Unnamed: 0
int64
0
4.52k
report
stringlengths
7
512
1,700
or the regulated community. For example, during 2007, FEC received a complaint alleging that a candidate had violated television advertisement disclaimer requirements by including an improper disclaimer in the advertisement. The complaint alleged that the printed disclaimer violated the requirements because it did not also state that the candidate approved the advertisement. FEC dismissed the case in an exercise of its prosecutorial discretion to not pursue a violation in part because of partial compliance
1,701
with disclaimer requirements. In doing so, FEC observed that the verbal disclaimer identified the candidate and informed the public of the candidate’s approval of the advertisement and the printed disclaimer informed the public that the candidate’s committee paid for the advertisement. FCC receives hundreds of thousands of complaints related to all areas it regulates but there have only been a small number of sponsorship identification cases. Of the sponsorship identification cases opened by FCC, only a han
1,702
dful have resulted in enforcement actions against broadcasters, and many of those enforcement actions were for fines of less than $100,000. Most broadcasters told us they generally have no problems meeting the sponsorship identification requirements because they have processes in place to review all content and ensure it has a sponsorship announcement if needed. However, FCC guidance for the sponsorship identification requirements has not been updated in nearly 50 years to address more modern technologies a
1,703
nd applications. We have previously reported that retrospective reviews of regulations can change behavior of regulated entities. Similarly, a review and update of FCC guidance that discusses outdated technologies could result in changes in behavior. One example discusses a broadcaster’s use of expensive kinescope prints as part of a story on a controversial issue. The example directs such a use should receive a sponsorship announcement because of the controversial issue being discussed and the cost of the
1,704
film. Yet, today, because the expense of providing film is no longer relevant, broadcasters may be unsure on whether the concern is the expense of the film or the controversial issues discussed in the film. FCC should clarify its guidance to clearly address how, when content is provided with no money or payment in-kind and it does not discuss a controversial issue, the situation should be treated. Furthermore, FCC should clarify its examples to direct broadcasters’ treatment of content provided with no mone
1,705
y or payment in-kind that does not highlight a product or brand beyond the “reasonably related” standard, such as a VNR. FCC has indicated VNRs must have a sponsorship announcement; however, FCC’s enforcement of VNRs has not found fault with the use of the VNR but rather when the VNRs focus on a specific product. Stakeholders disagree on the use of VNRs. FCC’s enforcement actions and guidance do not distinguish how to act when portions of VNRs are used or when a VNR does not disproportionately focus on a pr
1,706
oduct or brand. FCC indicated in 2005 that it would issue a report or take other necessary action regarding this issue and updating the guidance could serve this purpose. Unlike FEC in its enforcement of disclaimer requirements, FCC’s enforcement process for sponsorship identification cases generally does not inform the broadcasters or cablecasters named in the complaint when investigations have been closed. In cases where a letter of inquiry has been sent, the broadcaster or cablecaster must fulfill its re
1,707
sponsibility and provide FCC with the requested information. Yet, according to FCC, because they have no legal obligation to inform broadcasters that an investigation has concluded, it typically does not provide that information. By providing this information, for cases in which FCC conducts a full investigation and determines the broadcaster’s actions not to be a violation of requirements, the outcome could provide guidance to the broadcaster of allowable activities. Even in cases where FCC closed a case w
1,708
ith no investigation, informing the broadcaster that the case is closed, even if it may be reopened in the future, would support government-wide goals of greater transparency and sound oversight practices. Finally, while in 2011 FCC had specific goals related to the timeliness of completing sponsorship identification investigations, it was unable to provide data supporting how it met those goals, and in subsequent years it withdrew the goals. In an effort to achieve greater openness, the timeliness of repor
1,709
ting and publishing information has been identified as an essential component. By re-establishing goals about completing sponsorship identification investigations in a timely manner, FCC would support broader government goals of completing actions in a timely manner to better serve its constituencies and regulated entities. We recommend that the Chairman of the FCC take the following three actions: To provide clarity on how sponsorship identification requirements apply to activities not directly addressed b
1,710
y FCC’s current guidance, such as the use of video news releases, and to update its guidance to reflect current technologies and recent FCC decisions about video news releases, FCC should initiate a process to update its sponsorship identification guidance and consider providing additional examples relevant to more modern practices. To improve its transparency concerning which investigations are ongoing or have been concluded and to provide guidance on allowable activities, FCC should communicate the closur
1,711
e of all sponsorship identification investigations with the broadcaster named in the complaint after a letter of inquiry was sent. The letter should indicate the case has been closed, but in doing so, FCC could note that closing the case does not signify an endorsement of the actions that were being investigated and that the case could be reopened. To improve timeliness of investigations and ensure, when possible, that investigations are completed in an expeditious manner, FCC should develop goals for compl
1,712
eting sponsorship identification cases within a specific time frame and develop a means to measure and report on how well it meets those goals. We provided a draft of our report to FCC and FEC for review and comment. FCC provided comments in a letter dated January 23, 2013, that is reprinted in appendix II. Overall, FCC indicated that it will consider our recommendations and how to address the concerns discussed in our report. In response to our second recommendation—to communicate the closure of investigat
1,713
ions with the broadcaster named in the complaint when a letter of inquiry has been sent—FCC identified a number of issues, many of which were cited in our report. Specifically, FCC has concerns that reporting the closing of a case may be misinterpreted as an endorsement of a broadcaster’s actions. FCC further noted its limited number of Enforcement Bureau staff available to work on the large portfolio of cases could and that it could not dedicate the necessary time to craft closing letters tailored to each
1,714
case. However, we feel that FCC could create a standard letter—stating that a case has been closed, that the closing of the case does not endorse the actions of the broadcaster named in the complaint, and that the case could be reopened because of new evidence. We believe such a standard letter would require minimal resources to create and send, yet would contribute to greater transparency. FCC also noted that it is reluctant to single out sponsorship identification matters for special treatment in terms of
1,715
closure letters but are also concerned about the even greater impact on resources if closure letters are instituted on a broad basis. However, we believe that this could serve as a pilot program for greater adoption of closure letters for other types of FCC investigations. In response to FCC’s concerns, we updated our recommendation to demonstrate how a closure letter could be worded to indicate the closure did not indicate an endorsement of the actions being investigated and that a case could be reopened.
