Q: Could you explain the work you’ve been doing with formal and informal insurance in India?
I started by looking at the puzzle of why farmers, whose livelihoods are rainfall dependent, are not aggressively taking on rainfall insurance. As we gathered data, we started to tease out some of the ways the formal and informal insurance systems interact and how insurance impacts risk taking.
To put this in context, in Asian, African, and Latin American societies, informal networks are the most important way that people share risk, especially in rural areas. The way those work is, if you and I are part of a network, it’s understood that I’ll help you out if you’re going through some trouble, and you’ll help me out if I face bad shocks in a future year.
We know from the existing literature that farmers in developing countries are reluctant to purchase formal insurance contracts. Part of the reason is that they’re not entirely uninsured; it’s just that they rely on the informal networks.
Also, these farmers are often in an environment where there are very few consumer protections. It would be difficult for me to trust that an insurer will come back and pay up, if I have no recourse if they don’t.
We collaborated with the Agriculture Insurance Company of India (AICI), a government-run insurer that is probably the biggest agricultural insurer in the world, in terms of how many clients that they have directly on farms. We offered index insurance to farmers and hourly agricultural workers in 63 villages in three states.
Our project used a lot of door-to-door marketing and personal contact to convince farmers that a formal insurer is reliable and will be around if bad things happen. That makes things pretty expensive. To make it feasible at a large scale, even for a well-meaning insurance company, would require that a lot of very practical marketing, bureaucratic, and administrative challenges be solved.
Q: What was the formal insurance you offered?
We want to offer a product that lets farmers take on enough risk to allow for greater returns. At the same time, you don’t want to create moral hazard or adverse selection problems where once farmers take insurance, they take on too much risk or people who are more of a risk try to purchase it. Insurers don’t like that. Car insurance companies in the United States are continually worried about covering the riskiest drivers.
For the research project, we solved that by offering only index insurance, meaning what we insure is an event that affects the farmer but the farmer does not have control over that event. Rainfall insurance is an example. There is a centrally located rainfall gauge; if it shows drought conditions, there is an automatic payout.
We were really concerned about making the insurance very simple because we hypothesized that one reason people are reluctant to purchase insurance is that the contract is too complicated. For example, farmers are not used to thinking about millimeters of rainfall. If you start talking to them about if rainfall is up to x millimeters, you will get a big payout, between x and y, you would get a smaller payout, and above y, you get nothing, that may deter people from buying because they think that you’re trying to put something over on them.
Farmers care whether the monsoon arrives on time, so we defined payout on the basis of the expected onset date. If the monsoon is delayed by two weeks, you’ll get a payout. If it’s delayed by four weeks, you’ll get a larger payout. That’s really easy to talk about, and not easily manipulable.
Another benefit of this insurance is that you can make payouts early because you’re insuring against the first date of rainfall rather than waiting until the harvest. The harvest is when farmers have money anyway. If you can make payouts when they need money, they can use the money to make other investments. The same amount can be more valuable depending on the timing.
Q: Your paper seems to show that the farmers are making fairly sophisticated choices.
These are farmers whose entire income is coming from agriculture, so they cannot afford to not be very sophisticated about this. And it shows up in the results that we see. For example, even though index insurance solves the adverse selection/moral hazard problem, it introduces a new problem, which is that the rainfall gauge is located centrally. What the farmer cares about is the rainfall on his or her own farm and not the rainfall at the gauge. These two things are correlated, but not perfectly.
You might have a situation where at the gauge the rainfall is fine, so the insurance doesn’t make payouts, even though some farmers in outlying areas are experiencing a drought. In that design, we’ve made the worst case for farmers even worse than it was before. They face a shock, they need help but don’t get it, and on top of that we made them pay insurance premiums, so they are even poorer than they otherwise would be.
This imperfect correlation is called basis risk. To study whether basis risk is a problem, we randomly assigned rainfall stations at the village in some cases and in other cases at the block level, which is a significantly larger administrative district and where it’s normally placed. When the rainfall gauge is placed in the village itself, farmers are more likely to purchase insurance. So in all of their behavior, they do seem to be quite sophisticated.
Q: To put this in Robert Shiller’s terminology, is this an example where you can have a well-designed financial instrument that leads to a greater collective benefit?
I think so, but the situation is complicated enough that “well-designed” is a relative term. I think we can have better designed instruments. It’s complicated because if the monsoon comes and there is one heavy rainfall, and then farmers plant the crop, and then the rainfall disappears, that’s even more costly than if it doesn’t arrive on time because now, the seeds you planted go to waste. And so the optimal insurance contract covers that scenario. But if you start making the insurance contract complicated, farmers believe it’s going to be easier for the insurer to say, “Oh, you don’t really deserve payouts.”
There’s a balance that has to be struck, and I don’t think you can get to optimality here. That is partly because we’re in an environment where trust in a formal insurer is very low.
The system we used improves on the status quo, but lots of complementary, regulatory, bureaucratic, and institutional innovations have to come about to get anywhere close to optimal.
Q: Is there a tension between trying to mitigate risk on one level and encourage greater risk taking on another?
The way to think about it is we want people to move towards more highly productive or higher-growth occupations and technologies. There is a fundamental tradeoff—high risk offers the possibility of high returns; low risk offers low returns. The paradox is that the environment is already so risky that people find it difficult to take on more risk.
What we’re trying to do is improve the risk management institutions so that people find it easier to take on more risk, which on average, is going to pay off with higher growth. In some years and for some people, risk won’t pay off, but on average, people should have higher and higher income.
About 40% of the farmers bought insurance. Very important from a policy perspective, when the cultivators have insurance, they switch to the high-risk, high-return technology. Specifically, we saw rice farmers move away from the drought-resistant varieties that do well in bad years but produce less in other conditions in favor of other seeds that are better at returns, but also not as drought resistant.
