by Dean Karlan
Why Isn’t Microcredit More Popular?
If microentrepreneurs around the world can earn high returns like their counterparts in Sri Lanka, then it’s looking even better for microcredit. But hold on a moment. This apparent profusion of golden-egg-laying geese is actually a profound puzzle. If returns really were so high, traditional economics would expect people to funnel every available dollar toward their outrageously profitable businesses. Microentrepreneurs should be knocking down lenders’ doors.
Trouble is, they aren’t.
In the part of southern Sri Lanka where de Mel, McKenzie, and Woodruff conducted their study, microloans were widely available and reasonably cheap—interest rates were in the neighborhood of 20 percent APR, far lower than the returns the microentrepreneurs they had studied stood to make on average—so there seemed to be potential for profitable borrowing by these folks. But in fact there had been very little. Only one in nine had ever taken any kind of formal loan.
It’s not just the southern Sri Lankans who are strangely reticent. For, despite widespread enthusiasm in the developed world, it would appear that one very important group is not entirely sold on microcredit: the poor. At first blush the figure of 155 million clients worldwide is an impressive one, but let’s look a little closer. Compare that with the number of poor people. About half the world’s population—well over three billion people, or twenty times the number of microcredit clients—live on less than $2.50 per day. So even if every microcredit client was poor (and not all of them are), still less than 5 percent of the poor would be borrowers.
Realistically, 5 percent is a conservative estimate. Not every poor person is eligible for microcredit, or has access to it in the first place. But that number actually doesn’t seem too far off. A landmark study in Hyderabad, India, found a similar figure to the one from Sri Lanka—that somewhere between 10 and 20 percent of eligible borrowers choose to take loans. If people vote with their feet, microloans aren’t winning any elections. And that means the Ghanaian taxi driver we met at the start of this chapter is far from alone. What can explain this puzzle?
Maybe the eight in nine microentrepreneurs in the Sri Lanka study who failed to take out loans were simply ignoring the knock of opportunity. But let’s suppose they weren’t. There are still two plausible explanations for the low level of borrowing.
The first explanation is a mathematical quirk. Maybe the big average annual returns observed in Sri Lanka only tell half the story. After all, an average of 70 percent doesn’t mean that everybody saw a return of 70 percent on the nose. Let’s say that half had 140 percent returns and the other half had zero: Then the average return would still be 70 percent, but we wouldn’t be surprised to see the zero-percenters passing up microloans.
In fact, there was evidence to support this kind of story. Returns weren’t the same across the board; they were different for different types of people. Some of the differences were just what you’d expect. For instance, more-educated and smarter microentrepreneurs seemed to do better (although, statistically speaking, these weren’t robust results; the study wasn’t designed to get at such granular analysis with only 408 participants). An additional year of schooling increased returns by a quarter, and success on a simple test of cognitive ability was a strong predictor for high business returns.
But other differences in returns were surprising and troubling, most notably the disparity between the sexes. There was strong evidence that men had high returns from their businesses, and much weaker evidence that women did. Men in the study had average annual returns of about 80 percent, while women’s average returns were actually negative. Could it be that only men can run successful microenterprises?
Now, the general claim that women can’t succeed as microentrepreneurs seems patently false. Take a walk through a crowded market in just about any developing country: The voices you will hear calling out the prices of vegetables are women’s voices; the hawkers making their way down the aisles are announced by the rustling of their long skirts. Indeed, women are the lifeblood of microenterprise in much of the developing world. Moreover, much of the microfinance movement, from Yunus’s Grameen Bank on down, has emphasized lending to women—largely because they are believed to be more responsible as borrowers than men. But if their businesses are destined to be unprofitable, then clearly they are not the right people to take on entrepreneurial loans.
Is all the emphasis on women really misplaced? I hope not; but the findings from Sri Lanka force us to confront that uncomfortable question.
Fortunately, there is a second possible explanation for the low level of formal borrowing. Maybe people were driven away from microlenders by excessive restrictions on the use of borrowed money. In Sri Lanka as in the rest of the world, many microlenders require that loans be used exclusively to finance business activities. That means, for instance, that a tailor could take a loan to buy a sewing machine, but not to buy ready-made clothes for her children.
The problem was that the entrepreneurs in Sri Lanka didn’t want to finance just their business activities. They had other ideas.
De Mel, McKenzie, and Woodruff had designed their study to see just how far these other ideas went. The grants they made to entrepreneurs came in two flavors. Half were “in-kind”: Recipients could choose any business-related items up to the grant amount, and the researchers went with them to buy it. The other half were made in cash with no strings attached. Recipients were told they could spend the money on anything they liked.
The researchers found that recipients of no-strings-attached cash grants spent just over half (58 percent) on business purchases. The rest went toward savings, paying off debts, and everyday consumption items like food, clothes, medicine, and bus fare. If this is really how they wanted to spend their money, is it any surprise that they weren’t taking more entrepreneurial microloans?
