special attempts were made to contact refusals and convert 
them into interviews; refusals were sent a letter explaining the 
purpose of the study and an incentive. Data were obtained from 
telephone households and are weighted to the total number of households 
in the 2005 Current Population Survey adjusted for multiple phone lines.

As with all sample surveys, this survey is subject to both sampling and 
nonsampling errors. The effect of sampling errors due to the selection 
of a sample from a larger population can be expressed as a confidence 
interval based on statistical theory. The effects of nonsampling 
errors, such as nonresponse and errors in measurement, may be of 
greater or lesser significance but cannot be quantified on the basis of 
available data.

Sampling errors arise because of the use of a sample of individuals to 
draw conclusions about a much larger population. The study's sample of 
telephone numbers is based on a probability selection procedure. As a 
result, the sample was only one of a large number of samples that might 
have been drawn from the total telephone exchanges from throughout the 
country. If a different sample had been taken, the results might have 
been different. To recognize the possibility that other samples might 
have yielded other results, we express our confidence in the precision 
of our particular sample's results as a 95 percent confidence interval. 
We are 95 percent confident that when only sampling errors are 
considered each of the confidence intervals in this report will include 
the true values in the study population. All percentage estimates from 
the survey have margins of error of plus or minus 7 percentage points 
or less, unless otherwise noted. The 95 percent confidence interval for 
the estimate of the total number of U.S. households that subscribed to 
broadband service in 2005 is 28.5 million to 33.7 million households.

In addition to the reported sampling errors, the practical difficulties 
of conducting any survey introduce other types of errors, commonly 
referred to as nonsampling errors. For example, questions may be 
misinterpreted, some types of people may be more likely to be excluded 
from the study, errors could be made in recording the questionnaire 
responses into the computer-assisted telephone interview software, and 
the respondents' answers may differ from those who did not respond. 
Knowledge Networks/SRI has been fielding versions of this survey for 
over 20 years. In addition, to reduce measurement error, Knowledge 
Networks/SRI employs interviewer training, supervision, and monitoring, 
as well as computer-assisted interviewing to reduce error in following 
skip patterns.

For this survey, the 48 percent response rate is a potential source of 
nonsampling error; we do not know if the respondents' answers are 
different from the 52 percent who did not respond. Knowledge Networks/ 
SRI took steps to maximize the response rate--the questionnaire was 
carefully designed and tested through deployments over many years, at 
least five telephone calls were made at varied time periods to try to 
contact each telephone number, the interview period extended over about 
8 weeks, and attempts were made to contact refusals and convert them 
into interviews.

Because we did not have information on those contacted who chose not to 
participate in the survey, we could not estimate the impact of the 
nonresponse on our results. Our findings will be biased to the extent 
that the people at the 52 percent of the telephone numbers that did not 
yield an interview have different experiences with Internet access than 
did the 48 percent of our sample who responded. However, distributions 
of selected household characteristics (including presence of children, 
race, and household income) for the sample and the U.S. Census estimate 
of households show a similar pattern.

To assess the reliability of these survey data, we relied on a prior 
GAO report that made use of the Knowledge Networks/SRI 2004 survey for 
a similar purpose. In that prior assessment, we determined that the 
data were sufficiently reliable for our purposes. For this report we 
reviewed Knowledge Networks/SRI's documentation of survey procedures 
for 2005 and compared them to the procedures used in their 2004 survey. 
We determined that their survey methodology was substantively 
unchanged. Additionally, we performed electronic testing of the 2005 
survey data elements used in this report. We determined that the data 
were sufficiently reliable for the purposes of this report.

[End of section]

Appendix III: Broadband Deployment and Adoption Models:

This appendix describes our models of broadband deployment and 
adoption. Specifically, we discuss (1) the design of our models, (2) 
the data sources, (3) our methodology for assessing broadband 
deployment, and (4) the estimation methodology and results.

Design of Our Broadband Deployment and Adoption Models:

A company will deploy broadband service in an area only if the company 
believes that such a deployment will be profitable. Similarly, a 
household will purchase, or adopt, broadband service only if the value, 
or utility, to members of the household exceeds the price the household 
must pay to receive the service. In this section, we explain the two 
models we developed to examine the factors that influence the 
deployment and adoption of broadband service.

Deployment Model:

A company will deploy broadband service in an area only if the company 
believes that such a deployment will be profitable. Based on 
conversations with industry stakeholders, including companies deploying 
broadband service, we identified a number of factors that influence a 
company's decision to deploy broadband service. In particular, the 
following factors may influence the decision to deploy broadband 
service: population density, terrain, backhaul costs, existing or 
potential competition, the technical expertise of the population, the 
income of the population, and regulatory policies (such as rights-of-
way policies). We also reviewed relevant studies, and noted the same 
and additional factors that may influence the deployment of broadband 
service.[Footnote 33] Some of these factors, such as the population 
density and backhaul, will influence the cost of providing broadband 
service, while other factors, such as the income of the population, 
will influence the potential revenues that a company may hope to 
generate. Together, these revenue and cost factors will influence the 
potential profitability of providing broadband service, and ultimately 
the decision to deploy broadband service.

To empirically test these hypotheses, we estimated the following 
econometric model; since all the variables identified above were not 
available, we were unable to include some of the variables--such as 
terrain--in our model. The decision to deploy broadband service is a 
function of:

* the population in the area;

* the population density in the area;

* the percentage of the population residing in an urban area;

* the per-capita income in the area;

* the educational attainment of the population in the area;

* the population teleworking in the area;

* the age of the population in the area;

* the distance to a metropolitan area with a population of 250,000 or 
more; and:

* whether the state in which the area is located imposed a tax on 
Internet access in 2005.

Adoption Model:

Households will purchase, or adopt, broadband service only if the 
value, or utility, that members of the household receive from the 
service exceeds the price of the service. In conversations with 
industry stakeholders, we were told that several characteristics of 
households influence the extent to which households purchase broadband 
service; we also reviewed other studies, and noted characteristics of 
households that these studies associated with the purchase of broadband 
service.[Footnote 34] In particular, the following characteristics of 
households may influence the decision to purchase broadband service: 
income, education, age of household members, presence of children in 
the household, and the technological knowledge of members of the 
household. These characteristics may be associated with the extent to 
which a household would benefit from, and therefore value, broadband 
service, such as using broadband to telework, conduct research for 
school, and playing games. Industry stakeholders also noted that price 
influences a household's decision to purchase broadband service.

