Stimulus Keeping 6 Million Americans Out of Poverty in 2009, Estimates Show
September 9, 2009
Although meant chiefly to help the broad economy, the stimulus plan Congress enacted earlier this year (the American Recovery and Re-Investment Act of 2009, or ARRA) had the important secondary effect of significantly ameliorating the recession's impact on poverty.
This analysis, which comes one day before the Census Bureau will release updated poverty figures (for 2008), examines seven of the recovery act's provisions -- two improvements in unemployment insurance, three tax credits for working families, an increase in food stamps, and a one-time payment for retirees, veterans, and people with disabilities -- and finds that they alone are preventing more than 6 million Americans from falling below the poverty line and are reducing the severity of poverty for 33 million more. Those 6 million people include more than 2 million children and over 500,000 seniors. This analysis includes state-specific estimates for California, Texas, Florida, New York, and Illinois.
These estimates are conservative. They do not include the poverty impact of many of the stimulus bill's provisions for direct assistance to households -- such as increases in funding for medical services, Pell grants, child support collection, Temporary Assistance to Needy Families, and assistance to homeless individuals. Nor do they reflect the degree to which bill is ameliorating the increase in poverty by creating and preserving private- and public-sector jobs. According to a March CBO estimate, the recovery legislation "will increase employment by 0.9 million to 2.3 million jobs by the fourth quarter of 2009."
The estimates presented here are based on annual Census data, updated to better match recent economic conditions. To estimate poverty effects, we simulate each survey family's taxes and after-tax income in 2009, with and without the seven stimulus provisions in place, and compare its income with the poverty line. Because the government's official measure of poverty considers only cash income and would therefore miss many of the tax-based and non-cash income supplements in the stimulus bill, the analysis uses a broader poverty measure recommended by the National Academy of Sciences and favored by a wide array of analysts. We also correct the tendency of the Census data to undercount receipt of certain public benefits. 
Recession Puts Broad Spectrum of Americans at Risk of Poverty
The recession has put Americans at heightened risk of poverty from layoffs, involuntary part-time employment, and long spells of joblessness. From June 2007 to June 2009, the economy lost about 6 million jobs, and the number of workers unemployed half a year (27 weeks) or longer rose by 3.1 million. More than one-third of workers unemployed for half a year or longer in our data live below the poverty line.
Few groups have been spared. For example, the combined long-term unemployment rate in two occupations that have traditionally offered relative security -- management and legal occupations -- has quintupled in the past two years.
The job losses and resulting labor market problems have left rising numbers of formerly moderate- and middle-income Americans at risk of falling into poverty.
The Recovery Act
The American Recovery and Re-Investment Act of 2009 (ARRA) aimed to shore up a collapsing economy by halting the downward spiral of spending by consumers, state and local governments, and businesses. Congress enacted it in February amid mounting job losses; the three-month cumulative job loss peaked at 2.1 million in early February -- just prior to the bill's enactment -- and was the highest in over 60 years.
To a greater degree than stimulus bills enacted in response to prior recessions, the 2009 bill was designed to reach a wide spectrum of low-, moderate-, and middle-income Americans. Policymakers included low-income families not only because they stand the greatest risk of hardship during downturns but also because of evidence that they are the most likely to spend quickly whatever money they receive, thereby pumping more money back into the economy in a timely manner.
Our analysis considers seven of the Act's temporary provisions, totaling $205 billion over five years:
- a new Making Work Pay tax credit of up to $400 for workers ($800 for a couple) earning up to $95,000 (up to $190,000 for a couple);
- an expanded Child Tax Credit for lower-income working families with children;
- an expanded Earned Income Tax Credit, including up to $629 per year in increased tax-credit benefits for a working family with three or more children and larger credits for married families to lessen the marriage penalty the EITC can otherwise impose;
- additional weeks of emergency unemployment compensation benefits (paid after a worker's 26 weeks of regular state unemployment benefits expire);
- an additional $25 per week for unemployed workers to supplement their unemployment benefits;
- a $250 one-time payment to elderly people and people with disabilities who receive Social Security, SSI, or veterans' benefits; and
- additional food stamp assistance.
