Five hundred years of French economic stagnation: from Philippe Le Bel to the Revolution, 1280-1789

by Leonardo Ridolfi (IMT School for Advanced Studies Lucca)

In 2008, output per capita in France amounted to around $22,000 dollars per year. After the Second World War, in 1950, annual average income per capita reached $5,000 dollars, while in 1820, at the beginning of the first official national statistics, GDP per capita averaged $1,100 (Maddison, 2010). Nevertheless, precise knowledge of economic growth in France stops when we get back as far as 1820; before this date, the quantitative reconstruction of economic development is shrouded in mystery.

That mystery lies in the difficulty of uncovering sufficient resource material, devising adequate measures of economic performance in the past, and ultimately interpreting the complexity of the dynamics involved. These dynamics stretch far beyond just the mere economic sphere and concern the way a society is itself organised and structured. Nevertheless, several questions spring to mind.

What was the level of material living standards between the thirteenth and the late eighteenth century, from the early stages of state formation to the French Revolution? How did per capita incomes evolve over time? And were French workers richer or poorer than their European counterparts during the pre-industrial period?

This research provides answers to these questions by estimating the first long-run series of output per capita for France from 1280 to 1789.

The study reveals one important conclusion: the dominant pattern was stagnation in levels of output per capita. For the first time indeed, these estimates document quantitatively and in the aggregate what was previously known only qualitatively or for some regions by the classic works of French historiography (Goubert, 1960; Le Roy Ladurie, 1966): the French economy was an inherently stagnating growthless system, a ‘société immobile’, which at the beginning of the eighteenth century was not much different than five centuries earlier.

At the time of the death of King Philip the Fair in 1314, France was a leading economy in Europe and output per capita averaged $900 per year. Almost five centuries later, this threshold was largely unchanged, but the France of King Louis XVI now belonged to the group of the least developed countries in Western Europe. In the 1780s, per capita income was slightly above $1,000, about half the level registered in England and the Low Countries.

Nevertheless, stagnation was not the same as stability. The French economy was highly volatile and experienced multiple peaks and troughs. In addition, these results reject the argument that there was no long-run improvement in living standards before the Industrial Revolution, demonstrating that GDP per capita rose more than 30% between the 1280s and the 1780s.

Yet most of the rise was explained by a single episode of economic growth that took place prior to the Black Death between the 1280s and the 1340s and which shifted the trajectory of growth onto a higher path.

Overall, these estimates suggest that the evolution of the French economy can be suitably interpreted as an intermediate case between the successful example of England and the Low Countries and the declining patterns of Italy and Spain. Being neither a southern country nor a northern one, the growth experience of France seems to reflect this geographical heterogeneity.



Goubert, Pierre (1960) École pratique des hautes etudes, Laboratoire cartographique, Beauvais et le Beauvaisis de 1600 à 1730: contribution a l’histoire sociale de la France du 17e siècle, Sevpen.

Ladurie, Emmanuel Le Roy (1966) Les paysans de Languedoc Vol. 1. Mouton.

Maddison, Angus (2010) Historical Statistics of the World Economy: 1-2008 AD, Paris.

When political interests block new infrastructures: evidence from party connections in the age of Britain’s first transport revolution

New research shows how party politics and connections slowed the diffusion of much-needed improvements in river navigation in Britain during the early eighteenth century. The study by Dan Bogart (University of California Irvine), which is forthcoming in the Economic Journal, reveals that modern concerns about powerful interests coalescing to block infrastructure projects that will benefit the wider economy are nothing new.

Islington Tunnel in the early 19th century. Source: <;


The famous economist Adam Smith noted in The Wealth of Nations that landowners close to London petitioned Parliament against the extension of transport improvements because it threatened their rents. Was Smith right: do ‘downstream’ interests use their political connections to block ‘upstream’ transport improvements? The new research addresses this question in the context of river navigation, which before the development of canals and railways, was a key part of Britain’s early transport system.

A river navigation act established a company with rights to levy tolls and purchase land necessary for improvements in navigation. Through their statutory powers, navigation companies played a key role in the extension of inland waterways. Improved navigation lowered transport costs since freight rates by inland waterway were approximately one-third of the freight rates by road.

