The violent nature of the Mandsaur agitation has brought the agrarian crisis back to the forefront of the political discourse in India, and much of the discussion is focused on alleviating the farmers' immediate suffering. Crushing debt is forcing farmers to commit suicide every other day, compelling state governments to offer farmers debt waivers in a bid to prevent the situation from escalating into a political disaster. But while short-term measures ease the pain, policymakers need to focus on long-term measures that work.
After all, there is no shortage of ineffective policies—crop insurance schemes, for example—that are known to be marred by poor coverage and fraud. So what can policymakers do to ensure that the policies they promulgate actually work?
Some economists [conduct] experiments to evaluate the effectiveness of economic policies. Doing so enables researchers to reliably assess... what would happen if the policy was not in place and everything else was the same.
Experimental economics may provide some insight into this question. Randomised controlled trials (RCTs)—a central tenet of all experiments—were earlier considered to be the exclusive purview of the natural sciences. Often conducted in drug trials, such experiments involve separating a sample of people into treatment and control groups, where the treatment group is given the drug and the control group is given a placebo. Random assignment ensures that any significant difference in the outcomes of the two groups can be attributed to the drug and not any confounding factors.
Some economists have adapted the RCT into their own toolkit, conducting experiments to evaluate the effectiveness of economic policies, particularly in the area of development. Doing so enables researchers to reliably assess the counterfactual—that is, what would happen if the policy was not in place and everything else was the same.
Of course, before we jump the gun and start recommending RCTs for every policy impact evaluation, it is imperative we understand the limitations of the methodology. Conducting experiments in social science is notoriously difficult; the real world is not a laboratory where we can control all the variables at play. As such, RCTs can only be used in a small fraction of impact evaluations, and such studies can be very costly even for those.
There are also political deterrents that make such studies difficult. RCTs involve choosing treatment and control groups randomly. This means that some farmers may receive a purported benefit while others will be left out. We all know what happened when farmers in Uttar Pradesh received loan waivers and those in Madhya Pradesh did not. Needless to say, policies chosen for impact evaluations using randomisation should not be politically sensitive to the degree that they incite protests.
Given the plethora of limitations, where can RCTs actually help inform agrarian policy in order to significantly improve the lives of farmers in India? To answer this question we need to look in places where such impact evaluations have already been conducted. The World Bank conducted a meta-analysis of various impact evaluations which studied the efficacy of agrarian policies around the world. These studies, conducted between 2000 and 2009, were observational, experimental or quasi-experimental—they used statistical matching techniques to emulate an experiment. Not surprisingly, only 6% of the studies were actually experimental and used random assignment.
Needless to say, policies chosen for impact evaluations using randomisation should not be politically sensitive to the degree that they incite protests.
It is interesting to note, however, the type of policies these studies evaluated as it sheds light on the kinds of agrarian interventions which can be appraised using RCTs. Most of these studies focused on microcredit, marketing or farmer education interventions. This is not surprising as such interventions are geography invariant: they can be implemented consistently and uniformly in the treatment group of farmers irrespective of where they are located. Interventions based on irrigation, on the other hand, are difficult to implement uniformly as geography constrains the effectiveness of irrigation.
So what do these studies reveal? One study, conducted in Kenya in 2004, looks at the efficacy of a marketing intervention which is supposed to help farmers access the export market in order to raise incomes. Conducting a randomised controlled trial where some small farmers are provided access to export markets by a third party intermediary, the study found that while the intervention raised incomes in the short run, the benefits collapsed in the long run as the exporters chose not buy from farmers who did not meet EU guidelines.
Another study looks at the efficacy of group lending as a means of increasing repayment rates and reducing debt. Conducted in Philippines in 2009, the report finds, surprisingly, that solidarity lending does not provide any benefits over individual lending in terms of repayment rates. While this study did not specifically target farmers, this kind of an impact evaluation could be useful in the Indian context where farmers are debt-ridden. This is particularly true because solidarity lending is often considered a cornerstone of microfinance in India.
[W]e need to rigorously test policy choices before we implement them. Doing so may help improve farmers' lives, and that of all of us, in the long run.
Closer to home, there was a recent study which focused on the impact of providing rainfall index insurance on farmer's well being. Rainfall insurance provides payouts to customers if rainfall fails to meet a particular level. Since farming in India is often rainfall dependent, such a program could provide relief for farmers if rains failed.
A hundred villages in Gujarat were randomly sampled and partitioned into treatment and control groups. Farmers living in the villages in the treatment group were offered subsidised rainfall index insurance. Over the eight years in which the study was conducted, rainfall was poor which meant that payouts were greater than insurance premiums paid by the farmers.
Despite the intervention, there was no statistically significant difference in the well being of the farmers belonging to the two groups.
What does this tell us? Well, the results themselves are not directly useful to us; after all, what works in Kenya, Philippines or Gujarat may not work in Madhya Pradesh. However, the counterintuitive nature of some of the results presented above points to the fact that we need to rigorously test policy choices before we implement them. Doing so may help improve farmers' lives, and that of all of us, in the long run.