Abstract

The Great Recession which began in 2008 was unprecedented, constituting the largest downturn since the Great Depression. From 2007 to 2009, real GDP fell by 3.1 percentage points, real personal income per capita fell by 8.3 percentage points, and the national unemployment rate rose from 4.6 percent to 9.3 percent. Many individuals dropped out of the labor force, reducing the employment-to-population ratio from 63.0 to 59.3, a level not seen since the early 1980s. Like the Great Depression but unlike the other recessions since World War II, the Great Recession was set off by a financial crisis which spread to other sectors of the economy. Although economic activity has already recovered at a more rapid pace than it did after 1929, employment growth has been particularly slow relative to recent recessions, with an employment level of 143 million in November, 2012, still below the 2007 monthly average of 146 million. Most forecasts predict further recovery but continuing at a slow pace.

This paper addresses the responsiveness of the U.S. social safety net to this major economic downturn. The social safety net, as defined here, includes both all major means-tested transfer programs (or “welfare” programs) and all major social insurance programs. It is important to realize that not all programs are intended for countercyclical income replacement and, consequently, the responsiveness of the safety net as a whole is not clear a priori. Among means-tested programs, some are aimed at the retired elderly or the young and older disabled, for example, whose incomes may not change a great deal during a recession. Among social insurance programs, only the Unemployment Insurance program is explicitly aimed at countercyclical income replacement; the other social insurance programs (Disability Insurance and retirement benefits, for example) are based on past earnings histories and not directly on current unemployment status. Nevertheless, many programs do serve families during a recession and, while it should not be expected that the social safety net should replace all lost income in a major recession like the one the U.S. is currently experiencing, naturally most observers would think that significant replacement should occur in response to such a sizable decline in economic activity.

This paper examines the performance of the social safety net during the Great Recession in four separate ways:

1. How much did aggregate expenditures from all safety net programs rise? How did this compare to past recessions?

2. What were the most important programs responsible for the aggregate expenditure increase that occurred?

3. Since different programs serve different demographic groups, then, if the expenditure increase was different for different programs, did the increase in safety net expenditure favor certain demographic groups over others?

4. Since different programs serve families at different income levels--for example, means-tested programs typically serve families with lower income than do social insurance programs--did the increase in expenditure primarily benefit the very poor or those at somewhat higher income levels, including those with income above the poverty line?

The first two questions concern the aggregate performance of the safety net, where performance is judged by the strength of its aggregate response. The third and fourth questions concern its distributional performance. For the third, performance is measured by how well the safety net responded by benefitting all demographic groups equally, or at least similarly, rather than disproportionately benefitting certain groups and possibly leaving out other groups entirely. For the fourth, most observers would prefer a safety net that furnishes greater support to those at lower income levels than at higher income levels to one that provides greater support to those at higher income levels. The performance of the safety net in that respect therefore can be judged by the progressivity of its response to the Recession.

A different but equally important measure of performance is how the safety net affected the poverty rate during the Great Recession. This paper does not address that issue. The official government poverty rate for the U.S. population rose from 12.5 percent in 2007 to 15 percent in 2011 (De Navas-Walt et al., 2008, 2012) and rose even more strongly for the non-elderly, for the poverty rate among the elderly fell over that period. Since, as we shall see, transfers rose between those years, the poverty rate would have risen by more in the absence of that increase. This question is left for future work.

There has been very little prior work specifically focusing on the issues addressed here concerning changes in the magnitude and distribution of transfers during the Recession. Ziliak (2011) totaled up total transfers through 2009 using CPS data and found sizable increases in expenditures. Burtless and Gordon (2011) use aggregate statistics from administrative sources to estimate the increase in total federal government transfers, also finding a very large increase. Gabe and Whitaker (2012) focused specifically on UI, estimating the effect of federal stimulus legislation on the size of the program. Bitler and Hoynes (2010) conducted a state-level econometric analysis suggesting that cash transfers have become less responsive to recessions and in-kind programs have become more responsive.