1,716
Both FCC and FEC provided technical comments that were incorporated as appropriate. When providing its technical comments, FCC discussed the treatment of VNRs indicating that although the 2005 Public Notice states VNRs generally must have a sponsorship announcement, recent cases involving VNR complaints have resulted in FCC treating VNRs similar to other programming subject to the sponsorship identification requirements. We reflected this change in the report, and added a reference to FCC decisions about V
1,717
NRs to our first recommendation. As agreed with your office, unless you publicly announce the contents of this report earlier, we plan no further distribution of it until 30 days from the date of his letter. At that time, we will send copies of this report to the Chairman of the Federal Communications Commission, and the Chair of the Federal Election Commission. We will also make copies available to other on request. In addition, the report will be available on the GAO Web site at http://www.gao.gov. If you
1,718
or your staff have any question, please contact me at (202) 512- 2834 or [email protected]. Contact points for our Offices of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff that made major contributions to this report are listed in appendix III. We were asked to review requirements for identifying the sponsors of broadcast commercial and political advertisements and to determine the extent to which agencies responsible for administering them address comp
1,719
laints alleging violations of the requirements. Specifically, we (1) describe the sponsorship identification requirements and guidance for commercial and political content, the federal election disclaimer requirements, as well as stakeholders’ views of these requirements and guidance and (2) assess how and to what extent FCC and FEC address sponsorship complaints through each agency’s enforcement process. To describe the sponsorship identification requirements and guidance as well as stakeholders views on t
1,720
hese requirements and guidance, we reviewed sponsorship identification statutes and regulations, including the Communication Act of 1934, as amended, and the Federal Election Campaign Act. We also reviewed relevant academic literature and interviewed FCC and FEC officials to gather their understanding of the governing statutes and regulations, how they have changed, their purposes, and how they apply to various media platforms, including conventional and embedded advertising and video news releases, and pol
1,721
itical and controversial advertisements. We conducted interviews with over one-dozen stakeholders selected because they are either subject to sponsorship identification requirements, have monitored broadcasters’ compliance with sponsorship identification requirements, or contributed research to the topic. These stakeholders’ interviews were conducted with representatives from the four major television broadcast networks as well as two additional groups that own television and radio stations. We also intervi
1,722
ewed representatives from a consumer watchdog organization, academics, and trade associations that represent broadcasters and news directors and producers. These interviews were conducted to obtain views on the effect of statutes and regulations and possible changes to them. To determine how and to what extent FCC and FEC address sponsorship identification complaints, we interviewed FCC and FEC officials responsible for receiving, processing, and investigating complaints. In addition, we analyzed relevant F
1,723
CC and FEC documents describing agency methods and processes for identifying violations, receiving sponsorship identification complaints, communicating with the complainant and subject of the complaint, initiating and conducting investigations, and taking enforcement actions. We also analyzed relevant FCC and FEC data to describe sponsorship identification complaints, investigations, and enforcement actions. We analyzed FCC data showing all complaints received by FCC from 2000 through June 2012 to determine
1,724
the percentage of complaints that were sponsorship identification complaints, FCC actions in response to sponsorship identification complaints, and the time frames for resolving these complaints. We determined the FCC data to be sufficiently reliable for our purposes based on previous analysis of the FCC database and based on current interviews. To determine the extent to which FCC addresses sponsorship identification complaints, we analyzed all FCC enforcement actions pursuant to these complaints from 200
1,725
0 through June 2012. We also analyzed FEC data showing all complaints received by FEC from 2000 through May 15, 2012, to determine the percentage of complaints that were disclaimer statement complaints, FEC actions in response to disclaimer statement complaints, and the time frames for resolving these complaints. The FEC data were not materially relevant to our findings and so while we asked a series of questions about the internal controls and data reliability, we did not make a final determination of its
1,726
reliability. To determine the extent to which FEC addresses disclaimer statement complaints, we analyzed all FEC disclaimer statement cases, including cases dismissed by FEC, and all FEC disclaimer statement enforcement actions from 2000 through May 15, 2012. We also analyzed FCC and FEC strategic plans describing the agencies’ respective goals for sponsorship identification and disclaimer statement complaint processing and enforcement. In addition, we analyzed FCC and FEC data and documents describing whet
1,727
her the agencies met their respective goals. We interviewed FCC and FEC officials about their goals and their progress toward achieving them. We conducted this performance audit from March 2012 through January 2013 in accordance with generally accepted government auditing standards. Those standards require that we plan and perform the audit to obtain sufficient, appropriate evidence to provide a reasonable basis for our findings and conclusions based on our audit objectives. We believe that the evidence obt
1,728
ained provides a reasonable basis for our findings and conclusions based on our audit objectives. In addition to the individual named above, Ray Sendejas, Assistant Director; Eric Hudson; Bert Japikse; Michael Mgebroff; Amy Rosewarne; and Andrew Stavisky made key contributions to this report.
