Raising the Floor or Bottoming Out?
Raising the Floor or Bottoming Out?
Does the minimum wage reduce employment?
So what are the effects of increasing minimum wages? Any Econ 101 student can tell you the answer: The higher wage reduces the quantity of labor demanded, and hence leads to unemployment. This theoretical prediction has, however, been hard to confirm with actual data.
What are the economic effects of raising the minimum wage? The minimum wage, as the reader might know, is a legal floor on how much employees must be paid1. In the US, it has been 7.25/hour for over a decade, and there’s been discussion of raising it to anywhere between 9 and 15 dollars. The biggest objection proposed is that it would reduce employment and/or raise prices. Is that true?
Worth pointing out I’m primarily going to focus on the impact on employment, and exclusively on the impact on economic variables - so studies on, say, suicide rates, teen pregnancy, or whatever get set aside because I’m not familiar with those topics.
Any Econ 101 student can tell you the answer
The simplest model for what the minimum wage does to employment is perfect competition. If labor demand and labor supply determine a given wage for a given job, and the minimum wage is set above that level, then employment would have been set lower than otherwise. Right?
Wrong. Worth pointing out that this model predicts the demand for labor as measured in hours, resulting in the fact that the clearest adjustment channel is lower total hours, either resulting in lower employment in general (at the “extensive” margin) or fewer hours worked (at the “intensive” margin). Additionally, each worker has a level of productivity they contribute, so that could play a big role as well. Finally, there’s many other decisions firms make at the margin, even without deviating from the perfect competition set up (more on this in a minute). Some of these are non-employment adjustment channels are:
Increased demand: people who get make more also spend more, so the increased spending might ultimately crowd out the negative effects.2
Higher prices: costs go up, prices go up. Simple as.
Lower profits: companies just suck up the cost, and make less money.
Reductions in hours, benefits, or training: employers could reduce the number of hours they give their workers, the amount of training they provide, or cheap out on “fringe benefits” like discounts, parking, or health insurance.
Changes in composition: if forced to pay more than the average productivity of their workers, companies will hire workers with more skills (for example, more experience) whose productivity matches the new wage.
Efficiency wages: an increase in wages could boost employee morale and make them work harder, increasing output and cancelling out the cost.
Wage compression: instead of changing employment at the bottom of the payscale, companies just don’t raise wages for the rest instead.
Reduced turnover: hiring, finding, and training workers is a costly process. If the minimum wage is raised, workers might stay at any given position longer, and offset the increased “turnover” costs for their employer.
There is one more possible explanation of why raising the minimum wage might not decrease employment that has nothing to do with any of the above: monopsony power. Just like a monopoly occurs when a single company is the only supplier, monopsony is when there is a sole “demander” in a market. When a firm has monopsony power, it pushes prices down at the expense of suppliers. A non-labor example of this (there aren’t many) is the WIC program and baby formula.
Under a price floor, however, prices are set closer than the equilibrium level and therefore increase supply, since it incentivizes production. This effect goes up until the floor equals the perfectly competitive price, since any increase above would have negative effects on demand. If the labor market were more accurately described as monopsonistic rather than competitive, then a minimum wage increase that does not exceed the perfectly competitive level would at least not raise unemployment, while any increase that does exceed the competitive level would reduce employment.
Hard to confirm with actual data
What does the data say? Ohhhhh boy. This issue is super thorny and nobody really has any clear answer, which is always the most satisfying way to answer any important question. One of the big issues is constructing a counterfactual - i.e. which scenario where the minimum wage increase does not happen, and for whom, particularly since statistical techniques use assumptions that might or might not hold. The reason why is also the most fun one possible, complex methodological issues: do you focus on employment levels or job growth? Do you focus on certain groups of workers (and which ones?) or the whole workforce? What kind of data should you use? These are all complicated and ambiguous questions with major implications for results.
The best place to start3 would be Card & Krueger (1994), a very important paper. In it, the authors look at restaurants located in the border between New Jersey and Pennsylvania to verify whether an increase in the minimum wage in New Jersey had any impact in employment. Utilizing innovative techniques (for the early 90s), the authors found simultaneously no impact on employment, no impact on benefits, and very little impact on prices. Using “newer” and higher quality data, the authors didn’t find evidence that contradicted their original findings regarding the same event.
In response, many economists were very mad at Card and Krueger, for a variety of reasons, and many were very upset at the general understanding of the paper. A notable example was Nobel Laureate James Buchanan, who had this to say about a proposed change in minimum wage legislation that cited the paper’s results:
Just as no physicist would claim that “water runs uphill,” no self-respecting economist would claim that increases in the minimum wage increase employment4. Such a claim, if seriously advanced, becomes equivalent to a denial that there is even minimal scientific content in economics, and that, in consequence, economists can do nothing but write as advocates for ideological interests.
