Origins of the Sicilian Mafia:
The Market for Lemons
Arcangelo Dimico∗
Alessia Isopi
Ola Olsson
December 31, 2012
Abstract
In this paper, we study the emergence of an extractive institution that hampered
economic development in Italy for more than a century: the Sicilian mafia. Since its
first appearance in the late 1800s, the origins of the Sicilian mafia have largely remained
a mystery. Both institutional and historical explanations have been proposed in the
literature through the years. In this paper, we develop an argument for a market structurehypothesis, contending that mafia arose in towns where firms made unusually high
profits due to imperfect competition. We identify the production of citrus fruits as a
sector with very high international demand as well as substantial fixed costs that acted
as a barrier to entry in many places and secured high profits in others. We argue that
the mafia arose out of the need to protect citrus production from predation by thieves.
Using the original data from a parliamentary inquiry in 1881-86 on Sicilian towns, we
show that mafia presence is strongly related to the production of oranges and lemons.
This result contrasts recent work that emphasizes the importance of land reforms and a
broadening of property rights as the main reason for the emergence of mafia protection.
We also show that the local distribution of the mafia in the 1880s is a strong predictor
of mafia activity in recent years.
Keywords: mafia, Sicily, protection, barriers to entry, dominant position
JEL Codes:
1
Introduction
The institutional hypothesis of economic development, emphasizing the importance of private
property rights and constraints against the ruling elite, has recently received strong support
in empirical studies. Several cross-country studies have shown that countries with a firm
rule of law typically also have better economic performance. But how do socially inefficient
institutions arise and how can they sometimes survive for centuries?
The Sicilian mafia is arguably one of the most infamous and economically harmful institutions in the world. After its first appearance in Sicily in the 1870s,1 it soon infiltrated
the economic and political spheres of Italy and the United States and has at times been
∗
Corresponding author. Email: a.dimico@qub.ac.uk. Special thanks to Giuliano Isopi for excellent research
assistance. We gratefully acknowledge comments from Gani Aldashev, Jean-Marie Baland, Michael Bleaney,
Paolo Casini, Chris Colvin, Giacomo de Luca, Stefano Fenoaltea, Halvor Mehlum, Kalle Moene, Kyriacos
Neanidis, Jean-Philippe Platteau, Oleg Shchetinin, Yves Zenou, and seminar participants in Gothenburg,
Leuven, Birkbeck College, the EEA-ESEM meetings in Malaga, Manchester, Namur, Oslo, and Stockholm.
1
The first evidence of the presence of a secret sect (referred to as cosca) we have found is an account from
1872 by Dr Galati, a landlord who used to own a lemon grove just outside Palermo, who wrote about a “man
of honor” who made increasing use of violence and extortion to force him to sell his lemon grove (Dickie,
1
considered a serious threat to the rule of law in both countries. Although outcomes of the
mafia’s actions such as murder, bombings, and embezzlement of public money have been
readily observed during the last 140 years, its origins have largely remained a mystery.
In this paper, we provide a study of the origins of the Sicilian mafia using a unique dataset
from the latter part of the 19th century. The main hypothesis that we investigate is that the
origins of the Sicilian mafia are strongly associated with the unusually high profits made by
citrus fruit producers following a boom in international demand. We argue that the primary
source of these abnormal profits, especially for lemon producers, is found in a particular
market imperfection: the high and geographically varying fixed costs of production. These
high barriers to entry implied that citrus trees could only be cultivated in certain areas and
that a few local producers earned substantial profits from the boom.2 The combination of
high profits, a weak rule of law, a low level of interpersonal trust, and a large number of poor
people implied that lemon producers were a natural object of predation by thieves. Given
the poor ability of the government in protecting private property rights at the time, lemon
producers resorted to hiring mafia affiliates for private protection.
Using historical statistics and a formal model featuring households, producers, and the
mafia, we develop the argument that the market for citrus faced an exceptionally high demand during the mid- to late 1800s and that the high and regionally varying fixed costs
of production implied supernormal profits in certain locations. In the empirical section, we
then test the main implications of the model using a unique dataset from Sicilian towns and
districts gathered from a parliamentary inquiry conducted in 1881-86 (Damiani, 1886). Our
results indicate that mafia presence in the 1880s is strongly associated with the prevalence
of citrus cultivation, controlling for a number of other potential covariates. No other crop or
industry has a robust impact on mafia presence. We interpret these findings as being consistent with a market structure-explanation to the origins of mafia. We conclude by showing
that the historical distribution of mafia is a strong predictor for more recent mafia activity.
Our paper relates to several different strands of literature.3 First, it is related to the
literature on the historical emergence of “extractive” institutions that hamper economic
development and that can appear at critical junctures in a country’s history (Acemoglu et al.,
2006; Acemoglu and Robinson, 2012). The mafia is undoubtedly an example of an extractive
institution that emerged during a critical period in history (Italian unification). Our analysis
is, however, somewhat different from this tradition since we emphasize the economic or
market structure related factors behind mafia organizations rather than political origins
(such as a weak and oppressive Bourbon state in Sicily with substantial social inequalities,
as discussed further below).
2004). When the Minister of Home Affairs learned about Galati’s note, he soon asked for a written report
from the chief of police in Palermo and then ordered two parliamentary inquiries (the Bonfadini Inquiry in
1876 and later the Damiani Inquiry in 1881-5) to better understand the economic conditions and the situation
of crime in Sicily.
2
We argue that citrus production was associated with unusually heavy barriers to entry due to the high
fixed cost of planting trees and waiting several years for them to grow, the need to build protective walls to
keep thieves out, etc. Due to a large regional variation in the climate and soil suitability for growing lemons,
the fixed costs of starting up a cultivation was very different across towns.
3
Please see the working paper version for a more extensive literature review.
2
Our results are also strongly associated with research on the “curse of natural resources”
(see Van der Ploeg, 2011, for a recent overview). We find that the boom in citrus exports
in the late 19th century is a key factor behind the rise of mafia. This is consistent with
the more recent finding that windfall gains from natural resources are often associated with
intense rent seeking and patronage politics. For instance, Sala-i-Martin and Subramanian
(2003) argue that political corruption related to oil revenues hampered Nigeria’s growth for
decades. Acemoglu et al. (2004) show how mineral wealth in Zaire allowed President Mobutu
to buy off political challengers. A main theme in this tradition is that resource windfalls
might actually destabilize and deteriorate institutions if key groups in society believe that
predation is more profitable than production (Mehlum et al., 2006; Congdon Fors and Olsson,
2007).
Another literature that our analysis connects to is that on the economic analysis of organized crime, which focuses on weak institutions, predation, and enforcement of property
rights (Fiorentini, 1999; Grossman, 1995; Anderson, 1995; Skaperdas and Syropoulus, 1995;
Skaperdas, 2001). Grossman (1995) and Skaperdas (2001) both consider mafia as an alternative enforcer of property rights. Using a model with two actors (a self-governing community
and mafia) and potential robbers, Skaperdas (2001) shows that in the absence of an enforcer
of property rights, mafia can represent a sort of second best solution.4 Regarding the economic costs of organized crime, Reuter (1987) and Gambetta and Reuter (1995) analyze the
effect of organized crime on the enforcement of cartel agreements in legal markets.
Regarding the weak institutions-hypothesis, the loss of social capital and public trust may
also play a key role for the development of a private provider of protection. Putnam (1993),
for example, analyzes the loss of social capital in Southern Italy due to the repeated foreign
dominations experienced in the region. Gambetta (1996) also considers private trust rather
than public trust as very important for the development of mafia in southern Italy. Social
capital together with a loss of public trust may further affect the development of organized
crime because of kinship relations, corruption, etc. (see Fukuyama 2000; Gambetta, 2009;
Levi, 1996; Hardin, 1999; Newton, 2001).
Yet, our paper is most closely related to Bandiera (2003). Bandiera’s main hypothesis
is that it was the increase in land fragmentation following the Bourbon-era land reforms
(1816-1860) that provided the breeding ground for mafia protection. A higher number of
land owners increased the need for private protection. In Bandiera’s model, a key feature is
that protection of one producer generates a negative externality on other producers, since it
makes them more likely objects of predation. In an empirical section where she uses information from the Summary Report presented to the Italian parliament by Damiani (1886),
Bandiera (2003) finds that a variable capturing differences in land fragmentation is a significant determinant of mafia presence.5
4
The idea of a weak state and private protection is well illustrated by Don Calo Vizzini, one of the historical
bosses of the Villalba mafia. In an interview with Indro Montanelli, he said that ”...the fact is that in every
society there has to be a category of people who straighten things out when situations get complicated. Usually
they are functionaries of the state. Where the state is not present, or where it does not have sufficient force,
this is done by private individuals” (Montanelli, 1949).
5
The information available in the Damiani Inquiry (see Footnote 1) has previously been used also by
3
While our analysis also identifies landowners’ demand for private protection as the main
process through which the mafia was mobilized, we explicitly focus on the role of the market
structure rather than on land fragmentation. With respect to Bandiera (2003), we improve
the sample size by using the original survey where pretori (lower court judges) provided
answers related to crimes within the survey that Damiani sent out in 1883 (see Figures A1
and A2 in the Appendix). The original copy of the Daminai Inquiry permits us to increase
the sample size from the 70 towns located in the western part of the island in Bandiera (2003)
to almost all available Sicilian towns (143 in total) for which pretori provided answers. With
this more complete sample, we find that the land fragmentation variables indeed explain some
of the variation in mafia presence. However, we also find support for a link between mafia
and the prevalence of large scale plantations. The latter finding confounds the interpretation
that the mafia appeared as a result of land reform. Our main result is that the most robust
determinant of mafia activity is production of citrus fruits.
The working paper by Buonanno et al. (2012) also studies the importance of export
markets for mafia appearance using data from Cutrera (1900), a police officer in Palermo.
Cutrera uses as sources Colajanni (1900), Alongi (1887) and other data from local police
to elaborate a map of Sicily where the intensity of mafia activity is outlined for every city.
6
Even though the data provide figures on the level of mafia in almost all Sicilian cities
at the beginning of the 20th century, they capture the situation prevailing almost 20 years
later than the Inquiry. Over these twenty years, the mafia extended its activity to cities that
initially were not affected and hence the data from Cutrera are more useful for understanding
the temporal evolution of the mafia.7 Buonanno et al. (2012) find that sulphur production
had a strong association with mafia presence in 1900. However, they do not develop an
explicit argument for why export revenues were associated with mafia revenues in certain
sectors.
Besides the economic literature, our analysis is also related to a long tradition of works in
anthropology, sociology and history on the Sicilian mafia. The classical contributions include
early investigations such as Villari (1875), Sonnino and Franchetti (1877), and Colajanni
(1885, 1895). In recent years, the origin of the mafia has also been discussed in Gambetta
(1996), Dickie (2004), and Lupo (2009).8 While Lupo (2009) and Dickie (2004) consider
profits related to the lemon industry in the Western part of the island as a pre-condition
for the development of the mafia, Gambetta (1996) focuses on the division of land resulting
from the abolition of feudalism and other policies introduced by the Italian government after
1860 (i.e., sale of land owned by the church and the crown before the unification). These
other scholars studying the origins of the Sicilian mafia; see for instance Colajanni (1885, 1895), Hess (1973),
Arlacchi (1986), Catanzaro (1992), Gambetta (1996), Dickie (2004) and Lupo (2007).
6
Alongi (1887) and Colajanni (1900) in their turn use the information available from the original inquiry.
Therefore, their books represent a further elaboration of the results collected by the Damiani Inquiry, which
could potentially add a bias in the data provided in Cutrera (1900).
