The primary findings from this study are two-fold. 

First, across the multiple occupations and states, licensing (or stricter forms of it) consistently does not appear to produce greater service quality. In seven of the comparisons, there are no statistically significant differences in Yelp ratings. Quality in unlicensed or less burdensomely licensed states is essentially equivalent to that in licensed or more burdensomely licensed states. 

Second, where there is a measurable difference between states as indicated by statistical significance, the state with lighter regulation has higher quality. Specifically, cosmetologists in New York, with a less burdensome license, receive higher Yelp ratings than those in Connecticut and New Jersey, with more burdensome licenses. 

If we look at the results in more detail, based on the average Yelp ratings for businesses on either side of state borders, licensing regime does not appear to make a meaningful difference in service quality. As Figure 1 illustrates, the differences are small, never exceeding one point in the five-point Yelp scale. On average, the absolute difference in businesses’ ratings between unlicensed or less burdensomely licensed states and their licensed or more burdensomely licensed counterparts comes to approximately 0.27 points. 

We also observe no consistent patterns in the states with higher or lower Yelp ratings. In five of the nine comparisons, businesses in unlicensed or less burdensomely licensed states receive higher ratings, while in the other four comparisons, businesses in licensed or more burdensomely licensed states do. If licensing, or stricter forms of it, truly produces greater service quality, we would expect businesses in licensed, or more burdensomely licensed, states—the red bars in Figure 1—to consistently receive higher Yelp ratings. Yet our results show no such consistent pattern. 

Figure 1: Across Nine Comparisons, Licensing Never Produces Statistically Greater Quality

Even stronger evidence comes from our statistical analysis. All but two of the differences shown in Figure 1 are no greater than what we might expect to see through random chance. Across seven of nine comparisons in which businesses were effectively alike, save for their regulatory environments, licensing is not associated with differences in quality that exceed what we would expect from chance alone. And because the nine comparisons cover six occupations and nine states, our findings are unlikely to be an artifact of any particular occupation, type of occupation or state. 

Where we do find a statistically significant difference—the comparisons with asterisks in Figure 1—the state with less stringent requirements has higher quality. Cosmetologist ratings in less burdensomely licensed New York are, on average, approximately 0.34 points higher than those in Connecticut and 0.28 points higher than those in New Jersey, both of which have greater licensing burdens. (Appendix B provides our full results.)

If licensing proponents are right and licensing leads to greater service quality, we should see higher cosmetologist ratings in Connecticut and New Jersey—and with a clear, statistically significant difference. That we find precisely the opposite—plus no statistically significant differences in our seven other comparisons—strongly suggests licensing’s purported benefits for consumers are overstated for at least the occupations we studied and perhaps for others.