Why Johnny Can't Work
A theory of why countries do (or don't) restrict child labor. And what it may tell us about the true wealth of nations.
Published June 1, 2005 | June 2005 issue
Child labor. To those in industrialized nations, the phrase conjures Dickensian imagery of small children suffering as they toil ceaselessly in dingy factories in London. In the United States, as well, child labor was once widespread. In 1820, children constituted nearly 35 percent of the workforce in large manufacturing firms in the Northeast. In 1900, three-quarters of North Carolina's cotton mill spinners were under 14.
But as we enter the 21st century, these depressing images have faded to sepia. Child labor regulations were enacted decades ago in developed nations, and while children still work, their hours and activities have been severely curtailed by both custom and law. A modern-day Dickens would be hard-pressed to find Oliver Twist in contemporary England, and Horatio Alger's boys would be writing papers, not selling them.
In the developing world, however, child labor remains a pervasive reality. According to the International Labor Organization, the child labor rate is well over 50 percent in many West African nations. In countries as diverse as Vietnam and Guatemala nearly one-quarter of children are in the labor force. Worldwide, by UNICEF estimates, about 246 million children are engaged in regular labor, and of those nearly three-quarters work in what UNICEF considers hazardous situations.
Concerns about exploitation of children in Third World sweatshops have led to international campaigns to ban child labor. Other efforts have called for prohibiting trade in products made with child labor. This past Valentine's Day, for example, was marked by U.S. Congressional calls to action against child exploitation by chocolate producers in West Africa.
All would agree that vulnerable children should not be mistreated in the workplace, and there's persuasive evidence that such abuse does occur, just as it did in England and the United States in years past. "There are children who work in hazardous industries, risking accident and injury," write Cornell University economists Kaushik Basu and Pham Hoang Van in an American Economic Review article on the economics of child labor. "There are others working in conditions that take a slower but definite toll on the children's health."
Sheer survival is, of course, the fundamental reason many families in developing nations send their children to daily jobs, despite the hardships they'll endure. Wages earned by children can mean the difference between abject poverty and passable subsistence. An analysis of Venezuelan households, for example, by World Bank economist George Psacharopoulos finds that "working children contribute about 27% [of] total household income." In Bolivia, he finds, the wages "earned by the 13 years old amount to 13% of total household income."
These realities suggest that campaigns to end child labor will be unsuccessful to the degree such efforts ignore the economic forces behind it. In fact, some observers have suggested that boycotting products made with child labor may result in greater suffering for the families who depend on the food that children literally bring to the table. "Many of these well-meaning interventions can be counterproductive," write Basu and Pham. "This is a field of study where prescription has outstripped analysis by a wide margin."
A positive analysis
In a paper soon to be published in the American Economic Review, UCLA economist Matthias Doepke, a visitor at the Minneapolis Fed's research department, delves into the political economy of child labor regulations, developing a model to explore societal decisions about encouraging or forbidding children in the labor force.
With a deft blend of mathematical economics and political history, Doepke's model moves beyond the moral dimensions of child labor to better understand the demographic, economic and political factors that lead to it—and to its elimination. And for Doepke, this work, important in its own right, is part of a broad inquiry into a deeper problem: Why do some nations remain desperately poor despite decades of effort to find the right path to economic development?
In the words of Doepke and his coauthor, Fabrizio Zilibotti of Stockholm University's Institute for International Economic Studies, this paper (Minneapolis Fed Research Department Staff Report 354) presents "a positive theory of child labor regulations"—"positive" in the sense that it tries to describe and explain reality as it is, rather than prescribe what reality should be.
And indeed, Doepke explained in a recent interview, the lack of that perspective was one of the key motivations for the work. "If you read the existing literature on child labor, there often seems to be this assumption right from the outset that child labor is an evil, that it's a bad thing for children, that it should be ruled out just to protect the children," he says. "There are certainly cases where that's true. But if you look at child labor more broadly, it isn't necessarily exploitation." Doepke himself grew up on a farm where it was "perfectly natural" for children to help with light tasks. (Full disclosure: Doepke concedes that his childhood output as a farm laborer was limited. "I tried to get out of it," he says.)
