Fred Thompson: My Last Post on Minimum Wages (I Hope)

 Fred Thompson carries the water this week as I am in Washington, DC giving a talk and working at the World Bank.  

Why do folks get so worked up over the minimum wage? The best evidence shows that the net effects of increasing minimum wages are trivial (the figure on the right plots strength of results on the vertical axis and effect size on the horizontal axis). As Madeline Zavodny put it, the issue appears to be a “maximal conflict over minimal effects.”

Moreover, it is clear that state adoptions of minimums higher than the national standard are not due to differences in economic conditions: wage growth, the cost of living, income inequality, a high proportion of minimum-wage earners in the workforce, etc. That these things should matter seems intuitively plausible. For example, a recent article in Governing magazine categorically states that state minimum-wage increases are driven primarily by growth in the relative cost of living, especially where gentrification and amenity-based tourism amplify inequalities of income and wealth. But, when such claims are subjected to rigorous empirical testing no relationships are found. Minimum wage increases are apparently not driven by absolute or relative economic hardship or, absent interaction with political variables, economic conditions generally.

What, then, explains interstate variations in minimum wages? The best research on this issue done to date fingers politics. My colleagues, Kawika Pierson, and Tim Johnson, and I have replicated much of this research. Our analysis confirms most earlier findings: that minimums higher than the national standard predict subsequent state minimum-wage hikes, that timing matters, and that states, which enact minimum-wage hikes, are almost invariably under Democratic Party control (although not all states that are controlled by Democrats have higher minimums and neither the size of Democratic majorities nor shifts from Republican to Democratic control appear to be correlated with increases in state minimum wages.)

We have extended this line of investigation in three directions. First, we show that a disproportionate number of minimum-wage hikes are enacted during the even-numbered years between presidential elections (so-called mid-term election years). This variable turns out to be significantly more powerful than the number of years since the last boost (federal and/or state), the main timing-related mechanism considered in earlier studies. Second, simple partisan control matters more than ideology. Earlier studies looked at partisan control or ideology, but not both. And, third, after accounting for the election cycle and partisan control, state minimums are found to vary inversely with their cost to in-state consumers, either because states adopting them have few minimum wage workers or because of tax exporting (i.e., shifting the burden of the minimum wage to out-of-state consumers). Most economists believe that if a jurisdiction can export taxes, it will. But our attention to this factor was, in fact, triggered by the SeaTac initiative, which raised the minimum wage “for the airport’s hospitality and transportation workers.”  

We used mixture models with state minimums as our response variable and year fixed-effects and the interaction between Democratic control of the state legislature and low-wage or export employment as a percent of total employment as our predictor variables. Our interaction terms are consistently positive and significant. The interaction terms also dominate any effect that Democratic control (either as a percent or a dummy) or legislative ideology shows on its own. Interestingly, specifications featuring the interaction between Democratic control of the state legislature and export employment consistently outperform those featuring the interaction between Democratic control and low-wage employment. The interaction models involving ideology also work quite well, but are dominated by the interaction of a naive Democratic control dummy (1 when Democrats control the legislature) and the jobs measures. Stated statistically, there are good reasons to think that legislative ideology scores are measuring demand for minimum wages ‘with error’ in this setting, and concluding, therefore, that the real issue is party brand.

As improper as this is, we marvel at the fact that we can explain 40-50+ percent of the variance in state minimum wages net of fixed effects. This seems like a lot of explanatory power given that we are considering only partisan control, the jobs mix, and the election cycle. At the same time, we do not see how these results can be reconciled with the notion that minimum-wage battles are really about the policy’s material consequences. The variables included in our models are simply too feckless for that interpretation to be taken seriously.

Instead, they suggest that the primary function of proposed minimum-wage hikes is rallying the party faithful to fight mid-term elections and that when Democrats actually control legislatures they must often put up or shut up, whether they really want to or not. In other words, the minimum wage, like banning assault-weapons or abstinence-only sex education, is basically a political football, an otherwise largely inconsequential object, given symbolic importance by interparty rivalry.

Of course, minimum wage hikes are also symbolically important as expressions of public concern for low-wage workers. Nevertheless, this legitimate justification is difficult to reconcile with the fact the issue tends to reappear on the legislative agenda every fourth year even in states where the minimum wage is high and indexed for inflation.