An activist cheers at a minimum wage rally.

Economists care a great deal about the minimum wage because it is a policy prescription that increasingly affects a large portion of the workforce and because it is a clear case of government intervention, imposing a floor on the market price of labor. Minimum wages therefore offer a policy tool to test theories about how the labor market operates. In a new working paper, Alan Manning of the London School of Economics argues that a clear signal of the negative employment effects of the minimum wage is“elusive,” which should not be surprising if we think about the mechanisms underlying competition in the labor market.

Manning first reviews the response of teen wages and employment to increases in the minimum wage in the United States from 1979 to 2014. Teens are a declining portion of the workforce, but they are more likely to be earning the minimum wage than adults, and Manning indeed presents clear evidence that the minimum wage raises teen wages. When it comes to employment, however, most of the evidence does not suggest significant, negative employment effects of the minimum wage. Only one of seven models with various geographic controls that he presents has a statistically significant, negative effect; five out of the seven models yield small, positive point estimates.

The lack of clear evidence for teen disemployment, Manning explains, seems inconsistent with the predictions of the simple, perfectly competitive model of the economy, where the labor market is approximated by a downward-sloping demand curve for labor, and where the market wage would otherwise be at its market-clearing level. Under this view of the labor market, the direction of the effects of the minimum wage is unambiguous: Minimum wages must reduce the employment of low-wage workers. What’s unclear theoretically is the magnitude of the decline in employment.

Manning, however, spells out three channels through which the employment effects could be small. The first relates consumer demand to the products that minimum wage workers produce. After a rise in the minimum wage, employers pass some of the increased labor costs onto customers. But because the price of a hamburger, say, is only partly determined by low-wage labor costs, the price change may not be that large. Consumers, moreover, may not be that sensitive to a small price increase of a few cents rather than a few dollars. The smaller this elasticity of product demand—when consumers buy just as much fast food at slightly higher prices—the smaller the disemployment effects of the minimum wage. The redistribution of income from owners to workers with higher propensities to consume further attenuates reductions in consumer demand.

Second, when an increase in the minimum wage raises labor costs, employment falls as affected firms substitute away from low-wage employment toward the purchase of capital and intermediate goods produced by other firms. If the elasticity of substitution away from labor and toward other goods is low, as empirical results suggest, then the perfectly competitive model of the labor market need not generate large, negative employment effects. Péter Harasztosi of the Magyar Nemzeti Bank and Attila Lindner of the University College London argue that the low substitution between labor and intermediate goods is one reason why the perfectly competitive model of the labor market is consistent with the small employment effects that they found for what was a large minimum wage increase in Hungary.

Manning also raises a third possibility: Perhaps the labor market isn’t best described by the purely competitive model. In a world with no job-search frictions, no one would celebrate getting a job or mourn losing one if another can be readily obtained without cost or effort. Frictions in the labor market, however, mean that employers may have some degree of power in setting wages, and once employers exercise some choice in determining the wages they pay, it is no longer clear that increases in the minimum wage will always reduce employment. Instead of reducing the employment of low-wage workers, a minimum wage hike can make it easier for employers to recruit, possibly raising employment.

In an imperfectly competitive labor market, then, the direction of the change in employment due to a minimum wage increase is not uniformly negative, although at some point sufficiently high minimum wages will begin to reduce employment. Most of the evidence over the past 15 years does not suggest the United States has crossed that threshold. Manning adds that the current policy experiments in U.S. cities and states, as well as international increases, will shed more light both on the effects of the minimum wage and how labor markets work.