The economics profession stopped having the argument the public thinks it is having. The question is no longer whether a wage floor destroys jobs in the abstract. It is at what level, relative to local wages, a floor starts to bite, and how much wage-setting power employers hold in the meantime. On that question the federal floor of $7.25 an hour is not a contested case. It sits so far below any threshold under serious dispute that defending it on employment grounds means defending a number the empirical debate left behind years ago. That is my position, and here is the reasoning.
The model that produced the old consensus
The textbook case against a wage floor comes from a competitive labor market. If many employers compete for workers, wages already equal the value of what the marginal worker produces. Force the price above that and some workers become unprofitable to employ. Employment falls. The prediction is clean, and for decades it functioned as settled.
The model rests on an assumption that turns out to be doing enormous work. It assumes employers are price takers, facing a supply of labor so elastic that raising pay by a cent floods them with applicants and cutting it by a cent empties the building.
Anyone who has looked at an actual low-wage labor market recognizes that this is not the world. Workers face search costs, transport constraints, scheduling constraints and childcare constraints. A worker who cannot get to the other job across town is not choosing between employers on price.
What the natural experiments changed
The turn came from design, not theory. In the early 1990s David Card and Alan Krueger compared fast-food employment across the New Jersey and Pennsylvania border after New Jersey raised its state floor, using the state line as the boundary between treated and untreated firms in what was otherwise one labor market. They did not find the employment loss the competitive model predicted.
The result was contested hard, and correctly so. David Neumark and William Wascher challenged the data collection and reached different conclusions using payroll records. That exchange ran for years and produced better methods on both sides, which is what a functioning empirical dispute looks like.
What followed mattered more than who won any single round. The border-comparison design was generalized, most prominently by Arindrajit Dube, T. William Lester and Michael Reich, who compared every pair of contiguous US counties straddling a state line with different minimum wages. The approach controls for regional shocks that had contaminated earlier national comparisons.
The accumulated body of work did not establish that floors never cost jobs. It established that modest increases from low bases produce employment effects too small to detect reliably, which is a different and more useful finding.
Monopsony is the mechanism
The theory caught up by taking employer wage-setting power seriously. In a monopsony, a single buyer or a few buyers of labor face an upward-sloping supply curve and set wages below the value of what workers produce, because hiring one more worker means paying everyone more.
Under those conditions a wage floor set between the suppressed wage and the competitive wage raises pay and can raise employment, because it removes the employer’s incentive to restrict hiring. This is not an exotic result. It is in the textbooks alongside the competitive case, and it has been for decades. It simply never had the empirical support it now has.
Modern labor economics treats employer wage-setting power as a matter of degree present in most markets rather than a special case. Concentration in local hiring, non-compete agreements, information asymmetry and the frictions above all contribute. Once that is granted, the sign of the employment effect becomes an empirical question about a specific market at a specific level, which is exactly where the research went.
Why $7.25 is not the interesting number
According to the U.S. Department of Labor, the federal minimum wage has been $7.25 an hour since 2009. Seventeen years without an increase. Full-time at that rate is $15,080 a year.
Set that against the U.S. Census Bureau’s median household income of roughly $80,000 in 2023. The federal floor is under a fifth of it. The serious disagreement in the literature concerns floors set at meaningful fractions of local median wages, where the ratio determines whether the increase bites. A floor at this level is nowhere near that zone in any state.
The MIT Living Wage Calculator estimates what basic costs actually require county by county, and its figures run well above the federal rate across the counties it covers. Check your own county rather than take my word for it. The gap is not marginal.
Which is why the employment argument against the current federal floor is analytically empty. It invokes a threshold effect at a level far below where any contested threshold sits.
The argument that deserves more weight
The strongest case against a uniform national increase is not disemployment. It is geographic variation. A floor that is modest relative to median wages in a high-cost metro may be high relative to median wages in a low-cost rural county, and the ratio is what determines the effect. That objection is real and the research supports taking it seriously.
It argues for indexing and regional calibration, not for a frozen national number. A floor pegged to a share of local median wages and adjusted automatically would address the variation directly, while a nominal figure fixed in 2009 addresses nothing and erodes every year inflation runs positive.
The second serious objection is that a wage floor is a blunt instrument for household income, since it targets a wage rather than a household and interacts with benefit phase-outs. Also true, also an argument about instrument design rather than about whether the floor should move.
Where I think the argument actually sits
Here is my position stated plainly. The empirical literature has narrowed the disemployment claim to a question about ratios, the current federal floor is not close to the contested ratio anywhere, and the remaining serious objections concern indexing and geographic calibration rather than the existence or level of a floor.
That is a narrower claim than either side usually makes. It does not say floors are costless, and it does not say any particular national number is correct. It says the argument that has dominated public discussion for forty years is not the argument the evidence is actually about.
The bigger constraint
A wage floor also cannot carry the weight the debate assigns it. Housing, healthcare, childcare and education all rose faster than wages, and no plausible floor closes those gaps alone. National Association of Realtors and Census data put median home sale prices at roughly $400,000 to $420,000 in 2024, about five times median household income against about three times in the 1980s. A wage adjustment does not touch a price ratio like that.
Fight For A Living Wage, a nonpartisan grassroots 501(c)(3), makes the case for raising the wage floor while treating it as one symptom of an affordability problem spanning every major household cost rather than the whole diagnosis. That distinction is the right one. The floor is worth arguing about on its own terms, and a reader who wins that argument has still only addressed part of the arithmetic.
What should end is the reflexive invocation of a model whose central assumption the evidence stopped supporting. Argue about indexing. Argue about regional calibration. Argue about which instrument reaches households rather than wages. Those arguments are live, and they are more honest than the one currently standing in for them.
