Mulligan on Redistribution, Unemployment, and the Labor Market
What this covers
Casey Mulligan of the University of Chicago discusses his book The Redistribution Recession with Russ Roberts in a conversation that traces the 2007 labor market collapse not to aggregate demand failures or structural rigidity, but to a stacked set of policy changes in safety-net programs. Mulligan catalogs how expansions in Unemployment Insurance, food stamps, mortgage relief, and health insurance subsidies coincided to raise effective marginal tax rates on work substantially. The core mechanism is straightforward: when the cost of remaining unemployed falls through enlarged benefits, workers have less incentive to accept available work, employers find labor more expensive relative to output, and replacement rates climbing from roughly 70% to 85–90% move people across the work-on-the-margin into joblessness.
The conversation moves across several empirical and conceptual territories. Mulligan presents data on gross vs net job flows to challenge the "no jobs exist" narrative, and uses shifts in quit-versus-layoff classifications to argue that generous UI made separations be coded as involuntary rather than voluntary. He argues that endogenous stickiness reverses standard Keynesian causality: wages did not fall because workers receiving better unemployment terms refused lower pay, not because prices are sticky in some mechanical sense. He then turns to democratic vs non-democratic measures to explain why policies hammering only the rich might not reduce employment counts much while crushing GDP. The pair also contest whether the Keynesian multiplier argument for transfer spending holds water, whether international evidence supports or undermines his labor-supply story, and whether the scale of incentive change he measures is large enough to explain the observed labor market depression. Throughout, Mulligan insists on measure the magnitude rather than merely noting that incentives moved in some direction.
Mulligan argues that the depth and persistence of the post-2007 labor market collapse was substantially caused by expansions in safety-net benefits that raised effective marginal tax rates on work, eroding the incentive to work and to retain workers, rather than by aggregate demand failures or labor market stickiness imposed from outside.
- Many small simultaneous rule changes (UI to 99 weeks, food stamp expansions, mortgage relief, health insurance subsidies) added up to large incentive changes
- These changes pushed effective replacement rates from ~70% to ~85-90%, enough to move millions off the working side of the margin
- Endogenous wage stickiness and reduced spending follow from the incentive changes, reversing the Keynesian causal story
Safety-net rule changes directly reduced employment by lowering the cost of being unemployed.
- Employers and employees constantly weigh whether to continue an operation, and one factor is how painful a layoff would be; by making unemployment less painful, government benefits gave both sides less reason to take wage cuts, work harder, or cut prices to retain jobs, so the spike in 2008 layoffs reflected rational responses rather than accident.
“by making unemployment less painful, the government gave employers and employees less reason to retain the jobs that we already have. So, I don't think it's an accident that we had a spike in layoff in 2008.”
- The largest single expansion was extending unemployment insurance from the normal 26 weeks to almost two years (99 weeks), which put many people in a position to receive benefits they would otherwise have lost after 27 weeks under the old rules.
“The number one expansion would be the infamous 99 weeks of unemployment insurance. Normally it's 26 weeks, a half a year. And it went up to almost 2 years.”
- By quantifying changes in safety-net program rules (UI, Medicaid, food stamps) before versus during the recession, and applying established estimates of how benefit generosity affects labor supply, about half or a little more of the depression in the labor market can be linked directly to these new rules for helping the unemployed and low-income.
“it turns out about half, probably a little more of the depression of the labor market you can link directly to these new rules for helping people when they are unemployed or have low incomes.”
- Because government and private-sector help made being without a job less painful than it would have been five, ten, or fifteen years earlier, more people end up without a job, since when the cost of being unemployed falls, the quantity of unemployment rises.
“when it's less painful to be without a job there will be more people who are without a job.”
- Starting from a roughly 70% baseline replacement rate, the new recession-era rules pushed it up to 85 or 90%, which cannot be called trivial, and meant millions of laid-off people took less than a 20% pay cut—the kind of cut even forward-looking retirees voluntarily accept—while one or two million people actually had more to spend after being laid off.
“it pushes the 70% up to 85 or 90%. I don't think we can call that trivial any more.”
- Because there are always people near the margin between working and not working—roughly 140 million working and 140 million adults not working—even a small change in what they are paid on one side will move some off the fence, so it took only a modest benefit increase to move on the order of 10 million people, about 8-10% of the population, onto the not-working side.
“people who are on that margin and you change what you pay them on one side of the fence, they are going to fall off the fence.”
- The large drop in the quit rate during the recession is evidence for, not against, the incentive story: a separation is like a breakup whose labeling matters for benefits, since quitting forfeits UI while being laid off triggers benefits, so widening the benefit spigot caused more separations to be classified as layoffs rather than quits.
“if the employer steps forward and says, I initiated this breakup, that turns on a whole spigot of benefits for people. So it's no surprise to me that when we increase the spigot of benefits... more of these breakups were classified as layoffs”
Multiple program expansions compounded incentive effects: UI extension, food stamps, Medicaid, mortgage relief, and health insurance subsidies.
- There is no single program or rule that explains most of the labor market damage; rather, many relatively small rule changes occurred at the same time and all pushed in the same direction, so that even the largest single change (UI extension) would account for only a minority of the total effect.
