Unemployment Insurance, 1: Raj Chetty’s research presented at EPI

Allright, this goes out to the financial engineers in the audience.  How many of you caused a ton of damages in the financial economy because you never really got a solid handle on modeling liquidity in your theories and work?  Raise your hands.  It’s ok, this is a safe space.  (Slight hand raise here.)

Don’t feel too bad, because certain academics, economists and government ideologues are going to cause a ton of damages in the real economy by arguing against unemployment insurance without also having a decent theory of liquidity.   People are worried about the bad incentives of unemployment insurance in the abstract, but they are confusing bad work incentives with the effects of “cash-on-hand” (liquidity). So let’s look at some evidence.

Here’s two presentations from Raj Chetty at a recent Economic Policy Institute event on the long-term unemployed. He presented his research about where unemployment insurance stands not in a liquidity trap, not in a major recession, not when there are 5+ job seekers per job opening, but in normal times and through the efficiency lens of an economist:

Here’s his presentation as a pdf. If you are a liberal who wants to defend the social safety net, you should learn how to wield this argument like a katana blade. There’s three reasons why unemployment insurance is valuable: it’s valuable to people as when they are unemployed is when the marginal value of a dollar is very high. Many people have trouble borrowing when they are unemployed, and evidence (we will discuss tomorrow) shows that the people have to make very painful consumption cuts. As Raj notes: “Consumption-smoothing benefit of providing UI large because unemployed families are cash-constrained, median unemployed person has less than $250 in net savings prior to job loss and cannot borrow, and moreover, most families have many commitments that they cannot adjust.” People are liquidity constrained.

Now when you have higher unemployment benefits, you’ll see that people take longer to find a job. Conservatives point to this as a “work disincentive effect” – you are paying people to not work, and sure enough they aren’t working. Since they could be doing productive work but they are choosing not to, this hurts the larger economy.

However this effect seems to be concentrated among people with very low liquid wealth. See how people in states with high unemployment benefits take a longer time to find a job:

But if you look at the top of that graph, you’ll see that it’s for people with negative liquidity (the household holds short term debt that needs to be paid). What if we look at people with a fair amount of cash on hand:

If you check out the presentation there’s even less difference with people with more savings than that. From an economic viewpoint what we are concerned about is a work disincentive effect, that the insurance creates a moral hazard where people would not take a job that they otherwise would to get the insurance. But we don’t see this: there’s no reason that people with more savings should feel a work disincentive effect less than those with poor savings.

In fact, one of my favorite instrument experiments does just this. From Cash-on-Hand and Competing Models of Intertemporal Behavior: New Evidence from the Labor Market
, in Austria you get a lump sum of cash if you worked more than 36 month at your previous job. There should be no work disincentive effect from this; the cash doesn’t go away whether or not you get another job, as it does with unemployment insurance. And sure enough, we see the same issue:

On the left, there is no lump sum unemployment payment. Just giving people a bag of cash makes them take an efficiently longer amount of time to find a job, even though there isn’t a work-disincentive tying the payment to being unemployed. From the presentation of this paper: “Conclusion: 2/3 of the effects of UI effects on increased durations is a beneficial “liquidity” effect rather than a harmful work disincentive effect.” This is people searching for a job they fit into better, this is people making their basic payments and obligations, hedging against future risks and future financial ruin, this is people being able to efficiently make the choices for how to fix back into the economy.

So when we look at this, even in normal times, the benefits outweigh the work disincentive costs. Now the long-term unemployed, who have depleted their savings, who have a hard time finding a job period in the current economy, would gain the most from unemployment insurance. And that’s even before the macroeconomic effects. And it’s even before the interesting problem of subsidizing people to look for a job (which unemployment insurance requires) rather than join the long-term unemployed, where some current research shows that human capital depreciates at an alarming rate, which we will talk about shortly.

So really, every slice you look at it, extending unemployment insurance is a smart move. Good on policy, good for the macroeconomy, and good for people. It’s a shame Congress doesn’t see it this way.

This entry was posted in Uncategorized. Bookmark the permalink.

23 Responses to Unemployment Insurance, 1: Raj Chetty’s research presented at EPI

  1. People can get jobs to overcome liquidity constraints. In his analysis, Chetty assumes away the liquidity value of having a job (this is his perpetual job assumption where unemployed agents get one job and keep it forever). This produces his result. If agents were allowed to get temp jobs, their liquidity constraint would be overcome and the liquidity value of UI would be nil.

    Also, you say “there’s no reason that people with more savings should feel a work disincentive effect less than those with poor savings”. Isn’t this obviously false? In utility terms, Bill Gates would find a marginal change in UI benefits to be negligible.

    • Greg L says:

      If savings are a disincentive to work, then why didn’t the high savings people self-terminate a long time ago?

      Answer: Liquidity is not a disincentive to work. It is a disincentive to working at the wrong job.

      What? Don’t you believe that economic agents make rational decisions?

  2. Mike says:

    Right, but Bill Gates wouldn’t care about a lump-sum UI benefit versus a UI benefit linked to being unemployed. But, I bet Bill Gates takes all the subsidies he can get, even if he doesn’t feel them.

