Poor Economics

by Abhijit Banerjee

  If children are in part a way to save for the distant future, we would expect that when fertility drops, financial savings go up. China, with its government-enforced restriction on family size, provides us with the starkest example of this phenomenon. After encouraging high fertility rates immediately after the revolution, China started encouraging family planning in 1972, then introduced the one-child policy in 1978. Abhijit, with two Chinese-born coauthors, Nancy Qian (an only child born in the one-child policy era) and Xin Meng (one of four children born before it began) examined what happened to savings rates after the introduction of family planning.33 Households that had their first child after 1972 have one less child on average than those who had that child before 1972, and their savings rates are approximately 10 percentage points higher. These results imply that up to one-third of the phenomenal increase in savings rates in China in the past three decades (the household savings ratio increased from 5 percent in 1978 to 34 percent in 1994) can potentially be explained by the reduction in fertility induced by family-planning policies; the effect was particularly strong for households that had a daughter rather than a son at first birth, consistent with the view that sons are supposed to be the ones who take care of parents.

  This is a huge effect, but of course the Chinese “experiment” is somewhat extreme: It was a large, sudden, and involuntary reduction in family size. Something similar happened in the Matlab area in Bangladesh, however. By 1996, families in villages where contraception had been made available had significantly more assets of all kinds (jewelry, land, animals, house improvements) than families in the comparable villages where it was not available. On average, a household in the treatment area had 55,000 takas’ more worth in assets ($3,600 USD PPP, more than twice the GDP per capita of Bangladesh) than those in control areas. There is also a link between fertility and the amount of money given to parents by their children: Those in the treatment areas received on average 2,146 takas less in transfers from their children every year.34

  The very strong substitution between family size and savings may help us explain the surprising finding that having fewer children does not translate into healthier or better-educated children: If parents who have fewer children expect lower money transfers in the future, they also need to save more in anticipation, and this cuts into the funds they have available for investing in the children they have. Indeed, if investing in children tends to have a much higher return than investing in financial assets (after all, feeding a child is not that expensive), families may actually be poorer in a lifetime sense when they have fewer children.

  The same logic also tells us that if parents don’t expect their daughters to be nearly as useful in taking care of them as their sons—say, because they have to pay a dowry to get their daughters married or because women are expected to get married and once married, their husbands have economic control over them—parents will be less invested in the lives of their daughters. Families not only choose an optimal number of children, they also choose the gender composition. We typically think of our children’s gender as something we don’t get to decide, but that turns out to be untrue: Sex-selective abortions, which are now widely available and extremely cheap, allow parents to choose whether they would rather abort a female fetus. As the stickers pasted on the dividers in Delhi’s main road advertising (illegal) sex-determination services put it: “Spend 500 rupees now and save 50,000 rupees later” (on dowries). And even before sex-selective abortion was an option, in environments where a whole range of childhood diseases can easily turn fatal if not properly dealt with, there was always neglect, deliberate or otherwise, which can be an effective way to get rid of any unwanted children.

  Even if their children don’t die before or after birth, when parents prefer boys, they may have children until they have the number of boys they want. This means that girls will tend to grow up in larger families, and many of the girls will be born in a family that really wanted boys. In India, girl babies stop getting breast-fed earlier than boys, which means that they start drinking water earlier and have accelerated exposure to waterborne life-threatening diseases like diarrhea.35 This is mostly the unintended consequence of the fact that breast-feeding acts as a contraceptive. After the birth of a girl (particularly if she has no brothers), parents are more likely to want to stop breast-feeding earlier in order to increase the wife’s chances of getting pregnant again.

  Whatever the exact mechanics of discrimination against baby girls (or potential baby girls), the fact remains that the world has many fewer girls than human biology would predict. In the 1980s, in a now classic article in the New York Review of Books, Amartya Sen calculated that there were 100 million “missing women” in the world.36 This was before sex-selective abortion was available—and things have only gotten worse since. In some regions in China, there are today 124 boys for every 100 girls. Between 1991 and 2001 (the date of the latest census in India), the number of boys under seven per 100 girls the same age increased from 105.8 to 107.8 for India as a whole. In Punjab, Haryana, and Gujarat, three of India’s richest states but also three of the states where discrimination against girls is believed to be the greatest, there were respectively 126.1, 122.0, and 113.8 boys per 100 girls in 2001.37 Even according to self-reports, which almost certainly underestimate the phenomenon, the number of abortions is particularly high in those states: In families with two daughters, 6.6 percent of pregnancies ended in an induced abortion and 7.2 percent in a “spontaneous” abortion.