1,729
USPS has a universal service obligation, part of which requires it to provide access to retail services. It is required to serve the public and provide a maximum degree of effective and regular postal services to rural areas, communities, and small towns where post offices are not self- sustaining. USPS is intended to be a financially self-sufficient entity that covers its expenses almost entirely through postal revenues. USPS receives virtually no annual appropriations but instead generates revenue through
1,730
selling postage and other postal products and services. Retail alternatives are increasingly important to USPS; revenues from all retail alternatives—including self-service kiosks in post offices, USPS’s website, and CPUs, among others—increased by about $1.6 billion from fiscal years 2007 to 2011 while post office revenues decreased by $3 billion. (See fig. 1.) During this same period, USPS’s share of total retail revenues from all retail alternatives increased from about 24 percent to 35 percent. USPS pr
1,731
ojects that by 2020, retail alternatives will account for 60 percent of its retail revenues. Given this growing importance and USPS’s planned-retail-network restructuring, we recommended in November 2011 that USPS implement a strategy to guide efforts to modernize its retail network that addresses both post offices and retail alternatives. According to USPS officials, USPS is currently in the process of finalizing its retail strategy. The retail alternatives most similar to post offices are CPUs. They are p
1,732
rivately owned, operated, and staffed and are usually colocated with a primary business, such as a convenience store or supermarket. They provide most of the same products and services as post offices (see fig. 2) at the same prices. CPUs typically have a counter with prominently displayed official USPS signage, provided by USPS, giving the CPU the look of a post office. (See fig. 3.) According to USPS, CPUs offer potential service and financial benefits, and, as we have previously reported, some foreign po
1,733
sts have successfully used private partnerships similar to CPUs to realize such benefits. CPUs can enhance service by being located closer to customers’ homes and workplaces and operating at hours when post offices may not be open. They can alleviate long lines at existing post offices and provide postal services to areas with rapid population growth or where opening new post offices may be cost prohibitive. Regarding financial benefits, USPS has reported that the costs it incurs for CPUs are less than thos
1,734
e it incurs for post offices, relative to revenue earned. USPS estimated that in fiscal year 2011, it incurred $0.17 in costs for each dollar of revenue at CPUs and $0.51 in costs for each dollar of revenue at post Costs are lower, in part because CPU operators, and not USPS, offices.are responsible for their operating costs, such as rent, utilities, and wages for their employees. CPUs provide all their revenues from postal products and services to USPS, and USPS compensates CPUs for providing postal servic
1,735
es under the terms of their contracts. The amount of compensation USPS pays to a CPU operator depends in large part on the type of contract the CPU operates under. Currently there are two basic types of contracts: fixed-price, under which USPS compensates the CPU a contractually determined amount regardless of sales, and performance-based, under which USPS compensates the CPU a contractually determined percentage of sales. USPS’s compensation to CPUs—either the amount under a fixed-price contract or the per
1,736
centage under a performance-based contract—is specific for each CPU contract and the result of negotiation between USPS and the CPU operator. CPU hours of service are also negotiated for each contract, although USPS guidance on CPUs, in line with USPS’s goal for CPUs to provide increased access and convenience, states that their days and hours of service should exceed those at post offices. Other terms and conditions are standardized in all contracts. For example, all CPUs are required to offer the same bas
1,737
ic set of products and services such as stamps, Priority Mail, Express Mail, and Certified Mail. In addition, all CPUs are contractually prohibited from selling services, including private mailboxes and others, that are competitive with USPS’s products, and all CPU contracts specify USPS’s rights to inspect the CPU at any time during operating hours. CPU contracts are valid for an indefinite period, but CPU contracts specify that the CPU operator or USPS can terminate a contract and close the CPU at any tim
1,738
e with 120 days notice. USPS management and oversight of CPUs, including identifying and justifying the need for new CPUs, is done at the district and local levels. Staff at the district and local levels oversee day-to-day operations of CPUs and identify the need for new CPUs. When a district identifies the need for a new CPU, it approaches local businesses in the targeted area as potential partners and engages in a competitive application process. USPS has other partnerships with private entities to provid
1,739
e retail postal services, similar to CPUs. USPS launched a retail partnership called the Village Post Office in July 2011 in which existing small businesses provide a limited range of postal products and services in small communities where underutilized yet costly post offices may close, be consolidated with another nearby post office, or have their hours of service reduced. partnerships with national and regional retailers to provide postal services. These partnerships differ from CPUs in that they will no
1,740
t be subject to the same prohibitions as CPUs on selling competing services, and with them, USPS is attempting to expand access at a national or regional level as opposed to addressing specific local needs as CPUs do. Village Post Offices sell a more limited range of USPS products and services than CPUs do. Local USPS staff solicit a local business for a Village Post Office opportunity and a USPS contracting officer agrees to enter into a contract if they believe that the terms and conditions of that Villag
1,741
