VoxEU Column Development

The eradication of child labour in the United States in the early twentieth century: Lessons for developing economies

Widespread child labour may slow economic development in a number of ways, and legislation reducing child labour might break such a poverty trap. Why is such legislation rare? This column looks at the historical experience of the United States in eradicating child labour and suggests that industries highly dependent on child labourers may be the political stumbling block.

In recent years, there has been a renewed interest for the issue of child labour.1 An illustration of this trend is given by the sharp increase in the number of empirical papers addressing the topic (Edmonds, 2007).

Unfortunately, there are good reasons to be concerned: despite the persistent efforts to eradicate it, child labour is still pervasive. In 2000, an estimated 211 million children aged 5 to 14 worked worldwide, or about 17.6% of the children in that age group.

The incidence of child labour has clear geographical patterns. While activity rates for children reach nearly 30% in sub-Saharan Africa, they are slightly below 7% in the developed world (ILO, 2006). Indeed, there is a striking negative correlation between the level of economic development of a country and the level of child labour participation (see Figure 1).

Figure 1

Source: Edmonds (2007)

There are good reasons why economists think that the negative correlation between child labour and economic development is causal. In particular, a number of theoretical papers have suggested ways in which child labour may contribute to persistent poverty traps.

Child labour and poverty traps

In a seminal paper, Basu and Van (1998) proved that a poverty trap may emerge from the interaction effect between wages and child labour. Parents would like to keep their children from working but they cannot unless wages are sufficiently high. Interestingly, by sending their children to work, they further depress wages, making child labour and poverty persistent.

A number of other papers have proposed alternative channels through which child labour may perpetuate poverty in both static and dynamic set-ups. The interaction of child labour with social norms (Lopez-Calva, 1999), fertility and human capital accumulation (Hazan and Berdugo, 2002), coordination failures between firms and households (Dessy and Pallage, 2001), income inequality (Swinnerton and Rogers, 1999) and imperfections in credit markets (Baland and Robinson, 2000) have all been stressed.
A common feature of most of the theoretical papers on child labour is the existence of multiple equilibria. That is, depending on the initial conditions, the economy may end up in a “good” (i.e. low child labour participation) equilibrium or conversely, in a “bad” equilibrium.

A corollary of this is that social legislation may play a crucial role in the economy acting as an “equilibrium shifter”. Whether in the form of child labour bans or compulsory schooling laws, legislation allows economies stuck in the “bad” equilibrium to break free from the poverty trap.

Child labour lessons from US history

Consequently, there are good reasons for which we might want to inquire about the experience of economies that eradicated child labour in the past. Were child labour laws and/or compulsory schooling laws efficient? What were the short-run costs of the legislation? Who paid for them?

In a recent paper, I use the historical case of the United States at the beginning of the century in order to assess the effectiveness and economic consequences of child labour legislation. In the first two decades of the twentieth century, US states introduced labour legislation restricting the employment of children.

Like previous research (Moehling, 1999), my study exploits time variation in the implementation of the states’ child labour legislation. The main novelty is that I interact this variation with the industry’s level of technological dependence on child labour.

The degree of technological child labour dependence (measured by the proportion of children employed in the industry in 1900) reflects the potential for legislation to reduce the growth rate of industries. Put simply, when introduced, child labour laws are supposed to increase the costs and reduce the growth rate only of industries that were initially employing children.

The advantage of exploiting the degree of child labour dependence is that it becomes easier to argue that the effect of the legislation on growth rates is causal, due to the existence of a clear micro-level mechanism that explains which industries should be affected by the legislation.2 In addition, having observations for many industries in many states allows me to control for state and industry fixed effects.

The main finding is that the introduction of the child labour laws reduced the short-term growth rate of industries that were highly dependent on child labour. Even after controlling for state and industry fixed effects, the implementation of child labour legislation explains a reduction of up to 2% in the yearly growth of child-labour-dependent industries.

In sharp contrast, most of the industries of the economy did not suffer from the implementation of child labour legislation at all. The reduction in growth attributable to child labour laws for the average industry (in terms of child labour dependence) is just 0.1%.

An implication of the study is that the costs of eliminating child labour could be very large and fall disproportionately on some industries. This makes legislation difficult to enforce for various reasons. First, developing economies face a painful trade-off between short-run growth and a child labour ban. But since only a few industries would be affected by the legislation, it is not clear how important this trade-off would be.

Potentially more important is that, insofar as the policy has large effects on few industries, those industries have great incentive to organise effective lobbies. These lobbies could turn child labour legislation into a “dead letter”. Indeed, the evolution of US child labour legislation is full of anecdotal evidence about interest groups obtaining legislative exceptions or favourable court rulings (Trattner, 1970).

