The Federal Reserve’s FOMC just released another statement punting on both parts of their dual mandate. Many, including Ryan Avent here, are disappointed. I’m surprised that there are three (three!) dissenters. How often has this happened? Richard W. Fisher, Narayana Kocherlakota, and Charles I. Plosser all dissented, because they “would have preferred to continue to describe economic conditions as likely to warrant exceptionally low levels for the federal funds rate for an extended period.”
What’s their motivation? We looked at Richard Fisher before. In April 2011 his “gut tells [him] that [QE2] will result in some unpleasant general price inflation” – something that was absolutely wrong. From that link he was worried, presumably from his gut, about runaway inflation last fall as well as right before Bear Sterns collapsed. His gut has some major problems. I’m not a medical doctor but this could be a medical condition: when I put ‘my gut is always wrong’ into Webmd’s symptoms search there are 26 results.
But what is going on with Kocherlakota? Why is he dissenting in favor of tightening sooner? Last time we saw him, he was talking about job openings taking off, numbers that turned out to be exaggerated by modeling assumptions at the BLS. So that’s off the table.
What’s his deal now? Let’s go to his big discussion paper, Labor Markets and Monetary Policy, recently released through the Federal Reserve Bank of Minneapolis. In it, he uses a Diamond, Mortensen and Pissarides (DMP) model of unemployment, a model where unemployment is people searching for jobs, to understand why people aren’t creating jobs and we’ve hit a structural limit.
Kocherlakota uses the DMP model to explain why there is a lot of unemployment as a result of this previous recession. There’s been a lot of interest in the DMP model recently, both because the creators of it won a Nobel Prize last year and because of a series of research following Robert Shimer’s finding – in his seminal 2005 paper The Cyclical Behavior of Equilibrium Unemployment and Vacancies – that the DMP model is awful at explaining why there is a lot of unemployment as a result of a recession.
Kocherlakota uses this equation in his big paper to explain his reasoning for why there are so few jobs:
In this equation, job openings are a function of how many unemployed there are per job opening and the difference between productivity of the economy and the utility of not working. Kocherlakota thinks this model can explain why firms aren’t hiring after the recession we’ve been through.
First thing you should notice is that Kocherlakota of the Minneapolis Federal Reserve is basing his opinion on unemployment on an equation in which the Federal Reserve has no role. Lots of people think the idea that the Fed can’t set a negative interest rate – that there is a “zero lower bound” – has something to do with our current problems. Agree or don’t, there’s no possible way it impacts this model. Product markets not clearing doesn’t factor into this model, which is all about how lazy workers are. Some people, most notably Stanford economist Bob Hall, have tried to put the DMP model into a world of zero lower bounds and product markets not clearing and found results closer to our world – this isn’t mentioned in the report.
Second, many economists have tried to use this model to explain how unemployment spikes during a recession. The main thing that would drive changes in unemployment in this model are changes to productivity. As mentioned above, it was a major breakthrough when Shimer (2005) showed that there’s no way wild swings in productivity can cause the major swings in unemployment we’ve seen. This goes double for the recent recession, where productivity has increased. As Hall pointed out:
First, Shimer’s (2005) influential paper showed than it would take a gigantic drop in
productivity to cause the rise in unemployment in a typical recession, based on realistic values of the parameters of the DMP model. Second, productivity has increased in recent recessions…Productivity grew almost at normal rates during the huge contraction that started in 2008. To generate an increase in unemployment driven by productivity, an actual decline in productivity would be needed.
Shimer’s paper has stimulated an interesting literature surveyed in Rogerson and Shimer (2010) that alters the canonical DMP model to boost the response of unemployment to productivity. But with rising productivity in a recession, the stronger response is an embarrassment, making it even harder to square the behavior of the U.S. economy with the DMP model.
