Persistent Effects of Colonial Institutions on Human ... · the areas of Rio de la Plata until its split from the Viceroyalty of Peru in 1776. Our identification ... the viceroy in

1

Persistent Effects of Colonial Institutions on Human Capital Formation and

Long-Run Development: Local Evidence from a Regression Discontinuity

Design in Argentina

PLEASE DON’T CIRCULATE

Rok Spruk and Mitja Kovac1

University of Ljubljana, Faculty of Economics

e-mail: [email protected] , [email protected]

Abstract

We exploit the geographic discontinuity in the integration into the Spanish colonial empire as

a source of variation in human capital formation and long-run development across 527

departments in Argentina. A unique legal institution – the Audiencia Real – that ended more

than 2 centuries ago might be very important in explaining Argentinian regional development

down to the present day. To this end, we measure the department-level distance from the seat

of colonial audiencia in Upper Peru that used to be source of law and colonial institutions for

the areas of Rio de la Plata until its split from the Viceroyalty of Peru in 1776. Our identification

strategy exploits georeferenced spatial boundary splits with local quasi-randomization between

the former areas integrated in the Audiencia jurisdiction and the control areas as a source of

variation in development levels. Our results show that the effect of pre-1776 colonial law and

institutions imposed from the Colonial Audiencia in Upper Peru on the set of human capital

and development outcomes is both strong and remarkably persistent. The evidence suggests

departments outside the former Audiencia jurisdiction have lower rates of illiteracy, more

computer-literate population, better physical and digital infrastructure and more widespread

ownership of computers and cell phones. The established effects do not seem to be driven by

outliers, and remain robust to the battery of specification checks, placebo tests and pass a

number of falsification checks.

JEL classification: C23, C26, C51, K42, O43 Keywords: Audiencia real, colonial courts, human capital formation, empirical law and economics

1 We would like thank Daron Acemoglu, Bernard Black, Matias Cattaneo, Jens Damann and Sven Hoeppner for their precious comments, initiatives and suggestions.

1

1. Introduction and Background The notion that legal institutions matter for growth and development has acclaimed substantial scholarly attention. However, the identification of the effect of legal institutions on economic development using plausibly exogenous sources of variation is less clear and still awaits the conclusive quantification. This paper argues that a historical process, a unique legal institution – Audiencia Real – that ended more than 2 centuries ago might be very important to explain Argentinian regional development down to the present day. Levels of development and human capital vary widely within countries in South America (Dell and Acemoglu 2010, Bruhn and Gallego 2012, Acemoglu et. al. 2014). We argue that part of this variation in Argentina might have roots in colonial era and could be explained with persistent adverse effects of Audiencia Real on human capital formation and long-run development. Audiencia Real was the principal judicial body in Spanish colonial empire. It was the key institution that worked in tandem with the Crown and royal officials and had a fixed territorial jurisdiction. Audiencias were actually a prototype of a super-institution that had a combined powers of government including legislative, executive, judicial and administrative functions. The audiencia had, as commentators note, a general appellate jurisdiction over lesser courts within its borders, and an original jurisdiction for criminal and civil cases within a certain radius of the court (Cutter 1999). Their daily decision making, or the official hours have been three hours in the morning, followed by the afternoon’s decisions. Hence, they have spent 3 hours to hear the case and then they have pronounced their decision on the same day which indicates a very fast and efficient legal system indeed. They have also limited personal jurisdiction over civil disputes involving Indians within certain radius of the court (Soberanes Fernandez, 1980). However, was the audiencia an economic institution that should, while introducing certainty in to the system of sequential exchanges, boost economic activity and spur social wealth and progress? Several streams of institutional explanations of development outcomes have been proposed to account for deep differences in long-run development and human capital investment. First, historical circumstances and events can influence the state of formal institutional framework which tends to survive and, hence, matters for economic interactions and contemporary outcomes (North 1990). The survival of institutional framework can be explained by the resource endowments (Engerman and Sokoloff 1997), the differences in the historically-embedded legal system (La Porta et. al. 1998), persistence of property-rights institutions determined by the initial disease environment (Acemoglu et. al. 2001), access to Atlantic trade (Acemoglu et. al. 2005), the long-run effect of African slave trade (Nunn 2008), medieval trade access favoring institutions promoting religious tolerance (Jha 2013), and by the French invasion of Central Europe brining radical institutional changes and leaving a long-lasting impact (Acemoglu et. al. 2011). We propose the additional channel through which historically-determined institutions impact long-run development: institutional proximity to the sources of powerholding and property-rights institutions. Using the local variation in the set of development outcomes across Argentine departments, we exploit the colonial border discontinuity to consistently examine the contribution of institutions to long-run development. We assume that the distance from the colonial audiencia is a plausibly exogenous source of variation and discontinuity in long-run development that allows us to identify the contribution of institutions to development under a common institutional system. We show that the effect of pre-1776 colonial law and institutions imposed from the Colonial Audiencia in Upper Peru on the set of human capital and development outcomes is remarkably persistent.

2

Most royal audiencias were established across viceroyalties and captaincies general of Spanish America under Recopilación de Leyes do los Reynos de las Indias in 1680.2 As the supreme judicial institution, audiencias were empowered to hear complains against viceroys and executive officers, to take appropriate actions to curb the abuses of power, and to enforce law and order (Kahle 1951, Lockhart and Schwarz 1983, Rodríguez 1994). Audiencias were primarily charged with protecting the rights of Indians against the widespread abuses of power by the early settlers. Their primary function was judicial, having both civil and criminal jurisdiction. Appeals in major cases could be made directly from their decisions to the Council of the Indies in Madrid which oversaw the both viceroys and audiencias. Although primarily designed to take appropriate action against power abuses by viceroys and capitan generals, the latter served as the presiding officers of the Audiencias. Unless the viceroy or captain general was learned in the law, he was excluded from the judicial functions, and had no right to vote (Haring 1952). Three to five judges (oidores) and the presiding officers constituted the Audiencias outside Mexico City. Audiencia in Mexico City was designated as the important one on the Spanish America. From its foundation in 1527 to the 18th century, the number of judges had risen from four to ten. The development of audiencias in Spanish America differed markedly from the role and functions of audiencias in Spain (Elliot 1963, Remacha 1994, Cutter 1998, Dory-Garduño 2013).3 Unlike the peninsular audiencias, the overseas audiencias acted a prototype of super-institution (Mirow 2004). The Spanish Crown delegated legislative, executive and judicial functions to overseas audiencias which de facto represented the king as the lawmaker, dispenser of justice, and as the law enforcement body (Parry 1948). An overseas audiencias could issue local ordinances and served as a privy council to the viceroy or governor-capitain general (Benton 1999). An audiencia also oversaw the royal treasury. Legal historians often refer to it as a junta de hacienda (Del Castillo 1953, Salazar 1983, Mansilla 1985). The crown attorney (fiscal) had the right to correspond directly with the crown on treasury issues and weekly acuerdo decisions (Fisher 1926). Officials vested in the Audiencia were continuously subject to two distinctive forms of review (Coronas Gonzales 1981). After the expiry of the president’s term, the Crown carried out the judgement of the period in office known as juicio de residencia. The judgement reviewed the performance of the president and collected interviews with many stakeholders affected by the audiencia’s decisions (Elliot 1984). The second form of review comprised the unscheduled inspections known as visitas. These were carried out when the Crown perceived this as needed. The introduction of Bourbon reforms in early 18th century introduced additional limits on viceroys and capitain generals (Haring 1952). The particular set

2 The first Audiencia in Spanish America was established in Santo Domingo by 1511 and had a

jursidiction over the Caribbean islands and the adjacent mainland. It was abolished by the Spanish settlers but permanently reestablished in 1526. The first mainland Audiencia was set up in Mexico City in 1527 and had jursidiction over most contemporary Mexico and Central America. In 1538, Audiencia of Panama was established overseeing the Central America and the littoral regions of northern South America but was abolished in 1543. In 1564, the Crown reestablished the Audiencia covering the jurisdiction only over proper Panama. In 1543, two Audiencias were established simultaneously, i.e. one in Guatemala and another one in Lima. By the end of the 16th century, the Crown established six more Audiencias in Guadalajara (Nueva Galicia), Santa Fe de Bogota (Nueva Granda), Charcas (Upper Peri), Quito (Ecuador) and Concepcion (Chile). The latter was short-lived and lasted ten years until the dissolution in 1575. As a part of the administrative reforms introduced under the Bourbons, three additional audiencias were established in Caracas (1786), Cusco (1787) and Buenos Aires (1783) but these were short-lived

3 The first audiencas was founded in Valladolid in the Kingdom of Castile in 1371. It function as the highest court in Castille for two centuries after. After the union of crowns of Castile and Aragon in the Kingdom of Spain and after the fall of Granada in 1492, the Valladolid audiencia was split into two parts. The part north of Tagus River took cases north of the river while Royal Audiencia of Ciudad Real (established in 1494) dealt with cases from south of the river. In 1505, the Audiencia from Ciudad Real was moved to Granada.

3

of reforms created the regent office as a type of chief justice which largely removed the administrative functions from viceroys turning the presidential role into an honorary one. After the Bourbon reforms, the viceroys were charged by law and could correspond with the audiencia only in writing whilst the verbal command were prohibited, creating a record that could be inspected later. In their judicial functions, audiencias heard appeals from cases initially handed by first-instance courts (Lynch 1989). In some cases, the audiencia also served as the court of first instance, most notably for crimes committed in the immediate jurisdiction of the city that used to the seat of audiencias, and for cases involving crown officials (Bella et. al. 1992) The influence of audiencias exhibit a remarkable degree of historical persistence (Veliz 2014). Their importance in handling legal disputes is reflected in the fact that many modern Latin American countries have territorial boundaries that roughly correspond to those of the former audencias. The crown wanted the overseas officials and judges at the audiencias to be immune from the local pressures and pursue strictly judicial duties. Audiencias were heavily pressured by the mine owner and failed to react to the persistence of fraud in the Potosí mines even when the actions by the settler and mine owners led to the rampant abuse of power and widespread mistreatment of Indians and their pressure into forced labor camps such as mita and encomienda. One such notable example of audiencia’s cooptation with local elites comes from the adulteration of the silver used to make coins by adding the excessive amount of copper. (Hanke and Mendoza 1965, Villena 1976, Bakewell 1988). Even though the Spanish crown insisted on pursuing the perpetrators, the viceroy in Audiencia Real in Charcas (Upper Peru) was not to press local interest too hard even though adulteration inflicted a non-negligable damage to Spain’s creditors. Starting with the reign of Charles II, a dire financial situation triggered by the high costs of Spanish wars in Europe, the crown launched the sale of public offices, including the audiencias posts which started in 1687 (Cebrián 1976, Romero 1978). Office sale served as an important vehicle to select colonial governors by the end of the 17th century. The decision to sell local political posts had far-reaching consequences. Although the Crown was seemingly aware of the negative effects of office-seeling, the policy remain in place mostly due to the high likelihood of military defeat in Europe (Burkholder and Chandler 1977). The sale of provincial governorship positions (corregidor) was in high demand. As a check on the abuse of power, the crown paid low wages to the prospective governors but guaranteed future appointments. Although the crown maintained the selection of high-ability individuals of high status and military training, it often undermined its own standards in exchange for high bids which simply encouraged the purchase of the posts by power local elites across the territories of Spanish America even though their qualifications for the positions were dubious. The practice of offering high bids with meager qualifications for the position was widespread among the audiencias members themselves once they quitted the post, creating a widespread conflict of interest (Guarado 2016). The main incentive behind occupying the office was that governors were in charge of tax collection and justice administration. Such massive concentration of power in the hands of the governors invoked substantial leverage among the population to engage in the rent-extracting activities (Tapia 2009). Despite the de jure prohibition at the time office sales, a common and widespread de facto form of rent extraction was repatrimiento, the selling of goods and services by force. Repatrimiento rested on the forced distributions of goods and services to the local population at heavily inflated prices. Repatrimiento kept the local population in heavy and permanent debt, encouraged the exploitation through various forms of servitude, depressed human capital formation and

