The Beveridge curve and cultural evolution

Pascal Michaillat discusses the Beveridge curve in a recent labor market commentary. Beveridge curves are identified from a plot of job vacancies against unemployment like that shown in Figure 1. The dashed line connecting the points shows the flow of time, starting in December 2000, indicated by the purple asterisk in the lower left, all the way to February 2024 (green asterisk). The plot first flows right to the red asterisk, representing the June 2003 employment bottom of the post-2000 recession, then moving to the left with the employment expansion, tracking back approximately the same path the economy took going into recession. Unemployment moves back and forth on the Beveridge curve during each business cycle. Different periods give somewhat different curves (see Figure 5 here).

The 2008 financial crash sent the curve off the chart to the right, and then back onto the chart, heading left in the post crisis recovery. This expansion track defines a new Beveridge curve that is different than the previous one; it runs higher in job vacancies for a given unemployment rate. As a result, a point comes when vacancies (labor demand) exceeds unemployment (labor supply). The rising line in Figure 1 shows where vacancies and unemployment are equal. The region to the left if this line means a tight market, while that to the right means a slack one. As the labor market moves towards the tight side, (demand is rising relative to supply) the expected result is a higher price (wage) for labor, which is indeed what happened. The points colored in red reflect the 2016-2019 period when real wages were rising (see Figure 2). This period ends in February 2020, just before the pandemic (see black cross in Figure 1). With the onset of the pandemic, the Beveridge curve goes crashing to the right as the country shuts down. The months from March through August 2020 are off the chart to the right. The plot returns to the chart, heading left, in September 2020. It continues left, and then down to the present (green asterisk).

Figure 1. Job vacancy rate versus unemployment rate (Beveridge curves) since 2001.

The return curve following the pandemic is even higher than the return after the financial crisis. It passes into the tight region at an unemployment level just a bit below 6%. Yet no wage growth was observed, instead we got inflation. The standard explanation for this was supply disruptions and government stimulus-enhanced demand gave rise to the inflation and that offset nominal wage gains stemming from labor market tightness. The Fed raised interest rates dramatically in response to the bout of inflation, and annual inflation rate subsided from a peak around 9% in mid-2022 to around 3% a year later. After this it has remained largely unchanged. Wage increases did not commence once this happened. Rather vacancies have been shrinking, suggested that rather than raising wages to acquire the workers to fully staff their operations, businesses prefer to keep the higher profit margins they achieved during the inflation by hiking prices by more than costs rose. It is also possible that higher interest rates are increasing business costs, putting additional pressure to minimize growth in the wage bill.  

The balance model for inflation provides another explanation for this. This model is based on a money balance on the economy obtained by subtracting dividends, stock buybacks and trade deficit from growth in M3 money plus the government deficit, all expressed as a percent of GDP. The result is Balance, a measure of inflationary conditions, that is, how much money is chasing the goods and services produced. Balance is correlated with NAIRU, sometime called the natural level of unemployment. Balance can be converted into NAIRU using this correlation, and I usually express Balance as its NAIRU equivalent. NAIRU values obtained from Balance are estimates of what the actual NAIRU parameter is, which is usually not known in real time.

During the 2016-19 period (in red in Figure 1), unemployment rate relative to estimated NAIRU ranged from –0.4 to +1.0 points (average +0.1), implying an absence of inflationary forces.  In contrast, over the 2021-22 period (in blue) unemployment rate relative to estimated NAIRU ranged from –2.7 to –0.3 points (average –1.8), implying strong inflationary forces, which manifested as inflation. After the inflation peak unemployment rate relative to estimated NAIRU ranged from –1.1 to –0.7 points (average –0.9), implying weaker inflationary forces, which is consistent with both the decline in inflation rate following Fed interest rate hikes, and their failure to bring inflation back to the 2% target. What these hikes seem to have done is bring job offers down. Perhaps forces that suppress job offers serve the same inflation-suppressing function as higher unemployment rates do. At the present trend, vacancies will reach pre-pandemic levels in another year and exit the region of labor market tightness six months after that. Any force for higher wages will have been eliminated and the economy will return to the pre-2008 normal of permanent flat real wages for unskilled workers.

Figure 2. Entry-level wages and unemployment rate trends since 2008

Over the last couple of years, the economy has been in a rare period of sustained labor market tightness, with inflation under control and interest rates at historically normal levels. Yet there has been little rise in wages for unskilled labor. I think this is the reason why voters are so sour on the economy.

 I focus on unskilled labor partly because it eliminates questions of mismatch between worker skills and job skills. These are entry level jobs most every able-bodied person can do. Employers need workers, I see help wanted signs everywhere; most low-cost restaurants in my area are short staffed and provide subpar service, except for one. They pay $2-3 more than other places and they are fully staffed and provide excellent service. It seems obvious, under-staffed businesses who employ entry-level workers could get more workers if they paid more.