All the king’s graduate students and all the king’s junior faculty couldn’t put
humpty-dumpty the DMP model of explaining spikes in unemployment back together again. (p-z) doesn’t have the cyclical component necessary to generate our 9%+ unemployment, and it would have to overcome the massive increase in the u/v unemployment-to-vacancy ratio. Think about it – job openings drop from 3.2% pre-recession to 2.3% now; how could this model explain that with v/u jumping (and separations, not included in the equation, plummeting) and p increasing?
Here’s Kocherlakota, who explains a doubling of the unemployed with a (p-z) shift as follows (my bold):
Given the enormous rise in the benefits of creating job openings, why weren’t firms creating more of them? A common answer to this question is that firms face “insufficient aggregate demand.”….But the DMP model suggests two other possible reasons that firms are not creating job openings…As just discussed, u/v rose 165 percent from December 2007 to December 2010. What happened to the other two terms in the equation? There are good reasons to believe that expected after-tax productivity p fell. Over the past three years, the U.S. economy has experienced large increases in the federal budget deficits, contributing substantially to the overall federal debt…What about the utility that a person derives from not working? In response to the recession, the federal government extended the duration of unemployment insurance benefits…Now suppose that, for the reasons just mentioned, p fell by 10 percent in the past three years and z increased by 0.05 during this period. These are large changes, but they are not implausible…
While I won’t go through the details here, the DMP model provides a way to compute the natural rate of unemployment u*…If after-tax productivity p and utility from not working z have not changed since December 2007, then u* may be as low as 5.8 percent. However, if (p−z) has fallen by 0.15, then the implied u* is 8.7 percent.
There it is. Job creators hate future taxes, and unemployment insurance has left our workforce weak, so don’t expect unemployment to come down anytime soon.
For all the fancy math, this logic is very similar to Fisher. “Now suppose that, for the reasons just mentioned, p fell by 10 percent in the past three years and z increased by 0.05 during this period” is about as close to a “gut” feeling and “gut” reasoning as you can get. This appears to be how one of the most powerful people in the world for determining the future of the United States’ economy is determining his dissent from Bernanke’s position.
Where to begin? If unemployment insurance extensions are causing a rampant increase in the time people are unemployed, it should show up in aggregate data. There are a lot of ways to test this – for instance, you could compare the duration of unemployment for those who get unemployment insurance to quits and new entrants – or people that don’t get unemployment insurance. If UI was causing unemployment, you’d see very different results. In fact, Mary Daly, Bart Hobijn and Rob Valletta of the Federal Reserve Bank of San Francisco did this in January. What did they find?
They find that “the results of this analysis suggests that the availability of extended
unemployment benefits has increased the overall unemployment rate by about 0.4 to 0.8
percentage points.” But even this meager increase in “structural” unemployment is, as Scott Sumner noted, subject to the Lucas Critique: “the maximum length of unemployment insurance is itself an endogenous variable. If stimulus were to sharply boost aggregate demand it is quite likely that Congress would return the UI limit to 26 weeks, as it has during previous recoveries.”
And why don’t we assume that “z” has gone up in this recession? It is harder to find a job than in normal times per week of unemployment duration, outstanding debt loads hang larger over household net worth – having a job seems more important than ever.
If you believe that the natural rate of unemployment is near 9% because President Obama has terrified the job creators and the unemployed aren’t starving enough, I am unlikely to convince you otherwise using various forms of econometrics. But I’m also interested in the political dimensions of this.
I imagine that every day Kocherlakota interviews people – from banking, from the top 1%, from the corporate offices of our largest firms – who will kindly explain to him that the problem in the economy is that they just don’t get their demands answered quickly enough. If only they paid even less in taxes, if only regulations were weakened further, if only every pet demand they ever wanted was granted, then they would allow the economy to take off.
How often does he hear the opposite? How often do the unemployed disrupt his speeches, chanting about how they aren’t on a magical vacation but instead desperate to find a job? How often do students show up in huge numbers in his office explaining they they are terrified of entering this terrible job market, a job market likely to scar their careers for decades, instead of apathetic losers who’d rather just play on facebook all day enjoying their “z”? Maybe it is time that changed.