Our results show that there was a major response of the safety net to the Recession. Aggregate spending in the main safety net programs rose from $1.6 trillion to $2.1 trillion from 2007 to 2010. Caseloads over all programs rose from 276 million recipients to 310 million over the same period.¹ The largest contributors to this increase were from the Unemployment Insurance program, the Earned Income Tax Credit, and the Supplemental Nutrition Assistance Program which, combined, accounted for about a third of the rise in spending. Large increases in Social Security retirement and disability benefits and in Medicare and Medicaid also occurred. In addition, the increase was widely distributed, going to all major demographic groups and family types in the population. The increase in transfers went to all income strata within the low income population, although slightly more in percentage terms to those with incomes just below the poverty line or just above it. While there are some programs which displayed a disappointingly small response, the safety net as a whole exhibited a relatively successful performance during the downturn.

I. The Structure of the Social Safety Net

It is useful to begin with a brief discussion of the social safety net, program by program, assessing whether each should be expected to provide countercyclical income support and to discuss the particular demographic groups and income levels each is aimed to serve.

The transfer programs which constitute the social safety net are generally distinguished by whether they are means-tested or social insurance in character. The former base eligibility on having low income and assets, although often other eligibility criteria related to age, family structure, or disability are present. Not all are entitlements but are, instead, limited by funds available. Benefit levels are usually higher for those with lower income, although some programs included in this category in this paper (e.g., the Earned Income Tax Credit) sometimes pay higher benefits to those with higher incomes. Social insurance programs, on the other hand, base eligibility on employment and earnings histories and, generally speaking, those families and individuals with greater employment and earnings histories are more likely to be eligible and generally receive higher benefits when eligible. The distributional effect of social insurance programs, therefore, is quite different than for means-tested transfer programs because many of the most disadvantaged individuals in the country also have the lowest levels of employment and earnings and hence are ineligible for social insurance benefits.

III. Aggregate Trends

Figure 1 shows trends in real expenditure per capita from 1990 to 2010 in all major means-tested transfer programs except Medicaid. As expected, some programs showed very little change during the Great Recession, the best example being the TANF program, which has experienced a long-term decline since it was reformed during the mid-1990s. The small increase in TANF spending after 2009 was probably a result of the extra funds provided by the stimulus legislation. There were small increases in spending as well in the SSI and housing programs but these were not large, primarily for the reasons noted above. However, there were major changes in expenditure in the SNAP and EITC programs. Expenditure in SNAP grew the fastest during the Recession, with aggregate expenditures growing from $30 billion 2007 to $65 billion in 2010 and real per capita spending increasing from $136 to $287, both representing more than a doubling in magnitude. SNAP expenditures had risen previously from 2000 to 2007 as a result of reforms in the program that simplified eligibility procedures, lengthened and simplified recertification, increased vehicle exemption levels, and from a series of outreach programs to encourage families to see if they were eligible. These reforms made access to the program much easier than it had been prior to that time and no doubt made it easier for new recipients to enter when their incomes fell during the Recession. However, the increases in the benefit amount and relaxation of eligibility requirements during the Recession noted previously also no doubt played a major role. Nevertheless, data on the caseloads and benefits for the SNAP program show that most of its increase in expenditure arose from a rise in recipiency rather than benefits per recipient.

Open in a separate window

Figure 1

Expenditure per Capita, Non-Medicaid Means Tested Programs, 1990–2010 (real 2009 dollars)

Expenditure in the EITC program, while not rising at the same rate as for SNAP, also grew strongly, with aggregate expenditure rising from $49 billion to $59 billion and per capita real spending rising from $217 to $242, a 12 percent increase. This answers the question of whether there were more individuals whose earnings dropped to zero than individuals whose earnings dropped down into the positive or larger EITC range, for the latter appears to have dominated the former. While the stimulus bill increasing benefits for families with three or more children no doubt played a role, it is unlikely to be responsible for a major part of the growth. In fact, expenditures per recipient actually declined during the Recession; the growth was entirely an increase in the number of recipients.

Medicaid spending (not shown) also grew rapidly during the Recession, with aggregate spending growing from $327 billion in 2007 to $401 billion in 2010 and with real per capita expenditure rising from $1,459 to $1,658, a 14 percent increase. As noted above, this program should be expected to exhibit countercyclical trends.

