The Atlanta Fed's macroblog provides commentary and analysis on economic topics including monetary policy, macroeconomic developments, inflation, labor economics, and financial issues.

Authors for macroblog are Dave Altig, John Robertson, and other Atlanta Fed economists and researchers.

« Bob Hall Explains It All, Take Two | Main | More On The Labor Compensation Issue »

August 30, 2005

The Non-Mystery Of Labor Compensation (Bob Hall Explains It All, Take Three)

At Angry Bear, pgl joins Max Sawicky in his skeptical view -- the latest manifestation of it, at any rate -- of my lack-of-slack orientation.  Actually, this is a follow up to earlier comments from pgl, following up a blast from Brad DeLong, following up my response to Max, who was following up our Econoblog debate (with Tom Walker joining him on the prosecution's team).

I've been mulling over pgl's earlier post, as he honed in a argument that I did not adequately address in the Econoblog exchange, and haven't since.  Here's the money passage:

... David is arguing for a classical demand and supply explanation. But not only do we need to think in terms of quantity variables (EP and LFP), we should also think in terms of real compensation, which have not kept pace with productivity, and real wages, which have been flat. The introduction of price as well as quantity variables is where I think Brad has David cold.

Fair enough.  But I don't think that the facts on the price variables clearly bear witness to slack labor markets.  Here's a picture of real labor-compensation growth for the past fifteen years:


Much of the commentary on labor income has focused on the growth of salary and wages, but I'm not sure what theory of the labor market would make wages the correct series upon which to focus.  Benefits are a part of the returns to employment as well.  Because total compensation includes benefits as well as wages, it is that series that best represents payments from firms to workers.

If we focus on total compensation growth, it is not obvious that one would want to characterize growth in labor income as extraordinarily weak over the past five years.  For the period from the fourth quarter of 1996 through the fourth quarter of 2000, four-quarter growth in real compensation averaged 1.02 percent.  For the period from the fourth quarter of 2001 through the second quarter of this year the average was 1.55 percent.  This reverses the picture one would get from looking at wage growth alone, where the corresponding averages were 1.2 percent and 0.59 percent.  In terms of total compensation -- which I argue is the right thing to look at -- the recent past has actually been better than the go-go days of the latter 1990s.

It is true that the trend has been down over the past year-and-a-half, but that observation is not clearly out of line with productivity developments:


One might wonder why the labor-return and productivity patterns aren't even closer-- as Max and Tom did in the Econoblog discussion -- but here is where Bob Hall steps into the conversation again.  Hall's extended discussion of how to think about labor markets centers on the frontier of models in the tradition of Peter Diamond, Dale Mortensen, and Chris Pissarides.  Without going into too much detail about these models -- you can find much more in Hall's paper, or in the article by Richard Rogerson I referred to in the Econoblog feature -- these models treat employee/employer relationships as the outcome of costly search and negotiation processes, as opposed to the anonymous spot-market like constructs of typical macro models.  Here's a key passage: 

Our model delivers wage rigidity by disconnecting wage bargaining from conditions in the labor market. Once a qualified worker and an employer have found each other and determined that they have a joint surplus, costs of delay, not outside conditions, determine the bargain they make.

The wage does respond to productivity, but only half as much as in the standard model. The result is a strong response of unemployment to productivity and other driving forces. The wage no longer has a strong equilibrating role. If productivity falls, the part of the surplus accruing to employers falls sharply and they cut back on recruiting effort. The labor market softens dramatically.

That last part is particularly striking  in light of the picture (courtesy of knzn) I posted earlier showing the dramatic fall off in help-wanted advertising  post-2000. 

There are, of course, many variants of the Diamond-Mortensen-Pissarides model, with differing implications for the how various types of shocks alter equilibrium levels of employment, unemployment, wages, and so on.  I don't think anyone is claiming that any one of these models has all the answers just yet.  I certainly am not.  But they do lead the way to a view of the world where it is sensible think of labor markets as "soft" without the implication that there are obvious monetary policies that will make things all better.  That's what I'm talkin' about.

August 30, 2005 in Federal Reserve and Monetary Policy , Labor Markets | Permalink


TrackBack URL for this entry:

Listed below are links to blogs that reference The Non-Mystery Of Labor Compensation (Bob Hall Explains It All, Take Three) :


So when labor comp grows slower than productivity, labor's share falls. How does Hall's perspective explain things if we take labor's share as being a somewhat fundamental level in the economy. That is, it doesn't move around much, and shouldn't in some sense, but I'm not sure I see how Hall can explain its stability in light of the model he's sketching out.

Posted by: Ian D-B | August 30, 2005 at 04:57 PM

I'm enjoying this tremendously and will have more to say when I've got one or two more things straight. Basically I'm with you, especially over this:

"a view of the world where it is sensible think of labor markets as "soft" without the implication that there are obvious monetary policies that will make things all better."

Monetary or fiscal I would say. Incidentally:

"Robert Solow described the component of economic growth not explained in statistical models by labor or capital as productivity, as well as "the measure of our ignorance."

This was Moses Abramovitz talking about the residual :). There are some interesting points to be made about this, but again they will wait.

Solow having been mentioned, I have in my mind this plagiarism:

Spain - I can see the boom everywhere, except in the productivity numbers. (They are in fact in negative territory, those damn housing bubbles!).

