Ed Dolan's Econ Blog

Headline Employment Numbers for October Look Surprisingly Good but Fail to Reflect Full Impact of Government Shutdown

The Bureau of Labor Statistics today released its report on the October employment situation. Observers who have been apprehensive about the impact of the partial government shutdown will find the headline numbers surprisingly good. Nonfarm payroll employment increased by 204,000 jobs. The unemployment rate was up, but only by a whisker, rising from 7.24 to 7.28. However, for a number of reasons, these numbers do not fully reflect the impact of the shutdown on the economy.

Let’s begin with the payroll numbers, where the explanation is fairly simple. Payroll jobs are based on a survey of employers. They report how many employees worked or received pay for the pay period that included a certain reference date, in this case October 12th. Even though the reference date fell during the shutdown, all federal workers, whether on furlough or not, met this requirement, thanks to the generosity of Congress, who decided to award them back pay for days not worked. For that reason, the shutdown had no effect on the number of federal payroll employees. There could well have been effects in the private sector, for example, if a hotel near a national park sent workers home for the season who otherwise would have stayed through the end of October, but those effects were evidently small.

What is more, payroll job numbers for August and September were revised upward by a total of 60,000 jobs. When we take both the new data and the revisions into account (as shown in the following chart), recent payroll numbers do not look at all bad. The 605,000 job gain over the three months from August through October is a healthy improvement from the 452,000 for May through June. (Of course, the September and October numbers are still subject to final revision, so this result could change.)

Today’s report also appears to understate the impact of the government shutdown on the unemployment rate, but for a completely different reason. The BLS calculates the unemployment rate on the basis of a separate household survey in which its data collectors ask randomly selected individuals a series of questions. The classic definition of “unemployed” applies to people who answer “no” when asked if they worked for pay during the week in question and “yes” when asked if they actively looked for work. People who answer no to both questions normally count neither as employed nor in the labor force.

Unfortunately, it is hard to know how to apply this simple definition in certain borderline cases. One such case concerns people who have a job but did not work during the reference week. The BLS deals with this situation in one of two ways. People who have a job but are temporarily absent from work because they are on vacation, on sick leave, involved in a labor dispute, or prevented from working by bad weather count as employed. However, people who are on temporary layoff from a job to which they expect to return count as unemployed, remaining in the labor force even if they do not actively look for other work.

According to an explanatory box in the latest employment situation summary, this subtle distinction confused both for BLS survey workers and responders when the questionnaire was administered to federal workers on furlough. The BLS maintains that such workers should have been treated as unemployed on temporary layoff, that is, unemployed and in the labor force. Doing so would have raised the unemployment rate. However, in some cases, either BLS survey workers explained the question badly or respondents misinterpreted the question, with the result that they treated the furlough as if it were the result of a labor dispute. When that happened, they were wrongly counted as employed, thus lowering the unemployment rate.

The BLS chose not to attempt to correct such errors. Instead, as they explained, “according to usual practice, the data from the household survey are accepted as recorded. To maintain data integrity, no ad hoc actions are taken to reassign survey responses.”

So be it. The BLS makes the rules; we bloggers just report the data, which is fine as long as everyone understands what is going on. Here are the data, to which you may add an asterisk if you want.

In the long run, what happened to the unemployment rate in October is probably not all that important. If, miraculously, the ongoing budget negotiations get the nation’s fiscal policy back on the track, any uptick in October unemployment will, presumably, be reversed in coming months. If deadlock, stalemate, and government by crisis continue to be the norm, the results will show up in the unemployment data soon enough.



Comments are closed.