Here is a follow up to my earlier post on the April employment report.

The estimates of Washington employment that are reported monthly by the state Employment Security Department (ESD) are derived from a survey of employers (the establishment survey) conducted by U.S. Department of Labor’s Bureau of Labor Statistics (BLS). Because the estimates are based on a survey, they are subject to sampling error. These initial estimates are eventually reconciled to counts of employment derived from unemployment insurance tax returns through a process called “benchmarking”. BLS benchmarks once a year, at the time it releases employment estimates for the month of January. The benchmark is to unemployment insurance tax records for the preceding March. ESD benchmarks four times a year, when it releases estimates for January, April, July and October. These benchmarks are to UI tax records from four months earlier. For more on survey errors and benchmarking see this blog post.

The employment estimates ESD released today are the first to be benchmarked to December 2012 unemployment insurance tax records. The chart below compares these new estimates to those that ESD released last month, which had been benchmarked to September 2012 unemployment insurance records.

The new benchmarking added 12,900 to the estimate of nonfarm employment in December 2012, added 5,000 to the November 2012 estimate, subtracted 3,300 from the October 2012 estimate and added 1,400 to the September 2012 estimate. While the old numbers showed employment falling in the last two months of the year, the new numbers show steady growth.

The old and new series both show a gain of 24,200 jobs from December 2012 to January 2013. (By design, the benchmarking methodology does not affect estimated changes in employment for months subsequent to the benchmark month.) As I noted here and here, this gain is way too large to be credible. It is now clear that this gross overestimate of the December to January gain in employment was due in part to an underestimate of the December level of employment. The next round of benchmarking, due when ESD releases estimates of July employment, should reduce the employment numbers for January, February, March and April.

 Here is a chart showing paths of Washington employment in recessions since the second world war.