<<

Inherited culture, institutions and economic development in

M. C. Bramati,† A. Palestini,‡ and M. Rota§

Please do not quote this preliminary version

Abstract This paper establishes a precise channel of long run effect of inherited cultural traits on current development. We use data on 102 of Italy and we consider whether cultural traits from the past shape the today quality of formal institutions within country, proxied by the length of trials by . Our results point to a strong effect of inherited cultural traits on the different quality of the today formal institutions whereby different levels of development are determined within country. Those effects are robust to the inclusion of accumulation of human capital, past economic development, geographical location, and current civic capital. Moreover, we found that the effects of inherited cultural traits might be additional and not necessarily alternative to past formal institutions in influencing current economic behaviour and outcomes.

Jel classification: O10, F10, P10, N13 Keywords: inherited cultural traits, formal institutions, economic development, civil trials.

∗We acknowledge the Faculty of Economic, Sapienza University of for financial support, (under-40 researchers funding program 2012). †Sapienza, University of Rome, MEMOTEF Department, via del Castro Laurenziano, 9 000161 Rome ‡Sapienza, University of Rome, MEMOTEF Department, via del Castro Laurenziano, 9 000161 Rome §Sapienza, University of Rome, MEMOTEF Department, via del Castro Laurenziano, 9 000161 Rome, email:[email protected] corresponding author

1 1 Introduction

The cultural traits are often invoked as the determinants of different economic outcomes. Pref- erences, norms and beliefs are determined by those cultural traits which in turn are the results of events and institutional heritage from a very distant past (Bisin and Verdier 2001; Voigth- lander and Voth, 2012). Italy is a valuable laboratory to assess whether the large territorial disparities in terms of income per capita are generated by different cultural traits, historical heritages, and institutional set-ups. After more than 150 years since the political and administrative unifications (1861), the South has not yet converged to the North (Felice 2013) despite ad hoc economic policies, as in the early years of XX centuries or in the four decades after World War II, and despite of the common set of formal institutions governing the social and economic life from the Alpes to the Strait of . None of other European countries experiences larger and more persistent within country gap. Two explanations of the fail in convergence of the Italian have become popular in the past years calling into question the cultural traits: the attitude to satisfy self interests (the amoral familism of Banfield, 1958) and the lack of cooperative behaviour and trust (Putnam 1993). These hardly ever alternative perspectives could explain why in the South people react in a different way to the same set of incentives faced by Northern inhabitants. Yet, it is obvious that incentives can vary from North to South along with the endowment of infrastructures, presence of organised crimes, investments and many other conditions. A recent lab-in-field experiment (Bigoni et al. 2015) has revealed that a gap in cooperation between North and South persists even under the same set of incentives and that the measures of aggregated social and civic capital do not reflect the pattern in cooperation found into the intra-groups behaviour. These results leave an open question about the origin in a distant past of and the persis- tence of cooperation gap whose effects may still influence the level of development after 150 of unification. Moreover, there is lively debate through which channels, if any, the cultural traits shape the economic behaviour. In lines with a growing literature (Guiso et al. 2004, Ak¸comakand Ter Weel, 2009, Tabellini 2010; Algan and Cahuc, 2010) that searches in the past the roots of current behaviour of groups and societies, in this paper we explore whether some cultural traits, such as cooperative behaviour, trust, and inclusiveness as well as formal education inherited from a distant past influence the quality and efficiency of formal institutions today in Italy which in turn contribute to explain the different level of development across areas. Indeed, we argue that the transmission of historical events runs from the inherited past informal institutions and norms, determined by the cultural traits, to the current formal institutions, complementing the majority of the theoretical and empirical studies that have assessed a precise channel running from the past inherited formal institutions to both current formal institutions and cultural norms (Guiso et al., 2010; Voigthlander and Voth, 2012; Borowiecki, 2015). We have found the signs of past cultural traits in the diffusion of and in the gender composi- tions within the Societ`adi mutuo soccorso, the voluntary organizations that provided assistance to members and that emerged in the early years of unification. We assume the Societ`a as an indirect measure of trust, inclusiveness and cooperation of the past times. In 1873, six south- ern provinces had no Societ`a and the rest of the south had yet not developed the network of informal support that emerged in the north and in some areas of . It is commonly shared the idea that the civic tradition of the Italian cities has its roots in the communal life of late Middle Ages or in the Signoria of the Renaissance (Putnam 1993) whereby the cultural traits we are considering can be viewed as a wrong or biased perspective of the civic capital inherited. That could be true because the endowment of civic capital emerged in the XII-XV centuries may have produced economic benefits that perpetuated differences in