e Post Office present a best value to USPS. USPS plans to launch these partnerships in test markets in early 2013 and will evaluate the effectiveness of these partnerships before making decisions whether to expand the program. Although total number of CPUs has decreased in recent years, USPS continues to use CPUs to provide customers with access to postal services at additional locations and for more hours of service. CPUs are located in a variety of locations, both urban and rural, and range from very clos
1,742
e to far from post offices, demonstrating how USPS uses CPUs to provide customers with alternatives located near crowded post offices—which are often found in urban areas— and to provide service where post offices are not conveniently located or may not be cost effective for USPS, often in rural areas. In addition, CPUs allow USPS to provide customer access at times often beyond the hours of service at post offices. According to USPS data, the number of CPUs fell from 5,290 in 2002 to 3,619 in 2011. During
1,743
the past 5 fiscal years, USPS has opened new CPUs, but a higher number of CPUs have closed. (See table 1.) According to USPS headquarters officials who manage the CPU program, economic conditions forced many businesses that operated CPUs to close and declining mail volume and sales of postal products have been the primary factors behind the decrease in the number of CPUs. Although USPS does not track specific reasons for CPU closures in its contract postal unit technology (CPUT) database, retail managers in
1,744
eight USPS districts that we met with cited specific local issues resulting in CPU closures, including the following: The CPU operator retired or otherwise stopped working. For example, an Indiana CPU operator closed his primary business and moved out of the area. The CPU operator moved the primary business to a new location and did not retain the CPU. For example, the operator of a CPU in Texas moved his primary business across the street, but the new space was too small to host a CPU. The CPU operator so
1,745
ld the primary business. For example, a California CPU operator sold his self-storage business, and the new operators were not interested in maintaining the CPU. The CPU operator chose to close for financial considerations. For example, a CPU operator in Virginia closed the CPU because he felt it did not help his primary business. USPS initiated the closure because the CPU failed to meet the terms of the contract or USPS determined that the CPU was not cost effective. For example, USPS determined that a Mar
1,746
yland CPU that operated out of a private residence no longer brought in enough revenue to justify USPS’s compensation to the CPU, so USPS closed the CPU. Consistent with USPS’s goal to use CPUs to absorb excess demand at post offices, our analysis of the distance between CPUs and post offices shows that more than 56 percent of CPUs are less than 2 miles from the nearest post office and 26 percent are less than 1 mile. (See table 2.) For example, USPS opened a CPU in Frederick, Maryland, to better meet deman
1,747
d and reduce customer wait times in lines at the local post office about one-half mile away. Conversely, about 14 percent of CPUs are located 5 miles or more from the nearest post office, showing how CPUs can be used to provide services where post offices are not conveniently located, such as a CPU in rural Vigo Park, Texas, that is located 16 miles from the nearest post office. Similarly, USPS opened a CPU in Aubrey, Texas, located about 5 miles from the nearest post office, in order to serve customers in
1,748
a fast growing area. Consistent with the majority of CPUs’ being within 2 miles of a post office, CPUs are also more likely to be in urban than rural areas, and recent CPU openings further demonstrate this pattern. As shown in figure 4, more than 60 percent of CPUs active as of March 30, 2012, were in urban areas, as defined by the Rural-Urban Commuting Area codes we used for this analysis.CPUs to reduce the time customers have to wait in line at a post office, more often in urban areas. Furthermore, more t
1,749
han three-fourths of new CPUs in fiscal year 2011 were in urban locations. This suggests that CPUs may be most viable in urban areas with higher populations and customer traffic. Our analysis shows that CPUs are rarer in suburban, large-town rural, and small-town rural locations. In recent years, USPS has intentionally shifted its means of compensating CPUs from fixed-price contracts—in which compensation to CPUs is a fixed amount regardless of sales—to performance-based contracts—under which compensation t
1,750
o CPUs is a percentage of the CPU’s postal sales—resulting in potentially greater revenue and less financial exposure to USPS. (See fig. 7.) According to USPS officials, since 2002, USPS has entered into performance-based contracts for most new CPUs and has converted many fixed-price contracts to performance-based. The purpose of the shift is to incentivize CPU operators to market postal products and services to increase postal revenues. CPUs with fixed-price contracts have limited incentive to sell more po
1,751
stal products, since their compensation is the same regardless of their sales. Furthermore, since USPS compensates CPUs with performance-based contracts a percentage of the CPU’s sales, USPS does not compensate these CPUs more than it receives in revenues, a situation that can happen with CPUs with fixed-price contracts. The total revenues USPS received from sales of postal products and services at CPUsyear 2007 to $611 million in fiscal year 2011, as shown in figure 8. However, as mentioned earlier, USPS’s
1,752
revenues from post offices declined about 22 percent during this period. The decline in CPU revenues is part because of the decrease in the number of CPUs, as average CPU revenues decreased only 2 percent during this time. The downward trend in mail volume was also a factor, according to USPS officials. Several CPUs we visited experienced declining sales in recent years. For example, a CPU in Cedar Lake, Indiana, saw CPU revenues decline 17 percent from fiscal year 2007 to 2011. Several USPS district retai
1,753
