by Dean Karlan
Group lending offers an organic solution to this problem. As it was with screening, the key here is that clients know one another better than the bank knows any of them. Group members can monitor one another because they buy from the same distributors and sell at the same outdoor markets and they see one another at church. So they hear about it when someone spends her loan money on a new TV (or a rice cooker). They know if someone suddenly starts taking days off from work. And since they all have something to lose (namely, the possibility of borrowing again in the future), each member has a material incentive to keep all the others in line.
Even without active prodding by other members, group lending clients feel pressure to perform because they risk damaging their social standing if they miss payments. Business relationships with suppliers and customers become strained. Access to community assistance in times of need can wither quickly and irreversibly. Earning a reputation as a deadbeat often proves very costly in the long run. People usually do their best to avoid it.
By backing formal financial arrangements with social currency, the group lending model finds leverage to keep borrowers on track.
What If Things Go Badly?
American banks resolve the third issue—how to claw back lent money if the borrower fails to pay up—through legal action. Regulations vary from state to state, but in general lenders are well-protected against default. They are empowered to seize collateral, garnish wages, and recover other assets if need be.
If laws are the teeth of the system, then credit bureaus are the jaws. It’s hard for delinquent borrowers in developed countries to drop off the map and escape their obligations. Our records, both good and bad, follow us as surely as our shadows, held there by our names, social security numbers, and addresses—things that are hard to shake off.
By comparison, it is very easy to disappear in Ghana. Jake once tagged along with a loan officer who went out to track down a client who had missed two payments. It was a dispiriting trip. They went to the stall in the market where the woman had worked. All they found was an empty plywood table. Then they trekked across the city to her home, which was locked up. A neighbor said the woman hadn’t been around for a couple weeks. “She might be at Cape Coast for a funeral,” the neighbor offered; but that was all the information they could get.
The loan officer said he believed that, with some sleuthing, the woman could be found. But would it be worth the effort? Her loan was only a few hundred dollars to begin with, and she had paid off most of it. A trip to Cape Coast, three hours away by bus, or more days spent zigzagging around Accra would leave him with less time to deal with the rest of his roughly four hundred clients. Getting the police involved wasn’t very appealing, either. It would take even more time, was not much more likely to succeed, and would be quite frustrating to boot. Had the bounty been bigger, he might have bothered, but in the end he couldn’t justify it. He let the woman go.
Microlenders around the world face the same situation. Delinquents are hard to track down regardless of where they live or how much they’ve borrowed; but because microloans are usually fairly small, individual microcredit clients typically don’t owe very much. Chasing down a single delinquent may not be worth a loan officer’s time, and even if the client is found, the loan officer has limited muscle to recover money. In a group, though, as long as the threat is credible that everybody will be barred from further borrowing if anybody is late without good reason, the bank can likely get the group to take up a small collection to make up the missing payment.
Again, this just amounts to passing the enforcement buck down to the borrowers. Upstanding clients don’t want to take a hit any more than the bank does. When they cover for one of their own, they do everything they can to collect. The beauty of it is that clients are often more effective collectors than loan officers. Not because they are necessarily more tenacious, but because, living and working side-by-side with their debtors, they are in a better position to do it.
Other Advantages of Group Lending
Beyond giving lenders in developing countries a way to answer the three big questions above, group lending also makes credit a more appealing business proposition by controlling lenders’ costs. Where meeting with individual clients is time-consuming and expensive, banks benefit by meeting with entire groups—or groups of groups—at once. It is not uncommon for a loan officer to hold a single biweekly repayment meeting for ten groups of a dozen borrowers each. The whole thing might take two hours, or one minute per client.
One reason why the repayment process can be streamlined so much is that the bank does not need to keep detailed records about each individual client. Since it cares mainly about grouplevel repayment, it can leave the members to sort out exactly who gave how much each week. That’s what typically happens at the big groups-of-groups repayment meetings: The group treasurer records each member’s payment in her notebook, then presents a single payment to the loan officer on behalf of everybody. The officer might ask to see the individual records if the group’s payment is short; otherwise he usually just moves on to the next group. In the aggregate, loan officers save a lot of time—and manage to serve a lot more clients.
Microcredit advocates tout (at least) two more benefits of the group model: the opportunity to integrate complementary interventions into lending programs, and client empowerment. A hundred-plus clients attending a biweekly repayment meeting is a captive audience. As long as they’re stuck sitting there, why not give them some business training (like FINCA Peru, whom I mentioned in the last chapter) or a class on nutrition? Some banks have successfully piggybacked auxiliary programs onto their credit programs.