4

condemned the locals to the near absence of economic opportunities. The agreement allowed the governors and former audiencia appointees to enter the alliance with merchants in Lima and Sevilla to obtain goods and resell them at heavily inflated prices without the contractual consent of the locals. (Cebrián 1977, Andrien 1984, Baskes 2000). By default, the heads of the local household were forced to repay the debt to the repatrimiento since the governor performed the executive and judicial authority at the same time, and also presided the audiencia. Even though the Corregidor was de jure not allowed to vote in the absence of proper legal background, repatrimiento de facto permitted the exercise of unparalleled economic, political and judicial power of the local population captured in various forms of debt peonage and low-wage subsistence labor. At the same time, governors continuously reverted to imprisonment, flogging and torture of debtors, confiscating the goods in their possession if they failed to pay the debt. Golte (1980) estimates that rent-seeking via repatrimiento most likely exceeded the rents obtained from mita system or head taxation (tribute). Even though Spanish crown attempted to limit the power of the corregidores, their pressure on audiencia to annul the compliants against them was powerful and without parallel. Governors continuously engaged in the over-taxation of the population in collecting head tax. Governors oftentimes extorted sizeable amounts above the permitted legal amount even from those exempted, and forced the families to pay for those absent from home due to migration. As a chair of audiencias, the governors often depleted the treasury be diverting the funds designated for public education and hospitals, and using them for their own purposes (Cebrían 1977, de Velasco 1983). In addition, the governors enjoyed substantial benefits from a forced mobilization of labor in their provinces, for instance, by leasing the workers to neighboring haciendas in exchange for two thirds of the wage owed to the workers (Andrien 1984), by providing forced labor to the mita in Potosí mines and delay the implementation of native labor regulation set by the crown in exchange for greater gains from repatrimiento (Mukerjee 2008). Even though the crown granted audiencias the jurisdictions over the power abuses of governors, the oidores were reluctant to deal with such cases since their appointment vested in the relationship between the crown and the governor. The oidores were primarily reluctant to deal with such cases mainly because they were under the threat of intimidation and sanction by the governor or viceroy which implied that complaints against the abuse of powerholding were either dropped, not prosecuted, not treated under the legal principles or simply not enforced even when the complaint was resolved in favor of the defendant. Audiencias were the highest agency in Spanish America with combined judicial, legislative and executive functions. When the crown began to sell appointments to the oidores, the local elites seized the opportunity and began to view their own audiencia district as a homeland (patria) and even claimed that they had a legal right to hold all offices within their territories in spite of the near absence of intellectual, academic and even economic qualifications for the post. Strong creole domination was pertinent to all audiencias where the money transfers and embedded local influence jeopardized the competence of tribunals and their independence. Between 1678 and 1750, the share of creole appointees to the corregidor posts went from 44 percent to more than two thirds with the fraction of appointed govenors from Spain gradually diminishing to negligible levels (Burkholder and Chandler 1977). Such a remarkable shift of power entailed the dilution of royal authority as the ultimate check on the abuses of power by local strongmen, the absence of quality control, wealth creation by the creole elite through repatrimento and other forms of extraction, and the end of the consensus government. Creole oidores were deeply entrenched elites linked by kinship or the interests of economic elites. After the launch of office sales, the audiencia gradually evolved into a preservation of rich and powerful regional families where

5

intimidation, bribery and fraud formed the centerpiece of representation in the government, and laid both the intellectual and institutional seeds of the independence movements against Spain.4 In 1549, Spain created the Governorate of the Río de la Plata (Gobernación del Río de la Plata) as one of the governorates of the five governorates of the Spanish Empire in the Americas. The governorate was the successor territory of the New Andalusia (Nueva Andalucía) granted by Charles V prior to the establishment of the Viceroyalty of Peru.5 It encompassed all the land between 470 and 650 leagues south of the mouth of Río Santiago along the Pacific coast. The governorate fell under the jurisdiction of Audiencia Real in Charcas,6 Upper Peru from 1549 until 1776 when the governorate was dissolved and merged into the Viceroyalty of Río de la Plata.7 The jurisdiction of the Audiencia Real of Charcas encompassed the territories of the governorate but never extended its reach into the areas of southern Argentina since Spain never de facto conquered the territories outside Río de la Plata governorate in the same round as the rest of South and Central America although it de jure claimed these areas. Mirow (2004)

4 Eakin (2007: 154-155) emphasizes the distinctive conflict between the creole elites and the Bourbon reforms: “The Bourbons elevated a number of regions to the status of audiencias and captaincies-general. Regions on the

periphery of the empire … thrived as viceregal centers saw their autonomy diminished. Copying the pattern in 17th

century France, the Spanish Bourbons sent out intendants in the second half of the 18th century as directly

appointed royal authorities to rule over a number of regions and to report back directly to the king.

Understandably, the viceroys fought this move properly. Royal inspect began to make visitas generals (general

visits or reviews), checking up on colonial officials. José de Gálvez, the most famous of these inspectors, served

the king for two decades (1765-1787) and exerted enormous influence in Madrid. Like many other peninsular

Spaniards, he had great disdain for the creoles. By the 1760s, creoles had become the majority (among whites) in

a number of audiencias,… Throughout the previous century and a half, the percentage of creoles in the audiencias

of Spanish America had risen steadily, especially as imperial control relaxed in the seventeenth century.

Throughout the eighteenth century, the Bourbon kings systematically discriminated against the creoles when

making imperial appointments. This pattern was pronounced, enduring and wildly noticed. It would become one

of the principal grievances of the creole elites and a seed that would eventually blossom into open revolt against

Spain.” 5 Adelanto was a title held by Spanish nobles in service of their respective kings. Its origin dates back to

the Middle ages and was used as a military title held by the Spanish conquistadors. In overseas territories, adelanto encompasssed land use grants.

6 Real Audiencia of Chracas (Audiencia y Cancillería Real de la Plata de los Charcas) was established by a royal decree on September 4th 1559 during the reign of Phillip II. Law IX (Audiencia y Chancillería Real de la

Plata, Provincia de los Charcas) of Title XV of Book II of the Recopilacíon de Leyes de Indias of 1680 (which complies the original 1559 decree and three additional ones on August 29th 1563, October 1st 1563, and on May 23rd 1573) defines the borders and functions of the Audiencia: “In Ciudad de la Plata de Nuevo Toledo, Province

of the Charcas in Peru, shall reside another Royal Audiencia and Chancellery of ours, with a president; five judges

of civil cases [oidores], who shall also be judges of criminal cases [alcaldes del crimen]; a crown attorney [fiscal];

a bailiff [alguacil mayor]; a lieutenant of the Gran Chancellor; and the other necessary ministers and officials;

and which shall have for district the Province of the Charcas, all of El Collao, from the town of Ayabiri, along the

road of Urqusuyu, from the town of Asillo by the road of Umasuyu, from Atuncana, by the road of Arequipa,

towards the part of the Charcas, inclusive with the Provinces of Sangabana, Carabaya, Juríes y Diaguitas, Moyos

[see also Moxos people] and Ch'unchu, and Santa Cruz de la Sierra, sharing borders: in the north with the Royal

Audiencia of Lima and provinces not yet discovered; in the south with the Royal Audiencia of Chile; and in the

east and west with the two Seas of the North and South and the line of demarcation between the Crowns of the

Kingdoms of Castile and Portugal, along the Province of Santa Cruz of Brazil. All said territories are and shall

be understood to comply with Law 13, which deals with the founding and erection of the Royal Audiencia of

Trinidad, Port of Buenos Ayres, because our will is that said law be kept, complied with and executed precisely

and punctually.” 7 For the period 1661-1671, the governorate fell briefly fell underly the newly created Audiencia Real of Buenos Aires which had been short-lived and had little factual influence on the path of governorate's institutional development compared to the influence of Audiencia Real in Charcas which ruled the territory for more than two centuries.

6

For more than two centuries, Audiencia Real was the most important institution and the principal judicial body with combined legislative, executive and judicial power over the conquered territories integrated into the Spanish colonial empire. It was the key institution that worked in tandem with the royal official and had a fixed territorial jurisdiction. Legal historians agree that colonial audiencias in the Spanish empire were modeled closely on the most powerful Castillian tribunal of that period, Audiencia y Real Cahncillería de Valladolid. Similar to other audiencias in the Spanish empire, Audiencia Real in Charcas had a general appellate jurisdiction over lesser courts within its borders, and an original jurisdiction for civil and criminal cases (Taylor 2006). Audiencias were almost entirely judicial bodies serving as a court with a handful of judges (oidores). The judges often had multiple ancillary duties such as tax collection known as cruzada, serving as a probate judge, fleet inspection and serving as a judge for commercial appeals from commercial tribunals known as consulados. Crown legislation set out the ideal structure of the court but in audiencias were de facto understaffed. The enforcement of civil and procedural law over the jurisdiction held only within a certain radius of audiencia’s court and declined with greater distance from the seat of audiencia (Mirow 2004). Therefore, the ability of audiencia to enforce the legal institutions across its territory critically depended on the proximity of the intendant areas to its seat. The institutions established by the audiencias during the colonial period influenced the paths of institutional development in former Spanish America to a highly persistent degree (Miller 2003, Mirow 2004) even after it gained independence from Spain. Several sources of audiencia’s institutional persistence can be detected. First, the oiodores and the creole elite acquired a substantial de jure and de facto political power (Arroyo Abad 2016) The former is determined by the constitutions and electoral systems while the latter emerges from the ability to undertake collective action, use of brute force, bribery or lobbying. Acemoglu and Robinson (2006) propose a mechanism where equilibrium changes in the distribution of de jure political power offset the changes in the distribution of de facto political power. In such mechanism, the elites losing the de jure political power may still have a disproportionate influence in politics by expanding the intensity of collective action through lobbying, bribery, intimidation or brute force to ensure the persistence of the cluster of economic institutions yielding rents the social group set to lose the de jure political power. In equilibrium, the economic institutions are the net effect of the de jure and de facto sources of political power which implies that the established set of institutional arrangements will persist even in the wake of changes and shocks altering specific sets of institutions (Binger and Hoffman 1989). Extensive empirical evidence supporting the institutional persistence hypothesis can be found in Acemoglu et. al. (2001, 2002), Banerjee and Iyer (2005), Dell (2008) and Becker et. al. (2016) among others. Second, viceregal power was characterized by a certain amount of independence from the crown control because of the distance from Spain and difficult communication (Grafe and Irigoin 2006, 2012). Viceroys and oidores were known for applying orders with full discretion using the maxium “I obey but I don’t comply. [obedezco pero no cumplo]” The audiencias were captured by the viceroys since the term and appointment of oidores depended on the approval of the viceroy while the oiodores reported directly to the crown. Such an interplay generated persistent conflict most of the time, and the viceroys engaged in downright bribery or intimidation of the oidores to report favorably to the crown (Cañeque 2013). The expansion of colonial administrative apparatus and bureaucracy interplayed with the economic organization. As a part of viceroyalty, the audiencias were further subdivided into provinces or districts (corregimientos) and into municipalities (municipios). The latter were governed by the town council (cabildos) consisting of the prominent citizens (veccinos) mostly encomenderos and later hacienda owners (hacendados). The town council had a disproportionate influence on the recruitment of elite members onto prominent positions, such as audiencia, which critically contributed to the

7

disproportionate rise of the creole members to the oidores positions in the aftermath of Bourbon reforms (Phelan 1960, Irigoin and Grafe 2008) After the Bourbon reforms in Latin America, audiencia positions were occupied by the members of the creole elites. Economic historians agree that Latin America after the Bourbon reforms was a highly unequal society stratified bv race and wealth (Williamson 1999, Bulmer-Thomas 2003, Coatsworth 2005, Prados de la Escosura 2009, Milanovic et. al. 2011, Arroyo Abad et. al. 2012, Arroyo Abad and Van Zanden 2016). Each group had specific rights or privileges (fueros). Membership in the upper clases was open to whites only, particularly peninsulares born in Spain. Creoles tended to marry peninsulares for reasons for upward social mobility. As a highly unequal society, the lower classes were a mix of poor whites and the natives. They were protected from the Inquisition, paid head taxes, and could not own the property (Cutter 1986). The poor whites could not be considered for the jobs in the colonial administration. Even though the crown abolished encomienda by the end of 16th century, its role was replaced by haciendas were the lower classes performed labor duties to the hacienda owners in a similar but less restrictive fashion. The collapse of silver production by the mid-17th century led to widespread bankruptcies among the miners and hacienda owners which, as a response, extracted further tribute from the laboring class. (Campbell 1972). Repatrimiento further stifled the economic opportunites for the vast class of non-elites and condemned them to persistent poverty and underdevelopment compared to the wealthy creole elites (Rosenmüller 2016) The notion that economic institutions cannot be studied in isolation from the political institutions has become widely acclaimed (Haber 2013, Mariscal and Sokoloff 2000, Engerman and Sokoloff 2005). The economic institutions in Spanish America were primarily designed to transfer rents first to the Spaniards and then to the members of the creole elite (Lange et. al. 2006). The independence from Spain across the Spanish empire was essentially carried out as a coup d’etat by the creole elite to isolate Mexico from the current of political and economic liberalization in Spain following the 1812 Constitution of Cadiz (Coatsworth 2008, Engerman and Sokoloff 2000, Mirow 2014). Hence, even though the rent-extracting institutions such as encomienda and mita disappear by the end of the 16th century and had adverse effects of long-run development (Dell 2008), they were replaced by haciendas. In spite of the post-independence constitutions, the character of economic institutions established under the colonial system never disappeared from the surface of Latin America’s institutional development. High income and wealth inequality in colonial Latin America simply implied that affluent creole elite increased its de facto political power and enabled it to push for the set of economic and political institutional arrangements favorable to its interests which deeply influenced the paths of growth and development after the independence (Andrien 1984) We test the institutional persistence hypothesis for the effects of the Audiencia Real as a source of colonial law on long-run development across 527 departments in Argentina. To this end, we exploit the discontinuity in the integration in the Spanish colonial empire as a measure of the as a source of variation in human capital formation and long-run development. We assign the observations into treatment and control samples where the former comprises the areas under the jurisdiction of Audiencia whilst the latter comprises the departments outside the colonial jurisdiction in the south of Argentina. We examine the effect of the colonial law on various measures of human capital formation and long-run development using the department-level distance from the colonial court in Charcas that held civil jurisdiction over the areas of Río de la Plata until its split from the Viceroyalty of Peru in 1776. Using a geo-referenced spatial dataset along the lines proposed by Keele and Titiunik (2014), we reconstruct the former