To empirically test these hypotheses, we estimated the following 
econometric model; because we lacked data on the price of broadband 
service, we were unable to include this variable in our econometric 
model.[Footnote 35] The decision to purchase, or adopt, broadband 
service is a function of:

* the income of the household;

* the education attainment of the heads of the household;

* the age of the heads of the household;

* the presence of children in the household;

* the racial composition of the household;

* the occupation of the heads of the household;

* the number of people in the household;

* whether the household resides in an urban, suburban, or rural 
location;

* the number of companies providing broadband service in the area; and:

* whether the state in which the household resides imposes a tax on 
Internet access.

Data Sources:

We required several data elements to build the data set used to 
estimate our deployment and adoption models. The following is a list of 
our primary data sources. In addition, we list all of the variables, 
definitions, and sources for the deployment model in table 1 and the 
adoption model in table 2.

* We obtained data on a sample of households in the United States from 
Knowledge Networks/SRI, using Knowledge Networks/SRI's product The Home 
Technology MonitorTM: Spring 2005 Ownership and Trend Report. From 
February through April 2005, Knowledge Networks/SRI interviewed a 
random sample of 1,501 households in the United States. Knowledge 
Networks/SRI asked participating households a variety of questions 
about their use of technology, including questions such as whether the 
household purchased broadband service, and about the household's 
demographic characteristics.

* From the Federal Communications Commission (FCC), we obtained 
information on the companies providing broadband service in zip codes 
throughout the United States in December 2004. For each zip code, FCC 
provided the names of companies reporting, through the agency's Form 
477, that they provided broadband service to at least one residential 
or small business customer and the type of company providing the 
service (e.g., cable and satellite).

* We used the most recent information from the U.S. Census Bureau to 
obtain demographic information for the areas where the households 
responding to Knowledge Networks/SRI's survey were located.

Table 1: Deployment Model: Definitions and Sources of Variables:

Variable: Deploy; 
Definition: A binary variable that equals 1 if broadband service is 
available to the household responding to Knowledge Networks/SRI's 
survey; 
Source: FCC 2004 Form 477 and GAO analysis.

Variable: Internet taxation; 
Definition: A binary variable that equals 1 if the state where the 
household resides imposes a tax on Internet access; 
Source: GAO analysis.

Variable: Population, in thousands; 
Definition: The number of residents in the area where the household 
resides, in thousands; 
Source: Census Bureau.

Variable: Urbanization; 
Definition: The percentage of the population residing in an urban area; 
Source: Census Bureau.

Variable: Distance; 
Definition: The distance to a metropolitan area with a population of 
250,000 or more; 
Source: GAO analysis.

Variable: Percentage of work-at-home residents; 
Definition: The percentage of the population working from home; 
Source: Census Bureau.

Variable: Percentage of population under 16; 
Definition: The percentage of the population under the age of 16; 
Source: Census Bureau.

Variable: Percentage of population 17 to 24; 
Definition: The percentage of the population 17 to 24 years old; 
Source: Census Bureau.

Variable: Percentage of population 65 or older; 
Definition: The percentage of the population 65 or older; 
Source: Census Bureau.

Variable: Percentage of population with a high-school degree; 
Definition: The percentage of the population with a high-school degree; 
Source: Census Bureau.

Variable: Percentage of population with education beyond high school; 
Definition: The percentage of the population with education beyond high 
school; Source: Census Bureau.

Variable: Per-capita income, in thousands; 
Definition: The per-capita income in the area, in thousands of dollars; 
Source: Census Bureau.

Variable: Population density, in thousands; 
Definition: The ratio of population to square miles in the area, in 
thousands; 
Source: Census Bureau. 

Source: GAO.

[End of table]

Table 2: Adoption Model: Definitions and Sources of Variables:

Variable: Adopt; 
Definition: A binary variable that equals 1 if the household responding 
to Knowledge Networks/SRI's survey purchases broadband service; 
Source: Knowledge Networks/SRI.

Variable: Internet taxation; 
Definition: A binary variable that equals 1 if the state where the 
household resides imposes a tax on Internet access; 
Source: GAO analysis.

Variable: Number of broadband providers; 
Definition: The number of companies providing broadband service to the 
household; 
Source: FCC 2004 Form 477 and GAO analysis.

Variable: Income between $30,000 and $49,900; 
Definition: A binary variable that equals 1 if the household's income 
is between $30,000 and $49,900; 
Source: Knowledge Networks/SRI.

Variable: Income between $50,000 and $99,900; 
Definition: A binary variable that equals 1 if the household's income 
is between $50,000 and $99,900; 
Source: Knowledge Networks/SRI.

Variable: Income $100,000 or more; 
Definition: A binary variable that equals 1 if the household's income 
is greater than or equal to $100,000; 
Source: Knowledge Networks/SRI.

Variable: Race-white; 
Definition: A binary variable that equals 1 if the household reported 
its race as white; 
Source: Knowledge Networks/ SRI.

Variable: College graduate; 
Definition: A binary variable that equals 1 if either the man or woman 
of the household is a college graduate; 
Source: Knowledge Networks/SRI.

Variable: Age 34 to 49; 
Definition: A binary variable that equals 1 if either the man or woman 
of the household is 34 to 49 years old, and neither is younger than 34; 
Source: Knowledge Networks/SRI.

Variable: Age 50 or older; 
Definition: A binary variable that equals 1 if both the man and woman 
of the household are 50 years old or older; 
Source: Knowledge Networks/SRI.

Variable: Children; 
Definition: A binary variable that equals 1 if a child age 17 or 
younger resides in the home; 
Source: Knowledge Networks/SRI.

Variable: Household size; 
Definition: A binary variable that equals 1 if three or more people 
reside in the home; 
Source: Knowledge Networks/SRI.

Variable: Occupation-professional; 
Definition: A binary variable that equals 1 if the man or woman of the 
household reports working in a professional position; 
Source: Knowledge Networks/SRI.

