Does more schooling boost income for the educated, or are those able to gain college admission likely to earn more anyway? Does it matter where you go to school? Such causal questions are challenging because differences in life outcomes, such as earnings differences between university and high school grads, need not capture causal effects. Perhaps those more educated were destined to earn more anyway; they may have more educated parents, among other differences. This is the problem of selection bias; it’s why drug trials, such as used to evaluate COVID vaccines, rely on random assignment into treatment and control groups rather than more casual comparisons. Random assignment ensures that differences in infection rates between those receiving an experimental vaccine and those receiving placebo are due solely to the effect of the vaccine.
Unlike pharmaceutical trials, social scientists rarely have the chance to conduct randomized clinical trials (RCTs). But thanks to pioneering work by MIT economist Josh Angrist and his collaborators, we have new tools allowing us untangle knotty problems of cause using natural experiments rather than clinical random assignment.
Professor Angrist pioneered the use of natural experiments to answer causal questions ranging from the effects of childbearing on mothers’ employment and earnings to effects of military service on veterans’ health and wages later in life. Say you want to know whether military service affects your earnings later in life. For labor economists like Angrist, that’s an important question: the Pentagon is one of the world’s largest employers. If only we could randomize military service!
Remarkably, the US military came very close to this. From 1970-73, the order in which men were called up was determined by draft lottery numbers randomly assigned to birthdays. Men with low lottery numbers were called for possible service. Those with high draft lottery numbers needn’t have served if they didn’t want to.
The natural experiment this scenario suggests compares the earnings later in life of men with low and high draft lottery numbers. But it isn’t so simple: many called for induction managed to avoid service. Meanwhile, 20% of men with high lottery numbers volunteered for service. The draft lottery is not a real RCT: it’s messy—and we need an econometric framework to account for that.
In path-breaking studies published in the 1990s, Angrist and Guido Imbens of the Stanford Graduate School of Business showed how we can apply econometric strategies using events like the draft lottery to estimate causal effects. Their research links causal effects estimated using natural experiments to effects we would obtain were military service to be studied in an RCT. For their influential work, Angrist and Imbens shared half of the 2021 Economics Nobel, with the other half going to David Card of the University of California at Berkeley.
Angrist was born on 18 September 1960 in Columbus, Ohio. He received a BA in economics from Oberlin College in Ohio in 1982 and a Ph.D. from Princeton University in 1989. Angrist has taught at Harvard University, the Hebrew University of Jerusalem, and MIT. Since 1996, Angrist has been a Professor of Economics at MIT.