‘Causal Inferences’ Led to Nobel Prize

Come October, I eagerly look forward to The Royal Swedish Academy’s announcement of Nobel prizes, particularly in economics. And like many others, I was delighted to hear this year's prize going to economists who are still very active in their fields of research.  

This year, The Sveriges Riksbank Prize in Economic Sciences in memory of Alfred Nobel is awarded with one half of prize money to David Card of University of California, Berkeley, USA, “for his empirical contributions to labour economics”; and the other half jointly to Joshua D. Angrist of Massachusetts Institute of Technology, Cambridge, USA, and Guido W. Imbens of Stanford University, USA, “for their methodological contributions to the analysis of causal relationships”.

In the past—to be precise, till these three brilliant econometricians came up with their empirical studies to analyse the labour market effects of minimum wages, immigration and education—economists struggled to figure out whether an observed relationship between two variables is causal or coincidental. For, unlike in sciences, it is not possible in social sciences to conduct rigidly controlled randomised experiments to verify the causal relationships.

It is against this background that David Card along with his late co-author Alan Krueger analysed the labour market effects of rise in minimum wages and published the findings in their paper—Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania—that revolutionized empirical research in economics.

Taking advantage of the naturally occurring situations viz., the raised minimum wage in the State of New Jersey and no such raise in the neighbouring State of Pennsylvania, Card and Krueger, treating these policy changes as though naturally occurring random variations, studied the impact of rise in minimum wage on employment. Treating the fast-food restaurants in New Jersey as ‘treatment’ group and their counterparts in the neighbouring Pennsylvania State as ‘control’ group, they tested the hypothesis that raising minimum wages lowers labour demand by comparing changes in employment levels before and after the wage rise in New Jersey and found “no indication that the rise in the minimum wage reduced employment.”

The results of this study, challenging the conventional wisdom—textbook competitive model of the labour market which predicts that rise in minimum wage will have negative impact on employment—had simply offered a new way of doing economic research. Continuing with the new-found concept of ‘natural experiments’, Card studied a variety of other important policy questions such as: How does immigration affect employment levels and wages of native workers? How does a more number of years of education affect a student’s future income? His pioneering work that defied conventional thinking, simply revolutionised research not only in the field of economics but also in all other social sciences, besides bettering our understanding of how the labour market operates in real world situations. 

Taking this path breaking research forward, Prof. Imbens and Prof.  Angrist have made innovative methodological contributions to draw precise conclusions about cause and effect from the ‘natural experiments’ whose results were otherwise found to be difficult to interpret. For instance, it is said that in the US, graduates of private universities earn higher wages than public university graduates. This phenomenon tempts one to quickly jump to the conclusion that private universities cause wages to go up. But the research of these two econometricians offered correction for ‘selection bias’, i.e., adjusted for the fact that SAT scores and family incomes are higher for students of private universities. Thus comparing ‘like’ with ‘like'—apples with apples—they found that attending to private universities does not confer a wage premium. 

It is by developing such causal techniques basing on a comparison of observed outcomes with counterfactuals–the ‘what if’ scenarios or potential outcomes that are not observed–Prof. Imbens and Prof. Angrist have eventually proposed effective mathematical and statistical methods to disentangle causal effects from messy observed data. They proposed a simple two-step process to estimate causal effects: First, use “instrumental variables” to mimic the threshold difference between the two separate groups meant for comparison; and second, while evaluating the effects, explain the assumptions needed to be factored in—developing the Local Average Treatment Effect (LATE). Together, these two steps are supposed to boost transparency and credibility of empirical research. And these techniques were replicated by many researchers in different contexts validating the effectiveness of their contributions.   

Now, one may wonder if— in the present age of data science and AI—these econometric techniques would still hold good in interpreting causal relations. But Prof. Angrist argues that big data may help in ‘curve fitting’—in showing a pattern—but does not throw light on causation. Since it does not explain the reasons behind the pattern nor do offer any scope for evaluating counterfactual scenarios, we still need econometric tools.   

As the Nobel Committee observed, these three brilliant econometricians, laying foundation for the “design-based approach, have radically changed how empirical research is conducted over the past 30 years”, besides paving the way for a great improvement in our ability to answer causal questions that in turn enhanced effectiveness of economic and social policies.   

**