Variable: Occupation-clerical, sales, or technical; 
Definition: A binary variable that equals 1 if the man or woman of the 
household reports working in a clerical, sales, or technical position, 
and neither reports working in a professional position; 
Source: Knowledge Networks/SRI.

Variable: Occupation-blue collar; 
Definition: A binary variable that equals 1 if the man and woman of the 
household report working in a blue collar position; 
Source: Knowledge Networks/SRI.

Variable: Occupation-other; 
Definition: A binary variable that equals 1 if the man or woman of the 
household reports working in a position other than a professional, 
clerical, sales, technical, or blue-collar position; 
Source: Knowledge Networks/SRI.

Variable: Rural location; 
Definition: A binary variable that equals 1 if the household resides in 
an area outside a metropolitan statistical area (MSA); 
Source: GAO analysis.

Variable: Suburban location; 
Definition: A binary variable that equals 1 if the household resides in 
an area inside an MSA but outside the central city of that MSA; 
Source: GAO analysis. 

Source: GAO.

[End of table]

Assessing Broadband Deployment:

FCC's Form 477 data include information on companies providing 
broadband service to at least one residential or business customer in 
zip codes throughout the United States in December 2004. However, since 
zip codes can represent large geographic areas, companies providing 
broadband service in a zip code might not have facilities in place to 
serve all households in the zip code. Thus, while a household might 
reside in a zip code in which FCC's Form 477 indicates that broadband 
service is available, that service might not be available to the 
household. Additionally, as we note in the text, we identified other 
concerns with FCC's data. Therefore, we took additional steps to assess 
whether broadband service was available to households included in 
Knowledge Networks/SRI's survey. In particular, we took the following 
steps for each observation in our data set:

* removed firms providing only satellite service;

* removed firms that provided only broadband service to business 
customers, since residential households were the focus of our study;

* removed large incumbent local exchange carriers when the company was 
identified as providing service in areas that lay outside of its local 
exchange area, since these firms typically provide service only to 
business customers outside of their local exchange areas;[Footnote 36]

* removed firms when 2 or more of the 10 largest cable operators 
reported providing broadband service, since large cable operators 
rarely have overlapping service territories;

* removed cable operators if the responding household indicated that 
cable service did not pass the residence; and:

* removed companies providing telephone-based broadband service if the 
household's residence was greater than 2.5 miles from the central 
office facility, since DSL service is distance limited.

Estimation Methodology and Results:

For both the deployment model and adoption model, we are estimating a 
reduced-form, binary-choice model. That is, broadband service is either 
deployed in the area or it is not, and the household either purchases 
broadband service or it does not. Given the binary choice nature of the 
models, we employed the probit method to estimate the deployment and 
adoption equations.[Footnote 37] In this section, we present 
descriptive statistics and estimation results for the two equations and 
discuss the results.

Deployment Model:

In table 3, we provide basic statistical information on all of the 
variables included in the deployment model, and in table 4, we provide 
the results from the probit estimation of the deployment model. Of the 
1,501 respondents to Knowledge Networks/SRI's survey, we used 1,402 
observations in the deployment model; we were unable to match the zip+4 
code for all 1,501 observations with publicly available data, which was 
necessary to assess whether the residence was 2.5 miles from the 
serving central office facility.

Table 3: Deployment Model: Descriptive Statistics:

Variable: Deploy; 
Mean: 0.911; 
Standard deviation: 0.285; 
Minimum value: 0.000; 
Maximum value: 1.000.

Variable: Internet taxation; 
Mean: 0.546; 
Standard deviation: 0.498; 
Minimum value: 0.000; 
Maximum value: 1.000.

Variable: Population, in thousands; 
Mean: 26.022; 
Standard deviation: 17.982; 
Minimum value: 0.070; 
Maximum value: 113.935.

Variable: Urbanization; 
Mean: 76.154; 
Standard deviation: 33.240; 
Minimum value: 0.000; 
Maximum value: 100.000.

Variable: Distance; 
Mean: 34.361; 
Standard deviation: 42.743; 
Minimum value: 0.249; 
Maximum value: 572.803.

Variable: Percentage of work-at-home residents; 
Mean: 3.257; 
Standard deviation: 2.064; 
Minimum value: 0.000; 
Maximum value: 33.333.

Variable: Percentage of population under 16; 
Mean: 24.018; 
Standard deviation: 4.651; 
Minimum value: 6.225; 
Maximum value: 41.219.

Variable: Percentage of population 17 to 24; 
Mean: 10.585; 
Standard deviation: 4.785; 
Minimum value: 0.583; 
Maximum value: 55.113.

Variable: Percentage of population 65 or older; 
Mean: 12.995; 
Standard deviation: 5.688; 
Minimum value: 2.195; 
Maximum value: 59.057.

Variable: Percentage of population with a high-school degree; 
Mean: 29.546; 
Standard deviation: 9.002; 
Minimum value: 3.121; 
Maximum value: 68.966.

Variable: Percentage of population with education beyond high school; 
Mean: 51.395; 
Standard deviation: 16.092; 
Minimum value: 8.836; 
Maximum value: 95.348.

Variable: Per-capita income, in thousands; 
Mean: 44.466; 
Standard deviation: 15.597; 
Minimum value: 9.583; 
Maximum value: 164.479.

Variable: Population density, in thousands; 
Mean: 2.976; 
Standard deviation: 6.876; 
Minimum value: 0.002; 
Maximum value: 74.814. 

Source: GAO.

[End of table]

Table 4: Deployment Model: Estimation Results:

Variable: Intercept; 
Parameter estimate: -2.9299;
[p-value]: [0.0097][A].

Variable: Internet taxation; 
Parameter estimate: -0.1486;
[p-value]: [0.2275].

Variable: Population, in thousands; 
Parameter estimate: 0.0099;
[p-value]: [0.1140].

Variable: Urbanization; 
Parameter estimate: 0.0102; 
[p-value]: [0.0001][A].

Variable: Distance; 
Parameter estimate: -0.0012;
[p-value]: [0.3115].