Note that Buchanan could have been right if the posterior evidence would have been all against Card and Krueger’s findings - one false result is not enough to abandon a theory. On the contrary, the “teachings of two centuries” should be abandoned if the results came the other way. So, what has the evidence said about the minimum wage?
The meta of the matter
Let’s begin by focusing on the other channels. Does the evidence say anything about how the minimum wage would affect non-employment changes in outcomes?
Higher prices: low pass-through in the most affected sectors (like fast food).
Lower profits: seems to be real, based on limited evidence.
Reductions in hours, benefits, or training: there is not much evidence that hours worked go down, and there is not a lot of studies on benefits or training, but benefits are probably unaffected and training is a toss-up.
Changes in composition: extremely controversial, with some finding it disadvantages and others that it advantages young/low-skill workers.
Efficiency wages: some evidence in favor of employers demanding better performance, but not many tests of efficency wages have been conducted.
Wage compression: some evidence in favor of employers delaying wages or raises, and some evidence that the minimum wage reduces income inequality.
Reduced turnover: some evidence in favor of lower quits and lower hires.
Monopsony: monopsony market power seems to be real, and the more market power, seemingly fewer employment losses from a minimum wage hike.
But, truthfully, barely anyone wants to know if a higher minimum wage reduces spending on employee training. The question everyone wants to know the answer to is if it reduces employment, and more importantly, by how much.
Is there anything that “everyone” agrees on? Basically, that small increases have small effects, and large increases are unpredictable because they haven’t happened very often - particularly since most raises happen at a certain bound. Most studies report findings in the -0.25 to 0 range (aka “small” and negative), although it’s relevant to point out there is a publication bias that significantly reduces the range of estimates. The median outcomes from more precise studies turn out in the -0.15 to -0.05 range - that is, for every 10% the minimum wage increases, the number of people employed goes down ~1%. Some studies for sector-specific dynamics, like this one on the food sector after Minneapolis’ minimum wage hike, find bigger estimates (0.8).
But this is all looking at employment levels - what about employment growth? A disturbing possibility is that the minimum wage does no harm to workers who are already employed, but slows down hiring, harming those entering the labor market. Using high quality data, and looking at low skill, young, and low wage workers in particular, the results on job growth seem to be negative and larger than those for levels. This also might explain some of the disagreement between economists: looking at employment levels and not growth can yield directions in either consequences even when the results on employment growth are unanimously in one of them.
A common way of evaluating the consensus view in science are what are called “literature reviews”, which survey lots of papers and report on their overall findings. What do minimum wage lit reviews ultimately say? They’re somewhat all over the place, just like the literature itself. So let’s dive in.
The place to start is Dube (2019), a report commissioned by the UK Government. In it, using evidence from not just the US but also other developed nations, the evidence seems to point to the minimum wage having a frequently small, often statistically insignificant, but negative effect on overall employment. Studies looking at low wage workers specifically have consistently found no net disemployment effects. However, there is also some evidence that incumbent low-skilled workers face worse job market prospects further down the line, due to lower chances of finding another job, a higher probability of working for no pay, and less accumulated experience due to both. The effect is roughly in the -0.12 to -0.05 range, low but relevant.
However, literature reviews by David Neumark and coauthors in 2007, 2014, and 2022 find opposing results. Neumark’s objections to the results found by previous studies rely on differences over methodological choices, especially accounting for who the comparison group is (workers in other places or workers in the same place who made just enough to not be affected), how to account for differences between locations, the different methodologies of various studies aggregated in lit reviews, and most importantly, the fact that not all study results are the same. Focusing on Neumark & Shirley, which excludes non-significant results and only focuses on the preferred result (i.e. most important one) by the author of the original paper, or Neumark and Shirley otherwise. In their paper, they find a negative in the -0.1 to -0.17 range, which makes the whole thing kind of cute. Though it’s worth noting that excluding statistically insignificant papers is not uncontroversial, since “statistically insignificant” basically just means “nothing to see”, and might bias the sample towards bigger results.
Lab-or economics
Another way to look at this is through “experiments”, i.e. big changes in the minimum wage that for some reason or another can be compared to another group very accurately.
First, the US Congress set the same minimum wage for the mainland US and Puerto Rico, a significantly poorer associated territory (not getting into that one) of the United States, in 19835. Coverage of the minimum wage was around two thirds of the workforce, and the increase resulted in a decrease in employment, although this is not true for all industries, since employment increased in a few of them - at least providing some evidence that a large hike can have adverse consequences.
Secondly, in many cases states have changed their minimum wages at different times, and more recently cities have done so. Since many areas on the border between two states, or between a city proper and its metro areas, have somewhat similar economies, then it stands to reason that the effects of the minimum wage would be the main factor explaining the (lack of a) difference in employment levels or employment growth. For both counties bordering two states, and cities and their surrounding metro areas, there is very little evidence of disemployment effects, even after accounting for the fact that which states raise minimum wages is not exogenous to the average wage level.