7
This is confirmed by Gambetta (1996) who argues that in the period between the late 1870s and late
1890s, the mafia evolved quite markedly as a result of “economic and political conflicts among local factions, especially in connection with the institutional changes affected by the Italian state between 1869-1890”
(Gambetta, 1996, pg.83).
8
See also Lupo (2009) for a general history and Monroe (1909) for a description of the agricultural practices
in Sicily at the time.
4
policies opened a market for private protection where the mafia acted as an incumbent.
The extensive literature discussed above provides plausible explanations for the origins of
the mafia. Yet, with the exception of Bandiera (2003) and Buonanno et al. (2012), it is still
difficult to understand on the basis of these works why we observe a substantial variation
across regions experiencing very similar social conditions. If a weak state, a high regulatory
burden, and a lack of public trust are the factors that matter for the development of mafia,
then we should not observe any local variation within the territory at hand. However, this
is not the case as across counties and villages exposed to these same conditions there is a
notable variation in mafia presence. Actually, organized forms of crime initially appeared
only in a small number of localities before expanding over the entire region.
The market structure hypothesis advanced in our paper not only complements existing
theories of mafia emergence, for instance those focusing on political factors, but is also
extremely consistent with the timing of the origins of mafia. The scientific discovery in the
late 1700s that citrus fruits could cure scurvy led to a strong surge in demand during the
following century. By 1880 (the period for which we have data on mafia activity), Sicily
had become the largest supplier of lemons and lemon by-products in the world. In fact, the
island covered more than 78 percent of total lemon imports in the US, which at the time was
the largest importer from Sicily. This exogenous shock in the international lemon market
therefore provides us with the basis for considering profits in the sector as the factor able to
explain the cross-regional variation in the appearance of mafia.9
In summary, we believe our paper makes the following contributions to the existing
literature: First, it provides a formal model of how the market structure and the prevalence
of cross-sectional variation in fixed costs affect the demand for mafia protection. Second,
we offer the most comprehensive empirical analysis to date on the origins and persistence of
mafia since the 1880s and identify a novel explanation for the emergence of the mafia during
the period.
The paper is structured as follows: Section 2 gives a background to the Sicilian economy
and the role of citrus fruits. Section 3 outlines the formal model. Section 4 includes the
econometric specification and a discussion of the data, whereas the main empirical results
are found in Section 5. Section 6 concludes the paper.
2
The Sicilian Economy and the Role of Citrus Fruits
Sicily is the largest island in the Mediterranean. It is easily accessible by the sea from all
directions and has a varied and fertile terrain and, above all, an unparalleled location: the
island in fact has been both a gateway and a crossroads, on the one hand dividing the
eastern and western Mediterranean, on the other linking Europe and Africa as a stepping
stone (Finley et al., 1987).
Because of the island’s importance and strategic location, its past has been marked by
9
Following this exogenous shock, one might expect local producers to have increased supply. However,
this did not happen everywhere, but mostly along the coastline. The reason is related to the cross-regional
variation in the climate that we claim represents a barrier to the expansion of the sector inland.
5
continuous foreign domination and the list of migrants and colonizers is long: First the
Greeks, then the Romans followed by the Byzantines, then the Arabs, the Normans, the
Spanish, the French, and then the Spanish again. This long period of different foreign rulers
had an important impact on the the historical development of Sicily. The material advantages
of the foreign powers were created at the expense of the Sicilians through rents, taxes, and
plain looting. This parasitical interest inevitably did great harm to the countryside and to
the people.
At the beginning of the 19th century, overseas commerce was mostly carried out by the
French and a native industry in Sicily hardly existed except at household level. Sicily lacked
the main prerequisites for an industrial revolution: no iron, no coal, no navigable canals
and rivers and very poor internal communications. The middle class remained professional
and bureaucratic rather than commercial and industrial, while the leaders of society lived
off rents.
Before 1861, the vast majority of the population was employed in agriculture. The
production system in the agricultural sector reflected the typical feudal system. Villari
(1875) reckons that there were three main social classes in Sicily: 1) landlords, 2) a middle
class (gabelloti ), 3) and peasants who were normally exploited by the gabelloti. The gabelloti
leased the land from landlords and then rented small pieces of it to the peasants. Peasants
worked the land and gave back to the gabelloti a share of the harvest depending on the kind
of contract stipulated. The contracts were relatively short and generally designed to exploit
the peasants. When the yield was not enough, peasants usually had to borrow from the
gabelloti at interest rates that made it impossible to pay back the debt.
The situation did not change much after Sicily joined the Reign of Italy. In 1887 the share
of citizens that owned the land was still the lowest in Italy with an average of less than 2.05
owners per 100 citizens, compared to for instance 15 owners per 100 in Piedmont (Colajanni,
1885).10 In addition, almost 56 percent of the population employed in agriculture owned less
than one hectare of land and most of them were hired by the landlord on a daily basis at
less than one lira per day rate.
Despite its underdeveloped economy, Sicily was a leading producer of wheat, olive oil,
wine, and citrus fruits. The international demand for lemons started increasing from the
late 1700s when lemons and lemon juice became a standard preventive treatment against
scurvy.11 In the words of Baron (2009): ”The Sick and Hurt Commission agreed to supply
all naval ships on foreign service with lemon juice, extended in 1799 to all the ships on the
British coast. Between 1795 and 1814 the admiralty issued 1.6 million gallons of lemon juice.
Sweet lemons were imported, especially from the Mediterranean region turning Sicily into a
vast lemon juice factory” (Baron, 2009, pg 324).
In the late nineteenth century, Sicilian production of citrus fruits represented almost 73
percent of the total production in Italy (Pescosolido, 2010).12 Table 1 reports the distribution
10
In 1840, 127 princes, 78 dukes, 130 marquises, and an unknown number of earls and barons had complete
control over most of the land (Travelyan, 2001).
11
Scientific support for the theory that consumption of citrus fruits cured scurvy was established by James
Lind, a British naval officer and surgeon, in the latter part of the 18th century.
12
Other reasons why Sicily may have had a dominant position in this sector are historical (related to the fact
6
of lemon trees in southern Italy in 1898. Palermo and Messina were the two provinces with
the largest absolute number of lemon trees in Sicily, accounting for almost 59.5 percent of
the total trees, followed by Catania with approximately 12 percent. Outside Sicily, in the
Calabria region, Reggio Calabria was also a large producer of lemon with an absolute number
of trees of 1,232,675 (almost 18 percent).13
Table 1: Distribution of lemon trees in Southern Italy in 1898
In the late 19th century, the dominant position of Sicily was consolidated as a result
of a significant expansion in global demand. The total surface area devoted to the citrus
production went from 7,695 hectares in 1853 to 26,840 hectares in 1880 (Pescosolido, 2010).
The expansion was a result of the large returns associated with citrus productions. Monroe
(1909) estimates that revenues were almost $200 per acre (in 1908 US dollars), providing a
net profit of more than $150 per acre. The importance of these figures is further emphasized
by Dickie (2004) when he argues that “citrus cultivation yielded more than sixty times the
average profit per hectare for the rest of the island” (Dickie, 2004, p.39).
In the period 1881-85, the quantity of citrus exported was almost 949,000 units, compared
to 250,000 units in 1850 (Pescosolido, 2010). A large share of this production was exported
to the US, which was one of the largest markets for Sicilian citrus exports. A combination of
factors contributed to this outcome: a favorable international context, elimination of exports
duties, and a considerable improvement of transports.
Table 2 shows figures on the lemon trade between Italy and the US in 1898-1903. The
left-hand side of the table shows the total Italian lemons exports and the relative percentage
exported to the US. The right-hand side shows the total US lemons imports and the estimated
percentage coming from Italy.14 The average quantity of lemons exported from Italy (and
therefore mainly from Sicily) over this period amounts to 389 million pounds and the average
share of fruits imported from the US is almost 34 percent of the total Italian production.
On the other hand, most lemons imported form the US in this period came from Italy.
Calculating the total Italian exports to the US, using percentages to the right in the table,
we can estimate that almost 78.4 percent of the total US lemons imports between 1898-1903
came from Italy. Besides the US, the UK and Austria were also two large importers of
lemons. Over the decade 1898-1908, the UK imported 17.7-25 percent of the total Italian
lemons exports, and the Austro-Hungary imported 14.4-22.8 percent (Powell, 1909).
Table 2: Total Italian exports of lemon and total United States imports
Table 3 shows the total Italian exports of citrate of lime (which is a soft drink) as well
as total imports of it into the USA 1899-1903 period. The US is again the most important
that the plant developed in the Mediterranean starting in the Roman era) and geographical (the importance
of the Mediterranean in international trade).
13
Some trees were also planted in the central parts (almost 798,214 trees) and in northern Italy (almost
564,559). In total, there were 8,287,758 lemon trees across the peninsula, but Sicily and Calabria regions
accounted together for almost 91% of the total Italian production.
14
These data should to be evaluated with some caution given that the total Italina lemons exports refer to
the calendar year while the total US imports refer to the fiscal year (and the two do not coincide).
7
market for citrate. Besides the US, France and the UK are also important markets for citrate
from Italy and according to Powell (1909) the total quantity imported by these two countries
combined was at least as large as that for the US. A similar pattern can also be shown for
the export of essential oils from lemons and oranges.
Table 3: Total Italian exports of citrate of lime and total United States imports
Given the extent of the production and the international demand for lemons, the sector
was of strategic importance for the Sicilian economy.15 As shown in Table 4, the total export
revenue from the harbor of Messina in 1850 was approximately 21.6 million lire (at current
price). Revenues from citrus and derived products equaled almost 9.2 million lire, accounting
for almost 42.4 percent of the total export revenues. In 1873 the percentage went up to more
than 50 percent of the total exports (Battaglia, 2003).
Table 4: Exports from the Harbor of Messina in 1850
As mentioned earlier, these substantial revenues are mainly explained by Sicily’s geographic location and associated climate. In particular, the island’s hot coastal plains,
together with the exceptionally fertile soil made of limestone base with heavy coatings of
lava, were well-suited for growing citrus fruits. Lemon trees have a very poor tolerance for
extreme climates and require very specific climatic conditions. The average temperature required for them to grow and vegetate is 13-30 ◦ C, and flowers and fruits die after only a few
minutes of exposure to temperatures below 1-2 ◦ C. This high intolerance to frost explains
the local variation in places where the trees can vegetate: places slightly above the coastline
are more suitable because of the relatively low variation in daily (and annual) temperature,
while locations in the mountains with a larger daily (and annual) variation in temperature
are less suitable for cultivation. In addition, lemon trees require constant provision of water.
The irrigation system used in the plains surrounding Palermo worked thanks to the norie,
a quite advanced technology that enabled farmers to collect water very deeply in the soil.16
Yet, given that local institutions often failed to provide an adequate funding of the irrigation
system, local producers had to make costly investments in order to build and maintain it.
We argue that the high export revenues from citrus, combined with the sizeable initial
investments needed for the development of citrus cultivation, made local producers very
vulnerable to predation. In the absence of state protection, the mafia exploited this systemic
vulnerability to extort part of the profits made by the industry. Consequently, we consider
profits coming from imperfect market structures as a natural condition for the development
of the mafia.
15
The leading role of Sicily for the US lemons market continued until the end of the 19th century, when
the production in Florida became substantial.
16
The norie were initially moved by animals, but already in the 1872 they started to be moved by steam
pumps.
8
3
The Model
The model considers three active agents -households, lemon producers, and the mafia - as
well as a (latent) government that determines the strength of property rights. The aim of
the model is to explain how the structure of fixed costs affected the decision of whether to
produce lemons or not and, if production occurred, under what circumstances the lemon
producers chose to pay the mafia for protection against thieves. The model is meant to
describe the situation in the 1880s and is not necessarily relevant for understanding the
contemporary nature of mafia operations.
