Tuesday, September 19, 2006

Further notes on labor hours 

A week or so ago I blogged about Russ Roberts' article on how we were calculating workers' real wages. Russ writes, to repeat:
How do businesses report their total hours worked to get an accurate number? How much leisure takes place on the job today compared to 1964? How much work gets done at home that the company has no knowledge of? I have no idea. Neither does the BLS.

How does the BLS adjust the weights in the survey? When a manufacturing plant closes and a graphic arts company opens, how do they re-weight the sample? I'm sure they do re-weight, and I'll try and find out how they do it, but does their method result in any systematic bias?

I thought, since I know some folks at BLS -- and I need to, as a heavy consumer of their data -- I would ask. While I have nothing definitive to quote, I got a long, off-the-record answer to some questions which I can summarize for you. Of course, this will be only as good as my summary!

The National Compensation Survey (NCS) directly surveys worker hours. If you have a worker who is paid at some wage, it's not much of an issue -- each firm surveyed has someone who can tell you how many hours they worked. Full time is easy, part-time is estimated, but the examples I was given were reasonable. But, for example, it's difficult to deal with construction workers' seasonality in employment, or workers who work odd four-on, three-off type shifts, or public safety workers, etc. In all cases there's an attempt to build what the weekly schedule is of these workers. These estimates do not usually include overtime (those data come from the CES, more on which below.)

Roberts' point is to the "exempt worker" who doesn't punch hours. In this case one asks what is the expectation between employer and employee for hours to be delivered. Now sometimes they will code a special number of hours for some workers -- tax accountants for example work more early in the year -- but normally you code contracted hours. And you can't code exactly every hour a salesperson works, as successful sales people work more hours than unsuccessful ones.

My correspondent makes a good point, though, in the difference between NCS and CES (from which we get the monthly payroll employment figures as well as average hourly earnings). The former is a survey where someone spends maybe an hour talking to someone to get a feel for the data, whereas the latter is data collected by a field economist and evaluated off-site; the average wage is a number written down by a payroll clerk. There are changes coming to the series that would allow better measurement.