l managers cited CPUs that closed because of low sales. For example, a CPU in Texas closed because neither the CPU nor the primary business generated sufficient revenue for the operator to stay in business. Our analysis of USPS data found that CPUs with lower than declined about 9 percent from $672 million in fiscal average revenues were more likely to close than were those with higher revenues. On average, CPUs that closed from fiscal years 2008 to 2011 generated roughly 26 percent less revenue on average
1,754
in the year prior to closure than the average CPU revenue for that year. Individual CPU revenues vary widely, as shown in figure 9. On average, USPS’s revenue from individual CPUs averaged about $160,000 in revenue in fiscal year 2011, but a substantial number (41 percent) generated less than $50,000. Moreover, low revenue CPUs are more likely to be located in rural areas where population is sparse and demand for services is lower; 22 percent of small-town rural and large-town rural CPUs had revenues under
1,755
$5,000 in fiscal year 2011. High-revenue CPUs—such as the 7 percent that earned $500,000 or more in fiscal year 2011—are mostly located in urban areas where demand is likely higher and post offices are more likely to have long wait times. For instance, we visited one CPU in downtown Los Angeles with $1.8 million in revenues in fiscal year 2011. The ability to generate high revenues at this CPU led it to increase capacity by adding postal windows to keep pace with demand. USPS compensation to CPUs increased
1,756
about 6 percent from $75.4 million in fiscal year 2007 to $79.9 million in fiscal year 2011. However, USPS compensation to CPUs has decreased every fiscal year from 2008 to 2011. (See fig. 10). According to USPS officials, the increase in compensation from fiscal years 2007 and 2008 was because of larger numbers of performance-based contracts, fewer public service contracts, which are generally less expensive, individual CPUs’ petitions for increased compensation because of increased cost of doing business,
1,757
and economic conditions. The subsequent decline in USPS compensation to CPUs from fiscal years 2008 to 2011 was because of declining numbers of CPUs during the time. As with CPU revenues, USPS compensation to individual CPUs varies widely. (See fig. 11.) For example, 326 CPUs received no more than $100 in annual compensation in fiscal year 2011. On the other hand, that same year, 55 high-revenue CPUs with performance-based contracts received over $100,000 in compensation. In fiscal year 2011, USPS compensa
1,758
ted CPUs an average of about $21,000, but compensated more than a quarter of CPUs less than $5,000. As USPS undertakes actions to achieve a sustainable cost structure, it will be important to understand the implications of CPUs for USPS’s costs and revenues. Currently, USPS retains most of the revenues generated by CPUs, its major expense being compensation payments to CPU operators. As we described previously, in fiscal year 2011, USPS earned a total of $610.5 million in revenues from CPUs and, in return,
1,759
compensated CPUs a total of $79.9 million, allowing USPS to retain $530.6 million in CPU revenues. Measured in another way, after compensating CPUs, USPS retained $0.87 of every dollar of CPU revenues. However, for individual CPUs, the amount of revenues USPS retains after compensating the CPU varies significantly. USPS’s target for individual CPUs is to retain, after compensation, $0.80 for every dollar in revenues. percent of the roughly half of CPUs that have fixed-price contracts. (See fig. 12.) Moreove
1,760
r, for 23 percent of CPUs with fixed-price contracts in fiscal year 2011, USPS did not retain any revenues as it compensated the CPU an amount greater than the revenue USPS received from the CPU. Most of these CPUs were in rural areas. Forty-nine percent of small-town rural CPUs with fixed-price contracts generated less revenue for USPS than the compensation USPS provided in fiscal year 2011. According to USPS officials, while USPS does not retain any revenue from these CPUs after compensating them, operati
1,761
ng a post office in the same locations would be more onerous from a cost perspective. Because USPS compensates the roughly 45 percent of CPUs with performance- based contracts with a percentage of their sales—usually between 9 and 12 percent—USPS’s revenues from CPUs with performance-based contracts will, by definition, always be greater than the amount of USPS compensation to them. USPS officials said that they review CPUs in which USPS retains less than $0.80 per dollar of revenue and attempts to take act
1,762
ion to decrease CPU compensation or terminate the CPU if necessary. USPS is embarking on a substantial makeover of its retail network, including reducing hours of service at thousands of underutilized post offices and expanding the use of retail alternatives through partnerships with national and regional retailers. According to USPS officials, at this time there are no plans to strategically increase the number of CPUs to help enhance service in the changing postal retail landscape. USPS officials said tha
1,763
t they plan to continue to use CPUs to meet specific local needs identified by local and district officials. At the same time, pending legislation in the Senate would require USPS to consider opening CPUs as replacements for post offices that it closes.pared down its plans to close post offices by instead reducing their hours, to the extent that USPS closes post offices in the future, this requirement may put more pressure on USPS to open more CPUs. Furthermore, some district retail managers we spoke with s
1,764
aid that they see a potentially larger role for CPUs in the future as USPS transforms its traditional retail network. However, we identified a number of challenges USPS might face in its future use of CPUs: Limited Potential Business Partners. USPS may face limited private interest in opening CPUs in certain areas. USPS planned to open thousands of Village Post Offices, which, similar to CPUs, involve partnerships with private businesses, by the end of 2012. However, as of August 20, 2012, USPS has opened o
1,765
nly 41 Village Post Offices in part because of a lack of interested private parties. USPS officials said that this lack of interested parties is because in some rural areas, there may not be any businesses to host a Village Post Offices and in other rural areas, businesses may not want to partner with USPS in what some communities may perceive as a reduction in services they receive. In addition, some district retail managers told us there are a number of reasons that some interested businesses do not becom
