News & CommentaryArchive
Feb 08, 2012
Excerpt from Interview with David McKenzie, Part I
For my upcoming book, Experimental Conversations, I’m interviewing a variety of economists conducting field experiments on poverty interventions. Here’s Part 1 of 2 excerpts of my interview with David McKenzie, an economist at the World Bank (and now prolific blogger at the World Bank’s Development Impact blog) who has been studying the dynamics of microenterprises. David’s goal is to better understand how profitable these firms are, why they don’t grow, and how we may be able to help put them onto a growth path.
Tim Ogden: Tell me about where this research into microenterprises, entrepreneurs and returns to capital started.
David McKenzie: We started in Sri Lanka attempting to test this idea that people may be stuck in poverty because if you invest small amounts of money, the returns on those small amounts of money are just very low. That would help explain why, when there are so many microenterprises, so few of them grow, and so few of their owners seem to climb into the middle class.
So we gathered a sample of microenterprises and randomly assigned some of them to receive a cash grant larger than the lump sum they could typically accumulate on their own, either $100 or $200, and some to not receive a grant. Then we compared them and looked at their performance over the next 2 ½ years. In Sri Lanka we got these very surprising findings. We had very high increases in profits for male-owned businesses when we gave them grants. Their profits showed a real return on capital of about 11% per month which is incredibly high. But there were 0% returns to giving these grants to women.
TO: But returns by gender wasn’t what you started the project to look at it, was it? It came out of data to try to measure returns for microenterprises in general?
DM: Right. In Sri Lanka, we had a sample of men and women, but gender wasn’t the principal focus of it. When we went into Ghana we wanted to see if this finding would hold up in another setting and in particular in a different context. In South Asia we know that women have very low labor participation rates, but in Ghana women are actually the majority of small business owners. We purposely chose Ghana because it’s a country with this long history of women running businesses and is more gender equal than most countries in terms of labor force participation. In Ghana there’s this feeling that women can work and can do things. So that’s why we chose Ghana.
We replicated the experiment there and we gave these grants of about $120. We gave half of the randomly assigned grant recipients the grant in cash and half of them got the grant in-kind. With the in-kind grants we said to the owners, “We’ll go with you and buy you something for your business, you tell us what to buy.” The basic result in Ghana was again big returns on capital. On average for both men and women we find big increases in profits when we give the in-kind grants. Their profits went up about 30 Cedi a month, about a 20% return per month on the grant.
When we gave the women cash though, there was no increase in business profits. And when we look more closely at the data, even for the in-kind grants, the increase is really only happening for the top 40% of women. So women who were starting off in these subsistence businesses earning a $1 a day had no benefit in terms of business outcomes from getting more capital. The grants all seemed to get spent on household needs. For men across the board, with the in-kind grants we see these large effects on profits, and while noisier, there also seem to be some benefits to the men of cash grants. For the top 40% of women we also get big increases in profits if we push them to invest in their businesses via the in-kind grant but not if we just give them cash without any restrictions.
TO: The result that these grants only matter for the top 40%, looking at in hindsight, was that predictable? Were those 40% in industries that one would have expected to see higher returns?
DM: They seemed to not have been differentiated much in terms of the industry they were in from the women in the bottom 60%. The difference was in baseline profit levels.
The bottom 60% averaged about a $1 day in profits but the top 40% were earning about $5 a day in terms of profits. So it’s quite a difference in terms of size of profits. These women were better educated, they were wealthier to start with. They were more likely to have gone into business for business reasons rather than other reasons.
So there’s something different about the types of women who are running those businesses and were able to generate high returns from the grant, but it’s not that they are choosing different industries.
TO: In Sri Lanka there appears to have been a significant industry-related issue. The women were primarily concentrated in industries with low returns to capital like lacemaking.
DM: In Sri Lanka we found there to be two main reasons for the gender difference. The first seemed to be this industry difference, women who were in traditional female industries had the lowest returns. But even when we looked at retail trade where both men and women worked, men were doing better. The second thing, though we could only look at this suggestively because we hadn’t set out to look at this in the first place and our sample sizes became smaller, but it seems there was something to do with intra-household cooperation. Women who said their husbands were more supportive of their businesses seemed to be doing better. The data seem to suggest women perhaps were not investing optimally in their business for fear that the proceeds would just get captured by others. It’s very hard to distinguish how that happens—who is capturing that profit: people inside the household, people outside the household, or even whether it was captured from themselves. Maybe they were thinking, “I don’t trust myself not to spend loose cash,” so they overinvest in equipment and don’t buy enough working capital.
