Mankiw's Folly

Greg Mankiw was just tickled to death today after apparently catching Paul Krugman in a falsehood, when Krugman wrote:

We’re already closer to outright deflation than at any point since the Great Depression. In particular, the private sector is experiencing widespread wage cuts for the first time since the 1930s.

Mankiw then instructs his readers to look at the BLS report:

In January, average hourly earnings of production and nonsupervisory workers on private nonfarm payrolls rose by 5 cents, or 0.3 percent, seasonally adjusted. This followed gains of 7 cents in December and 6 cents in November. Over the past 12 months, average hourly earnings increased by 3.9 percent.

This is one of those times, however, when it helps to not be an imbecile.

First, why in the world does Mankiw care about the year-on-year total? Doesn't he realize that everything before September is ancient history? Yes, nominal wages rose quickly before September, but not since. From January to July of last year, the CPI-W rose about 4.0%, since then it's deflated by an alarming 4.5%, and so Mankiw takes comfort by the increase in the real wage of what, 1-2%? That gives us nominal wage deflation of about 3% for past three months... All the meager gain in the real wage tells me is that, yes, nominal wages are sticky. I signed my employment contract last year -- and my employer can't change it. Since prices have fallen, my real wage has increased, even though my employer is bleeding cash and will almost certainly try to renegotiate after this year...

Second, Mankiw doesn't get the difference between real and nominal. Krugman was arguing for nominal wage cuts, while the BLS is saying that real wages -- deflated with the CPI-W, which fell by .9% -- rose by .3%, which means that nominal wages did fall by .6% in one month. Extrapolated for the year, that's a dive of 7.2% per annum. It's an embarrassment for all economists that a top tenured macro-economist at Harvard can make such an elementary mistake.


Another issue is that I could not find whether the BLS compares the same workers when they do their average wage, or whether they take the average among all workers who work (which I suspect is more likely). In January, 600,000 workers lost their jobs, including 76,000 jobs lost in the "temporary help" industry. What do you bet that those in the "temporary help" industry do not make as much as workers on average? Even within industries, when times are tough I suspect you're more likely to lay off the new, inexperienced staff who help the business least. Since these people have the smallest salaries, firing them will raise average wages, regardless of what happens to everyone else. I suspect the average wage that the BLS provides is among all workers, which does not actually tell us much. If so, then even the real wage has likely fallen off a cliff the past three months...

I can't wait until Krugman responds...

(Update: OK, well Krugman never responded, and it appears that Mankiw's number

was

nominal. I got my data from the Bureau of Labor Statistics website though, here: http://www.bls.gov/news.release/empsit.t17.htm which gives the total change in "current dollars" as .3%, with a footnote at the bottom which says "The Consumer Price Index for Urban Wage Earners and Clerical Workers (CPI-W) is used to deflate this series. -- a footnote, which i now see was meant to apply to the line below it... the "current dollars" is also confusing... they should just write "nominal values" to avoid confusion... I still maintain looking at the year-on-year change is not informative, and that the change was likely do to low wage employees being laid off...)

(UpdateII: Krugman responded. The BLS's employment cost index is supposed to control for the fact that low-wage workers get laid off first. The issue, is that they only publish quarterly series. However, in downturns they are a bit lower than the hourly earnings figures... In the 4th quarter of last year, the BLS's employment hourly cost index increased by .5%, while it increased .88% for nominal wages. If the same spread held for January, then we would, indeed, have wage deflation of .08%...)