8

colonial borders into the modern department-level administrative division of Argentina and split the units into treatment and control areas. We argue that part of the extensive differences in human capital formation and development within Argentina is explained by the persistently negative effect of economic, legal and political institutions established by the Audiencia such as repatrimento, head taxation, tribute, office selling, intimidation, prohibition of property ownership, de facto disenfranchsiment through literacy and wealth-related voting qualifications and other forms of institutions that prevented the access to economic opportunities for the non-elites. Employing the regression discontinuity design with local quasi-randomnization, our evidence confirms the persistent adverse effects of the exposure to colonial law on human capital formation and long-run development. The evidence based on our preferred specification suggests that the departments historically under the jurisdiction of Audiencia Real tend to have 39.6 percent higher rate of illiteracy and 24 percent lower rate of computer literacy and 8 percentage points higher rate of landownership. The households in the departments outside the colonial Audiencia jurisdiction are 20 percentage points more likely to have sanitary facilities and water discharge, 12 percentage points less likely to lack water supply inside the house, 68 percentage points more likely to possess a fixed-telephone line, 4 percentage points more likely to own a cell phone, and 24 percentage points more likely to own a computer. Our results are robust to a battery of specifications checks, pass a number of placebo checks and do not appear to be driven by the alternative mechanisms such as the climate and disease environment. The key implication of our research is that historically established institutions tend to persist into the future well after they have been abolished long ago, and tend to exert a highly persistent and robust effect on human capital formation and development. Compared to the earlier evidence which suggests the beneficial effects of proximity to various layers of institutions on growth and development (North and Weingast 1989, Schäfer and Wulf 2014), we show that in the Latin American institutional setup, the proximity to the legal institutions determined at the onset of colonization tends to disfavor development outcomes and condemns to areas under the influence of colonial institutions imposed by Audiencia Real to underdevelopment in comparison to the departments in the south of Argentina which were not under the jurisdiction of the colonial officials in Upper Peru. The results imply confirm the evidence from a growing strand of “history matters” literature and indicate that, once established, the effect of institutional framework on long-run development appears to be remarkably persistent. The rest of the paper is organized as follows. In Section 2, we conceptualize the institutional persistent, discuss the Spanish colonial law, its judicial system and the institutional legacy across Spanish America with special reference to Argentina. Section 3 presents the data and the underlying methods. In Section 4, we present the results and to robustness checks. Section 5 concludes.

2. Institutional Persistence, Human Capital, and Long-Run Development

2.1 Persistent Effects of Historical Institutions

The notion that institutions critically matter for long-run development can hardly be disputed. Societies without the institutions promoting capital accumulation, human capital investment, strong property rights and low transaction costs are condemned to slow economic growth and long-run underdevelopment. North (1991) originally defined the set of institutional choices as “humanly devised constraints that structure political, economic and social interaction.” They consist of formal institutions (laws, constitutions, property rights, electoral systems), and informal institutions (customs, norms, reputation, traditions, codes of conduct, ethics, religion), both of which have been designed to establish law and order, and reduce

9

uncertainty in economic exchange (Huntington 1968, Greif 1989, Hodgson 2006, Ostrom 2005, Fukuyama 2015). Together with the basic framework of neoclassical economics, institutions shape economic choices, constrain economic actors (Acemoglu et. al. 2005b), directly affect production and transaction costs (Coase 1984), and hence determine the feasibility of engaging in productive and unproductive economic activities (Williamson and Masten 1995) and business opportunities (David 1994, Hall and Soskice 2001). Institutions establish incentives and thus shape the direction of economic change. Formal institutions can be modified at will (Williamson 2000) whereas informal institutions tend to persist for a long time and change only slowly, but not impossibly (Boettke et. al. 2004, Roland 2004). The historical precedence of institutions has been established by North (1989), North and Weingast (1989), and Mokyr (1990), and confirmed empirically by Hall and Jones (1999), Easterly and Levine (2003), Acemoglu et. al. (2001, 2002, 2011) and Van Zanden et. al. (2012) among many others. Historical and contemporary institutional choices tend to persist, and their legacy does not seem to disappear (Banerjee and Iyer 2010) which implies that the institutional environment matters a great deal for the long-run growth and development paths across and within countries.

Although the notion that institutions exhibit persistent effects on contemporary economic outcomes can seldom be disputed, (Acemoglu and Robinson 2008, Dell 2010, Putterman and Weil 2010, Nunn 2009) it remains less clear whether historically-determined institutions exhibit equally persistent effects on human capital formation as a proximate cause of economic growth and development. Existing studies have shown that the institutions are pivotal in creating the environment promoting human capital accumulation although the exact persistence channel is less clear. Some scholars suggest the inequality of landownership adversely affected the emergence of human capital-promoting institutions, and thus the pace and nature of transition from the agricultural to industrialized economy (Galor, Moav and Vollrath 2009). In contrast, some authors explicitly reject the notion of the importance of geography in shaping human capital formation Clark and Gray (2014), drawing on a micro-level data from England for the period 1810-1845, argue that geography-driven inequality fails to predict regional literacy rates which seem to have been influenced more by culture. In particular, the areas in the north of England with more exposure to highly literate Scottish society seem to have acquired a higher demand for education, independent of local inequality compared to the south of England where farmer elite dominated the society with many landless laborers. In addition, many scholars argue that culture is the decisive factor in human capital investment and long-run development, suggesting that “culture makes all the difference” (Landes 1998), and that culture played the fundamental role in economic development (Guiso et. al. 2006, Becker and Woessmann 2009, Tabellini 2010, Gorodnichenko and Roland 2011, Maseland 2013) since in the form of preferences, intergenerational transmission of norms and behavioral routes holds long-lasting implications for the economy. (Putnam 1993, Alesina and Fuchs-Schündeln 2007, Algan and Cahuc 2010, Cantoni 2015, Belloc and Bowles 2009, Spolaore and Wacziarg 2013, Chanda et. al. 2014). Some scholars are convinced that time-invariant geographical and ecological factors such as climatic zones, malaria prevalence, tropical agriculture, acid rainfall or the distance from international markets may have large and direct effects on the current economic development (Diamond 1997, Gallup et. al. 1999).

In contrast, Engerman and Sokoloff (2002, 2012) argue that the geography matters for institutions and human capital through the distribution of resource endowments which determines the degree of economic specialization. Drawing on the economic history of North and Latin America, the areas that were best suited for crops (sugar, coffee, cotton, bananas) developed large-scale plantation economies based on the exploitation of indigenous labor and the importation of African slave labor which led to resource-extractive institutions (Acemoglu and Robinson 2013), extremely low human capital and adversely affected Latin America’s

10

long-run development. In North America, the resource stock, climate and initial endowments favored the production of grain and dairy products on small-scale family farms. Low initial inequality between the settler elite and the rest of society encouraged a high demand for education, a large political class in favor of public education and led to high rates of literacy benefitting massive human capital accumulation, and long-run development. By contrast, Latin American political elites had almost no interest in the public provision of education because it did not serve their economic interest and undermined the political dominance. A huge class of unskilled workers was unable to afford education for their children through the exclusionary laws on public education or unaffordable private market. Such institutional choices created unequal, uneducated, and ultimately poor society in a world where human capital is the engine of growth. The inequality of political power across Latin America thus led to economic underdevelopment as the political and economic elites maintained growth-impeding institutional structure for their own gain at the expense of society at large not realizing its full human capital and growth potential. Engerman-Sokoloff hypothesis has been confirmed empirically Easterly and Levine (2003), Mitchener and McLean (2003), Easterly (2007) and Nunn (2008) although the results might be driven by Africa and Latin America (Islam and Montenegro 2002).

Not everyone agrees that high levels of inequality and endowment structure caused a lower investment in education and under-development. The evidence from the colonial São Paolo presented in Summerhill (2010) readily suggests the provision of education and public education outlays were not affected by historical and contemporary political and economic inequality. If anything, the expenditure and provision of education was higher in areas supposedly subjected to high wealth inequality and the extractive institutional arrangements such as colonial aldeamento. The argument suggests that immigrants to São Paolo often provided more private education and pushed for greater provisions publicly. Grafe and Irigoin (2012) challenge the traditional view of Spain as a predatory colonial state that extracted revenue from natural resources while offering little in return, and suggest Spain adopted the stakeholder control of revenue and expenditure, and thus ensured that much of the colonial revenue looped into the local economy with minimal enforcement cost. Arroyo-Abad and Van Zanden (2016), based on long-term estimations of GDP per capita for colonial Mexico and Peru in the period 1550-1780, refute the notion that colonial institutions impoverished Latin America. In contrast, they find vibrant economies characterized by rising wage rates, increasing urbanization and dynamic silver mining industry. In addition, endogenous adaptions of institutions moved towards increased market openness and a more balanced distribution of power between Spain and colonial elites which questions the notion of extractive institutions condemning Latin America to medicore growth and underdevelopment. On the other hand, Bruhn and Gallego (2012) exploit the variation in the type of colonial economic activities within countries in the Americas, and show that part of the large variation in income and development levels has roots in the colonial era where colonizers engaged in different economic activities. The activities based on labor exploitation led to underdevelopment compared to the areas not relying on labor exploitation. Differences in political representation, rather than the differences in human capital stock or inequality, appear to be the mediating channel in the relationship between colonial activities and current development. Rajan and Zingales (2006) show that the persistence of underdevelopment is not driven by bad institutions but by the existence of self-perpetuating constituencies demanding such institutions. Glaeser et. al. (2004) criticize the “institutions rule” view and suggest that most formerly poor countries achieved sustained economic growth through human capital investment, and often under authoritarian regimes supposedly blamed for bad institutions. Banerjee et. al. (2005) advocate the persistent effects of institutions and policy choices. They examine the colonial land tenure setup during the

11

British rule in India and demonstrate empirically that differences in historical property-rights institutions led to sustained differences in contemporary economic development outcomes. In particular, areas where the land property rights were historically given to landlords exhibit lower agricultural investment and productivity in the postwar period, and less investment in public health and education compared to the areas where the rights were given to cultivators.

2.2 History, Political Power and Institutional Persistence

The given set of institutional arrangements can endure even in the face of shocks and changes in the specific property-rights and contracting institutions. Acemoglu and Robinson (2006) construct a model where the equilibrium changes in the political institutions alter the de jure political power but may not offset the changes in the distribution of de facto political power. When the elites who monopolize the de jure political power lose particular kinds of privileges, they may still exert a disproportionate influence in politics by increasing their intensity of collective action such as bribery, lobbying or brute force, and thus ensure the continuation of the previous set of economic institutions. Economic and legal institutional arrangements might also persist due to (i) the interaction of social conventions and norms at the local level which eventually leads to persistence (Young 1998), (ii) specific investments by economic agents in activities where its value might be destroyed by changes in institutional arrangement (Coate and Morris 1999), or due to (iii) network externalities where a change in any activity may require a large degree of coordination which occurs only infrequently (Arthur 1989). Such circumstances imply that institutional arrangements, economic or policy outcomes might be invariant to the equilibrium changes in political institutions and the economic and legal institutions themselves tend to persist over time.

