Some New Problems With the Minimum Wage

Why it won't redistribute as effectively as we would like

I am not a very big fan of the minimum wage. No economist should have a positive instinctive reaction to a price control – they tend to lead to an inefficient allocation of resources. In a competitive market, they must necessarily lower welfare. Nevertheless, I can happily acknowledge that it is entirely plausible for a minimum wage to increase efficiency by correcting market failures.

The market failure is the same as with a monopoly, just in reverse. If a company has to pay everyone the same wage, and more people are willing to work for higher wages (there is an upward sloping supply curve) then there is a disconnect between wage and marginal product, called a markdown. It would be efficient for the firm to hire one more work at a price equal to what they produce, but in order to do that they must raise wages for everyone else. A minimum wage corrects this by simply mandating that the firm pay that efficient wage anyway.

The empirical literature on unemployment is well-trodden, starting with Card and Krueger (1994), and then going back and forth with different changes and different specifications and so on. I frankly have no interest in hacking through the weeds of whose control groups are correct; interested readers can consult Neumark (2018) and then work backwards through their citations, noting that Neumark has consistently been on the side finding negative effects on employment. I will also point out that our measurement will be biased against disemployment if firms can adjust compensation in more than just wages, such as by making the conditions of the job worse – see Clemens (2021) for more on this point.

In any event it is universally agreed that the short run employment effects are small. I would instead like to point out some undesirable implications of the minimum wage, once you grant the complications in the economic story which lead to it being efficient. First, the optimal minimum wage for efficiency purposes is quite different from the minimum wage one would choose if they were trying to maximize a social welfare function where poorer people have higher marginal utility. Second, the category of “low-income earners” does not map cleanly onto “poor people”; and since consumption patterns differ between rich and poor people it is conceivable for the minimum wage to lower the welfare of poor people even as it increases their wages. Third, under plausible theories of why firms have labor power, more productive workers will gain the most from a minimum wage, while the lowest wage workers will gain the least.

The first point comes from Berger, Herkenhoff, and Mongey (2025), who abstract away from measurement issues by giving ranges for different estimates of the effects of minimum wages. The federal minimum wage that maximizes productive efficiency is about $8 as of 2022; considering the differing marginal utilities of high and low earners increases this to around $15. I point this out not as evidence that the minimum wage is bad, per se, but that the arguments for redistribution and efficiency are mutually contradictory. If it is being advocated for to redistribute, then it must be compared to the efficiency and efficacy of other programs which redistribute, a comparison I believe it suffers by.

The second comes from a 2015 paper by Thomas MaCurdy. Take an archetypical minimum wage job – fast food employment. Casual observation will show that while there are some people doing it permanently, there are also many teenagers working a part-time job in high school. Those teenagers are part of a higher earning household, which makes them not poor. It’s the difference between being broke and poor. Thus, a minimum wage increase will in part redistribute some income to rich households.

But that’s not all. The preferences of rich and poor households differ, which is called “non-homotheticity”. (Homothetic preferences are when people consume the same proportions of goods as they get richer). The places where low-income workers work are also disproportionately the places which poor households buy from. When a minimum wage hike is partially passed through to consumers, it most affects poor households. Putting the two together, only a quarter of poor households actually benefit from a minimum wage hike, and the policy only slightly redistributes to actually poor households.

The third argument is due to Bils, Kaymak, and Wu (2025), in a paper out just this week. Let’s think about why firms have market power – after all, there are many, many firms, so they hardly have a monopsony. We can say that perhaps different workers are particularly suited for different jobs, for arbitrary reasons – maybe they get along well with their boss, or have become familiar with the routine at work, or alternatively there could be sets of firms which specialize in high and low skill workers respectively. Assume that absolute advantage coincides with comparative advantage. Thus, the larger the surplus they would get from working at a firm, the larger the markdown. Imposing a minimum wage benefits those with higher markdowns more, or in other words redistributes from the lowest productivity workers at a firm to the higher productivity ones. Applying this to data from Brazil, this substantially attenuates the gains from the minimum wage.

What is clear is that the minimum wage is only tenable as a method of redistribution when there is no better alternative available. Hurst, Kehoe, Pastorino, and Winberry (2021) find that expanding the Earned Income Tax Credit (EITC) dominates minimum wage hikes for worker welfare, especially in the long run. This accords with my intuition – in the long run, basically everything is substitutable – although I am a little concerned the model kind of assumes its results. (The key parameter is how substitutable different kinds of workers by education are. A low substitutability would mean that firms can’t change how they produce outputs, and have to pay more to workers forever. The estimates for substitutability are from 2001 and 1994, which – well, empirical economics has gotten a lot better since then!). Still, it is difficult for me to believe that minimum wage hikes beyond what improves efficiency are justified if it is at all politically feasible to simply give people money.