The 2021 Nobel Prize in Economics goes to David Card, Joshua Angrist and Guido Imbens due to their work on “natural experiments” and how this econometric technique can inform important research questions in labor economics. From the Nobel Prize press release:
Using natural experiments, David Card has analysed the labour market effects of minimum wages, immigration and education. His studies from the early 1990s challenged conventional wisdom, leading to new analyses and additional insights. The results showed, among other things, that increasing the minimum wage does not necessarily lead to fewer jobs. We now know that the incomes of people who were born in a country can benefit from new immigration, while people who immigrated at an earlier time risk being negatively affected. We have also realised that resources in schools are far more important for students’ future labour market success than was previously thought.
Data from a natural experiment are difficult to interpret, however. For example, extending compulsory education by a year for one group of students (but not another) will not affect everyone in that group in the same way. Some students would have kept studying anyway and, for them, the value of education is often not representative of the entire group. So, is it even possible to draw any conclusions about the effect of an extra year in school? In the mid-1990s, Joshua Angrist and Guido Imbens solved this methodological problem, demonstrating how precise conclusions about cause and effect can be drawn from natural experiments.
Difference-in-difference. One well-known natural experiment examined how minimum wage laws affects labor markets. A paper Card and Krueger (1994) compared nearby fast food restaurants in New Jersey and Pennsylvania. Fast food restaurants in the former state experienced a change in minimum wage laws whereas the latter did not. Card and Kruger look at the change in employment in NJ and compare it to changes in employment in PA. They stated that there was “no indication that the rise in the minimum wage reduced employment”.
Instrumental variables. Angrist and Imbens (1991) used instrumental variables (IV) to answer a number of questions such as examining returns to school. Alex Tabarrok provides a nice overview of how discrete cutoffs (e.g., ability to start school if born before Jan 1, but not if born after; ability to leave school at age 16), provide an exogenous source of variation on the months of education. Another canonical IV example pioneered by Angrist is the study [Angrist 1990] looking at the impact of military service on lifetime earnings. Since characteristics that predict military service may predict earnings as well, ideally one would randomize people into military services vs. not. Of course, this type of randomization is not feasible, but instead Angrist used information on an individuals’ Vietnam draft lottery number as an exogenous source of variation. The IV approach allows one to estimate the local average treatment effect (LATE) for compliers (i.e., those would would join the army if drafted, but would not join if not drafted). Using this IV approach, Angrist finds that “the earnings of white veterans were approximately 15 percent less than the earnings of comparable nonveterans”. Angrist also has a nice, short book on econometrics with Jörn-Steffen Pischke titled Mostly Harmless Econometrics: An Empiricist’s Companion.