A feature of the labor market in the US that has received little attention is its declining participation rate. The decline is especially remarkable since the end of the previous century and among workers with less than secondary education. In the BSE Working Paper Nº 1347, “Labor Force Participation, Wages and Turbulence,” Francesc Obiols-Homs and Virginia Sánchez-Marcos perform a quantitative evaluation of various channels proposed at explaining the decline in the labor force participation rate.
Participation rates decline for everyone, just not in the same way
One known fact about the labor market in the United States is that in the period between 1997 and 2017, there has been a steady decline in the participation rate of all workers. However, this has been especially the case for workers with at most upper secondary education. During the same period, such a sharp decrease has not been observed in workers with tertiary education.
These patterns have been seen especially for workers aged 25 to 34 and those between 45 to 54 years. Hence, this phenomenon has been mainly affecting prime-age male workers (see table 1).
These changes have also been accompanied by higher unemployment rates for this low-education group. Moreover, the data shows that these workers now face more extended unemployment and non-participation spells. This has resulted in wage stagnation for this demographic group and, importantly, a fall in tenure wage premia.
The potential destruction of human capital is key to lower participation
The authors pose that this reduction in participation for low-education groups comes mainly from an increase in the probability of skill loss during non-employment (unemployment and non-participation). This is the so-called “turbulence” hypothesis after Ljunqvist and Sargent (1998). In the data, being in an unemployment spell in recent times is more destructive for your skills than it was two decades ago. Furthermore, not being on the job makes it more likely for workers to lose human capital. This results in lower wages and weaker incentives to return to the labor force as fast as before.
To evaluate this hypothesis, it is required a theory that defines a meaningful role for human capital accumulation to explain these facts. Thus, the authors embed endogenous human capital dynamics in a search and matching model of the labor market with non-participation decisions.
There are two types of workers in their framework: high skill and low skill. When unemployed or non-participating, a high-type worker might become a low-type worker with some given probability. This is the channel through which they model the skill downgrading that might explain what we see in the data.
How damaging is in numbers this skill downgrading?
The authors evaluate the effect of increased turbulence in a quantitative exercise calibrating the economy to the US data. According to their results, the losses amount to about 19% after the increase in the probability. However, this seems to be a conservative number since the data shows that in the last years, earning losses after exiting employment might be as high as 30% when looking at the target group.
How does this transmit to lower labor force participation?
With a higher probability of a skill loss once you exit employment the fraction of low skilled workers tends, mechanically, to increase. Furthermore, participating in the labor market is less attractive than before precisely because it is more likely to become low skill. There are two important effects that point to the same direction and that is worth emphasizing: first, high skill workers are willing to accept lower wages as a way to protect their skills from deterioration (wages of high skill workers decline about 2%). This reduces the participation of high skill workers and it also lowers the value of participating for low skill workers (since for them the present value of participating is smaller than before). Second, from the perspective of the firm hiring a low skill worker is now more profitable than before, since they have the same probability of becoming high skill in which case their wage is smaller than before. As a result of this, wages of low skill workers tend to be about 1% larger than before. This final effect is too weak to overcome the reduction in the wages as a high skill worker. Therefore, the increase in turbulence reduces both high and low skill worker participation and the overall labor supply.
The authors also explore the effect of alternative hypothesis proposed in the related literature, such as an improvement of the value of non-participation (due, say, to a better unemployment and non-participation protection programs, and due to a reduction in the market value of low skill workers after a “China shock”, robotization, etc.). These alternatives are also able to provide a decline in participation, yet they do not account for the reduction in the premium to experience.
What are the social policy implications of this?
In the model with endogenous determination of human capital the equilibrium is likely to be suboptimal. In particular, firms do not internalize the social benefits of hiring low skill workers, which can be sizeable. Doing so would promote higher human capital accumulation, higher employment rates, and larger output. This creates room to think of subsidies and other policies to promote a more efficient outcome if skill downgrading is an important channel shaping workers’ choice.