For other versions of this document, see http://wikileaks.org/wiki/CRS-RL32111 ------------------------------------------------------------------------------ Order Code RL32111 CRS Report for Congress Received through the CRS Web Unemployment Compensation (UC)/ Unemployment Insurance (UI): Trends and Contributing Factors in UC Benefit Exhaustion October 10, 2003 Julie M. Whittaker Analyst in Applied Microeconomics Domestic Social Policy Division Congressional Research Service ~ The Library of Congress Unemployment Compensation (UC)/ Unemployment Insurance (UI): Trends and Contributing Factors in UC Benefit Exhaustion Summary This report examines trends and contributing factors in Unemployment Compensation (UC) benefit exhaustion rates. To counter the disincentive effect of benefit receipt, most state UC programs limit the duration of UC benefits to a maximum of 26 weeks and many require some evidence of a job search. The limited duration of UC benefits results in some unemployed individuals exhausting their benefits before finding work or voluntarily leaving the labor force. Furthermore, the availability of compensation (UC benefits) may create disincentives to search for and accept reemployment. These disincentives are alleviated by decreasing the generosity of the UC benefits and increasing the cost of benefit receipt through additional program requirements. The somewhat paradoxical policy of cutting UC benefits and increasing the burden of UC program requirements in order to help the unemployed leads to competing policy choices. Benefit adequacy is traded off against the disincentive effect; or, unemployment is increased at the expense of employment. Exhaustion rates increased over the last 30 years. There has been an important change in the relationship among unemployment and the exhaustion of benefits. Unemployment levels are lower in the 1990s than in the 1980s and 1970s for equivalent economic periods. However, these lower unemployment rates are associated with higher expected rates of benefit exhaustion. This trend has many complex contributing factors, including changing program benefit generosity and administration, workforce demographics, and general economic conditions. Research indicates that (1) increased UC coverage and lowered program requirements; (2) increased proportion of women, minorities and older workers in the workforce; (3) decreased manufacturing positions; and, (4) decreased temporary layoffs all contribute to an increase in UC benefit exhaustion. The UC benefit exhaustion rate also increases during economic recessions; after economic recovery, the UC benefit exhaustion rate decreases. This paper will be updated as events warrant. Contents Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Unemployment Compensation Benefit Exhaustion . . . . . . . . . . . . . . . . . . . . 2 Disincentive effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Trade-offs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Defining Unemployment Compensation Benefit Exhaustion . . . . . . . . . . . . 3 Conceptual Description of UC Exhaustion . . . . . . . . . . . . . . . . . . . . . . 3 Construction of UC Exhaustion Statistic . . . . . . . . . . . . . . . . . . . . . . . . 3 Caveats for UC Exhaustion Statistic . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Current Trends in UC Benefit Exhaustion . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Altered Relationship: Unemployment Rate, Exhaustion Rate, Average Benefit Duration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Lower Unemployment Rates but Higher UC Exhaustion Rates . . . . . . 7 Potential Causes of Increased UC Benefit Exhaustion . . . . . . . . . . . . . . . . . 9 Program benefit generosity and program administration . . . . . . . . . . . . 9 Workforce Demographics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Economic Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 List of Figures Figure 1. Average Annual Exhaustion Rate, Average Annual Duration of Benefit Collection, and Average Total Unemployment Rate, by Month, June 1972- June 2003 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 List of Tables Table 1. Average Annual UC Exhaustion Rate, Weeks of Benefit Duration, and Total Unemployment Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Unemployment Compensation (UC)/ Unemployment Insurance (UI): Trends and Contributing Factors in UC Benefit Exhaustion Background Unemployment Compensation1 (UC) benefits provide temporary income support to covered workers who have lost their jobs through no fault of their own.2 This temporary partial replacement of lost wages is intended to