Friday, July 20, 2012

Calling BS on Sticky Wages

There's a lot going around the economics blogosphere about sticky wages. A great deal has been made of various graphs that show wages are sticky downward (but not upward). Here's one from George Selgin, a researcher whose work I greatly respect: The graph does, in fact, show that average hourly earnings in the private sector have steadily increased every month from 2005 to now.

Unfortunately, and like most aggregates, that's only part of the story. I looked at the change in average hourly earnings from April 2006 to June 2012 for the following sectors: Construction, Durable Goods, Education and Health, Financial, Good-Producing in general, Information, Leisure & Hospitality, Manufacturing, Mining and Logging, Nondurable Goods, Other Services, Private Services, Professional and Business Services, Retail Trade, Total Private, Transportation and Warehousing, Utilities, and Wholesale Trade. I have graphs for all these, and they generally show upward trades, although slopes differ across the industries.

What's more important for the argument regarding 'downward stickiness' or the reluctance of firms and employees to settle on a decrease in wages is the frequency we observe negative changes (decreases) in the average hourly earnings. You'd think from the 'sticky' arguers that we never see negative changes. This is wrong. In fact, wages do decrease, about 22% of the time (average across all industries). The industry with fewest wage decreases is Private Services (4% of all changes are negative), and the industry with the most is Utilities (40% of all changes are negative). The magnitude of decreases is -0.27% across all industries, while the magnitude of increases is 0.40% across all industries. The industry with lowest decreases is Private Services (-0.06%) and the industry with highest decreases is Mining and Logging (-0.82%).

My point is this: wages do not appear to be sticky. At least, the evidence isn't all in favor of stickiness. Viscous, maybe, but not sticky. Also, here are three graphs drawn from the extremes of my wage change distributions showing the relationship between employment and wages. The first is Private Services, then Mining and Logging, and then Utilities. I see upward trends post-recession of both employment and wages. Riddle me that Batman.

For a strong theoretical point as to why wages are NOT STICKY, see my former Prof and blogger extraordinaire David Andolfatto:


  1. I am not sure how one can deny downward nominal wage rigidity based on those graphs. You note wages decrease 22% of the time, but it is more an issue of magnitude than frequency. In those graphs we see large decreases in employment and then a flattening of wages or only minor decreases in wages, and then a decrease in the rate of wage growth.

    The story of nominal wage rigidity also isn't that wages don't decrease. It is that rather than lowering workers' wages following a decrease in profitability (to maintain the level of employment), firms lay-off workers, and eventually hire new ones at lower wages, or at the same wage once it is profitable again to do so (though increasing the wage as sales and hence employment recovers).

    I am truly not interested in theoretical arguments about why there is no downward nominal wage rigidity, since it is clear from this and other graphics that there is (unless you are willfully blind). One should not privilege theory over evidence, and a theory is only useful to the extent it conforms to the observed regularities (to paraphrase Friedman).

    1. My responses:

      1) There is not much relationship between wage changes and employment changes. That means something else is driving changes in wages and employment.

      2) The magnitude of decreases not that much smaller than the magnitude of increases. And I would expect the general trend to be positive simply because of monetary inflation.

      3) You are truly not interested in theoretical arguments? Well then my dear just how do you propose to interpret the empirical data? Because it is simply not the case, ever, that 'the facts speak for themselves.'