Effects of the Seven Recovery Act Provisions on Poverty in 2009
This analysis examines how these policy changes will affect family income and poverty status for a nationwide sample of families. The estimates rely on Census data that are adjusted to approximate recent economic conditions, including labor market conditions in April through June of 2009 and food stamp participation levels in May 2009. For each family in the survey data, we estimate the family's 2009 income with and without the seven stimulus provisions.
Families are considered to be protected from poverty if their estimated income is below the poverty line without the stimulus provisions but above the poverty line with the provisions. We use a measure of poverty that adheres to National Academy of Sciences poverty measurement recommendations by including after-tax cash and non-cash income and subtracting child care and work expenses and out-of-pocket medical expenditures. (See Appendices A, B, and C for details of the calculation.)
The analysis finds that, in 2009, the seven stimulus provisions examined here keep an estimated 6.2 million Americans above the poverty line. They include:
- 2.4 million children under 18;
- 530,000 seniors age 65 and older;
- 1.9 million Latinos,
- 2.6 million non-Latino whites, and
- 1.4 million non-Latino blacks.
Antipoverty Effects of Individual Provisions
Some of the provisions we modeled have stronger poverty-preventing effects than others. The effects range from more than 100,000 Americans kept out of poverty by the $25-per-week increase in unemployment benefits to 1.6 million Americans kept out of poverty by the new Making Work Pay Credit. (See Table 1.)
The Child Tax Credit expansion keeps 600,000 children out of poverty. The other tax credits -- Making Work Pay (nearly 500,000) and the EITC (nearly 300,000) -- also protect many children from poverty.
|Making Work Pay Credit||1.6||0.5|
|Child Tax Credit Expansion||1.0||0.6|
|Food Stamp Increase||1.1||0.5|
|Unemployment Insurance Extension||0.7||0.2|
|Unemplyment Ins: additional $25/week||0.1||0.03|
For older Americans, the antipoverty effects of the stimulus come largely from two provisions, the $250 payment for recipients of Social Security, SSI, and veterans' compensation (which keeps nearly 350,000 seniors out of poverty) and the food stamp increase (which keeps more than 100,000 seniors out of poverty).
Act's Overall Poverty-Reducing Effects Larger Than Shown Here
The projections here of the number of Americans protected from poverty are conservative in two major ways.
First, the analysis leaves out most of the provisions, and the majority of the money, in the recovery act. The seven provisions we model account for only about 26 percent of the total funding in the act. We were unable to model billions of dollars in assistance in the form of Pell grants and education tax credits, funding for state health insurance programs, child care, child support enforcement, assistance to homeless individuals and to recipients of Temporary Assistance for Needy Families, and other assistance. We also ignore the effects of the bill in preserving or creating private- and public-sector jobs, which will also help reduce poverty.
Second, for the seven provisions we do examine, our analysis doubtless understates the amount of assistance provided due to limitations in the data. For example, by our projections, the law's additional weeks of unemployment compensation equal $19 billion. This falls short of CBO estimates of about $22 billion for calendar year 2009. (Moreover, CBO staff indicate these figures, estimated in February, would likely rise further if revised.) Other data problems include underreporting of unemployment among survey respondents and lack of information in the survey regarding the starting point of unemployment spells that began more than a year ago. (See Appendix B for data comparing our projections with official cost estimates for other provisions.)
Recovery Act's Impact on Large States
The following estimates are for the five states where we project the recovery act is protecting the largest numbers of individuals from poverty -- more than 300,000 residents in each state in 2009:
- California has 844,000 residents protected from poverty by the stimulus, including 335,000 children.
- Texas has 640,000 residents protected from poverty by the stimulus, including 271,000 children.
- Florida has 425,000 residents protected from poverty, including 154,000 children.
- New York has 419,000 residents protected from poverty, including 176,000 children.
- Illinois has 305,000 residents protected from poverty, including 109,000 children.
Limited sample sizes in smaller states make it difficult to make reliable projections for all 50 states.
