Wednesday 25 July 2012

Skills Shortage? US Evidence.

In the first post of this series, I briefly described Peter Capelli's thesis: knowing that the unemployed are looking for jobs, employers are driving hard bargains, the kind of bargain where more skills/experience get less (or at worst, no more) money in exchange. To the lowering effect on wages, this adds a slow job recovery.

Next, I showed that closely related ideas, while all but ignored in contemporary economic policy discourse, have a long and illustrious history, starting with Adam Smith, father of modern economics.

But those are all hypotheses. Where is the evidence supporting them?

Here I'll start considering this, in relation to the US.


Precisely this month, the "Chicago Fed Letter", published by the Federal Reserve Bank of Chicago, contained the aptly titled "Is There a Skills Mismatch in the Labor Market?" by Jason Faberman and Bhashkar Mazumder (h/t Next New Deal).

Apart from its timing, the paper has several attractive features. First, its authors have a mainstream economics background. Second, the paper is mercifully short and clear, while providing enough guidance on the main theories on skills mismatch/search theory

This is how the paper develops.

Figure 1, reproduced from the paper, shows the US Beveridge curve (for additional details, see Further Reading), cited as evidence for "mismatching skills":

I'll have much more to say about this in subsequent posts. For now, a brief explanation: if the US economy never changed (i.e. say a "Groundhog Day" equilibrium), neither unemployment nor vacancy rates (job openings divided by work force) would change. Figure 1 would show a single data point.

Alas, economies change. If things change within certain limits (say, cyclical demand variations), the likely outcome is for the vacancy and unemployment rates to follow historical patterns: to move along a curve. In Figure 1 this could mean that points occur only along the curve Jan-01 to Jun-09. In this case, one says that any resulting unemployment is involuntary and/or cyclical (due to the cyclical changes in the economy); demand-management policies are in order.

What if changes exceed some threshold? Say, structural ("large") changes in technology or in the willingness to work? Points appear outside the original curve. If points appear to the north/east of it, the conclusion is that unemployment is structural.

In the Figure, one sees that the data for 2012 do fall to the east of the curve. This is in essence the evidence adduced to say that there is a skills mismatch.


Although from the Figure itself one cannot conclude what causes this, one could think like this: for the same vacancy rate, there are now two unemployment rates, a more recent and higher one (more to the east); and the original, lower one (more to the west).

If the jobs are there for the taking, but unemployment is higher, what conclusion comes more naturally than the additional unemployed don't want those jobs? Unemployment is voluntary. As Michael Pascoe and countless others say, these are clearly "job snobs".

In this case, if there is a problem at all, demand-management policies will not address it. If one considers unemployment a problem (say, for moral or public order reasons), then maybe social welfare is the cause. One could just throw in labour market regulations, for good measure.

Alternatively, a more charitable view (represented by the "skills mismatch" theses) is that productive technology suddenly changed a lot. Well, then, maybe the newly unemployed are not skilled enough to get the new jobs. At least, unemployment is not voluntary.

Training could do the trick, in this case. But there is the inconvenient that it costs money and it takes time to train staff. Therefore, immigration policy should be used to fill this need cheaply and quickly.

This, in a nutshell, is what most neoclassical economists have to say about this. On this reasoning is that social welfare cuts, labour market deregulation, and immigration policy are predicated.

But "most" neoclassical economists are not all of them. Some have criticised this on the grounds that there is no single labour market: the labour market is segmented, either on a geographical basis (say, market in one State), by industry (market for construction workers) or by educative/qualification level. The conclusions are basically the same one reaches at national level, except that now one has "smaller" nations: islands. Needless to say, this is cutting-edge research, and it hasn't fully trickled down to policy-makers or pundits.

As I said, I'll have more to say about this. Here, my interest is on what two neoclassical economists say.


Faberman and Mazumder (FM) want to find out whether this simple diagnostic makes sense of American statistics. If it did, some three years since the end of the Great Recession, then one should expect employment for all educative levels to have recovered substantially and be near 2007 benchmarks and possibly considerably higher in a number of areas.

Think of it this way: Say's (or Walras') Law means that there cannot be general gluts. Excess supply in one industry means excess demand in another. But if there is excess demand for a good, then there is excess demand for the workers producing it.

As a side note, this, it seems to me personally, would be particularly suited to address the "island" view of the markets: we are considering "islands of education".

Figure 4 (panel A) is part of what FM found:

All groups are below their 2007 employment level. In fact, except for Group 5 (highest education), well below it.

In fairness, within Group 5, "employment among engineers was actually 2% higher in 2011 than in 2007, so it is plausible that engineers are currently scarce". However, the other side of the coin is that other highly educated workers' employment levels are lower than in 2007.

On the evidence provided, FM conclude, perhaps a tad cautiously: "Using this approach, we find limited evidence of skills mismatch".


Figure 4 (panel B), I believe, is even more telling. It shows data obtained from the Conference Board Help Wanted OnLine® (HWOL) series (similar to the Australian DEEWR and ANZ online job ads indexes). If employers are desperate to find workers, they must be advertising the job openings!

I don't think this chart needs much comment: employers don't seem that desperate. For workers with the highest skills level, they were more desperate in 2007. And for the lowest skills, they are about as desperate as they were in 2007. And, this is a chart of annual percentage growth, with some 3 years of negative growth in between...


This is how Faberman and Mazumder conclude:
"Since the end of the Great Recession, evidence of mismatch in the labor market has been mixed. Studies suggest that the degree of mismatch has abated since early in the recession, and there is evidence that many employers appear hesitant to fully commit to hiring. Our analysis of the supply and demand of workers by skill level points to some limited evidence of skills mismatch."
While the conclusion is not mistaken, I find it an understatement.

And, judging by Mike Konczal's post, I bet am not the only one:
"Their [FM's] answer: 'we find limited evidence of skills mismatch.' In other words, not really".

The first thing to consider is that one thing is the Beveridge curve as empirical phenomenon, another is the interpretation one chooses to give it. Faberman and Mazumder show that this interpretation is not consistent with other empirical facts.

But you be the judge.


"Wait a minute!" Attentive readers (and my regulars are all attentive!) may have noticed something: from Figure 1 we jumped to Figure 4...

"Aha! Gotcha! What about Figures 2 and 3? Is this a shameless sleight of hand, Magpie?"

Nope! How dare you? Well... uh... kinda.

You see, if you thought Figure 4 is dynamite, you are wrong. Figures 2 and 3 is where the really good stuff is. And as this post is already long, I'll leave you guys anxiously waiting for the next post, where I'll go in detail on them. After all, I said there was more, didn't I?

28-07-2012. From the 2011 Annual Report of the Federal Reserve Bank of Atlanta:
"Skill shortages not a primary cause of high unemployment
"Some analysts have argued that, in addition to the slow pace of growth in GDP, certain labor market impediments may also have contributed to the relatively slow employment recovery. Anecdotally, mismatches between the skills that job seekers possess and the skills that employers need were frequently cited as an example of such impediments."
(See here)
Further Reading:
For a didactical exposition on the Beveridge Curve, see Johnston. For a didactical and critical exposition, see Mitchell.
Johnston, Mark. (2010). "Structural Unemployment - The Beveridge Curve". Econofix.
Mitchell, William. (2010). "Nobel Prize - Hardly Noble". Billyblog.

Image Credits:
All charts come from Faberman's and Mazumder's excellent paper. I edited the charts only so as to present the full corresponding sources.

No comments:

Post a Comment