Over the last few months, I have repeatedly heard opponents of raising the federal minimum wage argue that "85 percent" of the "most credible studies" find that increasing the minimum wage has a negative effect on employment.
The main promulgator of this particular number is Michael Saltsman of the low-wage-employer-sponsored Employment Policies Institute. Earlier in the year, for example, at Forbes.com he wrote:
"In a comprehensive, 182-page summary of the research on this subject from the last two decades, economists David Neumark (UC-Irvine) and William Wascher (Federal Reserve Board) determined that 85 percent of the best research points to a loss of jobs following a minimum wage increase."
The original source of that statistic is a 2007 study by Neumark and Wascher, who examined 102 estimates of the employment effects of the minimum wage and from these selected 33 that were "most credible."
Of the 33 "most credible" studies, they determined that "28 (85 percent) ... point to negative employment effects." (p. 164)
But, this is a much more subjective exercise than is generally acknowledged by opponents of the minimum wage. The infographic below, prepared by my CEPR colleague Milla Sanes, can help to illustrate.
The chart lists, by author and year of publication, the 101 studies that produced the 102 estimates in Neumark and Wascher's review. The first column shows the 68 studies that did not make their cut as "most credible." By their judgment, basically two thirds of the extensive literature on the minimum wage is not "most credible."
Of the 33 "most credible" studies, Neumark and Wascher judge 28 to find negative effects on employment and 5 to show neutral or positive effects. This is the source of the "85 percent" of studies — 28 of 33 studies, not 85 percent of 102 studies as is sometimes implied.
One particularly problematic issue is the decision to include foreign studies of the minimum wage. Almost half (15 of 33) of the "most credible" studies — and more than half of the "most credible" studies that show negative employment results — are based on the experience of other countries. I am a firm believer that we can learn a lot about economics and economic policy from the experiences of other countries. However, if we want to know about the likely impact of a specific policy such as raising the minimum wage, done in the way that we typically raise the minimum wage in the United States, in an economy that has a supply and demand for labor like the United States, and labor-market institutions (unions, unemployment insurance benefits, health-care benefits, etc.) like the United States, I think we really want to limit our analysis just to studies of the U.S. experience.
The other countries where the minimum wage is associated with job loss may be implementing the policy in a different way and are certainly doing so in a different economic context, making comparisons difficult. These other countries may set the minimum wage much higher (relative to their economies) than we have here in the United States. Other rich countries also have more extensive social safety nets than what we have here and therefore may be more willing to raise the wages of the vast majority of low-wage workers at the expense of some job loss for a small share of low earners. We have extensive experience in the United States with the minimum wage at the federal, state, and, increasingly, local level, all of which gives a much better idea of the likely effects of the policy here.
Since all 15 of the foreign studies included in the 85 percent figure found negative employment effects and all five of the studies reporting neutral or positive effects were for the United States, excluding the foreign studies lowers the percentage of negative studies from 85 percent to 72 percent.
One final and obvious element of subjectivity concerns the designation of which studies are "most credible" and which are not. Of the 13 U.S. studies that demonstrate negative employment effects, five are studies by Neumark and Wascher themselves. The two have every right to designate their own studies as "most credible," but people citing the 85 percent statistic need to be upfront that the source is the subjective judgment of two researchers who may judge their own work and the work of their critics differently than their critics or other economists with a little more distance from the debate. If we limit the analysis to the "most credible" U.S. studies not performed by Neumark and Wascher, the 85 percent number falls to 62 percent.
Of course, tallying studies on one side or the other and using the final count is far from the best way to make economic decisions. Researchers have used meta-studies, which reduce the element of subjectivity involved in these kinds of exercises. After applying meta-study techniques to studies of the effect of the minimum wage on teen employment in the United States, for example, Hristos Doucouliagos and T. D. Stanley concluded that there is "little or no evidence of a negative association between minimum wages and employment."
But, ultimately researchers, policymakers, and the public should make their decisions on the quality of the arguments, not on a highly subjective shorthand statistic based solely on the number of studies pointing in one direction or the other.