Posted by: Edward Hugh | August 30, 2005 at 05:48 PM

One of the things that you already know I think is that the US went through a pretty profound demographic shock due to enormous immigration from the late eighties onwards. So the age structure of the population has changed from trend, you can also see this in the TFR which stabilised after years of going down (various factors at work here).

The issue is you can't simply talk about EP and LFP issues without addressing this. In a sense absolute numbers of jobs created and jobs lost are more revealing. In a weird way this whole debate reminds me of Searle's Chinese Box argument, an economic system is like the person over the other side of the interface receiving the coded information and interpreting. Economic systems don't peer over the screen and say jeez, there are a hell of a lot of young people lining up looking for work today, a lot more than yesterday, maybe I'd better adapt by doing 'x'. Things just don't work like that.I suspect there is some version or other of anthropomorphism knocking around.

What I'm trying rather to say in a rather awkward way is that economic systems run on signals, they don't peer beyond them to try to 'understand' the bigger picture.

Incidentally all this came up with Keynes after WWI. This was what the tract on monetary reform was all about. British demographics meant that an extra 250,000 or so workers hit the labour market every year, and unemployment could only be kept from rising by creating sufficient jobs, Japan, in contrast, has falling (except this month) unemployment based largely on a declining potential workforce.

What surprises me is that these issues have been around for so long, but they are only now begining to be addressed.

Posted by: Edward Hugh | August 30, 2005 at 06:06 PM

In my 1st RBC post, I noted Kash's argument that the rise in real compensation is substantially due to more costly health insurance. Not better, just more costly. Why did I mention it? It's a supply side. And yet you don't note that this is the reason for the divergence between real wage growth v. real compensation. Huh?

Posted by: pgl | August 30, 2005 at 09:16 PM

Ian -- Some versions of the DMP model replicate a constant share of labor easily enough. However, you are right that it is not so easy to see it in the variant I was emphasizing. If you look at the data, however, you do see that labor share looks quite a bit less constant than a fixed-coefficient Cobb-Douglas production technology would suggest.

Edward -- Your emphasis on demographics is clearly correct. However, I don't think pgl or Max or others are taking issue with the flattening of the trends post 1990, but the drop after 2000. It is easy to see population patterns in the former, less clear in the latter. However, it definitely is true that the post-2001 decline in participation rates and so on are heavily concentrated in the youn nger age groups. There is almost certainly information there -- I'm just not sure what it is.

pgl -- you have your very own post, above.

Posted by: Dave Altig | August 30, 2005 at 11:21 PM

I'm puzzled by the real labor compensation graphs. They don't look like what I think they should look like. What precisely is being plotted?

Since it ends up being plotted alongside labor productivity, I would have thought the compensation series to look at would be real hourly compensation for the nonfarm business sector. (The productivity series looks like labor productivity for the nfb, but maybe missing the last quarter and maybe from before the last round of GDP revisions.) But over the last four quarters, nfb real hourly compensation has increased by 3.6% (http://data.bls.gov/PDQ/servlet/SurveyOutputServlet;
, whereas the graph in this posting shows growth of 0.4% or 0.5%.

Real hourly compensation and labor productivity track each other much better than the two lines in the posting's second graph. But even the levels of those two measures slowly diverge, with productivity growing faster. But that is largely due to the fact that the BLS uses the CPI-U to deflate hourly compensation. Besides the usual drawbacks of the CPI, a consumptions goods measure of prices is wrong concetually. One should use something that measures all the prices of the goods produced by the sector. That is, one should use the price index for the nfb. The resulting series tracks productivity quite closely over reasonable periods of time. This is what one would expect from the fact that labor's share fluctuates in a fairly narrow band.

Posted by: dgsullivan | August 30, 2005 at 11:43 PM

Benefit increases are meaningful, but benefit COST increases are meaningless!
Methinks you are treating the later as if it was a measure of the former.
My benefit costs have gone up each year, yet I still have the same dental and medical services, life ins. To include benefits is to pretend that I now have my teeth cleaned 4 times a year instead of the same old 2 times, or that I now visit my doctor each quarter instead of yearly, or that my co-pay has gone down...ahahaha...funny stuff!

Posted by: RP | August 31, 2005 at 06:22 AM

Dan -- The series is the BLS' employee cost index (ECI). I am essentially replicating the pictures which appear in this month's Economic Trends publication from the Cleveland Fed: http://www.clevelandfed.org/Research/ET2005/0805/empcost.pdf

You are correct in guessing that I adjusted the series using the CPI. I did this because it tends to be the calculation that is emphasized by many of the people in this debate, who are focused on the return to workers. However, I think you are correct in asserting that, if I am going to compare these costs to productivity growth, I should be using an index which reflects product prices. I will follow up on that.

RP -- An increase in health insurance costs is conceptually no different from an increase in gas prices or the price of any other good or service. If your employer has to pay more to provide you a service, their compensation to you rises, pure and simple. The inflation adjustment controls - subject to all the usual caveats -- for cost-of-living increases.

Posted by: Dave Altig | August 31, 2005 at 07:00 AM


> "The inflation adjustment
> controls...for the cost-of-living
> increases".

If you believe the adjusted figure reflects reality, I suppose we'll have to agree to disagree. Never wrestle a statistical pig, right?

Posted by: RP | August 31, 2005 at 04:51 PM

RP -- Agreed.

Posted by: Dave Altig | August 31, 2005 at 05:19 PM

Post a comment

Comments are moderated and will not appear until the moderator has approved them.

If you have a TypeKey or TypePad account, please Sign in

Google Search

Recent Posts



Powered by TypePad