a week and a half ago, kevin drum points out Earlier, there was a disconnect between job creation measured by household surveys (private employment) and business surveys (nonfarm payrolls). My conclusion is that the empirical evidence suggests that most, if not all, weight should be placed on institutional surveys. Now, Torsten Slok compares the decline in response rates in the two surveys. This comparison further supports the case for using Enterprise Series.
Srock wrote:
Survey response rates for both employment and inflation have fallen sharply during the pandemic, introducing substantial measurement error and uncertainty, see chart below. One example is the widening gap in the jobs report between the establishment survey, which showed 2.6 million jobs created over the past eight months, and the household survey, which showed no job gains over the same period. …
Here is the graph (I added a red line 8 months ago in March 2022).
resource: T. Slok, Dec. 13, 2022. Red line for March 2022.
Note that starting around October of last year, the response rate of households relative to the establishment rate dropped sharply. From March onwards, the gap widens faster. Here’s a picture of the gap between nonfarm and civilian employment adjusted for the NFP concept (it’s a research series).
figure 1: Official nonfarm payrolls (sky blue), preliminary benchmark implied nonfarm payrolls (dark blue), and private payrolls adjusted for the NFP concept (green) are seasonally adjusted to the 1,000 mark. Numbers in (brackets) are net new jobs in millions since 2022M03 (shown by the dashed red line). Source: BLS from FRED and authors’ calculations.
Just as the household response rate fell, the gap in numbers widened.
So: (1) I put more weight on the institutional series, (2) the gap between the two series is more likely due to increased measurement error and bias in the family series than, for example, mainly is an increase part-timer.