provide sufficient income for the unemployed and the aggregate benefits help stabilize economic activity. The UC program is a federal-state partnership and is administered by the states. Federal standards are minimal and, within very broad guidelines, almost all decisions regarding UC are made by the states; they determine qualifying requirements, benefit amounts, and maximum duration of regular benefits. The program's compensation schedule varies by state, typically replacing approximately 50-60% of eligible workers' wages. Generally, maximum benefits are capped at approximately 50-67% of the average wage for workers in the state. Under all state laws, workers' benefits depend on work experience in covered employment. While almost all workers are covered by UC,3 the percentage of unemployed workers who receive UC benefits varies widely by state and by economic conditions, ranging on average from 30-50%.4 Most states limit maximum potential collection of benefits to 26 weeks; Massachusetts and Washington limit the maximum potential duration to 30 weeks. Eight states5 have uniform (26 weeks) benefit duration while the 1 In law, this program is called the Unemployment Compensation program. However, it is commonly referred to as the Unemployment Insurance program, reflecting its social insurance design. 2 Workers who have quit or have been fired may qualify for UC benefits after a waiting period in a few states. 3 Almost all wage and salary workers are covered by UC with the exception of self- employed workers. Veterans with recent service in the Armed Forces and civilian federal employees are covered by separate programs, with states paying benefits from federal funds as agents of the federal government. 4 Stephen Wandner and Andrew Stettner, "Why are Many Jobless Workers not Applying for Benefits?" Monthly Labor Review, June 2000, pp. 21-32. Wandner and Stettner (2000) find that most of those who do not file do so because they think they are not eligible or because they are optimistic about finding employment. 5 Connecticut, Hawaii, Illinois, Maryland, New Hampshire, New York, Vermont, and West Virginia. CRS-2 remaining states limit potential duration based on calculations that take into account work experience and earnings history. The limited duration of UC benefits results in some unemployed individuals exhausting their UC benefits before finding work or voluntarily leaving the labor force for other activities such as retirement, disability, family care, or education. Generally, the exhaustion of UC benefits is not considered to be a reliable indicator of UC abuse by an individual. In fact, empirical research suggests that workers who exhaust benefits search at similar, or higher levels, of intensity than workers who do find employment before benefit exhaustion.6 Unemployment Compensation Benefit Exhaustion Disincentive effects. The job-search behavior of the unemployed can be influenced by changing the timing, generosity, and duration of UC compensation.7 Economic theory suggests that unemployment compensation will have a negative effect on the recipients' willingness to search for and accept a new job. Higher benefit levels and easier program requirements for benefits will cause recipients to be less willing to accept jobs and may induce some of the employed to quit and become unemployed.8 Thus, the availability of compensation (UC benefits) may create a disincentive9 to search for and accept reemployment. Research in this area examines how these disincentives are alleviated by decreasing the generosity of the UC benefits and increasing the cost of benefit receipt through additional program requirements.10 Trade-offs. The somewhat paradoxical policy of cutting UC benefits and increasing the burden of UC program requirements in order to help the unemployed leads to competing policy choices. Benefit adequacy is traded off against the disincentive effect; or, unemployment is increased at the expense of employment. 6 Walter Corson and Mark Dynarski, A Study of Unemployment Insurance Recipients and Exhaustees: Findings from a National Survey, Unemployment Compensation Occasional Paper 90-3. U.S. Department of Labor Employment and Training Administration, 1990. (Hereafter cited as Carson and Dynarski, A Study of Unemployment Insurance.) 7 Cynthia Rogers, "Expectations of Unemployment Insurance and Unemployment Duration." Journal of Labor Economics, 1998, vol. 16, no. 3, pp. 630-666; Patricia Anderson and Bruce Meyer, "Time Varying Effects of Recall Expectation, a Reemployment Bonus and Job Counseling on Unemployment Durations," Journal of Labor Economics, vol. 10, Jan. 1992, 99-115; Jennifer Hunt, "The Effect of Unemployment Compensation Duration in Germany, Journal of Labor Economics 1995, vol. 13, no. 1, pp. 88-120. 