Poor Economics

by Abhijit Banerjee

In our eighteen-country data set, the majority of businesses operated by the poor have no paid staff, with the average number of paid employees ranging from essentially zero in rural Morocco to 0.57 in urban Mexico. The assets of these businesses also tend to be very limited. In Hyderabad, only 20 percent of the businesses have a dedicated room of their own. Very few have any machines or a vehicle. The most common assets are tables, scales, and pushcarts.

  Obviously, if these people had large and successful businesses, they wouldn’t be poor any longer. The problem is, notwithstanding the exceptional stories of the trash collector or Xu Aihua, the vast majority of the businesses run by the poor never grow to the point where they start having any employees or much in the way of assets. In Mexico, for example, 15 percent of those living on less than 99 cents per day had a business in 2002. Three years later, when the same families were visited again, only 41 percent of these businesses were still in operation. Out of those that were observed in both periods, one in five of the businesses that had zero employees in 2002 had one by 2005. But almost one-half of those that had one in 2002 had none by 2005. Similarly, in Indonesia, only two-thirds of the businesses of the poor survived five years. And out of those that did, the fraction that had one employee or more did not increase over the five-year period.

  Another characteristic of the businesses of the poor and the nearpoor is that, on average, they are not making much money. We calculated profits and sales of small businesses in Hyderabad: The average sales figure was 11,751 rupees ($730 USD PPP) per month, and the median, 3,600 rupees. The average monthly profit after deducting any rent paid but not including the unpaid time spent by household members was 1,859 rupees ($115 USD PPP) and the median, 1,035 rupees: It is as if the median businesses were generating just enough money to pay one member about 34 rupees per day, or about $2 USD PPP. In our Hyderabad data set, 15 percent of businesses had lost money in the last month, after subtracting rents. When we valued the hours spent by household members, even at the low rate of 8 rupees an hour (which would give someone close to the minimum wage for an eight-hour day), the average profits turned mildly negative. In Thailand, the median annual profit from a business of this scale was 5,000 baht ($305 at USD PPP) after deducting business costs but without accounting for family labor time. Seven percent of the household-run businesses had lost money in the last year, once again before deducting the value of family labor.6

  The low profitability of businesses run by the poor also explains why, as we saw in Chapter 7 (in our RCT of the Spandana program, for example), microcredit does not seem to lead to a radical transformation in the clients’ lives. If the businesses run by the poor are generally unprofitable, this may well explain why giving them a loan to start a new business does not lead to a drastic improvement in their welfare.

  The Marginal and the Average

  But wait. Didn’t we start off making the point that the return on investment in these small businesses is very high?

  What is confusing here are the two possible uses of the word return. Economists (for once, probably usefully) distinguish between the marginal return on a dollar and the overall return from a business. The marginal return on a dollar is the answer to the question “What would happen to your revenue net of all operating costs (but not interest costs) if you were to invest $1 less, or $1 more?”The marginal return is what is relevant when you ask whether you should cut your investment a little (or grow it a little): If investing $1 less allows you to borrow $1 less and therefore repay 4 cents less in principal and interest, you would want to do so if the marginal return is less than 4 percent and not otherwise. So when people borrow at a rate of interest of 4 percent a month, it must mean that their marginal return is at least 4 percent a month. The ability of the poor to borrow and repay and the high extra profits made thanks to the extra $250 in the Sri Lanka experiment show us that the businesses of the poor have high marginal return: Growing them a little would be worthwhile.

  The overall return on a business, on the other hand, is the total revenue net of operating expenses (the costs of materials, any wages you pay to your workers, and so on). This is what you can take home at the end of the day. You should look at the overall return to decide whether you should be in that business in the first place. If it is not high enough to cover the value of the time you are putting in the business, plus what it cost you to set up the business, and if you don’t expect things to improve dramatically, then you should shut it down.

  The paradox is explained by the fact that marginal returns can be high even though overall returns are low. In Figure 1 below, the curve OP represents the relation between the amount of investment in the firm (measured along the horizontal axis, OI) and its overall returns (measured along the vertical axis, OR), or what economists call the production technology. The overall return, for any invested capital of size K, is the height of the curve, whereas the marginal return is the change in height when you go from K to K+1. It tells us by how much the overall return increases when we increase the investment in the firm just a little bit.

  The curve in Figure 1 looks like the inverted L-curve we discussed in Chapter 1: The returns are first high, and then lower. OP is steepest when the investment is small (nearest to O) and slowly flattens out (as it gets closer to P)—which means that increasing the amount invested increases returns the most when the initial investment is small, and this increase eventually tapers off. In other words, the marginal return is high when investment is small.

  To see how this works, think of someone who has just started a shop in her home. She spends some money building shelves and a counter, but then she runs out of money and has nothing to sell. The overall return on her business is zero: not high enough to cover the cost of the shelves. Then her mother lends her 100,000 rupiahs ($18 USD PPP), and she buys a few packages of cookies to put on her empty shelves. The kids from the neighborhood notice that she has the brand of cookies they like and come and buy all of them. She makes 150,000 rupiahs. The marginal return is 1.5 rupiahs per rupiah of her mother’s loan, or 50 percent net, which is not bad at all for a week. But the overall return is nonetheless just 50,000 rupiahs—and that doesn’t cover the cost of her time and building the shelves and the counter.