2 civic capital until 1873 and beyond. We will discuss this aspect in the text showing in section 7 that controlling in our empirical model for current measures of civic capital the transmission channel from our measures of inherited cultural traits to current formal institutions is not altered yet it turns out to be complemented. Moreover, we add to the standard concept of trust and cooperation the literacy rate observed in a very distant past to get a snapshot of the overall cultural traits. If we want to track the inherited culture we must take into account that relatively more educated people contribute to the social life of their own community in a well-informed way. More educated persons join easier the community life, are less affected by stereotypes and prejudices, and are more prone to absorb the social norms and sentiments delivered by the diffusion of formal and informal education. Moreover, inclusion of literacy rate in our analysis helps to discount the potential degree of transmission of trust, cooperation and inclusiveness through generations especially when we are not able, as in our context, to directly measure their intergenerational transmission. Our choice to include literacy rate is therefore consistent with Tabellini (2010) who uses past illiteracy rate to causally identify the effect of today culture on current economic development. The current formal institution we consider is the degree of protection of the property rights. Measures of property rights protection are available at country level but at either regional or provincial level for Italy there are no comparable data. Hence, we look at the length of civil trials as a sign of institutional efficiency. The judicial system determines how much the property rights are secured by the law enforcement and how much the legal framework protects the potential investors and consumers. Trials are longer in the Southern courts suggesting a lower degree of protection of property rights. In 2006 the median length of a first degree trial for 102 Italian provinces considered in this study was 464 days (, ) and only eight northern provinces placed over the median while none southern provinces was below that value. The standard variation across Italy is high and equal to six months. What can account for the different length of trials across the Italian courts given that the judicial system in Italy is the same from North to South in terms of rules and personnel endowment and to the extent that the rules of law are established at national level? We relate the different efficiency across area of the same institution to the informal institutions that each area of Italy inherited from the past. In particular, the length of trials today are linked to the intrinsic degree of infighting among citizens. Because the informal institutions may influence the efficiency of formal institutions (Tabellini, 2010) the different degree of infighting can affect the working of judicial courts generating longer trials. A society endowed with a sufficiently high level of trust, cooperation and inclusiveness generates less judicial controversies than societies in which citizens do not trust and cooperate each other. The analysis shows that the inherited cultural traits may explain the different quality of the same institutions within countries. In our empirical analysis we find a strong and robust effect of inherited cultural traits on the quality of formal institutions proxied by the length of the civil trials which in turn contribute to explain the differences in the level of development of the Italian provinces. The results are robust to the inclusion of contemporaneous accumulated human capital, past level of development, geographical variables and the degree of openness. Most importantly, those effects are robust when we control for current measures of social or civic capital. A further robustness check we provide is the ”horse race” between inherited formal institutions and cultural traits. If we use past formal institution as excluded instrument the effects of cultural traits on current formal institution is strongly confirmed and, as we expect, past formal institutions influence current efficiency of the courts. Yet we do not find any effect of past formal institutions in the equation that explains the level of development when they are used as included regressor. Obviously, in this specification past formal institutions continue to be effective on current quality of formal institutions and the cultural traits effect is still in place.

3 Hence, our approach is new from three perspectives. First, we relate the quality of insti- tutions today to the informal inherited institutions trying to bridge the gap between generic assumptions and observable data. Second, we try to quantify the past informal institutions capturing territorial differences in a detailed way and trying to interpret as close as possible the Putnam’s idea of inherited civic capital. Finally, we complement the civic capital with the literacy rate to obtain an extensive measure of cultural traits. Our paper also aims at contributing to the literature on the role played by inherited cultural traits. First, compared to the existing literature that emphasizes that past cultural traits display permanent effect on current cultural traits, belief, preference and norms (Voigthlander and Voth, Guiso et al. 2010) we add a further effect that runs from the inherited cultural traits to formal institutions, at least in a within country context. Second, we provide evidence that past cultural traits are a separate and complementary channel to past formal institutions in shaping current formal institutions, and that they might be not an alternative channel. Third, we have also found robust evidence that the effects of past cultural traits cannot be mixed up to current civic capital either on current formal institutions or on the current level of development. This last result is consistent with the findings of Guiso et al. (2004) who distinguish between inherited and acquired social capital. Moreover, the other body of literature we aim to contribute is the extensive literature on income divergence and to that on North- South disparities in Italy, in particular. The empirical results support uncontroversially that history matters in the territorial divergence and that past cultural traits are still playing a role, distinct from the civic or social capital. In a very long run perspective Felice (2012) discharges the hypothesis that inherited social capital still affects convergence in the . The effect is found to be strong only in the last decades. We offer a different explanation than Di Liberto and Sideri (2015) who use measures of institutional quality proxied by the provision of several public services, including the length of justice in 2004, to explain income per capita differences of 103 Italian provinces. To the endogeneity problem they instrument the institutional quality with some historical variables. In particular, the causal identification is carried out using the different foreign dominations that ruled Italian regions between the 16th and 17th century and over seven hundred years before the creation of the unified Italian State. The way in which they identify causality is by the use of generic codification of past institutions. The South of Italy in XIX century was under the rules of Bourbons and they left in heritage the same set of formal institutions in and Sicily, where the last King Francis II never went during his reign. Despite this fact, the provinces of Campania and Sicily display wide differences in income per capita. The same is true for the provinces under the Papal State. is by far richer than any other province of the former Papal State (with the exception of Rome) and substantially richer than several northern provinces. The same argument can be used to doubt about the genuine effects of dominations before XIX century. Moreover, it could be questioned that some of past dominations delivered inefficient institutions that persisted. As example, let us consider the negative effect of the Norman domination of the XII Century in the South found in Di Liberto et al. (2015). It is difficult to reconcile this effect with the fact that the Reign of Frederick II of Swabia who ruled in and Sicily and is historically considered an innovator of bureaucracy and laws did not leave any significant impact on the quality of informal institutions in that area. Yet, a closer look at the duration of past domination suggests that were the Spanish rulers between the early XV century and the Treaty of Utrecht (1713) that deeply set the path of formal institutions. Precisely, those were the years in which the Gran Duchy of and the Centre-North of Italy were experiencing the epoch of the formation of civic tradition according to Putnam (1993). The sign of this heritage can be found, as our framework suggest, in the spirit of mutuality and trust that shaped social and economic preferences of the Italian populations. The first systematic snapshot of this heritage is, to our best knowledge, found in the voluntary associations of the early stages of Unification

4 and in the formal and elementary level of education of a distant past.

