Editors' note: This column has been updated following publication of the paper on which it was based (Palma and Reis 2021).
An active debate has taken place for some time concerning the determinants of schooling or of levels of human capital in different countries, particularly those where historically education has developed more slowly and more recently. The prevailing view is that political and institutional factors are of prime importance, although other variables – such as income and land inequality, ethnicity, religion, factor endowments, and GDP per capita – have been invoked to help account for these divergences. Studies have argued that countries that lacked democratic forms of government and where suffrage was not widespread were likely to have lower literacy rates and school enrolment as a consequence. The reason is that the supply of publicly funded mass education is a political decision and the elite that held political power did not favour a wide dissemination of human capital. Conversely, the majority, who aspire to having more education, lack the ‘voice’ that will change this situation (Engerman et al. 2009, Gallego 2008, Lindert 2004, 2010).
This literature tends to underplay the fact that human capital is not just the consequence of policy decisions but also of investments made by families whose decisions are strongly influenced by a variety of economic (Boucekkine et al.2007) as well as cultural circumstances. It consequently sometimes presents the provision of schools as the panacea for educational backwardness, forgetting that better access to education is only a necessary condition. While more favourable schooling policies can cause the appearance of more and better schooling facilities, which in the right context can lead to beneficial effects (Duflo 2001), these expanded opportunities will not necessarily involve students more, unless their families also want to send them to school, and are able to do so. This suggests a need to consider the positive incentives that can determine family decisions to invest in human capital.
In a recent paper (Reis and Palma 2021), we collect a random sample of 9000+ individuals from military archives in Portugal, and show that 20 year-old males were 50% more likely to end up literate under an authoritarian regime (the Estado Novo, until 1950) than under the democratic one (the Republic) that immediately preceded it (Figure 1). As the figure shows, the shortest individuals under the Estado Novo were about as likely to be literate as the tallest under the Republic.
Figure 1 Predictive margins for both regimes. The horizontal axis denotes stature quintiles
Why Portugal for our case study?
The choice of Portugal as a case study is justified by two circumstances. First, ever since official statistics have been gathered, Portugal has been one of the worst performers in the field of educational attainment in the West (e.g. Lindert 2004). At the beginning of the 20th century, its illiteracy rate of 75% of the population over 7 years of age was among the highest in Europe (in Spain it was 53%, in Italy 46%). But by 1950, child illiteracy had decreased to 24% (adult illiteracy only fell to a quarter as late as 1970). The enormous progress made during the second quarter of the century must have contributed to the golden age of Portuguese economic growth, together with catch-up to the European frontier that took place from c.1950.
The second consideration that justifies paying attention to this case study is the succession of political regimes that Portugal experienced during the 20th century. These assumed disparate stances, which are useful sources of policy variation. Specifically, Portugal’s regimes differed on the questions of schooling and the extent of permitted political participation. A significant contrast opposed the Republic (1910-26), a limited parliamentary democracy, to the more authoritarian Military Dictatorship (Ditadura Nacional, 1926-33), which morphed into the corporatist dictatorship of the Estado Novo (1933-74). The latter denied the population any possibility of freely exercising suffrage. For simplicity, we merge the last two together given that they purposed similar policies with respect to education.
For historians and public opinion in Portugal, the opposition between these regimes epitomises a major political and ideological struggle. We have chosen for the present exercise the period 1910-1950, a period during which there was a marked rise in literacy in Portugal. This enables us to compare the efforts to produce human capital by these two regimes over similar (and adjacent) time spans, while controlling for as many additional factors as possible.
Our results are robust controlling for a host of factors including economic growth, the disease environment, and regional fixed effects. Furthermore, our individual-level data on literacy, heights, and other covariates are not subject to sample selection because (unlike conscript data) they are taken directly from the population of interest, which we define here to be males of 20 years old (the nature of the data necessarily excludes women, but there is some evidence, which we discuss in the paper, that they benefited from the regime change as well).
The fact that the results are attributable to the Estado Novo is noticeable through the observation that its rise marked a literacy discontinuity – while the Republic was not making significant progress in eliminating illiteracy, the first generation exposed to the Estado Novo exhibited a noticeable improvement in literacy, a type of progress that furthermore continued as time went by (Figure 2). We furthermore show that the Estado Novo caused literacy improvements at two levels: it both decreased the probability of illiteracy, and it increased the probability of passing primary school exams for individuals at any given height quintile.
