Graton M. R. Gathright and Dr. Norman K. Thurston, Economics

The unambiguous prediction of the competitive labor market model is that a minimum wage serves as a price floor, decreasing the quantity of unskilled labor demanded by firms. In a landmark survey paper, Brown, Gilroy, and Kohen survey 30 years of minimum wage studies and recommend the widely cited generalization that a 10% increase in the minimum wage is estimated to result in a 1% to 3% reduction in total teenage employment.1

Contrarily, in a study of the employment effects of the April 1992 increase of the New Jersey minimum wage, David Card and Alan Krueger (CK) conclude that increasing the minimum wage from $4.25 to $5.05 led to a more than 13% employment increase in the state’s fast-food industry. CK conducted a “natural experiment” by surveying fast-food restaurants concerning their employment of low-wage workers in NJ and adjacent Pennsylvania prior to and following the increase in the NJ minimum.2

My initial intent had been to replicate the CK study in order to search for insights to reconcile its conclusions to the conflicting microeconomic theory predictions. However, David Neumark and William Wascher (NW) had already performed a re-analysis of the question with an improved dataset. NW contacted the restaurants from the survey to obtain administrative payroll data that would avoid alleged data problems in the CK survey. Based on the new data, NW conclude that the increased minimum wage in NJ led to a 4% decrease in fast-food industry employment in that state.3 In reply, CK argue that NW’s sample was not representative. CK improve upon the dataset again by obtaining access to Bureau of Labor Statistics unemployment insurance data for the very restaurants originally surveyed. From the re-analysis, CK conclude that the legislation likely had no effect on employment in the industry in that state.4

With these re-analyses in hand, my focus has turned to considering what further studies could test the robustness of the results. In their re-analyses, CK and NW both use ES-202 data from the Bureau of Labor Statistics that are based on unemployment insurance for “business establishments.” A mix of “establishments” more representative of the various segments of the market for unskilled labor for the two regions would provide a test of the robustness of the fast-food industry results. Finis Welch suggests that the generalization of outcomes in the fast-food industry to the entire low-wage labor market may represent a fallacy of composition.5

I have also considered a criticism made by Daniel Hamermesh of this “natural experiment.” Hamermesh points out that the initial survey followed the enactment of the legislation by nearly two years but preceded the change in the minimum by only one month. The survey may well have missed the change. Similarly, the second survey took place only a few months after the new minimum was implemented, whereas a 1993 study by Hamermesh demonstrated that in the short run, non-labor inputs adjust slowly, slowing the adjustment of complements in production, including labor. For this reason, even if we assumed that employers failed to anticipate the new minimum despite its having been legislated two years earlier, the second survey could likely have been administered too soon to capture the effects of the treatment. Finally, Hamermesh points out that it is unclear that PA and NJ would have experienced identical employment outcomes barring the NJ wage rate change. Any two economic outcomes that are similar at one time will not necessarily be similar at some other. Geographic propinquity does not imply equivalent economics propensities.5

An alternative study, which would retain much of the ‘experiment’ appeal, would address all three of Hamermesh’s criticisms and serve as a test for geographic robustness of the results. The methodology would require a much broader, perhaps national, cross-section of low-wage employment levels for “establishments.” This methodology, called optimal grouping for response surface analysis of observational data, would provide a controlled-experiment type study using this data. Ranking observational units (cities, for example) on minimum wage level, the units are then stratified into a specified number of groups. These groups become the observational unit, and a mathematical program makes slight adjustments to the groupings so as to maximize the range of the independent variable (minimum wage level) subject to the constraint that each group have the same average value on all control variables.6 The result is a sample of groups of cities each with identical average values on all control variables yet systematic differences in minimum wage. Any relationship between the minimum wage level and low-wage employment would be easily detected.

References

1. Brown, Charles, Curtis Gilroy, and Andrew Kohen. “The Effect of the Minimum Wage on Employment and Unemployment.” Journal of Economic Literature 20, June 1982, pp. 487-528.
2. Card, David, and Alan B. Krueger. “Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania.” American Economic Review 84(4), September 1994, pp. 772-93.
3. Card, David, and Alan B. Krueger. “Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania: Reply.” American Economic Review 90(5), December 2000, pp 1397-1420.
4. Neumark, David, and William Wascher. “Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania: Comment.” American Economic Review 90(5), December 2000, pp. 1362-96.
5. Welch, Finis, and Daniel Hamermesh et al. Review Symposium: “Myth and Measurement: The New Economics of the Minimum Wage.” Industrial and Labor Relations Review 48(4), July 1995.
6. Stone, Bernell K., and Donald L. Adolfson, and Tom W. Miller. “Optimal Data Selection and Grouping: An extension of traditional response surface methodology to observational studies involving non-controlled empirical data generation.” Advances in Mathematical Programming and Financial Planning, Vol. 3, pp. 39-68.