Figure 2 shows trends in the major social insurance programs, with all of them showing considerable growth during the Recession. As expected, both on the basis of normal countercyclical behavior as well as the large stimulus additions, the UI program grew the most. Aggregate spending between 2007 and 2010 increased from $34 billion to $142 billion and real per capita expenditure grew from $150 to $581, almost a four-fold increase. However, there was significant growth in OASI, Medicare, and DI as well, although the Medicare trend is mostly a continuation of long-term trends without a particular deviation during the Recession. Expenditure in the OASI and DI programs, while also trending upward over the long term, grew somewhat faster during the Recession. For the OASI program, this may have been partly the result of stimulus legislation providing for one-time increases in benefits. For the DI program, expenditure increases are best interpreted as an increase in the participation rate of medically eligible individuals, many of whom may have been employed but lost jobs during the downturn and applied for DI as a result. Thus, somewhat surprisingly, the social insurance programs other than UI also exhibited a fairly strong and noticeable positive response to the Recession. Indeed, taken together, the increases in spending on social safety net programs, both means-tested and social insurance, constitute a significant and major response to the Recession.

Open in a separate window

Figure 2

Expenditure per Capita, Social Insurance Programs, 1990–2012 (real 2009 dollars)

Aggregate per capita spending over a longer-period, from 1970 to 2010, is shown in Appendix Figure A-1. The figure shows the trend in expenditure for the programs shown in Table 1 and hence does not include all safety net programs but does include the most important ones. Despite the ups and downs of several of the individual programs, total spending has exhibited a continuous upward trend over time. In addition, the growth of spending during the Great Recession dominated the growth during the recession of the early 1990s. But that recession had a peak annual unemployment rate of only 7.5 percent, considerably below the Great Recession peak annual rate of 9.6 percent. Interestingly, the early 1980s recession, which had a peak unemployment rate of 9.7 percent had only slightly less growth in real per capita spending—14 percent from 1979 to 1982 compared to 18 percent from 2007 to 2010. The greater growth in spending in the Great Recession did not, in fact, stem from UI—in fact, social insurance spending grew by the same percent in both recessions. UI growth in the Great Recession was vastly greater than that in the early 1980s recession—about 280 percent in the former compared to 100 percent in the latter—but Medicare growth was 26 percent in the early 1980s recession compared to 13 percent in the Great Recession. Because Medicare spends about four times the amount spent on UI, even at its 2010 peak, Medicare growth dominates UI growth. Instead, the big difference in the two recessions was in means-tested program spending, which grew by 17 percent in the Great Recession but only 2 percent in the early 1980s recession. Most of the major means-tested programs today were much smaller thirty years ago.

IV. Distribution

While the safety net program as a whole exhibited strong aggregate growth during the Recession, that does not mean that all demographic or income groups benefitted equally, especially given the uneven growth across different programs that serve different such groups. We may define different demographic groups defined by family structure: single-parent families, two-parent families, and childless families and individuals, for example. Single-parent families have, historically, been the major recipients of the TANF program yet that was one of the programs that responded very little to the recession. Consequently, it is possible that those families did not share in the increases. The EITC provides support primarily to families with children. Another demographic distinction is between the elderly and disabled, on the one hand, and the non-elderly, non-disabled, on the other. Over the last thirty years, increases in safety net expenditures have grown much more rapidly for the elderly and disabled than for the rest of the population (Ben-Shalom et al., 2012), and it is of interest to know if this relative growth continued into the Great Recession. The OASI and SSI programs primarily benefit these groups, for example. Another important distinction is between employed and non-employed families. While the bulk of increased UI spending should be expected to benefit primarily the latter, the growth in EITC spending should benefit primarily the former.

It is also of interest to know the incidence of expenditure growth across different portions of the income distribution. Many of the programs in the safety net, particularly the social insurance programs, benefit families above the poverty line as well as those below, and the former may have received a significant portion, or even the bulk, of the increased spending. The EITC should also primarily affect those who are not at the very bottom of the income distribution, but rather those slightly higher up. Even UI benefits, which are not directly based on family income, may be received disproportionately by those who are not at the very bottom of the income distribution. Over the last thirty years, most of the growth of safety net expenditures in the U.S. has gone to those families with incomes above 50 percent of the poverty line and, indeed, per-family expenditures for those below that level have actually fallen for some groups (Ben-Shalom et al., 2012). Whether this trend has been maintained into the Great Recession is an important policy question.