Figure 2 The Estado Novo marks a literacy discontinuity
Note: the vertical line marks the first generation exposed to the Estado Novo. The years mark the time of observation at age 20.
We argue there is a political economy and cultural explanation for the success of the Estado Novo authoritarian regime in promoting basic education. In a sense, our paper is in the spirit of Edmund Burke’s argument that social change needs to be gradual, and cannot be implemented by top-down design overnight. Some of the Republic’s policies were too radical for its time. The evidence we obtained shows that both the economic elements that shaped family decisions on education and the state were critical. But while neither can be dismissed, the latter mattered most. The change in institutions and associated policy explains at least 80% of the considerable increase in literacy observed during 1910-1950.
The Estado Novo effect can be broken down into measurable institutional policy targeted towards raising human capital (i.e. the increase in the number of schools), versus immaterial factors. The latter mattered the most. Hence, it would be unwarranted to conclude that government institutional policy directed at extending the schools grid explains most of the observed decrease in illiteracy levels. Therefore, our result contrasts with the emphasis on school-building in other contexts (e.g. Duflo 2001).
Why was the Estado Novo successful in promoting basic education?
We argue that the success of the Estado Novo in educating the masses was due to three factors.
- First, its policies were more feasible than those of the Republic.
By this we mean that they were more cost-effective, more centralised, and there was better enforcement, especially near the capital.
- Second, the Estado Novo's educational policy was more in line with the cultural background of the masses.
For instance, following a secularisation program, the Republic banned crucifixes from the walls of the school, while the Estado Novo put them back. Co-education presents another example. It had been pursued by the Republic from its early years. By contrast, it was rejected by the Estado Novo from the start. There is evidence that parents were not comfortable with co-education. In a high-level official document written during the Republic, in 1921, for instance, it was recognised that enrolment numbers were falling. This was true especially for women, but also for men. The suggested reasons included parents' aversion to co-education (the report noting a discrete decline in the year this was adopted as a general policy), as well as the inefficient organisation of the Republic. By contrast, once the regime changed, it was decided, among other matters, that there should be separation of the genders in school buildings and playgrounds and enrolment as well as literacy outcomes duly improved.
- Third, because it was an autocracy, the Estado Novo had stronger political incentives than the Republic to educate the masses, because doing so did not threaten the regime's continuation, which was not the case with the Republic.
The Republican elites knew that much of the country – especially the illiterate rural masses – did not agree with the Republican ideals and policies, and would hence vote against them given a chance.
What can the 20th century Portuguese experience teach us about the interaction between political regimes and literacy attainment? Portugal fits poorly into the accepted scheme in the literature that links the existence of democracy and popular participation in politics to the spread of education. Its experience suggests that an autocracy can have greater success than a democracy if it has closer cultural alignment with the preferences of the masses. It also suggests that such effects can be stronger than those of building infrastructure. This has implications for educational and overall development policy in poor countries today – it means that the alignment of policy with the culture of the masses can lead to greater success than other, more financially costly policies that promote literacy attainment.
Boucekkine, R, D de la Croix, and D Peeters (2007), “Early literacy achievements, population density, and the transition to modern growth”, Journal of the European Economic Association 5(1): 183-226.
Duflo, E (2001), “Schooling and Labor Market Consequences of School Construction in Indonesia: Evidence from an Unusual Policy Experiment”, American Economic Review 91: 795-813.
Engerman, S, E Mariscal, and K Sokoloff (2009), “The evolution of schooling institutions in the Americas, 1800-1925” in S Haber (ed.), Political institutions and economic growth in Latin America, Hoover Institution Press, 159-217.
Gallego, F (2008), “Historical Origins of Schooling: The Role of Democracy and Political Decentralization”, The Review of Economics and Statistics92, 228-243.
Lindert, P (2004), Growing Public: Social Spending and Economic Growth since the Eighteenth Century, Cambridge University Press, 2 vols.
Lindert, P (2010), “The Unequal Lag in Latin American Schooling since 1900: Follow the Money”, Revista de Historia Economica-Journal of Iberian and Latin American Economic History 28: 375-405.
Palma, N and J Reis (2021), "Can autocracy promote literacy? Evidence from a cultural alignment success story", Journal of Economic Behavior and Organization 86: 412-436