Variable: Percentage of work-at-home residents; 
Parameter estimate: -0.0600; 
[p-value]: [0.0392][B].

Variable: Percentage of population under 16; 
Parameter estimate: 0.0335;
[p- value]: [0.1192].

Variable: Percentage of population 17 to 24; 
Parameter estimate; 0.0198;
[p- value]: [0.3027].

Variable: Percentage of population 65 or older; 
Parameter estimate: 0.0468;
[p-value]: [0.0271][B].

Variable: Percentage of population with a high-school degree; 
Parameter estimate: 0.0114; 
[p-value]: [0.3260].

Variable: Percentage of population with education beyond high school; 
Parameter estimate: 0.0121;
[p-value]: [0.1957].

Variable: Per-capita income, in thousands;
Parameter estimate: 0.0270;
[p- value]: [0.0074][A].

Variable: Population density, in thousands; 
Parameter estimate: 0.1706;
[p- value]: [0.0159][B].

Variable: Number of observations; 
Parameter estimate and [p-value]: 1,402.

Variable: 1-LogL/Log0; 
Parameter estimate and [p-value]: 32.0077. 

Source: GAO.

[A] Significant at the 1 percent level.

[B] Significant at the 5 percent level.

[End of table]

Results from our model indicate that several factors related to the 
cost of providing broadband service and the demand for broadband 
service influence the likelihood that service will be available in a 
particular area. Regarding the cost factors, we found that urban areas 
and areas with greater population density are more likely to receive 
broadband service. For example, urban areas are about 9 percentage 
points more likely to receive broadband service than are similar rural 
areas. These results are consistent with broadband service being less 
costly to deploy in densely populated, more urban environments, where a 
similar investment in facilities can serve a greater number of 
subscribers than is possible in rural areas. Regarding demand for 
broadband service, we found that areas with greater per-capita incomes 
are more likely to receive broadband service. Additionally, we found 
that areas with a greater number of people working from home are less 
likely to have broadband service and that areas with a greater 
percentage of people age 65 or older are more likely to have broadband 
service.

We did not find that taxation of Internet access by state governments 
influenced the deployment of broadband service. Taxes can raise 
consumer prices and reduce revenues and impose costs on providers, and 
thereby possibly reduce the incentive for companies to deliver a 
product or service. Since we used a binary variable to indicate the 
presence of taxes, this variable could also potentially capture the 
influence of other characteristics of the states, in addition to the 
influence of the tax. Results from our model indicate that Internet 
access taxes do not affect the likelihood that companies will deploy 
broadband service; while the parameter estimate has the expected sign, 
the estimate is not statistically significant.

Adoption Model:

In table 5, we provide basic statistical information on all of the 
variables included in the adoption model, and in table 6, we provide 
the results from the probit estimation of the adoption model. Since 
households can only chose to purchase, or adopt, broadband service 
where it is deployed, we only include households from Knowledge 
Networks/SRI's survey where we assessed that broadband service was 
available; based on our analysis, 133 respondents did not have 
broadband service available. Further, 355 respondents to Knowledge 
Networks/SRI's survey did not answer one or more demographic questions 
and 29 did not answer, or did not know, what type of Internet 
connection their household purchased. Therefore, we excluded these 
respondents. Thus, we used 901 observations in the adoption 
model.[Footnote 38]

Table 5: Adoption Model: Descriptive Statistics:

Variable: Adopt; 
Mean: 0.336; 
Standard deviation: 0.473; 
Minimum value: 0.000; 
Maximum value: 1.000.

Variable: Internet taxation; 
Mean: 0.553; 
Standard deviation: 0.497; 
Minimum value: 0.000; 
Maximum value: 1.000.

Variable: Number of broadband providers; 
Mean: 3.307; 
Standard deviation: 2.161; 
Minimum value: 1.000; 
Maximum value: 9.000.

Variable: Income between $30,000 and $49,900; 
Mean: 0.223; 
Standard deviation: 0.417; 
Minimum value: 0.000; 
Maximum value: 1.000.

Variable: Income between $50,000 and $99,900; 
Mean: 0.336; 
Standard deviation: 0.473; 
Minimum value: 0.000; 
Maximum value: 1.000.

Variable: Income $100,000 or more; 
Mean: 0.149; 
Standard deviation: 0.356; 
Minimum value: 0.000; 
Maximum value: 1.000.

Variable: Race-white; 
Mean: 0.858; 
Standard deviation: 0.349; 
Minimum value: 0.000; 
Maximum value: 1.000.

Variable: College graduate; 
Mean: 0.499; 
Standard deviation: 0.500; 
Minimum value: 0.000; 
Maximum value: 1.000.

Variable: Age 34 to 49; 
Mean: 0.378; 
Standard deviation: 0.485; 
Minimum value: 0.000; 
Maximum value: 1.000.

Variable: Age 50 or older; 
Mean: 0.424; 
Standard deviation: 0.494; 
Minimum value: 0.000; 
Maximum value: 1.000.

Variable: Children; 
Mean: 0.387; 
Standard deviation: 0.487; 
Minimum value: 0.000; 
Maximum value: 1.000.

Variable: Household size; 
Mean: 0.465; 
Standard deviation: 0.499; 
Minimum value: 0.000; 
Maximum value: 1.000.

Variable: Occupation-professional; 
Mean: 0.442; 
Standard deviation: 0.497; 
Minimum value: 0.000; 
Maximum value: 1.000.

Variable: Occupation-clerical, sales, or technical; 
Mean: 0.154; 
Standard deviation: 0.361; 
Minimum value: 0.000; 
Maximum value: 1.000.

Variable: Occupation-blue collar; 
Mean: 0.029; 
Standard deviation: 0.167; 
Minimum value: 0.000; 
Maximum value: 1.000.

Variable: Occupation-other; 
Mean: 0.244; 
Standard deviation: 0.430; 
Minimum value: 0.000; 
Maximum value: 1.000.

Variable: Rural location; 
Mean: 0.052; 
Standard deviation: 0.222; 
Minimum value: 0.000; 
Maximum value: 1.000.

Variable: Suburban location; 
Mean: 0.568; 
Standard deviation: 0.496; 
Minimum value: 0.000; 
Maximum value: 1.000. 