“there's not a single program or single rule you can point to and give a lot of the blame. There were a lot of relatively small changes happening at the same time, all in the same direction.”
- The food stamp benefit amount per family was increased twice—about 15% each in late 2008 and again six months later—for a combined increase of roughly 40%, and eligibility rules were loosened so that in most states only current income is checked, allowing people with substantial assets to qualify.
“They were, in percentage terms, about 15% or so each, for a combination on top of each other maybe around a 40% increase.”
- Several additional rule changes raised work disincentives: a $25/week Federal Additional Compensation bonus on UI, expanded UI eligibility for weakly attached workers, income-conditioned mortgage relief for underwater homeowners, and a 65% subsidy of health insurance for laid-off workers that millions took advantage of.
“they paid for the health insurance, 65% of the health insurance, for people who were laid off from their jobs. So, that's a big type of relief.”
- Recent recessions may behave differently because the safety net has been growing for 10-20 years (food stamps over a decade, disability over two decades), so the business cycle is increasingly superimposed on a larger welfare state, and moving from a 70% to 90% marginal tax rate is a much bigger deal than moving from 10% to 30%, potentially producing different cyclical dynamics.
“our business cycle has been increasingly superimposed on a larger welfare state. And I think that may lead to different business cycle dynamics.”
Wage stickiness and employer labor-retention decisions reflect policy-driven benefit generosity, not exogenous market facts.
- Wage stickiness is not an exogenous fact but a response to public policy: when benefits are more generous, workers will not accept lower wages because they were getting a better deal not working, so wages that would otherwise have fallen do not fall, producing the stickiness Keynesians observe.
“Why should people accept low wage jobs when they were getting a better deal when they were not working? So, in normal situations maybe wages would have fallen; they are not going to fall when there's more generosity on the unemployment side of things.”
- Employers are effectively forced to be compassionate by the market relationship: faced with falling demand, they may propose wage cuts to keep a factory running, but if workers can do well without working they will refuse, so the willingness to retain jobs depends on the generosity of benefits.
“The market relationship forces them to be compassionate.”
Unmarried and low-skill workers faced steeper incentive changes and suffered larger employment declines than married workers.
- Disaggregating the population into skill and marital-status groups shows that low-skill and especially unmarried people experienced significantly larger incentive changes (because married people's spouses lift them out of program eligibility), and their labor-market outcomes line up with those incentives—unmarried people's employment per capita fell about twice as much as married people's.
“Unmarried people... had employment per capita or hours worked per capita fall a lot more, about twice as much, than for married people.”
- The larger employment decline for unmarried people is not due to difficulty finding jobs, because employers do not check for wedding rings when hiring—if they did it would be profitable to rent rings for job interviews—rather it reflects that unmarried people get more government help and married people are in more of a hurry to return to work.
“I don't think that employers look at the wedding finger and say: You don't have a ring, I'm not going to hire you. If they did it would be a good business to rent people rings to take on job interviews.”
Safety-net design created perverse incentives including implicit tax rates above 100 percent that penalized work.
- In some instances the rebalancing was done very poorly, creating implicit tax rates above 100% so that people would actually have to pay to go to work rather than earn from it, which gives very little help for the large incentives it destroys and traps people such that the more they work the less they have.
“Instances where people would actually have to pay to go to work. In tax rate terms, their tax rate was above 100%.”
- Before the recession, the safety net was not ungenerous: a typical worker earning $600/week received about $300 in UI, but because payroll taxes and much of income tax are not paid when unemployed, the effective earnings replacement rate was about 70% even before counting food stamps or Medicaid.
“put it all together, before the recession about 70%, maybe a little more, of your earnings would be replaced. Not half.”
- Mulligan's large labor-market results come from the size of the incentive changes, not from assuming large behavioral responsiveness per unit of incentive change; he draws responsiveness from the literature including deliberately small elasticities, and finds median marginal tax rates rose 8 points (from 40 to 48), with zero change for some people and larger changes for others.
“The main driver in my book and the reason why I get such big labor market changes is because the incentives change so much. Not that the behavioral change per unit incentive change... is all that big.”
- The debt overhang is a much larger issue in states like Nevada, where underwater homeowners effectively work for their banker because earnings go toward digging out of a loan rather than to the family, creating particularly bad work incentives that are not uniform geographically.
“Nevada is full of people who... if they work, they are working for their banker. Their family is not going to get any of the proceeds of that.”
- Before the current highly politicized atmosphere, there was a consensus among economists that unemployment compensation increases the unemployment rate, such that five years earlier an economist denying any effect would not have held a credible position, even though magnitudes were debated.
“if 5 years ago an economist were to stand up and say, I don't think unemployment insurance has any effect on the unemployment rate, that wouldn't have been a credible position.”
- There is a long-recognized tradeoff between helping the poor and unemployed (which creates bad incentives and inefficiency) and holding back help (which preserves incentives but causes suffering); the book takes a positive, non-judgmental stance documenting that the United States rebalanced toward more help and less efficiency, without claiming the rebalancing was right or wrong.
