Recent Updates

  • New Zealand: BOP//IIP (2020)
  • Korea: Consumer Survey Index, Inflation Expectations (Sep)
  • US: Employee Benefits Survey (2020), New Residential Sales (Aug)
  • Belgium: Business Survey (Sep)
  • more updates...

Economy in Brief

BLS Failed Its Mandate: "Fearless Publication of the Facts" (Part 2)
by Joseph G. Carson ([email protected])  June 16, 2020

Mr. Cohen criticized my article of June 8, BLS Failed Its Mandate: "Fearless Publication of the Facts", stating my arguments were "exaggerated, misleading, and inaccurate". Let me respond to Mr. Cohen's key points.

First, the data in the June 8 article was taken from BLS's 15-page addendum. According to BLS, there were 8.1 million workers in April and 5.4 million workers in May that were classified as employed but absent from work. Mr. Cohen argues that the size of my misclassification is incorrect because there are always some people misclassified. That is true. BLS said that in the previous four years those numbers average 620,000 and 549,000 in May. But, misclassifications in April and May 2020 were 10X the norm---that's an accurate account of the data and not an exaggeration!!!

Second, Mr. Cohen says "misclassification" is not certain. That is fair. But so too are all of the numbers in the recent employment reports because of a change in gathering data and the low response rates. Mr. Cohen failed to mention that Census did not conduct in-person interviews, and two Census call centers were closed. Also, the response rate for households entering the survey for the first and second months was exceptionally low--30% below the average of the past 12 months. Surveys based on a smaller number of observations have higher error rates. That means the odds are high that the number of misclassifications is understated--that's not an exaggeration!!!

Third, Mr. Cohen argues that the data in May was an improvement from April "as the share of the adjusted unemployed on temporary layoff who were misclassified declined to 25% in May from 30% in April and 39% in March". A 25% 'error" does not mean the May data is accurate, only that it's less distorted than the prior two months.

Fourth, the June 8 article main argument was that BLS's statistical methodology failed to produce more accurate data. How can that point be argued? BLS was well aware of the problems with the statistical methodology in March. If BLS did more phone calling or extended the survey period to obtain more responses that would break from past practices. But it would have provided a more accurate picture of the level of joblessness during the pandemic.

The pandemic of 2020 will stand alone in American history triggering the sharpest contraction in economic activity without an accurate statistical record of job loss and the level of unemployment.

Viewpoint commentaries are the opinions of the author and do not reflect the views of Haver Analytics.
large image