APPENDIX A: Simulating the 2009 Economy
Accurately projecting the stimulus bill's effects requires accounting for the current state of the economy. The recession has affected family income and poverty status in two major ways. First, it has exposed more families to the risk of poverty by increasing unemployment and underemployment and thereby reducing their earnings. Second, it has increased participation in key income-stabilizing programs -- such as unemployment insurance and food stamps -- that the stimulus bill temporarily expands further. Failing to account for either of these changes would understate the role of the stimulus in protecting families from income loss and poverty.
The Census Bureau is not scheduled to release poverty data for 2009 until next year. Therefore, we project conditions in 2009 by updating earlier Census data. To do this, we start with data for three earlier years, 2003, 2004, and 2005. The data are from Census's March 2004, March 2005, and March 2006 Current Population Survey (CPS). We use these years because they are the latest for which corrections are available for underreporting of public benefits in the Census survey data. Three years are used to improve the sample size and stability of the estimates.
We update the data by "weighting up" the number of households with unemployed, underemployed, and, especially, long-term unemployed members until they match the numbers reported in monthly household survey data for April, May, and June of 2009. (Data on monthly employment status are available well before annual income and poverty data.) Numbers of fullyemployed households and of households with no adult members in the labor force are likewise re-weighted to match the figures for 2009. The calculation is done separately for each state. The resulting national totals are in Table A. 
The resulting approximation of the recent economy is imprecise. It does not capture changes in the relationship between household unemployment status and other characteristics -- such as program participation, poverty status, age, or ethnicity -- between 2003-2005 and 2009. As a result, it will not precisely reflect the number and characteristics of families in poverty, or the number of people who will benefit from the stimulus legislation.  It is, however, a much better representation of the recent economy than unadjusted data would provide.
|Unweighted households, Mar04-Mar06||Weighted households, Mar04-Mar06||Weighted households, Apr-Jun 09|
|Any Unemployed 15+ weeks||2,050||2.6 million||6.1 million|
|Other unemployed or involuntary part time||6,300||7.1 million||11.4 million|
|Only full-time workers||64,108||74.0 million||68.3 million|
|Elderly or other households with no adults in labor force||19,987||29.6 million||31.0 million|
|Total households||64,108||113.2 million||116.8 million|
|Date source used||March CPS||March CPS||March CPS|
A second step we take to replicate conditions in 2009 is to adjust the number of food stamp recipients. The monthly number of individuals receiving assistance from food stamps (recently re-named the Supplemental Nutrition Assistance Program or SNAP) has grown from 26 million in calendar year 2005 to 34 million in May 2009, according to the Agriculture Department. We reproduce this increase in our data through two steps. First, our reweighting by household employment status, described in Appendix A, increases the monthly number of food stamp recipients from 26 million to 30 million. This step fails to account for the full increase in food stamp participation levels,, in part because of limitations in our approach to modeling the changes in the economy and in part because of changes in state and federal policies that affect eligibility and accessibility of program benefits.
To match the May 2009 caseload of 34 million, therefore, we reclassify a share of eligible non-participants as participants. We do this using household-by-household data on the estimated eligibility and participation status of Current Population Survey households from the HHS/Urban Institute TRIM model. For each state, we re-assign eligible non-participating households as participants until the state's total number of participating individuals equals its actual total for May 2009.
APPENDIX B: Simulating ARRA
The analysis simulates selected direct income-support provisions in the American Recovery and Reinvestment Act of 2009. Because of limitations in the available Census data, we rely on a variety of simplifying assumptions to estimate which households receive these supports and how much they receive.
As noted, the analysis leaves out much of the assistance provided under the stimulus bill. It focuses on only seven easily allocated provisions that include about $205 billion of the estimated $787 billion cost of the bill. For the provisions we examine, our methods may understate the amounts received by some types of families and overstate it for others but on balance are likely to understate the total amount of benefits that families are likely to receive.
Key features of our simulation follow.
Extended Unemployment Compensation
This provision provides up to 33 additional weeks of Emergency Unemployment Compensation benefits for workers exhausting their 26 weeks of regular state unemployment insurance.
Assumed recipients: Persons with more than 26 weeks of unemployment in the year who appear to meet eligibility rules regarding job loss.