  But this is less of a problem where girls are more valuable either in the market for marriage or in the labor market. In India, girls are not supposed to get married within their own villages. Typically, there are designated areas, not too close to the village but not too far away, into which a majority of the girls will marry and move. As a result, it is possible to look at what happens when there is economic growth in this marriage “catchment” area, which presumably makes it easier to find a prosperous family to marry a daughter into. Andrew Foster and Mark Rosenzweig studied this and found that the mortality differential between boys and girls decreases when a girl’s marriage prospects are brighter; in contrast, economic growth in the village, which increases the value of investing in boys (because they stay home), leads to a widening of the mortality gap between boys and girls.38

  Perhaps the most striking illustration of how a household’s treatment of girls responds to the relative values of boys and girls comes from China, which has one of the largest imbalances between boys and girls. During the Maoist era, centrally planned agricultural production targets focused on staple crops. In the early reform era (1978–1980), households were allowed to produce cash crops, including tea and orchard fruits. Women tend to be more useful than men in the production of tea, which needs to be plucked with delicate fingers. In contrast, men are more useful than women in the production of orchard fruits, which involves lifting heavy loads. Nancy Qian showed that when we compare children born in the post- and pre-reform periods, the number of girls in the tea plantation regions (hilly and rainy) increased, but it went down in the regions that were more suitable for orchards.39 In regions that were not particularly suitable to either tea or orchards, where agricultural income increased across the board without favoring either gender, the gender composition of children did not change.

  What all of this underscores is the violence, active and passive, subsumed within the functioning of the traditional family. This was, until fairly recently, ignored by most (though not all) economists, who preferred to leave that black box closed. Yet most societies rely on the goodwill of the parents to make sure that children get fed, schooled, socialized, and taken care of more generally. Given that these are the same parents who contrive to let their little girls die, how much faith should we place in their ability to get this done effectively?

  THE FAMILY

  For the sake of their models, economists often ignore the inconvenient fact that the family is not the same as just one person.We treat the family as o
ne “unit,” assuming that the family makes decisions as if it were just one individual. The paterfamilias, the head of the dynasty, decides on behalf of his spouse and his children what the family consumes, who gets educated and for how long, who gets what kind of bequest, and so on. He may be altruistic, but he is clearly omnipotent. But as anybody who has been part of a family knows, this isn’t quite how families work. This simplification is misleading, and there are important policy consequences from ignoring the complicated dynamics within the family. We already saw, for example, that giving women access to a formal property title is important for fertility choices, not because it changes her view on how many children she wants but because it makes her views count more.

  The realization that the simplest model was missing important aspects of how the family works led to a reassessment in the 1980s and 1990s:40 Family decisionmaking came to be seen as the result of a bargaining process among family members (or at least between the two parents). Both partners negotiate over what to buy, where to go on vacation, who should work how many hours, and how many children to have, but do so in a way that serves both of their interests as well as possible. In other words, even if they disagree on how the money should be spent, if one of them can be made happier without hurting the other one’s well-being, they would make sure it is done. This view of the family is usually referred to as the “efficient household” model. It recognizes that there is something special about the family—its members, after all, did not meet just yesterday and are presumably tied together for the long term. It should therefore be possible (and in their interest) to negotiate over all their decisions to make sure that they do as well as they can, as a family. For example, if the family runs a small enterprise (be it a farm or a small business), it should always try to make as much money from it as possible, and only afterward find a way to split up the gains among its members.

  Christopher Udry tested this prediction in rural Burkina Faso, where each household member (the husband and the wife, or wives) works on a separate plot.41 In an efficient household, all inputs (family labor, fertilizer, and so forth) should be allocated to the various plots in a way that maximizes total family earnings. The data squarely rejected this view: Instead, plots farmed by women were allocated systematically less fertilizer, less male labor, and less child labor than plots farmed by men. As a result, these households systematically produced less than they could have. Using a little bit of fertilizer on a plot increases its productivity a great deal, but increasing the amount beyond that initial level does not do very much—it is more effective to use a little bit of fertilizer on all the plots than a lot of fertilizer on just one plot. But most of the fertilizer in the Burkina Faso households was used on the husband’s plot: By reallocating some of the fertilizer plus a bit of labor to the wives’ plots, the family could increase its production by 6 percent without spending an extra penny. Families were literally throwing money away because they could not agree on the best way to use the resources they had.

  The reason they were doing so also seems clear: Even though they are part of the same family, what the husband grows on his own plot seems to determine what he gets to consume, and likewise for his wife.42 In Côte d’Ivoire, women and men traditionally grow different crops. Men grow coffee and cocoa, whereas women grow bananas, vegetables, and other staples. Different crops are affected differently by the weather: A particular rainfall pattern may result in a good year for the male crops and a bad year for the female crops. In a study with Udry, Esther found that in good “male” years, more is spent on alcohol, tobacco, and personal luxury items for men (such as traditional items of clothing). In good “female” years, more resources are spent on little indulgences for women but also on food purchases for the household. What is particularly odd about these results is that spouses do not seem to be “insuring” each other. Knowing that they will be together for a long time, the husband could gift his wives some extra goodies in a good male year in return for some extras when the weather goes the other way. Informal insurance arrangements of this kind between households of the same ethnic group are not uncommon in Côte D’Ivoire,43 so why do they not operate within the family?