In light of their economic importance, it is significant that the diffusion of navigation acts was slow. It took nearly 50 years to extend navigation on most rivers in Britain. One immediate reason is that projects were proposed several times in Parliament as bills before being approved, and some were never approved.

In general, bills proposing infrastructure projects had high failure rates in Parliament. Opposition from interest groups was the most direct reason. Interest groups would appeal to their MP for assistance, and as this research shows, it was significant whether their MP was connected to the majority political party.

The Whig and Tory parties were in intense competition between 1690 and 1741, with the majority party in the House of Commons switching seven times. The two parties differed in their policy positions and their supporters. The Tories were favoured by small to medium-sized landowners, and the Whigs by merchants, financiers and large landowners.

This study is one of first to test empirically whether Britain’s early parties contributed to different development policies and whether they targeted supporters. The research uses new town-level economic, political and geographical data to investigate how party connections and interest groups worked in this important historical period.

The results show that the characteristics of river navigation supporters and opponents in neighbouring areas had a large effect on their diffusion. For example, more towns with roads in upstream areas (generally supporters) increased the likelihood of a town’s river bill succeeding in Parliament and more towns with harbours downstream (generally opponents) reduced the likelihood of the bill succeeding. Such factors were as important as project feasibility, measured by elevation changes.

Another important factor was the strength of majority party representation in neighbouring political constituencies. Having more downstream MPs in the majority party (a measure of opposition connections) reduced the likelihood of a town’s bill succeeding in Parliament and getting blocked from navigation acts. The identity of the majority party was also relevant. Whig majorities increased the probability of river acts being adopted.

These findings confirm the forces highlighted by Adam Smith and show that the institutional environment in Britain was not always favourable to rapid adoption of infrastructure. Interest groups were powerful and could block projects that went against their interest. The Whig and Tory parties contributed to the blocking power or bias from interest groups, although the Whigs appear to have been more pro-development.

More generally, this case focuses attention on the distributional effects of infrastructure and efforts to block projects. Political connections matter and can have important economic consequences.

‘Party Connections, Interest Groups, and the Slow Diffusion of Infrastructure: Evidence from Britain’s First Transport Revolution’ by Dan Bogart is forthcoming in the Economic Journal.

Globalisation and Economic Development: A Lesson from History

by Luigi Pascali (Pompeu Fabra University)

History teaches us that globalisation does not automatically translate into economic development.

How does globalisation affect development? This question has a long tradition in economics and has been much debated both in the academia and in policy circles. Neoclassical theories tell us that reducing trade barriers across countries should provide net benefits to individual economies by making markets more efficient and stimulating competition. Testing these theories, however, turns out to be difficult: rich countries are generally also those that trade the most, but is it trade that makes them rich, or do they trade more because they are rich to start with?

The ideal way to answer to these questions would be through an experiment, in which we randomly divide all the countries of the world into two groups and then we reduce trade costs for one group, while keeping trade costs constant for the other group. The difference in the observed GDP growth in the following years between the two set of countries would eventually provide us with an estimate of the impact of trade integration on development.

It turns out that history can provide us with such an experiment. The invention of the steamship in the late 19th century greatly reduced trade costs for some countries but not for others; whether a country was able to reduce its trade costs as a result of this innovation was the result of its geography, rather than economic forces. In a recent paper (Pascali, forthcoming), this natural experiment is used to assess the causal impact of trade on development.

The Experiment

Before the steamship, sea routes were shaped by winds. As an example, consider Figure 1, which illustrates a series of journeys made by British sailing ships in the 19th century, between England, the Cape of Good Hope and Java, and Figure 2, which depicts the prevailing sea-surface winds in the world.  Winds tend to follow a clockwise pattern in the North Atlantic; consequently, sailing ships would sail westward from Western Europe, after travelling south to 30 N latitude and reaching the ‘trade winds’, thus arriving in the Caribbean, rather than travelling straight to North America. The result is that trade systems historically tended to follow a triangular pattern between Europe, Africa, the West Indies and the United States. Furthermore, because in the South Atlantic winds tend to blow counterclockwise, sailing ships would not sail directly southward to the Cape of Good Hope; rather, they would first sail southwest toward Brazil and then move east to the Cape of Good Hope at 30 S latitude.

