August 06, 2014
In a series of recent blogposts, we updated a methodology used by Goldman Sachs to evaluate the impact of minimum-wage increases in 13 states at the beginning of this year. A number of publications, including USA Today and The New York Times, cited these blog posts in pieces on the minimum wage.
As we noted at the time, the Goldman Sachs’ methodology is not ideal for a number of reasons. Now, economists Saul Hoffman and Wai-Kit (Ricky) Shum from the University of Delaware have done a more formal analysis of these 13 states and found qualitatively similar results to the pieces using the Goldman Sachs method. Specifically, all of the analyses have found no evidence of negative employment impacts. If anything, the results have indicated that minimum wage increases are associated with more rapid employment growth, although this relationship is not statistically significant.
Hoffman and Shum improved on the Goldman Sachs approach by focusing on teenagers and less educated adults, which are the groups most likely to be affected by changes in the minimum wage. In contrast, Goldman Sachs looked at overall employment, where the likelihood of finding statistically significant negative employment effects is smaller. Furthermore, Hoffman and Shum used more rigorous controls for underlying macroeconomic trends. Goldman Sachs simply compared the average employment changes in the states that increased their minimum wage to those that did not.
The table below reproduces the key elements of Hoffman and Shum’s Table 4. The first column of the table reports their results for teenagers; the second column reports their results for adults (20-59) who have less than a high school education. (Note that Hoffman and Shum’s before-and-after comparison uses an earlier “before” period (March to May, 2011) than Goldman Sachs (December, 2013).
Excerpt from Hoffman and Shum’s Table 4:
“Effects of 2011-2014 State Minimum Wage Increases on Employment of Selected Low-Wage Groups”
| Age 16-19 | Not HS Grad, Age 20-59 | |
|
2011 Employment Rate in States with MW Increase (percent) |
20.4 | 51.09 |
|
2014 Employment Rate in States with MW Increase (percent) |
22.36 | 54.41 |
|
Change in Employment Rate in States with MW Increase (percentage points) |
2.32 | 3.32 |
|
2011 Employment Rate in States without MW Increase (percent) |
21.91 | 54.31 |
|
2014 Employment Rate in States without MW Increase (percent) |
23.48 | 57.28 |
|
Change in Employment Rate in States without MW Increase (percentage points) |
1.57 | 2.97 |
|
Difference-in-Difference |
0.75 | 0.35 |
|
Difference-in-Difference-in-Difference |
1.16 | 0.76 |
In order to evaluate how the states that raised their minimum wage are doing, economists compare them to the states that did not experience an increase in the minimum wage. A naïve way to do this would be to compare the current employment rates for teens in states with no minimum wage increase (23.48 percent) to states where the minimum wage went up (22.36 percent). This naive comparison would lead us to conclude that increasing the minimum wage decreases teen employment by 1.12 percentage points. But, this simple approach doesn’t take into account where employment was in each state before the minimum wage went up. If the states that raised their minimum wage started with a lower teen employment rate, then we might not be measuring the effect of the minimum wage increase but rather some other features of those states’ labor markets. In fact, as the table shows, teen employment was lower in the 13 states where the minimum wage went up–even before the 2014 increases went into effect.
We can control for the different starting points by comparing the change in employment within each of the two groups of states before and after the minimum-wage increase. The change in teen employment in the states where the minimum wage was constant acts as an experimental control group. The change in employment for teens in those states acts as a benchmark for what we think would have happened in the 13 states where the minimum went up–if the minimum wage had, instead, remained constant in those states. In this case, we can see that in the states where the minimum wage remained constant, teen employment rose 1.57 percentage points. In the states where the minimum wage increased, teen employment rose more, by 2.32 percentage points. If we subtract the first number from the second, we have an estimate of the impact that raising the minimum wage had on teen employment: a 0.75 percentage-point increase in employment. But this “difference-in-differences” is not statistically significant. So far, except for the use of teens instead of overall employment, this approach is similar to the methodology that Goldman Sachs used, and it comes to the same conclusion.
This second approach, however, does not control for underlying economic trends, which may differ across states. If, for reasons unrelated to the minimum wage, the states where the minimum wage went up were growing faster as a group, then the difference-in-difference estimates would paint an overly optimistic picture of the minimum wage. In order to control for these underlying trends, economists Hoffman and Shum add one additional level of benchmarks. They compare the change in teen employment to the change in employment rates for males, age 30-49 with at least some college, a group whose employment is unlikely to be affected by changes in the minimum wage. The employment patterns for this group will likely move with general trends, so they can serve as a good proxy for overall economic conditions in each state.
When Hoffman and Shum do this “difference-in-difference-in-difference” analysis, they find that the positive effect of raising the minimum wage on teen employment is even greater than when they don’t control for differences in employment growth across the states. In states where the minimum wage went up, teen employment actually did a bit better than would have been expected based on overall job creation, but, as before, this positive effect is not statistically significant.
The story is largely the same for workers age 20-59 who do not have a high school diploma. Hoffman and Shum find that minimum wage increases were associated with bigger employment increases for adults with less than a high school degree–up 0.35 percentage points relative to states where there was no increase–and that this positive effect was larger (0.76 percentage points) after controlling for differences in the business cycle across the states. As was the case with teens, these estimates are not statistically significant, but the consistency between the stories for teen workers and older workers without a high school diploma lends supports to the conclusion that moderate increases in the minimum wage don’t hurt state job growth even among the groups who are most affected.