  One finding in Côte d’Ivoire gives us a useful hint about why families are different. There is a third “player” in the family drama—the modest yam, nutritious and easy to store, a staple food in the area. Yams are typically a “male” crop. But as the French anthropologist Claude Meillassoux explains, it is not a crop that the husband can freely sell and spend.44 Yams are meant for the basic sustenance of the household. They can be sold, but only to pay for school fees or medical care for the children, not to buy a new blouse or some tobacco. And indeed, when there is a good year for yams, the family does consume more yams, which is perhaps not surprising, but spending on food purchased in the market and on education also increases. The yam makes sure that everyone in the family is properly fed and educated.

  Thus, what makes the family special is not that its members are effective in negotiating with each other: Quite the contrary—they operate by observing simple, socially enforceable rules such as “Thou shalt not sell thy child’s yam to buy new Nikes” that safeguard their basic interests, without having to negotiate all the time. Other results also make more sense viewed in this light. We saw that when women make more money on their plots, the family eats more. This may be a product of another rule that Meillassoux describes: It is the woman who is in charge of feeding the family; her husband gives her a fixed amount of money for that, but then it is her job to figure out how to do it best.

  The family is bound together then, not in perfect harmony or by the ability to always divide up resources and responsibilities efficiently, but by a very incomplete, very coarse, and often very loose “contract” that defines the responsibilities of each member toward the other members. It is likely that the contract has to be socially enforced, because children cannot negotiate with parents, or wives with husbands, on an equal basis, but society gains from all members of the family having something like a fair share of resources. The incomplete nature of the contract probably reflects the difficulty of enforcing anything more sophisticated. There is no way for anyone to make sure that parents feed their children the right number of yams, but society may be able to sanction or show disapproval of parents who are seen selling yams to buy sneakers.

  One problem with rules that rely on social norms for enforcement is that these norms change slowly, and therefore there is always the risk that the rules are entirely out of sync with reality, sometimes with tragic consequences. In Indonesia in 2008, we met a middle-aged couple at their house, a small white-and-green bamboo structure built on pillars. Right next to it stood another white-and-green house, much larger, airy, made of concrete. It belonged to their daughter, who worked as a maid in the Middle East. The couple was obviously very poor: The husband had a persistent cough and a headache that never seemed to go away, which made it hard for him to work. But he could not afford to see a doctor. Their younger child had dropped out of school after middle school because they could not afford his bus fare to the city. Suddenly, a four-year-old came into the room: She was visibly healthy, well fed, and dressed nicely in a pretty dress, with shoes that had little lights in them that went on and off as she ran around the room. It turned out that her grandparents were taking care of her while their daughter was away. Her mother sent money for the child, but nothing extra for the husband and wife. It seemed that they were the victims of some norm that had not yet shifted—married daughters were still not expected to take care of their parents, despite the obvious inequity it implied, but grandparents continued to feel obligated to take care of their granddaughters.

  Despite the many obvious limitations of the family, society does not have another viable model for bringing up children, and though one day social pension programs and health insurance might free the elderly in today’s poor countries from relying on their children for oldage care, it is not entirely obvious that it would
make them (or their children) happier. The right space for policy is not so much to replace the family as it is to complete its action and, sometimes, protect us from its abuses. Starting from the right understanding of how families function is crucial in being able to do so effectively.

  It is, for example, now widely recognized that public support programs that put money in the hands of women, like the Mexican program PROGRESA, for example, may be much more effective in directing resources toward children. In South Africa, at the end of apartheid, all men over sixty-five and women over sixty who did not have a private pension became eligible for a generous public pension. Many of these old people lived with their children and grandchildren, and the money was shared with the families. But it is only when a grandmother lived with a granddaughter that the granddaughter benefited: Those girls were significantly less likely to be stunted. Pensions received by a grandfather had no such effect. And there is more: Only if the pension was received by the girl’s maternal grandmother was this effect seen.45

  At least one of the two of us is inclined to interpret this evidence as saying that men are just a lot more selfish than women. But it may also be that this is where the norms and social expectations, which we argued play an important role in family decisionmaking, kick in. Perhaps women are expected to do things for the family when they get some windfall cash and men are not. If this is the case, not only who gets the money, but how it is earned, will also matter: Women may not feel that the money they have earned from their own work or their small business “belongs” to their family or their children. Paradoxically, it may be precisely because of women’s traditional role in the family that public policy can get some mileage by empowering them.