1,766
e CPUs, including financial instability and not wanting to meet the conditions of new CPU contracts, such as space requirements or prohibiting sales of competitors’ products and services. As a result, district staff are not always able to open as many new CPUs as they would like. Limited Staff Resources in USPS Districts. As we have previously mentioned, local and district-level USPS officials identify and justify the need for new CPUs, determining when and where to approach businesses as potential CPU part
1,767
ners. Some USPS district retail managers we spoke with told us that although there are unmet needs for CPUs in their districts, compared to prior years, they now have fewer staff and less time to seek out opportunities for new CPUs. Given the resources required to seek opportunities and open new CPUs, USPS may be unable to meet all local needs for CPUs with existing resources. Risk of Service Disruptions from CPU Closures. Because CPUs can close at any time—unlike post offices, which must undergo a lengthy
1,768
review process including a public comment period prior to closure—there is a risk in relying on CPUs to provide service, especially in underserved areas where there may be a limited number of potential CPU partners and other post office alternatives. As discussed earlier, CPU operators can decide to close their CPUs for a variety of reasons. Although CPU contracts require CPUs to provide 120 days notice to USPS before closing, some district retail managers we spoke with said that CPU operators often provide
1,769
much less notice, often as little as one week. Given the other challenges in opening new CPUs, USPS may have trouble replacing the lost service from unexpectedly closed CPUs. CPUs can play an important role in helping USPS provide universal service as it cuts costs to improve its financial condition—at times two conflicting goals. CPUs can help USPS reach customers in convenient locations during convenient hours at a potentially lower cost than through post offices. USPS data show that an increasing propor
1,770
tion of retail revenue is generated through channels other than post offices, which indicates a growing level of customer acceptance of these non-traditional means of accessing postal services. While USPS plans to continue to use CPUs as one alternative to post offices to fill local needs for postal services, it is exploring planned national and regional partnerships to more broadly expand access to convenient retail alternatives nation-wide. As USPS develops these regional and national partnerships, reduce
1,771
s hours of service at many post offices, and continues to use CPUs to fill specific local needs, it is important for USPS to consider CPUs’ continuing role in USPS’s evolving national retail network. We recommended in November 2011 that USPS develop and implement a retail network strategy that would address USPS customer access to both post offices and retail alternatives.officials told us that as of July 2012, the agency is in the process of finalizing this retail strategy. We continue to believe, as we st
1,772
ated in November 2011, that it is important that such a strategy discuss how USPS plans to increase its use of retail alternatives—including CPUs— while considering significant changes to its network of post offices and the means through which it provides access to USPS’s customers. As USPS continues to develop this retail strategy, we believe that USPS can capitalize on growing acceptance of retail alternatives by using information about CPUs to inform its decisions. For example, by considering factors, su
1,773
ch as the distance of CPUs to existing post offices, CPU hours and days of service, and USPS’s costs of compensating CPUs, USPS could better inform its retail strategy in order to make better strategic use of CPUs in its future retail network, which will likely include reduced hours at thousands of post offices. We provided a draft of this report to USPS for review and comment. USPS provided a written response (see appendix III) in which they discussed USPS’s efforts beyond CPUs to provide customers with su
1,774
fficient and convenient access to its products and services through other types of partnerships and alternatives to post offices. We are sending copies of this report to the appropriate congressional committees, the Postmaster General, and other interested parties. In addition, the report will be available at no charge on GAO’s Web site at http://www.gao.gov. If you or your staffs have any questions regarding this report, please contact me at (202) 512-2834 or [email protected]. Contact points for our Office
1,775
s of Congressional Relations and Public Affairs may be found on the last page of this report. GAO staff who made key contributions to this report are listed in appendix IV. To determine how contract postal units (CPUs) supplement the U.S. Postal Service’s (USPS’s) network of post offices, we analyzed data from USPS’s contract postal unit technology (CPUT) database. This database contains information for individual CPUs, including location, contract number, revenues, compensation, CPU contract type (fixed-pr
1,776
ice or performance-based), contract termination dates, and if the CPU is in active service. Location data for each CPU in CPUT includes a physical address including city, state, and ZIP+4 code. USPS provided us these data on March 30, 2012. In determining the number of active CPUs, we encountered some duplicate CPU records. To avoid double counting, we used the CPU contract number to keep the record for only the oldest contact associated with each CPU. Based on the physical address, including ZIP+4 code, we
1,777
determined the location type for each CPU by using the Department of Agriculture’s Economic Research Service’s Rural-Urban Commuting Area (RUCA) codes. RUCA codes classify a given location based on patterns of urbanization, population density, and daily commuting patterns. We classified CPU locations as one of four types: urban, suburban, large-town rural, and small-town rural. We also determined how many CPUs were located in each state using the state data in CPUT. Based on data from CPUT, we identified w