8 Anthony Atkinson and John Micklewright, "Unemployment Compensation and Labor Market Transitions: A Critical Review," Journal of Economic Literature, vol. 29, pp. 1679- 1727, Dec. 1991. In practice, an individual who quits a job may be disqualified or face reduced UC benefits depending upon state program rules. 9 See Gary Burtless, "Unemployment Insurance and Labor Supply: A Survey," In Unemployment Insurance, W. Lee Hansen and James Byers, eds., Madison: University of Wisconsin Press, 1990. 10 These may include weekly proof of job search, more in-person visits to the unemployment office, enrollment in job search or training classes. CRS-3 To counter the disincentive effect of benefit receipt, state UC programs limit the duration of UC benefits and many require some evidence of a job search. Recognizing that during economic recessions the unemployed find it more difficult to find new positions, many states employ one of three unemployment rate triggers to increase the maximum potential duration for these special periods through the permanent Extended Benefit (EB) program. In addition, Congress often has passed temporary programs to supplement the UC system during economic slowdowns.11 Defining Unemployment Compensation Benefit Exhaustion The Department of Labor (DoL) estimates the rate of UC benefit exhaustion and reports annual averages as well as a moving yearly average reported on a monthly basis.12 Conceptual Description of UC Exhaustion. Abstracting from actual available data, the exhaustion rate should measure the proportion of all UC benefit recipients who exhaust their receipt eligibility before finding a job. That is, (Hypothetical) UC Exhaustion Rate= UC Exhaustees week t UC recipients that would have been eligible to exhaust in week t However, current administrative constraints in data collection do not allow for this exact calculation. Each state collects and tracks their own UC benefit data. Aggregate weekly information from the state is passed on to the US Department of Labor which then constructs an estimate of the UC exhaustion statistic. Construction of UC Exhaustion Statistic. The UC benefit exhaustion rate is constructed from two statistics that are also reported separately. The first is the average number of claimants drawing the final payment of their original entitlement for a given program in the previous year (t). The second is the average number of first-week UC benefit recipients for the year beginning 6 months before the year t (t-6 months). According to the DoL this is to account for the normal flow of claimants through the program.13 Note that the tally of the number of first-week UC recipients is not the actual number of UC recipients that have the potential to exhaust their benefits in week t but rather represents a construct of what it would be if all UC recipients were eligible to collect benefits for exactly 26 weeks. The DoL calculation is UC Exhaustion Rate= UC Exhaustees t UC First-Week Recipients t-6 months. 11 See CRS Report RS 21397, Unemployment Benefits: Temporary Extended Unemployment Compensation (TEUC) Program, by Celinda Franco. 12 See statistics at [http://www.workforcesecurity.doleta.gov/unemploy/claimssum_us.asp]. 13 See at [http://workforcesecurity.doleta.gov/unemploy/content/data_stats/datasum01/ 3rdqtr/gloss.asp]. CRS-4 For example, the exhaustion rate reported for July 2003 will be the number of UC benefit recipients who exhaust their potential durations (UC Exhaustees) from August 2002 until July 2003 divided by the number of UC first-week recipients reported from February 2002 until January 2003. Caveats for UC Exhaustion Statistic. Since not all UC benefit recipients are eligible to receive a full 26 weeks of benefits, this statistic may overstate the rate of UC benefit exhaustion at the beginning of an economic recession (since the lagged estimated number first-week recipients may be smaller than the actual number of first-week recipients who would have been eligible for exhaustion in that week). Likewise, the statistic may understate the rate of UC benefit exhaustion as the economy begins to improve (since the lagged estimated number of first-week recipients may be larger than the actual number of first-week benefit recipients who would have been eligible for exhaustion in that week). This is because the relationship of UC exhaustions with the business cycle may vary and be substantially higher or lower at different points of the business cycle. In particular, as recessions end, the