Poor Economics

by Abhijit Banerjee

  Some types of risk ought to be easier to insure than others. Consider weather, for example. A farmer should value a policy that pays him a fixed amount (based on the premium he paid) when the rainfall measured at the nearby weather station falls below a certain critical level. Because no one controls the weather and there is no judgment to be made about what should be done (unlike in medical care, where someone must decide which tests or treatment is needed), there is no scope for moral hazard or fraud.

  Within health care, insuring catastrophic health events—major illnesses, accidents—seems much easier than covering outpatient care. Nobody wants to have surgery or chemotherapy just for the heck of it, and the treatment is easily verified. The danger of overtreatment remains, but the insurer can cap what it will pay for each treatment. The big issue that remains is selection: The insurance company does not want only sick people signing up.

  To avoid adverse selection, the trick is to start from a large pool of people who came together for some other reason than health—employees of a large firm, microcredit clients, card-carrying Communists . . . and try to insure all of them.

  This is why many microfinance institutions (MFIs) thought of offering health insurance. They have a large pool of borrowers who could be offered insurance products. And because catastrophic health problems sometimes drive the otherwise highly compliant microcredit clients into default, health insurance for them would be a little bit of insurance for the MFI as well. Moreover, it would be easy to collect premiums from the clients, since loan officers already meet with them every week—in effect, they could just fold the premium into the loan.

  In 2007, SKS Microfinance, then the largest microfinance institution in India, introduced “Swayam Shakti,” a pilot health-insurance program offering maternity, hospitalization, and accident benefits. It was made mandatory for the groups to which it was offered to avoid adverse selection. To deal with the potential for fraud, benefits were capped and clients were strongly encouraged to use those hospitals with which SKS had a long-term networking arrangement. To sweeten the deal, clients who went to these hospitals were offered a “cashless facility”: They would not need to pay anything as long as their treatment was for a covered illness—SKS would pay the hospitals directly.

  When SKS first introduced the product, the company tried to make it mandatory for its clients. But the clients rebelled, so SKS decided to make the product mandatory only at the first renewal. The result was that some clients decided not to renew the loans, and SKS started losing clients in the areas where they were offering the insurance. After a few months, renewal rates for SKS loans had fallen from about 60 percent to about 50 percent. A CEO of a competing microfinance institution was asking us about our work with SKS, and when we said we were working on evaluating the impact of offering mandatory health insurance to microcredit clients, she laughed and said, “Oh, I know the effect! Everywhere SKS made this product mandatory, we got many more clients. People are leaving SKS to join our organization!” About one-fourth of the clients, eager to continue borrowing from SKS while avoiding being insured, found a loophole. They prepaid their loan just before the end of the one-year premium. This way, when they renewed their loan, they still technically had coverage, and therefore they did not have to pay the new premium. Faced with this resistance, SKS decided to make the product voluntary. But a voluntary product taken by only a few clients is again susceptible to adverse selection and moral hazard. The charges per covered client exploded, and ICICI Lombard, the company on behalf of which SKS was offering the insurance, decided it was losing money and asked SKS to stop insuring new clients. Other organizations attempting similar schemes have encountered very similar problems with client resistance to mandatory enrollment.

  Micro health insurance is not the only form of insurance that has run into trouble. A group of researchers, including Robert Townsend, our colleague at MIT, tried to measure the impact of access to a very simple weather insurance scheme. Much like the one we described above, it pays a given amount of money when it rains less than a specific amount.19 The product was marketed in two regions in India—Gujarat and Andhra Pradesh—both dry and drought-prone. In both cases, it was sold through a well-respected and well-known microfinance organization. The company tried various ways to offer and present the insurance to farmers. Overall, the sign-up rates were extremely low: At most, 20 percent of farmers bought some insurance, and that level of sign-up only occurred when someone from those very well-known MFIs went door to door to sell the product. Moreover, even those who bought some insurance bought very little: Most farmers purchased policies that would cover only 2 percent to 3 percent of their losses if the rains did fail.

  Why Don’t Poor People Want Insurance?

  A first possibility for the low demand for insurance is that the government has spoiled the market. This is the familiar demand-wallah argument: When markets do not work, overprovision by the government or international institutions is probably to blame. The specific argument is that when disaster strikes, these kindly souls step in to help, and as a result, people actually don’t need insurance.

  It is true that during bad monsoon years, Indian districts compete to be designated “drought affected” because this opens the door for government help. Jobs are provided on government construction sites, food gets distributed, and so on. But it should be clear that this is a very small part of what the poor need. For one thing, the government intervenes only in cases of large-scale disasters, not when a buffalo dies or someone is hit by a car. And even disaster relief is, in most cases, vastly insufficient by the time it gets to the poor.