3.1
Households
Let us assume that the utility of a representative household is given by
U = α ln C + (1 − α) ln X
where U is utility, C is consumption of lemons from Sicily, and X is consumption of other
goods.17 The parameter α indicates the taste for lemons within the representative household.
For simplicity, we assume that the island of Sicily has a monopoly in the production of
lemon.18 The representative household might be thought of as the average Western household
in the last decades of the 1800s. The individual household’s budget constraint is Y =
pC + X, where Y is average household income at the time and p is the relative price of
lemon consumption (price of other goods X is normalized to unity).
From the first-order conditions for profit maximization, we can obtain the inverse demand
function for Sicilian lemons:
p (C) =
αY
C
As usual, there is a negative association between price and the total level of demand C,
whereas demand rises with income and with the preference for lemons α.
3.2
Lemon producers
There are in total I > 0 towns or municipalities in Sicily and 1 < n ≤ I towns where lemons
are produced. For simplicity, we assume that each town has one (representative) producer.
Total supply of Sicilian lemons is C =Ii=1 Ci where Ci is the local level of production in town
i. Total supply always equals total demand.
Profit of the local producer in town i is
πi = p (C) · Ci − γ (Ci ) − Fi =
αY
· Ci − ψCi − Fi
C
17
We will only refer to lemons in the model below, but as indicated in the section above, what we really
have in mind is the market for lemons, oranges, limes, and other citrus products.
18
This is a simplification. However, Table 2 shows that in the dominant US market, Sicilian lemon accounted
for 100 percent of all imports during certain years in the 1898-1903 period.
9
where p (C) is the price level (which depends on total demand), γ (Ci ) is a marginal cost
function such that γ 0 (Ci ) = ψ > 0, and Fi is the local fixed cost of entry into lemon
cultivation. Fi depends on local characteristics such as soil quality, water access, altitude,
and slope of the land, as well as non-community specific fixed costs such as costs of building
protective walls, etc. Typically, it takes several years before planted lemon trees have grown
to produce lemons. Once a lemon plantation has been established, the marginal cost ψ is
the same across localities.
The first-order condition for profit maximization can be written as
Ci
0
p (C) 1 + p (C) ·
= γ 0 (Ci ) .
p (C)
Since marginal cost ψ and inverse demand p (C) are the same everywhere, Ci must in
optimum be identical in every town. Hence, C = nCi . The expression above can therefore
be written as
αY
nCi∗
1
1−
=ψ
n
The fact that the number of towns n ≤ I is bounded from above implies that there will
be a positive mark-up over marginal cost and that the market is not fully competitive.
Solving for the Cournot equilibrium supply of lemon from town i gives us
Ci∗ =
αY (n − 1)
.
n2 ψ
(1)
Not surprisingly, equilibrium supply will increase with the typical income Y and decrease
with marginal cost ψ. Furthermore, it can easily be shown that Ci∗ will decrease with n for
all n > 2.
Inserting Ci∗ back into the profit function, we receive after some algebra the optimal
profit level
πi∗ =
αY
αY
· Ci∗ − ψCi∗ − Fi = 2 − Fi .
∗
nCi
n
In this very simple expression, profits increase with income and decrease with the number
of towns producing n. Obviously, lemons will only be produced in community i if πi∗ =
αY
n2
− Fi ≥ 0. Hence, fixed costs and the number of other producers are potential barriers to
entry into lemon production.
Let us assume that towns i ∈ {1, 2, 3, , , I} are ordered such that F1 < F2 < F3 ... < FI .
Let us further assume that fixed costs are uniformly distributed across towns and that they
are simply given by
Fi = a + bi
where a > 0 is a component common to all towns and where b > 0 is a parameter describing
the gradient of fixed costs across towns. One might for instance think of a as capturing the
cost of building protective walls, which is roughly the same everywhere, whereas b might
capture the difference in fixed costs that arises due to differences in soil quality or access to
10
surface water for irrigation which make it more costly in terms of time and effort to establish
a lemon plantation in some places rather than in others. Clearly, a b close to zero would
imply small differences between towns. The mean fixed cost across towns is F̄ = a + I · b/2.
With these assumptions, the last producer who will choose to produce lemon (i = n) will
be the one for which the following condition holds:19
πn∗ =
αY
αY
− Fn = 2 − a − bn = 0.
2
n
n
(2)
All potential producers i ∈ {1, 2, 3, , , n} will thus produce whereas i ∈ {n + 1, ...I} will
not. By using the implicit function theorem, we can deduce from the equation above that
the equilibrium level of lemon-growing towns is a function n = n (a, b) such that
∂n
= na =
∂a
∂n
−1
< 0;
= nb =
∂b
+b
2αY
n3
−n
< 0.
+b
2αY
n3
Although the explicit solution to n is mathematically messy, it is easily illustrated in a
graph as in Figure 1. The figure shows the two components of the profit level, αY /i2 and
a + bi, as a function of i when towns are ordered, starting from the one with the lowest fixed
costs to the left. The equilibrium occurs at the point where the two lines cross. At n, profits
for the nth firm are zero whereas they are given by the distance between αY /n2 and a + b
for the firm with the lowest fixed costs.20 The triangle D in the figure shows the total profits
of the lemon producing sector in Sicily.
Figure 1: Equilibrium number of lemon-producing towns
It is clear from the figure that an increase in a and/or b would shift the Fi curve to the
left, resulting in a lower n. Over time, it is likely that such barriers to entry have varied
in the lemon trade just as in other sectors. As a thought experiment, one might imagine
another agricultural good (perhaps wheat) with the same profit function except that it had
lower barriers to entry al < a and bl < b, as shown in the bottom of Figure 1. Such low
levels of fixed costs would imply that all towns (n = I) would produce the good and that
average profits would be quite small. Total profits in the sector are given by the distance
between the fixed cost curve Fi = al + bl i and the profit level αY /I (the area E).
Hence, the individual profit for an actual producer is
πi∗ =
αY
− a − bi ≥ 0
n (a, b)2
for all i ≤ n.
An increase in the fixed cost coefficients a and b thus has two effects on equilibrium
profits: it reduces the equilibrium number of lemon producers, which has a positive effect
on profits in town i, and leads to an increase in the fixed costs for all producers, which
decreases profits. The sign of the comparative statics will depend crucially on the level of
In the expression below, we assume for simplicity that there is always a level of profits where πn∗ = 0.
In reality, the equilibrium number
of lemon-producing towns n∗ would probably rather be defined by n∗ =
arg min max αY
−
a
−
bn,
0
.
n2
20
The profit level for the 1st firm is equal to b (n − 1) > 0.
19
11
i.21 In general, for a given n, profits fall with i. Profits always increase with household
demand αY . We can therefore express πi∗ = π (αY, a, b, i).
3.3
Government
As described above, Sicily in the 1880s was characterized by weak property rights institutions
and a substantial number of thieves who predated on agricultural production. An implicit
assumption in this section is that the “predation technology” in the lemon business was
particularly favorable for thieves. Compared to other agricultural goods like grapes or wheat,
lemons are very easy to collect quickly by a prospective thief, and the price per stolen bucket
is further very high. These factors contributed to lemon plantations being in particular need
of protection.
Let us assume that in each community there are d > 0 thieves.22 In the absence of
property rights and other forms of protection, thieves would steal the full profit from lemon
production and each thief would obtain an amount of πi∗ /d. Lemon production would then
make zero profits. The government in Rome offers some protection of property rights captured by the term θ ∈ [0, 1], where θ = 1 implies perfect enforcement of property rights
whereas θ = 0 implies total absence of government protection. For Sicily in 1880, θ was
presumably closer to 0.
The total proportion of profits saved from thieves by the individual lemon producer is
given by the ”predation success function”
ρ (mi ) =
mi
mi + d (1 − θ)
where mi is the level of private protection offered in i.
The functional form implies that if θ = 1, there is no need for private protection since
ρ (mi ) = 1 for any level of mi . If θ > 0, then ρ0 (mi ) > 0 and ρ00 (mi ) < 0, i.e., the proportion
of protected profits is a positive, concave function of the level of private protection. Lemon
producers then retain ρ (mi ) · πi∗ of their profits and lose (1 − ρ (mi )) · πi∗ to the thieves.
Lemon producers cannot provide protection themselves and hence need to employ people
to do this job for them. This is where the mafia comes in.
3.4
Mafia
The nature of the original organization of local mafia groups (cosca) remains largely a mystery. What we know about such groups is that they formed a secret society of sworn-in
men who managed to overcome the collective action problem through various measures (like
brutal punishments in the case of defection). Mafiosi were recruited among men with very
21
We can for instance see from Figure 1 that a rise in b with a unchanged will increase profits for the town
with the lowest fixed costs, whereas the previous nth firm will then have negative profits and should cease to
produce.
22
In a richer model, the number of thieves might be endogenized so that individuals self-selected into being
a mafioso, a thief, or a normal peasant in a process where marginal returns were the same in equilibrium.
The number of thieves in Sicily in the 1880s was reportedly very high due to a general release of prisoners
after the Italian unification and the breakup of feudal estates, which made many workers redundant.
12
diverse occupations in society, including peasants, sheep-herders, doctors, and politicians.
In these early days, mafiosi typically performed their daily jobs as an integrated part of
society while also undertaking mafia activities on the side. The key mafia activity was the
protection of businesses (Gambetta, 1996).
We assume that the local mafia organization in i has no influence over n (there was
no central coordinating mafia authority in the 1880s) and that a representative mafioso
considers the choice of allocating effort either to protecting local lemon producers or to
pursuing normal economic activity. A representative mafioso’s utility function in town i is
UiM = ωρ (mi ) πi∗ + (1 − mi ) A
where mi ∈ [0, 1] is available effort that can be spent on protecting the local lemon producers’
profits. The parameter A > 0 reflects productivity in normal production (farming, fishing,
herding sheep, etc). This type of production is one option available to mafiosi and is the
only available option for the majority of ordinary people. ω ∈ (0, 1) is the share of total
protected profits that the local producers offer to the mafia in return for protection. For
now, let us take ω as given.23 Note that ω must be somewhere within the interval (0, 1) for
any interaction to occur between the two.
The mafia maximizes the utility function
max U M =
mi
ωmi πi∗
+ (1 − mi ) A.
mi + d (1 − θ)
After manipulating the first-order conditions, we can solve for the optimal (interior solution) level of mafia activity in town i:
s
m∗i = d (1 − θ)
ωπi∗
−1
d (1 − θ) A
!
(3)
Clearly, m∗i > 0 will only be the case if the term under the square root sign exceeds 1.
The expression in (??) implies that we can express the following proposition:
Proposition 1: The mafia will be active in town i (m∗i > 0) only if ωπi∗ = ω
αY
n(a,b)2
− a − bi >
d (1 − θ) A.
This proposition offers some of the key insights of the model. If the opportunity costs of
being a mafioso A are very large, there will be no mafia. If the offer from the producers ω
is very low, the mafioso will not find protection worthwhile. Furthermore, it will obviously
be the case that there will be no mafia if property rights are fully enforced, i.e. if θ = 1.
It can be shown that m∗i is a decreasing, convex function of θ so that the mafia shrinks as
government-enforced property rights are strengthened. Similarly, there will be no mafia if
23
Please see Figure A3 in the Appendix for an extension and simulation of the model where ωi is endogenously determined in a two-stage game between lemon producers and the mafia. The main comparative
statics results below still hold after this extension.
13
there are no thieves so that d = 0. All these factors are assumed to be identical throughout
Sicily but might explain the varying presence of mafia over time.
