News & CommentaryArchive
Feb 13, 2012
Excerpt from Interview with David McKenzie, Part II
For my upcoming book, Experimental Conversations, I’m interviewing a variety of economists conducting field experiments on poverty interventions. Here’s the second excerpt from 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: Let’s talk about microenterprise profits. You’ve written a paper on how to measure profits of microenterprises. You found a lot of these small operations are profitable, but there’s a lot of question about that. Why is it so hard to measure profits?
David McKenzie: There’s a couple of things going on there. The first is how you value the time of the people participating in these enterprises. Our measure of profits includes any return to the labor of these enterprises. If you worked on this business and your business earned 3000 rupees this month, that includes a return to your own hours of work. If you start trying to value the opportunity cost of that labor and you calculate it at some sort of market wage rate quickly you’ll find that many of these businesses look unprofitable.
I think this is one of the key questions about how to think about microenterprises and what we should do to help them, especially when you are looking at women in these businesses. For these women there usually is no outside option to generate cash.
TO: The labor markets are so thin how do you assess the opportunity cost? Is that it?
DM: That’s one part of it. The second part of it, and that’s something that comes up in this profits paper, is that if you ask people detailed questions on revenues and detailed questions on expenses, there’s a lot of noise on each of those numbers. A small shopkeeper is buying stuff in one period and selling it in another period and you’re trying to match all those things together. When you do that, you’re going to find that revenue minus expenses is really, really, really noisy and a bunch of studies have found negative values on that measure.
Now what we do in the profits paper is try and better match expenses and revenue. Rob Townsend has done some work with his Thailand data to better understand whether you should you use accrual or cash methods for measuring these enterprises. Especially when you’re looking at a short time horizon [like how the business has done in the last month, or last three months] the mismatch between when things are bought and sold can make a lot of firms look unprofitable. But under a longer run view they would be profitable.
But when you start to ask people to recall lots of small transactions over six months or a year, that’s pretty hard to do. So in these studies of microenterprise profits we usually ask for recall over just a month span. But that’s where you get the mismatch between revenues and expenses. I think that’s why when you directly ask people about profits most of them say they are profitable.
TO: As long as you don’t factor in the cost of labor.
DM: Right. There’s this really nice paper by Shahe Emran and Joe Stiglitz on why it is that microfinance can get a woman to run somewhat profitable businesses with chickens and things but they never get those businesses to grow into something greater. And the whole thing is that the women in these Asian countries have no other options. When their time value is 0 they can do this but as soon as they have to hire somebody at market wage it becomes unprofitable to expand.
TO: In the profits study you asked people to use ledgers to record their costs and revenues, presumably to aid recall. But that seems like that would be a very helpful tool to the average small enterprise—it’s certainly part of the standard advice to new business owners: keep careful records. But you found that people didn’t typically use the ledger for very long and use didn’t seem to have much impact. Greg Fischer and Antoinette Schoar’s work in the Dominican Republic found not much impact from formal accounting training but some notable positive effects from teaching simple “rules of thumb.” How do you match up those findings?
DM: There’s a couple of themes there. One is that our study is from a general pool of microenterprise owners and not those who have already self-selected themselves or been selected by MFIs. So when we found that 50% of people will only keep up these ledgers, if all those people happen to also be microfinance clients then maybe the MFIs are doing a good job selecting for clients more inclined to do that. Secondly we weren’t giving them any training, we were just giving them these sheets of paper with five columns and asking them to write things down each day. Some people did and kept doing it, and others said there’s no real value to me and so they quit doing it.
With the rules of thumb, Greg and Antoinette are not finding it has much impact on ultimate business outcomes. They tell a nice story but if you look closely they have a bunch of sales measures. One of the six sales measures is significant at the 10% level—sales in a bad month. So it’s not clear that it’s really having a huge impact, even that training. What they’re pointing out is that it has more impact than formal training.
I think this is an issue with a lot of these experiments. The power these studies have to answer some questions is really low. And these profit measurement issues sort of feed into that. So if you look at the Banerjee and Duflo Spandana paper, for instance, there’s a huge amount of noise in their profits data. I did some calculations and I think it came out that they would need 2 million people to find an increase of 10% in profits given the take-up of microfinance and the noise in profit measurements Measuring these profits is crucial for our understanding for figuring out what makes sense but it’s incredibly hard to measure.
We have this other paper following up on measuring profits based on the work we’re doing in Ghana. There we used PDAs to do our measurements. Each wave after the first wave we put in the previous wave’s data and we checked their profits relative to last month and if the change was too large we would challenge them. So we would see that a business’s profits were 100 Cedi 3 months ago but this month the owner is reporting it’s 1000 Cedi. So we ask them, “Did we get things wrong?” Not implying that they’re lying to us, but that we made a data entry error. The remarkable thing to us is that in 85% of the cases they did confirm that their profits did change that much from one month to the next.
Part of it is just seasonality, but part of it is one of the things that comes up in Portfolios of the Poor. There’s a huge variation of incomes on a day-to-day and a month-to-month basis. You have good months and bad months. Some months you get sick and you don’t earn much, other months something else happens. It’s a huge challenge for trying to look at some of the impacts of our programs on profits if profits are jumping around this much. It’s not just all measurement error, some of it is general challenges that are facing the business.
So our solution to that is we can try to measure things more times. The current work in Sri Lanka has 11 waves of data on these firms so if we get one or two bad months we can average that out. In Ghana we have 6 waves. I’ve got a paper called “Beyond Baseline and Follow-up” where I’m trying to say that more people should be doing this. The standard approach to doing these experiments is to measure the baseline, run the program and then come back and do one more survey a year later or two years later. That works really nicely for things like health and education where the outcomes are highly correlated but it doesn’t work so well for things like business profits or consumption or things like that. The standard methodology needs to change.