They don’t do so probably because they see this labor tightness as a temporary affair and plan to just ride it out. Experience has shown this to be true. Eyeballing historical Beveridge curves in Figure 5 here, it seems that business cycles in the 1970’s through 2000’s rarely ventured into the labor-tight region while those in the 1940’s through 1960’s routinely did so. The 2010’s business cycle briefly returned to pre-1970 behavior, but any pain from having to pay more in wages was offset by the generous tax cut passed in 2017. Outside of that brief period, business owners have not had to deal with systemic labor shortage for their entire careers. And this shows up in real* unskilled wage trends, 0.6% growth over 1970-2010, compared to 2.3% over 1940-70 and 1.6% over 2010-9.

I’ve talked about all this before, Beveridge curves simply provide another way to characterize the effect of the shift from SC to SP business culture. Since business culture is a (delayed) function of the economic environment, which reflects government-set economic policies, ultimately the kind of economy we have and how it works is politically determined. The effect of politics on the economy is mediated through cultural evolution, which is slow. Major shifts in economic culture reflect major shifts in the political zeitgeist that I call dispensations. The current dispensation was established by President Reagan in 1981. It called for economic policies that promote shareholder primacy, such as legalizing share buybacks and tax cuts leading to permanent deficit spending. More share buybacks mean less investment of earnings; they act like higher interest rates to slow the economy, exerting a deflationary effect. Both tend to keep labor markets slack. Before buybacks became popular, higher interest rates were needed for inflation control until around 2000. The deflationary effect of buybacks can be overwhelmed with large enough deficits, which is what happened in 2021-2022, and so today high interest rate and buybacks are working together to suppress inflation.

Biden is getting hammered on the economy because he is overseeing an economy with unemployment at 3.6% with an estimated NAIRU of 4.5%, indicating modestly inflationary conditions and the need for high interest rates to keep a lid on inflationary forces. Voters are comparing Biden in 2023 with Trump in 2019, who had unemployment of 3.8% with a NAIRU of 3.9%, indicating little inflationary forces. Hence the Fed kept rates low, with wages rising, for Trump, but invoked high rates for Biden, with no wage growth. The difference is the higher NAIRU. And the reason for is this today’s 6% deficit compared to the 4% deficit in the late teens.

Does this mean that were Biden to cut the deficit back to 4% and the NAIRU fall back, rising wages would resume? Probably not. The economy is path dependent. The past experience of the inflation has boosted the profit level for business and they are going to try to keep it. This is why conditions that would normally result in rising wages (i.e. labor market tightness) are not having that effect. The higher prices instituted during the inflation have ensured adequate revenues in spite of a possible slowdown in unit sales growth resulting from inadequate staffing,

Had Biden combined large tax increases with the American Rescue Plan Act, he might have been able to reduce the impact of the inflation. Suppose Democrats had incorporated repeal of the 1997 and 2017 tax laws as well as the lower taxation of dividends instituted in the 2003 tax law into the Rescue Plan. Two things would have happened. One would be that revenues in 2022 would be larger and deficits smaller than they otherwise would have been, Second would be a wave of profit-taking on the stock market as investors realized that earnings would probably be lower in 2022 because of the tax change, and that profits taken in 2021 would be taxed less than those taken later. Each new rally would see a wave of selling into strength, reducing market gains for that year. A weaker market could raise recession fears, leading to lower expectations of future inflation. Business leaders might feel less confident in their ability to boost profits through larger-than-necessary price increases, without affecting sales. There would still he inflation, supply disruptions were a real thing due to shifts in demand resulting from the pandemic shutdown as well as direct effects of the pandemic on production, but these were largely restricted to the goods sector. Perhaps the inflation would not have spread had business thought a recession was nigh.

But raising taxes in that way is simply not politically feasible under the current Reagan dispensation. Similarly, economists like Pascal Michaillat write about how inefficiently tight labor markets can be rectified by regulated immigration:

Immigration has a negative effect on the welfare of native workers because, in the short run, it increases the unemployment rate of native workers. But immigration has a positive effect on the welfare of native firm owners because it makes it easier for firms to recruit. When the labor market is inefficiently slack, immigration reduces native welfare because then the reduction in native labor income swamps the increase in native firm profits. But some immigration improves welfare when the labor market is inefficiently tight, because then native firm profits increase more than native labor income drops

This seems valid but only in a purely economic, ceteris parabis sense. It assumes that worker and owner income have the same function. But they don’t. More worker income translates into more demand, which leads to business expansion to meet the demand, that is, economic growth. Since all earnings now go to dividends and buybacks, and so into financial markets, more profits means higher asset valuations rather than economic growth, and this can lead to financial crisis.

*Values adjusted for inflation using the GDP deflator.