Figure 1. 15 journeys made by British ships between 1800 and 1860

fig 1

Notes: These journeys were randomly selected from the CLIWOC dataset among all voyages between England and Java comprised in the dataset. 


Figure 2. Prevailing winds in May (between 2000 and 2002)

fig 2

Note: direction of wind defined by the direction of the arrow and speed by the length of the arrow. 


The invention and subsequent development of the steamship was a watershed event in maritime transport and was the major driver of the first wave of trade globalisation (1870-1913), an increase in international trade that was unprecedented in human history. For the first time, vessels were not at the mercy of the winds, and trade routes became independent of wind patterns.

The steam engine, however, reduced shipping times in a disproportionate manner across trade routes, depending on the type of winds that vessels used to face throughout their journeys. In some routes, shipping times were cut by more than half, while in some others the change was minimal.

These asymmetric changes in shipping times (and related trade costs) across countries are used as a natural experiment, to explore the effect of international trade on economic development.


Exploiting the random variation in trade costs, generated by the transition from sail to steam, this research documents that the consequences of the first wave of trade globalisation on development were not necessarily positive. On a sample of 36 countries, the average impact, in the short run, was a reduction in per-capita GDP, population density and urbanisation rates.

This average negative impact, however, masks large differences across different groups of countries.

Firstly, the initial wave of trade globalisation turned out to be particularly detrimental in countries that were already less economically developed to start with and it was probably the major reason behind the Great Divergence, the economic divergence observed between the richest countries and the rest of the world, in the second-half of the nineteenth century.

Secondly, trade turned out to be very beneficial for countries that were characterised by strong constraints on executive power, a distinct feature of the institutional environment that has been demonstrated to favour private investment.

Why should we expect institutions to be crucial to benefiting from trade? A common argument is that a country with ‘good’ institutions will suffer less from the hold-up under-investment problem in those industries that intensively rely on relationship-specific assets. In this sense, good institutions are a crucial source of comparative advantage in non-agricultural sectors, in which the hold-up problem is more binding. These results confirm this theoretical prediction: a reduction in trade costs increased the share of exports in non-agricultural products, and the share of the population living in large cities, only in those countries characterised by stronger constraints on the executive power.


How did the rise in international trade affect economic development? This research addressed this question using novel trade data and an historical experiment of history. It finds that 1) the adoption of the steamship had a major impact on patterns of international trade worldwide, 2) only a small number of countries, characterised by more inclusive institutions, benefited from trade integration, and 3) globalisation was the major driver of the Great Divergence.

Policymakers who are willing to learn from history are advised to consider that a reduction in trade barriers across countries does not automatically produce (at least in the short-run) large positive effects on economic development and can increase inequality across nations.


This article is based on research presented in the following paper: “The Wind of Change: Maritime Technology, Trade and Economic Development”, The American Economic Review, forthcoming. The associated working paper is available on the CAGE website

The long-term negative impact of slavery on economic development in Brazil

by Andrea Papadia (London School of Economics)

Jean Baptiste Debret (1826). From “The Atlantic Slave Trade and Slave Life in the Americas: A Visual Record”,


Slavery has been at the centre of many heated debates in the social sciences, yet there are few systematic studies relating slavery to economic outcomes in receiving countries. Moreover, most existing work on Brazil – which was the largest slave importer during the African slave trade and the last country to abolish the practice – has failed to identify any clear legacies of this institution.

This research overcomes this impasse by highlighting a distinctly negative impact of slavery on economic development in Brazil. More precisely, it illustrates that in the municipalities of the states of Rio de Janeiro and São Paulo, where slave labour was more prevalent in the nineteenth century, fiscal development was lower in the early twentieth century, long after slavery was abolished.