Technical notes: Analysis is based on individuals' reported weeks of unemployment in the previous calendar year in the Current Population Survey. Due to data limitations, all weeks of reported unemployment are counted as one spell. In calculating estimated weeks of unemployment, we account for the fact that weeks of unemployment appear to be significantly underreported in the annual retrospective survey questions on which we rely; that is, during 2003-2005, the CPS annual retrospective questions turned up only about 70 percent as many weeks of unemployment as respondents to the same survey reported at the time in response to contemporaneous questions about the previous week. We therefore divide weeks reported for the previous year by 70 percent. (In some cases, we have evidence that a worker was unemployed during the year based on a separate set of survey questions administered the following year, at survey time; these questions show whether a worker was unemployed at survey time and if so, how many months ago his or her unemployment began. If the answers to these questions conflict, we assume the longer spell of calendar-year unemployment is correct.)
We model the extension only for those who appear to be eligible for unemployment compensation -- either because they reported unemployment benefit income or because we can infer that they lost a job in that calendar year (based on their responses to questions about unemployment status and the duration and reason for their unemployment as of the time they were surveyed early the following year). This step excludes unemployed people who are likely to be ineligible because, for example, they quit a job voluntarily. We multiply weeks of unemployment by three-quarters to approximate the portion of the year (April through December) for which the benefit is in effect.
We also account for extended unemployment benefits that would have been available in each state without the stimulus bill under the permanent federal Extended Benefits (EB) program. In states where extended benefits would have become available by July 2009 (under laws in place on February 1, 2009) we added those EB weeks to the 26 weeks of regular UI benefits and only assigned benefits to workers whose estimated weeks of unemployment exceeded the total.
We assume that, for each week of extended unemployment compensation in April - December of 2009, workers in non-poor families are paid $297 and workers in poor families are paid $179. These amounts represent the national median amounts for these workers calculated from our data based on reported annual unemployment income (in 2009 dollars) per week of unemployment. (Dependent teenagers are assigned $179 as well.) This weekly benefit (which averages $213 per week across all recipients in our data) is a bit low in comparison with the actual national average benefit of ARRA's extended unemployment benefit payments, which was $292 per week in July 2009, according to Labor Department figures.
Note that in theory various data limitations mean that our estimates of eligibility for ARRA's long-term unemployment benefits could be either too high (for example, because we assume that all reported weeks of unemployment occur in one spell) or too low (for example, because the fact that we have only 12 months of data on each individual's work history means that we are probably missing a large share of long-term unemployment spells that began earlier). In practice, the resulting estimates of unemployment compensation appear to be quite conservative, relative to CBO estimates. Our simulations show $19 billon in extended unemployment compensation. CBO's February 2009 cost estimates show nearly $27 billion for fiscal year 2009 and beyond, of which staff estimate about $22 billion is for calendar year 2009.
Revised estimates that reflected the deeper-than-anticipated recession would be considerably higher.
$25 Increase in Weekly Unemployment Compensation
This provision provides an additional $25 for each week of regular or extended unemployment compensation.
Assumed recipients: Persons with one or more weeks of unemployment in the year who appear to meet eligibility rules regarding job loss.
Technical notes: As with our simulation of emergency unemployment compensation, we used last year's reported weeks of unemployment for individuals who appear to be eligible for unemployment compensation (based on receipt of unemployment benefits or reported job loss), and adjusted for underreporting by dividing by 70 percent (but not to exceed 52 weeks). We multiplied the resulting number of weeks of unemployment by $25 per week.
Making Work Pay Tax Credit
This is a new federal income tax credit providing up to $400 per worker ($800 per working couple). The credit phases in at a rate of 6.2 percent of earned income, and phases out for those with adjusted gross income between $75,000 and 95,000 ($150,000 and $190,000 for couples). Workers claimed as dependents by another tax payer are excluded.
Assumed recipients: Individuals reporting earnings, with AGI below $95,000 ($190,000 for a couple).