2 Judicial trials and the efficiency of formal institutions

The Italian civil judicial system is organized in three degrees of judgement. All citizens can claim their alleged rights in the first degree without any limit. The access to the second degree is unrestricted but it is expensive. The last degree is the claim at the High Civil Court which is restricted to some technical and legal circumstances. Hence, the first degree of judgement is the most informative in terms of efficiency of the judicial system as the access is actually free and unrestricted. Moreover, the second degree is held in the Corte d’Appello serving more than one province making it difficult to have data on the local degree of efficiency of the courts. As far as the first degree is concerned, the territorial distributions of the courts overlaps the administrative partition at NUTS-3 level with some exceptions. We have data on 144 courts which is more than the Italian administrative provinces. To make data comparable to economic variables we aggregate the courts according to the provincial administrative partition. As example, because the has three active courts, Rome, and we summed up the trials in each court assigning the resulting value to the province of Rome. Hence, we obtained data on judicial trials for 106 provinces. Available data show the existing trials at the end of the year, the new trials and the ex- hausted trials. The average length of trials is not observed yet can be estimated. Following the procedure of the national institute of statistics (ISTAT) we compute the length of trials as: P ending + P ending Length = t−1 t ∗ 365 (1) Closedt + Newt Figure ?? shows the geographical picture of the average length of the civil trials in 2006 the last year before the crisis. The average length was 478 days with a standard deviation of 184.5 days. The maximum value of 953 days has been registered in () the lowest, 181 days, in Mantova. The southern courts take a longer time to close a trial than the northern courts affecting the contracts enforcement and the protection of property rights in a negative way thereby determining a lower quality of the institutional set-up. A wider period to close a trial makes property rights less secure and increases uncertainty whereby the transaction costs increase and the economic exchanges are discouraged. Figure ?? indicate a clear pattern of the link between the length of trials and value added per capita. The unconditional association suggests that development and efficiency of institutions are closely and negatively linked although nothing can be said about causality. Thus, the geographical distribution of the length of civil trials is affected by several factors. One possible reason is that in the more prosperous areas the potential sources of litigation are mitigated by a wider availability of resources. If the pie is larger there is less probability that judicial conflicts might emerge relaxing the pressure on the courts. Yet in turn, it is also true that in the richer areas the number of the economic transactions is larger and the number of controversies is potentially higher than in the more backward areas in which less transactions occur. If this is the case, we expect that judicial claims could jam the working of the courts up. In this perspectives, though from very different sides, the length of trials is strongly influenced by the level of income. The last consideration recalls a common conundrum when institutions and income level are considered, namely the direction of the causality nexus. Moreover, because the judicial system is governed by national rules, the differences in the length of trials can be related to the efficiency of the human capital in the courts (judges and auxiliary personnel). It is difficult to estimate efficiency of the human capital working in the courts given that, in principle, judges and auxiliary personnel are selected according to standardized quality at national level through a competitive selection process. We could observe

5 Figure 1: The length of civil trials in 2006. Source: our computation on data provided by Ministry of Justice.

600 - 1.000

500 - 600

400 - 500

350 - 400

300 - 350

180 - 300

N.A.

Figure 2: Length of trials, appointed judges and value added per capita

(b) Correlation between the length of trials and the (a) Correlation between value added per capita and appointed judges (data on appointed judges were kindly provided civil trials. by the Consiglio Superiore della Magistartura.

Milano

1000

3.5 Vibo Bologna Roma ForliAostaParmaReggioFirenze TrevisoBergamoPadova 800 Mantova PordenoneVeronaBresciaVicenzaVenezia Torino VareseRavennaRiminiPiacenza MonzaVercelliLeccoUdine AlessandriaBiellaBelluno SondrioCremonaAnconaPesaroSienaPratoGenovaSavona ComoArezzo Latina LivornoLodiLucca RovigoFerraraMacerataGrosseto Asti Siracusa PaviaPistoiaImperia Ascoli CataniaAvellino Tempio Latina 600 FrosinonePalermo 3 Ragusa PescaraTeramo Aquila Napoli Isernia Perugia Tempio Aquila Massa CampobassoSassari Ogliastra Cagliari BellunoMatera Rieti Ragusa Pisa MacerataPescara AscoliAgrigento SassariSiracusa Viterbo Roma Chieti Catanzaro PotenzaBari Venezia Firenze

log of value adde per capita Nuoro 400 Padova MateraMessina BolognaLa Spezia Terni PalermoSalerno PiacenzaRimini Genova OristanoAvellino Taranto BresciaParmaPistoiaRovigo FerraraReggioPrato EmiliaMilano Lecce TrevisoModenaArezzoVeronaCremonaPesaro AlessandriaPavia Napoli Cosenza Brindisi Reggio Calabria PordenoneVercelli Length of civil trials in days (average 2006−2011) ForlìAncona TrapaniBeneventoOgliastra ComoNovaraRavennaVareseLivornoLodi Verbania Caltanissetta Vibo Asti FoggiaCaserta Torino Crotone Cuneo Enna

