What Can We Conclude from the Evidence on Minimum Wages and Employment? - Recent progress

Conflicting views on the benefits of minimum wage and the importance of comparative research

The debate over minimum wage is filled with conflicting evidence and differing opinions. Media statements and economists alike have presented arguments both for and against the idea. In particular, two comparable statements highlight the disagreement, with one suggesting that an increase to a $15 minimum wage would not harm jobs in New York, while another predicts a significant loss of jobs in California. To study the effects of minimum wage policies, researchers often compare changes in different regions, such as countries, states, or prefectures. The United States, with its diverse range of minimum wages, provides a valuable resource for understanding the impact of these policies on various regions.

In collaboration with Bill Wascher from the Federal Reserve Board, I conducted an extensive review of minimum wage research, finding that most studies suggest negative employment effects. Labor economists commonly express these findings using elasticity, with many studies reporting a range of minus 0.1 to minus 0.2. This implies that a 10% increase in the minimum wage could result in a 1%-2% reduction in employment. However, there has been a sustained critique arguing that relying solely on regional differences where the minimum wage changed might yield inaccurate results due to correlated economic shifts. These studies, viewed as simulations of experiments, led to two key assertions. First, when adjusting for these issues, the studies suggest the minimum wage may not significantly impact employment. Second, proponents argue that the analysis should focus on geographically close areas.

A notable study by David Card and Alan Krueger in the U.S. focused on the impact of minimum wage changes in New Jersey and Pennsylvania, specifically near the state border. Similarly, in developing countries, studies examine minimum wage changes between provinces or states, often involving major cities. One influential paper by Dube, Lester, and Reich (hereinafter “DLR”) explores counties on opposite sides of state borders, highlighting the variation in size and raising questions about the validity of certain studies.

The ongoing debate among economists is puzzling in that different conclusions are derived based on the same evidence. For example, a letter signed by renowned economists, including Nobel Prize winners, states that a decade of rigorous academic research suggests minimal or no negative effects on low-wage worker employment, challenging the notion that minimum wages lead to reduced employment.
In contrast, Card and Krueger’s statement on the reissuance of their seminal book ascertains that the subsequent literature on the topic has found a balance between positive and negative effects, so the evidence does not support a claim of a significant effect.
Finally, some economists have concluded that the evidence generally supports the claim that job loss is the natural result of a minimum wage increase.

Effects of minimum wages on employment: Evidence and findings

In our own survey of the literature, we captured findings from 70 published papers and determined the actual favored estimates from the authors themselves and their rationalization when possible or from deep analysis of the papers. In our summary of those findings, a significant majority of papers found a negative impact on employment from minimum wage increases.
This contradicts the opinions mentioned earlier that minimum wage literature suggests no effect or positive effects on employment. Interestingly, these findings remain consistent regardless of when the study was conducted.

The neoclassical perspective posits that minimum wages have the potential to decrease employment, particularly for individuals earning the lowest wages. While the least skilled workers would experience a wage increase if they manage to retain their jobs, they are also the ones who are most affected by the government’s intervention, as their wages become more costly, thus increasing their vulnerability to job loss. Extensive research focusing on less educated workers reveals that the majority of estimates lean towards the negative side.

A limited set of studies that directly examine workers affected by the minimum wage before it increases reveal that minimum wage has a greater effect on less educated individuals compared to those with higher education. This is because individuals with higher education are less likely to be minimum wage workers, and their job is less affected by the minimum wage increase. Studies show that the employment of ‘directly affected’ minimum wage workers is reduced the most due to the increase in minimum wage.

Examining low-wage industries and low-skilled workers

In a sub-set of studies that focus on low-wage industries, such as retail and restaurants, rather than low-skilled workers, the industries, which are the two with the largest share of low-wage workers in the U.S. (and likely Japan), on average do not really produce a large negative estimate of employment losses. However, industry studies can mask who is actually employed, as the lowest-skilled/lowest productivity workers are in fact replaced with slightly fewer numbers of higher-skilled workers, who are receiving slightly higher minimum wages, meaning that there is still significant job loss as the lowest-skill workers are in fact replaced.