What distinguishes towns is the level of profits in lemon production πi∗ . The central result
is of course that the likelihood of mafia presence increases with πi∗ . As discussed above, we
argue that one of the key distinguishing features of lemon production at the time was the
relatively high demand αY and the high barriers to entry due to high and geographically
differentiated fixed costs, represented by the parameters a and b. If these are high, then only
n < I towns will be able to produce and the average profit among these producers will be
relatively high. For other goods, we argue that a and b should be fairly low, implying low
profits in general and no large geographical variation in profits. The lower part of Figure 1
depicts such a scenario. Profits are then less likely to motivate a mafia to arise from (??).
The most likely place for mafia presence would be town i = 1 where fixed costs of lemon
production are lowest and profits are highest. In our empirical investigation, we do not have
data on profits from various types of production. What we do have data on is the presence
of sectors in each town. According to our model and the data discussed above, the presence
of lemon production in some towns should be an indicator of profitability and of low fixed
costs. Similarly, the presence of other types of production are interpreted as indicating that
profits in that sector were also positive. Holding the presence of other types of production
constant, we hypothesize that the prevalence of lemon production in a town should thus have
a positive association with the probability of mafia activity.
4
Econometric Specification and Data
4.1
Econometric Specification
From an econometric point of view, we can consider equation (??) as the latent equation
that will determine the probability of mafia presence. In this equation, the probability of
mafia presence depends on profits, the enforcement of property rights, and the number of
thieves. The latter are considered equally distributed across the region even though there
may be variation in the efficiency of the state at town level, which can explain the presence
of mafia across the region. For this reason, the efficiency of the enforcement of property
rights will represent part of the control variables.
The model to be estimated can be written as:
?
Mi,p
= ηp + β1 Πi,p + β2 Zi,p + µi,p
where
(
M af iai,p =
(4)
? ≥0
1 if Mi,p
? <0
0 if Mi,p
? represents the response variable
In the latent equation (??), the dependent variable Mi,p
that drives the probability of mafia in town i in province p. A response variable larger than
zero is associated with towns with a positive level of mafia presence. The probability of
14
? ) is smaller than zero.
mafia is zero if the response variable (Mi,p
The main explanatory variable is local profits from citrus production, which we denote
by Πi,p . Profits depend on fixed costs, which in our model represent a sort of barrier to entry.
As a result, the smaller the number of producers in the industry (n), the larger the profits
made, which in turn increases mafia activity. Even though we do not have data on profits,
we can consider the dominant position in the citrus market (73 percent of the total Italian
production and almost 78 percent of the total US lemon imports) as the result of a fixed
cost which prevented the entry of new competitors in the market. This dominant position
generated large profits for peasants and thus we expect the probability of mafia to increase
with the production of citrus.
Zi,p represents a set of control variables that may also affect the probability of mafia.
This set of variables includes controls for the degree of trust citizens have in the law and for
the peripherality of the town. These measures do not perfectly capture the enforcement of
property rights by the incumbent state, but should provide an idea of the efficiency of the
state in enforcing property rights. Finally, ηp represents provincial fixed effects that may be
correlated with the error term µi,p .
4.2
Data
Data both on town and district level (mandamento)24 for the entire island are collected from
the original Damiani Inquiry which represents one of the earliest and most important primary
sources concerning the economic and social conditions of Sicily in the 1880s.
25
Damiani’s
investigation is part of a larger inquiry, approved in March 1877 and proposed by the Member
of Parliament Stefano Jacini, aimed at assessing the conditions of the agricultural sector
and peasantry in every region of Italy. Abele Damiani was the MP in charge of surveying
Sicily. Our sample considers all seven provinces (Caltanissetta, Catania, Girgenti, Messina,
Palermo, Syracuse, and Trapani) providing a total of 143 observations.26
The section of the Inquiry that matters for our analysis is divided into two parts. The
first one discusses the situation of agriculture with particular reference to tax burden, wages,
the kind of crops produced, and the relations between peasants and landlords (i.e., tenancy
contract, fractionalization of land, etc.). Questionnaires relative to this first part were sent
out to almost 357 mayors, but less than half of them provided complete information. For
unknown reasons, the folder for this section on the province of Caltanissetta is not available
in the Archive of State in Rome. In order to get data for the agricultural conditions for this
province we use the information available from the Summary Report that Damiani presented
to the parliament, which has also been used by Bandiera (2003) and Buonanno et al. (2012).
The second part of the Inquiry provides information on the moral and social conditions
24
A mandamento is a judicial district of competence of the pretore.
The original handwritten copy of the Damiani Inquiry is still available from the Archive of State in Rome,
even though the conditions of manuscripts are far from perfect and some pages are very hard to read (see
figures in the Appendix).
26
Syracuse became a province in 1865 replacing Noto, which was a province at the time of the unification in
1861. Except for mafia activity, all the information regarding Caltanissetta comes from the Summary Report
that Abele Damiani presented to the parliament.
25
15
of peasants. In this case questionnaires were sent to 179 pretori (lower court judges).27 The
information provided gives us a unique picture of the moral and social conditions of Sicilian
peasants at the time. Questions related to this part of the survey regard the lewdness and
religiousness of people, corruption of the clergy, the rule of law, and the effect of introducing
a compulsory military service. However, the section mostly relevant to our analysis is the
one regarding the form and level of crime on the island. The question asked to the pretori
in the Inquiry is: “What is the most common form of crime in the district? What are their
causes?” There is a range of possible crimes that the pretori considered. Some of them relate
to rustling, robbery, murders, and of course the mafia. Our dependent variable is labeled
mafia and represents the criminal organizations’ activity.
28
Bandiera (2003) and Buonanno et al. (2012), on the other hand, use a different dependent
variable: an ordinal variable for the intensity of mafia collected from the Summary Report
that Abele Damiani wrote to the Italian parliament on the basis of the original Inquiry.
However, from the original Inquiry it emerges that very few pretori reported the intensity of
mafia (10 out of 143). Since the source of this additional information is unclear, we prefer
to use the information we have available from pretori in the original Inquiry for which we
are certain about the source.29
There are potential concerns with the data on mafia presence. Firstly, could the mafia
still be present in a district even though the pretore did not list it as the most common form
of crime? Because of the structure of the survey in the Inquiry, it is indeed possible that
some districts had mafia activity even though the pretore did not report it as being a major
crime. This problem may slightly affect our results.
Second, were pretori themselves mafiosi and hence likely to understate the presence of
mafia? The answer to this question is most likely no, although there is no conclusive evidence
in either direction. Pretori were directly appointed by the Minister of Justice with a Regio
decreto (royal order). Their appointment, and any other aspect concerning their career, was
subject to an evaluation by a local committee of experts of the local Court of Appeal. For
the first 10 years of their career, pretori changed districts (mandamento) very frequently,
which may have restricted their connection with the local environment. Another question
in the Inquiry reveals that many experienced difficulties administering justice without the
cooperation of local people since in several trials witnesses did not dare to say the truth due
to mafia threats or collusion with the mafiosi.30
Third, did the pretori have a common understanding of what the term mafia implied?
This indeed appears to have been the case. In the 1880s, the term mafia already meant a
criminal organization. The term was used to identify this sort of organized crime in Sicily at
27
There are much fewer pretori than mayors since the office of pretura is only present in larger provinces
so that one pretore often serves several towns.
28
Samples of these surveys are provided in the Appendix. Figure A1 shows the table summarizing information provided by pretori. Figure A2 shows a sample of the survey completed by the mayor of Cefalu.
29
Pretori were lower court magistrates. Because of their role, their information on criminal activity can be
considered the most indicative.
30
According to Pezzino (1990), the pretore in Bagheria said “There is a tendency to deny the truth. Not
only people does not answer truthfully, but they deliberately lie either because of mafia or because of money
or because they are scared.”
16
least since 1863, when a comedy titled “I Mafiusi di la Vicaria” was shown in Palermo. In
1865, the prefect of Palermo (Filippo Gualterio) used the term mafia in a private document
to identify the criminal organization. In addition, since 1871 mafia membership has been a
public law offence. We therefore regard it as highly unlikely that the term was misinterpreted.
These problems are common to most empirical analyses based on survey data, and cause
a measurement error that is part of the error term. However, as long as the error term is not
correlated with independent variables, there is no reason to believe that these problems will
affect estimates. This is important for our analysis given that the possible misinterpretation
of the question as well as the collusion between pretori and mafia (if any) are likely to be
distributed randomly across towns with mafia. Hence, it seems reasonable to assume that
the independent variable that we base our analysis on (whether the city produces citrus)
should not be correlated with such an error and should thus provide unbiased estimates.31
Table 5 reports descriptive statistics for our variables. The dependent variable, mafia, is
a binary dummy for whether the pretore of the town reckons mafia as the most important
source of crime in the district.
32
Figure 2 shows the local distribution of mafia in our sample. On average, 36 percent
of all towns were strongly affected by mafia, which means that almost 51 out of the 143
towns in our sample had mafia listed as the most common form of crime. Girgenti is the
province with the highest incidence of mafia, with almost a strong mafia presence in 14 out
of 17 towns. In Trapani, the mafia operates in 6 out of 15 towns, and in the Caltanissetta
province in 7 districts out of 16. Almost one third of the districts in Palermo (mainly those
in the Conca d’Oro) and Catania provinces are mafia infested. Messina and Syracuse are
the provinces with the lowest incidence. These summary statistics are consistent with the
description in Colajanni (1885), where Sicily is divided into three macro-regions, considering
Girgenti as the one with the highest rate of murders and convictions and therefore the one
with the highest incidence of mafia.33
Table 5: Table 5: Distribution of mafia and agricultural production across
provinces in 1881-86
Figure 2: Mafia and non-mafia towns in Sicily in the 1880s
To control for some of the most important sources of mafia presence discussed in the
previous section, we include three sets of independent variables. Consistent with our model,
Colajanni (1885), Dickie (2004), and Lupo (2009) all identify the profitable production of
31
Assume that the dependent variable is measured with error and that yi = yi + ψi . Then we can write
the composed error term in (??) as i = µi + ψi . If ψi is random, it should be uncorrelated with independent
variables.
32
Data on mafia are available for 162 districts, but when merged with independent variables the largest
sample covers 143 districts.
33
Using information available in Damiani (1886), Colajanni identifies three macro-regions on the basis of
economic and social conditions. The first region includes the province of Catania and Messina, where there
were good economic and social conditions. The situation was slightly worse in the provinces of Syracuse,
Trapani, Caltanissetta, and Palermo. The third region includes Girgenti, where both economic and social
conditions were very poor and there was a high level of crime.
17
goods, like citrus and sulphur, as important determinants of mafia presence. For this reason,
the first set of independent variables in Table 5 are related to agricultural production. Citrus,
Wheat, Olives, Grape, and Sulphur are the commodities we consider. In order to assess what
dominant crops were produced in each town, we use dummy variables coded 1 if the town is
a dominant producer of these crops and 0 otherwise. The relevant question we refer to in the
Damiani Inquiry is: “Which is the dominant crop produced in the city?” Mayors normally
listed more than one crop (for some cities they also reported quantities), and hence, the
dummies for crops are not mutually exclusive.
More information on actual quantities produced is provided from the Damiani’s Summary
Report. However, after comparing data in the Summary Report with data in the Inquiry, we
found that the Summary Report does not always match the original information provided
by mayors. For example, for Lercara Friddi, the Inquiry reports that grapes and wheat are
the predominant crops and both crops extend for almost 145 hectares. Yet in the Report
(Volume XIII Tomo II), the predominant crops are grape and wheat (and also Indian figs),
each extending for 2,000 and 220 hectares respectively, which is not consistent with data
provided by mayors. The same problem applies for Castelbuono. The extension of grapevine
and olive trees in the Inquiry is 142 and 193.27 hectares, respectively, while the Summary
Report records 490 and 640 hectares, respectively. Given that the Inquiry does not always
report data on hectares per crop, we prefer using a dummy variable (only recording whether
the crop is predominant or not) in order to minimize the potential measurement error of the
data available from the Summary Report.
