Sandwichman in the FT

Financial Times: "The minimum wage wars are heating up: A new study fails to prove its claim that Seattle wage floor hurts workers" by Martin Sandbu, at Free Lunch on FT Alphaville

First, the numerical result struggles to pass an intuitive “smell test”. As the Angry Bear blog [cross posted at EconoSpeak!] points out, employment in Seattle was booming throughout the period: average wages increased by 18 per cent (!) in the time covered by the study; as did the number of hours worked at all wage rates. It is important to note that the researchers have data on jobs, not on individual workers — so even if there are fewer low-paid jobs than before, it does not follow that workers have lost as many jobs rather than moved into better ones.

...

Fifth, there is a simple arithmetical issue with the methodology. If "low-paid" is defined as below a fixed threshold wage ($19/hour in this study), higher average wage growth in one group relative to an initially similar control group will necessarily take more people out of the low-wage bracket. Such wage bracket creep means that the higher wage growth will "cause" a "loss" of low-wage work through mechanical arithmetics even if nothing causal is happening (indeed even if everyone stays in the same job).