Source: GAO.

[End of table]

Table 6: Adoption Model: Estimation Results:

Variable: Intercept; 
Parameter estimate: -1.4919;
[p-value]: [0.0001][A].

Variable: Internet taxation; 
Parameter estimate: -0.1683;
[p-value]: [0.0745].

Variable: Number of broadband providers; 
Parameter estimate: 0.0118;
[p- value]: [0.6101].

Variable: Income between $30,000 and $49,900; 
Parameter estimate: 0.4531;
[p-value]: [0.0024][A].

Variable: Income between $50,000 and $99,900; 
Parameter estimate: 0.7429;
[p-value]: [0.0001][A].

Variable: Income $100,000 or more; 
Parameter estimate: 1.1331;
[p-value]: [0.0001][A].

Variable: Race-white; 
Parameter estimate: 0.2905;
[p-value]: [0.0405][B].

Variable: College graduate; 
Parameter estimate: 0.3525;
[p-value]: [0.0009][A].

Variable: Age 34 to 49; 
Parameter estimate: -0.2239;
[p-value]: [0.0759].

Variable: Age 50 or older; 
Parameter estimate: -0.3316;
[p-value]: [0.0217][B].

Variable: Children; 
Parameter estimate: 0.1318;
[p-value]: [0.3894].

Variable: Household size; 
Parameter estimate: 0.1241;
[p-value]: [0.3894].

Variable: Occupation-professional; 
Parameter estimate: 0.2610; 
[p-value]: [0.1409].

Variable: Occupation-clerical, sales, or technical; 
Parameter estimate: 0.2098;
[p-value]: [0.2867].

Variable: Occupation-blue collar; 
Parameter estimate: 0.2638;
[p-value]: [0.3879].

Variable: Occupation-other; 
Parameter estimate: 0.0212; 
[p-value]: [0.9086].

Variable: Rural location; 
Parameter estimate: -0.3234; 
[p-value]: [0.1892].

Variable: Suburban location; 
Parameter estimate: 0.0983; 
[p-value]: [0.3406].

Variable: Number of observations; 
Parameter estimate and [p-value]: 901.

Variable: 1-LogL/Log0; 
Parameter estimate and [p-value]: 16.2800. 

Source: GAO.

[A] Significant at the 1 percent level.

[B] Significant at the 5 percent level.

[End of table]

Our model results indicate that four characteristics influence whether 
households purchase, or adopt, broadband service. First, we found that 
households with greater incomes are more likely to purchase broadband 
service than are lower-income households. For example, the 25 percent 
of households with the highest income levels were about 39 percentage 
points more likely to purchase broadband service than the 25 percent of 
households with the lowest income levels. Second, households with a 
college graduate are about 12 percentage points more likely to purchase 
broadband service than are households without a college graduate. We 
also found that white households are more likely to purchase broadband 
service than households of other races. Finally, older households are 
less likely to purchase broadband service than are younger households.

As with the deployment model, we did not find that taxation of Internet 
access by state governments influenced the adoption of broadband 
service. As mentioned earlier, we used a binary variable to represent 
the presence of Internet taxation. As such, the variable may capture 
the influence of other characteristics of the states in which the 
households resided, in addition to the influence of the tax. Further, 
lacking a variable for the price of broadband service, we cannot assess 
how the imposition of the tax influenced the price of the service and 
thus the household's adoption decision. Using our model, we found that 
the parameter estimate had the expected sign--indicating that the 
imposition of the tax may have reduced the likelihood that a household 
would purchase broadband service. While the estimate was not 
statistically significant at the 5 percent level, it was statistically 
significant at the 10 percent level, perhaps suggesting that it is a 
weakly significant factor. However, given the nature of our model, it 
is unclear whether this finding is related to the tax or other 
characteristics of the states in which households resided.

[End of section]

Appendix IV: Additional Communications Technologies:

Based on our conversations with stakeholders, and our own research, we 
identified several emerging technologies that could further the 
deployment of broadband service.

Broadband over power lines. Broadband over power lines (BPL) is an 
emerging competitive source of broadband to the home. BPL transmits 
broadband by using existing electric distribution networks, such as the 
wires that deliver electricity to consumers. Although there are a few 
commercial deployments, most BPL efforts are currently at the trial 
stage. Trials and commercial deployments range across the urban-rural 
landscape, from Cullman County, Alabama, to Cincinnati.[Footnote 39] 
Currently, BPL can provide upstream and downstream speeds of 3 million 
bits per second (Mbps), and next generation equipment is being 
developed to provide speeds of 100 Mbps.

Industry stakeholders have identified several concerns with BPL 
service. First, while traveling across the electric network, BPL can 
emit signals that interfere with other users of the spectrum, such as 
amateur radio and public safety. The Federal Communications Commission 
(FCC) has taken steps to document, mitigate, and alleviate this 
potential problem. Second, some stakeholders also expressed concern 
that, due to the age or condition of the electric network, providers in 
some areas would be unable to transmit Internet data at high speeds. 
Finally, some stakeholders expressed varied opinions about the 
feasibility of BPL to bring broadband service to rural areas. Some 
stakeholders were optimistic about BPL's ability to serve these 
communities, while others expressed skepticism, pointing out that 
overcoming BPL's distance limitations would require more equipment and 
additional costs.

Wireless fidelity (Wi-Fi). Wi-Fi-enabled wireless devices, such as 
laptop computers, can send and receive data from any location within 
signal reach--about 300 feet--of a Wi-Fi-equipped access point. Wi-Fi 
provides data transmission rates, based on the current transmission 
standard, of up to a maximum of 54 Mbps,[Footnote 40] which is shared 
by multiple users. Wi-Fi equipment and services are based on the 802.11 
series standards developed by the Institute of Electrical and 
Electronics Engineers (IEEE) and operate on an unlicensed basis in the 
2.4 and 5 GHz spectrum bands. Several stakeholders we spoke with said 
that Wi-Fi service complemented, rather than substituted for, other 
broadband services.