  Figure 1: Marginal and Average Return

  Then our shopkeeper gets a loan of 3 million rupiahs and buys enough cookies and candy to fill up her shelves. The kids now tell their other friends and she sells a lot of her stock, but by the time all of the new customers get there, some of the cookies have gotten stale and can’t be sold. Still, she makes 3.6 million rupiahs in a week. The marginal return is now much lower than 50 percent—her investment was thirty times larger (3 million versus 100,000) but her revenues are only twelve times as much. Her overall return, though, is now a respectable 600,000 rupiahs ($107 USD PPP), enough to make staying in business a real possibility.

  This is exactly how it looks for many of the poor. The empty shelves, in particular, are not a figment of our imagination. The entire stock of a shop we visited in the outskirts of the town of Gulbarga in northern Karnataka, about a five-hour drive from Hyderabad, consisted of largely empty plastic jars in a dimly lit room. It did not take long to take inventory:Inventory of a General Store in a Village in Rural Karnataka, India

  1 jar of savory snacks

  3 jars of soft candies

  1 jar and 1 small bag of wrapped hard candies

  2 jars of chickpeas

  1 jar of Magimix instant stock

  1 packet of bread (5 pieces)

  1 packet of papadum (a snack made from lentils)

  1 packet of crispbread (20 pieces)

  2 packets of cookies

  36 incense sticks

  20 bars of Lux soap

  180 individual portions of pan parag (a combination of betel nuts

  and chewing tobacco)

  20 tea bags

  40 individual packets of haldi powder (turmeric)

  5 small bottles of talcum powder

  3 pack
s of cigarettes

  55 small packets of bidis (thin, flavored cigarettes)

  35 larger packets of bidis

  3 packs of washing powder (500 grams each)

  15 small packs of Parle-G biscuits (cookies)

  6 individual-size packets of shampoo

  During the two hours we spent with this household, we saw two customers. One bought a single cigarette, the other a few sticks of incense. Clearly, the marginal return of increasing the size of the inventory a little was potentially extremely high, especially if the family could try to buy something that the other shops in the same village did not supply. But the overall return of the activity was very low: With this volume of sales, it was not really worth the time spent sitting in the shop all day.

  There are countless such shops in developing countries, several in every village, thousands in alleys of big cities, all selling the same very limited inventory. And the same is true of the fruit vendors, the coconut sellers, and the snack stalls. Walking down the main street of the biggest slum in the town of Guntur at 9:00 AM, it is hard to miss the long line of women selling dosas, the rice-and-lentil pancakes that are South India’s answer to the morning croissant. Smeared with a spicy sauce, and wrapped in a piece of newspaper or a banana leaf, these sell for 1 rupee (roughly 5 cents, at PPP). By our count, one particular morning, there was one dosa seller for every six houses. The result was that at any particular time, many of these women were just waiting around for customers. It seemed obvious that if they could have merged three of the businesses together, and sent the others on some other endeavor, they could have made more money.

  This is the paradox of the poor and their businesses: They are energetic and resourceful and manage to make a lot out of very little. But most of this energy is spent on businesses that are too small and utterly undifferentiated from the many others around them. As a result, their operators have no chance to earn a reasonable living. The creative sand-driers of Mumbai had spotted an opportunity to make a profitable use of the resources available to them: some free time and the sand on the beach. But what the businessman’s uncle had failed to point out was that, for all their ingenuity, the profits from this activity were almost surely negligible.

  The very small scale of many of these businesses explains why their overall returns are often so low, despite the high marginal return. But it brings to light a new puzzle. The fact that marginal returns are high means that it is easy to grow the overall returns—just put more money into the business. So why aren’t all the small businesses growing really fast?

  One part of that answer we already know—most of these businesses cannot borrow very much, and what they can borrow is very expensive. But this is not the whole answer. First, as we saw, although there are millions of microcredit borrowers, there are many more who have the opportunity to borrow but choose not to. Ben Sedan was one of them. He had a business raising cows and could have grown it with a microcredit loan, but he decided against it. Even in Hyderabad, where there are several competing MFIs, the sign-up rate for any microcredit loan among families who were eligible to borrow was only 27 percent, and only 21 percent of those who had a small business had taken a microcredit loan.

  Moreover, even those who cannot borrow can save: Consider the shopkeeper family in Gulbarga. They lived on about $2 per day per person. In nearby Hyderabad, our data show that those with this level of consumption spend about 10 percent of their total monthly expenditures on health care, whereas those living on less than 99 cents a day spend about 6.3 percent. If, instead of spending an extra 3.7 percent of his budget on health care, our shopkeeper had used it to build up his inventory, he could have doubled his inventory in a year. Alternatively, the family could cut down completely on cigarettes and alcohol and save about 3 percent of their daily expenditure per capita: This would allow them to double their inventory in about fifteen months. Why don’t they?