The existing literature also fails to explain why attitudes about child labor vary widely and change dramatically. "If you go back in history, child labor was not only normal, but people thought it was actually a good thing," Doepke says. If idle hands are the devil's playground, American leaders argued not so long ago, keeping children busy with productive work builds character, provides discipline and inculcates standards that will serve them well as adults. "Most Americans still believed that hard work was good for youngsters," writes historian Walter Trattner in Crusade for the Children, his classic account of child labor reform in the late 1800s. "It prevented juvenile delinquency and female promiscuity, and it was the first rung on the ladder to success." Doepke and Zilibotti refer to debates in the Georgia state legislature in 1900 where opponents of child labor restrictions argued that the "danger to the child was not in work, but in idleness which led to vice and crime." Today, in contrast, Americans tend to view childhood as a time for gentle nurturance. Idle time in summer is cherished, if increasingly rare, and education—not labor—is seen as the best way to curb the sins of youth. Explaining why societal views about child labor have shifted so radically is therefore a central goal for the authors.
The starting point
As political economists, Doepke and Zilibotti start not by assuming that child labor is morally wrong but by asking whose economic position is improved by restricting it. Who might have a pecuniary interest in ruling out child labor?
Child labor activists aren't likely to start with this question, but it's not an unusual question for an economist. Governments often regulate commerce because interest groups have pushed for restrictions that will benefit them. Standard political economy models are based on the idea that the people who stand to gain from diminished competition (limits on immigration, for instance, or tariffs on imports) are those who will lobby for them. "The organizing principle is always the drive to limit competition," says Doepke. "And when you start there, you just have to ask yourself, in the labor market, who competes with children?"
By this logic, the most obvious instigators of child labor restrictions would be unskilled adult laborers, those who compete for jobs with unskilled children and whose wages would rise if children were excluded from the labor pool. Skilled laborers are less likely to be concerned since children are unlikely to have the ability to compete effectively with them.
But "what made the problem interesting," says Doepke, "is that child labor is different from all these other examples because many of these unskilled laborers also have families, and their children may be working." Unlike a labor union's effort to restrict immigration, then, or a steel industry's drive for quotas on imports, unskilled laborers face a
trade-off: If child labor is restricted, their wages may increase but family income fall because their children will no longer be employed. (They may, in addition, have to pay tuition for their children to attend school.) "And the entire paper, I would say, revolves around this trade-off."
A model world
Doepke and Zilibotti build a model to represent this world in which parents must decide whether to send their kids to work or to school. The model simplifies the real world—families are either large or small; adults have either high skill or low skill; people are either young or old. In order to render the mathematics and computation "tractable," demographic in—betweens are disregarded. But the model aptly describes reality in the features that matter most. "All people care about their children, and they care about their income," says Doepke. "They want to eat, but they also want to have happy children." So the model reflects that they derive utility from those sources. And the happiness of the children is assumed to flow from their future income, which will be higher if they attend school.
Making a decision about where to send children also has implications for the number of children parents might have. If the children will go to work, having lots of them is useful because they'll bring in more income. But if it's school, and school is costly, then it's more efficient to have a small family. This is a standard quality-quantity trade-off in family fertility decisions that economists have used since Gary Becker and labor economist H. Gregg Lewis developed it in 1973, elaborating on Becker's original discussion in a 1960 paper.
"In making this decision, what you care about is the relative return of those activities," Doepke says. "How much income can I get from my children working, or how happy will they be if they're educated?" The first depends on wages for working children; the second on returns to education. If schooling has a very high return, reflected in much higher wages for those with high school, college or advanced degrees, parents will lean toward fewer children. If child wages are high, schooling is less attractive and there's a high incentive to have more little wage-earners around the household.
Taking a vote
Up to this point, the model of fertility, education and child labor is fairly standard. But Doepke and Zilibotti then introduce the dimension of political economy. What if parents were able to vote—in a referendum, let's say—for a labor regulation that would prohibit children from working? The results become far more interesting.
The model—and common sense—predicts that people will vote with their children, so to speak. If they have lots of kids bringing in a significant share of family income, they'll have every reason to vote against child labor regulation. Even though the parents' wages will rise if the government forbids children from working, the other meal tickets at home will be unemployed. Small families, on the other hand, are more likely to vote for regulation—family income will increase because wages for unskilled adult parents will rise when there are no children in the labor market.
