Wednesday 8 August 2012

Skills Shortage? US Evidence (III)

"We have a lot of ways to observe how the unemployed behave. (...)  But none of that mental framework exists for employers and job openings. A cynic might note that economics, as practiced, is a machine for observing and disciplining labor." (Mike Konczal)

In the last two posts in this series on structural unemployment in the US (here and here) we dealt with empirical evidence challenging the view that "structural ('large') changes in technology or in the willingness to work" explain current levels of unemployment in the US. As these views have policy implications, policies advocated on this basis were challenged, too.

Today we'll revisit the two posts. Readers could find it convenient to open them in separate tabs, so as to follow the present exposition.

In the first post we described what the Beveridge curve was and how unemployment/vacancy data appearing to the north/east of it was to be interpreted: either jobless workers were less willing to work or a deep technological change made workers' skills obsolete (more "colourfully", the "job snob"/"technological change" view). In other words, we deduced the causes from the effect.

Before moving on, let's observe that in the labour market, workers represent the supply side of the market; employers, the demand side. Therefore, standard search theory claims that a high unemployment/vacancy ratio is caused by an inadequate labour supply: the story is all about workers' motivations and/or what they offer employers.

In the same post we presented evidence gathered by Faberman and Mazumder casting doubt on the "job snob"/"technological change" view. The specific piece of evidence I am referring to was the chart labelled "4. Labor market trends by skill group. B. Index of labor demand".

Given the previous paragraph, I hope readers will have noticed a subtle shift in focus: from considering the supply side of the market as the cause, we are considering now the demand side (that is, the employers); and, we find that the problem of slow job recovery may be caused by an inadequate demand, that is, by the employers.

For brevity's sake I'll bypass the examination of the MF evidence presented in the second post. Readers are invited to check by themselves: slow job recovery no longer is all about workers, but about prospective employers.

So, if we chose to focus on the jobless' characteristics in isolation, we deduce that they are responsible for joblessness; and we conclude this on the basis of an a priori reasoning: basically the only evidence adduced is the Beveridge curve itself, and, as already stated "from the Figure itself one cannot conclude what causes this".

If, on the other hand, we chose to focus on employers' characteristics, we find that, to a large degree, employers seem responsible for joblessness. And we conclude this, because the data say so.

From this characterization, it seems the choice of one explanation is an arbitrary decision. Next I intend to show it isn't.


In 2010 the Nobel Prize for Economics was awarded to 3 economists, Peter A. Diamond, Dale T. Mortensen and Christopher A. Pissarides for their work in search theory. In December 13, the same year, the recipients delivered their lectures at Uppsala University.

Among other subjects, Mortensen and Diamond treated Diamond's retail market model. Diamond considered a market where shoppers perform a sequential search; the search involves a cost (time, effort, money). In these circumstances retailers establish a unique monopoly price. As Mortensen explained in his oral presentation:
"The rationale is simple: if [the search] is sequential and you are the buyer and you know that all prices are the same, you buy the first time you have an opportunity at the going price. But if all the buyers are buying at the going price, then the sellers know that the workers are not going to search for a second price". [5]
At a monopoly price, retailers get all the benefits of trade, as monopoly prices are higher than competitive prices. Note that in this case, the supply side gets the upper hand.

The direct application of this model to the labour market only requires substituting jobseeker for shopper, employer for retailer. The conclusions remain the same. In the retailers' case, we speak of monopoly price; in the employers' case, of monopsony wages (which are lower than competitive wages). And in this case, it is the demand side, the employers who get the upper hand.

Eventually, researchers (not so much Diamond who moved on to other areas, but Mortensen, Pissarides and others) decided that these original assumptions were unrealistic and "relaxed" them. Models proliferated.

Why these assumptions were found unrealistic? Frankly, I don't know. I can speculate, though: maybe these researchers felt it unrealistic that employers had market power; perhaps they found it more realistic that the jobless are too picky when looking for jobs. This, for instance, could explain comments like those one sees in the media.

In any case, one can chose a specific model, with a given set of assumptions that guarantee one's preconceptions follow. From that, one "deduces" the causes (usually "unwillingness to work", "technological change"). And this is what the economic policy discussion boils down to.

This is where evidence comes to the fore and why the Faberman and Mazumder paper is vital: maybe these other models follow logically from their assumptions (and that in some cases could be a big if), but they are not supported by the evidence.


Regardless, even after all this relaxation was done, this is how Diamond concludes his own exposition (about 35:40 into it):
"I come away with the view (tentative view because we don't have nailed down all these effects) that what we really need is a lot more aggregate demand stimulation in the US. And of course any attempt to measure mismatch is gonna be sensitive to the tightness of the labour market. The idea that you can measure that as something independent of tightness and then say what's left is cyclical seems to me to be basically wrong". [2]
Again, the primacy of evidence comes to the fore.

And yet, all the Very Serious People in the US, Europe and Australia keep talking about skills shortage, mismatch and such, with assurance only matched by their evident ignorance and possibly by the vested interests speaking through them. And they do that, basically on one statistic: the Beveridge curve. That is, when they actually mention evidence.

And all Very Serious People keep talking about austerity.

In the previous discussion we mentioned that, depending on context and assumptions, sometimes supply, sometimes demand, gets the upper hand in search models. But in the economic policy discussions, one only sees references to the supply side of the labour market: the workers. Here I close this exposition, by invoking Konczal's quote opening this post.


As time permits (and I'll remind readers that I'm doing this on my spare time) I intend to proceed this investigation, focusing now on the Australian labour market.

Further Resources:

For an accessible exposition on search theory, see Andolfatto.

[1] Andolfatto, David. (2006). "Search Models of Unemployment". A preliminary document prepared for the New Palgrave Dictionary of Economics, 2nd. Edition.

[2] Diamond, Peter A. (2010). "The Nobel Prize Lectures in Uppsala 2010 - Economic Laureates - Peter A. Diamond".

[3] Diamond, Peter A. (2010). "Unemployment, Vacancies, Wages. Prize Lecture", December 8, 2010.

[4] Mortensen, Dale T. (2010). "Markets with Search Friction and the DMP Model. Prize Lecture", December 8, 2010

[5] Mortensen, Dale T. (2010). "The Nobel Prize Lectures in Uppsala 2010 - Economic Laureates - Dale T. Mortensen"

No comments:

Post a Comment