1,778
hich CPUs closed from fiscal years 2007 to 2011. We identified which CPUs opened during this time period based on contract start dates from USPS’s Contract Authoring Management System (CAMS). Contract start dates generally do not match the date that a CPU opens because, according to USPS officials, it usually takes 4 to 6 months for a CPU to open after USPS initiates a solicitation. We determined this date to be a reasonable approximation of when a CPU opens. We determined the number of CPUs that opened or
1,779
closed in each fiscal year by counting the number of contract start dates and closure dates for each year. In addition, we obtained data from USPS’s facilities database (FDB) on all post office locations including physical street address, city, state, and ZIP+4. USPS provided us these data on December 19, 2011. We determined the distance between the list of active CPUs as of March 30, 2012, and post offices as of December 19, 2011, using the latitude and longitude for each CPU and each post office and measu
1,780
ring the straight- line distance between the two points. We then determined which post office was closest to each CPU and by what distance. We also counted the number of post offices in each state and, along with using data on the number of CPUs in each state, to determine the number of CPUs per 100 post offices in each state. We analyzed FDB data to determine the number of hours of service per day and per week for each CPU and post office, and how many locations are open at certain times, such as on Sunday
1,781
s. USPS provided these data for CPUs on June 7, 2012, and for post offices on June 27, 2012. We also visited 10 CPUs in the following regions: Chicago, Illinois; Dallas- Fort Worth, Texas; Southern California; and Washington, D.C. We selected those regions and the 10 CPUs to ensure diversity in geographic location, location type (urban, suburban, and rural), CPU revenue levels, and type of CPU contract (fixed-price and performance-based). We also selected locations close to GAO office locations in order to
1,782
minimize the use of travel funds by GAO staff on this engagement. During our visits, we interviewed each CPU operator. We also interviewed district retail managers in each of the USPS districts responsible for managing these CPUs. During these interviews as well as interviews with USPS headquarters staff in charge of managing the CPU program, we discussed the reasons for and benefits from using CPUs, the reasons why CPUs have closed, and factors that affect CPU revenues and compensation. We also reviewed GA
1,783
O reports and USPS documents detailing the CPU program, including USPS guidance on CPUs and standard CPU contracts. To determine USPS revenue from CPUs and USPS’s compensation to them from fiscal years 2007 to 2011, we analyzed data from CPUT. We encountered numerous duplicate records for a single CPU address. To avoid double counting, we merged all financial data for a given contract number into a single record by summing up data for each unique contract number. As CPUT stores CPU revenue and compensation
1,784
data on a monthly basis, we summed monthly data to determine the total revenues and USPS compensation for each CPU for each fiscal year. We determined the amount of revenues USPS retains after compensating CPUs in each fiscal year by subtracting CPU compensation from CPU revenues and dividing that by CPU revenues. Finally, we linked data from CAMS on contract start dates to this financial data from CPUT by using the contract number for each CPU. As a result, we were able to determine the revenues and USPS c
1,785
ompensation to each CPU for CPUs that opened in each fiscal year. We did the same for closed CPUs by using CPU closure dates included in CPUT. We assessed the reliability of each of the data sources we used by interviewing responsible USPS officials about procedures for entering and maintaining the data and verifying their accuracy. We manually reviewed all data provided by USPS for any obvious outlying data. After reviewing this information, we determined that the CPUT data were sufficiently reliable for e
1,786
valuating revenue and compensation trends, closure dates, and CPU locations. We did find that CPUT reported outlying data on revenues in certain months for four CPUs in fiscal year 2007. To address these outlying data, we averaged the revenues for each of the four CPU in the other months, where reported revenues seemed normal, and assumed that the CPU earned the average level of revenue in the outlying months. We determined that the CAMS data were sufficiently reliable for evaluating CPU start dates. We det
1,787
ermined that the FDB post office and CPU hours-of-service data were sufficiently reliable for overall comparison purposes. As previously stated, the FDB included hours-of- service data for 3,320 CPUs as of June 27, 2012, 6.3 percent less than 3,542 CPUs indicated by our analysis of CPUT as of March 30, 2012. In discussing the discrepancy with USPS officials, we determined that there was no indication that the CPU records missing from the FDB differed from the general population and were therefore unlikely t
1,788
o affect the outcome of our analysis. To determine challenges USPS might face if it increases its use of CPUs, we reviewed relevant legislation, USPS documents related to managing CPUs, prior GAO reports, and USPS Office of Inspector General reports. We also interviewed USPS officials responsible for implementing the CPU program, CPU operators, and USPS district retail managers at the sites and districts discussed earlier regarding current CPU operations and challenges the CPU program might face going forwa
1,789
rd. Table 5 provides the number of Contract Postal Units (CPUs) and post offices in each state, as well as the number of CPUs per 100 post offices in each state as a measure of how reliant each state is on CPUs for providing access to postal services. Lorelei St. James, (202) 512-2834 or [email protected]. In addition to the individual named above, Heather Halliwell, Assistant Director; Patrick Dudley; John Mingus; Jaclyn Nelson; Joshua Ormond; Matthew Rosenberg; Amy Rosewarne; Kelly Rubin; and Crystal Wesco
1,790
made key contributions to this report.
