Today's Editorial - 16 December 2021
The credibility revolution
Source: By Pranav Patil: The Indian Express
This year’s Sveriges Riksbank Prize in Economic Sciences (the Nobel prize) has been awarded to David Card for his empirical contribution to labour economics and to Joshua Angrist and Guido Imbens for pioneering new methods to analyse causal relationships. The trio invented methods that have led to the so-called “credibility revolution” in empirical economics.
The scope of issues that economists examine has widened over the last three decades as the discipline began exploring answers beyond mathematical models and ideological discourse. Although neoclassical theories are elegant, questions were raised about their real-life evidence. Do economists have credible evidence such that policymakers and the public can take them seriously? Nobel laureates Abhijit Banerjee and Esther Duflo point out that the lack of evidence is one of the reasons economists were considered less credible.
For an evidence-based approach, understanding the causal relationship between different factors, therefore, becomes imperative. A classic example of a causal relationship is the impact of education on lifetime earnings — would one extra year of education increase earnings and by what magnitude? Economists embraced the experimental approach to tackle the credibility crisis and to assess the precise causal effect of policies. Like in medical science, development economists launched smaller randomised controlled trials in the hope of establishing causality between different variables and to investigate which policy interventions were effective.
In a randomised control trial, Duflo, along with others tested how monitoring and financial incentives reduced teacher absenteeism and improved learning in India. Based on experimentally derived causal inferences, economists can recommend more rigorous, objective and rational interventions to solve larger problems like poverty.
However, it is dreadfully challenging to conduct field experiments in many cases. They are expensive, time-consuming and ethically tricky. That is where the idea of “natural experiments” becomes illuminating which rely on random variation without any manipulation by researchers. Card and Alan Krueger designed their famous natural experiment based on the changes in the minimum wage in New Jersey and compared it with Pennsylvania, which has not experienced similar changes.
They studied employment in the fast-food industry in the two states before and after the wage changes in New Jersey. Contrary to the predictions of standard economic theory, they found a slight increase in employment in New Jersey compared to Pennsylvania. This finding was a massive blow to conventional supply and demand models. Angrist and Imbens have also designed many natural (quasi) experiments and have been developing a statistical toolkit to precisely estimate the causal effects of policies.
The study of causality is not novel to the research community. However, causal relations were not extensively studied with empirical methods in social sciences. Newton’s second law proposes that an object in uniform motion will continue its motion unless some external force is applied. Credibility revolutionists use this very principle to explain economic dynamics. Nonetheless, “causality is no correlation” is the most common catchphrase for these revolutionaries.
To distinguish causal links from correlation, economists rely on counterfactuals. For example, in the Card and Krueger study, they show that employment in two states had been evolving in parallel fashion before changes in the minimum wage. Based on that, they assume that employment would evolve similarly in both states without any intervention. Even if they did not observe what would have happened in New Jersey if there was not any intervention, they could observe the counterfactual situation in Pennsylvania.
Since economics closely deals with politics and the market, it is critical to identify which policy interventions are best (and cost-effective). It is worth considering two studies based on two flagship programmes of the Government of India — the Pradhan Mantri Gram Sadak Yojana and the Rajiv Gandhi Grameen Vidyutikaran Yojana.
The general assumption that policymakers make is that rural infrastructure programmes would increase farm and off-farm economic activities and reduce poverty. However, recent studies by Sam Asher, Paul Novosad, Fiona Burlig and Louis Preonas point out that while such programmes increase road and electricity connectivity, they do not cause significant economic development even four to five years after completion. It is thus meaningful to examine whether such interventions cause development, to what extent they increase welfare and where they fail.