· An interesting, but deeply flawed analysis from the Peterson Institute for Economics was recently released. This note addresses that analysis and builds on my prior work on Beveridge curve cycles.
· The surprising number of job openings is likely the result of the sudden unexpected unemployment shock (voluntary and involuntary) that hit the entire economy simultaneously in 2020.
· Firms are stuck with a pre-COVID capital to labor ratio, but they have a post-COVID labor force to man the machines available. Firms are desperately trying to right-size their labor force, but they cannot all do so because there simply is not the labor supply.
The magic of the movies is making the unreal seem real, aided to various degrees by the suspended disbelief of the audience. Perhaps the greatest trick motion pictures have achieved is the appearance of motion when a series of still images is recorded and played back at a rate of least sixty frames per second. At that rate the human eye cannot discern the gaps in motion that take place between frames, but we know that motion does take place in the real world.
Statistical economists view the real world like a strip of film but refuse to acknowledge the very existence of the time that takes place the frames. No matter how finely you define ticks of the clock, there is time from one tick to the next and time can be further subdivided into smaller and smaller increments.
The infinite divisiveness of time matters because statistical economics shows a series of images of different states of equilibrium with no effort to assign causality. It is as if the filmmaker chose a noticeably low frame rate and asked the viewer to assume the gaps were due to teleportation from spot to spot with no movement. Unfortunately, this assumption has resulted in a lot of low-quality economic analysis dominating the debate and so it is with the Beveridge curve analysis released by the Peterson institute.
The Shifting and Twisting Beveridge Curve
No doubt the job opening rate is extremely high, but the unemployment rate is also extremely low - arguably at a structural bottom (Chart 1). Therein lies the key point of disagreement between this writer and the authors of the Peterson Institute paper. Their argument is that the Beveridge curve has shifted but remains the same shape (Chart 2). One surprising shortcoming of the paper, given the number of renowned economics working on it, was the paucity of reference papers. If they had done just a bit more searching, they would have seen that in addition to shifting the Beveridge curve can also twist because of search costs (money and time). Indeed, the Beveridge curve goes through cycles along with cyclical tightness in the labor market[1].
The surprising number of job openings is likely the result of the sudden unexpected unemployment shock (voluntary and involuntary) that hit the entire economy simultaneously in 2020. That is highly unusual because normally the economy does not suddenly shut down for an entire month, so the usual cyclical shift in labor demand across sectors did not occur as it normally would (Chart 3). There is no reason to expect that situation to persist, or why it would permanently reduce the matching efficiency of the labor market. Without causality statistical analysis can easily lead to incorrect conclusions that happen to fit the data.
We are likely to find out the answer quite soon because the job openings rate across the country and across sectors has started falling (Chart 4). If unemployment begins to rise then we know that the Peterson Institute paper is correct, otherwise job openings will fall along the vertical axis. That vertical axis likely represents the structural bottom for unemployment, although we will know more as the quit rate falls (Chart 5).
If search costs begin to fall along with demand for labor the job openings rate could remain elevated (Chart 6). Obviously if searching is relatively cheaper employers will do more of it and they are showing signs of moderating. Search costs are undoubtedly high, and employers are reporting job openings are hard to fill, but neither indicator is out of line given the low unemployment rate. Fortunately, both are coming down which is necessary for the labor market to return to more usual conditions (Charts 7 & 8).
The State of Matching Efficiency
As mentioned above, the argument in the Peterson Institute paper is that matching efficiency has permanently deteriorated. No doubt, in 2020 the sudden disturbance to the economy and labor market along with cash largesse from the government temporarily reduced matching efficiency. However, that shock had worn off by 2021 and the labor market began operating with a normal level of efficiency but a high quit rate and low unemployment rate.
Ignored in the Peterson paper is the shifting of the slope of the yield curve (shown in Diagram A), from flat to increasingly steep, that takes place naturally and regularly as the labor market tightens. The labor market is unusually tight and demand for labor is extremely high, which makes the situation unusual but not necessarily different. Diagram B shows the shifting and twisting that the Beveridge curve undergoes across the course of the business cycle.