Lastly, there have been a few randomized controlled trials (RCTs) in the minimum wage. RCTs consist of separating a sample into two groups, one that gets an intervention performed (treatment group) and one that doesn’t (control group), and then outcomes are compared. Because of their pedigree from the natural sciences, RCTs are considered the gold standard in economics research - they are thought to very high quality evidence since they occur in a pristine environment where it is clear that the only difference between the control and treatment group, when properly designed, is the intervention itself. In 2017, an RCT on an online job market randomly assigned hourly minimum wages; in response, firms that had the wages assigned reduced hiring slightly, slashed the number of hours worked, and hired more productive workers. While i.e. it’s crystal clear that the change in employment was driven by changes in the minimum wage alone, because firms were generally considered equal, there is no reason to expect that the results be applicable to other contexts, since this particular online labor market is so unique: very flexible, high turnover, low training required, very short contracts, etc.
Conclusion
Instead of summarizing what already is a summary of a mountain of very complicated research, I’d like to focus on something else entirely: how can the research lead policy regarding the minimum wage. As stated above, while the impact of a small increase are well within the range normally considered within research, very large increases are not, and therefore caution must be key.
Both Dube (2019) and Dube (2014) find that the highest minimum wage without a big impact on employment seems to be between 50% to 60% of the median wage of any given area. This makes intuitive sense: if the productivity of labor in each low-skill sector is constant, then the wage corresponding to them should scale up with cost of living, implying that the right minimum wage for San Francisco would be much higher than of some city in Alabama. Therefore, setting a minimum wage that is too high for low-cost areas would yield bigger disemployment effects, and be in generally uncharted territory in “in between” areas. This is a blank area in the literature because very few places have tried setting the minimum wage above the 60% mark, so the “out of sample” problems could be severe: maybe nothing happens, or maybe you get a much higher elasticity. As a result, raising the minimum wage “too much” for low wage areas could have serious effects. A caveat is that smaller cities usually also have more monopsony power, but it might also be true that a minimum wage increase in excess of the equilibrium competitive wage might also have larger (negative) effects, as the monopsonist holds all the cards in the labor market.
All in all, what’s certain is that the minimum wage could be a good tool to raise incomes at the bottom of the distribution, but if it is handled carelessly, it might prove detrimental to welfare in excess of its benefits. Mucking around with relative prices isn’t really the best of ideas, after all, and might prove highly distortionary. The takeaway is that policymakers should err on the side of safety and stay in the “50-60 range” here, and rely on other methods such as direct transfers to reduce poverty or raise the poor’s incomes if that doesn’t prove effective.
Sources
(I’m not linking specific papers for each claim, just for general lit stuff and also some really important ones. Just click on the hyperlinks on each one)
Evan Soltas, “Everything you need to know about the minimum wage”, Vox, 2015
Noah Smith, “Why $15 minimum wage is pretty safe”, 2021
Noah Smith, “The minimum wage pushback”, 2021
The Econ 101 model - and more!
Schmitt (2013), “Why Does the Minimum Wage Have No Discernible Effect on Employment?”
Brian Albrecht, “How can price theory help us navigate the minimum wage debate?”
Data 101
The data
Adam Ozimek, “Why Do Economists Disagree So Much About The Minimum Wage?”, Forbes Magazine, 2014
Dube (2019), “Impacts of minimum wages: review of the international evidence”
Wolfson & Belman (2015), “15 Years of Research on U.S. Employment and the Minimum Wage”
David Neumark, “Employment effects of minimum wages”, 2018, IZA
Meer & West (2015), “Effects of the Minimum Wage on Employment Dynamics”
Conclusions
Arindrajit Dube (2014), “Designing Thoughtful Minimum Wage Policy at the State and Local Levels”
CBO (2019), “The Effects on Employment and Family Income of Increasing the Federal Minimum Wage”
The US, in particular, has a bunch of really weird laws around the minimum wage being different for tipped workers and for disabled people, and a bunch of countries have different minimum wages by establishment size and/or sector. Almost all of these are bad ideas with bad, confusing outcomes.
I’m not especially sure about this channel, and I wouldn’t particularly rely on it either. Massive issues with composition of spending, and direction depends on overall impact on employment anyways. Plus, it’s not really clear why, say, fast food workers would spend their money on fast food specifically anyways.
There were older papers about minimum wages and employment, but like almost all pre-Credibility Revolution econometrics, they had too much con and too little metrics.
Worth pointing out that Card & Krueger (1994) coincidentally find a small increase but didn’t actually attribute it to raising the minimum wage, and there isn’t much evidence it actually does anyways.
To be fair, the Puerto Rico parts of Fair Labor Standards Act were a kludgy mess that set various minimum wages with a very big brained logic behind them.
Hi Maia, what happen to your twitter account? Are you getting back?