A few days ago, Tyler Curtis at the Freeman (the Foundation for Economic Education’s flagship publication) posted a short piece on the minimum wage and health code violations in restaurants. Curtis based on a paper (unavailable in full) by Srikant Devaraj who asserted that increases in minimum wages in Seattle had led to increases in health code violations by restaurants (heavily affected by the increases).
Devaraj used a standard difference-in-difference econometric approach. The problem underlined by some was the choice of benchmark for the method : New York City. New York City and Seattle are two very different cities with different health codes. It is hard to make this claim stick even if the uncontrolled results (before statistical tests) show an increase in health code violations. Nonetheless, I have been able to find one other study that shows – based on Californian data – that there was a very small deterioration in health code violations (especially by the top restaurants) following increases.
Now, I am not convinced by the econometric design of both, but I am axiomatically convinced. This is where I think economists have made the error of relying too much on empirical methods. While, as an economic historian, I always favor more data, I also am trained to be skeptical about data does not say.
In the case of the minimum wage, the debate has raged between economists over the employment effects (i.e. the demand for labor). But this is a fraction of everything involved with the production of goods in industries affected by minimum wages. For an employer, a cost is a cost regardless of the form it takes. If an employer is forced to pay somebody above what they produce in value, then something has to give. For a 5% increase in the minimum wage, it is doubtful that an employer with three employees will be willing to sack one third of his workforce (and roughly one third of his output). So, he can cut costs differently. He may ask employees to buy their uniforms, he may refuse to provide them with free lunches, he may also even decide to cut on quality of his service – as is the case with the two studies outlined above.
The problem is that no study of the minimum wage has attempted to measure all these effects at once! There is no study that looks simultaneously at hours worked, people employed, type of people employed (substitution effects), prices for consumers, quality and marginal benefits (uniforms, free lunch, insurance, etc.) on both the short and long-terms levels and trends (they also rarely adjust the minimum wages for regional purchasing power parities and the under-reporting of tips)
The health code violations papers show how many channels employers can use to adapt – channels which some fail to account for when they proclaim that we can raise the minimum wage without adverse consequences. Maybe its time that we, as economists, try to be more cautious when we make claims about the minimum wage’s minimal effects.