The identification of this negative effect is tied to separating the true effect of slavery on fiscal development from the fact that the huge expansion of coffee production that Brazil underwent from the 1830s attracted large numbers of slaves to booming regions. In fact, the research shows that:

  • A naïve analysis of the data would suggest that for relatively low levels, more slavery in the nineteenth century was associated with higher successive fiscal development.
  • For population shares of slaves above 30-35%, more slavery was clearly associated with lower fiscal development.
  • Taking account of the impact of the coffee boom on both the demand for slave labour and development, slavery was unambiguously associated with worse developmental outcomes later on.
  • Comparing two hypothetical municipalities – equal in all respects except for their reliance on slave labour – one with 30% of slaves among its citizens would have had revenues 70% lower compared with one with 20%.
  • These results persist even when taking account of a wide variety of other factors that could explain difference in fiscal development across municipalities.

Fiscal development is widely considered as an essential building block in the creation of modern states able to foster economic growth by providing public goods and protecting the rule of law. While the historical process of fiscal development on the European continent is relatively well understood, in other parts of the world the study of the evolution of fiscal institutions is still in its early stages.

There are many reasons why a high incidence of slavery would hamper fiscal development and the provision of public goods:

  • First, a higher incidence of slaves in the population will translate into lower political representation for the masses, even in only partially democratic regimes such as nineteenth and early twentieth century Brazil.
  • Second, the provision of key public goods, such as education, will be less salient in areas that rely heavily on slave labour. These areas will also be less keen to attract workers from other areas of the country and abroad, thus making the provision of public services to their citizens less important.
  • Finally, slavery might make resource sharing though taxation more difficult due to increased ethnic, geographical and class cleavages in the population.

The history of Brazil, which was characterised by large-scale use of slave labour from the sixteenth century until the nineteenth century, provides an idea testing ground to investigate how this clearly extractive institution affected the developmental path of countries and their subdivisions.

The research shows that by accounting for confounding effects due to Brazil’s coffee boom, the pernicious effects of slavery on a key factor for economic growth – fiscal development – can be strongly identified.

British engineering skills in the age of steam

by Harry Kitsikopoulos (academic director, Unbound Prometheus)

Wiki Commons. The side-lever Engine, 1849 ca.


Engineering skills in Britain improved during the eighteenth century but progress was not linear. My research uses a novel approach to quantifying the trends from the first appearance of the technology of steam power (1706) through to the last quarter of the century (the Watt era), using a large amount of data on fuel consumption rates.

Britain was a very unlikely candidate for the invention of steam engines, as I argue in my 2016 book, Innovation and Technological Diffusion: An Economic History of the Early Steam Engines. It was French and Italians who first rediscovered, translated and published the ancient texts of Hero of Alexandria on steam power; they also discovered the existence of vacuum in nature, the main principle of a steam engine’s working mechanism.

But Britain had two advantages: first, a divorce-obsessed king who detached the island from the Catholic dogma and its alliance with the Cartesian epistemological paradigm, both denying the existence of vacuum in nature. The same king also brought a seismic institutional transformation by passing monastic properties under the ownership of lay landlords, a class far more keen on solving the water drainage problem plaguing the mining industry in its drive to exploit mineral wealth.

Britain was also fortunate in another respect: it was relatively backward in terms of mining technology! That proved to be a good thing. While mining districts in Germany and Liège used a technology that resolved the drainage problem, Britain failed to imitate them, hence forcing itself to seek alternative solutions, thereby leading to the invention of the steam engine.

Grand inventions earn glorious references in school textbooks, but it is the diffusion of a technology that contributes to economic growth, a process that relies on the development of relevant human capital.

The records reveal that there were not much more than a dozen engineers who were active in erecting engines during the period 1706-75, including Thomas Newcomen, the obscure ironmonger from Devon who came up with the first working model. The figure increased to at least 60 during the last quarter of the century through the action of the invisible hand: the initial scarcity of such skills raised wages, which, in turn, acted as stimuli transferring talent from related engineering occupations.

My new study traces the production and marketing strategies of this group, which ranged from the narrow horizons of certain figures concentrating on the erection of engines in one locality, a single model, or focusing on one industry all the way to the global outlook of the Boulton and Watt firm.

