Minneapolis FRB President Narayana Kocherlakota says,
I mentioned that the relationship between unemployment and job openings was stable from December 2000 through June 2008. Were that stable relationship still in place today, and given the current job opening rate of 2.2 percent, we would have an unemployment rate of closer to 6.5 percent, not 9.5 percent. Most of the existing unemployment represents mismatch that is not readily amenable to monetary policy.
Stephen Williamson has further comments. I will say more below the fold.The first thing to point out is that Kocherlakota might be making the relationship between vacancies and unemployment sound more precise and well established than it really is. How can one rule out non-linearities that would make the current combination of vacancies and unemployment seem quite normal? What sort of confidence interval should one draw around the predicted unemployment rate of 6.5 percent?
What other macroeconomic relationships are out of whack? Okun’s Law, relating the unemployment rate to GDP, is not working correctly. We are getting stronger output growth than employment growth, which suggests that the job openings are not for end-stage production workers.
There are some unfilled vacancies (fewer than during a boom, but more than one would expect in a deep recession), but these are not adversely affecting output. One way to read the numbers is that we are in a Garett Jones, or Labor is Capital economy.
However, if you look at where the openings are, there is a hint that it is production workers that are in demand. In the JOLTS data, the total private job opening rate (job openings divided by employment plus openings) was 2.3 percent in June, compared with 3.6 at the pre-recession peak and 1.9 at the trough. In manufacturing, the job opening rate was 1.9 percent in June, compared with 2.7 at the pre-recession peak and 0.9 at the trough. Relatively speaking, manufacturing has bounced further off its trough than the rest of the private sector. This view is supported by the manufacturing workweek of 41.1 hours for production and nonsupervisory workers supports, just 0.3 hours less than the pre-recession peak.
What about the Phillips Curve? I am not an expert on Phillips Curves. However, starting from a past inflation rate of about 2 percent, wouldn’t a standard Phillips Curve be predicting that an unemployment rate of 9.5 percent would be sending inflation well below zero? If that is what such an equation would predict, and we are not seeing it, that might be another sign that there is a labor matching problem.
The vacancy rate and the inflation rate may tell us that aggregate demand is not as weak as the unemployment rate might suggest, and instead we have more of a matching problem (yes, the Recalculation Story). Fine, but I still think that the state of aggregate demand is far from satisfactory. Even though I share some of Kocherlakota’s pessimism about the ability of monetary policy to bring down the unemployment rate, I think that the Fed should be making more of an effort to fight the recession. Even if you doubt that it would work, it seems a shame not to try.
READER COMMENTS
Ted
Aug 22 2010 at 7:24pm
The first thing to point out is that Kocherlakota might be making the relationship between vacancies and unemployment sound more precise and well established than it really is. How can one rule out non-linearities that would make the current combination of vacancies and unemployment seem quite normal? What sort of confidence interval should one draw around the predicted unemployment rate of 6.5 percent?
There is another problem also. Job opening figures are usually estimated more precisely than unemployment figures. We know that unemployment numbers are actually vastly underestimated – and the inaccuracy increases as business cycle conditions worsen. In fact, recent estimates suggests the true unemployment rate is closer to 15% right now – which would make a lot of this Beveridge Curve discussion disappear. (see: http://ideas.repec.org/p/iza/izadps/dp5057.html – you have to dig into the paper to see the 15% number).
What other macroeconomic relationships are out of whack? Okun’s Law, relating the unemployment rate to GDP, is not working correctly. We are getting stronger output growth than employment growth, which suggests that the job openings are not for end-stage production workers.
Okun’s Law is not actually really broken that much. Gross domestic product and gross domestic income should, in principle, equal one another. However, recently, they have diverged. And GDI has been matching Okun’s Law fairly nicely over the crisis. And recent research by Jeremy Nalewaik at the Fed suggests the GDI may be a more accurate measure of output than GDP, especially when considering real-time data. So maybe Okun’s Law isn’t broken. Maybe GDP measures go wild in recessions and GDI becomes a better metric – in which case, Okun’s Law may not be broken. (http://www.federalreserve.gov/pubs/feds/2007/200734/200734abs.html)
What about the Phillips Curve? I am not an expert on Phillips Curves. However, starting from a past inflation rate of about 2 percent, wouldn’t a standard Phillips Curve be predicting that an unemployment rate of 9.5 percent would be sending inflation well below zero? If that is what such an equation would predict, and we are not seeing it, that might be another sign that there is a labor matching problem.
I think this is mostly downward nominal wage rigidity. History suggests you need a really severe, persistent crisis to causes almost any wage cuts – think Great Depression, but even then though, we still saw substantial downward nominal rigidity. History, and evidence (a la Akerlof et al 1996) suggests the downward nominal rigidities tend to mess with the Phillips Curve. Also, there is some evidence that the long-term unemployed do not impose as much pressure on inflation. Perhaps this is because they exit the labor market or their human capital depreciates so they aren’t useful or something.
I think there is some degree of structural problem. But,most of it is still just expectations of sluggish demand. I think the Federal Reserve can fix most of the latter part, and by increased growth expectations and AD the Fed facilitates the re-allocation process of labor much more smoothly. If growth expectations increase, more small businesses open in response so recalculation happens much faster. The stories complement one another – but low aggregate demand is still the problem.
saltmanSPIFF
Aug 22 2010 at 10:28pm
I’m inclined to think AS/AD models can be used to analyze short-run economic activites for a given sustainable pattern of production and trade (if I may borrow your line). The old pattern has collapsed under the weight of its own fragility and a new one is just starting to emerge from the rubble.
You assert the current AD situation is “far from satisfactory.” Well, satisfactory compared to what? Previous AS/AD situations were properties of other patterns which are now outdated. Comparing AD now to AD then, it seems to me, is like comparing apples and oranges.
Comments are closed.