A Non-Partisan Review of the Upcoming (Feb. 2, 2024) Benchmark Employment Revision
Former JPMorgan Chase Global Chief Economist (Ph.D. in Economist) & Current BrightQuery Chief Economist
Each year, Mariah Carey makes an announcement saying, “It’s time.” Last year, she made her well-publicized pronouncement on November 1, 2023, to reintroduce her classic Christmas song “All I Want for Christmas is You!” For U.S. political pundits, the big day occurs each year around the end of August. On that day, the Bureau of Labor Statistics (BLS) tells us that as incoming data from individual states becomes available, it projects a revision in the non-farm payroll number of jobs created through the 12 months ending in March of that year.
Remember that the BLS announces a politically charged projection before it has collected all the data needed to compute its official annual benchmark revision six months later. If the projected revision is negative, it gives the opposition party ammunition to say that the administration has overestimated the number of jobs created. And if the revision is positive, the opposition will likely remain silent since it is difficult to critique good news.
For financial and labor market observers, the latest fateful day occurred on August 23, 2023, when the BLS announced that the employment benchmark revision would likely lower net job creation by -306 thousand for the 12 months ending in March 2023.
After Mariah’s big announcement each year, holiday observers start counting the days to Christmas. Similarly, after the BLS preliminary employment benchmark revisions are announced, political and market observers begin counting the days until they get the final benchmark revision, which includes more accurate data collected from all 50 states 6 months later!
Separately, each month, the BLS also revises its two previous monthly estimates of non-farm payroll estimates. These revisions occur simply because the government needs more data when releasing its initial non-farm payroll estimates. The lack of data has become more acute recently. The government’s response rate on its Establishment Survey (used to estimate the monthly non-farm payrolls) has declined from 64.9% in April 2017 to 41.8% in September 2023! That means that job revisions could be higher, while the reliability of those estimates could deteriorate.
A good rule of thumb is that when the economy grows faster, the employment revisions are often positive, and when the economy is slowing, the revisions are usually negative. This has nothing to do with politics, but it rarely prevents political pundits from suggesting that the revisions were politically driven!
Source: The Bureau of Economic Analysis and the Bureau of Labor Statistics
Historical Examples
After the government released its preliminary benchmark revision in August 2022, projecting an underestimate of +462 thousand, some political pundits remained silent since no one could criticize good news. But in December 2022, the Philadelphia Federal Reserve released a research report revealing that payrolls may have been overestimated by 1.1 million! Political pundits went into overdrive by suggesting that the administration orchestrated the overestimate for political purposes. The study’s warning was a false alarm since the -1.1 million overestimates of actual jobs created became an actual +506 thousand underestimate after the final benchmark revision was released two months later. Political pundits went into hibernation, and the widely cited research was so embarrassing that the Philadelphia Federal Reserve removed the webpage link to the study!
Source: The Bureau of Labor Statistics
In August 2023, the BLS reported that for the 12 months ending in March 2023, employment had been overestimated by 306 thousand jobs. While this figure is not as prominent as the incorrect -1.1 million overestimate projected by the Philadelphia Fed in 2022, it was still large enough to speculate that some political tinkering may have occurred.
To analyze whether there was any political tinkering by political parties, we conducted a non-partisan historical review of the data by reviewing the annual revisions under each President going back to the start of the BLS data in 2003.
Presidential Benchmark Revisions
· George W. Bush (2003-2009): The annual benchmark revisions during the tenure of President George W. Bush indicated a positive overall adjustment of 675 thousand jobs. This suggests that during his presidency, the initial employment estimates tended to underestimate the actual employment figures.
· Barack Obama (2009-2017): Contrasting sharply, the Obama administration saw a significant negative benchmark revision of -999 thousand jobs. This implies that, on average, the initial employment data during Obama's presidency tended to overestimate the actual employment figures. These results were the worst of any U.S. President since the BLS has tracked these annual benchmark revisions!
· Donald Trump (2017-2021): The Trump administration also faced a negative revision of 491 thousand jobs, signaling that, on average, employment estimates overstated the actual employment figures during his presidency.
· Joe Biden (2021-Present): Initial estimates during the Biden administration suggested an underestimation of 193 thousand, signaling that, on average, employment estimates were more robust than the preliminary estimates provided by the BLS. It is crucial to note that the data available at the time of writing reflects only part of President Biden's term.
Source: The Bureau of Labor Statistics
Employment Benchmark Revisions by Political Party
Finally, we examined the annual benchmark revisions and revealed some exciting results. Due to the outsized overestimation of jobs created during President Obama’s administration, Democrats, on average, overestimated job creation by a cumulative 806 thousand during their tenure in the White House, while Republicans underestimated the strength of U.S. labor markets by 184 thousand!
We leave it to the reader to decide whether these results result from the stages of the economic cycle or political tinkering, given that a careful review of the data revealed that during periods of slowing economic growth (irrespective of the political party in power), the BLS tends to overestimate job growth, while during periods of robust economic growth, the BLS tends to underestimate job growth.
Source: The Bureau of Labor Statistics
Summary and Concluding Thoughts
While it is tempting to conclude that benchmark revisions are politically motivated, to the non-partisan observer, the evidence is mixed. I could understand why some Presidents (namely, President G.W. Bush and President Obama) might wish to exaggerate labor market strengths during their administrations. Still, I am at a loss to explain why President Trump and President Biden may have understated the strength of U.S. labor markets during their tenure.
A more plausible explanation is to acknowledge the non-partisan nature of the BLS and accept that the annual benchmark employment revisions are based on more accurate and comprehensive data, which also reflects the challenges of estimating employment numbers accurately. This is especially true in an environment where the BLS has encountered a substantial decline in the number of establishments submitting monthly data to the BLS used to compute these employment statistics!