The degree to which the institutional choices from the past persist to the present is striking. Dell (2010) employed a regression discontinuity design (RDD) to study the long-run impacts of the forced labor mining system (mita) in Peru and Bolivia between 1573 and 1812 imposed by the Spanish colonizers, and found that households in mita-subjected districts face 25 percent drop in household consumption, and an increase in the prevalence of stunted growth of children by 6 percentage points. The adverse effects of mita on human capital formation and long-run development persisted through its impact on land tenure and public goods provision. Historically mita districts had fewer large landowners and lower educational attainment. Today, they face a markedly lower degree of infrastructural development and their residents are much more likely to be subsistence-based farmers. Drawing on the evidence from Habsburg Empire, Becker et. al. (2016) hypothesize that sharing common formal institutional environment should exhibit a beneficial effect on current institutional outcomes. Employing the two-dimensional geographic RDD with border specification, they find that the historical exposure to Habsburg institutions among individuals across seventeen Central and Eastern European countries tends to increase the contemporary trust in the bureaucracy, and tends to reduce the judicial and police corruption which advocates the long-last impact of historical institutional framework on contemporary trust and institutional outcomes.

Several streams of institutional explanations of development outcomes have been proposed to account for deep differences in long-run development and human capital investment. First, historical circumstances and events can influence the state of formal institutional framework which tends to survive and, hence, matters for economic interactions and contemporary outcomes (North 1990). The survival of institutional framework can be explained by the resource endowments (Engerman and Sokoloff 1997), the differences in the historically-embedded legal system (La Porta et. al. 1998), persistence of property-rights institutions determined by the initial disease environment (Acemoglu et. al. 2001), access to

12

Atlantic trade (Acemoglu et. al. 2005a), the long-run effect of African slave trade (Nunn 2008), medieval trade access favoring institutions promoting religious tolerance (Jha 2013), and by the French invasion of Central Europe brining radical institutional changes and leaving a long-lasting impact (Acemoglu et. al. 2011). We propose the additional channel through which historically-determined institutions impact long-run development: institutional proximity to the sources of powerholding and property-rights institutions. Using the local variation in the set of development outcomes across Argentine departments, we exploit the colonial border discontinuity to consistently examine the contribution of institutions to long-run development. To this end, we measure the department-level distance from the seat of colonial audiencia in Upper Peru that used to be source of law and colonial institutions for the areas of Rio de la Plata until its split from the Viceroyalty of Peru in 1776. We assume that the distance from the colonial audiencia is a plausibly exogenous source of variation and discontinuity in long-run development that allows us to identify the contribution of institutions to development under a common institutional system. We show that the effect of pre-1776 colonial law and institutions imposed from the Colonial Audiencia in Upper Peru on the set of human capital and development outcomes is remarkably persistent. Employing a regression discontinuity design with local randomization, our evidence suggests departments in Argentina not initially integrated in the Spanish colonial empire and subject to the policies and institutions of Colonial Audiencia from its foundation have higher rates of illiteracy, less computer-literate population, better physical and digital infrastructure and more widespread ownership of computers and cell phones. The established effects do not seem to be driven by outliers, remain robust to the battery of specification checks, placebo tests and pass a number of falsification checks. Compared to the earlier evidence which suggests the beneficial effects of proximity to various layers of institutions on growth and development (North and Weingast 1989, Schäfer and Wulf 2014), we show that in the Latin American institutional setup, the proximity to the legal institutions determined at the onset of colonization tends to disfavor development outcomes and condemns to areas under the influence of colonial institutions imposed by Audiencia Real to underdevelopment in comparison to the departments in the south of Argentina which were not under the jurisdiction of the colonial officials in Upper Peru. The results imply confirm the evidence from a growing strand of “history matters” literature and indicate that, once established, the effect of institutional framework on long-run development appears to be remarkably persistent.

3. Data and Methods

3.1 Basic OLS Regression We tackle the department-level relationship between the Spanish colonial law, human capital and long-run development by exploiting the variation in the distance of each department from the seat of the colonial audiencia and estimating the OLS model specification for the cross-section of departments:

, , 0 1 , , 1 , ,1 1

J KSucre

i j k j j k k k i j k i i i j k

j k

H I I D Gδ ϕ θ λ ω β ε∈= =

= + ⋅ + ⋅ + ⋅ + ⋅ + +∑ ∑ 'X

(3.1)

where H is the measure of human capital formation or long-run development for i-th department in j-th province from k-th region, 0δ is the constant term,

j kI ∈ is the indicator

variable for j-th province which captures the unobserved province-level heterogeneity bias captured by the set of province-specific effect

jϕ ,

kI is the indicator variable for k-th region

13

which captures the set of unobserved region-specific effects, indicated by kθ

. The variable SucreD denotes the geographic distance of i-th department from the seat of the colonial audiencia

in Sucre, i

G is a dummy variable indicating whether i-th department has been integrated in the

Spanish colonial empire, and X is the vector of the department level human capital and long-run development covariates such as population density, fraction of foreign-born population, population size, and the distance from Buenos Aires. The stochastic disturbances are captured by ε . The key coefficient of interest is 1λ

which represents the baseline effect of the institutional

legacy of Spanish colonial law, captured by the proximity to the seat of colonial audiencia, on the set of human capital and long-run development outcomes. A valid inference on 1λ

can be

undermined if the stochastic disturbances in Eq. (3.1) are correlated across departments, provinces and regions which renders the estimated parameter inconsistent. The failure to control for within-cluster residual correlation at multiple levels can lead to substantially underestimated standard errors and over-rejection of the null hypothesis triggering Moulton bias (Moulton 1986, 1990). Serially correlated stochastic disturbances within the single cluster do not disappear with the inclusion of province-fixed effects of region fixed-effects (Davis 2002, Pepper 2002, Bertrand et. al. 2004, Kezdi 2004). Controlling for the province-level and region-level unobserved effects, a non-nested multiway clustering scheme (Cameron et. al. 2011) used to tackle the within-cluster serially correlated stochastic disturbances simultaneously at province- and regional level based on multiway error component model under i.i.d residual distribution assumption. Compared to the traditional one-way cluster schemes, such as the robust OLS variance matrix estimator (Huber 1967, Eicker 1967, White 1980), the non-nest multiway clustering estimators facilitates the valid inference in the presence of within-cluster residual correlation at multiple levels of panel dimension (White 1984, Pfefferman and Nathan 1981, Liang and Zeger 1986, Arellano 1987, Wooldridge 2002, Cameron and Trivedi 2005, Hansen 2007). The non-nested multiway clustering scheme on the underlying empirical distribution function ensures that the model parameters are robust against arbitrary heteroscedasticity and serially correlated stochastic disturbances at department-, province-, and regional dimension of our panel.

3.3 Regression Discontinuity Design The underlying OLS relationship in Eq. (3.1) allows us to establish the aggregate correlation between the legacy of Spanish colonial law, and the set of human capital and long-run development outcomes but fails to provide the evidence of causality. The established OLS relationship merely provide evidence of correlation but, at the same, it fails to identify the causal effect of the Spanish colonial law on human capital and long-run development. In a similar vein, the IV techniques might also be conceptually unsuitable to identify the causal effect of Spanish colonial law as a measure of institutions, on human capital and long-run development. As emphasized by Helland and Klick (2011) and Klick (2010), the estimated OLS and IV relationship might be contaminated by omitted variable bias. The former is fundamentally unsuitable for making causal inferences on the effect of the set of historical institutions on various outcomes while the latter relies on the natural experiment where it is highly unlikely that the underlying IV is orthogonal to the set of outcomes. Since the institutions and institutional legacy tend to change very slowly over time, any fundamental changes in institutions as a result of external shocks render the isolation of the effects of institutions from other factors at play almost impossible.

14

We attempt to overcome the particular identification dilemma, and estimate the causal effect of institutions on human capital and long-run development by constructing a regression discontinuity design for our sample of departments by exploiting the variation in the distance from the seat of colonial audiencia, which held the jurisdiction over the Argentine territory until 1776. The departments are assigned to the treatment group if they had been integrated into the Spanish colonial empire, and to the control group if they were not under the administrative and territorial control of the Spanish Crown. Such a setup allows us split the cross-section of departments into treatment and control group, to observe the behavior of human capital and long-run development outcomes in the formerly colonized departments and non-colonized departments and provide some further evidence on the causal effect of institutions on human capital and long-run development.

3.2 Assignment Mechanism Suppose the influence of the Spanish colonial law is a deterministic, spatial and discontinuous function of the geographic distance from the seat of colonial audiencia in Sucre, and define the treatment status:

, , 0

, , 0

1 if

0 if

Charcas

i j kTreatment

i Charcas

i j k

D DG

D D

∈=

∉

(3.2)

where 0D is the normalized colonial border cutoff variable separating the treatment and control

group. Specifically, the departments integrated into the colonial empire are characterized by 1Treatment

iG = which implies that the distance from the colonial real audiencia in Characas is a

discontinuous function of the cutoff. In addition, 0Treatment

iG = implies that i-th department was

not integrated into the Spanish colonial empire given its distance from colonial audiencia which ensures that the department falls outside the treatment cutoff. For i-th department, we compute its distance from the seat of Audiencia Real by using the high floating-point precision for latitude and longitude map points to mitigate the rounding errors for small distances and the following central angle between from the spherical law of cosines:

( )( ), , , , , ,arccos sin sin cos cos cosSucre Sucre Sucre

i j k i j k i j kσ∆ = Θ ⋅ Θ + Θ ⋅ Θ ⋅ Ω − Ω

where , ,i j kΘ is the i-th department’s radian latitude and i j k∈ ∈ and SucreΘ is the latitude of

Sucre in present-day Bolivia where the official seat of Audiencia Real was established during the Spanish colonization for the area of Upper Peru, , ,i j kΩ is i-th department radian longitude,

and SucreΩ is the longitude of Sucre. The distance between i-th department and the Audiencia Real is given by the arc length which is computed from the Earth’s surface radius and the central angle expressed in radians:

, ,Sucre

i j kD R σ= ⋅∆

where R denotes the earth’s radius. The latitude and longitude coordinates in radians are converted to simple degrees using the following formula:

15

( ) ( )

( ) ( ) ( )

, , , ,

, , , ,

arccos rad 90 cos rad 90

sin rad 90 sin rad 90 cos rad

Sucre Sucre

i j k i j k

Sucre Sucre

i j k i j k

D R = ⋅ ° − Θ ⋅ ° − Θ

+ ° − Θ ⋅ ° − Θ ⋅ Ω − Ω

which allows us to match the latitude/longitude cells in the standard degree notation to the cells on the map and thus compute the underlying distance between i-th department and Sucre. We collect the data on the latitude and longitude coordinates of each department using the standard GPS Coordinates from Google Earth and construct the underlying distance therefrom.

3.3 Regression Discontinuity Setup

Since the potential outcomes between treatment and control group are best described by the linear model with constant effects with the underlying relationship:

, , 1 , , 1 , , , ,1 1

ˆJ K

Sucre Treatment Sucre

i j k i j k i j j k k k i j k i i j k

j k

H D G I I Dλ π ϕ θ λ β ε∈= =

= ⋅ + ⋅ + ⋅ + ⋅ + ⋅ + +∑ ∑ 'X

(3.3)

where π indicates the causal effect of the Spanish colonial law on human capital and long-run development outcomes. In comparison to the OLS model in Eq. (3.1), Treatment

iG is not only

correlated with SucreD but is also its deterministic function which allows us to rule out the possibility that the relationship between the distance from the Audiencia Real is a non-linear function potentially mistaken for the discontinuity. Controlling for the non-linearities in the distance from Audiencia Real with respect to the set of human capital and long-run development outcomes at both sides of the border cutoff, we construct the fundamental flexible RD setup that takes place:

( ), , , , , , , ,1 1

J KSucre

i j k i j k i j k j k i i j k

j k

H f D G I Iπ ϕ θ β ε= =

= + ⋅ + ⋅ + ⋅ + +∑ ∑ 'X

(3.4)

where ( ), ,Sucre

i j kf D denotes the polynomial set of non-linear effects of the geographic distance

from Audiencia Real at both sides of the border cutoff, where , , 01Treatment Charcas

i i j kG D D = ∈ is

discontinunous in , ,Charcas

i j kD at , ,Charcas

i j kD , and continuous around the neighborhood of 0D .The

flexible RD setup in Eq. (3.4) specifically allows us to net out the causal effect and identify the true contribution of Spanish colonial law and its long-term persistence by comparing the distribution of human capital and long-run development outcomes at both sides of the border cutoff where π

indicates the potential causal effect. Such effect might be undermined by the

different trends in the set of outcome variables at both sides of the border. Our empirical strategy attempts to mitigate such concerns and the flexible RD setup by expanding the baseline specification in Eq. (3.4) with differential trend functions for the

expected outcomes between treatment and control samples, denoted by 1 , , , ,| Sucre

i j k i j kE H D and

0 , , , ,| Sucre

i j k i j kE H D which leads to the following re-specified core flexible RD setup:

16

( ) ( ) ( )0 , , , , 0 , , 0 01 , , 02 0, , , ,

, ,1 1

| 2 ...Sucre SucreSucre Sucre Sucre

i j k i j k i j k i j k pi j k i j k

J K

j k i i j k

j k

E H D f D D D D p

I I

δ β β β

ϕ θ β ε= =

= = + ⋅ + ⋅ + +

+ ⋅ + ⋅ + +∑ ∑ 'X

ɶ ɶ ɶ

(3.5)