There is in fact a conflict in the research literature between the panel data estimators approach and the geographically close controls approach, which is the approach that attempts to compare areas that have similarity due to geographic proximity, assuming that the geographically similar areas will be subject to the same economic shocks. This seems to be a positive development, but if that assumption is not correct, then this approach is not reasonable.

Exploring the impact of minimum wages on employment: Analysis of the close-control approach

There are around five or six main studies in the larger literature that use a close-control strategy to study minimum wages. The first study is the DLR study, which focuses on paired counties on opposite sides of the state border. The other studies generally agree that minimum wages do not reduce employment, regardless of the focus on restaurant workers or teenagers.

The Liu, et al. study is different as it incorporates the Bureau of Economic Analysis’ Economic Areas into the close-control approach, studying areas that, in addition to being geographically similar, are in the same economic area with similar industrial composition. They get a minus 0.17 elasticity for teenagers, which is similar to the previous, existing studies. This suggests that the study should consider what’s on either side of the border in addition to the geographical proximity.

The title of the paper, “What’s Across the Border?”, emphasizes the importance of exploring the area’s economic composition before deciding on the data. This approach helps to understand the impact of minimum wages on employment and the economic conditions in different regions.

To increase our understanding of whether the finding that increased minimum wage does not increase employment found in the close-control studies is indeed correct or not, it is also important to understand whether these changes are linked to other economic changes. Four studies take the approach of adding further detail, including one technical approach, one that seeks exogenous variation of minimum wages, and one that adds another level of control, focusing on the lowest wage workers but in comparison to other low-wage workers whose wages are nonetheless higher than minimum wage. All four of the studies that account for other economic effects found quite strong negative effects of increases in the minimum wage.

Minimum wage increases have been found to have significant negative effects on employment. The Liu, et al. study uses regions with integrated economies as an ideal setting for a close controls approach. This approach considers economic shocks and it should satisfy critics of other minimum wage studies.

Effectiveness of cross-border counties and commuting zones for border control and minimum wage impact on employment

In addition to using cross-border counties with a single economic zone to account for confounding factors, commuting zones are regions within the U.S. where almost everyone who lives in that area also works, meaning that there is meaningful economic linkage. Two authors of the DLR paper actually suggest using commuting zones as a way to control for shocks. We re-evaluate the DLR evidence by comparing commuting zones to the cross-border counties used in the DLR study, and found that in fact, the zones used in DLR also did not represent commuting zones, so their findings were not as resilient as some may imply.

Comparison of regression tables focusing on elasticities and their significance shows that when the DLR study compared counties across state borders, they did not find significant negative numbers, but when we limited the data to cross-border regions with a single commuting zone, we again found large negative effects of increased minimum wages.
The cross-border county design used by DLR may give misleading estimates. It is only an assumption that the economic shocks are similar on both sides of the border, which may not be true. While this argument may be intuitively appealing, it may not be the most accurate representation of the situation.

Economic shocks can be more similar in states close to each other, such as Florida and Georgia, than in states far away like Florida and California. However, minimum wage differences between close states are typically smaller. This variation can be high in some regions and low in others. The importance lies in the size of the shocks relative to the minimum wage differences. The idea of using cross-border counties to control for shocks could potentially make the bias worse or even lead to a positive bias if not fully controlled for.

To determine the effect of incorporating commuting zones into the analysis, we compared the DLR’s findings with a smaller data set that only includes the counties in their study that were in the same commuting zone, and finally with another larger data set comprised of commuting zones across the nation. Our findings were clear that when the commuting zones are incorporated, the estimates of employment effects are clearly negative, meaning that the DLR is incorrect. Increasing minimum wage does in fact have negative effects on employment for lowest-wage workers.

In conclusion, I want to emphasize that raising the minimum wage is not a bad policy, but rather that there are trade-offs. Raising minimum wage may have benefits that the policy may achieve and it is important to consider the potential benefits and costs of government regulations if the benefit is greater than the cost. For example, enacting climate change policies may result in job losses for oil workers, but the benefits may outweigh the costs. However, policymakers must weigh the trade-offs, consider who gains and benefits, and weigh the importance of the social goal. Forcing higher wages has consequences that may or may not deter policymakers from attempting to achieve other goals.