Technical notes: We model the effect of this and other tax credits by estimating each family's federal income and payroll taxes. Where possible, we start with estimates of the family's taxable income and adjusted gross income created by the Census Bureau; where this is not possible (for example, due to top-coding), we calculate taxable income and AGI from the family's income data. We inflate these figures to 2009 dollars and then apply IRS rules for tax year 2009 to estimate taxes and credits, with and without the Making Work Pay Credit and other ARRA provisions.
(Note: By law, families who receive the $250 payments described below have their Making Work Pay Credit reduced by the amount of the payment, in order to eliminate overlapping payments. For ease of calculation, we address the issue of overlapping payments by removing them from the $250 payments rather than here on the tax side.)
The credit is in effect in tax years 2009 and 2010 and is valued by the Joint Tax Committee at $116 billion. For calendar year 2009 alone, as would be expected, our simulation shows half that amount ($58 billion).
Child Tax Credit Expansion
This provision expands the federal Child Tax Credit to reach families with lower earnings. Without ARRA, the credit would provide up to $1,000 per qualifying child under 17 for tax filers with earned income of at least $12,550 in tax year 2009. ARRA lowers this earnings threshold from $12,550 to $3,000.
Assumed recipients: working families with children under 17 making more than $3,000.
Technical notes: Families earning between $3,000 and $12,550, who would have been ineligible in the absence of ARRA, may now qualify for at least a partial credit. Many families earning above $12,550 will be helped as well. For example, two-child families earning between $12,550 and $25,883 will qualify for as much as $1,433 more from the Child Tax Credit than they would otherwise have received.
We model families' taxes, as described above, with and without the tax credit expansion.
The Joint Tax Committee scores this provision at $15 billion over five years, or about $7 billion per year. Our model values the credit at $6 billion in 2009.
Earned Income Tax Credit Expansion
Expands the federal Earned Income Tax Credit for families with three or more children and provides relief from potential tax-credit marriage penalties for married couples, with or without children, in the EITC's phase-out range.
Assumed recipients: working families with three or more qualifying children with AGI up to $48,279 in 2009, as well as married parents making between $19,540 and $48,297; and childless couples making between $10,590 and $18,440 in 2009.
Technical notes: We model families' taxes, as described above, with and without the tax credit expansion.
The Joint Tax Committee estimates this provision is worth $4.7 billion over five years or about $2.3 billion per year for the two tax years in which it is in effect. Our model overshoots this at $4 billion in 2009.
Increase in Food Stamp Benefits
This provision provides a 13.6 percent increase beginning in April 2009 in the maximum monthly benefit for the Supplemental Nutrition Assistance Program, formerly known as food stamps. The benefit level thereafter remains frozen until it would have caught up, without ARRA, through annual inflation adjustments.
Assumed recipients: all food stamp recipients.
Technical notes: We start with baseline TRIM estimates of food stamp participation and eligibility for 2003-2005. As noted, we take two steps to adjust these data in order to approximate the actual May 2009 numbers of food stamp participants by state based on Agriculture Department data: we "weight up" households with unemployment and underemployment, and we re-assign some eligible food stamp households from 2003-2005 TRIM data that were originally coded as non-participants. (See Appendix A for details.)
For each participant (including re-assigned participants), we estimate the number of months of food stamp participation based on TRIM data on months of eligibility. Benefits depend on family size and months of eligibility. A family of four participating all 12 months of the year would not be eligible for any additional benefits for the three months (January through March) before ARRA took effect but would be eligible for $80 per month in extra benefits from April through September, and a projected $84.60 per month for October through December, for a total of $734. For those eligible for food stamps for only a fraction of the year, we reduce these figures proportionally.
CBO's February estimates showed this provision to be worth $19 billion over five years, of which it appears about $6 billion would come in calendar year 2009. Our projection is likewise $6 billion.
Payment to Recipients of Social Security, SSI, and Veterans' Disability Benefits.
This provision provides a one-time Economic Recovery Payment of $250 to retirees, disabled individuals, and SSI recipients receiving benefits from the Social Security Administration, Railroad Retirement beneficiaries, and disabled veterans receiving benefits from the U.S. Department of Veterans Affairs.
Assumed participants: All persons receiving annual income from Social Security, SSI, or veterans' disability compensation.
