Can school centralization foster human capital accumulation? A quasi-experiment from early twentieth-century Italy

By Gabriele Cappelli (University of Siena) and Michelangelo Vasta (University of Siena)

The article is available on Early View at the Economic History Review’s link here


The issue of school reform is a key element of institutional change across countries. In developing economies the focus is rapidly shifting from increasing enrolments to improving educational outputs (literacy and skills) and outcomes (wages and productivity). In advanced economies, policy-makers focus on generating skills from educational inputs despite limited resources. This is unsurprising, because human capital formation is largely acknowledged as one of the main factors of economic growth.

Related to education policy, reforms have long focused on the way that the school systems can be organized, particularly its management and funding by local v. central government. On the one hand, local policy makers are more aware of the needs of local communities, which is supposed to improve schooling. On the other hand, school preferences might vary considerably between the central government and the local ruling elites, hampering the diffusion of education. Despite the importance of the topic, there is little historical research on this topic.

In this paper, we offer fresh evidence using a quasi-experiment that aims to explore dramatic changes in Italy’s educational institutions at the beginning of the 20th century, i.e. the 1911 Daneo-Credaro Reform. Due to this legislation, most municipalities moved from a decentralized school system, which had been based on the 1859 Casati Law, to direct state management and funding, while other municipalities, mainly provincial and district capitals, retained their autonomy, thus forming two distinct groups (Figure 1).

The Reform design allows us to compare these two groups through a quasi-experiment based on an innovative technique, namely Propensity Score Matching (henceforth PSM). PSM tackles an issue with the Reform that we study, namely that the assignment into treatment (centralization) of the municipalities is not random: the municipalities that retained school autonomy were those characterized by high literacy. By contrast, the poorest and less literate municipalities were more likely to end up under state control, implying that the analysis of the Daneo-Credaro Reform as an experiment will tend to overestimate the impact of centralization. PSM tackles the issue by ‘randomizing’ the selection into treatment: a statistical model is used to estimate the probability of being selected into centralization (propensity score) for each municipality; then, an algorithm matches municipalities in the treatment group with municipalities in the control group that have an equal (or very similar) propensity score – meaning that the only different feature will be whether they are treated or not. To perform PSM, we construct a novel database at the municipal level (a large sample of 1,000+ comuni). Secondly, we fill a gap in the historiography by providing an in-depth discussion of the way that the Reform worked, which has so far been neglected.

Figure 1 – Municipalities that still retained school autonomy in Italy by 1923. Source: Ministero della Pubblica Istruzione (1923), Relazione sul numero, la distribuzione e il funzionamento delle scuole elementari. Rome. Note: both the grey and black dots represent municipalities that retained school autonomy by 1923, while the others (not shown in the map) had shifted to centralized school management and funding. 

We find that the municipalities that switched to state control were characterized by a 0.43 percentage-point premium on the average annual growth of literacy between 1911 and 1921, compared to those that retained autonomy (Table 1). The estimated coefficient means that two very similar municipalities with equal literacy rates at 60% in 1911 will have a literacy gap equal to 3 percentage points in 1921, i.e. 72.07% (school autonomy) vs 75.17% (treated). This difference is similar to the gap between the treatment group and a counterfactual that we estimated in a robustness check based on Italian provinces (Figure 2).

Screen Shot 2019-07-22 at 20.34.56
Table 1 – Estimated treatment (Daneo-Credaro Reform) effect, 1911 – 1921.
Figure 2 – Literacy rates in the treatment and control groups, 1881 – 1921, pseudo-DiD. Source: see original article

Centralization improved the overall functioning of the school system and the efficiency of school funding. First, it reduced the distance between the central government and the city councils by granting more decision-making power to the provincial schooling board under the supervision of the central government. Thus, the control exercised by the Ministry reassured teachers that their salary would be increased, and the government could now guarantee that they would be paid regularly, which was not always the case when the municipalities managed primary schooling. Secondly, additional funding was provided to build new schools. The resultant increase appears to have been very large and its impact was amplified by the full reorganization of the school system. The funds could be directed to where they were most needed. Consequently, we argue, a mere increase in funding without institutional change would have been less effective in increasing literacy rates.
To conclude, the 50-year persistence of decentralized primary schooling hampered the accumulation of human capital and regional convergence in basic education, thus casting a long shadow on the future pace of aggregate and regional economic growth. The centralization of primary education via the Daneo-Credaro Reform in 1911 was a major breakthrough, which fostered the spread of literacy and allowed the country to reduce the human-capital gap with the most advanced economies.


