Let's Read A Paper: The Miracle of Microfinance? Evidence From A Randomized Evaluation
Using microloans like a credit card
Microcredit—providing small loans to members of the global poor—was long heralded as the solution to global poverty. The praise hit its peak when Muhammad Yunus, founder of the first microfinance bank, was awarded the Nobel Peace Prize. Even today, when I pitch people on GiveDirectly, I’m asked why I’m not giving to microfinance charities instead. Don’t donations to microcredit organizations go farther? People invest the money, earn more, and pay back the loan—so one small donation to a microfinance organization can help maybe ten times as many people as an unconditional cash transfer.
The 2015 paper The Miracle of Microfinance?, by Abhijit Banerjee and Esther Duflo, is a report on one of the largest evaluations of microfinance so far, following up with almost seven thousand households in India. The global-poverty media often presents it as an anti-microfinance paper which shows that microcredit is useless. However, a close reading shows that that’s not the case. Microfinance is a useful tool which improves the life of the global poor; it’s just not a panacea. Disproving the overblown claims of microfinance advocates is far from saying that microcredit does nothing.
Banerjee and Duflo worked with Spandana, a microfinance organization. Spandana agreed to randomly expand into half of the neighborhoods they were considering expanding into and to delay its expansion into the other neighborhoods by two years. Thus, this randomized controlled trial is of marginal neighborhoods—neighborhoods we know were relatively low-priority for Spandana to expand into, because they hadn’t done it yet. If you looked at all neighborhoods, you’d expect microcredit to have a larger effect.
Only about a quarter of eligible households had taken out a microcredit loan1 eighteen months after Spandana’s expansion, which suggests that demand for microcredit is fairly low. Most eligible households do borrow, but they often borrow from informal moneylenders instead. That’s another reason to believe this study underestimates the effect of microcredit in many areas: people in the study already had plentiful access to credit.
The overall level of borrowing wasn’t different between treatment and control groups. However, treatment groups were more likely to borrow from Spandana and less likely to borrow from informal moneylenders. Informal moneylenders usually charge extremely high interest rates. Some people may have chosen to borrow from Spandana because its interest rates were lower. Other people may have “refinanced”: they borrowed money from Spandana at a lower interest rate to pay off their higher-interest informal loans. It also seems likely that the Spandana expansion actually increased the rate of borrowing and the study was just underpowered to detect the effect: remember, only a quarter of households had a microloan at all.
Treatment households don’t spend more money in general. However, they’re more likely to spend money on durable goods: “gold and silver, motorcycles, sarees (purchased in bulk, presumably mainly for weddings or as stock for a business), color TVs, refrigerators, rickshaws, computers, and cellphones” (page 48). They’re less likely to spend money on temptation goods: “alcohol, tobacco, betel leaves, gambling, and food consumed outside the home” (ibid). They also spend less money on parties and festivals.2
The obvious explanation is that treatment households use microcredit the way that people in the developed world use credit cards. When you want to buy something large—a car, a vacation, a wedding—you might take out a loan for it and then pay it back over time. If you had to pay cash for a car, you might spend the money on impulse and never be able to afford it. You can also enjoy your car now instead of having to wait years to save up.
For many people, microcredit actually ends up functioning like a bank account. The global poor save too little money for it to be worthwhile for large banks to offer them interest-bearing savings accounts, so they typically save by paying someone in the neighborhood to keep an eye on the money. Most microcredit organizations allow participants to take out larger loans if they successfully pay off smaller loans. If you pay off your loan and then immediately take out another one, mathematically this is equivalent to a savings account you have to pay for: in the long run it doesn’t matter if you got the money at the beginning or the end of your first month. Microcredit organizations may charge lower fees than local money guards. They also have better security and are less likely to flee in the night with the money. The loan officer will show up at your door and complain if you miss your payment, so there’s an incentive to save.
But people don’t think that microfinance will end global poverty by enabling poor people to buy color TVs and have a cheaper basically-a-savings-account. They think that microcredit will let the global poor start and invest in small businesses, which will make them and their families richer.
This is… sort of true? After eighteen months, the top 5% of businesses are earning much, much larger profits in the treatment than in the control group. After three years, the top 15% of businesses are much more profitable. But the average business in both treatment and control is still small and unprofitable. In fact, the new businesses founded because the owners had microloan access tend to be smaller and less profitable than the average business, which makes sense. If it were a very profitable business, someone would have started it already!
Again, this finding makes sense if you think about your own life. Most people do not want to start new businesses. But people in developed countries get to go work for Walmart or Amazon. Because there are fewer large businesses in developing countries, people in developing countries are more likely to have to start their own tiny stores that sell cigarettes and newspapers. But they don’t want to do it, aren’t any good at it, and have neither clever ideas of what to do with more money nor any motivation to put them into practice. Microloans enrich the small percentage of the population with entrepreneurial spirit, but have no effect on the wealth of the average person.3
The evaluation finds that microcredit doesn’t improve health outcomes, or make children more likely to go to school, or empower women, or anything like that. You would not expect it to. The existence of credit cards doesn’t have a huge effect on health or women’s empowerment either.
None of this is saying that microloans are bad. You would be pretty annoyed if someone took away your access to credit cards and car loans, not to mention saving accounts. Even if most people don’t want to start or invest in a business, they should be free to do so if they like. No one should be unable to expand their profitable business because they live in the wrong neighborhood in India. Lots of things are good without single-handedly ending global poverty. Nearly every good thing, in fact!
We can also see that GiveDirectly and microcredit solve different problems. Microcredit solves the problem of people not having access to credit, which they use basically the same way that anyone else uses credit: refinancing loans, making large purchases, and investing in businesses if they are a startup sort of person. In the vast majority of cases, people have about the same amount of money before and after taking out a microloan. Cash transfers solve the problem of people not having enough money by giving them money. If you want people to have more money, you can’t achieve it by giving them credit cards.
From any provider; non-Spandana microcredit providers were available in all neighborhoods. Nevertheless, the number of people who had a microloan was 46% higher in treatment areas than in control areas.
These are small festivals, not big life-cycle events like weddings that people might endorse spending lots of money on.
The Spandana study shows no increase in how much money people have at all, but in my opinion it was just underpowered to detect the likely effect of people having more profitable businesses.
Do those top 5% of businesses that benefit a lot create many jobs, directly or indirectly? I.E. does microfinance benefit a lot of people but economies of scale/division of labour mean that only a few people directly interact with it so they count as reaping all the benefit from the perspective of the study?
Also if what people need is banking should we just be providing banking services? Is that what mobile phone transfer systems like M-PESA in Africa do?