The number of areas that can access Wi-Fi service, known as "hot 
spots," has grown dramatically and, according to one equipment 
manufacturer, may exceed 37,000. Wi-Fi hot spots include such diverse 
entities as airports, colleges, retail establishments, and even entire 
towns. Increasingly, municipalities are planning or deploying larger 
area or citywide hot spots; some municipalities considering or 
deploying a Wi-Fi network include Atlanta, Philadelphia, San Francisco, 
and Tempe, Arizona. While Wi-Fi service is widely deployed in urban and 
suburban areas, some stakeholders identified a few problems with the 
service. Because Wi-Fi hot spots operate in unlicensed spectrum, 
interference can be a problem. Several stakeholders we spoke with 
mentioned congestion or limited distance capability in Wi-Fi as a 
potential limitation of the service.

Worldwide Interoperability for Microwave Access (WiMAX). With WiMAX 
service, the distance covered and data transmission speeds can exceed 
those found with Wi-Fi service. WiMAX can provide data transmission 
speeds of 75 Mbps with non-line-of-sight service--that is, the signal 
can pass through buildings, trees, or other obstructions--or up to 155 
Mbps with line-of-sight service. In a non-line-of-sight environment, 
WiMAX can provide service in an area with a radius of 3 miles or more; 
in a line-of-sight environment, WiMAX can provide service up to 
approximately 30 miles. WiMAX equipment and services are based on the 
IEEE 802.16 series of standards and operate in unlicensed and licensed 
spectrum.

WiMAX networks are being deployed on a trial commercial basis, but some 
challenges remain for further deployment. More than 150 pilot and 
commercial deployments of WiMAX networks are currently in use. Because 
of its greater capabilities in terms of distance and speed, WiMAX can 
extend wireless broadband to less densely populated communities, where 
wired solutions may be more expensive to deploy. Stakeholders we spoke 
with serving smaller, less densely populated areas indicated that they 
were testing or interested in WiMAX to serve their communities. 
However, concerns have been raised about spectrum availability, 
interference, and the ability of different manufacturers' equipment to 
support the same level of broadband applications. FCC has several 
initiatives under way to increase the availability of spectrum for 
WiMAX services. While the WiMAX Forum Certification Lab certifies WiMAX 
equipment, the standard allows manufacturers of equipment various 
options, such as different levels of security protocols, and thus, not 
all equipment may support the same level of service, such as carrying 
voice over the Internet (VoIP) and security.

Third generation (3G) cellular broadband. Recently, several major 
commercial wireless companies have introduced broadband service based 
on advances in cellular technology and data protocols. Focused 
primarily on the business customer and more expensive than cable modem 
and DSL services, 3G services permit consumers to receive broadband 
service while mobile. 3G services typically provide data transmission 
speeds of 400 to 700 kilobits per second (Kbps). There are two 
competing technologies: EV-DO service, introduced by Verizon and 
Sprint; and HSDPA, introduced by Cingular. Currently, Verizon Wireless 
reports that its service is available nationally in 181 major 
metropolitan markets, covering approximately 150 million people. Sprint 
reports providing EV-DO service in major airports and business 
districts in 212 markets, covering approximately 140 million people. 
For HSDPA service, Cingular reports that its service is available to 
nearly 35 million people in 52 communities. Industry stakeholders 
expressed concerns about the ubiquity of service, data transmission 
speeds, and the monthly costs associated with 3G service. Opinions 
varied as to whether cellular broadband services would be a competitive 
threat, or a complementary service, for consumers of other broadband 
services.

Fiber to the home (FTTH). FTTH provides a high-speed, wire-based 
alternative to traditional cable and telephone networks. According to 
the FTTH Council, as of September 2005, 2.7 million homes were passed 
by fiber and over 300,000 homes were connected to fiber in 652 
communities in 46 states. Stakeholders expressed concerns about the 
high cost associated with deploying FTTH, and also that FTTH deployment 
was concentrated in urban and suburban communities, or in newly 
developed communities (known as "greenfields").

[End of section]

Appendix V: Comments from Industry Participants:

We provided a draft of this report to several associations representing 
industry trade groups and state and local government entities for their 
review and comment. The following associations came to GAO headquarters 
to review the draft: Cellular Telecommunications and Internet 
Association (CTIA), National Association of Regulatory Utility 
Commissioners (NARUC), National Association of Telecommunications 
Officers and Advisors (NATOA), National Cable and Telecommunications 
Association (NCTA), National Telecommunications Cooperative Association 
(NTCA), Satellite Industry Association (SIA), US Internet Industry 
Association (USIIA), United States Telecom Association (USTA), and 
Wireless Internet Service Providers Association (WISPA).

Officials from CTIA, NARUC, and NTCA did not provide comments. 
Officials from NATOA, NCTA, SIA, and USIIA provided technical comments 
that were incorporated, as appropriate.

USTA officials noted that our discussion of the effects of local 
franchising on deployment imply that franchise agreements have helped 
to ensure broad deployment of broadband, but that, in the view of USTA, 
franchise buildout requirements can deter entry and thus reduce 
deployment.

WISPA officials expressed concern about our findings regarding the 
taxation of Internet access and noted that it is important, in their 
view, that wireless Internet access provided by small providers not be 
taxed, and in fact, WISPA officials noted that small providers should 
be provided a tax incentive to encourage investment and expansion in 
underserved areas. Additionally, these officials expressed concern 
about the presentation of data on how households currently access the 
Internet from their homes. WISPA stated that these data understate the 
importance that wireless access will have toward the goal of universal 
broadband coverage both within and outside of users' homes. WISPA 
stated that the report accurately depicts that wireless Internet 
service providers (WISP) currently hold a minority market share, and 
WISPA officials note that without certain government policies to foster 
growth in the wireless industry, WISPs will be at a competitive 
disadvantage. WISPA officials also expressed concern that the report 
understates factors that are hindering the growth of the wireless 
Internet industry--most notably, the need for additional spectrum under 
1 Ghz, such as the TV white spaces. Further WISPA noted that the data 
showing broadband penetration rates in urban, rural, and suburban areas 
should not be interpreted as indicating that access to broadband is 
lower in only rural areas. They suggested that differences in broadband 
penetration rates across these types of locations are not that great 
and that pockets of areas with no access exist in many areas. As such, 
WISPA suggests that policy response regarding spectrum availability, 
USF funding, and Rural Utilities Service be focused on engaging smaller 
providers that can bring broadband to areas not currently served by the 
larger incumbent providers.