Technical notes: We assign $250 in additional income to those with income from these programs. For SSI, we identify participants using TRIM data that adjust for underreporting; for Social Security and veterans' benefits, we do not have TRIM data so we use unadjusted CPS data.
Under ARRA rules, the $250 payment is not meant to overlap with the Making Work Pay Credit; any amount one receives from the $250 is to be taken out of one's Making Work Pay Credit. In our calculations, for ease of calculation, this overlap is removed from the $250 rather than the credit; the combined result of the two provisions, however, is the same.
Joint Tax Committee estimates value this provision at just over $14 billion. Our model appears to understate its value, at $10 billion.
APPENDIX C: Expanded Measure of Poverty
Because the official Census Bureau poverty measure is a cash measure, it does not include the value of non-cash and tax-based benefits, which constitute much of the assistance in the stimulus act.
To better measure the effects of the stimulus, this analysis takes the following steps, following wherever possible the recommendations of the National Academy of Sciences' 1995 expert panel on poverty measurement:
- Measuring income on a post-tax, post-transfer basis. As recommended by the NAS panel, we include in the measure of income the value of non-cash benefits (other than health benefits such as Medicaid and CHIP) and the net impact of taxes. Specifically, we include the market value of food stamps and housing subsidies, the family's share of low-income home energy assistance (from the CPS), and estimates of what each tax filer owes in federal and state income taxes and payroll taxes as calculated by the Census Bureau (available in the CPS file). We do not subtract state sales tax because the data needed to do that are not available.
- Correcting for underreporting of benefits. The CPS significantly underreports receipt of certain benefits. (Census's counts of program participants typically fall well short of the totals shown in actual administrative records.) Therefore, we go beyond the NAS recommendations by using data on TANF, SSI, and food stamp receipt from the Transfer Income Model (TRIM) developed for the Department of Health and Human Services. The TRIM data start with the CPS data and add to it by assigning extra benefit income to some individuals in the CPS who did not answer all of the CPS questions. For example, someone who skipped a question about whether he or she received food stamps might be assigned food stamp income. (Unlike in the CPS, the total number of recipients in the TRIM data is designed to approximate actual administrative totals.) TRIM data files are currently available through 2005. We also incorporate updated estimates of the value of housing subsidies.
- Subtracting out-of-pocket medical expenses and work expenses (including child care) from income. Following the NAS recommendations, we subtract these expenses from the income measure. To estimate the size of these expenses, we follow procedures provided by the Census Bureau and Bureau of Labor Statistics.
- Counting unmarried partners. Unlike the official poverty measure, we treat unmarried partners as part of the family, as recommended by the NAS panel. (We do not do this for relatively transient partners. Some boyfriends or girlfriends may not have been present in the household for most of the previous year and may not have been regular contributors to the family's budget. We count partners who have been present for 12 months.)
- Adjusting the poverty line itself. We use alternative poverty lines based on the NAS recommendations. These alternative poverty lines are based on the cost of very basic needs -- food, clothing, shelter, utilities, and (in the words of the NAS panel) "a little more" -- and vary geographically based on housing costs. The poverty line is initially set at about the 33rd percentile of expenditures on food, clothing, shelter, and utilities -- that is, the minimum amount that all but the bottom one-third of families spend on these basic needs. The NAS-style poverty line that we use for a family of four, consisting of a couple with two children living in a community with average housing costs, was $21,395 in 2005, compared with $19,806 for the official federal poverty threshold. We calculate variations for different family sizes and geographic locations following procedures provided by the Census Bureau.
For additional methodological details and data using this measure of poverty, see CBPP, "Safety Net Effective At Fighting Poverty But Has Weakened For The Very Poorest."
 The figures start with detailed Census survey data collected in the Current Population Surveys for March 2004, March 2005, and March 2006. As described in Appendix A, we project these data forward to match the number and employment situation of households as of April, May, and June 2009. We also assign food stamps to additional eligible households so that food stamp levels for each state match the actual May 2009 levels as reported by the Agriculture Department.
 The following data are less precise than the overall 6.2 memo estimate because we are unable to fully account for variations in the recession's effects on different age, racial-ethnic, and other groups. The following numbers thus should be viewed as approximations.
