In a post in February, I made the claim that in the 1981-82 recession, a fall in real wages helped employment recover. Menzie Chinn took issue with that, pointing to a time series showing real wages not falling. On Thursday, David Leonhardt produced another time series that supports my claim.
HT to Alex Tabarrok.
READER COMMENTS
John Hall
May 15 2010 at 7:18pm
I think in the back of your mind you’re thinking about the relationship between hours worked and wages. It seems that your line of thinking is that weak growth in real wages means that firms will increase hours worked. However, from the perspective of the recovery, it makes more sense to think about aggregate spending on wages and salary.
To do this, I took the BEA’s series on private wages and salary and deflated it by the PCE deflator (data available since 1959). Most recessions show private wages and salaries fall 3-5% during the recession and then recover around 5% in the year after the recession ends. The exception are the last two where the rebound was flat. If you just look at the 1982 and earlier recessions, the rebound in personal income was consistent with history.
With wages that weak in Leonhardt’s graph, it would be impossible to have the rebound (in personal income) they had out of the 1982 recession if hours worked didn’t substantially improve. Indeed the unemployment rate fell 2.3% (10.8% to 8.5%) in the first year after the recession. Maybe weak real wage growth was responsible for the decline in the unemployment rate. However, the recovery in personal income following the recession was primarily due to robust job growth.
For an econometric approach, I used the data that Menzie charted, along with the HOABS series from that same menu he got his series from at St Louis fred. I regressed percent changes in the hours worked series, against its lag, percent changes in the real compensation per hour, and its lag. The three independent variables are all significant and the R^2 is around 34.3%. While the sign on real compensation per hour growth is negative, the lag has a positive sign (with quite close values). So rising real compensation today depresses hours worked, but rising real compensation yesterday depresses hours worked. Hence, if you just completely exclude compensation from this regression, the R^2 only declines to 32.7%. In other words, adding in compensation to this model doesn’t really add much.
To me this should be obvious, all else being equal, a decline in real wages would be consistent with increasing hours worked. However, declines in real wages (say more than 0.5% per quarter) are pretty rare. If the do occur frequently, they will likely be during recessions. Firms are probably concerned just as concerned about weak expected sales growth as they are about the cost side. Since expected sales growth is much more volatile than real hours, this is probably more likely what’s driving the increase in hours worked over these periods.
Doc Merlin
May 15 2010 at 7:21pm
Ok, discrepancy, why the discrepancy?
Jacob Hedegaard
May 24 2010 at 10:24am
@ David Henderson:
You wrote:
So, in effect, you are arguing, that inflation helps the economy recover, and, indirectly, that NGDP is a good measure for economic growth?
Comments are closed.