    Rephrase that with a median person. He or she gets subsidized for being unemployed. In the bellman world, what does it matter what he or she’s savings are? Work is a strict disutility, and to get paid not to do is going to impact your decision regardless of savings. Unless we aren’t in that world, or we are in that world and liquidity is the driver, then savings matter. Savings may in theory do something like increase the reservation wage, but it does that regardless of UI, but from the eyeball of the graphs in the pdf it doesn’t appear to increase duration.

    I think your first point is Raj’s point too – that engineers aren’t forced to get part-time jobs at McDonalds in order to keep the lights on. (that’s what you mean, right?) He argues that this is good for the economy, better matching and better consumption smoothing. I agree. Why would you not?

  3. Yes, the effect will exist even for rich people, but that doesn’t mean its size will be the same no matter wealth.

    Chetty assumes away the possibility of getting a hold over job. I’m not sure what’s best for the economy, but getting the “McDonald’s job” and having on-the-job search would undo the liquidity constraint Raj finds to be so important. My point is that his model shuts down this obviously important channel.

  4. On the last point, is there evidence that job search on the job is less efficient that job search while unemployed? Card’s evidence suggests not and my guess is that, if anything, workers find better matched jobs when on the job. This would be due to better access to information.

  5. sraffa says:

    Will- how can getting a job at McDonald’s possibly help an engineer get a better job as an engineer? This makes no sense to me.

    The information access might be better with on-the-job search with an engineering job, but with a minimum-wage job? It just means less time for job search.

    These kind of “bridge” jobs don’t happen much for several reasons. One is that firms know that these workers will leave soon and thus they will just have to retrain another worker. Another is that this kind of job looks terrible on a resume. Would you take a job at McDonald’s if you were unemployed without UI benefits?

    Besides, overall job growth is terrible, not just high skilled jobs. There’s just no hiring going on. Even if all the unemployed workers with college degrees took minimum wage jobs, would there be enough? Of course not.

  6. Mike says:

    “Card’s evidence suggests not and my guess is that, if anything, workers find better matched jobs when on the job. This would be due to better access to information.”

    Will, if so, why the jump in Austria? Workers would rationally know they find better matched jobs while working and thus would take the first job they could get ASAP.

    And what is “information” here? Like access the internet over lunch? Or hearing about opportunities from customers and co-workers (socializing?)? Or movement within the firm?

    Also sraffa and Will, I miss you guys!

  7. sraffa says:

    Miss you too, but somebody’s got to teach the youth of American about the Great Vacation, don’t they?

  8. The engineer/McDonald’s job is a silly example; we’re talking about marginal changes in reservation wages. And people get bridge jobs all the time, why do you think not? At least in the PSID at yearly granularity, job changes are clustered.

  9. Mike, I don’t understand why you’d think there’d be no jump. The value of on-the-job search is already baked into both sides of the discontinuity. That paper is about differentiating the wealth and the substitution effects of UI.

  10. Pingback: Links 7/9/10 « naked capitalism

  11. sraffa says:

    Sure people get bridge jobs- this makes them employed. There are very few vacancies of any kind right now- that’s why extending unemployment insurance is essential.

    Right now, there are 5 unemployed for every vacancy as shown in the jolts data. Even with a totally efficient matching technology where every match is met by an appropriate worker, this would only reduce unemployment by 20%. If any new workers enter the job market, this makes it even lower.

    Making this unemployment crisis about a lack of search just doesn’t describe the situation. This is all about a lack of vacancies, not in a lack of people looking for vacancies.

  12. matthew says:

    So what if higher or lower UI has no effect on time off the job? The current discussion regards increasing the length of benefits (and therefore total sum), not monthly amount. The example from Austria proves that the greater the total amount of benefits are, the longer people will stay unemployed.

    What about the incentive UI gives people to not save in the first place? UI is just a form of social security to compensate for worker’s lack of savings (There are other factors that contribute to the low savings rate).

    *I did not watch the presentation

  13. Pingback: Matthew Yglesias » Improving UI With Lump Sum Payments

  14. Pingback: Josh Barro on Raj Chetty’s UI Research « Rortybomb

  15. Pingback: SF Fed on Labor Force Participation and Older Workers « Rortybomb

  16. Pingback: Older Workers: Between a Rock and Slashed Benefits » New Deal 2.0

  17. Marilyn says:

    Has Raj written a book, published papers? If so, how does one purchase them?

  18. Pingback: The Off-Whites » Joblessness v. Uselessness

  19. Pingback: » Joblessness v. Uselessness C-Town Mag

  20. Pingback: Read the Whole Thing « Anthony Collebrusco

  21. Pingback: Examining the Limitations of a Neoliberal Safety Net: Romney’s Unemployment Insurance Savings Accounts | Rortybomb

  22. Pingback: One Less Bell to Answer: Further Thoughts on Neoliberalism By Way of Mike Konczal (and Burt Bachrach) « Corey Robin

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s