One of the most remarkable achievements of the past decades has been the spectacular collapse in the share of the world’s population living in extreme poverty (Lakner et al. 2016, Ravallion 2021). This trend is now well-documented, yet much remains to be understood about its underlying drivers. A natural candidate is education, which has improved dramatically in the past decades and is often mentioned as a key driver of economic development (Andrabi et al. 2023, Égert et al. 2022, Hendricks and Schoellman 2018). But in the absence of a quantitative assessment of the income gains that education has generated and how they have been distributed within countries, we lack a proper understanding of how large the benefits from schooling have been for the global poor. Addressing this gap is of fundamental importance for policy, in a world where the vast majority of children from low-income households are enrolled in public schools.
In a new paper (Gethin 2023), I construct the first estimates of the aggregate and distributional effects of worldwide educational expansion since 1980. Leveraging a unique micro-database on individual incomes, a ‘distributional growth accounting’ model of education and the wage structure, and complementary evidence from three education policies, I quantify the contribution of schooling to growth for different groups of the world distribution of income. I find that education explains a considerable fraction of global economic growth, extreme poverty reduction, and gender inequality reduction since 1980, putting education policies at the forefront of economic progress.
Distributional growth accounting: Education and the world distribution of income
To estimate the effect that education has had on the world distribution of income, I combine three methodological ingredients.
- The first ingredient is a simple model of education and the wage structure à la Goldin and Katz (2007). In this ‘distributional growth accounting’ framework, expanding education increases aggregate labour income by the private return to schooling. It also tends to reduce inequality, as increasing the relative supply of skilled workers exerts a downward pressure on their relative wage.
- The second ingredient is a new micro-database representative of 95% of the world’s population. It is based on household surveys fielded in 150 countries around 2019, which I collected from various data repositories and country-specific sources. This database allows me to observe the joint distribution of education and income and to estimate the returns to schooling in each country and how they vary by education level (primary/secondary/tertiary).
- The third ingredient consists in three in-depth case studies of education policies implemented in Indonesia, India, and the US, which I use as a validation exercise. Figure 1 illustrates the results in the context of Indonesia’s 1970s school construction programme, which has received considerable attention in the economics literature (e.g. Duflo 2001, Akresh et al. 2023). I find that the policy led to large aggregate income gains, which were disproportionately concentrated among low-income groups. Crucially, the model reproduces these findings with a remarkable degree of accuracy, indicating that it provides a good methodological foundation to study the aggregate and distributional effects of education.
Figure 1 Actual versus simulated distributional effects of Indonesia’s INPRES school construction programme
Notes: The figure compares actual causal effects of educational expansion on the average income of each quintile with simulated effects predicted by the model. Capped spikes correspond to 95% confidence intervals. The dashed line shows the estimated effect of average regional years of schooling on the average income. Estimates combine SUSENAS 1993-2019 microdata with INPRES programme intensity from Duflo (2001). Interpretation: a one-year increase in average years of schooling of the working-age population is associated with a 20-25% increase in average income among the poorest 20%.
Combining the data and the model, I quantify the contribution of education to real income growth for different groups of the world distribution of income. To do so, I use estimates of returns to schooling and the model to construct a counterfactual world distribution of income, that is, by how much lower would incomes be today if education had not improved (while all other factors of production had evolved the way they have). I then compare this counterfactual to the actual evolution of global incomes, yielding an estimate of the contribution of education to growth for different income groups within each country. This approach is analogous to the one usually used by macroeconomists in growth accounting (e.g. Barro and Lee 2015), with the critical difference that it accounts for the inequality-reducing effects that education has had within each country.
Education and global poverty reduction
I find that education has had large effects on the world distribution of income. Figure 2 plots real income growth for all individuals in the world, ranked from the poorest 1% to the richest 0.01%. The lower shaded area shows growth that can be explained by educational expansion, while the upper shaded area corresponds to residual growth coming from other factors (such as physical capital or technology).
Real income gains have been greatest at the middle and the very top of the global income distribution, generating what has often been referred to as the ‘elephant curve’ of global inequality and growth (Lakner and Milanovic 2016). This pattern reflects the conjunction of trends in inequality between and within countries, including the rise of China and India (middle of the distribution), sluggish economic growth in low-income countries (bottom of the distribution), weak income gains for most households living in high-income countries (upper middle of the distribution), and skyrocketing top income inequality in many parts of the world (top end of the distribution).
The main contribution is to isolate gains from education, represented by the lower shaded area. These gains have been enormous. Under conservative assumptions, education accounts for 50% of global economic growth and as much as 70% of income gains among the world’s poorest 20% individuals since 1980. It can also explain 40% of the reduction in the share of the world’s population living in extreme poverty.
Figure 2 Education and the distribution of global economic growth, 1980-2019
Notes: The figure plots total real income growth by global income percentile from 1980 to 2019, decomposing it into a part that can be explained by private returns to schooling and an unexplained component. The upper shaded area represents the growth rates that would have prevailed absent any improvement in the education of the world’s working-age population since 1980. The lower shaded area represents the corresponding contribution of education to economic growth. Taking the ratio between this contribution and actual growth rates, education explains about 70% of growth for the world’s 20% poorest individuals. The income concept is pretax income per capita.