To contact the author: Gabriele Cappelli


Twitter: gabercappe

Learning for life? Comparing miners’ education and career paths in Chile and Norway 1860-1940

by Kristin Ranestad (University of Oslo)


Is formal education relevant and useful for industry? Do trained workers acquire relevant knowledge outside the school setting, and if so, where and how?

Much research has been done on technical education, industrial performance and economic growth. But we still lack knowledge of the content of teaching, and the direct use of formal education in daily work tasks and innovation processes. Moreover, our knowledge of the limitations of formal education is scarce.

This research seeks to complement previous work with a detailed investigation of the connections between formal education, ‘learning by doing’, networking and innovation in mining from around 1860 to 1940. Analysing the connection between education, learning and innovation in mining is particularly interesting because mining education was one of the first technical training programmes aimed at a specific industry.

The reason it is possible to study this subject in detail is because of unique source material for the period. Student yearbooks from Norway for the years between 1855 and 1943, and for some years for Chile, provide exclusive information about the life and work of secondary school graduates after they completed their formal education.

This allows to follow the graduates from school into their practices, work and travels, and it is possible to make in-depth analyses of the functions of formal education and of knowledge and skills learned outside school settings.

The student yearbooks for Norway were published each year by the university and are collections of reports made by the graduates themselves about scholarships, continuing education in Norway and abroad (technical and higher), study travels, trainee positions, companies they worked at in Norway and abroad, working positions and personal experiences.

From these yearbooks and additional sources, we find that the formal mining education was relevant and useful for positions in a broad spectrum of mining organisations. Moreover, the radical technological changes that were happening in mining at the time were supported by increased diversification in workers’ educational background and an increase in the proportion of trained workers.

Workers with formal education were increasingly used by the industry. At the same time, we find that practice, work experience and especially study travels abroad, are key examples of essential supplementary knowledge to the formal and theoretical mining instruction, which was acquired outside a school setting.

Workers, technicians and engineers from Norway had a long tradition of travelling abroad. Out of 341 Norwegian mining engineers, 256 (75%) went abroad between 1787 and 1940, normally to Germany, Sweden, France, England, and the United States from the turn of the twentieth century – all countries with important mining industries. They went to study at a foreign universities or schools, to do geological surveys or acquire information about specific techniques, or to work for a longer period at a foreign company.

During these trips abroad, the engineers created networks, acquired knowledge about up-to-date mining technology and contacts and took specialised courses at universities. To understand all dimensions of technology, and especially how to select, transfer, adopt and modify techniques, hands-on experience and learning by doing on-site was key.

The trips abroad were vital to learn how to use, repair and maintain new mining machinery, tools and techniques and enabled knowledge transfer. They functioned as a form of networking and sometimes led to new investments and business opportunities in Chile and Norway. The knowledge acquired during these trips was different than the knowledge learned in school, but not less important.



How new technology affects educational choices: lessons from English apprenticeships after the arrival of steam power

by Alexandra de Pleijt (Utrecht University), Chris Minns and Patrick Wallis (London School of Economics)


Many workers today worry whether robots will do away with their jobs. Most economists argue that the effect of automation is likely to depend on what workers do. Robots may replace some types of manual work, but new jobs will also be created to design, maintain and manage automated production.

A shift towards ‘new jobs’ would mean that different skills will be valued in the future, and many policy experts have argued that secondary and post-secondary education will have to change in response. But if young people and their parents anticipate how automation will affect their job prospects, the choices made among current educational opportunities could shift ahead of any changes in what is offered.

The effects of automation on educational choice will be seen in the future. But past experience can offer some ideas as to whether the arrival of new technology affects these choices, even before the technology is widespread.

This research examines how the arrival of a new production technology affected educational choices in late eighteenth century England. The period between 1760 and 1810 is at the beginning of the largest shift in history from hand- to machine-powered production, through the invention and spread of the steam engine that powered the British Industrial Revolution.

Our research combines detailed evidence on the location and timing of the adoption of steam engines with the records of over 300,000 English apprenticeships from the rolls of the Commissioner of Stamps.

The main finding is that the arrival of steam power changed the willingness of young people to pursue apprenticeships, which for centuries had been the main route to acquiring the skills required for the production of manufactured goods. Counties saw a fall of 40-50% in the share of population entering into textile apprenticeships once a steam engine was present.

Despite the possible association with machine design and maintenance, mechanical apprenticeships also saw a decline of just under 20% following the arrival of steam. Merchant and professional apprentices, who were trading the goods produced by craft or industry, were mostly unaffected.

These findings show that the workforce responded to the emergence of technology that would dramatically change the nature of production and work in the future, but that much of the response was local. Apprenticeships fell first in northern counties where industrial towns and cities with factory-based production had emerged earlier. A similar decline in how workers were trained was not seen in southern and eastern England in the early part of the Industrial Revolution.