That’s important. It suggests that farmers aren’t thinking of the insurance as a lottery ticket. Having insurance actually changes their production decisions.
Q: You also sold insurance to hourly workers?
Yes. To do it outside the experiment would require some legal and administrative changes—the AICL isn’t currently allowed to directly sell insurance to agricultural wageworkers. From the perspective of the legal systems in India, because the wageworkers don’t own land, they don’t have an insurable interest; therefore, they aren’t eligible for insurance.
However, if you think about it, if there’s a drought, the landed cultivators don’t have much to harvest, so they’re not going to be hiring as many laborers. So the landless agricultural wage workers need insurance against rainfall shocks even more than the cultivator because unlike richer landed people, they don’t have other ways of diversifying risk. They don’t have financial assets. They are less likely to have informal networks that extend to cities.
Imagine a situation where the cultivators in the village are offered insurance, but the laborers are not, which is how insurance is traditionally sold. If the cultivators take on more risk, then the environment becomes more risky for the laborers. If there’s good rainfall, the cultivators hire a lot of labor. If there’s bad rainfall, they even hire less than before because they’ve now switched to riskier, high-return technologies. It’s possible for the workers to end up worse off than if insurance did not exist at all.
What happened with the data is that, if you look across all 63 villages we worked in, average hourly wages increased. But they were much more volatile. If rainfall was good, the laborers were better off. If rainfall was bad, the workers were even worse off. And the fact that it’s more volatile makes the wageworkers worse off.
One clear policy lesson that comes out of this is that it’s a bad idea to prevent agricultural laborers from buying insurance. Even though the landless wageworkers are a lot poorer, so you’d expect them to purchase less insurance, they actually have just as strong a demand for insurance as cultivators do—about 40 percent.
Q: Why prevent selling to the landless?
The practical worry of selling index insurance to somebody without insurable interest is that, sitting in New Haven, I could go online and buy it as a bet on what kind of rainfall Tamil Nadu will experience this year. This is particularly worrisome if the Indian government is going to be subsidizing these insurance contracts, which we know is probably a good idea given people’s low demand, currently.
But I imagine there’s a happy middle between letting anybody buy the insurance online and keeping people who are agriculture laborers physically resident in that village from purchasing it.
The Indian government already requires some proof of identity and residence to access mobile bank services that extend to these villages, so it’s possible to solve that problem.
Q: How do the formal and informal markets interact?
One reason we did this research in India is because the informal network is really easy to identify and characterize, because it’s caste-based. People are just born into a caste, and it’s hard to leave. In fact, every third Bollywood film seems to be a Romeo and Juliet story where somebody tries to marry somebody from a different caste, and all the problems that causes. From a research perspective, we get a clean look at impacts.
We started out wondering whether the existing informal networks crowd out formal insurance. And when an informal network already provides coverage for rainfall, it does. But not all networks are able to do this. It’s only possible for informal networks to cover rainfall if they are sufficiently dispersed across geography; otherwise, everybody is being hit by the same rainfall shock, and they can’t really help each other.
However, when there’s occupational diversification among the members of a network, and farmers believe they have very good informal coverage because not all participants are likely to suffer the same shock at the same time, that actually allows farmers to purchase more formal insurance.
These interactions between formal index insurance contracts and informal networks are not as simple as one just crowds out the other. In some circumstances, those who are informally insured become even more likely, not less likely, to purchase formal insurance. The systems interact in complex ways.
Q: How applicable is this work to other parts of the world?
I think the general ideas here are actually quite applicable broadly, especially to agrarian economies in developing countries. The issue may not be the monsoon, but all harvests are weather dependent. In the Philippines, the big risk is typhoons. Elsewhere it will be something else.
There are going to be some key differences in the details, so before we take the results of one study and try to apply it more broadly, we do need to understand what those are. For example, depending on the characteristics of a network, they may be more or less able to indemnify against large area shocks. In India, the caste system links you directly to people you know well who are in the same caste and same village and indirectly to people who live very far away. So some of these castes are very good at indemnifying against large shocks.
In other countries where you don’t have such a salient identity for the network, that may not be as prevalent. The informal network may not hold back purchase of formal insurance as much just because the formal insurance is designed to cover area shocks, and the network isn’t really doing anything for that.
Q: What is the bigger-picture context for this work?
Douglas North won the Nobel Prize in 1993 for his work on the history of institutional development. He showed that as countries get richer they develop good institutions endogenously. Higher GDP per capita and higher productivity allows you to invest in your institutional quality, but it takes a long time.
As economies modernize, it’s important to understand that there is the possibility of clashes between these modern institutions and modern markets, and the traditional structures.
These traditional institutions are entrenched for a very good reason; they solve really important problems, and if we tried to get rid of reassuring networks to make room for formal insurance, at least in the interim period, you’re going to have people be worse off.
Q: What are your next steps with this line of research?
One of the things that would be really interesting to look at is now that formal insurance has penetrated some of these areas, how does that change the informal relationships? Have the frequency and nature of gifts and informal transfers changed at all? Are there spillovers from the formal insurance back to the informal network?
We also want to look at the different ways in which people manage risk. For example, right now in these areas, people not only use informal networks to manage risk, they also use self-insurance in the form of savings.
Beyond that, we’ll look at another form of risk mitigation, migration. We’ll continue to offer formal insurance in the same villages, but we’re also going to offer migration opportunities. Migration allows people to go to the city and diversify the sources of income. If agriculture is really badly hit, then a construction job in the city will help them to survive that year until the rains come back, and they can go back to agriculture.
Edited by Ted O'Callahan.