[End of section]

Appendix VI: GAO Contact and Staff Acknowledgments:

GAO Contact:

JayEtta Z. Hecker, (202) 512-2834 or heckerj@gao.gov:

Staff Acknowledgments:

Individuals making key contributions to this report include Amy 
Abramowitz (Assistant Director), Eli Albagli, Stephen Brown, Michael 
Clements, Sandra DePaulis, Nina Horowitz, Eric Hudson, Bert Japikse, 
John Mingus, Sara Ann Moessbauer, Karen O'Conor, Lindsay Welter, and 
Duffy Winters.

(544102):

FOOTNOTES

[1] Throughout this report, we refer to high speed Internet access over 
broadband technologies as broadband Internet access. 

[2] See GAO, Internet Access Tax Moratorium: Revenue Impacts Will Vary 
by State, GAO-06-273 (Washington, D.C.: Jan. 23, 2006). 

[3] While FCC states that its zip-code information is not meant to be a 
measure of broadband deployment, some parties have used it in this 
manner because there are no other official data on deployment of 
broadband across the country. 

[4] The 3-mile limit applies to the path taken by the telephone wire, 
not necessarily a straight line between the central office and the 
customer's residence. 

[5] With fiber feeders, DSL service can be extended beyond three miles 
from the central office. 

[6] FCC defined "advanced service" as exceeding 200 Kbps both upstream 
and downstream and "high-speed" service as exceeding 200 Kbps in at 
least one direction, in order to distinguish these from existing data 
services based on widely available analog telephony and ISDN technology.

[7] Spectrum is a natural resource used to provide an array of wireless 
communication services. FCC regulates commercial entities' use of 
spectrum. With unlicensed spectrum, a number of users without licenses 
share a portion of the spectrum, adhering to certain technological 
specifications. In contrast, with licensed spectrum, FCC provides 
entities with a license to use a specific portion of the spectrum. 

[8] In the past, companies with less than 250 broadband connections 
were not required to submit information to FCC through Form 477. FCC 
officials told us that many of the companies that are now required to 
report are very small and in rural areas. These officials stated that 
many of these companies are not reporting and that therefore the data 
may not fully represent broadband deployment.

[9] FCC requires providers to report on their broadband lines or 
wireless channels. While this may not exactly equate to subscribers, 
the number of lines and subscribers is related, and we use the word 
subscribers throughout this report as we refer to the 477 filings of 
companies. 

[10] We used survey data from Knowledge Networks/SRI's The Home 
Technology MonitorTM: Spring 2005 Ownership and Trend Report.

[11] A very small number of respondents to the survey accessed the 
Internet over a satellite connection, but none of the respondents 
reported any other means of wireless access.

[12] We refer to rural areas as areas outside metropolitan statistical 
areas (MSA); suburban areas as areas within an MSA but not a central 
city; and urban areas as a central city of an MSA.

[13] Our findings are not substantially different from those of other 
organizations. Based on 2003 data, the Census Bureau reported that 62 
percent of American households had a computer--see U.S. Census Bureau, 
Computer and Internet Use in the United States: 2003 (Washington, D.C., 
2005). Additionally, the Department of Commerce reported that 20 
percent of households--or 37 percent of online households--had 
broadband service, with DSL becoming increasingly popular. This study 
also found that broadband service was less commonly purchased in rural 
areas--see U.S. Department of Commerce, A Nation Online: Entering the 
Broadband Age (Washington, D.C., September 2004). Similarly, using 
survey data from 2005, the Pew Internet and American Life Project 
reported that 53 percent of Internet users subscribed to broadband 
service, that much of the growth in broadband service in recent years 
arose from DSL subscriptions, and that broadband service was less 
prevalent in rural areas when compared with broadband subscribership in 
suburban and urban areas.

[14] In a recent report, we also noted that the 477 data do not provide 
a full description of broadband services for certain segments of the 
population, such as Native Americans residing on tribal lands. See GAO, 
Telecommunications: Challenges to Assessing and Improving 
Telecommunications for Native Americans on Tribal Lands, GAO-06-189 
(Washington, D.C.: Jan. 11, 2006). 

[15] See FCC, High-Speed Services for Internet Access: Status as of 
December 31, 2004 (Washington, D.C., July 2005).

[16] The problems related to tracking data on subscribership versus 
deployment/availability in Form 477 is not an issue with mobile 
wireless operators. Because mobile wireless broadband services are 
designed to be used while subscribers are mobile, those operators are 
directed to report the zip codes covered by their mobile wireless 
broadband networks, rather than the zip codes of the billing addresses 
of their subscribers.

[17] As noted earlier, some households might not be able to actually 
receive broadband over satellite if they do not have a clear view of 
the southern sky.

[18] UNEs are physical and functional elements of the telephone 
network, such as the telephone line, or loop, which, under the 
Telecommunications Act of 1996, incumbent telephone companies must make 
available to competitors for lease or purchase.

[19] In particular, we removed satellite providers, removed any 
companies we determined only provide service to business customers, 
removed a cable provider if we found that more than 1 of the largest 10 
cable providers served the zip code, removed a cable provider if the 
respondent said that cable does not pass their residence, and removed 
telephone-based providers if the residence was further than 2.5 miles 
from the central office that served the respondent's home.

[20] See Tony H. Grubesic and Alan T. Murray, "Waiting for Broadband: 
Local Competition and the Spatial Distribution of Advanced 
Telecommunication Services in the United States," Growth and Change 
(2004), 139-165.

[21] We did not contact officials in Alaska and Hawaii, since the 
survey data from Knowledge Networks/SRI did not include households from 
these two states.

[22] This program also can provide loan guarantees, but to date, no 
loan guarantees have been requested.

[23] A recent GAO report, Telecommunications: Challenges to Assessing 
and Improving Telecommunications for Native Americans on Tribal Lands, 
GAO-06-189 (Washington, D.C.: Jan. 11, 2006) discusses how leadership 
in a community can help to improve telecommunications services on 
tribal lands. The report provides several examples of tribes addressing 
the barriers to deployment of telecommunications networks by partnering 
with private entities, providing technical training, and taking 
initiative to access federal grants.