Can we extrapolate from the US experience to modern developing economies?

Although it would be probably too far-fetched to argue that the United States in 1900 was just like a contemporary developing economy, there are some aspects in which they are similar. As can be seen from Table 1, the labour force participation of children in the United States in 1900 did not differ substantially from those of modern developing economies.

Although more difficult to compare, the available per capita GDP measures are also similar. Furthermore, the industries that employed children in modern economies and in the United States at the beginning of the century are essentially the same. The textile and glass industries were and are still the major employers of children (evidence from India, Tripathi, 2008).

Table 1 Labour force participation and GDP – historical and modern economies


Child labour (Participation)

GDP (per capita)

United States (1900)

$ 4,921

World (2000)

$ 5,245

Latin America (2000)

$ 3,923

Asia and the Pacific (2000)

$ 952

Sub-Saharan Africa (2000)

$ 508

Other regions (2000)

$ 26,156

Notes: Child labour refers to activity rates. For the 1900 figure see Puerta (2008), for the rest see ILO (2006). GDP per capita in 2000 in US dollars are from World Bank, World Development Indicators (2000 figures) and Johnston and Williamson (2008) for the US in 1900.’Other Regions’ corresponds to high-income countries in 2000 as classified by the World Bank.

The comparison between the US experience and modern economies may be reasonable also because the results of the paper are shown to hold under quite general conditions. In particular, the results are not dependent on the particular definition of “technological” child labour dependence or child labour laws. I test this by changing the definition of child labour laws to include alternative combinations of working-age restrictions, compulsory schooling laws, and factory inspection legislation.

In addition, I alternatively define the dependence on child labour using earlier census dates (1880 and 1890) as a test for the robustness of the child labour dependence measure. Regardless of how we measure child labour legislation or the child labour dependence, the main result of the paper that introduction of the law slows down the growth rates of highly dependent industries holds up.


The main lesson of the paper is that the short-run implications of child labour legislation may be crucial in order to understand why child labour legislation is unevenly enforced in developing countries today. Despite wide consensus that the eradication of child labour has positive effects in the long run, the fact that the cost in the short run may be highly concentrated in a few industries may make it difficult to implement the policy, thereby contributing to persistent poverty and underdevelopment.


Baland, Jean Marie and James Robinson (2000), ‘Is Child Labour inefficient?’, Journal of Political Economy, Vol. 108, No. 4.
Basu, Kaushik and Van, Pham Hoang (1998), ‘The economics of Child Labour’, American Economic Review 88, Vol. 3, June.
Dessy, Sylvain and Stephane Pallage (2001), ‘Child Labour and Coordination Failures’, Journal of Development Economics 65, No. 2.
Edmonds, Eric V. (2007), ‘Child Labour’, National Bureau of Economic Research (NBER) Working Paper 12926, February.
Hazan, Moshe and Binyamin Berdugo, ‘Child Labour, Fertility and Economic Growth’, The Economic Journal 112, October.
ILO (2006), Global Child Labour Trends 2000 to 2004, Geneva, SIMPOC, ILO/IPEC, 2006
Johnston, Louis D. and Samuel H. Williamson (2008), ‘What Was the U.S. GDP Then?’ MeasuringWorth.Com, 2008.
Lopez-Calva, (1999), ‘A Social Stigma Model of Child Labour’, mimeo.
Moehling, Carolyn (1999), ‘State Child Labour Laws and the Decline in Child Labour’, Explorations in Economic History 36.
Puerta, Juan Manuel (2008), ‘Child Labour Laws and the End of Child Labour in the US: Evidence from American Manufacturing Censuses (1900-1920)’, Economic History Society's 2008 annual conference at the University of Nottingham, Friday 28 March to Sunday 30 March.
Rajan, Raghuram and Luigi Zingales (1998), ‘Financial Development and Growth’, American Economic Review, Vol. 88, No. 3
Swinnerton, Kenneth A. and Carol Ann Rogers (1999); ‘The Economics of Child Labour: A Comment’, American Economic Review, Vol. 89, No. 5
Trattner, William (1970) , Crusade for the Children: A History of the National Child Labour Committee and Child Labour Reform in America, Chicago: Quadrangle Books.
Tripathi, Shruti (2008), ‘Child Labour as an Institution in India’ (February 4, 2008).

1 This column is part of a broader on-going project I am co-authoring with my advisor Hans-Joachim Voth.
2 The identification strategy is similar in spirit to that used by Rajan and Zingales (1998).

7,874 Reads