Buonanno et al. (2012), on the other hand, use a measure of crop suitability to proxy
citrus, arguing that it is preferred to actual data on crop production because of possible
reverse causality. Even though this is generally true there are at least two reasons why in
this case a measure of actual production is preferable (given that data are available). First,
the risk of reverse causality between mafia presence and citrus production in this occasion
is minimal, given that production of citrus in Sicily responded to an external shock in the
demand related to the use of the fruit as a treatment for scurvy. According to all available
accounts, the boom in the citrus sector preceded the emergence of the mafia.
The second reason why a measure of actual production is preferable relates to the huge
revenues in the sector inducing producers to undertake large investments in water irrigation
and similar technologies (for example, the norie previously mentioned) that increased the
efficiency of production. These investments made producers vulnerable to violence, predation
and therefore mafia retaliation.
Since data on citrus production are of particular relevance to test our hypothesis, we
compare our data with the ones available in Di Vita (1906).34 We can confirm that the
original Damiani Inquiry provides an accurate distribution of local citrus production.35
Data on sulphur mines are also provided by Di Vita (1906).36 As argued by Colajanni
34
Di Vita (1906) wrote a statistical bulletin reporting summary statistics for all Sicilian cities.
In a couple of cases the Inquiry reports that the lemons industry is relevant in the city but then does not
list lemons among the relevant crops. These cases are compared with Di Vita (1906) and coded according to
this latter source.
36
See Figure A4 in the Appendix for a more recent distribution of crops and fruit trees in Sicily.
35
18
(1885), sulphur mines are almost exclusively concentrated in the province of Girgenti (12 out
of 17 towns). Outside Girgenti, there are 5 mines in the province of Catania, 3 in Palermo,
and 1 in the province of Messina and Trapani. Wheat is produced in the entire province
of Girgenti and the production is the lowest in the province of Messina. Grapes and olives
are almost equally distributed across the island. Our summary statistics seem to match the
picture provided in Colajanni (1885) quite well.
The second set of explanatory variables is related to the political status of towns and to
other policies aimed at increasing the small-scale ownership of land (Table 6). Since the 13th
century, Sicilian towns have had three different types of political organization. The feudal
system was the first type of political organization, where a small minority (elite) owned the
land and a large peasant population passively accepted the subservient role. The second
form of political organization was labeled ecclesiale (church-ruled cities), where bishops used
to act as typical lords. Finally, there were the crown-ruled towns (demaniali ). Demaniali
towns were independent of local lords and bishops and were self-regulated. Using data from
Di Vita (1906), we find that feudal cities represent almost 62 percent of our sample, i.e., 79
out of the 127 had a feudal heritage. Crown-ruled (demaniali ) cities account for one fourth
of the sample (32 cities), and 10 cities were instead ecclesiali cities (church ruled).
The last column of the table shows the effectiveness of fractionalization policies implemented 1812-1870. We consider three types of policies: i) the abolition of feudalism and
the auction/allocation of land to small holders; ii) the enfiteusi, a perpetual lease that allowed farmers to use the land as if they were owners; and iii) the seizure of church ruled
terrains after the unification of Italy and the consequent land auction. These policies were
more effective in the provinces of Caltanissetta, Girgenti, and Catania and less effective in
the provinces of Palermo and Messina. However the distribution of land in the latter two
provinces has always been highly fractionalized and hence, there was little room for such
policies. In contrast, in Caltanissetta, Catania and Girgenti the latifund was much more
common and hence policies were more successful, even though in relative terms, land was
still more fractionalized in Palermo and Messina than in Girgenti and Catania. This was
mainly due to the fact that the policy increased fractionalization but not among peasants:
rich landowners were the only ones who could afford to bid for land in auctions and therefore
the policy had little effect in increasing the private ownership among peasants.
Table 6: Descriptive statistics for political organization across provinces before
the Italian unification in 1860
Table 7 provides descriptive statistics for the scale of plantations and the fractionalization
of land.37 The question asked for the scale of plantation and fractionalization is: “What is
the dominant scale of the plantation? And what is the fractionalization of land?” Most of the
time, mayors answered that a large, a medium, and a small scale are dominant, therefore the
sum of the three variables is larger than 1. Regarding land distribution, we note that there is
37
The Damiani Inquiry (1886) is the source of these data, except for Caltanissetta whose data are from
Damiani’s Summary Report.
19
high fractionalization in almost 50 towns whereas fractionalization is relatively low in almost
45 towns. The scale of plantation tends to be relatively high in Palermo, whereas in the other
provinces the percentage of high scale plantations tends to be around 33 percent. Small-scale
plantations are relatively frequent in all provinces and particularly so in Trapani, Messina,
and Catania. Girgenti and Caltanissetta are the provinces with the lowest frequency of
small-scale plantations.
Table 7: Land fractionalization and scale of plantations by province in 1881-86
Finally, in Table 8 shows the pairwise correlation among selected variables. Mafia seems
to be positively correlated with Citrus (0.39 correlation) and Large-scale plantation (0.25
correlation) as well as with the effectiveness of the fractionalization policy (0.26 correlation).
The presence of sulphur mines is also positively correlated with Mafia (0.19 correlation).
A feudal origin and the high fractionalization of land, which also have been considered
important for the diffusion of organized crime, are weakly correlated with Mafia (0.05 and
0.04 correlation, respectively). Finally, Population density, capturing the wealth of a locality,
is weakly correlated with Mafia. All these variables are weakly cross-correlated, preventing
problems of multicollinearity.
Table 8: Pairwise Correlations
5
Empirical Analysis
5.1
Mafia presence in the 1880s
Table 9 presents probit estimates with mafia presence in the 1880s as the dependent variable.
We start by estimating a simple model where mafia presence in the 1880s depends only on
variables capturing the economic activity in the town. We then proceed by introducing
additional variables to control for observables. In Column 1, the diffusion of the mafia
significantly depends on production of citrus. At the mean, the production of citrus increases
the probability of mafia by 45 percent. In Column 2, we re-estimate the same model but now
control for province dummies to capture regional fixed effects. The previous results still hold.
In Column 3 we drop the observations for the province of Caltanisetta in order to detect any
potential bias due to a different source of information. The estimated effects in Column 3
are almost unchanged. Coefficients and z-statistics suffer from negligible variation.
In Column 4, we change specification by dropping the variables that are not significant
(to prevent an excessive reduction of the degrees of freedom) and we enter two additional
controls. We use a Fractionalization policy dummy and the variable Population density
to account for policies that may have affected the private ownership of land and wealth.
Population density is not significant, whereas the Fractionalization policy dummy has a
strong significant effect on the probability of mafia presence. In Column 5, we include the
dummy for the high fractionalization of land to see whether the fractionalization policy
dummy truly captures the effect of increasing the fractionalization of land (as argued by
20
Bandiera, 2003) or something else. As argued before, fractionalization policies had little
effect in increasing private ownership among peasants. Hence, the Fractionalization policy
dummy could capture an increase in the use of the gabella because existing landowners
bought most of the available land.38 Therefore, even though mayors consider these policies
effective in increasing land fractionalization, land still remained quite concentrated in the
hands of a few, thus preventing peasants from acquiring rights. In line with our hypothesis,
the dummy variable High land fractionalization turns out to be non-significant.
Because of the fixed costs that farmers had to sustain in order to expand citrus production
(and also other crops that could generate higher profits such as grapes), we expect profits
to be higher in towns with a relatively low land fractionalization. To test this hypothesis,
Column 6 includes the dummy variable Large-scale plantation. The idea is that investments
for expanding the sector were more likely made in towns with a large-scale plantation (because of the decreasing costs), making producers more vulnerable to a potential loss due to
extortion. The variable turns out to be highly significant, increasing the probability of mafia
presence by 39 percent. The set of covariates specified in Column 6 represents our preferred
specification. In this model, the presence of mafia is significantly determined by the citrus
production, the effect of policies for private ownership, and by the scale of the plantation.
Table 9: Mafia Probit Model
In Table 10, we proceed by testing the robustness of our preferred specification. We start
by including in Column 1 control variables to account for the distance from Palermo and
from Mazzara del Vallo. The first variable should capture the distance from a well-known
mafia-base center and also a key port, and the variable Distance from Mazzara del Vallo is
used to capture the possible diffusion of citrus. In the 10th century, lemon plants were in
fact originally introduced by the Arabs entering Sicily from Mazzara del Vallo. The former
variable turns out to be non-significant, whereas the latter is marginally significant, i.e., the
presence of mafia decreases if the distance from Mazzara del Vallo increases.
In Column 2 we introduce the dummy variable Feudal to test whether this political system, often associated with patronage and kinship relations, had any impact on the presence
of mafia. The variable turns out to be non-significant.
In Columns 3 and 4, we control for the level of law enforcement. We use the distance
from the nearest railway station (collected from Di Vita, 1906) as a measure of peripherality
and three dummies for whether citizens trust, mistrust, or do not care about the law (the
excluded group is whether they fear the law). These dummies are coded using information
available in the Damiani Inquiry (1886). The results show that being close to the railway
station has a negative effect on mafia presence (consistent with the peripherality hypothesis).
Towns where citizens do not care about the law have almost 60 percent higher probability
of mafia presence.
38
Gabella is synonymous of tax used in the Medieval era.
21
Finally, in Column 5 we control for the length of the tenancy contract which also turns
out to have a significant and negative effect on the probability of mafia presence. The Citrus
dummy still increases the probability by 57 percent and the Fractionalization policies dummy
increases the probability by an average of 23 percent.
Table 10: Mafia Probit Model with additional controls
Given the potential downward bias due to the entry barriers that mafia presence could
have represented for possible entrants, we decide to compare our estimates with a consistent
estimator like the IV model where Citrus is instrumented using data on the altitude above
sea level. This instrument is chosen because of the particular climatic conditions required
by this plant and their influence on fixed costs, as outlined in the theoretical model. The
causality link can be represented as follows:
Altitude → Citrus→ (Profits) → Mafia
where Altitude affects the climate and both its daily and annual variation. The high
variation in climate and the high probability of frost at high altitudes represent an important
fixed cost for the production of lemons since, as we discussed earlier, lemon plants are totally
intolerant to frost. Therefore, lemon production will occur in towns characterized by a mild
climate with low seasonal variation. High profits will be generated by this activity and more
protection against potential losses will be required from local producers, hence affecting the
level of mafia activity. The choice of the variable Altitude as an instrument is also consistent
with the categorization of the fruit industry suggested by Monroe (1909) who argues that
the fruit industry in Sicily may be divided into three zones: “The marine, or lemon belt,
from sea-level to fifteen hundred feet. The middle, or orange zone, from fifteen hundred to
three thousand feet. The forest belt above three thousand feet” (Monroe, 1909, pg. 190-91).
Therefore, we expect to have a larger concentration of lemons below 1,500 feet.
In Table 11, we check whether the variable Altitude can be excluded when entered together with our proxy for citrus. Even though an omitted variable bias might still be an
issue, the results support the potential exclusion of the variable Altitude, which turns out to
be significant in Column 2. When entered together with Citrus, Altitude loses significance
as a result of multicollinearity. However, the effect of Citrus dominates that of Altitude
(p-values 0.103 against 0.311 for Altitude).