200 Mantova 2.5 200 400 600 800 1000 .2 .4 .6 .8 1 length of civil trials (first degree) in days Judges appointed each 10000 inhabitants

6 or estimate the individual effort and efficiency if detailed data would be available which is not the case. An obvious concern is that the number of judges in the courts accounts for the differences in the length of trials. In particular, the southern courts could be refused by the appointed judges because the living standards in the South are less attractive. This circumstance seems not verified for two order of reasons. First, the number of judges is predetermined, according to the number of inhabitants, each year by the Ministry of Justice who replaces the retired judges or those appointed to other services by a competitive selection process. Second, the available data show that the length of trials and the number of judges relative to population appointed in each court are positively correlated meaning that the Ministry, observing the length of trials or other information on the degree of litigation, reacts by appointing more judges. Yet, the reverse is not true. The different length of trials is not the result of understaffed courts. In turn, the observed longer trials and the resulting poor quality of formal institutions in the south could be the outcome of the poor endowment of civic capital, of the quality of human capital as well as of some predetermined historical facts. We relate the length of civil trials to the degree of infighting and litigation which can be higher in the south because a society endowed with a sufficiently high level of trust inclusiveness and cooperation generates less judicial controversies than societies in which citizens do not trust or cooperate each other. It is the civic tradition of Northern and part of the central Italy rooted in their history that still influence the degree of infighting and litigation today even after the social and economic events, many of them common to the whole Italy, have raised physical, human, and civic capital.

3 Cultural traits from the past

Past civic capital or cultural traits because direct measures that could be assembled using surveys such as World Value Survey or in-field experiment are obviously unavailable for a very distant past. We consider the Societ`adi Mutuo Soccorso that emerged in the XIX century in each province as a proxy of trust and cooperation in the past. The Societ`adi Mutuo Soccorso were voluntary organizations among citizens emerged in order to provide mutual assistance in case of adverse events such as physical diseases and poverty, and the support of widows and sons of the de- ceased members. They were diffuse in Europe since about the medieval times, although in Italy as well as in the rest of the continent, their institutional profiles evolved in the XVIII and XIX centuries. The medieval and early modern period congregations were charitable - ciation organized according to hierarchical roles of members. The sympathetic behaviour was to some extent vertical and the membership was restricted to a particular profession. In turn, the associations after XVIII century were free from any hierarchical relationship among the members (horizontal) and open to all the citizens in the majority of the cases. Apart from the effectiveness in providing assistance and mutuality which was by far more efficient in the modern associations, the Societ`adi Mutuo Soccorso testify to the existence of cooperation and trust in the areas where they developed. Hence, the membership in the Society did not provide a direct economic pay-off. Being a mem- ber entailed to support other members in event of sickness, widowhood, unemployment, and orphanhood. Nobody at the moment of becoming a member could know if he/she would in the future be in one of the previous conditions. If were the economic rational expectations to inspire the membership we expect that in the South cooperation ought to be higher, because the mortality rates, life expectations and income per capita were lower than in the North. As it shown in the next paragraph, in the South cooperation was effectively lower signalling that joining a Society was the mirror of civic attitude rather than of economic pay-off. Moreover, in 1873 the Societ`a were not yet regulated by any law and they conveyed the genuine attempt to organize and provide assistance to members. In the Reign of Italy in 1862 there were 443

7 organizations which origins date back to the Unification for 234 of them. Before the unification in the south only 13 societies were active. Between the unification and 1873, 146 societies were created in the south and 1053 in the centre-north without any particular effect of the unification in their geographical distribution. The first comprehensive census of the society were taken in 1873 (MAIC, 1875) after the ultimate defeat of what remained of the Papal State. The census reported important information on the Societ`adi Mutuo Soccorso, and in particular on their exact location, the number of members and the gender composition of the membership. In 1873 were recorded 1448 Society and 220818 members whose 199816 were men and 21002 were women. The exact location of each society allows us to take account of changes in the provin- cial boundaries and to match the societies with the current boundaries of each Italian province finally obtaining a distribution of membership in the past comparable to the distribution of the current formal institutions and income per capita. We are forced to exclude the provinces of , , and that joined the Reign of Italy after and for which we do not have data. We use this information to build two measures of cooperation, trust and inclusiveness from the past. The first is the number of members registered in the Societ`a each ten thousands inhabitants. The second is the ratio of female members to male members. A higher value of the ratio may signal a higher degree of inclusiveness of a society. In computing our measures we have taken into account the response rate to the survey. In 1873, just 280 out of 1448 did not send data for various reasons and without any geographical regularity. Hence, we re-weighted our measures of civic capital on the response rates.

Figure 3: Components of past cultural traits in the early 1870s. Members of the Societ`adi mutuo soccorso each 10000 inhabitants (Panel A), associated female to male ratio (Panel B), and illiteracy rate (Panel C). Source: our computation on MAIC (1875a and 1875b)

(a) Members of the Societ`aeach (b) Female to male ratio in the 10000 inhab. Societ`a. (c) Illiteracy rate.

0 - 10 0 0,88 - 0,923

10 - 30 0 - 0,05 0,84 - 0,88

30 - 60 0,05 - 0,075 0,78 - 0,84

60 - 100 0,075 - 0,1 0,7 - 0,78

100 - 200 0,1 - 0,15 0,5 -0,7

200 - 425 0,15 - 0,72 0,4 - 0,5

N.A. N.A. N.A.