Q&A

Q:
The Japanese labor economy white paper estimates that a 1% pay increase for all workers leads to a ¥2.2 trillion boost in productivity and ¥0.5 trillion in employee compensation. What are your thoughts on this analysis? Is boosting minimum wage likely to boost inflation by increasing consumer demand in low-paid workers? Raising the minimum wage may raise workers’ skills or encourage them to upskill, which could boost labor productivity and the economy overall. What are your opinions on this possibility?

David NEUMARK:
Regarding the impact of minimum wages on industries where minimum wage workers are employed, when the minimum wage increases, it is likely to pass through some of the cost of the higher wages into higher prices, which is not the full amount of the minimum wage, however based on the economic definition of inflation (the rate of growth of prices), it does not have such an inflation-increasing effect, but some prices may increase. However, there is always a trade-off between the minimum wage and other adjustments made by employers.

There is a reasonable amount of research on whether minimum wages create incentives to get more skills and if they lead to higher output, but it does not find that minimum wages increase skills, and may in fact decrease skill level. Instead, it suggests that skills can be affected by two channels: schooling decisions and employer training. One channel is that people might decide to stay in school to increase their earning potential in the future or drop out of college to take advantage of current higher wages, leading to lower levels of skill in the workforce. Another channel is that employers may do less of their training after the minimum wage goes up because the worker now costs more.

The argument of whether a higher minimum wage boosts spending in output is no longer seen and almost no one believes it. If a higher minimum wage leads to more money in the economy and increased spending and output, then the money comes from somewhere else. If there are higher prices, some people demand a little less, and if some businesses are closed or fewer businesses open, output shrinks.

The argument that low-income people spend a higher share of their income than high-income people do, or “have a higher marginal propensity to consume,” and people who earn higher incomes have a higher savings rate. Based on that idea, in the short run, there could be a minor boost in demand, but in the long run, investment by higher earners leads to more capital and a larger economy, so the effect wouldn’t apply anyways.

Q:
How does minimum wage affect family formation? Do you believe rising youth minimum earnings and stabilizing employment will help individuals who want to marry?

David NEUMARK:
The hypothesis suggests that economic prospects impact marriage, and poor prospects can reduce incentives to marry, but I am unaware of specific evidence of this. Raising wages while also stabilizing employment might boost marriage, but it doesn’t necessarily happen. In fact, increasing wages seems to lower employment so there is a trade-off. In the U.S., half of minimum wage workers are under 24 or under, making minimum wage a teenage phenomenon. Teenagers have low skills and may not be paid much due to their lack of knowledge. Therefore, wages at these ages may not have a significant impact on family formation. However, governments worldwide are interested in boosting marriage and fertility, but are unsure of how to do so.

Q:
About industry-specific minimum wages, do you think raising minimum wages in certain industries works?

David NEUMARK:
Most governments around the world consider minimum wage floors as beneficial. However, there is a trade-off between paying higher wages to some workers and potentially losing jobs to others. To achieve a balance, governments should consider the specific industry they want to target. For example, in California, minimum wages are higher than in Mississippi due to higher wages overall. Additionally, a higher wage industry can afford a higher minimum wage than a low-wage industry. For example, a manufacturing plant and a restaurant might have different labor costs, with the retail store having many low-wage workers. Raising the minimum wage would increase labor costs, especially in labor-intensive places like restaurants. In contrast, a manufacturing plant would have more workers earning more, and raising the minimum wage by the same amount would not significantly impact the overall cost of labor, so the restaurant could see a 20% increase in costs while a manufacturing plant only 1%.

In Japan, minimum wage increases are done by region to some extent, but it’s also possible to do it by industry. This may be a political issue, but as an economic matter, it’s not a bad idea.

Q:
What is your opinion on the issue of publication bias? Do you think if only statistically significant figures are published, they may be biased?

David NEUMARK:
Regarding publication bias, studies that are not cited often may not be seen as credible. This is of course not necessarily true. Publication bias is a potentially real problem. This is a common issue among graduate students who believe that if their paper doesn’t yield significant results, they won’t be able to publish it. However, recent research on minimum wages doesn’t necessarily indicate publication bias. Two recent papers by Belman and Wolfson in Labor conclude weak evidence of publication bias and Casey in the American Economic Review concluded there is no evidence of publication bias in estimates, but that there may be some in terms of significance suggesting that it is not problematic.