Maybe we should think of this question another way: Why were microlenders making so many rules about how loan money could be spent? (The answer, ultimately, is that we as donors tell them to. Look at Kiva.org. We like the idea that our loans go to microentrepreneurial types. Would people give as much to Kiva if the appeals read: “Help fund this person’s loan so she can buy a new television or pave her floor”?) We will see more about the wisdom—and the futility—of these rules in the next chapter.
A Little Meat on the Bare Bones
Two years after we had worked with Credit Indemnity to measure the impact of credit on borrowers in South Africa, Jonathan Zinman and I had a chance to replicate our Credit Indemnity study with a Filipino lender that made entrepreneurial loans. It was a great opportunity to see whether a more traditional version of microcredit (i.e., one that targeted entrepreneurs) could produce the same positive impacts as its consumer-oriented cousin. Maybe we would even find evidence to support those uplifting success stories we read in the brochures: the bread baker whose business takes off when she buys a new oven, FINCA’s Mrs. Potosí and her thriving sweater concern.
As is the case with much of IPA’s microfinance work in the Philippines, we have microfinance guru John Owens to thank for introducing us to our partner. He led us to Reggie Ocampo, president of First Macro Bank, a lender in and around the capital city of Manila with some seven thousand clients.
Much more so than Credit Indemnity, First Macro sits squarely in the microcredit universe. It lends exclusively to entrepreneurs, and loans are supposed to be spent solely on borrowers’ businesses. Most of its clients have no formal employment, no credit history, and no collateral to secure their loans. And while First Macro is a for-profit business, it does have an explicit social agenda. Its mission statement talks about “community development,” “customer-driven products,” and “sustainable growth.” First Macro’s loans are also about two-thirds cheaper, at 63 percent APR, than the ones we studied in South Africa.
But operationally there were a lot of similarities between the two lenders. Like Credit Indemnity, First Macro made loans to individual borrowers, typ
ically with maturities of a few months, and they fed information about each applicant into a computer program that instantly produced a basic creditworthiness recommendation. So it was fairly straightforward to adapt our earlier experiment.
The replication with First Macro worked almost exactly like the Credit Indemnity study. Zinman and I modified the creditworthiness software so that some first-time applicants whose scores were just on the edge (i.e., the “maybes,” who actually made up about three-quarters of the applicants) were randomly approved. Over the following two years, we surveyed everyone, including those who were rejected, to see how their lives had changed. Had the ones who got loans prospered?
Yes and no. Looking at all the applicants together, the results were unremarkable. Business profits were 10 percent higher for those who got loans, but statistically the increase was not significant, so we can’t say with any confidence that the change in profits had anything to do with getting loans.
Looking at specific groups of applicants, it turned out there were some striking things to say, after all—but they were not the things microcredit evangelists wanted to hear. Not everybody succeeded. First, as in Sri Lanka, men did much better than women. They saw three times the increase in business profits as their female counterparts. Second, better-off borrowers proved much more adept at putting their loans to work: for the (relatively) wealthier half of applicants, getting a loan led to a 25 percent jump in business profits, whereas for those less well-off, we could not say with confidence that the loans had any effect on profits at all. So poor women, the stock heroes of microcredit lore the world over, did not steal the show in Manila.
For that matter, there was something else about the story emerging from First Macro that didn’t square with conventional wisdom. The endpoints of the narrative arc matched—in general, businesses that received loans went from less to more profitable—but the middle part of the plot was a surprise. We had thought we might find an excuse to dust off those durable bromides about microcredit allowing enterprises to grow, spreading outward like great hydrangea blossoms, bursting with the life and vivid color of a magnolia in early spring, et cetera. No such luck. We found that where businesses actually improved, most did so not through a process of unfettered growth, but through pruning.
That’s right: Increases in profits were driven mostly by shrinking firms, not expanding ones. Applicants who (randomly) received loans consolidated and pared down their operations. They had fewer businesses overall, and the businesses they did have employed fewer paid workers. Costs fell and profits rose. It was that simple.
Simple, but also unexpected. After all, nobody pitches microcredit with stories about closing down businesses and laying off workers. But maybe they should try; if nothing else, they would have evidence to back them up.
Can Microcredit Transform Communities?
What we’ve seen so far suggests that at least some people can and do prosper from access to microcredit. But the stories one hears from staunch advocates—think again of the adapted proverb “Give a man a fish, he’ll eat for a day. Give a woman microcredit, she, her husband, her children, and her extended family will eat for a lifetime”—imply something much stronger: not only that a huge breadth of individuals around the world can benefit directly from these loans, but that the rising tide lifts all boats. The great promise of microcredit is that it can be plunked down almost anywhere and be expected to pull entire communities up out of poverty.