  Another possibility is that the poor do not understand the concept of insurance very well. It is true that insurance is unlike most transactions that the poor are used to. It is something that you pay for, hoping that you will never need to make use of it. When talking to SKS clients, we met many people who were upset when their health insurance premiums were not reimbursed even though they hadn’t made any claims over the past year. It is certainly possible to explain the concept of insurance better, but it is hard to imagine that a population that ingeniously found a loophole in the SKS system couldn’t figure out the basic principle of insurance. Townsend, as a part of his effort to sell weather insurance, carried out an exercise to figure out whether people understand how the insurance works. While visiting each farmer, the salesman read aloud a brief description of a hypothetical insurance product (temperature insurance) and then asked the potential client several simple hypothetical questions about when the policy would pay out. The respondents had the correct answers three-fourths of the time. It is not clear that the average American or French person would do much better. It is therefore no surprise that the attempts to explain the rainfall insurance product better had no impact on farmers’ willingness to purchase.20

  The farmers were able to understand the main concept of insurance and how it functions, but they were simply not interested in buying it. They were, however, swayed in their decision by relatively small things. A simple home visit, without any particular effort at marketing, raises the fraction of people who buy weather insurance by a factor of four. In the Philippines, households that were randomly selected to complete a baseline survey containing many questions on health were more likely to eventually subscribe to health insurance than comparable households that had not completed the baseline survey. Presumably, answering all these questions about the possibility of health problems had reminded them of what could happen.21

  Given the very high stakes, why aren’t poor people more enthusiastic about the advantages of being insured, even without these little nudges?

  The key problem, we think, is that because of the problems we mentioned earlier, the type of insurance the market can offer only covers people against catastrophic scenarios. This creates a number of issues.

  Credibility is always a problem with insurance products: Because the insurance contract requires the household to pay in advance, to be repaid in the fu
ture at the discretion of the insurer, the household must trust the insurer completely. In the weather insurance case, the team marketing the product sometimes went with someone from Basix, an organization that the farmers know well, and sometimes they went on their own. They found that the presence of a member of Basix had a fairly large effect on sign-up rates, suggesting that trust is an issue.

  Unfortunately this lack of credibility may be endemic, given the nature of the products and the way insurance companies react to any possibility of fraud. In winter 2009, we visited some of the SKS clients who had decided not to renew their health insurance. One woman said that she decided not to renew after SKS refused to reimburse her when she went to the hospital with a stomach infection. Since the policy covered only catastrophic events, a stomach infection, horrible as it can be, did not qualify. But it was not clear that she understood the distinction—after all, she went to the hospital and was treated there. She also talked about a woman from another borrowing group (like most MFIs, SKS has its clients organized into groups) whose husband died of a severe infection, but not before his wife spent quite a bit of money on medicines and doctors. After his death, she submitted her bills to the insurance company, but the company refused to pay up on the grounds that he had never spent a night in the hospital. Appalled by the incident, an entire group of women decided to stop paying the premium. From a purely legal point of view, the insurer was clearly within its right to refuse payment. On the other hand, what could be more catastrophic?

  Weather insurance has many of the same problems. The crop may have dried up and the farmers may be starving, but if the rainfall is above the cutoff at the rainfall station, no one in that area will get any payment. Yet there are many microclimates: In any year when the average rainfall in the area is just above the drought cutoff, many individual farmers must face droughtlike conditions, just by the laws of chance. It is not going to be easy for suffering farmers to accept the verdict of the weather station, especially in an environment where corruption is not unknown.

  The second issue is the problem of time inconsistency, which we already encountered in our chapter on health. When deciding whether or not to buy the insurance, we need to do the thinking now (and pay the premium), but the payout, if any, would take place in the future. We have already seen that this is a type of reasoning human beings are particularly bad at doing. The problem is made even harder when the insurance is against a catastrophic event: The payout would take place not only in the future, but in a particularly unpleasant future that no one really wants to think about. Not spending too much time anticipating these events may be a natural protective reaction, and this may explain why people were more likely to buy insurance after they were forced to think about it by answering a survey.

  For these reasons, micro insurance may not become the next billion-client market opportunity: There seem to be deep reasons that most people don’t yet feel very comfortable with the kinds of insurance products that the market is willing to offer. On the other hand, the poor clearly bear unacceptable levels of risk.

  There is thus a clear role for government action. This does not mean the government needs to substitute for a private insurance market, but for a real market to have a chance to emerge, the government will probably need to step in. Private companies could continue to sell exactly the kinds of insurance they are currently willing to sell (catastrophic care with a strict cap, indexed weather insurance, and so forth). But for the time being, the government should pay a part of insurance premiums for the poor. There is already evidence that this could work: In Ghana, when weather insurance was offered to farmers with a large subsidy on the premium, almost all farmers to whom it was offered took it up. Because the fear of bad shocks leads the poor to costly mitigation strategies, subsidizing insurance could pay for itself in terms of higher incomes for the poor. In Ghana, farmers who had received cheap insurance were more likely to use fertilizer on their crops than those who had not received it, and they were better off as a result. They reported, for example, being much less likely to have missed a meal.22 It is possible that over time, as people start to see how insurance works and the market starts to grow, the subsidy could be phased out. But even if that is not possible, given the enormous potential gains that could be achieved if the poor did not need to be the hedge-fund managers of their own lives, this seems like a great place to use public funds to promote the common good.