 We start with Census data on a nationally-representative sample of about 90,000 households, each one of which is presumed to represent a larger number households in the nation as a whole at the time it was surveyed. Researchers and government analysts use such data to calculate national estimates multiplying the data for each sample household by a sample "weight" provided by the Census Bureau and summing the results. Thus, our data, based on about 278,000 household records, yields a "weighted" total of about 113 million households for 2003-2005. Our method adjusts those weights upward (or, where appropriate, downward) in order to approximate the population size and household employment conditions for 2009.
For example, raw "unweighted" Census survey data from the March 2004-2006 CPS show an average of 2,059 households where household members reported 15 or more weeks of unemployment prior to the survey. Following standard Census procedures, these households would represent a weighted total of 2.6 million households nationwide in March 2004-2006. However, recent monthly data indicate that there are now 6.1 million such households nationwide. Therefore, we increase the weights accordingly until they reach the new total. (As noted in the text, we do this adjustment separately by state. Thus, in Massachusetts we multiply the weights for all households with household members unemployed 15 weeks or more by 1.67; in Michigan, we multiply by 2.67.)
 Another potential source of imprecision relates to the timing of the survey data. The March CPS asks households about their annual income, poverty status, and program participation as of the previous year. When we "weight up" unemployed and underemployed households, however, we do so with respect to their March employment status, not their status during the previous year. Because there is only a partial correspondence between March employment status and employment status during the previous year, we may not always be "weighting up" the families with the greatest risk of poverty during the previous year.
 For a description of the HHS/Urban Institute TRIM model, see footnote 8 below.
 To the extent that the increase in food stamp participation has been caused by changes in eligibility we may not accurately represent the income and demographic information of the additional food stamp participants.
 National Research Council, Measuring Poverty: A New Approach (National Academy Press: 1995).
 TRIM is developed and maintained by the Urban Institute under contract with the Office of the Assistant Secretary for Planning and Evaluation at the Department of Health and Human Services. Documentation of the TRIM Model is at http://trim3.urban.org/T3Technical.php. While the model was developed in large part to allow users to compare current policies with proposed policies, we use data only for TRIM's "baseline" (i.e., current-policy) scenario.
 In producing the CPS files, the Census Bureau, like TRIM, also assigns benefits for some people with missing data. Unlike TRIM, however, Census does not use this process of assigning missing benefits to try to match the actual number of recipients shown in program records.
 Because medical expenditure data are not available directly from the CPS, they must be estimated using a formula. In this analysis, we estimate each family's out-of-pocket medical spending by starting with the medical share of the poverty line (specifically, a version of the NAS poverty line referred to in Census publications as "MIT" or "Medical-out-of- pocket-expenditures In the Thresholds") for a standard family of two adults and two children. We adjust that amount up or down using a set of ratios provided by the Census Bureau that depend on the family's size, health insurance status, and age. That yields the family's own estimated medical out-of-pocket expenses. We subtract those expenses from family income prior to comparing family income to the poverty line. (The NAS poverty line we use is the appropriate one for this comparison, namely, the MIT line minus the medical share. The result is by definition identical to using the Census Bureau's MIT method. As a mechanical matter, however, we follow the original NAS recommendations by treating these medical expenses as an item to be subtracted from income rather than as part of the bundle of basic needs.)
Our estimates of work expenses also rely on formulas provided by the Census Bureau. The formulas are based on data on median weekly out-of-pocket child care expenses and other work-related expenditures from the Survey of Income and Program Participation. Values vary depending on year, weeks worked, number and ages of children (in the case of the child care formula), and other family characteristics. The estimated value of work expenses is capped at the value of the worker's earnings. A couple's child care expenses are further capped based on the earnings of the lower-earning spouse (although, in practice, few couples have estimated child care expenses large enough to be affected by this cap).
 Following the approach used in the TRIM model, we also treat unrelated household members younger than 15 living with no relatives as de facto family members of the head of household. This affects fewer than 200,000 poor children.
 As described in footnote 10, the version of the NAS threshold used here is essentially what Census publications call the MIT line with the medical share removed.