We complement the measures of cultural traits drawn by the membership in the Societ`a with the level of literacy rate in the past. Although the literacy rate can be viewed as a proxy of the human capital accumulation, it is an active part of the building of the cultural traits of a society, as the frequently cited piece of Cipolla (1969, p.102) argues: ”widespread literacy meant not only an elastic supply of literate workers but also a more rational and more receptive approach to life on the part of the population”. This is true for the acquisition of techniques and professional skills as well as of the customs and beliefs delivered by the formal education. An individual in absence of access to formal education acquires his cultural traits typically from

8 Figure 4: Geographical distributions of the inherited cultural traits

-1 - -0,6

- 0,6 - -0,3

-0,3 - 0

0 - +0,6

+0,6 - +1,2

+1,2 - +2,4

N.A.

parents, determining the long term persistence of those traits (Bisin and Verdier, 2001). The imperfect empathy is responsible for the persistence given that parents transfer to children the cultural traits and have a preference for their own cultural beliefs. Yet, many historical studies argue that changes in cultural beliefs are also possible. For example, the feelings about Jewish population, though persistent in Germany, evaporated in the German Hansa cities because of their particular geographical and economic evolution showing that cultural traits are malleable under some circumstance. Hence, the inclusion of literacy rate is in turn a potential predictor of future changes in cultural traits that we do not observe. Moreover, Tabellini (2010) includes past literacy rate as a key determinant of current cultural traits. Given the imperfect measurement of the different indicators of cultural traits, we suggest a continuous latent variable (CT ) to capture the historical cultural traits as the results of three components: the number of members of the Societ`adi Mutuo Soccorso each ten thousand inhabitants in each province (Members), the associated female to male ratio in membership (F toM) and the illiteracy rate in the past (Illrate):

IC = f(Members, F toM, Illrate) (2) The factor analysis returns a clear pattern of how the previous component determine the cultural traits. Membership in the Societ`a and related measures increase the cultural traits and the illiteracy rates lowers them. The latent variable patterns shown by the map in Figure ?? is clear: moving from North to South the cultural traits worsen. It is interesting to note that the areas with the best cultural traits matches with the provinces that Putnam emphasized in his celebrated book in 1993. Other area of high quality of cultural traits were , , , Lecco, and La Spezia. The distribution in the past of cultural traits seem to be largely independent by the level of past development. If we consider the urbanization rate in 1871 as a proxy of economic development, its correlation with our latent variable of cultural traits is virtually zero, In the empirical analysis of the next sections we control for the urbanization rates in the past to clean possible mutual effects between inherited cultural traits and past economic development.

4 The effects of formal institutions and inherited cul- tural traits on the level of development

In this section we estimates how much the quality and effectiveness of formal institutions influence the territorial disparities in Italy. Although the judicial system is the same within

9 the country, local courts perform on average better in the North than in the South with few exceptions. We argue, in first instance, that the differences in the functioning of the courts partly reflects the difference in the effectiveness of formal institutions as a whole. We start with a very simple model such as:

yi = α + δ ∗ Li + β ∗ y0 + γ ∗ si +  (3) where y is the value added per capita in 102 Italian provinces L is the length of process, y0 is a measure of development in the far past and i is a measure of human capital accumulation, i.e. the years of schooling,  is the error term. Our model includes regional fixed effects in the shape of fixed capital formation. We rely on this measure to capture unobserved heterogeneity across provinces. Generic dummy variable are avoided in our specification because of the loss in the degree of freedom. Moreover, the only reason to control for regional fixed effect is the role that regional government can play in the fixed capital formation. In fact, the other possible channel is precisely the quality of institutions that is the focus of the analysis. We will introduce other interesting fixed effects such as the presence of organised crimes (dummy variable equals to one if provinces are in the historical area of influence of Camorra, Mafia, and Ndrangheta and the membership to the historical Mezzogiorno club (provinces in the former Kingdom of the Two Sicilies and the provinces of ) . Our model is very similar to the basic specification in Tabellini (2010). The differences are that here we are interested to the role of formal institutions in income divergence and not to the current cultural traits that, to our best knowledge, is not possible to compute at provincial level for Italy. In a later section we will include current measures of civic capital which are close to, although different from, the current cultural traits. We anticipate since now that our entire reasoning is not invalidated by the current measure of civic capital. Table ?? investigates the correlation of formal institutions, namely the length of civil trials, on value added per capita in columns 1 and 2, the reduced form including the inherited cultural traits in columns 3 and 4, and the first stage regression of inherited cultural traits on current formal institutions in columns 5 and 6. Value added per capita is expressed in log whereas length of trials and the years of schooling are in level. All specification include regional fixed effect in shape of investment rate, south dummy variable and organised crime dummy variable. We look at 2006 which is the last year before the crisis. The OLS results in Table ?? indicate that the level of income per capita is inversely corre- lated with the length of trials, and, as we expect the years of schooling and the past development are positively correlated. The coefficient of the length of trials has a powerful economic mean- ing. A reduction in the length of trials of one standard deviation (184 days) is associated with an increase in the value added per capita by 4.8% in columns 1 and 2. Most importantly, the inclusion of inherited cultural traits in the reduced form equation states that historical vari- ables matters in the current economic performance and that our purported effect via formal institutions might work. Finally, columns 5 and 6 show that inherited cultural traits are cor- related with the quality of current formal institutions. The interpretation of the coefficients of inherited cultural traits is straightforward. In the reduced form if the provinces with the worst cultural traits in 1871 (Ogliastra in Sardinia) had had the best cultural traits in the sample (Biella in Piedimont), its value added per capita in 2006 would have been 7.2% higher. In the first stage regressions inherited cultural traits is negatively associated to the length of trials. If the province with the worst cultural traits had had the best in the sample the length of trials would be today shorter of by 102 days.