To conduct this investigation, we extend the work of Scholz et al. (2009), Moffitt and Scholz (2010), and Ben-Shalom et al. (2012) to the Great Recession. Using the 2004 panel of the Survey of Income and Program Participation (SIPP), these authors documented the degree to which families of different demographic and economic characteristics, and at different places in the income distribution, received safety net transfers. Their results for the year 2004 will constitute the pre-Recession benchmark for the purpose of this paper. The national unemployment in the months of the SIPP interviews used in their 2004 analysis averaged 5.7 percent. Here, we will use the 2008 SIPP panel, covering both the early period of the Great Recession, September 2008 to March 2009--when the unemployment rate averaged 7.3 percent but the safety net expansions arising from Congressional legislation had just barely begun--and a later period, September 2010 to March 2011--when the unemployment rate averaged 9.3 percent and the expansions were fully in place. We will call the first period “2008” and the second period “2010.”

The SIPP is a nationally-representative household survey which periodically enlists a random sample of the U.S. nonstitutional population and follows them for several years. The survey interviews families every four months, collecting information on the demographic and economic characteristics of the family as of the interview date, as well as receipt of private income and public income for each of the four months prior to the interview. Because the survey is intended to measure program participation, a reasonably complete set of questions on receipt of different transfer programs is included. Using the data for the two periods in the 2008 panel noted above, we define, for each person in the SIPP, their “market” income in each of the four months prior to interview, constructed as the sum of their wage and salary income, self-employment income, capital income (interest, dividends, and rent), and defined benefit pension income. We then average that market income over the past four months and sum across individuals in the family to obtain an average family amount. We use this income amount to classify families into different portions of the pre-tax, pre-transfer income distribution: income less than 50 percent of the official government poverty line for their family size; income greater than 50 percent but less than 100 percent of that poverty line; and income greater than 100 percent but less than 150 percent of that line.³

Post-transfer income is computed by adding together the reported receipt of each transfer program in each of the four months prior to interview, averaging over the four months and summing over all individuals in the family. The transfers we include are the EITC, Child Tax Credit, SSI, Housing subsidy amounts, SNAP, TANF, General Assistance, WIC, General Assistance, Veterans programs, Foster Child payments, OASI, DI, UI, and Workers’ Compensation. We exclude Medicaid and Medicare because the families in the survey did not know the total amount being spent on them from these programs.⁴

For demographic groups, we define five groups: the elderly (reference person 62 and over), the disabled (receiving SSI-disability or DI), and three non-elderly non-disabled types--one-parent families with children under 18, two-parent married families with children under 18, and childless families and unrelated individuals.⁵ For the disabled, we simply define the category as receiving disability-related incomes, for the SIPP has insufficient information to determine medical eligibility for disability programs. In addition, we defined two other mutually exclusive demographic groups in the non-elderly, non-disabled population defined by employment in the family: “employed” families are those with at least one adult who worked all of the four months prior to interview, and “nonemployed” families are those with no adult who worked all of those four months.

Findings

We first show trends in safety net expenditure for the different demographic groups and different portions of the income distribution. Figure 3 shows the real average monthly expenditure per family in each of the seven demographic groups in 2004, 2008, and 2010, with expenditure summed over all transfer programs and the average taken over all families in the population in each group, regardless of income. In 2004, expenditures fell into an expected pattern given what is known about relative incomes and program eligibility--higher for single-mother families ($414) than for two-parent families ($242) and childless families and individuals ($101), and higher for the nonemployed ($353) than for the employed ($161).⁶ The elderly and disabled received much higher amounts, however ($1,245 and $1,277, respectively), mainly from the greater rate of receipt of OASI for the elderly and the relatively high benefit levels for DI and SSI.