This is fantastic. That equation shows how you can get both of the wacky claims about the cause of the recession that I’ve seen from “Real Business Cycle” style economists: the “technological shock” where productivity mysteriously disappears all of a sudden, and the “great vacation” argument where workers suddenly decide en masse that they don’t really want jobs. Other, more sensible explanations are conveniently written out of the model.
Am I dumb? This is supposed to be an equation that explains whether a business will decide to create an opening, right? By assessing the benefit to the firm. And says benefit depends in part on whether a lot of people are available for given openings, which is an environmental variable that affects the compensation that would have to be paid. And says benefit depends in part on how much value this position will add to the firm, which is something the firm controls and can assess. And says benefit depends in part on the utility of someone refusing to take a job.
What is this last term doing in this equation? It’s unknowable. And it’s beyond the control of any firm. And it’s not an element of the environment in which a firm does business. It’s something individuals decide.
It seems to be a stand-in for the probability of hiring someone at a given salary, or something like that. But it strikes me as being totally incommensurate with any of the other terms in this equation because it expresses something no firm can possibly know, something that no individual can know except the one who makes the decision to take the job. I’m not even sure it can be abstracted, but would depend on a specific person and a specific job.
The people who use this equation are entitled to argue that unemployment compensation decreases the probability that someone will prefer unemployment to taking a specific job (or any job, I guess), but they can’t read minds. It would have to be an empirical measure to have any validity, wouldn’t it?
Just a naive question from a non-economist.
Altoid – Oversimplify to understand. Suppose you have a job available, and there are four possible people who could fill it perfectly. (Lifetime loyalty, perfect skillset–you can wish for a pony, too.)
Now two of those are employed currently (u/v relatively low). One was recently fired from a similar job because he’s a cocaine addict and just entered recovery (p too low).
The fourth variable (z) is unknowable. She may be about to inherit a small fortune. She may just have spent her last UI check, but her partner says she’ll keep them both going for the next few months while she looks. Or she may be willing to take any salary, but if you offer her less than what she needs, she’ll be on the job market again in six months to a year, and your perfect-match employee will be imperfect.
Zed isn’t directly knowable; on the worker side, it’s equivalent to the “reservation wage” concept.
But you know how much you are willing to pay a person in that position. And if your willingness to pay is too low, you’ll either (1) hire her and lose her or (2) not even get an interview, depending on her circumstances.
After which point, p becomes marginally less high (not to mention that there will be a future “menu cost” in relisting the job–checking c.v.s, interviewing, contracts, offers, negotiations, etc.), until you reach the point where what you’re willing to pay matches with how perfect a fit for the job someone is.
If you (also) accept that “the market” is reasonable, you’ll know roughly the tradeoff you’re making, and decide from there that it is(n’t) worth posting an advert.
It’s a simple model, and a fairly reasonable proxy in a well-operating, near-full-employment economy. But when you have to speak in absurdities (p down 10% in the current, productivity-increasing environment? z up 5% when UI is generally at best 50% of previous salary, capped at around [iirc] 70% of the median US wage?) to make it work, it’s best to find another model.
Pingback: Riots, Recessions and Fear « Left Outside
Pingback: FT Alphaville » Further further reading
Two of the 3 dissenters are from “fly over country”. Coincidence??? And I think our boy Thomas Hoenig’s vote dropped off on this one??? Frankly it’s just like the college football and college basketball team ranking polls, 85% of the voting weight goes to Eastern establishment basturds who think human life only exists within a 5 mile radius of NYC or 2 mile radius of Greenwich Connecticut or whatever sh*thole suburb they inhabit, hiding from minorities and getting nervous shivers when they found the one upper class black in the back of their kid’s math class.
By the way, in case you couldn’t tell, I’m on the dissenters side. Not because I don’t like Bernanke either.
