Regional Research

Expect St. Louis and Missouri Construction Employment Declines To Be Revised Away.

Economic conditions in local construction are not as dire as they might appear. This FRED graph shows that mining, logging, and construction employment in St. Louis declined almost 7 percent from November 2017 to November 2018. Such a drop appears to be the largest of any metro area nationwide. This development caught our attention, as it presents a much dimmer view of the construction sector than reported by our Beige Book contacts in recent months. It's important to keep in mind that the recent data on employment are not final.

As we have written about in previous articles,¹ these regional data are subject to substantial but predictable revisions.   For example, if we look back just one year using our Archival FRED database (ALFRED) we get this graph, which shows that construction employment was reported at the time as declining by an average of 0.8 percent between January and October 2017. However, the data reported today indicate construction employment was increasing by 1.9 percent during the same period. With history as a guide, one should be skeptical of these data, particularly given the contradictions with anecdotal reports.

These revisions come from a "benchmarking" process that revises the survey based data reported above with administrative data. While the labor department doesn't publish the St. Louis MSA construction series to calculate this early benchmark, the administrative data for the first half of 2018 are available for Missouri. Following the method in the previous report, an early benchmark for Missouri suggests an upward revision to the data in a few months. Given the high correlation between the St. Louis and Missouri series, it is likely St. Louis construction employment will also be revised upward as well.

NOTE: The figure plots the published employment data for Missouri and St. Louis, along with our early benchmark measure. The data used to compute the early benchmark are available only up until June 2018. The published data were used to extrapolate the series to the end of the sample.

¹ See, for example, "St. Louis Fed Steps in to Provide More-Timely Jobs Data" and "Don't be surprised: Employment data get revised."