President Obama’s State of the Union recent speech called for a higher minimum wage:
We know our economy is stronger when we reward an honest day’s work with honest wages. But today, a full-time worker making the minimum wage earns $14,500 a year. Even with the tax relief we’ve put in place, a family with two kids that earns the minimum wage still lives below the poverty line. That’s wrong. That’s why, since the last time this Congress raised the minimum wage, nineteen states have chosen to bump theirs even higher.
Tonight, let’s declare that in the wealthiest nation on Earth, no one who works full-time should have to live in poverty, and raise the federal minimum wage to $9.00 an hour. This single step would raise the incomes of millions of working families. It could mean the difference between groceries or the food bank; rent or eviction; scraping by or finally getting ahead. For businesses across the country, it would mean customers with more money in their pockets. In fact, working folks shouldn’t have to wait year after year for the minimum wage to go up while CEO pay has never been higher. So here’s an idea that Governor Romney and I actually agreed on last year: let’s tie the minimum wage to the cost of living, so that it finally becomes a wage you can live on.
The proposal here assumes that all of the increase in minimum wages will flow to poor workers with no adverse consequences. The academic literature on this subject is tangled and highly controversial. Still, is worth thinking through the evidence on the effects of a massive proposed hike in the minimum wage.
The key paper setting off the first round of recent debates was written by David Card and Alan Krueger (the current Chairman of the Council of Economic Advisors) in 1994. The key issue to grapple with in studying minimum wages in the data is the ability to construct a valid control group or counterfactual. For instance, suppose the federal minimum wage went up and aggregate unemployment rose. While this could be consistent with the minimum wage having a negative impact on employment, it is also possible that unemployment would have risen anyway absent any policy change due to unrelated factors.
Card and Krueger addressed this issue by focusing on state minimum wage hikes. In 1992, New Jersey hiked its minimum wage, while Pennsylvania held its minimum wage the same. To the extent that employment conditions in these two states were comparable, the difference in employment outcomes between the two states after the minimum wage hike can be seen as the pure effect of the minimum wage on employment.
Card and Krueger focused on four fast food chains—Burger King, KFC, Wendy’s, and Roy Rogers—located in both states. They called up managers at over 400 restaurants and had them report employment both before and eight months after the minimum wage hike. Surprisingly, they found that employment actually increased in New Jersey restaurants, relative to those in Pennsylvania.
This result was contested by David Neumark and William Wascher, who were particularly concerned that simply calling up fast food restaurants would not result in reliable data. Together with EPI, they contacted franchisees directly for payroll information at the four chains chosen by Card and Krueger. On this new dataset, they found that the minimum wage hike in New Jersey resulted in higher unemployment.
Card and Krueger hit back with yet another paper using data taken the Bureau of Labor statistics data on fast-food employment. On this new dataset, they found that the higher minimum wage in New Jersey led to a statistically insignificant change in employment. To be sure, this finding is more positive for advocates of a higher minimum wage than Neumark and Wascher’s estimates. However, it is worth emphasizing that this better data source didn’t replicate the original Card and Krueger result (which is still referenced frequently) that employment could even rise after a minimum wage increase.
Card and Kruger also display the following graph in the paper, showing overall employment changes in both New Jersey and Pennsylvania:
Recall that the statistical strategy from both sets of researchers relies on the assumption that the employment situation in both New Jersey and Pennsylvania is identical except for the impact of the minimum wage. With this assumption, any differences between the two states will be interpreted as representing only the employment effects of the minimum wage. Instead, it looks as if both states appear to have highly variable employment trends that strongly diverge across many time periods. As a result, any effects identified as being the result of the minimum wage hike may instead reflect spurious other fluctuations in employment markets.
Partially to address the issues resulting from these state-level studies, Arindrajit Dube, T. William Lester, and Michael Reich have recent influential work arguing that the employment effects of the minimum wage hike are minimal. The focus here on contiguous counties that straddle state lines, and are hopefully more similar to each other than entire states. When one state raises its minimum wage, the counties along the state line experience a minimum wage shock that does not affect the counties on the other side of the state border, so any difference in employment outcomes between the two sets of counties can be interpreted as reflecting the impact of the minimum wage hike. Their overall conclusion is that “These estimates suggest no detectable employment losses from the kind of minimum wage increases we have seen in the United States.”
Striking back; Neumark, Salas, and Wascher have a recent paper critiquing this approach—again focused on the issue of which statistical procedures result in reasonable counterfactuals. They point out that the Dube et al. work is primarily focused on teenagers and restaurants; when many restaurants do pay above minimum wage, and so are unaffected by the minimum wage increase. Also, other studies on minimum wages that find negative employment effects in the aggregate sometimes do not find them teenagers specifically.
They also dispute the issue of whether other states or nearby counties provide better controls in response to a hike in the minimum wage. While it would seem reasonable that nearby areas offer good controls; just looking at the graph of overall New Jersey and Pennsylvania trends suggests that might not be the case. Instead, the employment pattern in other states elsewhere in the country may actually provide more reasonable counterfactuals.
The precise critiques get complicated here. Neumark et al. find that their preferred models result in a negative employment impact of a negative wage. No doubt Dube and company will respond back with a model that finds the opposite result. Though the degree of disagreement in this literature is huge, there are several things to keep in mind:
Finally, it’s worth considering the federal minimum wage hike proposal in light of long-run trends in low income employment. Low skilled employment remains low, and the long-run unemployed in particular find it very difficult to get new jobs. Overall labor force participation rates have remained poor, and many workers have used disability benefits to exit the workforce. Large swaths of employment for low-skilled workers, men in particular, have disappeared—such as manufacturing and construction.
At the same time, policies have made low-skilled employment even harder. Low income workers face high implicit marginal tax rates, as their higher incomes result in lower benefits. Firms face escalating health care costs, and PPACA heightens the burden of paying for the health costs of workers. While it’s difficult to establish the effects of a minimum wage hike with any certainty, it certainly would not help a troubling dynamic in which a larger chunk of the population is divorced from the workforce entirely due to escalating costs and diminishing benefits of employment.