The economic world is not flat. All advanced economies are characterised by core areas, with high levels of income and wealth, and peripheral regions often specialised in low value-added sectors. One possible explanation for these patterns is that places are inherently different in terms of productivity: core urban areas are frequently characterised by the presence of natural advantages (i.e. first-nature advantages such as large and deep harbours, or a central location in extensive plains with a highly productive agriculture system) that eventually generate agglomeration economies (i.e. second nature).
The distribution of population is also subject to (non-economic) historical shocks that may have had long-lasting effects (Rosenthal and Ross 2015, Schumann 2014). A possible consequence is that the distribution of population is mislocated, i.e. not optimally allocated across areas, and some high-productivity sites end up being sub-optimally under-populated – and low-productivity sites over-populated – with consequences for aggregate growth (Hsieh and Moretti 2015).
In a recent paper, we analysed the extent and the effects of the population mislocation in Italy due to pirate attacks launched from north-African shores against coastal locations. Exposure to piracy and armed attacks started in the 16th century, when north-western Africa (today’s Morocco, Algeria, Tunisia, and Libya) started to be governed by local rulers who chose to live by plunder (Encyclopædia Britannica 1911), and lasted until the early 19th century, when France conquered modern-day Algeria.
Figure 1 provides a historical representation of the Barbary Coast and also illustrates why the south-west coast of the Italian peninsula, Sicily, and Sardinia were the territories most exposed to the attacks.
Figure 1 The Barbary Coast
Source: Antonio Zatta, “Le coste di Barbaria, ovvero i Regni di Marocco, di Fez, di Algeri, di Tunisi, di Tripoli”, Altante Novissimo Vol.4, Venice 1794.
Figure 2 shows the coastal municipalities in Italy that were subject to slave-taking raids between 1516 and 1798 as reported by Davis (2003).
Figure 2 Pirate raids in Italy (1516–1798)
Note: The red dots are the Italian municipalities attacked by pirates between 1516 and 1798, as reported by Davis (2003).
Pirate attacks were frightening. Coastal areas with a high probability of being raided lost their attractiveness for residents. People moved to safer places that were far from the coasts and difficult to assault.
We show that the likelihood of being attacked and the security features of the havens (altitude and ruggedness) explain why low-productivity areas became relatively overpopulated. As a placebo, we also show that the concentration of population in space was very different before the attacks, for instance during Roman times, and when we consider the distance to other trading ports (like Marseille).
Our results are difficult to explain by other causes: neither agriculture-based explanations for urbanisation– such as the adoption of the heavy plough (Andersen et al. 2016) or the start of potato cultivation (Nunn and Qian 2011) – nor population dynamics triggered by earthquakes (Belloc et al. 2016) seem to have a role in our findings.
The effects of pirate attacks on the distribution of Italy’s population were long-lasting (Rosenthal and Ross 2015). The magnitude of the impact measured for 1871 (the first year for which a complete census of Italian cities is available) is quite large even though the attacks had ceased more than 40 years before (Figure 3); the effect is still evident in 1951, although its magnitude diminishes by almost one-third. Thus, the effect persisted notwithstanding two world wars and the exceptional wave of outward migration from the end of the 19th century to the 1920s. The impact ceases to be visible after 1971, erased by the massive South-to-North and rural-to-urban migration, a salient feature of the Italian industrialization process until the mid-1970s.
Figure 3 The impact of pirate attacks on population
Notes: The points connected by the blue line in the graph represent the standardised coefficients for the impact of pirate attacks on population. The explanatory variable is defined as the interaction between a measure for the probability of being a target of pirate attacks (calculated as the inverse distance between the municipality and the port of Tunis) and an index for the protection features of a given location. Vertical bars denote 95% confidence intervals.
We now consider the consequences of the spatial mislocation of the population for a number of local economic outcomes. Due to data constraints, all estimates refer to post-WWII censuses, when measured mislocation was quantitatively smaller than that registered 80 years before. We find that the relative overpopulation driven by pirate attacks in areas less suitable for economic activities correlates with a slower accumulation of human capital, higher migration flows mostly involving younger people, and over-specialisation in subsistence agriculture.