“you can help people a lot; that will make for some bad incentives, some inefficiencies, but you are helping people. Or you can hold back the help, and then people will suffer, but you'll have more efficiency and more incentives. It's a tradeoff.”
- The macroeconomic claim that marginal tax rates do not change over the business cycle is simply wrong, since UI extensions varied across recessions (52 weeks, 73 weeks, now 99 weeks); Mulligan constructed a monthly marginal tax rate series for 2007-2011 and argues it should be extended back several decades to test the theory against past recessions.
“There's a big area in macro that... said marginal tax rates do not change over the business cycle... It's just wrong.”
- The Affordable Care Act involves many apparently small changes that add up to big changes in work incentives, on the order of those seen in 2008-2009, so if behavior depends on incentives as Mulligan argues, the labor market may contract significantly over roughly the next four-year period.
“there are big changes in incentives. On the order of the changes that we saw... in 2008, 2009. And so, if I'm right about that... we may see the labor market contracting significantly in the next, say, 4 year time frame.”
- A common reaction from economists is that they had heard of individual program changes but had not realized the sum total of all the expansions could be so large, agreeing that someone needs to account for the cumulative effect of government policy changes on incentives.
“I had heard about this program or that program. But now that you tell me, it makes sense. But I didn't realize that there were so many expansions happening that the sum total could be so big.”
- Some of the failure of the employment-population ratio to recover reflects forces unrelated to the recession, especially the baby boom generation reaching retirement age, so the ratio has bounced back partly to expected levels even though it remains down significantly.
“absent recession, wherever it came from, there were forces pushing that ratio down anyway. Especially the baby boom coming into retirement age.”
- Because employment statistics are democratic—everyone counts once—a policy that hammers only the rich (a small fraction of the population) cannot move employment much, but since spending and GDP are not democratic and the rich contribute disproportionately, the same policy can have very large effects on GDP and overall efficiency.
“employment statistics are, by construction, democratic. Everybody counts once... something that's happening just to rich people--well, rich people are a small group... Their direct effect on the employment statistics can't be that big.”
Keynesian multiplier theory lacks empirical support and misapplies wartime research to transfer payments.
- Economists who claim on television that putting money in the hands of the poor expands demand via the Keynesian multiplier have no research models supporting it, because all modern Keynesian models recognize that money must come from somebody who then spends less and that consumption is not the only activity; the only related research concerns wartime or defense buildups, from which they take an illegitimate leap to paying people not to produce anything.
“They said that. They don't have models to that effect. I'm not aware of a Keynesian model that shows that when you put money into the hands of the poor, that expands demand in any nontrivial way.”
- The Director of the Congressional Budget Office justified applying a fiscal multiplier to unemployment insurance simply by saying 'that's Keynesian economics' and 'everyone else says it,' without any study of a country that expanded its UI and enjoyed a boom, because no such study exists.
“the Director of the Congressional Budget Office (CBO) told me that. I said: Why are you using the multiplier for unemployment insurance? He said: Well, that's Keynesian economics.”
- Contrary to the Keynesian view that safety-net spending injects demand into the economy, transfers merely take money from one group and give it to another while reducing total work and income, so the net effect is less spending and a real reduction in economic activity.
“It does put money in a group of people's hands; it takes it out of another group of people's hands. And the net reduction in the economy is actually less spending. Because, you know, you have less work going on.”
Productivity rose despite employment falling, showing employers treated labor as costlier due to redistribution and reduced wage-cut acceptance.
- Unlike the Great Depression and many recessions where output per worker fell, in this recession output fell less than employment so productivity rose and stayed above trend into 2012, consistent with employers treating employees as more expensive (due to redistribution falling on employers and on workers' reduced willingness to take wage cuts) and producing in ways that use less labor.
“Output per worker went up during this recession and as of the beginning of 2012, output per worker was still higher than it was before the recession, even above kind of a prior trend”
Job availability was abundant; the claim that jobs disappeared mischaracterizes what actually occurred.
- Although construction and manufacturing shrank dramatically in percentage terms and the sand states suffered heavily, those industries were small to begin with so the fraction of unemployed coming from them is modest, and the regional differences are far smaller than the differences between married and unmarried or elderly and non-elderly people, so geography fits the safety-net story rather than refuting it.
“the percentage shrinkage of the construction industry was huge... But the fraction of unemployed people who came from those industries is much smaller. Because those industries were not that large to begin with.”
- JOLTS data show that during 2008-2009 about 240 million job-starts occurred (some people starting jobs more than once), so the claim that there were 'no jobs to be had' is a terrible description of what happened, even though it would have been better had more jobs been created.
“There were 240 million people who started new jobs during that period. So, there were lots of people starting new jobs.”
- Critics who compare U.S. GDP to European GDP are cherry-picking, because Mulligan's theory is about the labor market and U.S. employment contracted far more in 2008-2009 than in any major country he could find, which is consistent with his claim that the U.S. situation should differ to the degree its safety-net changes differed.
“employment went down a lot more in the United States in 2008 and 2009 than it did--I was not sure I was able to find any country where it went down--any major country--where it went down that much.”