What's most significant, though, is the persistence of these patterns. Adults who might otherwise favor child labor restrictions will tend to oppose them because years ago they decided how many children to have based on the status quo of child labor; it's not a reversible decision. "You already have these eight children, and you're stuck with them because they take 15 years, 20 years to grow up," says Doepke. "You can't sell them," he jokes. "They're a given."
It's a lock-in that explains why many nations persist in high levels of child labor, and as Doepke points out, policy lock-ins aren't unique to child labor. People often make decisions based on existing policy that lock them into supporting that policy in the future. Think of your mortgage, a 30-year decision based in part on a tax code that allows you to deduct interest payments from your taxable income. Few U.S. homeowners would favor eliminating the mortgage interest deduction even if they feel it intrinsically unjustified as a matter of fiscal policy.
"We're just saying that in the case of child labor, it's particularly important because children are long-lived and they're also a huge part of the family budget," says Doepke. Fertility decisions are irreversible, and children are important economically both for families who derive income from child labor and for those who must pay for their child's education.
And this lock-in—established mathematically—is the feature of their model that allows it to replicate the two steady-state patterns seen in many countries, either high fertility, high child labor and low support for child labor regulation, or low fertility, low rates of child labor and high support for child labor bans.
In Mexico and Brazil, for example, child labor is widespread and fertility rates are high. "It's very persistent," says Doepke. "Countries that had lots of child labor and high fertility rates in the 1960s still look like that 20, 30 years later. That, to me, is suggestive that some kind of lock is happening."
Lock-in also exists for nations with child labor restrictions, nations like the fast-growing economies of East Asia that ruled out child labor early on and face no pressure to reverse the decision. "In each case," write Doepke and Zilibotti, "the existing political regime induces fertility decisions that lock parents into supporting the status quo."
"If you look at developing countries today, the question is why is there so much variation for similar income levels, and not just variation but also persistence?" observes Doepke.
If these patterns are so durable, what has the power to tip the balance so that child labor regulations might actually be passed in a nation with a long tradition of children in the workplace? All societies began without codified restrictions, of course, but in the last two centuries an increasing number of nations have been convinced to carefully circumscribe child labor. Some might attribute the change to shifting cultural mores about the nature of childhood; in addition, international political pressure has been building for such change in recent years. For Doepke and Zilibotti, however, economics holds greater explanatory power.
Assuming that parents are altruistic—they care about the welfare of their children—they will be in favor of child labor restrictions when they gauge that the income lost from child wages will be out-balanced by that gained from future streams of income that their kids will earn after they attend school. And this means that there must be an independent increase in the return to education relative to child labor—a rise in the "skill premium." While the economists don't focus on this aspect, they surmise that technological change which favors skilled labor is a probable source of increasing returns to education. Alternatively, the loss of child-specific tasks would reduce the returns to child labor.In any case, these shifting fortunes will influence the fertility decisions of young married couples, those who aren't yet locked in. If returns to education are high, they'll have fewer children and favor child labor restrictions. Over time, the number of those supporting restrictions will outweigh those whose prior fertility decisions locked them into a preference for no child labor regulations, tipping the vote. "A key prediction of the model," write Doepke and Zilibotti, "is that the change in workers' attitudes to CLR [child labor regulations] occurs gradually. During the early stages of the transition, the working class does not back CLR unanimously, since families with many children continue to depend on child labor."
The reality test
The ultimate test of any model is how it stands up to reality. While the model is crude in a sense—for example, it relies on a simple referendum on child labor regulations, when in reality any such legislation would be subject to years of debate, interest group lobbying, logrolling and compromise—it duplicates the general mechanisms and trends with remarkable verisimilitude.
One empirical implication of their theory is a high correlation between fertility rates and child labor rates, other things equal. And when they run a regression for 125 countries from 1960 to 1990, looking at the effect of fertility on child labor rates, holding constant per capita income and share of population engaged in agriculture, they're able to explain nearly 90 percent of the variation in child labor (which ranges from 0 to 59 percent in their sample). Doepke considers this an interesting finding, supportive but not conclusive proof of their theory's power. "This result, the correlation with fertility, is a bit of a weak point in the sense that other models would make similar predictions for different reasons," he says.
