One of the flagship policies in the Biden presidential platform is a $15 national minimum wage - indeed, many see it as one of the wins of the Sanders campaign in shifting the Democratic consensus. But increasing a specific wage is not a policy goal - and so the more pertinent question is whether this is an effective way of improving living standards?
What is the Economics 101 of a minimum wage?
The 101 explanation of a minimum wage involves thinking about it as a price floor. In a simple diagram of the supply and demand of labour, a price floor higher than the equilibrium price will mean that at that price, the quantity demanded is less than the quantity people are willing to supply. In other words, there is unemployment. Suppose a $15 minimum wage - functionally what is happening is that for those who are willing to work for less than $15 an hour and who companies are only willing pay less than $15 an hour, they are being deprived of a job.
So does that mean we should have no minimum wage?
Of course not - and the reason this blog post doesn’t end here is because this is a deeply simplified model. It assumes a perfectly competitive market where firms have no market power and where there are no search costs associated with looking for a job. In practice, labour markets are much more accurately described by a model that is not perfectly competitive1.
What happens when markets aren’t perfectly competitive?
It may well be the case that the labour market is monopsonistic - that is, because one buyer of labour dominates the market, they have the market power to pay workers below the wages they would be paid in a perfectly competitive market. In this case, it would make sense that an appropriate minimum wage could correct this market failure.
A monopsonistic labour market is created when there are few employers available to you. This could be because there just aren’t that many firms hiring in an area or sector, because there are large search costs associated with finding a new job2 or because the jobs available are sufficiently far away that the travel costs are non-trivial3.
For example, a study has looked at over 8,000 labour markets in the USA, each defined by their location and sector4. It found that the average labour market was highly concentrated - that is, there were only a few firms that dominated the hiring in that market. Furthermore, it found that going from the first to third quartile in concentration is associated with a 17% decline in wages, corroborating the idea that monopsonistic labour markets exist and push down wages. Around 23% of the national workforce in the USA works in highly or moderately concentrated labour markets5, with research on wage stagnation confirms that employer concentration has a significant effect on depressing wages6. In fact, wages are suppressed by 17% on average due to monopsony power7. To be clear, this is not an effect that is unique to the United States - monopsony power has been observed in labour markets across the globe, such as in Mexico8.
The consequence is that a minimum wage might not cause any disemployment effects, since the wages being paid are currently depressed below what a competitive equilibrium would produce due to monopsony power.
What does the empirical work on minimum wages say?
The groundbreaking empirical work on minimum wage that began a debate around its disemployment effects began in 1994, when David Card and Alan Krueger found no significant effects from a change in the minimum wage by comparing the fast food industry in New Jersey and Pennsylvania9. Neumark and Wascher re-evaluated the evidence, finding a significant disemployment effect10, leading to a response11 over methodology that was eventually mediated by a recognition that these differing effects applied to different types of firms12.
What has happened since this seminal debate?
One of the workhorse studies in recent times replicated Card and Krueger’s technique of comparing the borders between states where one state increased the minimum wage and the other didn’t, applying it to every instance where this occurred from 1990 to 2006 - they found no disemployment effect on low-wage work13. Another study looks at the number of jobs just above and below the minimum wage, both before and after an increase in the minimum wage from 1979 to 2016 - likewise, they show that the employment effect is basically non-existent14. In both cases, they find that there are increased wages for low wage workers.
However, what these studies consistently find is that there is a small but statistically significant disemployment effect for the tradable goods sector - this could be because these are more price elastic goods where passing the costs onto the consumers is difficult.
What are the welfare implications of a higher minimum wage?
There is a relative consensus that a small to moderate rise in the minimum wage has statistically negligible employment effects15. That doesn’t mean it is a good anti-poverty policy. One critique is that minimum wage workers tend to be young - around 40% of those paid the minimum wage or less are under 2516. And because 62% of those workers are still in education, they tend not to be the main breadwinners, but rather living in middle-class households17.
However, empirical evidence suggests that even if imperfect targeted, the minimum wage remains useful towards increasing family incomes at the bottom end - with clear positive effects on family incomes below the 20th percentile. [^H]: Dube 2017
This may be because the minimum wage increases can lead to wage increases for people currently earning above the minimum wage in order to maintain the compensating wage differential. Suppose a cashier earns $7.5 an hour and a mechanic with a less comfortable working environment and a higher risk of bodily harm earns $10 an hour - if the minimum wage rises to $10 an hour, many mechanics may become cashiers if their wages don’t change, since they are no longer being compensated for the less pleasant working conditions. Consequently, a rise in the minimum wage is associated with wage rises for workers up to the 25th percentile, or up to those earning 175% of the minimum wage18.
Furthermore, recognise that even if there were no monopsony power and there may be unemployment effects, that must be put into context. We are concerned with people’s overall earnings- it is seems plausible that even if people spend less time employed, they earn more in total due to the higher wages. Of course, the welfare implications of this depends on how those periods of unemployment are distributed - is unemployment concentrated to a few people who lose their job for a substantial period of time or does it mean that everyone spends a few more weeks off the job? Because of the high turnover rate in low wage jobs, the reality is that the minimum wage, even if it did cause unemployment in a non-monopsonistic market, is simple causing everyone to be unemployed for a week or two more in between jobs without reducing their annual earnings, rather than causing some people to be permanently unemployed19.
What are the policy implications?
It is clear that even without monopsony, the low wage labour market is likely to structured in a way that the loss of income due to unemployment from a small rise in the minimum wage is relatively limited. Because there is evidence of monopsony power, the disemployment effect observed up till now is negligible, while the benefits accrue not just to those paid the minimum wage but may also ripple to some other low-wage workers.
In practice, the benefits of the minimum wage depend on the monopsony power within a specific labour market. As such, a minimum wage should vary based on regional conditions. For example, one proposal involves setting the minimum wage to half the median wage in a particular locality - this seems feasible, given that the minimum-to-median ratio stood at 48% in the OECD from 1960 to 201220. It is also suggested that this should be indexed to regional inflation indicators in order to keep up with the costs of living.
Ultimately, there is a lot we don’t really know about the minimum wage - labour markets are messy and confusing. What we do know is that a $15 national minimum wage would be a large increase, one beyond the size described by the empirical work done so far and beyond the half-of-the-median metric established above - as such, the disemployment effects are likely going to be noticeable. Its lack of regional disaggregation is also a problem in terms of minimising the downside. And perhaps most importantly, it is unclear why we want one - although the minimum wage does benefit low-wage workers, it also benefits lots of middle-class teenagers and is an incredibly blunt instrument. If we want an coherent anti-poverty strategy, we can do better than a $15 national minimum wage.