The last question I pose is perhaps the most interesting: did British engineers get better during the eighteenth century in managing these engines?

Measuring skill is not a straightforward affair. Two well-respected experts at the time came up with tables that specified what the ideal fuel rates ought to have been for engines of different hp. When plotted in a graph these two variables depict a curve of ideal rates.

My analysis uses two distinct datasets with 111 fuel rate observations recorded in working engines – one for the older Newcomen model and another for the newer Watt engines. These actual fuel rates were plotted as bullet points around the respective ‘ideal’ curves. A progressively narrower distance between the curves and the bullet points would indicate higher efficiency and improved engineering skills.

The results reveal that for the first 25 years following the appearance of both models, there was no consistent trend: the bullet points alternated coming closer and moving away from the ideal curves. But the data also reveal that these initial patterns gave way to trends revealing consistent progress.

In an era of practical tinkerers lacking a formal educational system when it comes to this particular skill, British engineers did get better through a classic process of ‘learning-by-doing’, But this only happened after an initial stage of adjustment, of getting used to models with different working mechanisms.

Industrialisation and the origins of modern prosperity: evidence from the United States in the 19th century

by Ori Katz (Tel Aviv University)

Wiki Commons. Market scene by Pieter Aertsen, c.1550


The largest economic mystery is the modern prosperity of humankind. For thousands of years since the Neolithic revolution, most humans lived in small communities, working as farmers, and their average standard of living did not change much.

But in the nineteenth century, things changed: large parts of the world become industrialised. In those parts, people moved to live in huge cities, where they worked in manufacturing and commerce, had fewer children, invested more in schooling, and their standard of living began to rise, and then to rise dramatically, and it has never stopped since. Whether you look at life expectancy, birth fatality, income per person or any other measure, the trend is the same. And we don’t really know why.

We have a lot of theories. Some believe that this dramatic change has something to do with a geopolitical environment that encouraged competition and maintained stability in property rights. Others talk about a change in human preferences, maybe even in human biology. But in every theory, two of the main ingredients are the dramatic reduction in fertility and the increasing investment in human capital during the late nineteenth century.

This research examines the effect of industrialisation on human capital and fertility in the United States during the period from 1850 to 1900. This effect is hard to identify, for example because human capital also affects industrialisation, or because other variables such as ‘culture’ may affect both.

To deal with those problems, the study uses the westward expansion of the country as a ‘natural experiment’. The appearance of new large cities such as Chicago and Buffalo led to the development of new transport routes, and the study looks at counties that happened to be close to those new routes.

Those counties experienced industrialisation only because of their geographical location, and not because of the human capital of the local population or other variables. This means that analysing them is similar to a laboratory experiment, where it is possible to change only one parameter and leave the others intact.

Results show a very large effect of industrialisation on both fertility and human capital. These results are in contrast with an old theory according to which industrialisation was a ‘de-skilling’ process that increased the demand for unskilled labour. It seems that industrialisation was conducive to human capital.

They also find that the effects of industrialisation on both fertility and human capital were larger in counties that were already more developed in the first place. This led to a divergence between them and less developed counties. Indeed, when we look at the country level, we see increasing gaps between the industrialised countries and the rest of the world, starting in the nineteenth century, just like the gaps shown at the county level.

The modern period of growth is still a mystery, but these research results tell us that the effects of industrialisation on fertility and human capital are an important piece of the puzzle. These effects might be the reason for the great divergence between nineteenth century economies that created the modern wealth gaps between nations.

From The NEP-HIS Blog: Fifty Years of Growth in American Consumption, Income, and Wages

Fifty Years of Growth in American Consumption, Income, and Wages By Bruce Sacerdote (Darmouth) Abstract: Despite the large increase in U.S. income inequality, consumption for families at the 25th and 50th percentiles of income has grown steadily over the time period 1960-2015. The number of cars per household with below median income has doubled since […]

via Is the Glass Half Full?: Positivist Views on American Consumption — The NEP-HIS Blog

The impact of new universities on regional growth: evidence from the United States 1930-80

by Alexandra López Cermeño, Lund University / Universidad Carlos III de Madrid

From ODU  Twitter account

Universities generate growth spillovers beyond simply the local market. Analysing data on the universities founded in the United States between 1930 and 1980, my research shows that these drove growth of GDP and population not only in the counties that hosted them, but also in their neighbouring regions. But analysis of their longer-term impact suggests that although there are growth spillovers, the positive effect wears out if it is not periodically renewed.