10 Table 1: Effects of formal institutions on value added per capita and reduced form of the effect of past cultural traits. OLS regressions

value added per capita 2006 length of trials 2006 (1) (2) (3) (4) (5) (6)

length of -0.000265*** -0.000265*** civil trials (7.46e-05) (7.16e-05) schooling 0.175*** 0.175*** 0.184*** 0.184*** -29.96 -29.96 (0.0273) (0.0224) (0.0245) (0.0211) (30.24) (36.13) urbanization 0.0775* 0.0775** 0.0525 0.0525 65.22 65.22 rate in 1871 (0.0431) (0.0365) (0.0413) (0.0432) (57.53) (72.60) inherited 0.0485*** 0.0485** -69.21*** -69.21*** cultural traits (0.0171) (0.0172) (21.91) (22.01) Constant 1.048*** 1.048*** 0.771*** 0.771** 961.4*** 961.4** (0.326) (0.306) (0.286) (0.290) (343.4) (365.3)

Observations 102 102 102 102 102 102 R-squared 0.890 0.890 0.886 0.886 0.674 0.674 *** p<0.01, ** p<0.05, * p<0.1 Robust standard errors in parentheses in cols 1-3-5, robust and clustered standard errors by region in cols 2-4-6.

5 Causal identification

However, the length of process are likely to be correlated with the error term  in equation ??raising standard problem of endogeneity. The potential sources of endogeneity are the reverse causality with the value added per capita and the omission of important variable. One way to overcome endogeneity is to model the length of process as a stochastic process in which the same regressors in the (??) appear together with some excluded variables able to capture the exogenous variation in the length of trials across provinces. We imagine that the length of trials is potentially influenced by the degree of litigation in each area which in turn depends on the inherited cultural traits. More cooperative and inclusive societies have a lower degree of infighting among citizens. Moreover, the cooperative behaviour matters also when a dispute arrives to the court. The judicial controversy is solved quickly if the sides do not try to circumvent the legal procedures or to cheat using improperly the legal instruments. However, we do not observe the degree of infighting among citizens. To cope with this issue we have proposed in the previous section the inherited cultural traits from a distant past. Because the cultural traits are long term persistent we expect that they still influence the behaviour of economic agents today. Therefore, the chain of causality ought run from inherited cultural traits to today formal institutions which in turn determine the differences in the level of development. Using the results of columns 5 and 6 of Table ?? we imagine that the length of trials can be approximated by the following stochastic process:

Li = a + d ∗ ICi + b ∗ y0i + g ∗ Xi + ηi (4)

where L is the length of civil trials today in the i − th provinces, ICi are the inherited cultural traits and Xi include the same set of regressors of equation ??. By combining equations ?? and ?? we might estimate the causal effect of the formal insti- tutions on the level of development today using as excluded instruments the inherited cultural traits. The following Table ?? shows the result of our exercise. Let us note that in columns 1 are

11 Table 2: Effects of formal institutions on value added per capita. IV regressions.

(1) (2) (3) (4) IV First stage IV First stage length of trials -0.000700 -0.000645 (0.000296)** (0.000273)** [0.000196]*** [0.000188]*** school 0.163 -29.96 0.160 -23.70 (0.0299)*** (30.24) (0.0274)*** (30.65) [0.0257]*** [36.13] [0.0250]*** [38.02] urb. rate 1871 0.0981* 65.22 0.0788 64.92 (0.0539)* (57.53) (0.0514) (58.44) [0.0488]** [72.60] [0.0449]* [71.41] inherited culture -69.21 -68.58 (21.91)*** (22.14)*** [22.01]*** [22.24]*** openess 0.0800 -25.75 (0.0349)*** (29.84) [0.0310]*** [28.21] Constant 1.445*** 961.4 1.422 906.5 (0.397) (343.4)*** (0.367)*** (350.9)** [0.330] [365.3]*** [0.320]*** [392.9]** Observations 102 102 101 101 R-squared 0.854 0.674 0.871 0.675 All specification include regional fixed effect in shape of investment rate, south dummy variable, and organised crime dummy variable. Robust standard errors in parentheses and clustered robust standard errors in squared brackets *** p<0.01, ** p<0.05, * p<0.1 reported the results of the instrumental variable regression, that columns 2 is, obviously, iden- tical to columns 5 (and 6) of Table ??, and that columns 4 and 5 include an addition regressor, import+export the degree of provincial openess ( grossvalueadded ), in both first stage and IV regressions. Our causal identification is supported into the data. By instrumenting the length of trials with the inherited culture we find that the effect of our proxy of property rights protection is confirmed. Compared to OLS estimation the coefficient is higher. A one month reduction in the length of trials increase value added per capita by 2.7%. Schooling maintains its positive and significant effect on value added per capita as well as past urbanization rate. The results are robust to the inclusion of a further regressor, the openess degree, in columns 4 and 5. Indeed, an excessively long time passing from submission to the first decision of a judicial controversy is a symptom of institutional inefficiency. Economic agents perceive the length of trials as an inadequate and insufficient level of protection of property rights, the transaction costs increase and the economic activity turns out to be depressed. It is also plausible that foreign direct investments are likely to occur in the areas with the lowest lengths of trials, inducing non- domestic agents to select the location of investments.However, we do not explore the effects on FDI in this research.