One way to find out whether this is true is to go ahead and try: See what happens when microfinance arrives in a community for the first time. In 2005, Abhijit Banerjee, Esther Duflo, Rachel Glennerster, and Cynthia Kinnan, four economists from J-PAL and IPA, converged on Hyderabad, India, to conduct an RCT. They partnered with Spandana, an Indian microlender serving some 1.2 million clients with group loans. At the time, Spandana was planning to expand by opening branches in new neighborhoods. Working with the researchers, they identified about a hundred neighborhoods where microloans were not yet available, and randomly selected half to receive a branch during the following year.
In late 2007, about a year after the branch openings, the researchers surveyed widely in all hundred neighborhoods. Where branches had been built, they spoke not only with those who had borrowed, but also with those who hadn’t. They were interested in the experience of the community as a whole—not just of the go-getters who were first in line at the new branches.
The first thing they noticed was that there actually weren’t all that many go-getters. Just as we saw earlier in the Sri Lanka study, fewer than one in five people who met Spandana’s eligibility criteria were sufficiently enticed to come in and take loans. Fewer still invested those loans in a microenterprise. In fact, the most common reason for borrowing was to pay off other debt—usually of the high-interest, moneylender variety.
In light of those facts, maybe it is not so surprising that, despite the shiny new Spandana branches on their streets, communities were not transformed overnight. The surveys found no marked changes in women’s empowerment, children’s school enrollment rates, or spending on health, hygiene, and food. Another way to see this was by tracking the total amount households spent each month—on everything from dinners to diapers, school fees to cigarettes. A year after the branch openings, total expenditures hadn’t increased. It looked like people were, on the whole, no wealthier than before.
So for poor communities in Hyderabad, the introduction of microcredit didn’t mean instant prosperity for all; but this was not the whole story. As with First Macro in Manila, the interesting dynamics here were below the surface. They had to do with different kinds of people and the different ways they responded to increased access to credit.
The researchers broke down the residents of the hundred Hyderabad neighborhoods. First they separated out everyone who already owned a business. Then they used a model to predict whether a person was likely to start a new business based on some demographic information about his or her entire household—how much land it owned, how many working-age women it included, and whether the wife of the household head was literate and had a paying job. They used the model to split the remaining people into two groups according to the strength of their entrepreneurial bent. Once each person had been labeled as an actual entrepreneur, a likely entrepreneur, or an unlikely entrepreneur, the researchers could compare the groups to see how credit impacted each of those groups.
This three-way division gets at the heart of our question. Do the poor share a common and equal capacity to exploit microloans for their own—and their families’, and their communities’—benefit? Or does that ability belong to some more than others?
Comparing the three groups side by side, the differences were striking. And they told a coherent story.
Business-minded folks did well. Actual entrepreneurs tended to funnel money into their existing enterprises. Likely entrepreneurs cut back on consumption—especially consumption of so-called “temptation goods” like alcohol, cigarettes, lottery tickets, and roadside cups of tea (the Indian equivalent of Starbucks)—and ramped up spending on durable goods. They bought exactly the kinds of things you’d need to start a business: sewing machines if they were tailors, ovens if they were bakers, refrigerators if they were grocers.
All this business-related spending meant people were building and fueling economic engines. Despite the researchers’ finding that people were not wealthier on the whole, they appeared to be heading in that direction. And the cutback in spending on temptation goods suggested that, with their entrepreneurial dreams now in reach, people were making smart sacrifices to achieve their goals. So far, the classic story about microcredit was safe.
But the unlikely entrepreneurs threw a major wrench into the works. They didn’t buy durable goods or invest in businesses; they just consumed more. More of everything, from clothes to food to cigarettes and cups of tea. And at the end of the day, they weren’t any wealthier than when they started. All they had left was their obligation to Spa
ndana. So they wound up looking more like characters from a cautionary tale about, say, credit card debt than like the inspiring figures of microcredit literature.
The Means, Not the End
Now, we really need to be clear about the evidence we’ve reviewed in this chapter. It does not mean that microcredit is a failure, or that the enormous amount of enthusiasm it has generated is necessarily misplaced. All it means is that the jury is still out. Since the initial trials haven’t come up all roses, the burden shifts slightly on to microcredit’s advocates to make a strong case for it, and on to researchers to dig deeper and learn in what contexts it works best—or works at all. No single study, conducted in one place and at one point in time, can generate sufficient evidence to make worldwide prescriptions. One of the biggest challenges in development is to replicate evaluations in enough places and contexts that we finally arrive at universal lessons. This challenge is part of what motivated me to found IPA, an organization dedicated to the hard, slogging work that will ultimately bridge the gap between single studies and comprehensive, consistent evidence that we can take to the bank, so to speak.
The punch line is that what we’ve learned in this chapter about the limitations of microcredit is not a tragedy! It just means that not everybody is a born entrepreneur—or a born microcredit client—any more than everybody is a born fisherman. In the next chapter, we’ll have more to say about why.
So the problem with microcredit is not microcredit. The success of microlenders, as viable businesses serving the poor, is genuinely impressive. And, more important, thanks to the explosion of the industry in the past three decades, millions of people around have more choices than they did before. These are truly great things.