Table 11: Exclusion of Altitude
In Table 12, we start with a Linear Probability Model to evaluate the potential bias with
respect to OLS estimates. In Column 1, all variables in the baseline model are statistically
significant at least at the 5 percent level. Citrus increases the probability of mafia presence by almost 32 percent, Fractionalization policies by almost 25 percent, and Large-scale
plantations by almost 23 percent.
22
In Column 2, we show results from the IV estimator to control for endogeneity issues.
The results provide us with a picture quite similar to the OLS estimates, except for the
effect of Citrus on mafia activity, which increases by almost 13 percent (compared with the
OLS estimates).39 The scale of plantation is still significant at the 5 percent level or lower.
In Column 3, we estimate our model by IV-Probit and the results still hold. Finally, in
Column 4, we use a Spatial Linear Probability Model to control for spatial autocorrelation
in the error term. The idea is that mafia presence in one town could positively influence the
probability of mafia presence in a neighboring town as a local spillover effect. In this model,
the dummy Citrus significantly increases the probability of mafia presence by 32 percent,
the scale of the plantation by 21 percent, and the fractionalization policies dummy by 23
percent.
Table 12: Robustness Check – Alternative Estimators
5.2
Mafia persistence
Over the 140 years since its origin, the presence of mafia has spread all over Sicily. Repeated
clashes between clans have further affected the territorial distribution. However, the districts
where mafia originally developed still remain hot spots for mafia activity. Table 13 shows
data, at province level, on mafia murders reported to the police in 1999 according to the
416 bis article of the Italian penal law and data on the number of asset confiscations from
mafiosi by the Italian government during 1997-2008.40 Except for the province of Palermo,
where numbers are inflated by the fact that the city of Palermo itself is the most relevant hub
for the Sicilian mafia’s activity, it is important to note that Trapani and Agrigento (former
Girgenti) are the two provinces with the highest number of per capita assets confiscated.
They also used to be the two provinces with the largest presence of mafia according to the
Damiani Inquiry (1886).
Table 13: Mafia persistence, 1880-2008
In order to get a better idea of this persistence, we estimate the impact of our dummy
variable for mafia presence in the 1880s from Damiani (1886) on the per capita number of
assets confiscated from the mafia at town level. As Table 13 shows, the presence of mafia
in the 1880s explains a significant portion of the variation in the dependent variable. Towns
with mafia in the 1880s have a significantly higher per capita number of assets confiscated
in the 1997-2008 period (the marginal effect is almost 0.38 as seen in Column 1). After
controlling for a full set of province dummies (Column 2), the result still holds. In our view,
this suggests a strong persistence of mafia presence throughout centuries.
Table 14: Historical mafia presence as a determinant of current mafia activity
39
For completeness, in Table A2 in the Appendix we also provide estimates for the sample of provinces
excluding Caltanissetta. The results still hold.
40
The 416 bis article prescribes 3-6 years of prison for members of the mafia criminal organization.
23
6
Conclusions
The issue of how socially inefficient institutions arise and persist is currently a very active
research topic in economics. In this paper, we have developed a market structure hypothesis
for understanding the origins of the Sicilian mafia. Unlike existing works that emphasize
political and historical factors, our analysis studies the importance of a boom in international
demand for citrus fruits, combined with the presence of fixed costs in lemon production, as a
source of market imperfections and very high profits in certain localities. This profitability,
combined with a general lack of rule of law prevailing at the time, provided an ideal breeding
ground for a mafia thriving on providing private protection to lemon producers.
In the empirical analysis, using data from a parliamentary inquiry from the 1880s, we
show that the presence of mafia is strongly related to the production of citrus fruits. The
results continue to hold when we include several control variables and use alternative estimators. The historical presence of mafia is further a strong predictor of current mafia
activity.
We believe that our results contribute to shedding light on the emergence of one of the
most harmful economic institutions in Europe during the modern era. Additional studies
on country or regional level along the same lines could potentially uncover the micro level
mechanisms for how and why other inefficient institutions arise and often persist for extended
periods.
24
References
[1] Acemoglu, D. and J. Robinson, 2012. Why Nations Fail: The Origins of Power, Prosperity, and
Poverty. New York: Crown Publishing Group.
[2] Acemoglu, D., S. Johnson, and J. Robinson, 2006. ”Institutions as a Fundamental Cause of
Long-run Growth” chap 6 in Handbook of Economic Growth, Vol 1, Part A, 385-472.
[3] Acemoglu, D., T. Verdier, and J. Robinson, 2004. ”Kleptocracy and Divide-and Rule: A Model
of Personal Rule” Journal of the European Economic Association 22(2-3), 162-192.
[4] Agenzia del Demanio, 2008. Beni Confiscati Alle Mafie, Minister of Home Affairs Report.
[5] Alongi, G. 1886. La maffia nei suoi fattori e nelle sue manifestazioni: studio sulle classi pericolose
della Sicilia, Bocca Editori, Torino.
[6] Anderson A., 1995. Organized Crime, Mafias and Governments, in Fiorentini, Gianluca and
Peltzman, Sam (eds), The Economics of Organized Crime, Cambridge, Cambridge University
Press.
[7] Arlacchi Pino, 1986. Mafia Business: The Mafia Ethic and the Spirit of Capitalism, London:
Verso.
[8] Bandiera, Oriana, 2003. Land Reform, the Market for Protection, and the Origins of the Sicilian
Mafia: Theory and Evidence. Journal of Law, Economics, and Organization , 19(1), 218-44.
[9] Baron, Jeremy H. (2009). Sailors’ Scurvy Before and After James Lind - A reassessment. Nutrition Reviews: 67(6):315-332.
[10] Battaglia, Rosario, 2003. L’ultimo Splendore: Messina tra Rilancio e Decadenza, Rubbettino
Editore.
[11] Bonfadini, Romualdo, 1876. Relazione della Commissione Parlamentare, Inchiesta sulle Condizioni Sociali ed Economiche della Sicilia, Roma.
[12] Buonanno, Paolo, Ruben Durante, Giovanni Prarolo, Paolo Vanin, 2012. ”On the Historical and
Geographical Origins of the Sicilian Mafia”, Munich Personal RePEc Archive.
[13] Catanzaro, Raimondo, 1992. Men of Respect: A Social History of the Sicilian Mafia, New York:
The Free Press.
[14] Colajanni, Napoleone, 1885. La Delinquenza della Sicilia e le sue Cause, Giornale di Sicilia.
[15] Colajanni, Napoleone, 1895. Gli Avvenimenti di Sicilia e le Loro Cause, Remo Sandrom,
Palermo.
[16] Colajanni, Napoleone, 1900. Nel regno della mafia, Remo Sandrom, Palermo.
[17] Congdon Fors, H. and O. Olsson, 2007. Endogenous Institutional Change After Independence,
European Economic Review 51(8), 1896-1921.
[18] Cutrera, Antonio, 1900. La Mafia e i Mafiosi, Paermo, A Reber.
[19] Damiani, Abele 1886. Atti della Giunta per l’Inchiesta Agraria e sulle Condizioni della classe
Agricola, Forzani Tip. Del Senato tomo 2 fasc 4 (Archivio Centrale Roma).
25
[20] Di Vita, Giuseppe, 1906. Dizionario Geografico dei Comuni della Sicilia e delle Frazioni Comunali: Con Notizie Storiche, F. Pravatà.
[21] Dickie, John, 2004. Cosa Nostra: A History of the Sicilian Mafia, Palgrave Macmillan.
[22] Finley M.I., Mack Smith, Denis, Duggan, Christopher; 1987. A History of Sicily. Elisabeth Sifton
Books, Viking Penguin INC.
[23] Fiorentini, Gianluca; 1999. Organized Crime and Illegal Marteks. Dipartimento di Scienze
Politiche. Universita’ di Bologna’ Wp 8400.
[24] Fukuyama, F., 2000. Social Capital and Civil Society, IMF working paper WP/00/74.
[25] Gambetta, D. and Reuters, P. 1995. Conspiracy among the many the mafia in legitimate industries in Fiorentini and and Peltzman, Sam (eds), The Economics of Organized Crime,
Cambridge, Cambridge University Press, pp.116-136.
[26] Gambetta, Diego, 1996. The Sicilian Mafia: the Business of Private Protection, Harvard University Press.
[27] Gambetta Diego, 2009. Codes of the Underworld: How Criminals Communicate. Princeton
University Press.
[28] Grossman, Herschel I., 1995. Rival Kleptocracies: The Mafia versus the State, in Fiorentini,
Gianluca and Peltzman, Sam (eds), The Economics of Organized Crime, Cambridge, Cambridge University Press.
[29] Hardin, Russell, 1999. Social Capital. In Alt, James E., Margaret Levi, Elinor Ostrom: Competition and Cooperation. New York: Russell Sage Foundation, 170-189.
[30] Hess, Henner, 1973. Mafia and Mafiosi: The Structure of Power, Farnborough: Saxon House.
[31] Italian National Institute of Statistics - ISTAT, 1999. Delitti Denunciati dalle Forze di Polizia
all’Autorita’ Giudiziaria.
[32] Levi, M., 1996. Social and Unsocial Capital: a Review Essay of Robert Putnam’s Making
Democracy Work. Politics and Society, 24: 45-55.
[33] Lupo, Salvatore, 2009. History of Mafia, Columbia University Press.
[34] Mehlum, Halvor, K-O Moene, and R. Torvik, 2006. ”Institutions and the Resource Curse”.
Economic Journal, 116(508), 1-20.
[35] Monroe, Seymour Will, 1909. The Spell of Sicily: The Garden of the Meditarranean, Boston,
The Page Company.
[36] Montanelli, Indro, 1949. Interview with Don Calo Vizzini. Corriere della Sera, October 30.
[37] Newton, Kenneth, 2001. Trust, Social Capital, Civil Society, and Democracy, International
Political Science Review April 2001 vol. 22 no. 2 201-2.
[38] Pescosolido Guido, 2010. L’Economia Siciliana nell’Unificazione Italiana, paper presented in the
Convegno Nazionale di Studi: La Partecipazione della Sicilia al Movimento dell’Unificazione
Nazionale.
26
[39] Pezzino Paolo, 1990. Una certa reciprocita’ di favori. Mafia e modernizzazione violenta nella
Sicilia postunitaria. FrancoAngeli.
[40] Powell, Harold G., 1909. Italian Lemons and Their By-Products, Bulletin No. 160, Bureau of
Plant Industry, US Department of Agriculture.
[41] Putnam R., 1993. Making Democracy Work: Civic Tradition in Modern Italy. Princeton University Press. Princeton.
[42] Reuter, P. (1987). Racketeering in Legitimate Industries: A Study in the Economics of Intimidation, Santa Monica, CA: The Rand Corporation.
[43] Sala-i-Martin, X. and A. Subramanian (2003) ”Addressing the Natural Resource Curse: An
Illustration from Nigeria” NBER Working Papers, No 9804, NBER.
[44] Sonnino, Sidney and Leopoldo Franchetti, 1877. La Sicilia nel 1876, Firenze, Tip. di G. Barbera.
[45] Skaperdas, S., 2001. The Political Economy of Organized Crime: Providing Protection When
the State Does Not, Economic of Governance 2: 173-202.
[46] Skaperdas, S. and Syropoulos, C. 1995, ”Gangs as Primitive States”, in Fiorentini, Gianluca and
Peltzman, Sam (eds), The Economics of Organized Crime, Cambridge, Cambridge University
Press.
[47] Travelyan, Raleigh, 2001. Principi Sotto il Vulcano. Storia e Leggenda di una Dinastia id Gattopardi Anglosiciliani dai Borboni a Mussolini, Milano BUR.
[48] van der Ploeg, Frederick, 2011. ”Natural Resources: Curse of Blessing?” Journal of Economic
Literature 49(2), 366-420.
