Does Angi Recommend Occupational Licensing?

Peter Blair and Mischa Fisher have a clever new paper on occupational licensing that uses data on millions of leads generated by Angi’s HomeAdivsor. Consumers on HomeAdvisor search for services, the platform knows whether the service requires a license in the consumer’s state and attempts to match the consumer with an appropriate local provider, the local provider can then choose whether to accept or reject the lead. If a lead is accepted, the consumer and provider then negotiate on the price and services–as the negotiation is mostly handled offline, the main measure of interest is the probability a lead is accepted.

Many occupations are licensed in one state but not in another (as I pointed out in my talk on occupational licensing to the Heritage Foundation this is odd if you think there are strong arguments for occupational licensing on safety or quality grounds). Thus the authors compare the accepted lead rate in states that require a license to complete a task to the rate in states where the same task is unlicensed. To better control for other factors, the authors only compare the accepted lead rate in counties that border different states, as shown below. The authors also control for state, month and task fixed effects.

The bottom line is that the accepted lead rate is 12.3 percentage points or 21% lower in a county/state that license an occupation/task compared to a similar county/state where the task is unlicensed. In other words, if you live in a state that requires a license to complete a task it’s considerable more difficult to find a contractor than if you lived in nearby state that does not license the task.

Not surprisingly, the authors find that the accept rate declines not because there is a surge in demand for the licensed service but because supply declines when there are fewer licensed providers. In the long run, we also know that prices rise in licensed industries (e.g. my paper with Pizzola on licensing in the funeral services industry).

The authors combine their cross-sectional study with an event study that shows that after New Jersey required a license for pool contractors it become more difficult to find a pool contractor in New Jersey relative to other states.

The authors conclude:

The existing literature on licensing on digital platforms, which consists of three other papers, has carefully measured the impact of licensing on consumer satisfaction and safety by demonstrating that customer self-reports of service quality and objective platform measures of service provider safety do not increase in the presence of licensed service providers, despite the positive impact of licensing on prices (Hall et al., 2019; Farronato et al., 2020; Deyo, 2022).

…Taken together, our findings and those from the three others papers studying licensing in digital labor markets indicate that the traditional view of licensing espoused in Friedman (1962)
about licensing in offline markets, i.e, licensing is a labor market restriction with limited benefits, also holds in digital labor markets (Hall et al., 2019; Farronato et al., 2020; Deyo,
2022). Our work provides a clear example where labor market regulations developed to govern the analogy economy work against the efficiency gains that technological innovation promises to bring in a digital economy (Goldfarb et al., 2015).