severity of the problem of increased benefit exhaustion may be underestimated. Current Trends in UC Benefit Exhaustion Over the past 30 years the percentage of the unemployed receiving unemployment compensation benefits has declined steadily.14 During this period, there also has been an increase in the percentage of UC benefit recipients exhausting their benefits. Figure 1 displays the monthly average total unemployment rate,15 monthly annual average weeks of benefit collection, and monthly annual average UC exhaustion rate from 1972-2003. The `monthly' annual average should not be misused as `real time' information. Instead, the monthly statistic is an average of the exhaustion rates for the previous year.16 In Figure 1, the average annual exhaustion and benefit duration rates are plotted at the midpoint of their span. (That is the annual exhaustion rate for February 2001-January 2002 is placed at its midpoint, August 2001.) The shaded areas in Figure 1 represent recessionary periods (as determined by the National Bureau of Economic Research). 14 Daniel McMurrer and Amy Chasanov,"Trends in Unemployment Insurance Benefits" Monthly Labor Review, Sept. 1995, pp. 30-39. While the percentage of the unemployed who receive UC benefits generally increases during economic recessions, these increases have also lowered over time. For monthly trends, see Section 3, "Unemployment Compensation," Table 4-2 -- Insured Unemployment as a percent of total unemployment, by month, selected years 1967-2002 in the Greenbook, 2003, at [http://waysandmeans. house.gov/Documents.asp?section=813.] 15 The average total unemployment rate is formed from estimates that include all workers who are unemployed and is not based upon UC eligibility. The statistic has been seasonally adjusted and may be found at [http://www.bls.gov/cps/] by selecting "most requested statistics." 16 While this has the unfortunate result of `lagging' the exhaustion rate, actual monthly exhaustion rates are extremely volatile and calculation of actual monthly changes in rates are of little value for policy decisions. CRS-5 Figure 1 charts the declining average rate of total unemployment as well as the increasing average exhaustion rate. Figure 1 also depicts how the relationship among exhaustion rates, average benefit duration, the total unemployment rate and the economic recessions and recoveries underwent substantial change. In the 1970s and early 1980s as the recessions drew to a close the average exhaustion rate peaked as did the unemployment rate. The unemployment rate soon began to decline after the end of the recession, and was generally followed by declines in the exhaustion rate and average benefit duration. Altered Relationship: Unemployment Rate, Exhaustion Rate, Average Benefit Duration. However, after the 1980s the relationship among these statistics and recessionary periods was substantially altered. Unemployment rates continue to rise or remain stable after the end of the recessionary periods rather than fall immediately after the end of the recession. Likewise, this trend is repeated with exhaustion rates and average benefit durations in this period. Furthermore, while in the 1970s and 1980s exhaustion rates eventually declined to under 30%, in the 1990s rates were always above 30%. CRS-6 Figure 1. Average Annual Exhaustion Rate, Average Annual Duration of Benefit Collection, and Average Total Unemployment Rate, by Month, June 1972- June 2003 50 Recession Annual Ave. Exhaustion Rate Annual Ave. Weeks Benefit Duration Total Unemployment Rate 45 40 35 30 Percent 25 20 15 10 5 0 Jun-72 Jun-73 Jun-74 Jun-75 Jun-76 Jun-77 Jun-78 Jun-79 Jun-80 Jun-81 Jun-82 Jun-83 Jun-84 Jun-85 Jun-86 Jun-87 Jun-88 Jun-89 Jun-90 Jun-91 Jun-92 Jun-93 Jun-94 Jun-95 Jun-96 Jun-97 Jun-98 Jun-99 Jun-00 Jun-01 Jun-02 Jun-03 Month Source: CRS table compiled from DoL data. Note: Recession dates are as determined by National Bureau of Economic Research (NBER): Nov. 1973 -- Mar. 1975; Jan. 1980 -- July 1980; July 1981 -- Nov. 1982; July 1990 -- Mar. 1991; and Mar. 2001 -- Nov. 2001. CRS-7 Table 1 displays the average total unemployment rate, average weeks of benefit collection, and average UC exhaustion rates from 1972-2002. The average total unemployment rate was at its peak at the end of the 1981-82 recession and has generally been declining with small increases during the recession of the early 1990s and the current recession. The average total unemployment rate for 2002 was 5.8%, as compared to an average of 6.25% during 1972-2002. The average duration of UC benefit receipt slowly increased over this period. The average annual duration of benefits