1,791
The voucher program is not an entitlement program. As a result, the amount of budget authority HUD requests and Congress provides through the annual appropriations process limits the number of households that the program can assist. Historically, appropriations for the voucher program (or for other federal housing programs) have not been sufficient to assist all households that HUD has identified as having worst-case housing needs—that is, unassisted households with very low incomes that pay more than 50 pe
1,792
rcent of their incomes in rent, live in substandard housing, or both. In 2009, 41 percent of the more than 17 million very low-income renters had worst-case housing needs, according to HUD. The primary problem affecting these renters was rent burden— approximately 97 percent paid more than 50 percent of their incomes in rent. To be eligible for assistance, in general, households must have very-low incomes—not exceeding 50 percent of the area median income, as determined by HUD. Under the Quality Housing and
1,793
Work Responsibility Act of 1998 (P.L. 105-276), at least 75 percent of new voucher program participants must have extremely low incomes—not exceeding 30 percent of the area median income. Under the voucher program, an assisted household pays 30 percent of its monthly adjusted income in rent; the remainder of the rent is paid through a HUD-subsidized “voucher,” which generally is equal to the difference between (1) the lesser of the unit’s gross rent (generally, rent plus utilities) or a local “payment stan
1,794
dard” and (2) the household’s payment. The payment standard is based on the HUD-determined fair market rent for the locality, which generally equals the 40th percentile of market rents (including utilities) recent movers paid for standard-quality units. HUD annually estimates fair market rents for metropolitan and nonmetropolitan areas. Housing agencies—the state and local agencies that administer the voucher program on HUD’s behalf—can set payment standards (that is, pay subsidies) between 90 percent and 1
1,795
10 percent of the fair market rent for their areas. By determining fair market rents and setting payment standards at a rate sufficient to provide acceptable choices for voucher program participants, HUD and housing agencies essentially set the upper and lower bounds on the cost of typical, standard-quality units that voucher holders rent. Participants in the voucher program can choose to live in units with gross rents that are higher than the payment standard, but they must pay the full difference between
1,796
the unit’s gross rent and the payment standard, plus 30 percent of their income. In 2011, more than 2,400 housing agencies administered more than 2.2 million vouchers—their programs ranged in size from more than 96,000 vouchers to fewer than 5. Housing agencies are responsible for inspecting units, ensuring that rents are reasonable, determining households’ eligibility, calculating and periodically re-determining households’ incomes and rental payments, and making subsidy payments to landlords.functions, su
1,797
ch as establishing and maintaining a waiting list, processing tenant moves, conducting landlord and tenant outreach, and reporting to HUD. HUD disburses appropriated funds to housing agencies for subsidy payments to landlords and administrative expenses. In addition, housing agencies perform basic program Each year, Congress appropriates funding for subsidies for renewal (existing) and incremental (new) vouchers and administrative expenses. As part of the appropriations process, Congress outlines a formula
1,798
that determines the amount of renewal funding for which housing agencies are eligible (“eligible amount”). However, the amount Congress appropriates to the voucher program may not equal the total amount for which housing agencies are eligible under the formula. HUD is responsible for allocating program funding (“appropriated amount”) among housing agencies based on their eligible amounts. To the extent that the appropriated amount does not fully fund housing agencies at their eligible amounts, HUD reduces t
1,799
he funding housing agencies receive to fit within the appropriated amount. Housing agencies are expected to use all the subsidy funding HUD allocates for authorized program expenses (including subsidy and utility payments). However, if housing agencies’ allocated amounts exceed the total cost of their program expenses in a given year, they must maintain their unused subsidy funds in NRA (reserve) accounts. Housing agencies may use their NRA balances (subsidy reserves) to pay for authorized program activitie