12 Table 3: Effects of formal institutions on value added per capita: Inherited institutions vs current social capital. IV regressions

(1) (2) (3) (4) IV First stage IV First stage length -0.000700 -0.000645 (0.000296)** (0.000274)** [0.000195]*** [0.000188]*** school 0.164 -11.20 0.160 -2.246 (0.0319)*** (28.42) (0.0298)*** (28.96) [0.0247]*** [33.93] [0.0251]*** [35.83] urbrate10k 0.100 102.8 0.0785 105.3 (0.0628) (58.99)* (0.0591) (59.73)* [0.0547]* [69.58] [0.0506] [68.31] cartocci -0.000851 -17.09 0.000121 -18.44 (0.00902) (6.930)** (0.00853) (6.859)*** [0.00696] [4.722]*** [0.00654] [4.604]*** factor -69.00 -68.24 (21.13)*** (21.36)*** [19.81)*** [20.09]*** openess 0.0800 -30.59 (0.0351)** (29.37) [0.0308]*** [30.31] Constant 1.429 637.3 1.424 546.1 (0.386)*** (331.7)* (0.359)*** (340.3) [0.325]*** [344.5]* [0.330]*** [377.3]

Observations 102 102 101 101 R-squared 0.854 0.692 0.871 0.696 All specification include regional fixed effect in shape of investment rate, south dummy variable, and organised crime dummy variable. Robust standard errors in parentheses and clustered robust standard errors in squared brackets *** p<0.01, ** p<0.05, * p<0.1

6 Inherited informal institutions vs current civic capital

The cultural traits from the past result in an invalid instrument if they affect income per capita through other channels and through accumulation of civic capital in particular. One possible risk in our causal identification is the omission of current measures of civic capital. It would be the case that civic capital today is capturing the degree of infighting that we are trying to measure using the length of civil trials. Under this circumstance, the effects of duration of legal trials on the value added per capita should evaporate or, at least, substantially shrink. Moreover, the effect of inherited informal institutions on the protection of property rights should disappear or reduce whenever the current civic capital has capturing the inherited cultural traits from the past. We carry out the exercise in Table ?? where a current measure of civic capital is introduced. At provincial level, the series of Cartocci (2007), the most popular in this field, captures under one variable four proxies of civic capital, namely the diffusion of newspaper, the political participation, the blood donations and the membership to sport association. In columns 3 and 4 we add also the usual indicator of openness.

13 The inclusion of current social capital does not change the results either in the IV regression or in the first stage. The length of process are still positive and significant for the level of value added and the inherited cultural traits do not evaporate in the first stage regression. Moreover, the current civic capital has no direct effect on the value added yet it could affect economic outcome via efficiency of formal institutions. If current capital reflected the past cultural traits we would find that our measure were no longer effective on the length of trials. This finding may figure out two sources of civic capital, that is one coming from the past and another accumulated in recent years.

7 Inherited informal institutions vs past formal institu- tions.

It might be possible that formal institutions from a very distant past shaped the informal institutions and that the pattern of formal institutions today are driven by them. In Tabellini (2010) the formal institutions from the past are used to causal identify the cultural traits today. We follow a similar logic introducing in the first stage equation the past formal institutions that prevailed in the pre-unification Kingdoms to check whether our framework is robust to this control. Our measure of formal institutions is the average of the polity score (Polity IV project) achieved by the Italian states between 1848 and the unification. In 1848 the uprising in several pre-unification kingdoms culminated in constitutional reforms of the executive constraints and political participation that with varying degree improved the quality of formal institutions in the pre-Unification kingdoms. Tables ?? compares the effects of both past formal and inherited informal institutions on the length of civil trials and, in turn, on the value added today controlling for schooling, past level of urbanization, current social capital and openess. Let us to note that the coding of past formal institution is highly collinear with one of our fixed effects, i.e. the dummy variable of south regions. Hence we compare the results with and without the dummy south into the specification, maintaining in both cases all the other fixed effects. The results are difficult to interpret. First of all our measure of inherited cultural traits is powerful in explaining the length of process today, i.e. the efficiency of current institutions in protecting property rights. Controlling for past institutions its coefficients is significant and positive whether we include or not a south dummy variable. In turn the effects of the length of trials on the value added per capita are still in place and they are highly significant. Yet the magnitude of the coefficient in the specification without the south dummy is 43% larger than with the south dummy. This could suggest that the specification in columns 3 and 4 are somewhat upward biased. However, the results are clearly driven by the high collinearity of past formal institutions with the south dummy variable because the provinces in the south were under the same set of formal institutions in XIX century with the exception of the Sardinian provinces. When the south dummy is excluded the formal institutions from the past display the expected effect on the efficiency of formal institutions today. However, as far as the robustness of our hypothesis is concerned we do not find any competitive effect of past formal institutions on our cultural traits being possible that their influence on current economic behaviour and outcomes works throughout different channels.