Open in a separate window

Figure 3

Average Monthly Government Expenditures

Average per family total transfers rose for all groups in the early Recession (2008) and later Recession (2010). From 2004 to 2008, all non-elderly non-disabled groups experienced increases in transfers, ranging between 16 and 20 percent. The elderly received the smallest increase, 6 percent, and disabled families experienced a 10 percent increase. As shown in Appendix Table A-1, the increases for the non-elderly non-disabled population were primarily from increases in expenditure from three programs: SNAP, the EITC, and UI. All demographic groups benefitted from these programs, including the EITC and UI programs which are aimed at those either currently working or with work histories. Because the stimulus legislation had only barely begun in 2008, these increases can be interpreted as resulting mainly from the automatic countercyclical features of the programs.

From 2008 to 2010, transfers strongly rose again for all non-elderly non-disabled groups, increasing by from 16 to 20 percent for the different groups. The elderly and disabled experienced much smaller gains (5 percent or less). As shown in Appendix Table A-1, the sources of the gains for the non-elderly non-disabled families were different for different types of families. Single and two-parent families benefitted from all programs, while childless and non-employed families primarily benefitted from SNAP and UI increases and not the EITC. Families with at least one employed member also benefitted from all three programs. Interpreted as resulting from an increase in automatic countercyclical increases as well as the stimulus expansions, the benefits from the social safety net as a whole were remarkably widespread across all demographic groups; none was left out.

Figure 4 shows the pattern of average transfers received by place in the pre-tax pre-transfer income distribution, focusing only on families with incomes less than 150 percent of the poverty line but averaging over all demographic groups. Average transfers increased for the three income groups in the low income population noted previously: those in deep poverty (pretax, pre-transfer income less than 50 percent of the poverty line), those in shallow poverty (50 percent to 100 percent of the poverty line), and those in near poverty (100 percent to 150 percent of the poverty line). Interestingly, average transfers to those in shallow poverty are not much smaller than those in deep poverty, even prior to the Recession, indicating that the U.S. transfer system is not strongly progressive within the poor population.⁷ Independent of this, however, the increase in average transfers were approximately 8 percent from 2004 to 2008 and from 2008 to 2010 for those in deep poverty, while the corresponding rates of increase for those in shallow poverty were 9 percent and 10 percent, respectively; and those for families in near poverty were about 11 percent over both time intervals. Thus the Great Recession safety net response was not particularly progressive.

Open in a separate window

Figure 4

Average Monthly Government Expenditures

Table A-2 shows the contributions of SNAP, the EITC, and UI to the transfers for each income group. While those in deep poverty saw increased transfers from the SNAP program and from UI, they saw little increases from the EITC. That program’s support mostly went to families in shallow poverty, who also receive the highest transfers from this program, with those in near poverty receiving quite a bit more than those in deep poverty as well.

A finer-grained examination of these distributional trends can be gleaned by focusing on both demographic groups and place in the income distribution jointly. Table 2 shows average expenditures broken out in this way and Table 3 shows the corresponding percent increases. All types of families, both demographic and in every income stratum, experienced some increases in transfers after 2004 and after 2008 in almost every case. However, the relative increases for those in shallow poverty versus those in deep poverty shows up differentially across demographic groups in the way that would be expected if the relative targets of the EITC, on the one hand, and SNAP and UI, on the other, are behind the patterns of transfers. Thus single-parent and two-parent families in shallow poverty benefit particularly from the EITC, while the childless do not, leading to a smaller differential by income for the latter group. Likewise, employed families benefit more from the EITC than nonemployed families, leading to particularly large increases in transfers for those in shallow poverty, especially from 2004 to 2008 (20 percent versus 6.8 percent for those in deep poverty). On the other hand, nonemployed families in deep poverty saw greater increases in transfers than those in shallow poverty, for this group receives little in EITC payments and mainly relies on UI and SNAP.