[49] Villari, Pasquale, 1875. Lettere Meridionali, L’Opinione.
27
Table 1: Distribution of lemon trees in Southern Italy in 1898
Region
Number of trees
Reggio Calabria
1,232,675
Messina
1,634,231
Palermo
2,488,475
Catania
828,640
Syracuse
460,125
Calatanissetta
8,210
Girgenti (Agrigento)
56,379
Trapani
216,610
Total
6,924,985
Source: Powell (1909)
Table 2: Total Italian exports of lemon and total United States imports
Year
Total Italian exports
Value (US $)
Exports to
the US (%)*
3,419,486
41.3
1898
Quantity
(pounds)
325,504,061
1899
359,473,041
3,234,489
36.7
1900
331,563,577
3,000,286
1901
368,801,294
1902
1903
Quantity
(pounds)
133,374,95
Total US imports
Value (US $)
2,521,985
Imports from
Italy (%)**
100^
298,634,448
4,399,160
44.1
29
159,384,389
3,655,926
60.3
3,328,610
29.2
148,334,112
3,516,877
72.5
490,033,260
3,432,677
35.3
162,962,091
3,318,909
100^
459,622,020
3,218,948
31.2
152,775,867
3,087,244
93.8
*Percentages provided by Powell (1909)
** Percentages estimated using percentages on quantity exported from Italy above. For example for the year 1900
the quantity exported to the US is 331,563,577*0.29=96,153,473 which divided by 159,384,389 provides a
percentage equal to 60.32 percent.
^ In 1898 and 1902 the percentage exported from Italy to the US exceeds the total import into the US. This is
because Italian figures refer to the calendar year, while USA figures refer to the fiscal year.
Source: Powell (1909)
Table 3: Total Italian export of citrate of lime and total United States imports
Year
Total Italian exports
Value (US $)
Exports to
the US (%)*
151,295
32.8
1899
Quantity
(pounds)
3,142,248
1900
3,743,448
196,826
37.6
1901
3,120,202
147,502
1902
7,517,541
1903
7,229,647
Quantity
(pounds)
1,577,804
Total US imports
Value (US $)
157,482
Imports from
Italy (%)**
65.3
1,944,803
204,243
72.4
35.3
2,416,658
209,583
45.6
329,965
32.5
3,056,904
293,293
79.9
632,905
38
2,657,110
240,486
100^
*Percentages provided by Powell (1909)
** Percentages estimated using percentages on quantity exported from Italy above. For example for the year 1901
the quantity exported to the US is 3,120,202*0.353=1,101,431 which divided by 2,416,658 provides a percentage
equal to 45.57 percent.
^ In 1903 the percentage exported from Italy to the US exceeds the total import into the US. This is because Italian
figures refer to the calendar year, while USA figures refer to the fiscal year
Source: Powell (1909)
Table 4: Exports from the harbor of Messina in 1850
Product
Quantity
Units
1,100
Balle
4,469,850
Oz 645 x Balla
Silk
Lire
Description
Olive Oil
320,000
Cafissi
2,419,200
Approx 3 Litres x Cafisso
Oranges
500,000
Casse
2,726,325
Approx 240 oranges x cassa.
Lemons
600,000
Casse
3,779,622
Approx 360 lemons x cassa
1,000
Barili
503,986
Oz 40 per barile
Oz 6 per barile
Lemon Juices
Salted Lemons
200
Barili
151,200
400,000
Libre
2,014,740
Sulphur
90,000
Quintali
Wheat
50,000
Salme
2,477,790
Flax
20,000
Salme
1,007,937
Wine
2,000
Salme
37,800
Nuts
4,000
Salme
655,169
4 salma = 3 hl
20,000
Cantaia
1,763,370
Oz 7 x cantaio
Citrus Parfumes
Almond
Pistacchio
302,211
4 salma = 2 hl
1/2 salma= 801 hl
200
Cantaia
30,240
Oz 12 x Cantaio
Walnuts
2,000
Salme
50,387
4 salma = 3 hl
Liquorice
16,000
Cantaia
680,400
Oz 9 per cantaio
Sardines
4,000
Barili
151,162
Oz 2 per Barile
Carob
4,000
Sacchi
90,720
24 sacchi = 90Kg
Wool
2,000
Cantaia
453,600
6 cantaio = 80Kg
Linen
7,000
Quintali
264,600
Oz 3 x quintali
Cotton
4,000
Quintali
30,240
Source: Battaglia (2003)
Table 5: Distribution of mafia and agricultural production across provinces in 1881-86
Dominant production of:
Province
Towns
Mafia
Citrus
Grape
Olive
Wheat
Sulphur
0.471
1
0.471
1
0.526
0.478
0.826
0.347
0.782
0.226
0.625
Caltanissetta
16
Catania
22
0.437
0.318
Girgenti
17
0.823
0.5
0.611
0.388
1
Messina
25
0.24
0.608
0.739
0.521
0.478
0.07
Palermo
27
0.296
0.37
0.777
0.444
0.74
0.133
Siracusa
21
0.142
0.35
0.85
0.35
0.75
0
Trapani
15
0.4
0.4
0.8
0.6
0.8
0.066
Total
143
0.357
0.454
0.797
0.441
0.776
0.23
Numbers in the table refer to the share of towns within each province with mafia presence and/or with dominant
production of each commodity. Each variable in the table is a binary dummy. Mafia=1 if mafia is perceived to be
the most common form of crime in the town and 0 otherwise, as explained in the text. Citrus, Grape, Olive, Wheat
and Sulphur are also binary dummies taking on the value of 1 if the commodity is listed by the pretore as one of the
key agricultural goods produced in the town, as explained in the text.
Source: Damiani (1886)
Table 6: Descriptive statistics for political organization across provinces before the Italian unification in
1860
Political organization:
Province
Towns
Feudal
Crown-ruled
Church-ruled
Frac. policies
Caltanissetta
12
0.833
0.167
0
0.714
Catania
31
0.452
0.29
0
0.625
Girgenti
24
0.583
0.125
0
0.714
Messina
28
0.357
0.28
0.214
0.285
Palermo
30
0.566
0.267
0.133
0.346
Siracusa
22
0.818
0.181
0
0.5
Trapani
15
0.666
0.333
0
0.285
Total
162
0.574
0.216
0.0617
0.413
Numbers in the table refer to the share of towns within each province that were characterized by feudal
organization, was crown- or church-ruled, or had a substantial degree of fractionalization polices. Each of the
variables of Political organization is a binary dummy.
Source: Damiani (1886)
Table 7: Land fractionalization and scale of plantations by province in 1881-86
Fractionalization
of ownership
Province
Scale of plantation
Towns
High
Low
Large
Small
Caltanissetta
12
0.545
0.454
0.333
0.5
Catania
23
0.409
0.227
0.26
0.695
Girgenti
14
0.461
0.461
0.357
0.461
Messina
20
0.277
0.444
0.35
0.789
Palermo
26
0.52
0.16
0.461
0.576
Siracusa
20
0.35
0.5
0.35
0.666
Trapani
14
0.333
0.333
0.285
0.923
Total
129
0.413
0.347
0.348
0.661
Numbers in the table refer to the share of towns within each province that were characterized by high or low
fractionalization in land ownership and large or small plantations.
Source: Damiani (1886)
Table 8: Pairwise correlations
Mafia
Citrus
Wheat
Olive
Grape
Sulphur
Feudal
Frac.
policy
Pop.
density
High
frac.
Mafia
1
Citrus
0.3863
1
Wheat
0.0946
-0.239
1
Olive
0.0014
0.2289
-0.185
1
Grape
0.068
0.0424
-0.015
0.1391
1
Sulphur
0.1914
-0.0124
0.2903
-0.036
-0.058
1
Feudal
Frac.
policies
Pop. density
High land
fract.
Large scale
plant.
0.0556
-0.1212
0.0598
-0.011
-0.132
0.0519
1
0.2637
0.0713
0.2214
-0.056
0.0552
0.1771
0.0531
1
-0.05
0.1145
-0.341
-0.039
0.0492
-0.182
-0.237
-0.057
1
0.0413
0.0491
-0.104
0.0946
-0.065
-0.048
-0.038
0.073
0.1596
1
0.2465
0.1623
0.1054
-0.020
0.1406
0.0299
0.0642
-0.15
-0.2317
-0.1886
Large
scale
plant.
1
Table 9: Mafia probit model
Dependent variable:
Mafia (in 1880)
Citrus
Grape
Olive
Wheat
Sulphur
(1)
1.164**
(0.453)
0.273
(0.371)
-0.322
(0.209)
0.455
(0.321)
0.416
(0.303)
(2)
1.401**
(0.567)
0.620
(0.422)
-0.451
(0.291)
0.269
(0.402)
-0.195
(0.387)
(3)
1.488**
(0.701)
0.603
(0.431)
-0.325
(0.315)
0.342
(0.423)
-0.351
(0.533)
Population density
(4)
1.363**
(0.581)
-0.0291
(0.0676)
0.831***
(0.246)
Fractionalization policies
High land fractionalization
(5)
1.479***
(0.540)
(6)
1.342**
(0.606)
0.970***
(0.198)
-0.128
(0.293)
1.100***
(0.197)
Large scale plantation
-1.468***
(0.411)
-1.311***
(0.354)
-1.461***
(0.416)
-0.891***
(0.203)
-1.144***
(0.304)
0.962***
(0.173)
-1.311***
(0.281)
Predicted probability
Area under ROC curve
0.16
0.76
0.28
0.83
0.31
0.85
0.34
0.86
0.35
0.86
0.39
0.88
Provinces
All
All
Without
Caltanissetta
All
All
All
Province dummies
Observations
No
138
Yes
138
Yes
123
Yes
124
Yes
113
Yes
119
Constant
Notes: The estimator is binomial probit in all specifications. Clustered robust standard errors in parentheses.
*** p<0.01, ** p<0.05, * p<0.1.
Table 10: Probit regressions with additional controls
Citrus
Fractionalization policies
Large scale plantation
Distance from Palermo (in log)
Distance from Mazzara del Vallo (in log)
(1)
1.327**
(0.612)
1.003***
(0.217)
0.987***
(0.232)
-0.311
(0.295)
-0.220*
(0.116)
Feudal
Dependent variable:
Mafia (in 1880)
(2)
(3)
(4)
1.183**
1.167**
1.472***
(0.598)
(0.547)
(0.545)
0.823***
0.970***
1.037***
(0.229)
(0.258)
(0.245)
0.902***
0.903***
0.832***
(0.180)
(0.225)
(0.256)
-0.156
(0.512)
Distance from railway station (in log)
-0.216*
(0.129)
Not care of law (dummy)
1.654**
(0.642)
0.149
(0.528)
0.584
(0.809)
Mistrust in law (dummy)
Trust in law (dummy)
Length of contract (in log)
Constant
Predicted probability
Area under ROC curve
Province dummies
Observations
(5)
1.318**
(0.571)
0.988***
(0.234)
0.397
(0.244)
0.729
(1.528)
0.4
0.89
Yes
119
-1.671**
(0.756)
0.35
0.87
Yes
117
-1.618**
(0.778)
0.36
0.87
Yes
112
-2.424***
(0.906)
0.41
0.89
Yes
112
-0.606***
(0.121)
-0.631
(0.628)
0.34
0.86
Yes
100
Notes: The estimator is binomial probit in all specifications. Clustered robust standard errors in parentheses. ***
p<0.01, ** p<0.05, * p<0.1.
