Poverty and Financial Distress Would Have Been Substantially Worse in 2010 Without Government Action, New Census Data Show
 Details may not sum to the total because of program overlap. That is, some people are kept above the poverty line by more than one program, but in the total they are counted only once; for other people, no single program is enough to alter their poverty status, but they are counted in the total because two or more programs combined kept them above the poverty line. In addition, as explained below, the 6.9 million total is net of the effect of certain unemployment insurance benefits that would have become available if the Recovery Act had not been enacted.
 This figure also includes some smaller unemployment insurance improvements made in the 2009 Recovery Act, including “UI modernization” provisions, that were not part of the two unemployment initiatives whose effects we were able to quantify separately.
 These figures assume, as seems likely in most cases, that people who suffered lost income during the recession due to job loss, investment losses, or other reasons generally would not have been able to avert those income losses if government assistance had not been available. The figures also do not include additional private-sector employment and earnings created by government assistance as a result of its boost to consumer purchasing power during the economic downturn.
 NAS-based poverty measures use a number of methods of updating from year to year. The main NAS measure we use here (called MSI-GA-CE) is generally updated based on the increase in the typical (median) two-adult two-child family’s consumer expenditures on basic needs (food, clothing, shelter and utilities) averaged over three years. A number of other NAS measures released by the Census Bureau use a poverty line updated only for inflation using the Consumer Price Index (CPI). Here, to be conservative, we use the 2010 poverty threshold adjusted to 2007 dollars using the CPI method. If we use the consumer expenditure (CE) method preferred by the NAS panel, our findings on the efficacy of the safety net are even more dramatic — poverty would have risen by only 1.6 million from 2007 to 2010 — rather than 8.5 million without the effect of the six initiatives — implying that the initiatives averted an even larger share of the increase in poverty.
If we use a non-NAS measure, we again find similar results. Defining poverty as cash income after taxes plus the value of SNAP benefits and housing assistance, we find that poverty rose by 3.0 million (from 11.0 percent in 2007 to 11.7 percent in 2010). Without the six initiatives, it would have risen by 8.8 million (to 13.6 percent), or about three times as much.
 The enacted payroll tax holiday would have kept 1.1 million people above the NAS poverty line if it had been in place in 2010, based on our analysis of the 2010 data. The Obama Administration has proposed an enlarged version of the payroll tax holiday that we estimate would have kept 1.7 million people above the poverty line in 2010.
 Erica Williams, Michael Leachman, and Nicholas Johnson, "State Budget Cuts in The New Fiscal Year Are Unnecessarily Harmful," Center on Budget and Policy Priorities, updated July 28, 2011 ; Liz Schott and LaDonna Pavetti, "Many States Cutting TANF Benefits Harshly Despite High Unemployment and Unprecedented Need," Center on Budget and Policy Priorities, May 19, 2011.
Census tables include eight alternative NAS-based based poverty measures, in addition to the Supplemental Poverty Measure (SPM), which is a newer refinement of those measures. We cannot use the Supplemental Poverty Measure in this analysis because the Census Bureau has not yet released the detailed data needed for reproducing the measure and the measure is not available for years before 2009. Therefore, we use the NAS measure that most closely corresponds to the original NAS recommendations, issued in 1995. That measure is referred to in Census tables as "MSI-GA-CE," which stands for Medical-out-of-pocket expenditures Subtracted from Income, with Geographic Adjustment of the poverty threshold and year-to-year updating of the poverty line in accordance with Consumer Expenditure data on basic needs expenditures of typical households rather than with the Consumer Price Index.
 The Census Bureau counts tax credits as income in the tax year for which they are filed, even if the family receives them in the following year, a practice we follow.
 The survey data are merged with an additional data file from the Census Bureau's Experimental Poverty Measures website that includes geographic adjustments and other poverty measurement information.
 We identify unemployment compensation recipients eligible for EUC benefits between January and December of 2010, based on complete data on the length of the worker’s current unemployment spell, reported in the monthly basic CPS file for February, March, and April 2011. In general, we count the participant’s 27th through 80th weeks of unemployment in 2009 as eligible for EUC assistance, because regular state unemployment benefits run out after 26 weeks. (In Massachusetts and Montana, regular state unemployment benefits last longer, a difference we account for.) We consider a person’s EUC benefit to be a percentage of their reported annual total unemployment income, based on the share of their reported unemployment weeks that appear to be eligible for EUC. For example, if a person reports an unemployment spell that began in the first week of April, 2010, we assume that his or her regular unemployment benefits expired 26 weeks later, and the EUC benefits lasted the remaining 13 weeks of the year through December 31, so the EUC benefits were 13 / (26+13) or 33 percent of his or her total unemployment benefits for the year.
 For a family of four, the increase during calendar year 2010 was $85 per month from January to September and, after the start of a new federal fiscal year on October 1, $86 per month from October to December, yielding an average of $85.25 per month. We multiply the appropriate monthly figure for each family by the number of months of SNAP benefits reported by the household and cap the result so that it does not exceed the family’s reported total SNAP income.
 We also check that our method is not overstating the average duration of long-term unemployment benefits. In our method, the average EUC spell is 24 weeks, similar to or slightly lower than what program records suggest (25 weeks).
 BEA records suggest the credit provided about $38.6 billion in reduced income tax withholding in 2010 (for tax year 2010) while Treasury Department records show $13.9 billion in outlays during CY 2010 (corresponding with credit refunds paid for tax year 2009).
 According to BEA, the combined value of all Recovery Act refundable tax credits paid out in 2010 (generally for the 2009 tax year) plus the tax reduction under the Making Work Pay credit in 2010 equals $68 billion. Our estimates produce a slightly lower figure, $65 billion, for the value of the three Recovery Act credits in tax year 2010.