Table 2

Average Government Expenditures by Income Range and Demographic Group

Income Range	Under 50% of Poverty Line			Between 50% and 100% of Poverty Line			Between 100% and 150% of Poverty Line
Year	2004	2008	2010	2004	2008	2010	2004	2008	2010
Nonelderly, nondisabled
Single-parent families	618	726	769	565	741	869	442	488	561
Two-parent families	530	688	775	559	757	827	414	522	607
Childless families and individuals	217	239	326	191	224	294	131	161	204
Employed families	345	370	464	384	480	575	271	323	376
Nonemployed families	369	433	506	528	575	581	292	---1	---1
Elderly families and individuals	1199	1295	1365	1373	1444	1517	1362	1449	1493
Disabled families and individuals	1219	1362	1459	1479	1613	1679	1429	1524	1605

Open in a separate window

Source: Author’s calculations using 2004 wave 1, 2008 waves 2 and 8 SIPP.

Notes: Elderly families and individuals are those families and unrelated individuals headed by an individual age 62 or older. Disabled families and individuals are those with anyone in the family who received SSI or DI. The Nonelderly, nondisabled families and individuals are mutually exclusive from the Elderly and Disabled. Single-parent families are families with children under 18 in the household and with one parent present. Two-parent families are families with children under 18 in the household and two married parents present. Childless families and individuals are those without a child under 18 in the household, and include what Census definitions call unrelated individuals as well as families. Employed families are those with at least one person over 15 who worked in all four months prior to the interview. Nonemployed families are those without any such person. The single-parent, two-parent, and childless groups overlap with the employed and nonemployed groups; they are not mutually exclusive.

¹Sample sizes too low for reliability

Table 3

Percent Change in Government Expenditures by Income Range and Demographic Group

Income Range	Under 50% of Poverty Line		Between 50% and 100% of Poverty Line		Between 100% and 150% of Poverty Line
Year	2004– 2008	2008– 2010	2004– 2008	2008– 2010	2004– 2008	2008– 2010
Nonelderly, nondisabled
Single-parent families	14.8	5.6	23.7	14.7	9.3	13.0
Two-parent families	22.9	11.2	26.2	8.4	20.8	14.0
Childless families and individuals	9.0	26.9	14.7	24.0	18.4	21.1
Employed families	6.7	20.2	20.1	16.4	16.1	14.3
Nonemployed families	14.8	14.4	8.0	1.2	---1	---1
Elderly families and individuals	7.4	5.2	5.0	4.8	6.0	3.0
Disabled families and individuals	10.5	6.6	8.3	4.0	6.3	5.0

Open in a separate window

Source: Author’s calculations using 2004 wave 1, 2008 waves 2 and 8 SIPP.

Note: See Table 2 notes for definition of groups.

¹Sample sizes too low for reliability

Tables A-3, A-4, and A-5 in the Appendix confirm these interpretations by showing transfers from the SNAP, the EITC, and UI for the different demographic and income groups. The largest transfers for any of the groups arise from the EITC for single-parent and two-parent families in shallow poverty. The next largest average transfers are those from the SNAP for single-parent and two-parent families in deep poverty. Employed families in shallow poverty and single-parent and two-parent families in near poverty also receive high levels of EITC transfers. The impact of stimulus legislation is undoubtedly partly behind the increases in UI transfers to nonemployed families from 2008 to 2010, and to childless individuals and families in poverty as well.

These patterns are to a large degree expected, based on the structure of the programs that experienced the largest expansions during the Recession, both from automatic triggers of higher transfers as incomes fell and from the stimulus legislation, which further expanded the three main programs most responsible for the increase (SNAP, the EITC, and UI). The SNAP program directs greater support to families, the lower their income; the UI program, while eligibility partially directed to those somewhat higher in the permanent income distribution and not basing benefits or eligibility on family income per se, provided its greatest support during the Recession to those with the lowest family incomes; but the EITC, which provides greater assistance to those higher up the income distribution even in normal economic environments, did so as well during the Recession. On net, these structural features are responsible for the relatively even increases in transfer to all families below the poverty line, both those far below it as well as those with incomes closer to it.