[24] "High-income households" were defined as those having incomes in 
top 25 percent of all households, while "low-income households" were 
defined as those having incomes in the bottom 25 percent of all 
households.

[25] We define "young adults" as people between the ages of 18 and 33.

[26] See Debra J. Aron and David E. Burnstein, "Broadband Adoption in 
the United States: An Empirical Analysis" (paper presented at the 31st 
Annual Telecommunications Policy Research Conference, Arlington, Va., 
2003).

[27] We recognize that our model does not fully include all the 
variables that would influence the adoption decision. As such, the 
parameter estimates will be biased. We are unable to assess the 
possible extent of this bias.

[28] U.S. Department of Agriculture, Office of Inspector General, Audit 
Report: Rural Utilities Service Broadband Grant and Loan Programs, 
Audit Report 09601-4-Te (Washington, D.C., Sept. 30, 2005).

[29] According to RUS officials, demand for the grant program exceeds 
available funding under the current program requirements.

[30] RUS officials noted that many different technologies were used in 
these 19 communities, so that the cost per household varied 
considerably across the grant recipients.

[31] For rural ILECs, the cost of service is based on "embedded costs"-
-or the historical costs of infrastructure, which is used to provide a 
variety of communications services. For nonrural ILECs, the cost of 
service is based on the forward-looking costs of providing only certain 
telecommunications services.

[32] See GAO, Telecommunications: Strong Support for Extending FCC's 
Auction Authority Exists, but Little Agreement on Other Options to 
Improve Efficient Use of Spectrum, GAO-06-236 (Washington, D.C.: Dec. 
20, 2005).

[33] For example, see Tony H. Grubesic and Alan T. Murray, "Waiting for 
Broadband: Local Competition and the Spatial Distribution of Advanced 
Telecommunication Services in the United States," Growth and Change, 
vol. 35, no. 2 (2004): 139-165; and James E. Prieger, "The Supply Side 
of the Digital Divide: Is There Equal Availability in the Broadband 
Internet Access Market?" Economic Inquiry, vol. 41, no. 2 (2003): 346- 
363.

[34] For example, see Scott Wallsten, Broadband Penetration: An 
Empirical Analysis of State and Federal Policies (Washington, D.C.: AEI-
Brookings Joint Center for Regulatory Studies, 2005); Scott J. Savage 
and Donald M. Waldman, "United States Demand for Internet Access," 
Review of Network Economics, vol. 3, no. 3 (2004): 228-247; and Debra 
J. Aron and David E. Burnstein, "Broadband Adoption in the United 
States: An Empirical Analysis" (paper presented at the 31st Annual 
Telecommunications Policy Research Conference, Arlington, Va., 2003).

[35] As such, we recognize that the parameter estimates will be biased. 
We are unable to assess the possible extent of this bias.

[36] We did not remove Verizon, since thought its acquisition of GTE, 
it serves a wide variety of locations as an incumbent exchange carrier.

[37] An alternative method to estimate these equations is the logit 
model. In a binary choice model, the differences between the logit and 
probit models are generally not significant. Differences can arise in 
the multinomial model, where there are three or more choices, because 
the logit model imposes independence conditions that sometimes do not 
reflect the conditions being modeled. Such was not the case in our 
models, since we are estimating binary choice equations.

[38] We did not find that the households that failed to answer the 
demographic questions were more or less likely to be online than were 
the households that answered these questions.

[39] BPL companies, such as Current Communications, offer the ability 
to improve the monitoring, detection, and management of the electrical 
distribution network through improved communication capabilities 
inherent in BPL equipment. These features increase the attractiveness 
of BPL to electrical companies, as the companies receive the benefits 
of improved network operation and potential revenues from broadband 
service. 

[40] Discussions are underway for newer standards for Wi-Fi that would 
dramatically increase the transmission speeds.

GAO's Mission:

The Government Accountability Office, the investigative arm of 
Congress, exists to support Congress in meeting its constitutional 
responsibilities and to help improve the performance and accountability 
of the federal government for the American people. GAO examines the use 
of public funds; evaluates federal programs and policies; and provides 
analyses, recommendations, and other assistance to help Congress make 
informed oversight, policy, and funding decisions. GAO's commitment to 
good government is reflected in its core values of accountability, 
integrity, and reliability.

Obtaining Copies of GAO Reports and Testimony:

The fastest and easiest way to obtain copies of GAO documents at no 
cost is through the Internet. GAO's Web site ( www.gao.gov ) contains 
abstracts and full-text files of current reports and testimony and an 
expanding archive of older products. The Web site features a search 
engine to help you locate documents using key words and phrases. You 
can print these documents in their entirety, including charts and other 
graphics.

Each day, GAO issues a list of newly released reports, testimony, and 
correspondence. GAO posts this list, known as "Today's Reports," on its 
Web site daily. The list contains links to the full-text document 
files. To have GAO e-mail this list to you every afternoon, go to 
www.gao.gov and select "Subscribe to e-mail alerts" under the "Order 
GAO Products" heading.

Order by Mail or Phone:

The first copy of each printed report is free. Additional copies are $2 
each. A check or money order should be made out to the Superintendent 
of Documents. GAO also accepts VISA and Mastercard. Orders for 100 or 
more copies mailed to a single address are discounted 25 percent. 
Orders should be sent to:

U.S. Government Accountability Office

441 G Street NW, Room LM

Washington, D.C. 20548:

To order by Phone:

Voice: (202) 512-6000:

TDD: (202) 512-2537:

Fax: (202) 512-6061:

To Report Fraud, Waste, and Abuse in Federal Programs:

Contact:

Web site: www.gao.gov/fraudnet/fraudnet.htm

E-mail: fraudnet@gao.gov

Automated answering system: (800) 424-5454 or (202) 512-7470:

Public Affairs:

Jeff Nelligan, managing director,

NelliganJ@gao.gov

(202) 512-4800

U.S. Government Accountability Office,

441 G Street NW, Room 7149

Washington, D.C. 20548: