Pasar al contenido principal

Preparing for the Next Recession: Lessons from the American Recovery and Reinvestment Act

Executive Summary

Recessions generate significant risks for both people and the economy.  At the human level, they increase hardship for families that lose jobs or businesses.  For the economy, they cost output that’s never recovered.  Moreover, recent research has found that recessions can cause considerable “scarring” as temporary symptoms such as joblessness morph into permanent disconnections between the jobless and the labor market.[1]  If such effects are large enough, they can significantly reduce both the future level of gross domestic product (GDP) and the economy’s growth rate.

Consequently, measures that can quickly respond to a recession by bolstering the economy and at least moderating the downturn’s negative impacts are important.  While the Federal Reserve lowers interest rates and expands access to credit, the President and Congress can tap various “stabilizers” through budget and tax policy that can offset some of the financial losses that households experience and help them maintain higher levels of consumer spending.

In many cases, stabilizers already take effect automatically.  The federal income tax is a good example.  As people lose income and fall into lower tax brackets, their tax burden falls by still-larger percentages, enabling them to retain more of their income.  On the spending side, more people become eligible for SNAP (formerly called food stamps) and unemployment insurance (UI), putting money in their hands to spend and, thus, relieving their hardship while boosting the economy.  In states with particularly high unemployment, jobless workers who exhaust their regular state UI benefits can get more weeks of benefits through the Extended Benefits (EB) program.

The depth of the Great Recession and the slow recovery, however, serve as poignant reminders that monetary policy and automatic stabilizers don’t always do enough.  Meanwhile, state balanced-budget requirements present a serious obstacle to recovery efforts.  Because recessions drain state revenues as people and businesses become less prosperous and pay fewer taxes, states must cut spending, raise taxes, or do both to avoid deficits, which puts a further drag on the economy.  These realities strengthen the case for countercyclical federal stimulus measures beyond the existing automatic stabilizers, even during recessions less severe than the most recent one.

For decades, policymakers have enacted temporary countercyclical measures during recessions to supplement the automatic stabilizers.  These measures have included additional weeks of unemployment insurance,[2] like the Emergency Unemployment Compensation (EUC) program that policymakers have enacted during or after every major recession since the late 1950s and which was in effect most recently from mid-2008 through 2013.  Policymakers also have cut taxes temporarily, provided temporary SNAP benefit increases, and allocated temporary fiscal relief for states.  These additional measures have reduced the depth and length of downturns and further alleviated the suffering associated with joblessness and income losses, especially for people with limited savings or access to credit.

Countercyclical stimulus measures were especially important during the Great Recession, which ran from December 2007 through June 2009 but, due to its depth and sluggish aftermath, has affected Americans for much longer.  The 2009 American Recovery and Reinvestment Act (ARRA) “provided more than $830 billion in stimulus measures, much of it in the first three years after its passage in February 2009; about three-fourths of this was temporary spending increases, and the other fourth was tax cuts,” economists Alan Blinder and Mark Zandi wrote in a recent paper for our Full Employment Project.[3]  “It worked.  The job losses started to abate immediately, and the Great Recession officially ended in June.”[4]

Blinder and Zandi’s analysis provides evidence that ARRA not only helped shorten the recession and lessen its severity, but that it also created jobs and raised output.[5]  They estimate that in 2010, for example, ARRA saved 2.6 million jobs and raised real GDP by 3.3 percent above what it otherwise might have been.  It also protected millions of people from falling into, or deeper into, poverty.[6]

But while ARRA was clearly effective, many of its interventions ended too soon, as the economic need for them persisted both at the macroeconomic level (growth and unemployment) and the household level.[7]  The “fiscal impulse” from government at all levels (federal, state, local) turned negative in 2011[8] — that is, fiscal policy impeded the recovery — even though unemployment remained well over 8 percent and the labor market had considerable room to grow.  Though much of the fall in countercyclical spending was due to state and local government actions, federal fiscal policy also pivoted to deficit reduction too soon, toward the end of 2010, while the job market was still weak, long-term unemployment was historically high, and a large gap remained between actual and potential GDP.

Moving forward in anticipation of further recessions, a stronger set of automatic stabilizers would help, though severe economic slumps like the Great Recession would still require policymakers to enact additional discretionary stimulus.  As outlined below, we recommend that policymakers:

  • Make UI’s EB program more responsive to economic conditions by having it take effect more quickly and remain in effect until hardship and labor market weakness are alleviated sufficiently, encourage “worksharing” among employees by creating incentives for it through UI, strengthen basic UI benefits, and bolster UI’s financing system;
  • Have temporarily higher SNAP benefits (and perhaps higher SNAP administrative funds for states) take effect automatically when a trigger, possibly tied to state unemployment rates, reaches certain thresholds;
  • Make state fiscal relief, in the form of higher federal payments to help states cover their Medicaid costs, take effect automatically, possibly via the same mechanism that is used to trigger a temporary increase in SNAP benefits; and
  • Prepare for additional discretionary steps during downturns by establishing a dedicated fund for subsidized jobs and job creation programs and considering one-time housing vouchers that can help struggling families keep their homes, pay their rents, and avoid homelessness.

Some policymakers, including some who support stronger countercyclical measures, may hesitate to enact them due to concerns that such measures would increase our debt as a share of GDP (which, at 76 percent, is high by historical standards).  In fact, some commentators worry that we already lack the “fiscal space” — the leeway to use fiscal policy — to fight the next recession.[9]  All else being equal, we agree that a lower debt-to-GDP ratio is better for the economy over the long run.  All else is not equal, however, especially during recessions.

The benefits of more effective fiscal stimulus measures to fight recessions outweigh the potential drawbacks of higher debt.  That’s particularly true if interest rates remain low, leaving the Federal Reserve less room to use monetary policy effectively and making it less costly to service the debt than it would otherwise be.  In a recent paper on fiscal policy, former Congressional Budget Office (CBO) Director Douglas Elmendorf and the Brookings Institution’s Louise Sheiner examined the long-term fiscal outlook and the need for fiscal space but nonetheless concluded, as we argue here, that “efforts should be made to build automatic fiscal stabilizers that are more powerful than existing stabilizers and that respond very rapidly.”[10]

Strengthening Automatic Stabilizers

Unemployment Insurance

Unemployment insurance (UI), a federal-state partnership that dates back to 1935, shows both how different types of countercyclical measures work and why strengthening automatic stabilizers can reduce the need for discretionary countercyclical measures that depend upon the actions of policymakers.  It provides income support to people who lose their jobs through no fault of their own.  It’s run by states, overseen by the U.S. Labor Department, and financed by taxes on employers.[11]  While states typically provide eligible unemployed workers with about half of their previous earnings (up to a specified maximum in the state) for up to 26 weeks, states exercise considerable discretion over eligibility, benefit levels, and benefit duration.  The EB program referenced above, which triggers additional weeks of benefits during economic downturns, is part of the UI system.[12] 

UI mitigates hardship for individuals and stimulates the economy when unemployment swells during downturns.  Because UI recipients may lack savings to tide them over until they find their next job, they’re likely to spend most of their UI benefits quickly, thus adding to overall consumer demand and helping to keep businesses afloat.  Blinder and Zandi estimate that each $1 of UI benefits during tough economic times adds $1.61 to GDP.[13]  In addition, one analysis found that the full spate of UI benefits in place during the second quarter of 2009 (the trough of the Great Recession) boosted employment by about 1.75 million jobs.[14]

Reforming EB.  Analysts have long recognized that policymakers should reform EB to make it more responsive to need — specifically, to long periods of high unemployment of the type that many states experienced during the Great Recession.[15]   EB’s inadequate responsiveness is an important reason why policymakers have enacted temporary UI expansions on top of EB in the aftermath of every recession since EB’s creation in 1970.

During the Great Recession, the President and Congress enacted Emergency Unemployment Compensation (EUC), which ran from June 2008 to December 2013 and offered additional weeks of UI in several tiers.[16]  Policymakers failed to extend EUC in 2014, however, though the need persisted: in the last quarter of 2013, long-term unemployment remained significantly elevated.[17]  That reality underscores the potential benefit of making EB automatically more responsive to economic conditions.  If a more robust EB system takes effect in a more timely fashion and remains in effect until hardship and labor-market weakness are alleviated sufficiently, it will mitigate the harm of recessions and reduce the need for policymakers to act (which they often fail to do on a timely basis).

Such a system could build on EUC’s successes by, among other improvements, basing multiple tiers of benefits on state economic conditions.  President Obama’s 2017 budget proposes an updated EB system which, similar to EUC, would have tiered benefits based on state unemployment rates to ensure that more weeks of benefits are available in states with higher rates.  The proposal also includes a provision that would allow EB to take effect in states with lower levels of but rapidly rising unemployment.[18]

These proposals would substantially improve EB and mitigate the need for discretionary measures in moderate recessions.  Still, using the unemployment rate as a trigger for EUC has its limits: it would only count people who have actively looked for work in the past four weeks as unemployed.  People who want a job but haven’t searched for one recently don’t count as part of the labor force and don’t show up in the unemployment rate.  Nor do people who are underemployed (working part-time despite their desire for a full-time job).  As a result, a low, flat, or declining unemployment rate doesn’t necessarily reflect a healthy labor market.  During the current recovery, even as unemployment has approached levels that many economists and policymakers believe are consistent with full employment, other indicators point to remaining labor market “slack” (that is, untapped potential).[19]

We therefore suggest a modified version of the Obama proposal, one that uses either a more comprehensive measure of underemployment that counts each category listed above (known as U-6) or the rate of long-term unemployment (a more direct indicator of the need for extra weeks of UI), or both.[20]  As Figure 1 shows, if one possible set of modified U-6 triggers had been in place before the most recent recession, benefits would have ended more slowly after the recession than they would have under the President’s proposed triggers, better reflecting remaining labor market slack.[21]

Whatever EB trigger policymakers eventually implement, they should include multiple tiers that account for both high levels of labor market slack and large increases in slack (for states with below-average slack during normal times).  Such triggers won’t necessarily eliminate all need for policymakers to enact additional discretionary UI measures — as noted earlier, that’s especially true during deep recessions — but they will improve UI’s basic function, thus helping to stimulate the economy and reduce hardship even when policymakers fail to act (or to act on a timely basis).

Encouraging Worksharing.  Lawmakers should encourage worksharing (as the President’s budget proposals do),[22] a policy in which firms reduce workers’ hours instead of laying off workers.  Workers who see their hours reduced could become eligible for prorated UI benefits. 

The 2012 Middle Class Tax Relief and Job Creation Act provided some worksharing incentives, but they expired in 2015 and worksharing usage remains low due to low employer awareness of the program, a cumbersome signup process, and participation bans on certain employers.  Former Bureau of Labor Statistics Commissioner Katharine Abraham and the Upjohn Institute’s Susan Houseman recommend several ways to increase the use of worksharing, including requiring all states to offer worksharing plans and guaranteeing federal funding for half of the UI benefits under approved worksharing plans whenever EB takes effect.[23]  As American Enterprise Institute economists Kevin Hassett and Michael Strain note, while the German and U.S. economies performed similarly during the Great Recession, Germany experienced significantly less unemployment in part due to its widely used worksharing program.[24]

Improving the Basic UI Program.  To maximize UI’s effectiveness as a countercyclical tool, policymakers should adopt additional improvements to the basic UI system.  Some states have restricted UI considerably in recent years, and cuts to the basic number of weeks available and to benefit levels, along with more restrictive eligibility requirements, inhibit UI’s ability to function as effectively as possible as an automatic stabilizer.  The President’s 2017 budget would address these issues by requiring all states to offer at least 26 weeks of benefits in their basic program and to expand coverage to part-time workers, newer labor market entrants, certain low-income and intermittent earners, and workers who have to leave their jobs for compelling family reasons.[25]  In a forthcoming paper, researchers from the Center for American Progress, Georgetown Center for Poverty and Inequality, and National Employment Law Project propose additional measures such as strengthening employment services, reforming eligibility criteria, and educating workers on their rights.

Finally, to help prepare for the next downturn, policymakers should strengthen UI’s financing system. Thirty-six states depleted their UI trust funds during the Great Recession and had to borrow from the federal government.  That borrowing was not unexpected given the recession’s length and depth (and, initially, a “jobless” recovery), but states would have borrowed far less if they had been better prepared.  Moreover, though the recovery has progressed for well over six years now, many states are again poorly prepared for the next recession.  For instance, the UI trust funds of three states remain in the red from the last downturn.  As of late February 2016, California owes the federal government $6.5 billion, Ohio owes $773 million, and Connecticut owes $101 million.[26]  In addition, many other states have trust funds that are far from adequate.  Some 34 state trust funds currently don’t meet DOL's minimum standard for being prepared for a recession.[27]

One problem is that many states only apply the unemployment insurance tax to a very small share of workers’ wages and haven’t adjusted the level for many years, for either inflation or wage growth.  In California, the tax applies only to the first $7,000 of wages, the minimum required under federal law, and 17 other states set their taxable wage base under $10,000.  The federal government should raise the $7,000 wage base minimum, which it hasn’t done in over 30 years.  But absent federal action, states with low taxable wage bases should raise them and also index them to keep pace with wage growth over time.

States also should avoid sharp cuts in UI benefits.  To address the strain in their UI systems in recent years, states such as Georgia, Florida, and North Carolina substantially reduced the maximum number of weeks that jobless workers may receive UI, cut the size of benefits, or both.  These harsh cuts diminish UI’s power to limit the depth and severity of future recessions; they also impose unnecessary added hardship on workers who lose their jobs through no fault of their own.  Besides increasing the taxable wage base, the federal government can help by establishing new incentives for states to build up their trust funds adequately during good economic times, such as by offering a higher interest rate on trust fund balances that exceed the Labor Department’s minimum standard for adequacy.[28]


SNAP is the nation’s most important anti-hunger program, helping 46 million Americans afford an adequate diet during a typical month in 2015.[29]  It reaches a broad range of the poor and near poor, including families with children, senior citizens, and other individuals; 85 percent of those eligible participated in 2013 (the latest year for which we have estimates).[30]

Along with providing vital nutritional support to low-income families and improving children’s long-term outcomes,[31] SNAP is a powerful automatic stabilizer.  During recessions, as unemployment rises, incomes fall, and poverty increases, more people become eligible for SNAP.  Open to everyone who meets eligibility criteria, it’s designed to expand during times of increased need and contract when the economy improves.  After UI, it’s the most responsive federal program during downturns.[32]  

SNAP is also well situated to provide additional countercyclical stimulus.  It has an efficient administrative infrastructure and can get benefit increases into low-income families’ pockets very quickly.  States provide monthly benefits through electronic benefit transfer (EBT) cards, and SNAP is targeted to low-income families — those most likely to spend additional income quickly and thereby inject more consumer purchasing into the economy.  Four-fifths of SNAP benefits are redeemed within two weeks of receipt, 97 percent within one month.  While SNAP benefits can buy only food, they free up resources with which families and individuals can meet other needs.

Capitalizing on these SNAP features, ARRA provided a temporary 13.6 percent increase in the maximum monthly benefit.  A three-person household, for example, received a benefit boost of $63 per month beginning in April 2009, less than two months after ARRA’s enactment, while a one-person household received a benefit boost of $24 per month.

ARRA also gave states additional administrative funds to handle growing caseloads during the recession.  Because SNAP administrative costs are shared with states, and state budgets are typically very tight during recessions, this measure proved a helpful countercyclical tool as well.

The $42.8 billion in additional SNAP benefits that ARRA provided between 2009 and 2014 had their intended effect, the evidence suggests.  Blinder and Zandi estimate that every $1 of higher SNAP benefits during the first quarter of 2009 generated $1.74 of economic activity, a higher bang-for-the-buck than that of any other tax or spending action they examined.[33]  The higher SNAP benefits also substantially reduced food insecurity and kept close to 1 million people out of poverty in 2010 alone (along with the several million people that regular SNAP benefits kept out of poverty that year).[34]

Due to their success, SNAP provisions similar to those enacted under ARRA — particularly the temporary benefit increase — should take effect automatically when future recessions hit.  Specific thresholds for when that would occur will always be a bit arbitrary, but here’s one idea that would have worked well during past business cycles: an automatic benefit increase if at least 26 of the 51 states (counting Washington, D.C.) experience rising unemployment in four consecutive months.  Had this trigger been in place in the last downturn, the SNAP benefit increase would have taken effect in June 2008.[35]  Another possibility is a trigger that the Government Accountability Office (GAO) examined for Medicaid, under which a temporary benefit increase would take effect when more than half of all states see falling three-month-average employment-to-population (EPOP) ratios (relative to the previous year) in two consecutive months.[36]  Although such a trigger wouldn’t have taken effect in as timely a manner as the unemployment trigger cited above during the 1990-1991 and 2001 recessions, the higher SNAP benefits would have taken effect in January 2008 during the most recent recession.[37]

Due to SNAP’s powerful stimulative features, policymakers should make the size of its benefit increase proportional to the severity of the downturn.  As SNAP has a uniform national benefit structure, higher countercyclical SNAP benefits should take effect equally in all states, as they did in 2009.

An automatic, formulaic end to these higher benefits could be problematic.  ARRA was designed to address this problem by freezing SNAP benefit levels after the higher benefits took effect, allowing inflation to gradually erode their value over time until policymakers could apply regular annual food price inflation adjustments to them again.  We believe this approach is sound.  During the Great Recession, however, food price inflation rose much more slowly than CBO had predicted.  At one point, CBO estimated that, rather than phasing out by 2013 as originally expected when ARRA was enacted, the temporary SNAP benefit increase would remain in partial effect for several additional years.  That possibility prompted lawmakers to accelerate the end of the SNAP benefit increase.  As a result, SNAP recipients endured an immediate benefit cut in November 2013, with one-person households losing $11 a month and three-person households losing $29 per month. 

In future recessions, policymakers could provide a temporary SNAP benefit increase that inflation will erode over time, but also provide a schedule for phasing down the benefit increase if food price inflation ends up substantially lower than CBO had anticipated when the benefit increase took effect.

Block Grant Proposal Would Negate SNAP’s Countercyclical Role

Some policymakers propose to convert SNAP into part of a large block grant, under which the federal government would send states a fixed amount of money each year to run SNAP and a number of other programs.  Among its multiple problems, this misguided approach would negate SNAP’s countercyclical role.

SNAP, which provides benefits to all eligible recipients, expands to reflect greater need when the economy falters and then contracts as need shrinks during an economic recovery.  By definition, block grants do not respond automatically to economic fluctuations because their funding is fixed, not flexible.

Block grant proponents respond that federal policymakers could adjust the annual state grant amounts when state unemployment rates rise, but that wouldn’t solve the problem.  Because policymakers would set block grant levels at the beginning of the year, the levels would quickly become out of date if the economy sputters.  Moreover, even if federal policymakers enacted in-year changes to boost block grant levels, states would find it difficult to administer those changes.  Policymakers would find it virtually impossible to design a trigger for higher benefits under a block grant that responds to rising poverty and need as well as the current SNAP program.a

To further recognize the limits of a block grant for countercyclical purposes, consider the stark difference between the responsiveness of SNAP and that of the Temporary Assistance for Needy Families (TANF) block grant during the Great Recession.  While SNAP’s rolls increased by nearly 50 percent as more people became eligible for it, TANF’s caseloads increased by only 13 percent — and only because states reduced spending in other areas, tapped reserves, and/or received additional funding through the TANF Emergency Fund, a temporary revenue stream created by ARRA that might not be replicated in a future recession.b  

a For more information, see Robert Greenstein, “Commentary: Ryan ‘Opportunity Grant’ Proposal Would Likely Increase Poverty and Shrink Resources for Poverty Programs Over Time,” Center on Budget and Policy Priorities, July 25, 2014,

b LaDonna Pavetti, Ph.D. and Liz Schott, “TANF’s Inadequate Response to Recession Highlights Weakness of Block-Grant Structure,” Center on Budget and Policy Priorities, July 14, 2011,

State Fiscal Relief

State fiscal relief — temporary federal aid to state governments — is another essential countercyclical tool.

Unlike the federal government, states must balance their operating budgets each year.  Thus, in an economic downturn, with their personal and corporate tax receipts falling, states cut spending and raise taxes to meet their balanced-budget requirements, impeding federal efforts to revive the economy through federal tax cuts and spending increases.  In 2010, for instance, budget tightening at the state and local levels worked against the federal stimulus.[38]

State fiscal relief reduces the need for states to cut vital services or raise taxes, or both, in a recession.  As with UI and SNAP, economists have found state fiscal relief an effective stimulus tool.[39]

Under ARRA, state fiscal relief came in two main forms.[40]  First, it boosted the Federal Medical Assistance Percentage (FMAP), the federal share of Medicaid spending.  That share varies from state to state — from a high of nearly 75 percent in the poorest states to a low of 50 percent in the wealthiest states — and averages 57 percent across the states.[41]  ARRA provided an across-the-board bump to the federal share for each state and another increase for states with particularly high unemployment compared to pre-recession levels.  (Policymakers also boosted the FMAP to provide fiscal relief to states after the 2001 recession.)  Through the FMAP boost, the federal government allocated approximately $87 billion to the states between October 2008 and December 2010.  In exchange, states were required to maintain their Medicaid and Children’s Health Insurance Program (CHIP) eligibility levels and not tighten eligibility procedures in ways that affected enrollment (or renewal) among eligible beneficiaries.  Second, ARRA included a “State Fiscal Stabilization Fund” that provided two block grants to states — a $39.5 billion one for education and an $8.8 billion one for other key services.[42]

All told, the roughly $135 billion that ARRA provided[43] to states filled “between 30 percent and 40 percent of projected state shortfalls for fiscal years 2009, 2010, and 2011.”[44]  Though states made substantial budget cuts — including to long-term investments such as education — and some raised taxes, those budget cuts and tax increases would have been significantly larger in the absence of fiscal relief. 

Unfortunately, although policymakers extended the FMAP boost through June 2011, “federal assistance largely ended before state budget gaps had fully abated.”[45]  Consequently, states were forced to cut spending or raise taxes — or both — even more, further impeding the recovery.  To improve the timing of state fiscal relief in future recessions, policymakers should enact permanent triggers for FMAP increases that would tie closely to the economy’s performance.  Like UI’s EB program and a temporary SNAP benefit increase, FMAP increases would then serve an automatic stabilizing function during times of economic weakness.[46]

The options discussed above for triggering enhanced SNAP benefits also could work for triggering enhanced FMAP.  Under the proposal that the GAO designed for a countercyclical FMAP trigger, for example, the relief would end when half or fewer of the states experienced declining three-month-average EPOPs (again, compared with the previous year) during the two preceding months.[47]

Under that option, enhanced FMAP would have taken effect in January 2008 and ended in September 2011, as opposed to the October-2008-to-June-2011[48] period when it was actually in place under ARRA.  Another option would be to use the four-consecutive-months-of-rising-unemployment trigger discussed above and have enhanced FMAP end automatically when unemployment returns to pre-trigger levels in at least ten states.  Such a mechanism would have ended the enhanced FMAP benefits in May 2014, better reflecting the need for stimulus beyond 2011.

Finally, states should get advance notice before a temporary FMAP increase ends so they can budget efficiently, enter into contracts in a timely fashion, and run their programs effectively.

Discretionary Measures

Though we should strengthen our automatic stabilizers to better anticipate and respond to recessions, as per the proposals described above, no one can design such stabilizers for all future recessions.  The downturns will differ in their cause (e.g., a housing crisis, an energy shock, and so on), their duration, and their depth, and each will bring its own particular kind of hardship.  As a result, policymakers must be open to taking steps beyond the automatic stabilizers to confront the hardship of a future recession and help prod the economy back to recovery.

Subsidized Jobs and Job Creation

Subsidized jobs and job creation programs can serve two major purposes.  First, they provide income to people who need it and will spend it, thereby helping the economy move from recession to recovery.  Second, especially for those disconnected from the labor market, the jobs can potentially provide long-term benefits by boosting workers’ skills and job experience, thereby increasing their employability and earnings potential.

ARRA sponsored $1.3 billion of subsidized jobs programs in 39 states, Washington D.C., Puerto Rico, and the Virgin Islands through the TANF Emergency Fund (TANF EF), which operated in 2009 and 2010.  The 260,000 subsidized slots that the funds financed were mainly in the private sector; some were targeted to adults, others to youth.

After studying programs in Los Angeles, San Francisco, Mississippi, Wisconsin, and Florida, the Economic Mobility Corporation — which focuses on programs and policies designed to help the disadvantaged find jobs — concluded that most employers who participated in TANF EF said they created jobs that they would not have otherwise created.[49]  The corporation also found evidence that TANF EF provided longer-term benefits for participants after their subsidized job placements ended.

In Florida, people who participated in a subsidized jobs program earned $4,000 more, on average, in the year after the program than they had in the year before it, whereas those who applied for but didn’t get a subsidized job earned $1,500 more.  The benefits were particularly pronounced for the most disadvantaged.  In Mississippi, average earnings for the long-term unemployed who participated had risen about twice as much as average earnings for unemployed participants as a whole a year after their subsidized placements ended.  And for a group of “targeted non-custodial parents” in Wisconsin, employment rose 15 percentage points between the year before and the year after their time in the program.  These results bolster the case for such job creation programs not just as a countercyclical tool, but also as an ongoing investment (on a smaller scale) during better economic times. 

Consequently, policymakers should provide a consistent, dedicated funding stream (an “employment fund”) that can support job creation efforts and expand when the economy is weak.  That arrangement would achieve two objectives.  First, it would provide a continuing level of job creation for the needy, especially those whom employers might not otherwise hire (particularly for a first job), even in good times.  Second, it would create a job-creation infrastructure that policymakers could ramp up in times of high need and ramp down to normal levels during better economic conditions. An employment fund with a dedicated funding stream could also help address important national needs. 

The Commission on Inclusive Prosperity, co-chaired by former U.S. Treasury Secretary Larry Summers and former U.K. Shadow Chancellor of the Exchequer Ed Balls, proposed such an idea in a January 2015 report.[50]  National service programs like AmeriCorps, Volunteers in Service to America (VISTA), or the National Civilian Community Corps (NCCC), the authors argued, are perfect targets for countercyclical job creation.  These programs build leadership capacity and job skills while simultaneously helping participants make a difference in the lives of individuals who are struggling to make ends meet — and often in the communities in which those individuals live.  The programs also expand and contract relatively easily.  The commission proposal suggests 500,000 fully funded national service jobs as a constant baseline, which would be twice as many as the Serve America Act now authorizes.

We recommend a hybrid approach: an employment fund that supports a set of national service jobs on an ongoing basis and includes a flexible funding stream that can ramp up in economic downturns.  This initiative should enable states to try different approaches to subsidized jobs, encouraging them to experiment to learn more about what works best and for whom.

To maximize its effectiveness, the employment fund should reflect the following four principles:

  • Funding should cover three types of costs: wages and other payroll costs, services (like child care) that support participation, and administrative costs.  The program should cap administrative and support services costs each at a relatively low share of the funds, possibly 15 percent, ensuring that most of the funds go to direct compensation.
  • Countercyclical funds should be allocated for at least two years.  Job creation programs require advance planning and development.  By establishing a constant infrastructure, a dedicated employment fund will shorten the time needed to create countercyclical jobs.  But the federal government, states, localities, and nonprofits will still find it difficult to plan how to use countercyclical funds effectively without knowing how long they will last.
  • During normal economic times, job placements should go first to people with the greatest need.  These individuals may include the long-term unemployed, residents of high-poverty neighborhoods, members of families that receive TANF cash assistance, those with criminal records, youth, or others — with states having some discretion over which groups to target.
  • Program rules should protect against displacing existing workers: the program should bar employers that receive funds from the flexible stream from replacing existing employees with subsidized workers.


During downturns, whether or not the economic contraction is driven by housing market problems, families that lose income face a greater risk of housing instability and homelessness.

For that reason, ARRA’s Homelessness Prevention and Rapid Re-Housing Program (HPRP) was important in helping very poor families who were homeless or at risk of becoming homeless.  Between July 2009 and September 2011, it drew on $1.5 billion to serve about 1.15 million people in about 470,000 households.[51]  These numbers are particularly striking since the program lacked adequate administrative infrastructure in many places when it was enacted.  HPRP also provided countercyclical stimulus, enabling those at risk of losing their housing to spend more of their income on other basic needs like food, clothing, transportation, and medical care.

HPRP now has a more established infrastructure and, as with an employment fund, can and should ramp up in times of need and scale down during normal times.  Additionally, the program can use funds not just for renters, but also to prevent homeowners from losing their homes and to help families and individuals without a home secure a stable place to live.  HPRP-type funds, for instance, could help families temporarily make their mortgage payments without the complexity of some of the service-intensive programs that the Obama Administration has tried.

At the same time, HPRP limited assistance to 18 months.  During recessions, needy families, particularly those that can’t find jobs that pay enough to cover rent, need another tool.  One-time housing vouchers would be a good option.  Under the current housing choice voucher program, low-income families receive subsidies they can use to help pay for rental housing in the private market.[52]  These vouchers are very effective at helping to address homelessness.[53]  Recent research by Raj Chetty and colleagues also suggests that good neighborhoods, to which housing vouchers help low-income families gain access, have substantial and lasting positive long-term impacts on younger children.[54]

One-time housing vouchers would work like regular housing vouchers with one exception: they wouldn’t become an ongoing part of a state or local housing agency’s supply of vouchers.  They would help only the families issued the one-time vouchers during the recession-response period; when these families left the program, the vouchers would disappear (rather than go to other families).  The federal government has used this approach for vouchers issued after major disasters, and the Department of Housing and Urban Development has such a policy now for certain other special types of vouchers.  Through this approach, the federal government can wind down its stimulus efforts without hurting current beneficiaries.


The current recovery is well into its sixth year, making it older than the average economic expansion of recent decades.  We don’t know when the next recession will hit, but we would be wise to consider the lessons learned from the last recession to prepare the most effective countercyclical responses to the next one.

UI and SNAP are important automatic stabilizers that ramp up during times of hardship.  By leveraging their effective countercyclical features and high bang-for-the-buck and creating additional automatic stabilizers within these programs, our fiscal policy can become still more helpful in stabilizing the economy and more responsive to need.  Introducing automatic triggers for state fiscal relief also would be beneficial.

Still, recessions differ in their causes, severity, and duration.  The President and Congress will often need to supplement Federal Reserve actions and the automatic stabilizers with discretionary stimulus measures — even with the enhanced stabilizers that we recommend here.[55]  Policymakers can lay the groundwork for such steps, for example, by acting now to establish the infrastructure for a job creation program.  Then, when the need for discretionary measures strikes, they can draw upon lessons from TANF EF and HPRP, among other programs.

Whether or not policymakers agree with the specific trigger mechanisms that we suggest, we encourage them to think proactively about actions they can take in advance of the next downturn.  Given our knowledge of countercyclical policies that can reduce the economic pain that recessions cause, there is no reason not to be ready for the next one when it hits.


The authors thank our colleagues at the Center on Budget and Policy Priorities, including Matt Broaddus, Stacy Dean, Joel Friedman, Bob Greenstein, Larry Haas, Brynne Keith-Jennings, Mike Leachman, Liz McNichol, Edwin Park, LaDonna Pavetti, Dottie Rosenbaum, Barbara Sard, Isaac Shapiro, Chad Stone, and Paul Van de Water, for much useful feedback. 

We also appreciate the contributions of Maurice Emsellem, Mitchell Hirsch, and Claire McKenna from the National Employment Law Project, David Kamin from NYU, Rachel West from the Center for American Progress, and Indivar Dutta-Gupta and Kali Grant from the Georgetown Center for Poverty and Inequality.

End Notes

[1] Olivier Blanchard, Eugenio Cerutti, and Lawrence Summers, “Inflation and Activity — Two Explorations and their Monetary Policy Implications,” International Monetary Fund, November 2015,

[2] Julie M. Whittaker and Katelin P. Isaacs, “Extending Unemployment Compensation Benefits During Recessions,” Congressional Research Service, May 2, 2013,

[3] Alan S. Blinder and Mark Zandi, “The Financial Crisis: Lessons for the Next One,” Center on Budget and Policy Priorities, October 15, 2015,

[4] Note that stimulus measures were also enacted in 2008, before the full effects of the financial crisis had become manifest.

[5] Compared with a situation in which the federal government enacted no fiscal measures to stimulate the economy.

[6] Arloc Sherman, “State-Level Data Show Recovery Act Protecting Millions From Poverty,” Center on Budget and Policy Priorities, December 17, 2009,

[7] A discussion of the need for more adequate baseline benefit levels for households during normal times is important but outside the scope of this paper.

[8] Luke Kawa, “Goldman: Government Spending Is About to Boost the U.S. Economy for the First Time Since 2010,” Bloomberg Business, November 3, 2015,

[9] The International Monetary Fund states that “fiscal space” can be defined as room in a government´s budget that allows it to provide resources for a desired purpose without jeopardizing the sustainability of its financial position or the stability of the economy. Peter Heller, “Back to Basics — Fiscal Space: What It Is and How to Get It,” Finance and Development, International Monetary Fund, June 2005,

[10] Douglas Elmendorf and Louise Sheiner, “Federal Budget Policy with an Aging Population and Persistently Low Interest Rates,” Hutchins Center on Fiscal and Monetary Policy, Brookings Institution, February 5, 2016, p. 48,

[11] As noted in the introduction, the federal government has historically provided additional funding for the UI program during economic downturns, beyond the revenue that employer taxes provide.

[12] Chad Stone and William Chen, “Introduction to Unemployment Insurance,” Center on Budget and Policy Priorities, updated July 30, 2014,

[13] Blinder and Zandi, “The Financial Crisis: Lessons for the Next One.”

[14] Dr. Wayne Vroman and Jacob M. Benus, “The Role of Unemployment Insurance As an Automatic Stabilizer During a Recession,” IMPAQ International, July 2010,

[15] The basic EB trigger depends on a state’s insured unemployment rate (IUR), or the number of UI recipients as a percentage of all people in the state who (if they lose their jobs) are potentially eligible for UI.  (The IUR is a different metric than the total unemployment rate, or TUR, which is the traditional unemployment rate reported every month by the Bureau of Labor Statistics.)  It is fairly complicated and its threshold is rarely breached.  A state may also adopt an optional set of EB triggers based on its TUR, but states don’t always take this option and, even when they do, benefits still only last a maximum of 20 weeks.

[16] At EUC’s peak, 53 additional weeks of UI were available to states in a top TUR tier, and all states qualified for at least 34 additional weeks of UI.  While EB is generally funded through a combination of federal and state dollars, the stimulus also temporarily committed the federal government to paying for the full EB costs, leading many states to adopt EB’s optional TUR triggers.

[17] The long-term unemployment rate is the number of people looking for work for at least half a year as a share of the labor force.  The long-term unemployment rate was 2.5 percent in the last quarter of 2013.  The last time this rate was around 2.5 percent, in 1983, the federal government was providing additional benefits for jobless workers in many states.

[18] “State Unemployment Insurance and Employment Service Operations,” FY 2017 Congressional Budget Justification, Employment and Training Administration, p. 21,

[19] Jared Bernstein, “What is the ‘natural rate’ for u6?,” On The Economy, November 23, 2015,

[20] Currently, the Bureau of Labor Statistics only tracks state U-6 rates quarterly, which is a notable disadvantage and due in large part to small sample sizes, but if the rates could be tracked monthly, they would provide an advantage over state TURs.

[21] Figure 1 is based on quarterly data and shows the percentage of states in which EB would have taken effect during a particular quarter (the graph doesn’t account for data lags); the U-6 triggers used are 11.5 percent, 12.5 percent, 13.5 percent, and 14.5 percent, while the Administration’s proposed triggers (6.5 percent, 7.5 percent, 8.5 percent, and 9.5 percent) use the TUR, which is also called U-3.  Both proposals incorporate a provision that activates a trigger threshold in a state with below average but rapidly rising unemployment.  As mentioned in the previous footnote, an important limitation is that these data are quarterly, so the modeling in the graph is based on quarterly data for each series, but it shows the advantage of U-6 if it could be tracked on a monthly basis.

[22] The White House, Office of the Press Secretary, “FACT SHEET: Improving Economic Security by Strengthening and Modernizing the Unemployment Insurance System,” January 16, 2016,

[23] Katharine G. Abraham and Susan N. Houseman, “Proposal 12: Encouraging Work Sharing to Reduce Unemployment,” The Hamilton Project,

[24] Kevin A. Hassett and Michael R. Strain, “Worksharing and Long-Term Unemployment,” Center on Budget and Policy Priorities, April 2, 2014,

[25] See “FACT SHEET: Improving Economic Security by Strengthening and Modernizing the Unemployment Insurance System” and “State Unemployment Insurance and Employment Service Operations,” p. 20.

[26] Department of Labor, Employment and Training Administration, “Trust Fund Loans,”

[27] Defined as having an “Average High Cost Multiple” (AHCM) of at least 1.0 (the number of states that fail to meet this standard is based on data from the third quarter of 2015; see Department of Labor, Employment and Training Administration, “Unemployment Insurance Data Summary,”  The AHCM provides an estimate of how long a state’s trust fund balance could pay UI benefits if a recession hit; an AHCM of 1.0 implies that a state could pay full benefits for a year.

[28] See Michael Leachman et al., “Rebuilding the Unemployment Insurance System: A Deficit-Neutral Plan That Limits Tax Increases and Maintains Benefits,” Center on Budget and Policy Priorities, February 8, 2011, for more discussion.  Another option is the President’s 2017 budget proposal to penalize states with inadequate trust fund balances by reducing the credit that employers in those states receive against federal UI taxes (see Department of the Treasury, “General Explanations of the Administration’s Fiscal Year 2017 Revenue Proposals, February 2016, p. 197,

[29] Department of Agriculture, Food and Nutrition Service, Supplemental Nutrition Assistance Program (SNAP), National Level Annual Summary, “Participation and Costs, 1969-2015,”

[30] Department of Agriculture, Food and Nutrition Service, “Trends in Supplemental Nutrition Assistance Program Participation Rates: Fiscal Year 2010 to Fiscal Year 2013 — Summary, August 2015,

[31] Hilary W. Hoynes, Diane Whitmore Schanzenbach, and Douglas Almond, “Long Run Impacts of Childhood Access to the Safety Net,” National Bureau of Economic Research, November 2012,

[32] Chad Stone, Arloc Sherman, and Brynne Keith-Jennings, “No Mystery Why SNAP Enrollment Remains High: It’s Still the Economy,” Center on Budget and Policy Priorities, Updated March 18, 2015,  Note that non-disabled childless adults between the ages of 18 and 50 qualify for only three months of SNAP benefits (in a 36-month period) while they’re unemployed or not enrolled in a work or training program for at least 20 hours per week.  States are not required to offer these individuals a place in a work or training program, and most states don’t; individuals who search for a job but can’t find one face the three-month cutoff.  One aspect of SNAP’s countercyclical design is that states may request waivers of this time limit for areas of high unemployment.  As unemployment grew during the recent recession, almost every state became eligible to suspend the time limit statewide.  (ARRA waived the time limit in all states temporarily between April 2009 and September 2010; five states opted to implement the time limit anyway in at least parts of the state.)

[33] Blinder and Zandi, “The Financial Crisis: Lessons for the Next One.”

[34] Arloc Sherman, “Poverty and Financial Distress Would Have Been Substantially Worse in 2010 Without Government Action, New Census Data Show,” Center on Budget and Policy Priorities, November 7, 2011,

[35] Unemployment rose in 26 or more states from February through May of 2008, but the benefit increase would have turned on in June, as the data are reported with a one-month lag.  This trigger also would have worked reasonably well in the previous four recessions dating back to 1980; the stimulus would have kicked in by the fifth month of the recession in each case.

[36] The standard EPOP is the share of the population aged 16 and over with a job.  That population will include an increasing number of retirees over time as baby boomers move into their golden years, and the EPOP will likely be on a declining path even in a strong economy.  If the data were reliable, this trigger could be improved by using the prime-working-age (25-54) EPOP rather than the age-16-and-over EPOP.

[37] Government Accountability Office, Report to Congressional Committees, Medicaid, “Prototype Formula Would Provide Automatic, Targeted Assistance to States during Economic Downturns,” November 2011,

[38] Luke Kawa, “Goldman: Government Spending Is About to Boost the U.S. Economy for the First Time Since 2010.”

[39] Gabriel Chodorow-Reich et al., “Does State Fiscal Relief During Recessions Increase Employment? Evidence from the American Recovery and Reinvestment Act,” American Economic Journal: Economic Policy, 2012,

[40] None of the discussion in this paragraph depends on whether or not a state has adopted the Affordable Care Act’s Medicaid expansion.

[41] Department of Health and Human Services, Office of the Secretary, “FY2017 Federal Medical Assistance Percentages,” December 29, 2015,

[42] Iris J. Lav, Nicholas Johnson, and Elizabeth McNichol, “Additional Federal Fiscal Relief Needed to Help States Address Recession’s Impact,” Center on Budget and Policy Priorities, updated March 1, 2010, ARRA also provided additional programmatic funding for education.  

[43] Additional state fiscal relief enacted in the 2010 jobs bill raised the total above $135 billion.

[44] Phil Oliff, Chris Mai, and Vincent Palacios, “States Continue to Feel Recession’s Impact,” Center on Budget and Policy Priorities, updated June 27, 2012,

[45] Oliff, Mai, and Palacios, “States Continue to Feel Recession’s Impact.”

[46] One concern with state fiscal relief is that it can to some extent supplant rather than supplement existing state resources.  If states reduce their own contributions to needed services (either by cutting taxes or diverting state funds to other purposes) when they receive federal funding, federal aid won’t necessarily prevent cuts to important programs.  As noted earlier, maintenance of effort (MOE) provisions during the Great Recession, under which states were required to maintain Medicaid eligibility levels and procedures in place in the fall of 2008, can help (though MOE requirements related to state spending levels are much harder to enforce).  Robert Greenstein, “Commentary: The Ryan ‘Opportunity Grant’ Proposal – A Reply to Scott Winship,” Center on Budget and Policy Priorities, September 9, 2014,

Another concern is that state aid could create a “moral hazard” problem wherein states that didn’t prepare for recessions (that is, those that failed to build adequate rainy day funds) are rewarded for irresponsible budgeting.  However, evidence from past business cycles suggests that states have not systematically “overspent or slashed taxes since then in the expectation they would be bailed out during future downturns.” (See Chad Stone and Kris Cox, “Principles for Fiscal Stimulus Economic Policy in a Weakening Economy,” Center on Budget and Policy Priorities, revised January 11, 2008,

[47] As noted above, the trigger would turn on following two consecutive months of falling three-month-average EPOPs (relative to the previous year) in more than half of states.  The FMAP increases under this proposal would initially depend on changes in state unemployment rates and salaries and wages (relative to their minimum and maximum levels, respectively) over an 8-quarter “look back” period and would be updated each quarter.

[48] Though the funds weren’t available until ARRA’s enactment in February of 2009, states could get retroactive reimbursement dating back four months.

[49] Displacement, wherein a subsidized worker replaces an incumbent worker, can be a problem in these programs, but a strong program design can reduce its likelihood.

[50] Lawrence H. Summers and Ed Balls (Co-Chairs), “Report of the Commission on Inclusive Prosperity,” Center for American Progress, January 2015,

[51] Department of Housing and Urban Development, “Homelessness Prevention and Rapid Re-Housing Program (HPRP): Year 2 Summary,” February 2013,

[52] “Policy Basics: The Housing Choice Voucher Program,” Center on Budget and Policy Priorities, updated September 29, 2015,

[53] Douglas Rice, “Major Study: Housing Vouchers Most Effective Tool to End Family Homelessness,” Center on Budget and Policy Priorities, July 14, 2015,

[54] Douglas Rice, “Groundbreaking Studies: Good Neighborhoods Help Low-Income Children Succeed,” Center on Budget and Policy Priorities, May 4, 2015,

[55] Enhanced automatic stabilizers, of course, would reduce the scale of the discretionary measures needed.