V. Work Disincentives

An important issue in assessing the response of the safety net in the Recession is whether the increase in transfer payments to the low income population was pushed up by work disincentives of those programs. To the extent that the increase in expenditure documented here was the result of voluntary reductions in work effort, those observed expenditure increases overstate the income replacement effect of the increases because pre-transfer income may have been reduced as a result of the programs. In theory, the effect of an increase in program generosity during a downturn has ambiguous effects on work disincentives. On the one hand, a more generous program would be expected to increase whatever disincentives were already present. On the other hand, a recession reduces job opportunities or, in the language of economic research, reduces the arrival rate of job offers (if not the offered wage itself), and this leads to a lower level of work even in the absence of any increase in work disincentives. Thus the work disincentives of a program during a recession could be larger or smaller than they are during more normal economic times.

For the most part, direct econometric evidence on work disincentives of transfer programs during a major recession are lacking. An exception is unemployment insurance, where at least two studies have been conducted. Rothstein (2011) studied the effects of differential UI extensions across states and over time to ascertain the effect of those extensions on lengths of unemployment spells. Rothstein found only a very small effect, and most of the effect arose from individuals’ extending their unemployment spells rather than dropping out of the labor force. Schmieder et al. (2012) studied the German unemployment insurance system and estimated work disincentives in both recessions and normal economic periods, finding that work disincentives were smaller during recessions than during normal periods.

Aside from these two studies of UI programs, there have been no studies specifically examining work disincentives of other programs during downturns. The literature on work disincentives during normal periods is moderately large, however. For the two most important programs in the Recession, the SNAP program and the EITC, the literature shows them to have very small effects on work effort. Currie (2003) surveyed the literature on the effects of Food Stamps and found that most studies estimate the program to have only small or zero effects on labor supply, with the maximal estimate suggesting a one-hour-per-week disincentive. These small estimates are generally attributed to the small size of the benefit relative to income. In a more recent contribution, Hoynes and Schanzenbach (2012) found work disincentives of the Food Stamp program in its early years to be non-trivial for single mothers but small and insignificant for the population as a whole. For the EITC, the evidence suggests that the program has had a positive impact on the employment rates of single mothers but no effect on their hours of work if working, little or no effect on the labor supply of married men, and small negative effects on the employment rates and hours of work of married women of approximately 1 percent and 1-to-4 percent, respectively (Hotz and Scholz, 2003; Eissa and Hoynes, 2006). While studies of the effects of these programs in the Recession may show different effects, these studies provide little evidence that the expenditure increases reported here attributable to work disincentives to any significant degree.

VI. Summary

The social safety net has responded in significant and favorable ways during the Great Recession. In aggregate, per capita expenditures have grown in total over all major programs, with particularly strong growth in the SNAP, EITC, UI, and Medicaid programs. Expenditures have also grown by smaller degrees for the DI, SSI, OASI, and Medicare programs. Distributionally, the increase in transfers was widely shared across multiple demographic groups, including families with and without children, single-parent and two-parent families, and, to a lesser degree, the elderly and the disabled. Transfers grew as well among families with more employed members and with fewer employed members. The increase in transfer amounts was, however, not strongly progressive across income classes within the low-income population. The percent increase in transfers was positive for all income groups but was slightly larger for those just below the poverty line and those just above it, compared to those at the bottom of the income distribution. This is mainly the result of the EITC program which, looked at as a distributional program rather than one to induce work incentives, is regressive within low income ranges and provides greater benefits to those with higher family earnings. The expansions of SNAP and UI benefitted those at the bottom of the income distribution to a greater extent.

The most important legislative extension in the Recession was that in the UI program, and the stimulus expansions have already begun to be withdrawn as the economy recovers. However, the EITC program should remain more or less in place as the recovery occurs, regardless of whether the benefit increase for larger families is maintained. The future of the SNAP program primarily depends on whether the eligibility requirements which were relaxed during the Recession and whether the SNAP reforms of the 2000s are maintained. The withdrawal of the UI benefits will primarily affect those at the bottom of the income distribution, according to the figures here, and any contraction in SNAP benefits will as well. The maintenance of the EITC more or less in its permanent structure will continue to benefit those with somewhat higher incomes. The progressivity of the safety net is therefore likely to decrease after the recovery.

Acknowledgments

Appendix

Figure A-1: Total Real per Capita Spending, 1970–2010

Footnotes

References