peaked at 17.5 weeks (with a total unemployment rate of 9.5%) as the 1981-82 recession ended. UC benefit duration in 2002 averaged slightly lower, at 16.5 weeks (but with a total unemployment rate of 5.8%). This exceeds the 14.6 week average for the period 1972-2002 by almost 2 weeks. The average rate of exhaustion for 2002 was 42.5%, the highest rate recorded since 1972. This compares to an overall average rate of exhaustion experienced during the 1972-2002 period of 32.2%. Lower Unemployment Rates but Higher UC Exhaustion Rates. As previously outlined during the discussion of Figure 1, Table 1 demonstrates that there has been an important change in the relationship among unemployment and the exhaustion of benefits. Unemployment levels are lower in the 1990s than in the 1980s and 1970s for equivalent economic periods. However, these lower unemployment rates are associated with higher expected rates of benefit exhaustion. CRS-8 Table 1. Average Annual UC Exhaustion Rate, Weeks of Benefit Duration, and Total Unemployment Rate Average annual UC Average annual weeks Average annual total exhaust rate of UC duration unemployment rate Year (percent) (weeks) (percent) 1972 30.0% 14.2 5.5% 1973 27.7% 13.4 4.8% 1974 31.0% 12.6 5.5% 1975 37.8% 15.7 8.3% 1976 37.8% 14.9 7.6% 1977 32.5% 14.2 6.9% 1978 26.7% 13.3 6.0% 1979 26.7% 13.1 5.8% 1980 33.2% 14.9 7.0% 1981 32.4% 14.4 7.5% 1982 38.5% 15.9 9.5% 1983 38.4% 17.5 9.5% 1984 34.2% 14.4 7.4% 1985 31.2% 14.2 7.1% 1986 32.2% 14.5 6.9% 1987 30.6% 14.6 6.1% 1988 28.5% 13.7 5.4% 1989 28.0% 13.2 5.3% 1990 29.4% 13.4 5.5% 1991 34.8% 15.4 6.7% 1992 39.9% 16.2 7.4% 1993 39.2% 15.9 6.8% 1994 36.3% 15.5 6.1% 1995 34.3% 14.7 5.6% 1996 33.4% 14.9 5.4% 1997 32.8% 14.6 4.9% 1998 31.8% 13.8 4.5% 1999 31.4% 14.5 4.2% 2000 31.8% 13.7 4.0% 2001 34.1% 13.8 4.8% 2002 42.5% 16.5 5.8% Source: U.S. Dept. of Labor, Employment and Training Administration, Office of Workforce Administration CRS-9 Potential Causes of Increased UC Benefit Exhaustion There are many factors with complex interactions contributing to exhaustion rates. These factors include program benefit generosity and program requirements, workforce demographics, and economic conditions. Program benefit generosity and program administration. More generous state benefits and lower levels of program requirements increase UC benefit exhaustion rates. Generally, the more generous the UC benefit and the less onerous the process for receipt, the greater likelihood of benefit receipt and likewise benefit exhaustion. The most recent significant increases in UC coverage were in the 1970s when state and local government employees, many household (domestic) workers, and employees of small businesses were covered.17 Although many state policy changes have restricted eligibility, individual (nominal) wages have increased due to inflation, allowing more individuals to reach the minimum earnings thresholds. In the 1980s numerous state UC laws were changed to restrict eligibility and reduce benefit levels, partially in response to the 1982 requirement to repay federal loans to state UC trust funds with interest. In 1986 all unemployment benefits became subject to federal income taxation. In the 1990s, in response to lowered federal administrative funding, program rules that required intensive monitoring such as weekly proof of job search and weekly attendance requirements were relaxed or replaced with telephone responses or internet reporting systems in some states. In addition, some states began to expand benefits to workers with more limited work histories through the alternative wage bases for UC benefit calculations. The variance in potential durations of receiving UC insurance benefits has implications for the exhaustion rate. Eight states have uniform duration of benefits18 of 26 weeks while others have varying duration19 based upon an applicant's earnings and labor market history. Research estimates find that states with higher average potential duration of benefits have a lower percentage of exhaustees -- after controlling for unemployment levels.20 For example, in a non-recessionary period of the late 1980s, Carson and Dynarski (1990) find that 26% of exhaustees had 17 Daniel McMurrer and Amy Chasanov, "Trends in Unemployment Insurance Benefits," Monthly Labor Review, Sept. 1995, pp. 30-39. 18 Connecticut, Hawaii, Illinois, Maryland, New Hampshire, New York, Vermont, and West Virginia. 