TO: You’re now looking in Sri Lanka at helping women shift industries. How are you randomizing that? Are you trying to judge the issue of intrahousehold cooperation?
DM: What we’ve done there is we’ve taken a group of 600 women who currently have low profit businesses clustered in the types of industries which are mostly female dominated. We’ve used our three years of survey data to find out what type of industries women seem to be earning more in and have more prospects for growth. Then we’ve given them a five day business training course, based on the ILO curriculum, and as part of that training we’ve also provided them with this information about what the opportunities are in different industries in terms of how much money women like them earn. Those tend to be industries where both men and women work but we also tell them about some more female dominated industries that seem to have higher prospects. So for example, bakeries seem to do pretty well in comparison to making lunch packets for neighbors. Baking cakes is something not as many people do—you need a little more capital to be able to do it—but the returns seem to be pretty high. So we’re giving them that information and seeing whether that will push them into higher potential return industries. Some of the sample also received training with additional capital grants to see if you need additional capital in addition to the information and the training to switch industries and start achieving higher levels of profit.
TO: This is particularly interesting because it’s a vexing problem in developed economies too that most people starting small businesses go into industries with low barriers to entry, low capital requirements and low returns.
DM: Exactly. What we’re doing also in Sri Lanka is trying to understand what it really takes to make this jump to hiring employees. That’s the big distinction between being self-employed and starting to grow a business: when they start hiring workers outside of the family. So we’re doing this with men also. We have a sample of just over 1500 men, where we’re trying to—well, if you think about in terms of a production function, we’re trying to hit A [total factor productivity], K [Capital] , and L [Labor] together and separately. But basically what we’re trying to do is look at what the constraint to growth is and do we need combinations of things to overcome those constraints.
So we’re giving some of them business training, we’re using a [commitment] savings program to build capital for some of them and then something we think is really innovative, that we haven’t seen anyone do before, we’re using a wage subsidy program. We subsidize, by giving them a grant of about half the market wage, these sole proprietors to hire their first worker for six months and then we phase the subsidy out. The idea is that this gives you time to learn whether your business has what it takes to support an additional worker and whether you have what it takes to manage a worker. It also sort of subsidizes that training period for both the owner and the employee. So for instance we talked to a guy who runs an aquarium for tropical fish. He says it takes him a month of having somebody work with him before he can be trusted to be left alone with the fish and six months to get him up to level that he can produce at an export standard. So we’re subsidizing that training process, but he’s likely to keep the worker on once the worker is trained—at least we think so.
So we’re trying to learn what is really going on that’s keeping these businesses from growing and hiring. Is it a labor market matching problem and the need to learn about your ability as an employer or is it a credit constraint problem or is it a human capital and business skills gap? So we have people who have been randomly chosen to receive one of these programs—the skills training, the employee subsidy or the savings—or a combination of the three. We’ve been doing this for about 2 years now and because of the possible complementarity of these things—you might need capital and training—we’ve staged the interventions. We had people who had nine months in the matched savings program to build up some savings, and then they got the business training or the wage subsidy treatment, and then the 9 months when they could use the wage subsidy and then you have to wait to see what happens after that. Over six months to a year is when we’ll have data that start to give an answer to whether this worked, and what combinations mattered.
I think a lot of us who have been working on microenterprises think these are the key critical questions. What are the constraints to getting some microenterprises to grow, to have a level increase in profits? With grants and microfinance it seems like you can lift people a little out of poverty; you can raise their incomes a little bit but then they level off. We’ve done these grants and we’ve found they lift profits at least in male businesses and they stay higher for three years at least, so this one off grant does at least semi-permanently raise their incomes but it’s just a level increase. It doesn’t shift their growth patterns. They don’t get better and better over time, they just move up a level and reach a new equilibrium almost.
Can we put them on a growth trajectory? I think its still valuable on a mass scale if we can lift people’s income for a long period of time with these one-off interactions but we’d like to know whether we can alter the growth pattern at least for some of them.
TO: Lots of people in aid would love to find a one-off interaction that has 3 year effects, but it still doesn’t get us where we actually want to be.