  7

  The Men from Kabul and the Eunuchs of India: The (Not So) Simple Economics of Lending to the Poor

  The sight of countless fruit and vegetable sellers standing side by side on street corners is common to cities in most developing countries. Each of the sellers (usually a woman) has a small cart or just a sheet of tarp on the pavement on which she has piled tomatoes, onions, or whatever she happens to be selling. The vendors buy their stock in the morning from a wholesaler, usually on credit, and sell it during the day, reimbursing the wholesaler at night. Sometimes, the cart that they use to carry and display the vegetables is also rented for the day.

  This is the way many businesses in rich countries operate, too: They get a working capital loan to produce and purchase goods and then repay the loans out of their revenues. What is striking is how much the poor repay, compared to the rich. In Chennai, India, when the typical fruit seller reimburses the wholesaler at night for the 1,000 rupees’ ($51 USD PPP) worth of vegetables she got in the morning, she gives him 1,046.9 rupees on average. This interest payment is 4.69 percent per day.1 To see what this means, try the following calculation: If you borrowed 100 rupees ($5.10 USD PPP) today and kept it until tomorrow, you would need to repay 104.69 rupees. If you kept this amount a further twenty-four hours and repaid it the following day, you would need to repay 109.6 rupees. After thirty days, you would owe almost 400 rupees, and after a year, 1,842,459,409 rupees ($93.5 million USD PPP). So the equivalent of a $5 loan, if it goes unrepaid for a year, leaves a debt of nearly $100 million.

  These very high interest rates were the call to action for the founders of microfinance. For instance, Padmaja Reddy, the CEO of Spandana, one of the largest microfinance institutions (MFI) in India, told us that she got the inspiration for starting Spandana after striking up a conversation with a ragpicker in the city of Guntur, in Andhra Pradesh. She realized that if only the ragpicker could come up with the funds to buy one cart, she could be in a position to buy “scores of carts” in just a few weeks with the money saved from not having to pay the daily rental fee. But the ragpicker did not have enough money to buy a cart. Why, Padmaja asked herself, is no one lending her the money to buy one cart? According to Padmaja, the ragpicker explained that the bank would not lend to someone like her. She could have gotten a loan from a moneylender, but the rates would have been so high that it would not have been worth it. In the end, Padmaja decided to give her a loan. The ragpicker reimbursed it faithfully and flourished. Soon after, people were lining up at Padmaja’s doorstep for loans, and she decided to quit her job to start Spandana. Thirteen years later, in July 2010, Spandana had 4.2 million loan clients, with an outstanding portfolio of 42 billion rupees.

  The story Padmaja tells is not very different from that told by Muhammad Yunus, hailed as the father of modern microfinance: Banks are unwilling to touch the poor. Into this banking void step exploitative moneylenders and traders who charge outrageously high interest rates. Microfinance, in this narrative, is a wonderfully simple idea. Someone who is not out to make money off the poor can enter the market, charging the poor enough in interest to be financially sustainable, and perhaps make a modest profit, but no more. By the power of compounding, a small decrease in the interest rate can transform the clients’ lives. Consider the fruit sellers: Imagine they can get a 1,000-rupee ($51 USD PPP) loan, even at a relatively hefty rate of, say, 10 percent monthly. They can now buy the vegetables in cash, rather than on credit. In one month, they would each already have saved 4,000 rupees ($203 USD PPP) in interest paid to the wholesaler, more than enough t
o repay the microfinance agency. They could grow their businesses and escape poverty in a matter of months, at least in theory.

  Yet even this simple story raises questions. There are many fruit wholesalers in Chennai. Why didn’t one of them, or an enterprising moneylender, decide to slightly drop the interest rate charged to the women? That individual should have been able to capture the entire market, still keeping a reasonable margin. Why did the fruit sellers have to wait for people like Muhammad Yunus or Padmaja Reddy?

  In this sense, the advocates of microfinance are being too modest: They must be doing something more than introducing competition where there was a monopoly. On the other hand, they may also be too sanguine about the potential of small loans to lift people out of poverty. For all the individual anecdotes of fruit sellers turning into fruit magnates that can be found on the various Web sites of microfinance institutions, there are still many poor fruit sellers in Chennai. Many of them do not borrow from microfinance institutions, even though there are several in their town. Are they forgoing their tickets out of poverty, or is microfinance less of a miracle than we have been told?

  LENDING TO THE POOR

  Very few poor households get loans from a proper lending institution like a commercial bank or a cooperative. In the survey we conducted in Udaipur, in rural India, about two-thirds of the poor had a loan. Of these, 23 percent were from a relative, 18 percent from a moneylender, 37 percent from a shopkeeper, and only 6.4 percent from a formal source. The low share of bank credit is not due to the lack of physical access to banks, because a similar pattern occurs in urban Hyderabad, where households living below $2 a day primarily borrow from moneylenders (52 percent), friends or neighbors (24 percent), and family members (13 percent). Only 5 percent of their loans are with commercial banks. In all the countries we have in our eighteen-country data set, less than 5 percent of the rural poor have a loan from a bank, and less than 10 percent of the urban poor do.