( ) ( ) ( )1 , , , , 1 , , 0 11 , , 12 1, , , ,

, ,1 1

| 2 ...Sucre SucreSucre Sucre Sucre

i j k i j k i j k i j k pi j k i j k

J K

j k i i j k

j k

E H D f D D D D p

I I

δ π β β β

ϕ θ β ε= =

= = + + ⋅ + ⋅ + +

+ ⋅ + ⋅ + +∑ ∑ 'X

ɶ ɶ ɶ

(3.6)

where the distance from the Audiencia Real in the assignment mechanism is centered around

the border cutoff, , , 0 , ,Sucre

i j k i j kD D D≡ −ɶ . Such normalization critically ensures that the effect of

Spanish colonial law in the treatment jurisdiction is the coefficient of the treatment variable with interaction effects. Centering the geographic distance from the Audiencia Real also allows us to compare the distribution of outcomes at both sides of the border more in-depth in the presence of potentially non-linear relationship between the distance from audiencia and the set

of outcomes. Using the fact that Treatment

iG is a deterministic function of , ,Sucre

i j kD , the causal effect

of Spanish colonial law on human capital and long-run development can be estimated from the following specification:

( ), , , , 0 , , , , , , 1 , , , , 0 , , , ,| | | |Sucre Sucre Sucre Sucre

i j k i j k i j k i j k i j k i j k i j k i j k i j kE H D E H D G E H D E H D = + ⋅ − (3.7)

Once the non-linearities are allowed, we construct a more parsimonious specification by substituting the polynomial terms for the set of conditional expectations which leads to the general variant of the model used to establish the true effect of Spanish colonial law on the set of outcome variables:

( ) ( )

( ) ( )

*, , 0 01 , , 02 0 1 , ,, , , ,

* *2 , , , ,

1 1

2 ...

2 ...

Sucre SucreSucre Sucre

i j k i j k p i i i j ki j k i j k

J KSucre Sucre

i p i j k i ii j k i j kj k

H D D D p G G D

G D G D p I I

δ β β β η β

β β ϕ θ β ε= =

= + ⋅ + + + + ⋅ + ⋅ ⋅ +

⋅ ⋅ + + ⋅ + ⋅ + ⋅ + +∑ ∑ 'X

ɶ ɶ ɶ ɶ

ɶ ɶ

(3.8)

where *1 11 01β β β= − , *

2 22 02β β β= − , and *1 0p p pβ β β= − . The validity of RD estimates in Eq.

(3.8) critically depend on whether the p-th order polynomial form of the assignment variable provide a well-behaved and adequate representation of the outcome in the control sample to prevent the non-linearity in the counterfactual conditional mean function from taking place instead of the true discontinuity used to determine the effect of colonial law on the set of outcomes. Our strategy attempts to mitigate the likelihood of such mistakes and address the validity of the estimates and from the flexible RD setup by looking only at assignment mechanism around the cutoff to balance the full sample for some small but positive number

denoted by ∆, which leads to ( ) ( ), , 0 , , 0 0 , , , , 0| |Sucre Sucre

i j k i j k i j k i j kE H D D D E H D D− ∆ < < =≃ and

( ) ( ), , 0 , , 0 1 , , , , 0| |Sucre Sucre

i j k i j k i j k i j kE H D D D E H D D≤ < + ∆ =≃ .

Comparing the average outcomes between treatment and control samples in a small neighborhood left and right of the border threshold 0D is thus equivalent to estimating the true

treatment effect in the presence of model specification bias. We use the locally weighted linear

17

regression model at both sides of the cutoff to correct for the specification bias (Cleveland 1979) that can potentially contaminate the effect of Spanish colonial law on human capital and long-run development outcomes across the departments in treatment and control samples and use Calonico et. al. (2014a, 2014b, 2015, 2016), Cattaneo et. al. (2015, 2016a, 2016b, 2016c) inference techniques to test our randomization assumption in the determination of the border cutoff in the assignment mechanism which is crucial for the validity of RD estimates in our setup.

3.4 Outcomes and Descriptive Statistics We construct the measures of human capital and long-run development from the 2010 Population and Housing Census (Censo Nacional de Población, Hogares y Viviendas) in Argentina. Our set of human capital and long-run development outcomes consists of nine αdepartment-level variables which capture human capital formation and degree of development from the household survey: (i) illiteracy rate (% adults aged 15 and above), (ii) computer literacy rate (% adults aged 3 and above), (iii) housing and land ownership (% households), (iv) access to public water network (% households), (v) absence of water supply inside the house (% households), (vi) absence of sanitary facilities and water discharge inside the house (% households), (vii) connection to the fixed telephone line (% households), (viii) cell phone ownership rate (% adults), and (ix) computer ownership (% households). The total sample consists of 12,300,706 households collapsed into 528 departments spanning 23 provinces and the City of Buenos Aires across 6 geographical regions (Northwest, Gran Chaco, Littoral, Cuyo, Pampas, Patagonia). Although these measures are by default imperfect, they largely reflect the cross-department differences both in the household-level human capital formation and department-level long-run development and allow us to identify whether these differences can be explained by the persistent effect of Spanish colonial law.

In Figure 1, we construct a latent variable measuring combined human capital formation and development level by taking the first component from the factor analysis of the full set of outcomes. The dominant component from the factor rotation matrix exhibits a high internal consistency as indicated by Cronbach (1951) Alpha (α = 0.9) and correlates strongly with each of the nine outcome variables. The rotated factor loadings suggest that the underlying latent component is strongly positively associated with computer literacy, access to public water network, cell phone ownership, computer ownership, sanitary facilities and water discharge, and fixed telephone line, strongly negatively correlated with illiteracy rate and the absence of water supply, and somewhat negatively correlated with the landownership rates. The distribution of the latent variable is broken down into quartiles to capture the variation in its level across space. Darker colored areas correspond to the lowest level of human capital formation and development whereas as the white colored departments indicated a very high level of development.

18

Figure 1: Department-Level Human Capital and Long-Run Development Across Argentina

Such a pattern succinctly demonstrates the relevance of the underlying factor variable as a composite latent measure of human capital formation and long-run development. The red line in the figure corresponds to the former colonial border line separating the departments colonized by Spain from the departments not integrated into the Spanish colonial empire. This particular subset of departments is located south of the border line whereas the colonial subsample is located north of the border line. The first glimpse of the descriptive pattern advocates a large gap between the departments in the level and distribution of human capital and long-run development. The departments formerly integrated into the colonial empire tend to exhibit a markedly lower level of human capital formation and development compared to the departments south of the former colonial border. The value of the latent variable is particularly high in the area of Buenos Aires, it tends to decline somewhat at intermediate distance from the capital city, and jumps substantially in the temperate southern region. Focusing on the area north of the border, the value of the latent variable appears to drop persistently in the closer proximity to Audiencia Real. The value of the index is particularly low in the areas in a close circle of the Charcas where the seat of the Audiencia Real was located.

In Table 1, we present the descriptive statistics for the full set of outcomes and the covariates used throughout the empirical analysis. In Panel A, the descriptive statistics is presented for the set of outcome variables. Each outcome exhibits a strong tendency of substantial variation

19

across the departments. The observed illiteracy rate is the range between 0.2 percent in Comuna 14 in Buenos Aires City, and 17.9 percent in department Gastre in Chubut Province. For computer literacy, we observe the lowest value in the department Mitrè in the Province Santiago del Estero at 6.3 percent compared to the highest value at 100% in the southern part of Tierra del Fuego. The highest observed rate of landownership is found in the department General Angel V. Peñalosa in the Province La Rioja whereas the lowest rate is observed in the departments La Poma (Province Salta), Comuna 1 (City of Buenos Aires), and Lago Argentino (Province Santa Cruz). The lowest fraction of households without the water supply inside the house is found in the three communities in the City of Buenos Aires whereas the fraction jumps to 95 percent the department Mitré in Province Santiago del Estero. The departments from the Provinces Chaco and Buenos Aires are significantly more likely than the rest of the sample to lack the connection to the public water network whereas the share of households with access to public water network gradually approaches 100 percent moving toward the provinces in the more temperate zone (Chubut, Santa Cruz,Tierra del Fuego) and the City of Buenos Aires. A very similar pattern is found for the cross-department distribution of households with sanitary facilities and water discharge inside the house where the lowest value is observed for the department Ramón Lista in the Province Formosa, and the highest value is found for the communes in the City of Buenos Aires. The descriptive evidence outright suggests that the level of human capital formation and long-run development is much higher in the departments located across southern temperate provinces such as Santa Cruz and Tierra del Fuego situated far away from the jurisdiction of the colonial Audiencia Real. Considering the cell phone ownership, the departments falling outside the jurisdiction of Audiencia Real have greater cell phone ownership rates and more computer literate population. As a preliminary test of our hypotheses, we employ the paired-sample t-test on mean comparison between the treatment and control sample. On average, the departments falling outside the scope and jurisdiction of Audiencia Real have 1.4 percentage points lower rate of illiteracy (p-value = 0.000), and 16 percentage points higher rate of computer literacy (p-value = 0.000). The households in the departments outside the jurisdiction of Audiencia Real are, on average, 18.9 percentage points more likely to have water supply inside the house, and 17.5 percentage points (p-value = 0.000) more likely to have sanitary facilities and water discharge installed in the house. The departments outside the administrative and imperial jurisdiction of the Audiencia are also 16 percentage points more likely than Audiencia-affected departments to have a fixed telephone line (p-value = 0.000) , have almost 5 percentage points higher rate of cell phone ownership (p-value = 0.000) and have 14 percentage points higher rate of computer ownership (p-value = 0.000), but no meaningful differences in the housing and landownership rate, and the frequency of access to public water network.

20

Table 1: Descriptive Statistics Mean by Sample Treatment

-Control Difference

Mean StD Min P25 P75 Max Treatment Control

Panel A: Outcomes

Illiteracy Rate (%)

3.3 2.6 0.2 1.4 4.5 17.9 3.8 (2.4)

2.4 (2.7)

0.000

Computer Literacy (%)

43.6 15.3 6.3 32.5 53.8 100 37.2 (15.4)

53.2 (8.9)

0.000

Landownership Rate (%)

68.2 7.3 32.2 64.2 72.6 88.6 68.4 (7.9)

68.0 (6.3)

0.501

Access to Public Water Network

(%)

77.6 19.8 0.60 69.6 91.2 100 77.4 (19.9)

77.9 (19.6)

0.798

Households without Water

Supply Inside the House (%)

21.0 19.5 0.50 6.0 30.7 95.3 28.5 (21.0)

9.6 (8.4)

0.000

Households with Sanitary Facilities

and Water Discharge

79.2 18.8 8.6 70.6 93.4 100 72.2 (20.3)

89.7 (8.9)

0.000

Households with Fixed Telephone

Line (%)

34.7 21.2 0.2 17.2 49.9 100 28.2 (21.2)

44.4 (17.0)

0.000

Cell Phone Ownership (%)

82.6 11.1 11.6 81.6 88.1 100 80.7 (11.2)

85.4 (10.3)

0.000

Computer Ownership (%)

35.1 15.4 3.5 22.9 45.4 100 29.4 (15.3)

43.9 (10.9)

0.000

Panel B: Treatment Status

Integrated into Spanish Empire

0.6 0.4 0 0 1 1

Panel C: Assignment Mechanism

Distance from the Jursidiction of

Audiencia Real

1585.1 632.4 335.8 1136.50

1897.81

6292.0 1210 (358.5)

2147.8 (528.6)

0.000

Panel D: Covariates

Population Density

(# per km2)

772.5 3166.8 0.1 2.8 28.3 29302.7

910.23 (3846.3)

563.03 (1655.6)

0.155

Foreign-Born Population (%)

2.4 3.2 0.1 0.5 3.1 24.7 1.86 (3.2)

3.3 (3.0)

0.000

Log Population Size

10.2 1.4 2.8 9.3 11.1 14.3 10.3 (1.2)

10.1 (1.6)

0.156

Distance from Buenos Aires

728.8 481.6 2.7 353.5 1014.6

4517.2 831.4 (341.6)

574.9 (605.5)

0.000

Notes: the table presents the means and standard deviations for the full set of outcomes and covariates used in the empirical model setup. In mean comparison between treatment and control samples, the standard deviation of each covariate in both subsamples is displayed in the parentheses.