The empowerment angle comes from the simple fact of clients coming together on a regular basis. They talk about their work and home lives, share knowledge, and help one another deal with problems. They inspire one another. Some microlenders—most notably, Yunus’s own Grameen Bank—use the social dynamic among borrowing groups to pursue an empowerment agenda explicitly. Grameen does it by having clients commit as a group to a list of sixteen life-affirming behaviors, from subsistence farming to children’s education. When the group signs on together, the reasoning goes, they will work hard to honor their commitment, not just out of respect for the behaviors themselves, but out of solidarity with one another.
The Problem with Group Liability
So, in a nutshell, that’s the good news about the theory behind group lending. It solves the three basic problems facing lenders (problems that, before Yunus’s bold experiment in Bangladesh thirty years ago, had proved intractable in the developing world), gives banks a way to streamline their operations, and serves as a vehicle for other socially beneficial endeavors.
Now for the bad news. For all their merits, these defining features of group lending are really just compromises—trade-offs that ease the burden on lenders so they can operate. Those burdens don’t just disappear. When banks cast them off, clients end up shouldering the extra load.
The most obvious way group loan clients pick up the slack is by covering one another’s late payments—as Mercy did, to the tune of a thousand dollars. But there are other, subtler ways that individual clients are taxed by the system.
One is simply a time issue. As we saw, holding big repayment meetings is a great boon to the bank, as it allows one loan officer to process scores of clients’ payments in a couple hours. It’s not so hot from the client’s perspective. Instead of making your payment at a teller window, which might take five or ten minutes, you have to wait two hours while your credit officer meets with other clients. Add in the travel time to and from the meetings, and you might be missing half a day of work every week or two. Over the term of your loan, that amounts to a lot of wasted time.
Another disadvantage of the group lending system is the fact that it quietly encourages the clients with modest needs to borrow excessively. Here’s why: Suppose that each client applies for her ideal loan amount. Now think about the group member whose loan request is
smallest. Because everyone else is more indebted than she is, she’s often taking on more than her fair share of risk by committing to cover for them. In the extreme, imagine a group with just two members—one borrowing ten dollars and the other borrowing a hundred. If the ten-dollar borrower is liable for the hundred-dollar loan, she’s getting a bad deal. Borrowers level the playing field by taking larger loans than they need, which leads to overindebtedness and, in the long run, trouble. (One potential solution is to make each borrower’s liability proportional to her share of the group’s overall debt. Some microlenders actually do this, but the vast majority that I have seen do not.)
When trouble does strike, a borrowing group becomes a precarious chain of dominoes. As more members miss payments and disappear, those left behind sink deeper and deeper into debt. Nobody—not even the best client—wants to be the last sucker standing. Eventually even the best borrowers owe more than they can give, and everybody stops paying. The result is a loss for the bank, which might have recovered at least some of the outstanding loans, and a loss for the good clients, whose reputation is ruined through no fault of their own.
Of course, some clients take a more proactive approach. Rather than waiting around for their coborrowers to put them in a bad situation, they pay off one last loan and get out. That’s what Mercy was doing. Some people go a step further and avoid microcredit altogether.
The irony is that these folks—the ones who don’t borrow for fear of being dragged down by the bad apples in their group—are precisely the people who should be taking microloans! By their very refusal to sign up, they demonstrate how much more confidence they have in their own ability to repay than in that of their fellow clients.
Less Is More: Simple Individual Loans Could Be the Answer
A version of microcredit that punishes or scares away people like Mercy isn’t ideal for banks. More important, it isn’t the best way to help the poor.
Ironically, the major obstacle to fixing microcredit is that, in the eyes of many, it doesn’t appear to be broken. Thousands of microlenders around the world, working exclusively within the classic model, have all the business they can handle and see almost no default. Influencers of global thinking from the UN to U2 have stamped it with their seals of approval, which sends money pouring into lenders’ coffers from governments, philanthropic foundations, and private donors.
Ask a microlender to change and they might reasonably fire back: Why mess with a good thing?
You mess with it because, if you’re losing clients like Mercy, it still isn’t quite right. The microcredit world is slowly coming to this conclusion on its own. The past decade has seen many variations on the plain-vanilla group lending scheme. Most visibly, the original architect of (and great advocate for) group-based microcredit himself, Muhammad Yunus, started tinkering.
His idea was to build a new version of microcredit that retained the benefits of group lending—the social dynamism of the borrowing group and the cost-saving practice of large repayment meetings—without overburdening good clients. In 2002 Yunus unveiled his creation and dubbed it Grameen II. On the surface, it looked a lot like its predecessor: Clients formed groups, took loans, and met weekly to make payments together. But behind the scenes there was a very big difference. Grameen II made only individual-liability loans. Clients were no longer forced to cover for one another.
There were concerns that the whole thing would flop. After all, group liability was, in theory, the glue that held the group lending model together. It provided a powerful material incentive for clients to screen, encourage, monitor, prod, and ultimately help one another to make their payments on time. What would happen without it?