More persuasive is the model's ability to replicate the historical path followed by the United Kingdom, a nation with an almost painfully well-documented labor history. According to Doepke and Zilibotti's theory, remember, an increase in skill premium should persuade some young parents to have fewer children and send them to school so they'll earn more in the future, rather than have many kids and depend on the income from their labor. Fertility rates will fall as more parents make this decision. As the number of small families grows, support for child labor restrictions will rise, since the unskilled parents will realize their wage rate will increase if children are prohibited from competing with them. As skilled labor becomes more plentiful and unskilled labor less so, the premium will actually reverse its upward trend. Eventually, small families will outnumber large families, and voters will pass child labor regulations.
The United Kingdom followed precisely this path. From 1820 until about 1850, the skill premium (as measured by the ratio of skilled to unskilled wages) rose dramatically, but over the next six decades it gradually declined. By 1910, the skill premium was no higher than it had been nearly a century before. Doepke and Zilibotti's mathematical model—after inserting realistic values for schooling costs, family sizes, production technologies and the like—duplicates the humped shape of this wage-ratio trend with uncanny accuracy. (See charts.)
* Defined as the ratio of skilled to unskilled wages.
Source: Doepke and Zilibotti, 2005, "The Macroeconomics of Child Labor Regulation," Research Department Staff Report 354, Federal Reserve Bank of Minneapolis
In the United Kingdom (and the model), fertility rates also began to fall as new families decided to have fewer children. Even before regulations were adopted, child labor rates declined since these new families sent their kids to school rather than jobs. But once small families became the majority and regulations were adopted, child labor dropped sharply. The graphs of the child labor trends are again quite similar—that is, the model mimics the historical data—but the drop in the model's child labor rate is far more precipitous than that seen in reality. (See charts.)
Source: Doepke and Zilibotti, 2005, "The Macroeconomics of Child Labor Regulation," Research Department Staff Report 354, Federal Reserve Bank of Minneapolis
"This discrepancy," explain the economists, "may be due to the fact that in the simulation, CLR are introduced and perfectly enforced instantaneously, whereas in the data, this happens progressively." In other words, many British parents didn't comply initially with the child labor regulations. The model doesn't have this problem.
Doepke explains that the conflict over prohibition of child labor in the United Kingdom was not just between social reformers and factory owners, but also within the working class—between older, large families and newer, small families. The latter favored and complied with child labor regulations. The former, dependent on child wages, sent their kids to work in defiance of the law. "Hundreds of thousands of cases of truancy were prosecuted at the time," notes Doepke. "It was a very popular offense that really fades out only 20 years later or so."
Doepke observes that similar trends in fertility and child labor occurred in other European countries, even though their economies were structurally quite different. France, Germany and Italy all experienced initial declines in fertility, followed by introduction of child labor regulations and then by further fertility reduction. The fact that all three countries were considerably poorer and more agriculturally based than England yet also restricted child labor gives further credence to the Doepke-Zilibotti argument that fertility decisions, not income levels, are the force behind adoption of child labor regulations.
In the United States, child labor regulations were first adopted at the state level, not as federal law, precluding a national analysis. But the variation among states in both fertility and child labor trends offers further support for the Doepke-Zilibotti model. Comparing the states that were early adopters of minimum-age laws for employment to those that adopted only after 1910, the economists find that birth rate declines began earlier and progressed more quickly in the early adopters. While birth rates were similar in the two groups in midcentury, "by 1890, the average birth rate had fallen to 25 in the early group, but was still at 30 in the late group," they write. "So again," adds Doepke, "we see this correlation between demographic change and political change."
The big picture
This last is a telling comment from Doepke. While his interest in child labor regulations is real and remarkably deep, it isn't his primary motivation—even for this piece of research. More profound is his desire to understand the interrelationship between demographic and economic trends, on the one hand, and sociopolitical change, on the other.