The role of universities in generating growth is rarely contested. But most research tend to associate the presence of a university with long-term path dependency. In the era of knowledge and information, the role of universities as producers of new ideas and technologies is crucial to productivity. New light on this subject is required not only to understand the role of cultural amenities but also to explore the spatial dynamics around them.

Long-term analysis that compares recipient counties of their first universities between 1930 and 1980 with statistically similar counties that never got an institution shows that the effect of these new universities implies 20% more growth in terms of GDP. Moreover, the analysis shows that the new amenities eventually had an impact neighbouring counties. These dynamics seem to be related to population migration.

This sizeable increase of GDP in these counties is corresponded by a similar size increase in population: new universities generate migratory movements of workers, which eventually lead to higher housing prices and costs to use other infrastructures. Higher costs motivate many workers to relocate to nearby areas where housing and infrastructures are less expensive and access to the amenity is still feasible.

The positive effect of new universities is therefore neutralised in the longer term unless further investments reduce congestion costs. Indeed, the role of infrastructures such as roads seems to explain a large share of the effect of universities.

But the interaction of universities and infrastructure seems to be defined by the decreasing importance of the latter: whereas physical access to infrastructure seemed to constrain the impact of new amenities before the 1950s, more recently established institutions seem no longer dependent on face-to-face contact.

There is further evidence on the role of knowledge dynamics in my study: in the earlier half of the period 1930-80, all that mattered was getting a new university in the county, whereas in the latter half of the period, the quality of the institution seems to have become much more relevant. Counties where research-intensive institutions were established during the period 1950-80 grew almost 40% more.

My analysis shows that the effect of new academic institutions during the twentieth century induced regional spatial dynamics in terms of migration and GDP. But it indicates that the impact of these new amenities was seriously constrained by the congestion of utilities, which limited the extent of growth to the short run.

Thus, it questions the extent of the impact generated by these institutions that is so praised in recent literature since it suggests that their growth dynamics are not self-sustaining: further investments are needed to keep up with the agglomeration forces that attract population and firms to these counties.

Extractive Policies and Economic Outcomes: the Unitary Origins of the Present-Day North-South of Italy Divide

by Guilherme de Oliveira (Columbia Law School) and Carmine Guerriero (University of Bologna)


Italy emerged from the Congress of Vienna as a carefully thought equilibrium among eight absolutists states, all under the control of Austria except the Kingdom of the Two Sicilies, dominated by the Bourbons, and the Kingdom of Sardinia, ruled by the Savoys and erected as a barrier between Austria and France. This status quo fed the ambitions of the Piedmontese lineage, turning it into the champion of the liberals, who longed to establish a unitary state by fomenting the beginning of the century unrest. Although ineffective, these insurrections forced the implementation, especially in the South, of the liberal reforms first introduced by the Napoleonic armies, and allowed a rising class of bourgeoisie, attracted by the expanding international demand, to acquire the nester nobility’s domains and prioritize export-oriented farming. Among these activities, arboriculture and sericulture, which were up to 60 times more lucrative than wheat breeding, soon became dominant, constituting half of the 1859 exports. Consequently, farming productivity increased, reaching similar levels in the Northern farms and the Southern latifundia, but the almost exclusive specialization in the agrarian sectors left the Italian economy stagnant as implied by the evolution of the GDP per capita in the regions in our sample, which we group by their political relevance for the post-unitary rulers as inversely picked by Distance-to-Enemies (see upper-left graph of figure 1). This is the distance between each region’s main city and the capital of the fiercer enemy of the Savoys—i.e., Vienna over the 1801-1813, 1848-1881, and 1901-1914 periods, and Paris otherwise—and is the lowest for Veneto, which we then label the “high” political relevance cluster. Similarly, we refer to the regions with above(below)-average values as “low” (“middle”) political relevance group or “South” and to the union of the high-middle relevance regions and the key Kingdom of Sardinia regions—i.e., Liguria and Piedmont—as “North.”