I actually think Bernanke has done relatively well (under the conditions he has been thrown into of a horrid economy). People expect monetary policy to do the job of fiscal policy. This economy needs targeted fiscal policy (examples: our nation’s crumbling bridges, the establishment of a home based, domestic rare earth minerals industry). People are expecting that putting more money into the TBTF bankers’ reserves (quantitative easing) to do what Republican congressman have been blocking the last 2 years—-targeted fiscal spending.
I think zero% interest rates is an unnatural condition. Kind of like a river with no current. Low rates are desired here, but I think zero rate is going to f*ck things up long-term, and doesn’t solve the core problem. Which is what Republicans want.
I think DMP model can answer a question, how many employees you will hire, if there is demand for your products. Given sticky prices none increase in productivity will help.
So I agree with Ken – in full employment economy you can use that model.
I look at that equation and I think ‘Those economist guys are just plain nuts, someone got Nobel for that!?’.
So in the equation I guess you have got ‘Can I produce more or better stuff to sell if I hire this guy?’ that’s your productivity, but where is ‘Can I actually sell the more stuff, or charge higher prices for the better stuff, hiring this guy will produce?’.
Maybe you say that doesn’t matter, this new guy is twice as good at making stuff than than my current guys and therefore I still should hire him. Fine, of course that means I’ve got layoff two of my existing guys, if I’ve even got two existing guys instead of one…
I think it’s funny that some economists like Kocherlakota are so dead-set that only supply-side factors can determine recessions, and have such a disbelief in changes in aggregate demand. I mean, doesn’t their reasoning just fall apart as soon as you consider a global economy?
I mean, let’s say that the developed countries are in recession, and the global price of copper crashes. If copper miners in Chile start working fewer hours, is it because they are less productive? It seems much more sensible to talk about it in the Econ 101 way– there was a shift in the aggregate demand curve, and a movement along the aggregate supply curve.
Pingback: Employment Math, or How to Make Prejudice Look Scientific « thecurrentmoment
You know the best part about this post — eventually, it will be forgotten.
Pingback: Pre-Marketing: S&P blames default skeptics – gpkendall.com
You forgot to mention Hall is also affiliated with the Hoover Institute at Stanford.
I have been struggling with Hall’s article in the April, 2011 issue of the American Economic Review, and I haven’t quite understood why. He uses the DMP model for the labor market, but the overall model itself really looks Keynesian, including having households preferring to hold a lot of cash under certain circumstances, that lower consumption without increasing investment. (Liquidity Trap anyone). Yet he claims not to have a model of aggregate demand, he says the “closest he comes to it” is a figure that has interest rates on one axis and employment on the other. This is what I struggle with since is model seems to have demand as part of it. It even has a financial sector, yet it is within a conventional sort of supply side modeling framework. In any case there is a model out there, by someone from the Hoover Institute, using a sort of supply side model, including the same DMP model, that argues a really good thing for the Fed to do would be to increase its inflation target. Could Kocherlakota have missed it? Or is he afraid that reading it might cause his head to explode?
I forgot to mention there are 30 some odd equations in Hall’s paper and 67 in the appendix, so for those of you out there Dr. K is using a single equation to determine employment in a supply and demand word. Another problem he has. (And also why I have been struggling with Hall).
43 actually, but most of those are subequations to get to the five “overviews.”
Which still doesn’t mean that the equation Dr. K. is using isn’t inaccurate and inappropriate, not to mention that his assumptions are, uh, unsubstantiated.
But he has fans in the Midwest who are happy with 10% and rising unemployment, so don’t tell them that my hedge fund buddies specifically and the NYC-area financial services industry in general are doing great while they are shat upon, which was when Dr K. was happy.
“B=u/v × (p-z) × constant”
Profit depends on prices, not just productivity. Silly model
the iPod effect (img)
Pingback: Discussing Rick Perry and the Federal Reserve on Last Word | Rortybomb