"The general topic is not so much child labor but this transition perspective," Doepke explains, referring to the evolution from a premodern state of stagnant living standards to the era after the Industrial Revolution, when living standards began to rise steadily. In trying to understand this transition, economists have looked closely at fertility trends and the demographic transition to low birth rates and smaller families. "This child labor paper is a bit of a bridge" from that work, explains Doepke, "to what's kind of the heart of my interest right now, the political aspect of this transition."
If you compare the early 19th century to the early 21st century, higher incomes and smaller families are certainly one part of the transition, Doepke says. "But one of the biggest changes really is the political landscape," he points out. "Back then, the government had perhaps 5 percent of GDP, very small. There was almost no social insurance; you were basically on your own. The government financed a nice court and a very small army. Even military expenses were minuscule compared to today."
Nowadays, government expenditure in the United States and Europe accounts for about one-third and one-half of GDP, respectively. Social insurance for old age, free public education and universal child labor laws are common to all rich countries. "So there's really huge change in the political structure and the role of the government," Doepke notes. "The overall thrust of my research right now is to look at the political aspect of this grand transition and try to understand why that happens and why it doesn't."
The development path of now-industrialized nations holds intrinsic interest from a historical perspective, of course, but the lessons drawn may also help explain why so many nations have yet to make the same transition, why the developing world remains so poor. "If you think of development policy in general," observes Doepke, "people have fought a lot about capital accumulation, human capital accumulation, corruption, things of that nature, and haven't really gotten that far. What I'm trying to understand is whether these social policies that are really so central to our environment right now are perhaps also essential parts of understanding why this whole economic transition came about."
"This work on child labor is, in some sense, the first piece of that."
Perhaps it was the promise of springtime, but Minneapolis Fed economists have been particularly active in recent months on questions concerning human fertility.Doepke and Zilibotti's February paper was followed in March by Staff Report 359 by Michele Boldrin, Mariacristina De Nardi and Larry Jones on the relationship between old-age pensions and fertility. And in April, Alessandra Fogli issued Staff Report 361, written with New York University economist Raquel Fernández, examining the impact of culture on decisions about work and fertility.
Old age and babies
Discussions about Social Security these days often mention that low birth rates will limit the financial viability of government pension systems: too few workers supporting too many senior citizens. Boldrin, De Nardi and Jones reverse the looking glass in their examination of how pension plans affect fertility.
They build a model around the assumption that parents have children because those children will care for them in their old age and then see if the model can successfully explain 20th century trends in fertility and government pension programs. In other words, could the expansion of public pension programs in Europe and the United States have curbed fertility in those regions by decreasing parental motivation to have children? Their results suggest that, indeed, the growth of government pension systems can account for as much as 65 percent of variations in fertility among nations and over time.
"These findings give indirect support for a strong role for the 'old age security' motive for fertility," write the authors, who suggest a broader hypothesis: "Any social or institutional change that affects the economic value of other components of the retirement portfolio will have a first order impact on fertility choices."
Does culture matter?
Fernández and Fogli investigate the role played by culture in women's fertility and work decisions. Economists tend to be skeptical of cultural explanations for economic phenomena, in part because culture is nearly impossible to quantify. But Fernández and Fogli contend that the large cross-country and time variations in women's behavior cannot be explained solely by economic and institutional factors. Cultural factors—including the formation and evolution of preferences and beliefs about women's roles—are critical, they suggest, to understanding women's behavior.
To quantify the effect of culture on women's fertility and work decisions, the authors examine the 1970 labor force activity and fertility behavior of American women whose parents were born in another country. "It can be argued," write Fernández and Fogli, "that these [American-born] women share similar markets and formal institutions but have possibly different cultural heritages as reflected in their parents' country of origin." This cultural heritage is captured in their analysis by the aggregate labor force participation and total fertility rate in the country of ancestry in 1950, a proxy for the cultural environment where the parents grew up before moving to America.
After accounting for other relevant variables, the economists find that culture does matter. Women whose parents came from countries where women worked more, work more themselves. Similarly, higher fertility rates in the country of ancestry in 1950 are associated with higher fertility among the American-born women in 1970. "We find that culture, as reflected in our proxy variables ... , is a quantitatively and statistically significant determinant of women's work and fertility outcomes."