21

4. Results

4.1 Basic OLS Estimates In Table 2, we present the basic OLS estimates of the effect of distance from colonial Audiencia Real on the set of human capital and development outcomes. Each empirical specification of the underlying model is estimated twice. First, the baseline model specification is estimated using a simple OLS technique. But since the level of human capital formation and development outcomes might exacerbate measurement error due to the varying sizes of the departments, the department-level outcomes might exhibit excessive sampling variation that might lead to outliers or excessive heterogeneity. Second, our strategy to mitigate such concern is to weigh the department-level observations by the household count included in the survey. Such an approach partially alleviates the measurement error by overweighing the departments with a lower variance-to-signal ratio in the construct human capital and development outcomes. In columns (1) through (4), the effects of the geographical proximity to Audiencia Real on department-level human capital formation are examined. The OLS estimates readily indicate a deeply persistent effect of the distance from the Audiencia on human capital formation. In the non-weighted model specification, doubling the distance from the Audiencia is associated with 69 percent decrease in the rate of illiteracy, and 62 percent increase in the rate of computer literacy. The point estimate is marginally significant at 10%. In the weighted specification, the effect does not disappear, and tends to gain further strength in terms of magnitude and significance even when potentially confounding covariates are switched on. In the weighted specification in column (2), halving the distance to the seat of colonial Audiencia is associated with an increase in the rate of illiteracy by the factor of 1.5 while the computer literacy tends to increase by 71 percent, respectively. Both point estimates are statistically significant at 1%. In both weighted specifications, region-fixed effects and province-fixed effects are controlled for to further tackle the spatial heterogeneity bias which might invoke the omitted variable bias. In the base setup, the distance from Audiencia Real accounts for up to 18 percent of the illiteracy gap, and up to 43 percent of the variation in computer literacy across departments. In columns (5) and (6), the OLS estimates suggest that the proximity to the seat of Audiencia Real tends to boost landownership substantially in the weighted specification whereas no such effect is indicated in the baseline setup. The distance from the Audiencia explains no more than 4 percent of the variance in the landownership rate between the departments in the non-weighted specification. In columns (7) through (14), the effects of the distance from the Audiencia on the set of household-level development outcomes are examined. The OLS estimates arguably indicate that doubling the geographic distance from the seat of the colonial Audiencia court is associated with a marked and persistent improvement in the level of development. In economic terms, doubling the distance from the court is reflected into higher wealth and income levels and better infrastructural development. Take the differential between the 75th and 25th percentile of the distance from the court. In our dataset, the representative departments for the 25th and 75th percentile of the distance variable are La Paz (Catamarca) and Ensenada (Buenos Aires), respectively. Evaluating the effect of the distance from the court at both percentile thresholds implies that the departments around the 75th percentile experience 65 percent higher access rate to water supply, are 1.2 times more likely to possess sanitary facilities and water discharge inside the household, 3.4 times more likely to have a fixed telephone line, and are 3.8-times less likely to lack water supply inside the house. In the weighted specification, controlling for the confounding influence of population density, population size, distance from Buenos Aires, and the fraction of foreign-born population confirms the importance of the distance from Audiencia Real in explaining the human capital formation and development outcomes.

22

Table 2: Persistent Effects of Colonial Audiencia on Human Capital and Development Outcomes Department-Level Human Capital Formation Landholding

Concentration Household-Level Development Outcomes Household-Level Human Capital Outcomes

Illiteracy Rate Computer Literacy Land Ownership Access to Public Water Network

Absence of Water Supply Inside the House

Sanitary Facilities and Water Discharge

Fixed Telephone Line Cell Phone Ownership Computer Ownership

Basic OLS

Weighted Basic OLS Weighted Basic OLS

Weighted Basic OLS Weighted Basic OLS Weighted Basic OLS

Weighted Basic OLS

Weighted Basic OLS

Weighted

Basic OLS Weighted

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18)

Panel A: Treatment Effect

Spanish Colonial Empire

.125 (.328)

-1.568*** (.339)

-.059 (.121)

.624** (.255)

-.047 (.049)

-.176*** (.055)

.110 (.102)

2.050*** (.595)

.207 (.384)

-4.473*** (1.313)

.0008 (.061)

.999*** (.302)

-.008 (.235)

1.607*** (.566)

.109 (.074)

.154 (.188)

-.050 (.153)

.538** (.253)

Panel B: Assignment Mechanism (Distance from Colonial Audiencia)

Log Distance from Sucre

-.690* (.405)

-1.561*** (.179)

.629*** (.212)

.717*** (.134)

-.085 (.072)

-.070** (.028)

.226 (.187)

.918*** (.341)

-1.353** (.567)

-2.956*** (.651)

.446*** (.171)

.628*** (.162)

1.185*** (.326)

1.221*** (.296)

.284*** (.096)

.201** (.101)

.771*** (.226)

.779*** (.127)

Constant Term 5.868** (2.961)

13.233*** (1.523)

-.856 (1.614)

-2.279* (1.284)

4.871*** (.553)

5.135*** (.204)

2.573* (1.439)

-4.856* (2.661)

12.311*** (4.230)

27.080*** (4.609)

1.075 (1.290)

-.932 (1.418)

-5.362** (2.387)

-7.259*** (2.727)

2.257*** (.733)

2.788*** (.880)

-2.149 (1.710)

-3.060** (1.375)

Exogenous Covariates

(p-value)

NO YES (0.000)

NO YES (0.000)

NO YES (0.000)

NO YES (0.000)

NO YES (0.000)

NO YES (0.000)

NO YES (0.000)

NO YES (0.000)

NO YES (0.000)

# Departments 528 518 529 518 520 518 526 518 520 518 529 518 521 518 521 523 521 518

# Regions 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6

# Provinces 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24 24

# Households 12,300,706 12,300,706

12,300,706 12,300,706

12,300,700 12,300,700

12,300,700

12,300,700

12,300,7000

Province Fixed-Effects

NO YES (0.000)

NO YES (0.000)

NO YES (0.000)

NO YES (0.000)

NO YES (0.000)

NO YES (0.000)

NO YES (0.000)

NO YES (0.000)

NO YES (0.000)

Region-Fixed Effects NO YES (0.000)

NO YES (0.000)

NO YES (0.000)

NO YES (0.000)

NO YES (0.000)

NO YES (0.000)

NO YES (0.000)

NO YES (0.000)

NO YES (0.000)

Adj. R2 0.18 0.78 0.43 0.71 0.04 0.66 0.01 0.18 0.36 0.71 0.31 0.55 0.28 0.68 0.18 0.33 0.39 0.72

Wald Test (p-value)

0.106 0.000 0.000 0.000 0.495 0.000 0.430 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.012 0.000 0.000 0.000

Notes: the table reports the OLS estimated effect of distance from colonial Audiencia Real on the set of human capital and development outcomes. Panel A reports the treatment effect of integration in the Spanish colonial empire while Panel B reports the coefficient on the log distance from colonial Audiencia as the assignment mechanism used throughout the analysis. Standard errors are adjusted for heteroscedasticity and within-country serially correlated stochastic disturbances into 528 department-specific clusters, 24 province-specific clusters and 6 region-specific clusters using Cameron-Gelbach-Miller non-nested multi-way clustering scheme for finite-sample adjustment of the empirical distribution function and cluster-robust parameter inference to remove the inconsistencies arising from biased OLS covariance matrix estimator and serially correlated intra-cluster residuals. Multiway cluster-robust standard errors are denoted in the parentheses for each empirical specification. Asterisk denote statistically significant parameter estimates at 10% (*), 5% (**) and 1% (***), respectively.

23

In columns (15) through (18), the basic OLS estimates of the effect of the distance from the seat of Audiencia on household-level development outcomes are presented. The size and magnitude of the effect is similar between weighted and non-weighted model specification. The proximity to the colonial court is associated with a marked drop in both cell phone ownership and computer ownership. In particular, the point estimates indicate that increasing the distance from the seat of Audiencia by 500 km is associated with 76 percent increase in the rate of cell phone ownership, and with an increase in the frequency of computer ownership by 1.3 times. The distance from the seat of Audiencia in the baseline OLS setup explains up to 18 percent of the between-department differences in cell phone ownership, and up to 39 percent of the corresponding differences in the frequency of computer ownership. Full-sample OLS estimates in Table 2 indicate substantial department-level aggregate correlation between the set of human capital and development outcomes and the distance from the seat of colonial Audiencia. However, the estimates susceptible to spatial heterogeneity since already indicated in Figure 1 where the exceptionally low levels of human capital and developed are clustered in the North and Northwest are clustered, and potentially neglect the outliers that could critically affect the stability of OLS estimates. In Table 3, we replicate the baseline OLS specification but split the departments in each province off the full sample to address the spatial contingency in the distribution of human capital and development across departments. In column (1), the replicated effects of the distance from Audiencia on illiteracy rate are presented. The effect become noticeably weaker once the departments in the Northwestern cluster are excluded from the full sample, and only marginally significant within p-value=0.15 significance threshold. Excluding the Northwest from the full sample does not appear to affect the effects on the rate of computer literacy in column (2) but somewhat renders the effect on the access to public water network insignificant. In column (3), the Northwest spatial split suggests that greater distance from the colonial Audiencia is associated with a notable drop in the rate of landownership in the range between -.071 and -.124 for each log point increase in the distance from the court compared to the weak OLS effect in Table 2. In Panel B, the departments from the provinces in Gran Chaco (Chaco, Formosa, Santiago del Estero) are split of the full sample. This particular spatial split-off is essential mainly because the departments in these provinces were fully integrated into the Spanish colonial empire and were under direct jurisdiction of the Audiencia Real in a close proximity. Similar to column (1), the effect of the proximity to the colonial court on the prevalence of illiteracy becomes noticeably weaker whereas the effects on computer literacy, landownership rate, household-level infrastructural development and human capital outcomes across columns (2) through (9) remain intact, and quantitatively comparable both to Panel A and Table 1. In Panel C, excluding the departments from the Littoral (Corrientes, Formosa, Santiago del Estero) groupwise from the full sample does not seem to affect the stability of the OLS estimates. In Panel D, the spatial split of the departments from the Cuyo Region (La Rioja, Mendoza, San Juan, San Luis) does not seem to influence the results either. In Panel E, the departments from the provinces in the Pampas are split off the core sample. Since the departments in the Pampas are characterized by above-median human capital and development outcomes, this particular split can indicate whether the spatial heterogeneity can be effectively ruled out. In column (1), the effect of the distance from the colonial court on the rate of illiteracy tends to become noticeably stronger at marginal significance levels. For instance, excluding the departments from the Province of Buenos Aires from the full sample suggests that one basis log point increase in the distance from the Audiencia Real is associated with 45.1 percent drop in the rate of illiteracy, 45.5 percent increase in the rate of computer literacy, 6.5 percent drop in the rate of landownership, 64.8 percent decrease in the likelihood of the absence of water supply inside the house, 37.4 percent rise in the likelihood of sanitary facilities and water discharge inside the house, 78.3 percent increase in the fixed-telephone line,

24

25.7 percent increase in the cell phone ownership rate, and 54.8 percent increase in the computer ownership rate, respectively. Table 3: Distance from Colonial Audiencia, Human Capital Formation and Development Across Subsamples

Department-Level Human Capital Formation

Landholding Concentration

Household-Level Development Outcomes Household-Level Human Capital Outcomes

Illiteracy Rate

Computer Literacy

Land Ownership

Access to Public Water

Network

Absence of Water Supply

Inside the House

Sanitary Facilities and

Water Discharge

Fixed Telephone

Line

Cell Phone Ownership

Computer Ownership

(1) (2) (3) (4) (5) (6) (7) (8) (9) Panel A: Excluded Subset: Northwest

Catamarca -.424 (.290)

.550*** (.200)

-.071*** (.027)

.235 (.170)

-1.164** (.546)

.431** (.182)

.923*** (.124)

.249*** (.078)

.649*** (.209)

Jujuy -.560 (.372)

.708*** (.245)

-.071 (.050)

.347* (.209)

-1.539** (.680)

.512** (.254)

1.206*** (.225)

.150** (.076)

.799*** (.250)

Salta -.490 (.352)

.623*** (.256)

-.124*** (.028)

.284 (.216)

-1.354** (.675)

.502** (.238)

.972*** (.304)

.250*** (.100)

.731*** (.256)

Tucumán -.420 (.278)

.543*** (.197)

-.074** (.031)

.234 (.170)

-1.145** (.537)

.433** (.182)

.934*** (.206)

.256*** (.076)

.637*** (.208)

Panel B: Excluded Subset: Gran Chaco

Chaco -.377* (.225)

.537*** (.180)

-.076** (.035)

.196* (.105)

-1.117** (.482)

.409*** (.153)

.924*** (.254)

.251*** (.078)

.635*** (.188)

Formosa -.382 (.355)

.507*** (.159)

-.080** (.037)

.214 (.155)

-1.112** (.517)

.371*** (.125)

.865*** (.208)