Figure 1: Income, Political Power, Land Property Taxes, and Railway Diffusion

1  Note: “GDP-L” is the income in 1861 lire per capita, “Political-Power” is the share of prime ministers born in the region averaged over the previous decade, “Land-Taxes” is the land property tax revenues in 1861 lire per capita, and “Railway” is the railway length built in the previous decade in km per square km. _M (_H) includes Abruzzi, Emilia Romagna, Lombardy, Marche, Tuscany, and Umbria (Veneto), whereas KS gathers Liguria and Piedmont. The North (_L) cluster includes the M, H, and KS groups (Apulia, Basilicata, Calabria, Campania, Lazio, and Sicily). See de Oliveira and Guerriero (2017) for each variable sources and definition.


Despite some pre-unitary differences, both clusters were largely underdeveloped with respect to the leading European powers at unification, and the causes of this backwardness ranged from the scarcity of coal and infrastructures to the shortage of human and real capital. Crucially, none of such conditions was significantly different across groups since, differently from the Kingdom of Sardinia, none of the pre-unitary states established a virtuous balance between military spending and investment in valuable public goods as railway and literacy. Even worst, they intensified taxation only when necessary to finance the armies needed to tame internal unrest, which were especially fierce in the Kingdom of Two Sicilies. The bottom graphs of figure 1 exhibit this pattern by displaying the key direct tax, which was the land property duty, and the main non-military expenditure, which was the railway investment.

Meanwhile, the power of the Piedmontese parliament relative to the king grew steadily and its leader Camillo of Cavour succeeded to guarantee an alliance with France in a future conflict against Austria by sustaining the former in the 1856 Crimean War. The 1859 French-Piedmontese victory against the Habsburgs then triggered insurrections in Tuscany, the conquest of the South by Garibaldi, and the proclamation of the Kingdom of Italy in 1861. Dominated by a narrow elite of northerners (see upper-right graphs of figure 1), the new state favoured the Northern export-oriented farming and manufacturing industries while selecting public spending and the Northern populations when levying the taxes necessary to finance these policies. To illustrate, the 1887 protectionist reform, instead of safeguarding the arboriculture sectors crushed by 1880s fall in prices, shielded the Po Valley wheat breeding and those Northern textile and manufacturing industries that had survived the liberal years thanks to state intervention. While indeed the former dominated the allocation of military clothing contracts, the latter monopolized both coal mining permits and public contracts. A similar logic guided the assignment of the monopoly rights in the steamboat construction and navigation sectors and, notably, the public spending in railway, which represented the 53 percent of the 1861-1911 total. Over this period indeed, Liguria and Piedmont gained a 3 (4) times bigger railway spending per square km than Veneto (the other regions). Moreover, the aim of this effort “was more the military one of controlling the national territory, especially in the South, than favouring commerce” [Iuzzolino et al. 2011, p. 22]. Crucially, this infrastructural program was financed through highly unbalanced land property taxes, which in turn affected the key source of savings available to the investment in the growth sectors absent a developed banking systems. The 1864 reform fixed a 125 million target revenue to be raised from 9 districts resembling the pre-unitary states. The ex-Papal State took on the 10 percent, the ex-Kingdom of Two Sicilies the 40, and the rest of the state (ex-Kingdom of Sardinia) only the 29 (21). To further weigh this burden down, a 20 percent surcharge was added by 1868 creating the disparities displayed in the bottom-left graph of figure 1.