Methodologically, accounting for distributional effects within countries is essential to quantify the role of schooling in global poverty reduction. Because education increases the supply of skilled workers in the economy, it reduces their relative wage while increasing that of low-skilled workers. Ignoring this channel would lead to strongly underestimating the benefits of education for the world’s poorest 20% individuals.
Education and global gender inequality
I also extend distributional growth accounting to the study of another major historical transformation: the decline of global gender inequality. To do so, I quantify how large gender labour income gaps would be today if education had not improved. I account for differential educational expansion by gender, but also for heterogeneous effects of schooling on earnings and labour force participation. This counterfactual is then compared to the actual evolution of female labour income shares, on which data is available since 1991.
The main conclusion is that education can explain a large share of reductions in gender inequality observed in the past decades. As shown in Table 1, the share of labour income received by women worldwide increased modestly, from 29.3% in 1991 to 32.1% in 2019. The second row estimates by how much lower the female labour income share would have been if the distribution of educational attainment had remained unchanged, assuming returns to schooling are the same for men and women and no differential effect of schooling on employment. This compositional factor alone explains about half of global gender inequality reduction: the female labour income share would have increased by 1.3 percentage points instead of the 2.8 observed. The third row incorporates heterogeneous returns to schooling by gender. Because returns to schooling are higher for women than for men in nearly all countries, this raises the contribution of education to over two-thirds. The last row incorporates the effect of education on female labour force participation. By this last measure, education accounts for approximately 80% of global gender inequality reduction since 1991, and about 59% of gender inequality reduction in the average country.
Table 1 Education and global gender inequality, 1991-2019
Notes: The table reports actual versus counterfactual global female labour income shares under different assumptions. Global female labour income: total share of labour income received by women in the world as a whole. Change in education: only accounts for differential trends in schooling by gender, applying the same returns to schooling for men and women to build the counterfactual. Heterogeneous returns: accounts for differential returns by gender. Extensive margin: account for differential effects of schooling on employment by gender. Cross-country average: population-weighted average of the share of gender inequality reduction explained by education in each country.
Conclusion
My study represents a first attempt at estimating the role played by education in the historical reduction of global poverty and gender inequality. Under conservative assumptions, private returns to schooling can explain a large fraction of real income gains among the world’s poorest individuals, in the order of 60-70% and potentially more. It can also account for over half of the rise in the share of labour income accruing to women. This puts public education policies at the center of the remarkable reduction of poverty and gender inequality observed in the past decades.
These results call for future research in two directions.
First, there is a need to better understand through which channels education can be most effective at reducing poverty. This calls for more macro-micro perspectives on the relationships between education, technical progress, innovation, and other ingredients of economic development. In particular, more research should be conducted on which combinations of policies not only improve growth, but also make the growth process more inclusive. This also implies improving our understanding of inequalities in access to public services and how they have changed in the past decades (e.g. Fisher-Post and Gethin 2023, Gethin 2023b).
Second, future research could adopt comparable methodologies to extend the study of the world distribution of income to other key transformations observed in the past decades, such as trade globalisation, structural change, financialisation, and even democratisation and changing gender norms. The microeconomics literature provides ample and growing empirical evidence on the economic effects of these factors in specific contexts. Combined with the microdata collected in this paper, additional data collection efforts, and adequate theoretical frameworks, this evidence could be aggregated to shed light on the role played by these long-run processes in the reduction of global poverty and inequality.
References
Akresh, R, D Halim and M Kleemans (2023), “Long-Term and Intergenerational Effects of Education: Evidence from School Construction in Indonesia”, The Economic Journal 133 (650): 582–612.
Andrabi, T, N Bau, J Das and N Karachiwalla (2023), “The education spending multiplier: Evidence from schools in Pakistan,” VoxEU.org, 11 June.
Barro, R, J-W Lee (2015), Education Matters: Global Schoolig Gains from the 19th Century to the 21st Century, Oxford University Press.
Égert, B, D Turner and C de La Maisonneuve (2022), “A new aggregate measure of human capital: Linking education policies to productivity through PISA and PIAAC scores”, VoxEU.org, 28 April.
Duflo, E (2001), “Schooling and Labor Market Consequences of School Construction in Indonesia: Evidence from an Unusual Policy Experiment”, American Economic Review 91(4): 795–813.
Fisher-Post, M and A Gethin (2023), “Government Redistribution and Development: Global Estimates of Tax-and-Transfer Progressivity, 1980-2019”, Working Paper.
Gethin, A (2023), “Distributional Growth Accounting: Education and the Reduction of Global Poverty, 1980-2022”, Working Paper.
Gethin, A (2023), “Revisiting Global Poverty Reduction: Public Goods and the World Distribution of Income, 1980-2022”, Working Paper.
Goldin, C and L F Katz (2008), The Race between Education and Technology, Harvard University Press.
Hendricks, L and T Schoellman (2018), “Human Capital and Development Accounting: New Evidence from Wage Gains at Migration”, The Quarterly Journal of Economics 133(2): 665–700.
Lakner, C, M Negre and J Cuesta (2016), “Know your facts: Poverty numbers”, VoxEU.org, 7 November.
Lakner, C and B Milanovic, “Global Income Distribution: From the Fall of the Berlin Wall to the Great Recession”, The World Bank Economic Review 30(2): 203– 232.
Ravallion, M (2021), “A historical perspective on China’s success against poverty,” VoxEU.org, 4 February.