.234*** (.086)

.615*** (.179)

Santiago del Estero

-.393 (.253)

.505*** (.130)

-.069*** (.026)

.215 (.138)

-1.107** (.474)

.389*** (.123)

.896*** (.206)

.247*** (.079)

.599*** (.135)

Panel C: Excluded Subset: Littoral

Corrientes -.454* (.249)

.548*** (.189)

-.071** (.034)

.220 (.162)

-1.175** (.522(

.425*** (.169)

.941*** (.263)

.249*** (.079)

.655*** (.194)

Entre Ríos -.424 (.283)

.539*** (.194)

-.071** (.032)

.213 (.160)

-1.134** (.549)

.417*** (.172)

.940*** (.273)

.258*** (.077)

.635*** (.205)

Misiones -.459* (.246)

.569*** (.182)

-.077** (.032)

.248* (.154)

-1.209** (.512)

.441*** (.179)

.972*** (.251)

.252*** (.081)

.675 (.187)

Panel D: Excluded Subset: Cuyo

La Rioja -.389 (.287)

.535*** (.199)

-.084*** (.028)

.217 (.165)

-1.150** (.552)

.415** (.176)

.919*** (.281)

.248*** (.077)

.625*** (.213)

Mendoza -.428 (.282)

.533*** (.189)

-.068* (.036)

.219 (.158)

-1.123** (.549)

.411*** (.166)

.941*** (.262)

.249*** (.085)

.636*** (.199)

San Juan -.396 (.275)

.557*** (.190)

-.068*** (.037)

.199 (.149)

-1.156** (.555)

.412*** (.163)

.955*** (.261)

.240*** (.086)

.658*** (.202)

San Luis -.316 (.284)

.529*** (.187)

-.077*** (.034)

.227 (.160)

-1.151** (.540)

.423*** (.171)

.928*** (.267)

.252*** (.081)

.622*** (.196)

Panel E: Excluded Subset: Pampas

Buenos Aires

-.451** (.221)

.455*** (.188)

-.065*** (.031)

.114 (.165)

-.648 (.411)

.374** (.179)

.783*** (.225)

.257*** (.086)

.548*** (.200)

Buenos Aires City

-.298 (.218)

.510*** (.181)

-.065** (.031)

.173 (.155)

-.898** (.372)

.416** (.176)

.894*** (.248)

.258*** (.078)

.591*** (.183)

Córdoba -.383 (.257)

.530*** (.183)

-.074** (.033)

.232 (.161)

-1.112** (.501)

.413*** (.167)

.906*** (.246)

.257*** (.079)

.627*** (.192)

La Pampa -.421 (.271)

.548*** (.189)

-.073** (.032)

.230 (.158)

-1.174** (.512)

.426*** (.171)

.938*** (.259)

.250*** (.078)

.649*** (.199)

Santa Fe -.397 (.264)

.535*** (.186)

-.076** (.034)

.231 (.160)

-1.133** (.521)

.419*** (.169)

.920*** (.255)

.259*** (.076)

.633*** (.195)

Panel F: Excluded Subset: Patagonia

Chubut -.546** (.277)

.563*** (.197)

-.062*** (.032)

.241 (.168)

-1.239** (.561)

.444*** (.179)

.967*** (.213)

.301*** (.063)

.669*** (.207)

Neuquén -.451* (.273)

.549*** (.190)

-.072** (.034)

.233 (.161)

-1.183** (.536)

.429*** (.173)

.945*** (.264)

.249*** (.077)

.647*** (.200)

Rio Negro -.463* (.275)

.552*** (.191)

-.077** (.033)

.237 (.160)

-1.201** (.539)

.440*** (.173)

.942*** (.266)

.263*** (.075)

.662*** (.201)

Santa Cruz -.316 (.284)

.536*** (.194)

-.078*** (.033)

.239 (.167)

-1.069** (.542)

.422** (.179)

.911*** (.201)

.238*** (.080)

.623*** (.202)

Tierra del Fuego

-.360 (.276)

.547*** (.191)

-.076** (.033)

.240 (.162)

-1.157*** (.531)

.430*** (.175)

.938*** (.260)

.245*** (.083)

.641*** (.199)

Notes: the table reports the OLS estimated effect of distance from colonial Audiencia Real on the set of human capital and development outcomes on multiple spatially-contingent subsamples. Standard errors are adjusted for heteroscedasticity and within-country serially correlated stochastic disturbances into 528 department-specific clusters, 24 province-specific clusters and 6 region-specific clusters using Cameron-Gelbach-Miller non-nested multi-way clustering scheme for finite-sample adjustment of the empirical distribution function and cluster-robust parameter inference to remove the inconsistencies arising from biased OLS covariance matrix estimator and serially correlated intra-cluster residuals. Multiway cluster-robust standard errors are denoted in the parentheses for each empirical specification. Asterisk denote statistically significant parameter estimates at 10% (*), 5% (**) and 1% (***), respectively.

Since the departments in Buenos Aires Province are characterized by relatively high levels of human capital formation and development outcomes compared to the rest of the sample, the evidence from the replicated core OLS model specification suggests that the distance from the colonial court and the department-level proximity to its jurisdiction matter a great deal in explaining the disparities in human capital formation and development levels. Similar results

25

are obtained once the departments from Buenos Aires City, Córdoba, and Santa Fe which share above-average levels of human capital formation and development, are excluded from the full sample. In Panel F, the departments from each province in Patagonia (Chubut, Neuquén, Río Negro, Santa Cruz, Tierra Del Fuego) are split off the core sample. The departments in the Patagonian cluster are equally characterized by relatively high levels of development and human capital formation which tends to rise with the increasing distance from the colonial Audiencia. Excluding the oil-rich provinces Chubut, Río Negro, and Neuquén which can potentially comprise the set of outliers confirms the baseline OLS estimates whereas excluding Santa Cruz and Tierra del Fuego provinces, characterized by exceptionally high levels of human capital formation and development, does not seem to render the results unstable. Henceforth, the evidence clearly suggests that once spatial heterogeneity and contingency are addressed in the baseline model specification, the establish effect of the proximity to the colonial Audiencia on human capital formation and development tends to be stable and within expected confidence bounds.

4.2 RD Estimates

In Table 4, we present the core RD estimates of the effect of the proximity to the colonial Audiencia where discontinuous jump from the non-affected departments to the departments under the jurisdiction of the former colonial court is exploited to identify the causal effect of colonial law on human capital formation and long-run development. An important ex-ante issue in RD setup is the selection of the smoothing parameter used to estimate the regression function on both sides of cutoff. In selecting the bandwidth parameter, we first follow Calonico et. al. (2014b) and use bias-corrected local polynomials to derive the optimal bandwidth size with optimal finite-sample properties. Second, following Imbens and Kalyanamaran (2011), we allow the optimal bandwidth to depend on prior unknown functional specification of the RD model and adopt a data-driven algorithmic approach towards bandwidth selection. And third, following Ludwig and Miller (2007), we use a boundary-optimal kernel-weighted local linear regression left and right of the cutoff to estimate the persistent impact of the proximity to the colonial Audiencia Real on human capital and long-run development. Using the three distinctive procedures for optimal bandwidth selection allows us to address the aggregate uncertainty regarding the magnitude of the smoothing parameter, compare the estimates across different bandwidths, and leverage the precision of our estimates. In column (1), the core RD estimates of the effect of the proximity to the Audiencia on human capital and development outcomes are presented in the flexibile RD setup without the control variables. The evidence readily suggests that the proximity to the colonial Audiencia Real is associated with persistently worse development outcomes and disadvantageous human capital formation in the long run. Compared to the departments under the jurisdiction of the Audiencia, the departments falling outside the colonial jurisdiction of the Audiencia Real have 39.6 percent lower rate of illiteracy (p-value = 0.000), 24.2 percent higher rate of computer literacy (p-value = 0.002), and 8.1 percent lower landownership rate (p-value = 0.001). The households in the departments in the non-treated sample outside the Audiencia’s jurisdiction are 20.2 percentage points more likely to have sanitary facilities and water discharge installed in the household (p-value = 0.002), 12.7 percentage points less likely to lack water supply inside the house (p-value = 0.095), 68.3 percentage points more likely to have a fixed-telephone line (p-value = 0.000), 4.5 percentage points more likely to own a cell phone (p-value = 0.283), and a little more than 24 percentage points more likely to own a computer (p-value = 0.002). The estimated RD effects are constructed using the robust confidence intervals with non-parameteric procedure suggested by Calonico et. al. (2014a). The evidence from column (2) suggests the inclusion of the relevant covariates in the vector of control variables does not affect the precision of our estimates.

26

In column (3), the set of departments from the Buenos Aires City is split off the full sample to check for the potential outlier-driven discrepancy. The point estimates in the sample without the departments in the Buenos Aires City do not appear to deviate markedly from the baseline RD estimates in columns (1) and (2). In particular, the effect of the distance from the Audiencia Real rises from -.396 to -.417 without any change in the underlying p-value. The exclusion of Buenos Aires City from the full sample renders the relationship between computer literacy and the distance from the Audiencia Real substantially weaker whereas no such weakening of the baseline RD estimates is found for the remaining set of development outcomes and human capital formation. A notable objection to the RD estimates posited by columns (1) through (3) might be the uncertainty about the precise colonial border line in Figure 1 which might include potentially outlying departments on either side of the border which could have critical implications for the stability and validity of RD estimates. To mitigate such an important concern, we switch the assignment cutoff at the existing border line separating the treatment and control group one standard deviation to left, and one standard deviation (S.D) to the right of the baseline cutoff. In column (6) and (7), the baseline RD estimates are replicated by switching the cutoff value in the assignment mechanism one standard deviation to the left and right of the existing border cutoff. The point estimates remain essentially unchanged for the effect of the proximity to colonial Audiencia on department-level human capital formation, i.e. illiteracy rate and the computer literacy rate. Switching the assignment cutoff one S.D left of the colonial border cutoff leaves the parameter estimates essentially unchanged compared to the baseline RD estimates in columns (1) and (2) except for the set of department-level development outcomes which confirms a slightly more robust relationship between the proximity to Audiencia and the set of human capital formation outcomes. Since the treatment impact of the colonial law under Audiencia Real is invariant across departments, changing the cutoff must by default alter the estimates especially if the departments near the border cutoff possess higher levels of development compared to the rest of the treatment sub-sample closer to the seat of Audiencia Real. As indicated in Figure 1, outcome comparison between treatment and control group in Figure 1 indicates exceptionally higher levels of human capital formation and development outcomes in the treated departments near the border cutoff, particularly in the provinces such as Cordóba and Santa Fe. Switching the assignment mechanism to either side of the border cutoff is by default expected to render the estimates much different from the ones in column (1) and (2) where the existing border cutoff is used on the running variable. In a similar vein, the invariance of the border cutoff implies that different bandwidth selection should matter for the stability of parameter estimates since the regression line near the cutoff might change substantially. To illustrate the pitfalls of border cutoff switch and the alternative bandwidth selection from the one used at the existing cutoff, columns (8) through (10) confirm that the relationship between the proximity to the Audiencia Real and the set of outcomes is persistently unstable and weak, and also indicates why the existing border cutoff used throughout columns (1) through (3) should comprise the core of the RD analysis. In Appendix A1, the figure graphically depicts the relationship between the distance to the colonial jurisdiction of the Audiencia Real and the set of human capital- and development outcomes with the corresponding 95% confidence bounds. In Appendix A2, we replicate the identical set of relationships with the locally weighted linear regression using the smoothing parameter to allow for kernel-weighted non-linearites at both sides of the cutoff. Both figures indicate that the effect of the proximity to the Audiencia Real on human capital and development disappears in the control sample where a clearly stationary relationship is observed even when non-linearities in the running variable are allowed. On the other hand, the relationship turns strongly negative in the treatment sample subject to the colonial jurisdiction of the Audiencia Real indicating a strongly persistent negative effect of the proximity to the court on the full set of

27

outcomes, except for the access to public water network where no such treatment-control differences are observed.