The 1886 cadastral reform opened the way to more egalitarian policies and, after the First World War, to the harmonization of the tax-rates, but the impact of extraction on the economies of the two blocks was at that point irreversible. While indeed a flourishing manufacturing sector was established in the North, the mix of low public spending and heavy taxation squeezed the Southern investment to the point that the local industry and export-oriented farming were wiped out. Moreover, extraction destroyed the relationship between the central state and the southern population by unchaining first a civil war, which brought about 20,000 victims by 1864 and the militarization of the area, and then favouring emigration. Because of these tensions, the population started to display a progressively weaker culture as implied by the fall in our proxy for social capital depicted in the bottom-left graph of figure 2.

The fascist regime’s aversion to migrations and its rush to arming first, and the 1960s pro-South state aids then have further affected the divide, which can be safely attributed to the extractive policies selected by the unitary state between 1861 and 1911.

Empirical Evidence

Because the 13 regions remained agrarian over our 1801-1911 sample, we capture the extent of extraction with the land property taxation and the farming productivity with the geographic drivers of the profitability of the arboriculture and sericulture sectors. In addition, we use as inverse metrics of each region’s tax-collection costs (political relevance) the share of previous decade in which the region partook in external wars (Distance-to-Enemies).

Our fixed region and time effects OLS estimates imply that pre-unitary revenues from land property taxes in 1861 lire per capita decrease with each region’s farming productivity but not with its relevance for the Piedmontese elite, whereas the opposite was true for the post-unitary ones. Moreover, post-unitary distortions in land property tax revenues—proxied with the difference between the observed and the counterfactual ones forecasted through pre-unitary estimates (see upper-left graph of figure 2)—and the severity of the other extractive policies—negatively captured by the tax-collection costs and the political relevance (see below)—positively determined the opening gaps in culture, literacy (see bottom-right graph of figure 2), and development, i.e., the income in 1861 lire per capita, the gross saleable farming product, and the textile industry value added in thousands of 1861 lire per capita.


Figure 2: The Rise of the North-South Divide

2Note: “Distortion-LT” are the land property tax distortions in 1861 lire per capita, “Distortion-R” is the difference between Railway and the forecasted length of railway built in the previous decade in km per square km, “Culture-N” is the normalized share of the active population engaged in political, union, and religious activities, and “Illiterates-N” is the normalized percentage points of illiterates in the population over the age of six. See figure 1 for each cluster definition and de Oliveira and Guerriero (2017) for each variable sources and definition.


These results are consistent with the predictions of the model we lay out to inform our test. First, because of limited state-capacity, the pre-unitary states should reduce extraction if confronted by a more productive and so powerful citizenry, whereas the extractive power of the unitary state should be sufficiently strong to make taxation of the South profitable at the margin and so crucially shaped by his relevance. Second, it should also induce the Southern citizenry to prefer private to public good production and his investment and welfare to rise with factors limiting taxation, i.e., marginal tax-collection costs and political relevance.

Since our proxies for the drivers of extraction are driven by either geographic features independent of human effort or events outside the control of the policy-makers, reverse causation is not an issue. Nevertheless, our results could still be produced by unobserved heterogeneity. To evaluate this aspect, we control for the interactions of time effects with the structural conditions differentiating the two blocks in 1861 and considered key by the extant literature (Franchetti and Sonnino, 1876; Gramsci, 1966; Barbagallo, 1980; Krugman, 1981), i.e., the pre-unitary inclusiveness of political institutions, the land ownership fragmentation, the coal price, and the railway length. Including these controls has little effect on our results. Finally, two extra pieces of evidence rule out the possibility that extraction was an acceptable price for the Italian development (Romeo, 1987). First, it did not shape the manufacturing sector value added. Second, while the pre-unitary length of railway additions was only affected by the farming productivity, the post-unitary one was only driven by the political relevance, resulting useless in creating a unitary market (see upper-right graph of figure 2).


Although the North-South divide has been linked to post-unitary policies before (Salvemini 1963; Cafagna, 1989), nobody has formally clarified how the unitary state solved the trade-off between extraction-related losses and rent-seeking gains. In doing so, we also contribute to the literature comparing extractive and inclusive institutions (North et al., 2009, Acemoglu and Robinson, 2012), endogenizing however the extent of extraction in a setup sufficiently general to be applied to other instances, as for instance the post-Civil War USA.