The Peterson paper does not address this possibility, likely because it would be difficult to express mathematically. The twisting of the Beveridge curve occurs as the search intensity, quality, and quantity of unemployed workers changes as the business cycle progresses and more and more of the available labor supply is made use of. That is a dynamic process that is not driven by causality and not easily addressed by a statistical analysis.
The job finding rate for those unemployed less than five weeks has recovered to and surpassed the rate observed prior to the COVID shock and the 2007-2009 recession before that (Charts 9 & 10). There is no reason to believe yet that the rate will not continue to climb if the labor market remains stable. The simple answer to the question of what is happening in the labor market is that nobody wants to work. The participation rate has not recovered to it’s pre-COVID rate and is indeed trending downward at this point (Chart 11). Indeed, the number of participants in the labor market might have topped off below the 2019 rate (Chart 12).
That sudden shock to the participation rate explains the intense demand for labor we are currently observing. Firms are stuck with a pre-COVID capital to labor ratio, but they have a post-COVID labor force to man the machines available. Firms are desperately trying to right-size their labor force, but they cannot all do so because there simply is not the labor supply. The so-called “conundrum” or unexplained distortion reducing matching efficiency is actual an economic disequilibrium that cannot be explained by a methodology that jumps from one quasi-equilibrium to another with no accounting for what takes place in between. This is economics, not rocket science, and written or spoken language is a much more effective tool for understanding economic phenomenon and projecting future outcomes.
For a Few Dollars More
Americans love meat and, despite the extremely tight labor market and undesirable working conditions in the knackering profession, employment quickly recovered and surpassed pre-COVID levels. However, that came at an extremely steep cost as wage growth in the industry accelerated to rates not seen since the 1970s. As mentioned above, the situation is unusual but not necessarily different from the past.
The shock to the labor market caused a sudden, although not unprecedented, steepening of the wage Phillips curve just as occurred after the shock of the financial crisis of 2008 (Charts 15 & 16). That means an easing of pressure is likely to reduce the rate of wage growth and inflationary pressure. It does not mean that inflationary pressure will have permanently eased unless the Phillips curve flattens again. That will only happen if inflationary pressure eases and workers stop worrying about their lifetime cost of living.
Conclusion
The matching efficiency of the labor market has likely not fallen permanently. Undoubtedly there was a bad shock to efficiency caused by businesses closing, workers moving, and cash distributions by the government to households. But that shock had clearly faded by mid-2021 and the labor market returned to the tight conditions observed in 2019. The economy has likely moved past the pre-COVID level of tightness because the participation rate, indeed the number of workers, is lower than it was in 2019. That has serious implications for businesses because their collective capital/labor ratio, determined by the capital currently in place, does not match the demand patterns of the household sector or the availability of labor. That is particularly the case in low skill service industries that simply cannot keep their positions staffed at the current level of demand.
Although the situation in the labor market is unusual, it is not abnormal and there is no indication of a permanent decline in labor market matching efficiency. There is a cyclical aspect to matching efficiency known as the Beveridge cycle. At the end of a recession there are plenty of workers available for cheap but finding the right one requires sorting through hundreds of resumes. During an expansion as the labor market tightens participants become self-sorted because they are employed and will only look for jobs that are a better fit than their current job. Conversely, when the labor market is extremely tight firms begin to lose interest in recruitment because try as they might there simply aren’t the workers available. That is where the U.S. labor market is right now.
[1] See my notes “Recessions, Beveridge Cycles and Labor Market Adjustment” Part One of 6 August 2020 and Part 2 of 3 October 2020.
Related Notes
“Recessions, Beveridge Cycles and Labor Market Adjustment” Part One of 6 August 2020 and Part 2 of 3 October 2020
“U.S. Labor Market: Neutral is Negative” of 15 July 2022
“The Underpinnings of a Nonlinear Phillips Curve” of 25 September 2018