8 Conclusions

Cultural traits and informal institutions affect economic outcome and behaviour yet it is not clear the channel through which they shape economic results. Culture is persistent through generations, although it is malleable and subject to change over time along with development

14 Table 4: Effects of formal institutions on value added per capita: Inherited institutions vs inherited formal institutions. IV regressions

(1) (2) (3) (4) IV First stage IV First stage length -0.000763 -0.00109 (0.000271)*** (0.000208)*** [0.000167]*** [0.000179]*** school 0.161 -9.150 0.161 -13.69 (0.0309)*** (28.84) (0.0372***) (30.31) [0.0267]*** [33.85] [0.0373***] [32.66] urbrate10k 0.0891 74.93 0.102 80.43 (0.0651) (67.37) (0.0801) (68.07) [0.0549] [75.22] [0.0721] [74.81] cartocci -0.00208 -19.94 -0.000512 -29.31 (0.00888) (6.657)*** (0.0108) (6.025)*** [0.00724] [4.892]*** [0.00769] [6.300]*** openess 0.0752** -22.07 0.0772 -31.19 (0.0360)** (30.50) (0.0449)* (29.38) [0.0302]** [27.93] [0.0373]** [26.18] factor -64.47 -82.38 (20.32)*** (19.50)*** [17.94]*** [15.49]*** formalinst1848 -12.43 -19.63 (7.658) (7.912)** [9.220] [9.847]* Constant 1.482 458.2 1.743 308.0 (0.363)*** (346.3) (0.424)*** (381.3) [0.352]*** [407.9] [0.404]*** [452.6] South dummy Y Y N N

J-stat p.value (0.258) (0.235) [0.384] [0.388] Observations 101 101 101 101 R-squared 0.851 0.686 0.769 0.702 All specification include regional fixed effect in shape of investment rate and organised crime dummy variable. Robust standard errors in parentheses and clustered robust standard errors in squared brackets *** p<0.01, ** p<0.05, * p<0.1

15 and accumulation of human and physical capital. However, the inherited culture from the past is an important piece of the current cultural traits and might be responsible for the different reaction of citizens to the same incentives. We have tested, in the context of the Italian provinces, whether cultural traits form a very distant past still play a role in income divergence. Because of its historical events, Italy is a suitable laboratory to carry out this exercise. The different institutional arrangements in the past and the proceedings of history have rooted different cultural traits in the population. Those traits in turns have shaped the attitude towards cooperation, trust, and inclusiveness of which we found systematic and quantitative evidence in the early years of Unification. They have proven to be a good predictor of different efficiency and quality of formal institutions today. Thus, we have established a precise channel of influence of inherited cultural traits and a causality nexus running from past culture to formal institutions today finally to divergence in income per capita. Our model complements the possible and plausible channels of how history influences the current economic outcomes.

16 References

[1] Ak¸comak,I. S., and Ter Weel, B. (2009). Social capital, innovation and growth: Evidence from Europe. European Economic Review, 53(5), 544-567. [2] Algan, Y. and Cahuc, P. (2010). Inherited trust and growth. The American Economic Review, 2060-2092. [3] Banfield, E. C., (1958). The moral basis of a backward society. New York: Free Press. [4] Bigoni, M., Bortolotti, S., Casari, M., Gambetta, D., and Pancotto, F. (2015). Amoral Familism, Social Capital, or Trust? The Behavioral Foundations of the Italian North- South Divide, Economic Journal, forthcoming. [5] Bisin, A., and Verdier, T. (2001). The economics of cultural transmission and the dynamics of preferences. Journal of Economic theory, 97(2), 298-319. [6] Borowiecki, K. J. (2015). Historical origins of cultural supply in Italy. Oxford Economic Papers, forthcoming. [7] Cartocci R., (2007). Mappe del tesoro. Atlante del capitale sociale in Italia, Bologna: Il Mulino. [8] Cipolla C. M. (1969). Literacy and Development in the West (Vol. 35). Harmondsworth: Penguin Books. [9] Di Liberto, A., and Sideri, M. (2015). Past dominations, current institutions and the Italian regional economic performance, European Journal of Political Economy, online version. [10] Felice, E. (2012). Regional convergence in Italy, 1891–2001: testing human and social capital. Cliometrica, 6(3), 267-306. [11] Felice, E. (2013). Perch´eil Sud `erimasto indietro, Il mulino. [12] Guiso, L., Sapienza, P., and Zingales, L. (2004). The Role of Social Capital in Financial Development, The American Economic Review, 94(3), 526-556. [13] Guiso L., Sapienza P., and Zingales L. (2010). Civic capital as the missing link (No. w15845). National Bureau of Economic Research. [14] ISTAT (2010). Conti Economici Territoriali, Version October 2010. [15] ISTAT (2014). Conti Economici Territoriali, Version October 2014. [16] MAIC-Ministero di agricoltura industria e commercio (1875a). Statistica delle societ`adi Mutuo Soccorso, Roma, Regia Tipografia. [17] MAIC-Ministero di agricoltura industria e commercio (1875b). Popolazione classificata per et`a,sesso, ed istruzione elementare, vol. II, Roma, Tipografia Cenniniana. [18] Putnam, R. D., Leonardi, R., and Nanetti, R. Y. (1994). Making democracy work: Civic traditions in modern Italy, Princeton university press. [19] Tabellini, G. (2010). Culture and institutions: economic development in the regions of Europe, Journal of the European Economic Association, 8(4), 677-716. [20] Voigtl¨ander,N., and Voth, H. J. (2012). Persecution perpetuated: the medieval origins of anti-Semitic violence in . Quarterly Journal of Economics, 127(3), 1339-1392.

17