The internal distribution of economic activities affects the growth potential of a country (Baldwin et al. 2001, Fujita and Thisse 2002, Baldwin et al. 2003, Duranton and Puga 2019). If knowledge spillovers are localised, aggregate growth may benefit from the spatial concentration of economic activities in core areas. In our case, the documented relative overpopulation in peripheral areas occurring in the early stages of Italy’s development may have generated a lower steady-state equilibrium. Piracy may have hampered the emergence of larger urban areas, with possible negative effects on steady-state GDP per capita levels.
To test this idea, we proceeded in two steps. First, we computed a counterfactual distribution of city-sizes in 1951 if piracy had not existed in the past; we calculated agglomeration indices on both actual and counterfactual distributions. Second, we used the elasticity of GDP per capita growth with respect to agglomeration (measured by urban primacy) computed by Castells-Quintana (2017) to assess the loss in terms of growth and steady-state GDP per capita level due to under-agglomeration.
Our results are striking. For instance, in 1951 Rome would have been 15% more populous without piracy. The share of the Italian population living in the biggest city (urban primacy) would have been 0.5 percentage points larger. Owing to piracy, the Italian economy converged on a steady-state GDP per capita that was 5% lower than the counterfactual scenario: according to our calculations, the difference in GDP per capita between Italy and Germany would have diminished by one-fourth if Italy had not experienced pirate attacks in the past.
These results may have significant implications in both positive and normative terms. From the positive perspective, we have shown that in the areas affected by pirate attacks, the advantages of relative overpopulation in terms of agglomeration economies were largely outweighed by the disadvantages in terms of productivity; as a result, once the historical event that had determined concentration was concluded, those locations slowly depopulated.
From the normative perspective, our results cast doubt on the economic foundations of the policies intended to halt population outflows from places characterised by low productivity and weak fundamentals. Places experiencing de-population patterns may have been ones slowly reverting to previous historically-driven overpopulation. Policies intended to keep people in bad locations could generate negative consequences on welfare at not only local but also aggregate levels.1
Andersen, T B, P SJensen and C V Skovsgaard (2016), “The heavy plow and the agricultural revolution in Medieval Europe”, Journal of Development Economics 118: 133–49.
Baldwin, R E, P Martin and G Ottaviano (2001), “Global income divergence, trade, and industrialization: The geography of growth take-offs”, Journal of Economic Growth 6(1): 5–37.
Baldwin, R, R Forslid, P Martin, G Ottaviano and F Robert-Nicoud (2003), Economic geography and public policy, Princeton: Princeton University Press.
Belloc, M, F Drago and R Galbiati (2016), “Earthquakes, religion, and transition to self-government in Italian cities”, The Quarterly Journal of Economics 131(4): 1875–926.
Castells-Quintana, D (2017), “Malthus living in a slum: Urban concentration, infrastructure and economic growth”, Journal of Urban Economics 98: 158–73.
Davis, R (2003), Christian slaves, Muslim masters: White slavery in the Mediterranean, the Barbary Coast, and Italy, 1500–1800, London: Palgrave-McMillan.
Duranton, G, and D Puga (2019), “Urban growth and its aggregate implications”, NBER Working Paper w26591.
Encyclopædia Britannica (1911), “Barbary pirates”, Encyclopædia Britannica Vol. III.
Fujita, M, and J-F Thisse (2002), Economics of agglomeration: cities, industrial location, and globalization, Cambridge: Cambridge University Press.
Hsieh, C, and E Moretti (2015), “Why do cities matter? Local growth and aggregate growth”, NBER Working Paper w21154.
Nunn, N, and N Qian (2011), “The potato's contribution to population and urbanization: evidence from a historical experiment”, The Quarterly Journal of Economics 126(2): 593–650.
Rosenthal, S, and S L Ross (2015), “Change and persistence in the economic status of neighborhoods and cities”, The Handbook of Regional and Urban Economics 5.
Schumann, A (2014), “Persistence of population shocks: Evidence from the occupation of West Germany after World War II”, American Economic Journal: Applied Economics 6(3): 189–205.
1 For example, the Italian government has recently proposed a programme for the development of internal areas (‘Aree Interne’), with the aim of resisting depopulation and attracting economic activities (see http://www.dps.tesoro.it/aree_interne/ml.asp). Many of the areas eligible under this policy are exactly those that – because of the pirate threat – were overpopulated until the mid-1970s.
The views expressed in this column are those of the authors only. They do not necessarily represent the views of the Bank of Italy