As it turned out, there was no need to worry; Grameen II was a hit. Borrowers liked the new look of microcredit, and they proved it by signing up in record numbers. The bank’s client base, which had taken about twenty-five years to reach 2.1 million borrowers in 2002, jumped to 3.7 million by 2004.
Grameen’s makeover was an important and visible step, but it has not singlehandedly transformed the microcredit landscape. To make a global change, thousands of banks have to be on board. And while microlenders worldwide draw inspiration and encouragement from Grameen’s success, they still have legitimate doubts. They operate in vastly different settings and face different challenges; why should they assume that Grameen II, a solution tailored to the realities of Bangladesh, will work for them?
The Evidence
Fortunately, they don’t have to bank on an assumption. The success of Grameen II demonstrated that individual-liability microloans could work; that was enough to get people excited. Soon, microlenders began partnering with economists to nail down the important questions and pursue them in the field.
Could individual-liability loans attract more, or better, borrowers than group-liability loans? Would the loans have different default rates? Would clients behave differently under the different regimes?
These questions had to be answered before anyone could say what kind of lending was best. So in 2004 Xavier Giné of the World Bank and I went looking for a partner organization to put individual-liability loans to the test. Actually, we had been hunting for some time already, but hadn’t found any takers. It had surprised us how divided the world was. Lenders either did it one way or the other, and very few were interested in finding out rigorously which system worked better for them.
There was one potential partner in the Philippines who I thought might be game. Omar Andaya, president of the Green Bank of Caraga, a family-run bank in Mindanao and one of the fastest-growing rural banks in the Philippines, is a remarkable entrepreneur with an uncanny knack for charging ahead on growth and expansion. He is also an inveterate tinkerer. He is constantly intrigued and motivated by learning what really works and what doesn’t, both for his business and for his clients.
Sure enough, Omar was interested. We got to work on an RCT that would put individual- and group-liability loans head-to-head. It helped to break down the comparison into the three big questions that group lending was supposed to answer in the first place: (1) Will group liability attract better clients? (2) Will it ensure that clients use the funds how they said they would and put enough effort into their businesses? (3) Will it help us recover from clients in any case?
All three questions are essential, but they don’t come into play simultaneously. To get a good look at each one, we had to compare group- to individual-liability loans at a few different stages in the lending process. So the RCT we designed with Green Bank of Caraga was really two experiments rolled into one.
First, we worked with 169 existing borrowing groups on the island of Leyte in the central Philippines. We randomly selected half the groups to convert to individual-liability for future loans, and had the other half continue under group liability as usual. Whether they were selected for conversion or not, the day-to-day process of borrowing remained the same for everyone. Clients still met once a week and made a single payment as a group, even if the members were no longer forced to cover for one another.
Second, we marketed loans to potential new clients on the island of Cebu, some fifty miles west of Leyte across the Camotes Sea. When the study began in 2004, Green Bank was not yet lending in Cebu, but it had already identified sixty-eight communities where it planned to open operations the following year. We piggybacked on its expansion plan and had it advertise different types of loans in different areas. A third of the communities got plain-vanilla group-liability loans; a third got individual-liability loans; and the remainder got a hybrid—the group’s first loan would be group liability, and all future loans would be individual liability. As in Leyte, the individual-liability loans in Cebu didn’t look, on the surface, much different from the standard offering. Clients formed groups and made their payments at group meetings. The motivation for this setup, similar to Yunus’s in creating Grameen II, was that the social support of other borrowers and the operational benefits of group lending could still be beneficial, even in
the absence of group liability.
After months of monitoring, the basic story that emerged was simple: Dropping group liability lifts a burden off clients’ shoulders. Clients like that; their patronage proves it. In Leyte, the groups that converted to individual liability attracted more new members and had fewer dropouts than the ones that continued with group liability.
The move to individual liability also strengthened the social ties between group members in two ways. First, since members no longer had to dig into their own pockets to cover for delinquents, borrowers began to cut one another some slack. Clients were less likely to force one another out of groups. Second, the prospect of having to prod and even punish fellow borrowers, which had long dissuaded clients with strong social ties from signing up together, disappeared. People started inviting their close friends and relatives to join Green Bank.
As expected, individual-liability clients did less monitoring, less hounding, less collecting, and less meting out of discipline than their group liability counterparts; and there was some evidence that loan officers picked up the slack. Individual-liability borrowers were more likely to be thrown out of the group by bank staff, and loan officers in Cebu reported that weekly repayment meetings with individual-liability clients took about ninety minutes longer.
For their part, Green Bank’s loan officers were understandably apprehensive about the switch. First and foremost, by abandoning group liability they were giving up their first line of defense against client default. Second, it looked as if individual-liability loans would mean doing more dirty work. So it makes sense that they were reluctant to open individual-liability lending operations. Anyone who has lent money to someone for his rent, only to learn that he went out and bought a rice cooker, can sympathize.