19 Alabama, Arkansas, Arizona, California, Colorado, Delaware, District of Columbia, Florida, Georgia, Idaho, Indiana, Iowa, Kansas, Kentucky, Louisiana, Maine, Massachusetts, Michigan, Minnesota, Mississippi, Missouri, Montana, Nebraska, Nevada, New Jersey, New Mexico, North Carolina, North Dakota, Ohio, Oklahoma, Oregon, Pennsylvania, Rhode Island, South Carolina, South Dakota, Tennessee, Texas, Utah, Virginia, Washington, Wisconsin, Wyoming. It is important to note that Indiana, Kentucky, and North Carolina also have a minimum duration specified in law. 20 Stephen Woodbury and Murray Rubin, "The Duration of Benefits" in Unemployment Insurance in the United States: Analysis of Policy Issues, Christopher O'Leary and Stephen Wandner, eds, W.E. Upjohn Institute for Employment Research: Kalamazoo, Michigan, 1997, pp. 211-283. CRS-10 potential UC durations of less than 20 weeks as compared with only 12% of nonexhaustees having such low potential durations. Workforce Demographics. In general, over the last 30 years the changing demographics of the workforce have contributed to the increased likelihood of UC benefit exhaustion. While the recession ending in November 2001 is too recent for new research to have been completed, previous research indicates that UC benefit exhaustees differ from other workers who receive UC benefits.21 UC benefit exhaustees are more likely to: have lower levels of education; be female, Hispanic, or Black; have a less stable work history and therefore qualify for lower benefits and shorter potential durations; be older; not belong to a union; not be employed in manufacturing; and not have a recall date or be more likely to be on permanent layoff. During the last economic boom, Needels et al., (2002) find that those UC recipient characteristics that increased exhaustion rates during the most recent economic boom include: an aging recipient population, an increase in Hispanic recipients, and a decrease in the proportion of UC recipients who were previously in manufacturing jobs and those who had definite recall dates. The characteristics of UC exhaustees mirrored early studies to the extent that they were more likely than non-exhaustees to be females and racial or ethnic minorities, had lower-paying, part-time jobs, or were in temporary jobs. Carson and Dynarski (1990) find the major distinction between exhaustees and nonexhaustees is the workers' expectation and ultimate likelihood of recall by their previous employer. They also find that the UC exhaustees were more likely to be female, older, black or Hispanic and to have less than a high school degree. Their research also suggests that there are two diverging demographic trends among exhaustees: those with low skills and low wages and those with high skills and high wages. Workers with low skills experienced decreasing demand for their labor, increasing the likelihood that they exhaust their UC benefits. Generally, higher skilled workers experience lower rates of unemployment than other workers. During recessions in the 1980s however, higher skilled workers experienced higher rates of unemployment than experienced during the economic expansions. Empirically, higher skilled workers require more time to find appropriate jobs. As a result, this longer search time for higher skilled workers increases the overall UC exhaustion rates during economic recessions. Economic Conditions. During economic recessions, UC benefit exhaustion increases; after economic recovery the UC benefit exhaustion rate declines. There is substantial evidence to suggest that the economic cycle does matter in which type of workers are more likely to exhaust the UC benefits. Carson and Dynarski (1990) find that the characteristics of the exhaustees depends on whether there is a recessionary economy; the proportion of exhaustees who have substantial attachment to their job is likely to rise during recessions. 21 See Karen Needels, Walter Corson, and Walter Nicholson, Left Out of the Boom Economy: UI Recipients in the late 1990s, ETA Occasional Paper 2002-03, May 2002. US Department of Labor. Also see Corson and Dynarski, A study of Unemployment Insurance. CRS-11 General economic changes have also increased UC exhaustion rates. In particular, the decline in the manufacturing sector (more likely to use temporary lay- offs than other sectors), the decline in unionization (union members are more likely to experience temporary lay-offs and are more likely to apply for benefits for shorter unemployment spells -- decreasing the overall UC exhaustion rate), and the increase in the service sector (service sector lay-offs are more likely to be permanent) all have contributed to the overall increase in exhaustion rates. ------------------------------------------------------------------------------ For other versions of this document, see http://wikileaks.org/wiki/CRS-RL32111