Arnold Kling points to labor cost data that he sees as being inconsistent with sticky wage models of the recession. I have a different view of macro than most economists, for instance I’ve often argued that inflation data is basically meaningless, and should be removed from all macro analysis (except perhaps “dorm room” discussions of long run changes in living standards.) Instead NGDP is the nominal variable of interest, and hours worked is the real variable that best picks up the business cycle. Kling’s post looks at the ratio of prices to unit labor costs, a variable that plays no role in my business cycle analysis.
In my view the ratio of wages to NGDP per capita (NGDPPC) is the best way of picking up nominal shocks that might throw the labor market out of equilibrium. Normally these two variables will grow at about the same rate, although there can be gradual changes in the ratio of W/NGDPPC if the share of income going to capital increases, or if the share of total compensation going to non-wage benefits increases.
The St. Louis Fred has wage data only for the period since 2006. Here’s the change in NGDP, NGDPPC, nominal hourly wages, and W/NGDPPC over the past 5 years:
Time period: NGDP growth NGDPPC growth Wage growth W/NGDPPC growth
2006:2-2007:2 4.85% 3.95% 3.72% -0.23%
2007:2-2008:2 3.14% 2.24% 2.92% 0.68%
2008:2-2009:2 -3.90% -4.80% 2.93% 7.73%
2009:2-2010:2 4.43% 3.53% 1.94% -1.61%
2010:2-2011:2 3.77% 2.87% 1.99% -0.68%
This is a very limited sample, but I bet if you went back to 1990 you’d see a similar pattern. During normal times, nominal wages grow at roughly the same rate as NGDPPC, maybe a bit less for the reasons mentioned above. Then from mid-2008 to mid-2009 wages soared relative to NGDPPC. Why did this occur? In my view it happened because the Fed let NGDPPC fall 9% below trend. Nominal wage growth is quite sticky in the short run, so a sudden large change in NGDPPC will usually change the ratio of wages to NGDPPPC, which can be seen as the funds available to pay salaries. If those funds drop sharply, and the hourly wage is stable, the number of hours worked will fall sharply. If the Fed had engineered 3% or 4% NGDPPC growth in 2008-09, nominal wages would still have risen by about 3%, and there would have been far less unemployment.
Since 2009 wages have risen more slowly that NGDPPC, but at current rates it will take years to restore equilibrium in the labor market. One might ask why nominal wages don’t fall quickly to restore equilibrium. Paul Krugman presented data a few months back showing workers rarely receive nominal pay cuts. So in “hard times” you have some workers getting zero raises, and others (in healthy parts of the economy) getting 3% or 4% raises. The average wage increase falls to slightly under 2%.
The last 5 years fit this version of the sticky wage model almost perfectly. If there are other versions of the sticky wage model (perhaps using W/P) that don’t fit the data very well, then perhaps we should consider discarding those non-useful models.
PS. Readers may notice that I estimated NGDPPC simply by subtracting recent US population growth (0.9%) from the NGDP growth numbers. Normally I don’t even talk about NGDPPC, as NGDP is close enough. But for this exercise I thought the per capita numbers would make it easier to see the intuition. NGDP shocks are like a game of musical chairs. Remove 9% of NGDP relative to trend, and you’ll have 9,000,000 unemployed workers sitting on the floor. It’s that simple.