  The experiment in Sri Lanka provides another striking illustration of the fact that financing is not the only barrier to expansion. Recall that entrepreneurs who got $250 made a lot of money—far more, per dollar invested, than most successful firms in the United States. But here is the catch: The profits of micro-entrepreneurs who received the $500 grant did not increase any more than the profits of those who received the $250 grant, in absolute terms. In part, it is because those who received the $500 grant did not choose to invest all of it in their business: They invested about half of it, and used the rest to buy things for their home.

  What is going on? Could the owners really have something better to do with this free money, given how high the marginal return is?

  The notable fact is that the Sri Lankan micro-entrepreneurs did invest the first tranche of dollars. If they chose not to invest the second tranche, it is perhaps because they thought that their businesses would not be able to absorb it: Investing the entire amount would have meant tripling the capital stock of the average business, and a step like that might well require hiring a new employee or finding more storage space, which then would cost much more money.

  Figure 2: Two Technologies

  So part of the reason the businesses of the poor don’t grow, it seems to us, goes back to the nature of the businesses they operate. Recall the inverted L-shape in Figure 1, showing that overall returns could be low even when the marginal returns are high. Figure 2 shows two versions of the curve in Figure 1: One, denoted OP, is very steep when it starts out and then flattens very quickly. The other, OZ, goes up less quickly to start with but keeps going up a long way.

  If, in the real world, the profits of poor businesses look like the curve OP, then it is easy for a very small firm to grow, but the growth potential tapers off quite fast. This is similar to the shopkeeper example: Once you have set aside some room in your home for a shop and have committed to working there a few hours a day, your profits will be much higher if you have enough goods to fill up the shelves and keep you busy than if you have next to nothing (as many shops seem to). But once your shelves are full, any further expansion probably would not have enough marginal return to pay off the very high interest rates on the loan you might use to make it happen. So all the businesses will stay small. If the shape is more like OZ, on the other hand, there is much more scope for growing the business. Our reading of the evidence is that for most poor people, the world is more like OP.

  Of course, we know that it can’t all be like OP—otherwise there would be no large firms anywhere. Perhaps the businesses of shopkeepers, tailors, and sari sellers look like OP, but it must be possible for some other kinds of businesses to use more productive capital. It is clearly possible to run large retail chains or textile factories if one can buy the right equipment, but doing so must require either some special skill or a much larger up-front investment. You can start Microsoft in a garage somewhere and keep scaling up, but to do so you need to be the kind of person who is at the absolute cutting edge of some new product. For most people, that is not really an option. The alternative is to invest enough to get a production technology that allows your business to operate on a large scale. Recall Xu Aihua, the Chinese woman who started her business with one sewing machine and built a garment empire. Her big break came once she got an export order. Without that, she would have soon hit the limits of the local market. However, in order to be considered for the export order she needed to have a modern factory with automatic sewing machines. This required her to invest more than 100 times the initial capital in the firm.

  Figure 3 represents the idea of these two production technologies. There is OP on the left, but way over on the right there is a new production technology, QR, which generates no returns whatsoever until a minimum investment is made, but high returns thereafter. Notice also the way we have marked parts of OP and QR in bold to make one connected line, OR—this represents the actual return on investing a given amount. When you invest just a little, you invest in OP; you have no reason to invest in QR because QR produces no return at first. When you invest more, OP
becomes a bad deal, so for a while, the marginal returns are quite low. However, once you have enough money, you may switch to QR. This represents Xu Aihua’s history: She started with OP with her secondhand sewing machines and at some point managed to switch to QR and the automatic machines.

  Figure 3: Combining Technologies and the S-Shape of Entrepreneurship

  What does OR look like? Kind of like the S-shape, right? There is a big hump in the middle, which is the point you need to reach to make serious money. OR brings back the usual S-shape dilemma: Invest little, make little money, and remain too poor to invest much more, or invest enough to cross the hump and then become rich and invest even more and get even richer. The point is that for most people, crossing that hump is not an option. Although small loans may be available, no one (not even the MFIs, which, as we have seen, like playing it safe) will lend these small entrepreneurs enough money. Moreover, getting there might also need some management and other skills that they don’t have and cannot afford to buy. Therefore, they are stuck staying small. Sometimes, the initial flattening off of returns comes so soon that the same person ends up running three different businesses rather than trying to grow any one of them, for instance, selling dosas in the morning, trading saris during the day, threading beads to make necklaces in the evening.

  But then how did Xu Aihua do it? Remember that she increased her stock of machines by 70 percent a year for eight years through reinvesting her profits. Therefore, her profits must have been at least 70 percent of the value of her machines, after paying her workers, and her overall returns must have been even higher. That is unusually profitable—the average small business in the Hyderabad survey, we noted, would actually lose money if it were to pay even minimum wages. We suspect that this reflects in part that Xu Aihua is an especially talented woman, and in part the fact that in those early days of China’s opening up, there was very little competition and lots of demand, so she was at the right place at the right time.