Table 11: Exclusion of altitude
Citrus
Fractionalization policies
Large scale plantation
(1)
1.342**
(0.606)
1.100***
(0.197)
0.962***
(0.173)
Altitude (in log)
Constant
Predicted probability
Area under ROC curve
Province dummies
Observations
-1.311***
(0.281)
0.34
0.86
Yes
119
Dependent variable:
Mafia (in 1880)
(2)
1.012***
(0.172)
1.135***
(0.187)
-0.270***
(0.0882)
-0.0846
(0.221)
0.27
0.83
Yes
119
(3)
1.031
(0.633)
0.960***
(0.195)
0.920***
(0.222)
-0.122
(0.121)
-1.173
(1.041)
0.34
0.86
Yes
119
Clustered robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1.
Table 12: Robustness checks with alternative estimators
(1)
Estimator
Citrus
Fractionalization policies
Large scale plantation
Panel A
Dependent variable:
Mafia (in 1880)
2nd stage estimates
(2)
(3)
LPM
IV
IVPROBIT
0.319**
(0.126)
0.254***
(0.0470)
0.230**
(0.0650)
0.431**
(0.204)
0.249***
(0.0812)
0.209**
(0.0893)
1.714***
(0.333)
0.825**
(0.326)
0.679**
(0.305)
0.176***
(0.0410)
0.377
Yes
119
0.44
0.144
(0.155)
0.376
Yes
119
0.42
Spatial autocorrelation
Constant
Predicted probability
Province dummies
Observations
R-Squared
Excluded Instrument: Altitude (in log)
Anderson LR Statistics
Cragg-Donald F-Statistics
Stock and Yogo 10% critical value
Partial F-statistics
Endogeneity Test (p-values)
-1.894***
(0.669)
0.319
Yes
119
0.902
Panel B
1st stage estimates for
Citrus
-0.160*** -0.156***
(0.030)
(0.044)
21.921
22.027
16.38
28.21
0.5424
(4)
SPATIAL
LPM
0.316***
(0.0749)
0.210**
(0.0816)
0.225***
(0.0859)
-0.143
(0.409)
0.0196
(0.0934)
0.379
Yes
116
0.44.
Notes: The estimator is a Linear Probability Model (LPM) in column 1, IV in column 2, IVPROBIT in column 3,
and Spatial LPM in column 4. In columns 3-4, we run two-stage estimations with Citrus as the endogenous variable
and with Altitude as the excludable instrument. Robust standard errors in parentheses. *** p<0.01, ** p<0.05, *
p<0.1.
Table 13: Mafia persistence, 1880-2008
Province
Mafia indicators
Mean
Assets
mafiosi
confiscated
per capita
from the
in 1999
mafia 19972008
0.071
5.368
Mafia
(in 1880)
Mafiosi
reported
in 1999
Assets
confiscated
per capita in
1997-2008
Caltanissetta*
0.437
31
Catania
0.318
56
0.051
4.566
0.132
Grigenti
0.823
9
0.019
5.125
0.447
Messina
0.24
27
0.041
2.384
0.079
Palermo
0.296
30
0.024
70.900
0.855
Siracusa*
0.142
43
0.059
4.045
0.128
Trapani
0.4
5
0.018
17.266
0.679
0.346
201
0.040
19.027
0.377
0.535
(Agrigento)
Total
Numbers in the first column refer to the share of towns within each province that had mafia in 1880. The second
column shows the number of mafiosi cases reported in 1999 per province whereas the third column shows the
number of reported cases per capita in each province. Columns four to five show the number of assets confiscated
from the mafia during 1997-2008 in absolute and per capita terms. See table A1 in the Appendix for an explanation
of the variables.
*Caltanissetta is merged with Enna and Siracusa with Ragusa as in the Damiani Inquiry.
# Population is in thousands
Table 14: Historical mafia presence as a determinant of current mafia activity
Mafia (in 1880)
Province dummies
Observations
R-Squared
Dependent variable:
Assets confiscated per capita
1997-2008
(1)
(2)
0.303**
0.247*
(0.140)
(0.141)
No
Yes
143
143
0.04
0.19
Notes: The estimator is OLS in both specifications. Robust standard errors in parentheses. *** p<0.01, ** p<0.05,
* p<0.1. Trapani is the excluded province in column 2.
Figure 1: Equilibrium number of lemon producing towns
αY
Fi=a+bi
αY/i2
αY/n2=a+bn
D
a+b
(al+bl)
αY/I
E
(Fi=al+bli)
1
n
I
Figure 2: Mafia and non-mafia towns in Sicily in 1880s
Appendix (online publication only)
Table A1: Variable descriptions
Variables
Description
Source
Mafia
Dummy variable for whether the pretori
reports mafia as one of the most
common form of crimes
The relevant question in the Inquiry is:
What is the most important form of crime?
Dummy Variable proxying the most
important crop produced in the town.
The question in the Inquiry is:
What is the dominant crop in the town?
Damiani (1886). Original
surveys completed by pretori
Dummy variable for whether there is a
sulphur mine.
Dummy variable coding the political
organization within the town before the
Unification of Italy.
Di Vita (1905)
Scale of the Plantation
Dummy variable for the scale of the
plantation.
Question: Is the large, medium, or the small
scale of the plantation prevalent?
To be noticed that there is not a
universal criterion to classify the scale.
For example the mayor of Caltagirone
considers large a plantation extending
for more than 2000 hectares and small a
plantation extending for less than 20
hectares. On the other hand the mayor
of Paterno’ considers large a plantation
extending over 50 hectares and small
one extending over less than 1 hectare.
Most of the times this difference reflects
the extent of the town.
This difference is important because
affects the answer on the
fractionalization of land below
Damiani (1886). Original
surveys completed by mayors
excepted for the Province of
Caltanissetta for which we use
the summary Report (Damiani).
Fractionalization of Land
Dummy variable for the level of
fractionalization of land.
Question: What is the fractionalization of the
land and which factors have affected it?
As mentioned above this is affected by
the classification of the scale of the
plantation above
Dummy variable coded one if policies
aimed at increasing peasants’ ownership
have been effective in fractionalizing the
land. Policies considered are: Enfiteusi,
Abolishment of Feudalism, and
Privatization of the church’s assets.
Question: Which factors have affected the
fractionalization?
Altitude above the sea level (in metres)
Damiani (1886). Original
surveys completed by mayors
excepted for the Province of
Caltanissetta for which we use
the summary Report (Damiani).
Citrus, Grape, Olive, Wheat
Sulphur mines
Feudal, Church-Ruled, CrownRuled.
Fractionalization of Policies
Altitude and Distance from the
Damiani (1886). Original
surveys completed by mayors
excepted for the Province of
Caltanissetta for which we use
the summary Report (Damiani).
Di Vita (1905) is used to check
the information in Damiani.
Di Vita (1905)
Damiani (1886). Original
surveys completed by mayors
excepted for the Province of
Caltanissetta for which we use
the summary Report (Damiani).
Di Vita (1905).
Railway
Rule of Law
Length of Tenancy Contract
Mafia Crimes
Assets Confiscated
and distance in Km from the closer
railway station.
Dummy variable for the trust people
have in the application of law.
Question: Do people trust, mistrust, care, of
fear the law?
Question: What is the average length of
the tenancy contract?
In some cases the mayor reports that the
contract lasts for the entire life. In this
case we assume an average life
expectancy of 60 years and an average
duration of the contract of 40 years. We
also try with different numbers (20, 30,
50 years) but results do not change.
Number of mafia crimes reported to the
police which are punishable according to
the 416bis article of the Italian Penal
Law.
Number of assets (i.e. apartments,
buildings, etc) confiscated from the
mafia in the town.
Damiani (1886). Original
Surveys completed by pretori.
Damiani (1886). Original survey
completed by mayors excepted
for the Province of Caltanissetta
for which we use the summary
Report (Damiani).
National Institute for Statistics ISTAT (1999)
Agenzia del Demanio (2008)
Table A2: IV without the province of Caltanisetta
Estimator
Citrus
Fractionalization Policies
Large Scale Plantation
Constant
Province Dummies
Caltanissetta Excluded
Observations
R-squared/Predicted Prob
Excluded Instrument: Altitude
Anderson LR Statistics
Cragg-Donald F-Statistics
Stock and Yogo 10% critical value
Partial F-statistics
Endogeneity test (p-values)
Panel A
Dependent variable:
Mafia (in 1880)
nd
2 stage estimates
(1)
(2)
IV
IVPROBIT
0.455**
1.779***
(0.213)
(0.331)
0.225***
0.731**
(0.0818)
(0.328)
0.232**
0.759**
(0.0922)
(0.339)
0.136
-1.901***
(0.157)
(0.707)
Yes
Yes
Yes
Yes
109
109
0.4
0.86
Panel B
1st stage estimates for Citrus
-0.157***
-0.154***
(0.030)
(0.0458)
21.095
21.353
16.38
27.24
0.4285
Figure A1: Variables in the original Inquiry Damiani (1886) related to mafia
Note: Figure A1 shows a picture of the second part of the table from which we got data on mafia. The first row reports variables. Starting from the left, the first
variable is Scostumatezza (lewdness). There are three possible causes among which the prefect can choose. The first one is Adulterio (adultery), the second Incesto
(incest), the third is Nascite Illegali (illegally born child), and the last one is Varie (various). The second variable relates to the religiousness of people in the town,
the third one relates to the clergy (corrupt or exemplary), and the fourth variable regards perjury. The last variable is Vagabondaggio accattonaggio (vagrancy and
begging). The fifth variable is the one we use to get information on mafia. The variable is labelled Reati (crimes) and then it is asked what the most common crimes
are and the extent of these crimes. Most of the time, pretori only answered providing information on the sort of crime committed. The most common forms of
crime were rustling, mafia, bloody crimes, and bloody crimes for passion. In addition poverty was described as the most common cause of crime.
Figure A2: Inchiesta Agricola (Agricultural Inquiry) in Damiani (1886)
Note: Figure A2 shows the first page of the Agricultural Inquiry for the city of Cefalu. The column on the left reports the question (Quesiti). The column on the
right reports the answer of the mayor. For example the first question asks “What is the surface area of the city”, “In how many areas the territory can be divided?”
and “What is the extent of each area?” The mayor answers that the surface land of Cefalu is 52,943,492 sq.mt. and the territory is divided in three zones: 1) plain; 2)
hills; 3) mountains. The first zone extends for almost 2,500,000 sq.mt.; the second for almost 20,000,000 sq.mt.; and the third for almost 30,443,492 sq.mt. The
second question relates to the physical and chemical characteristics of the territory. This first part of the inquiry which is titled the Condition of Agriculture also
reports information on the kind of crops produced, the sort of manufactures developed in the city, and so on. The second part of the Inquiry relates to the
relationship between peasants and landlords, while the third part regards the moral conditions of peasants.
Figure A3: Optimal mafia effort m* at varying level of profits π*
ω**=0.297
ω*=0.526
m
m**
0.625
Profits
increase
0.5
0.375
0.25
m*
0.125
0
0
0.25
0.5
0.75
1
om ega
Note: The initial example above (giving the curves m* and ω*) assumes the following parameter levels: d=0.4, θ=0.5,
π*=8, A=10. In the second example, we alter only the level of profits to π*=32 which yields the new curves m** and
ω**. The equilibrium level of mafia activity thereby increases from m*=0.09 to m**=0.236.

Per leggere la pagina clicca qui - Campionato di Giornalismo il

Comunicato Stampa INAIL SuperAbile presidio in Prefettura

Intro - WebLearn

Untitled - A riveder le stelle

17 12 14 Il Messaggero Truffa alla Marina

Programma convegno.pub

21 12 14 Corr della Sera Roma e i 300 milioni di rimborsi

Tesina di Maturità – La mafia – Ragioneria

09-02-2014 Il Cittadino

Un - Virtualnewspaper