4.2.1 RD Estimates with Sample Splits

In Table 5, we replicate the core RD estimates of the effect of proximity to the

jurisdiction of Audiencia Real from column (2) in Table 4 with the full set of covariates switched on. In each replication, each individual province is split off the core sample with the full set of departments. This particular kind of replication allows us to examine whether the underlying effect of the colonial law on long-run development and human capital formation is driven by the spatial idiosyncrasies which could influence the persistent effect of the proximity to Audiencia’s jursidction especially in within the treatment sample. In column (1), the effect of the proximity to Audiencia’s jurisdiction is re-examined by splitting the departments within each province in the Northwest region off the full sample. The evidence is perhaps the most crucial one since this part of Argentina was not only integrated into the Spanish colonial empire but also lies in the closest proximity to the Audiencia compared to the rest of the sample. The evidence largely indicates that excluding the departments within each province off the full sample does not affect the stability of the core RD estimates. In particular, the switch off the jurisdiction of the Audiencia Real is associated with a drop in the illiteracy rate between 37 percent and 45 percent, respectively. The established effect is statistically significant at 1%. In Panel B through Panel E, the exclusion of departments from different provinces and regional clusters does not seem to affect the baseline RD estimates at all both in terms of the magnitude and significance level. In Panel F, where the Patagonian departments are split of the full sample, the switch from the Audiencia’s jurisdiction to the non-jurisdiction translates into the drop in the illiteracy rate between 31 percent and 36 percent, respectively. We find similar evidence in column (2) where the jump from the Audiencia’s jurisdiction across the colonial border cutoff tends to improve the computer literacy rate between 14.3 percent, and 35.4 percent at 1% significance level, respectively although the exact point estimate depends on the excluded subset of departments from the full sample. In column (3), the jump from the Audiencia’s jurisdiction to the non-jurisdiction is associated with a markedly lower rate of landownership in the range between 3.7 percent and 14.6 percent across different spatial subsamples. In column (4), the evidence suggests a notably weaker effect on the access to public water since the jump outside Audiencia’s jurisdiction is associated with a marginally higher rate of access to public water network but only at borderline significance levels. In column (5), the estimates confirm a marked drop in the likelihood of absence of water supply inside the house in the range between 41.8 percent and 73.7 percent whilst in column (6) a quantitatively smaller effect but statistically significant (at 1%) is confirmed with respect to the likelihood of having sanitary facilities and water discharge inside the house. Spatial sample splits do not seem to affect the likelihood of having a fixed-telephone line inside the house since the cross-subsample point estimates indicate that the departments under the former jurisdiction of colonial Audiencia are between 52 percent and 79 percent less likely to have a fixed-telephone line compared to the rest of the sample. On the other hand, spatial sample splits do affect the RD estimates with respect to cell phone ownership where, as indicated in column (8), we find no relationship with the proximity to the Audiencia’s jurisdiction. In column (9), the evidence confirms a large increase in computer ownership in the departments outside the Audiencia’s jurisdiction in the range between 28.2 percent and 47.4 percent, respectively.

28

Table 4: Effects of Distance from Colonial Audencia on Human Capital Formation and Development in Regression Discontinuity Design (1) (2) (4) (6) (7) (8) (9) (10)

Assignment Cutoff Switch

Bandwidth Selection Procedure

Without Controls With Controls Excluding Buenos Aires City

-σ +σ Calonico-Cattaneo-Titiunik

Imbens-Kalyanaraman

Ludwig-Miller

Panel A: Illiteracy Rate

Point Estimate -.396 (.113)

-.396 (.113)

-.417 (.113)

-.453 (.106

-.453 (.106)

-.276 (.198)

-.253 (.199)

-.229 (.203)

p-value 0.000 0.000 0.000 0.000 0.000 0.165 0.204 0.259 95% CI [-.706,-.389] [-.706,-.389] [-.641,-.324] [-.756,-.460] [-.756,-.469] [-.576, .082] [-.511, .065] [-.512, .065]

95% CI Robust [-.618,-.174] [-.618,-.174] [-.638,-.195] [-.662,-.244] [-.662,-.244] [-.665, .113] [-.643, .137] [-.628, .168] Bandwidth 304.46 304.46 304.46 304.46 304.46 0.195 0.260 0.259

Treatment Sample Size

174 174 174 205 205 54 84 84

Control Sample Size

354 354 338 323 323 104 125 125

Panel B: Computer Literacy

Point Estimate .242 (.076)

.242 (.076)

-.090 (.095)

.180 (.096

.159 (.060)

.049 (.090)

.128 (.165)

.111 (.094)

p-value 0.002 0.002 0.344 0.062 0.009 0.589 0.439 0.237

95% CI [.228, .444] [.228, .444] [-.161, .179] [.163, .413] [.200, .383] [-.083, .226] [-.109, .196] [-.115, .182]

95% CI Robust [.092, .392] [.092, .392] [-.278, .097] [-.008, .369] [.039, .278] [-.128, .226] [-.196, .453] [-.073, .296]

Bandwidth 304.46 304.46 304.46 304.46 304.46 0.263 0.306 0.374

Treatment Sample Size

174 174 174 150 205 85 103 115

Control Sample Size

355 355 339 379 324 126 142 226

Panel C: Land Ownership

Point Estimate -.081 (.025)

-.081 (.025)

-.084 (.025)

.111 (.058)

.009 (.025)

.063 (.054)

.071 (.057)

.051 (.051)

p-value 0.001 0.001 0.001 0.055 0.717 0.241 0.214 0.320

95% CI [-.057, .007] [-.057, .007] [-.050, .014] [.107, .274] [-.042, .024] [-.043, .137] [-.045, .106] [-.052, .073]

95% CI Robust [-.131, -.031] [-.131, -.031] [-.134, -.034] [-.002, .226] [-.040, .058] [-.042, .169] [-.041, .183] [-.049, .152]

Bandwidth 304.46 304.46 304.46 304.46 304.46 0.211 0.272 0.366

Treatment Sample Size

174 174 174 49 427 61 89 113

Control Sample Size

354 354 338 479 101 110 128 227

Panel D: Access to Public Water Network

Point Estimate -.505 (.127)

-.505 (.127)

-.528 (.127)

.620 (.234)

.093 (.185)

-.183 (.167)

-.277 (.205)

-.307 (.203)

29

p-value 0.000 0.000 0.000 0.008 0.612 0.272 0.177 0.131

95% CI [-1.069, -.701] [-1.069, -.701] [-.992, -.625] [-.021, .586] [-.165, .329] [-.506, .054] [-.510, .053] [-.508, .054]

95% CI Robust [-.755, -.254] [-.755, -.254] [-.779, -.278] [.160, 1.079] [-.269, .457] [-.511, .143] [-.681, .125] [-.705, .091]

Bandwidth 304.46 304.46 304.46 304.46 304.46 0.261 0.258 0.259

Treatment Sample Size

174 174 174 49 427 84 84 84

Control Sample Size

354 354 338 479 101 126 125 125

Panel E: Sanitary Facilities and Water Discharge

Point Estimate .202 (.066)

.202 (.066)

.202 (.066)

-.172 (.162)

-.062 (.023)

.016 (.060)

.068 (.069)

-.038 (.059)

p-value 0.002 0.002 0.002 0.289 0.008 0.787 0.324 0.512

95% CI [.130, .328] [.130, .328] [.124, .323] [-.348, .127] [-.093, -.012] [-.068, .137] [-.089, .116] [-.044, .170]

95% CI Robust [.072, .333] [.072, .333] [.071, .333] [-.491, .146] [-.107, -.016] [-.102,.135] [-.067, .204] [-.155, .077]

Bandwidth 304.46 304.46 304.46 304.46 304.46 0.221 0.280 0.504

Treatment Sample Size

174 174 174 49 427 65 94 130

Control Sample Size

355 355 339 480 102 116 128 289

Panel F: Absence of Water Supply Inside the House

Point Estimate .127 (.076)

.127 (.076)

.125 (.076)

-.155 (.124)

-.049 (.055)

-.198 (.114)

-.163 (.128)

-.159 (.089)

p-value 0.095 0.095 0.098 0.212 0.366 0.083 0.202 0.076

95% CI [-.007, .287] [-.007, .287] [-.019, .276] [-.386, -.031] [-.072, .083] [-.358, .035] [-.346, .009] [-.092, .163]

95% CI Robust [-.022, .276] [-.022, .276] [-.023, .275] [-.399, .088] [-.158, .058] [-.421, .025] [-.414, .087] [-.334, .016]

Bandwidth 304.46 304.46 304.46 304.46 304.46 304.46 304.46 304.46

Treatment Sample Size

174 174 174 49 427 55 68 145

Control Sample Size

355 355 339 480 102 105 116 333

Panel G: Fixed Telephone LIne

Point Estimate .683 (.189)

.683 (.189)

.680 (.189)

-.038 (.557)

-.444 (.102)

.234 (.197)

.438 (.387)

.336 (.209)

p-value 0.000 0.000 0.000 0.945 0.000 0.236 0.258 0.108

95% CI [.326, .856] [.326, .856] [.290, .820] [-.912, .631] [-.620, -.285] [-.066, .610] [-.090, .579] [-.099, .548]

95% CI Robust [.313, 1.054] [.313, 1.054] [.309, 1.050] [-1.130, 1.053] [-.645, -.243] [-.153, .621] [-.321, 1.198] [-.073, .746]

Bandwidth 304.46 304.46 304.46 304.46 304.46 0.269 0.300 0.374

Treatment Sample Size

174 174 174 49 427 89 102 115

Control Sample Size

355 355 339 480 102 127 141 126

Panel H: Cell Phone Ownership

Point Estimate .045 .045 .045 .109 .037 -.013 -.022 -.014

30

(.042) (.042) (.042) (.110) (.026) (.029) (.039) (.034)

p-value 0.283 0.283 0.284 0.322 0.159 0.640 0.565 0.660

95% CI [-.141, -.012] [-.141, -.012] [-.142, -.014] [-.198, .161] [-.167, -.079] [-.064, .029] [-.063, .022] [-.062, .019]

95% CI Robust [-.037, .129] [-.037, .129] [-.038, .129] [-.107, .325] [-.014, .089] [-.072, .044] [-.101, .055] [-.081, .051]

Bandwidth 304.46 304.46 304.46 304.46 304.46 0.250 0.290 0.308

Treatment Sample Size

174 174 174 49 427 80 96 103

Control Sample Size

355 355 339 480 102 123 135 144

Panel I: Computer Ownership

Point Estimate .242 (.076)

.242 (.076)

.240 (.076)

-.114 (.188)

-.216 (.031)

.049 (.090)

.128 (.165)

.111 (.094)

p-value 0.002 0.002 0.002 0.542 0.000 0.589 0.438 0.237

95% CI [.228, .444] [.228, .444] [.210, .427] [-.316, .174] [-.162, -.054] [-.083, .226] [-.109, .196] [-.115, .182]

95% CI Robust [.092, .392] [.092, .392] [.090, .391] [-.484, .254] [-.277, -.155] [-.128, .226] [-.196, .453] [-.073, .296]

Bandwidth 306.46 306.46 306.46 306.46 306.46 0.263 0.306 0.374

Treatment Sample Size

174 174 174 49 427 85 103 115

Control Sample Size

355 355 339 480 102 126 142 226

Notes: the table presents the effects of the distance from the jurisdiction of Audiencia Real on the set of development outcomes and human capital formation in regression discontinuity design setup. Each Panel indicates the estimates of the effect of the proximity to the jurisdiction on each outcome. Local first-order polynomials are used to construct the confidence interval for the treatment effect of the colonial law on long-run development and human capital formation with the bias-corrected discontinuity estimator. Calonico-Cattaneo-Titiunik standard errors, computed from the conditional variance of the linear combinations of weighted linear least-squares estimators, are denoted in the parentheses.

31

Table 5: Stability of the Effect of Proximity to the Audiencia Jurisdiction Across Subsamples (1) (2) (3) (4) (5) (6) (7) (8) (9) Department-Level Human

Capital Formation Landholding

Concentration Household-Level Development Outcomes Household-

Level Human Capital

Outcomes

Department-Level

Human Capital

Formation Outcome Illiteracy

Rate Computer Literacy

Land Ownership

Access to Public Water

Network

Absence of Water Supply

Inside the House

Sanitary Facilities and Water Discharge

Fixed Telephone

Line

Cell Phone Ownership

Computer Ownership

Panel A: Northwest

Catamarca -.459*** (.111)

.259*** (.078)

-.069*** (.025)

.145* (.082)

-.549*** (.127)

.224*** (.067)

.690*** (.186)

.051 (.042)

.373*** (.100)

Jujuy -.386*** (.118)

.202*** (.078)

-.076*** (.027)

.084 (.075)

-.450*** (.128)

.148*** (.064)

.646*** (.200)

.013 (.028)

.293*** (.095)

Salta -.342*** (.114)

.192** (.079)

-.037 (.025)

.077 (.072)

-.451*** (.129)

.153*** (.070)

.561*** (.191)

.043 (.048)

.313*** (.103)

Tucumán -.378*** (.113)

.228*** (.077)

-.085*** (.025)

.114 (.074)

-.485*** (.127)

.186*** (.067)

.658*** (.193)

.042 (.044)

.345*** (.099)

Panel B: Gran Chaco

Chaco -.317*** (.113)

.257*** (.080)

-.092*** (.025)

.054 (.065)

-.470*** (.133)

.197*** (.069)

.715*** (.192)

.049 (.041)

.367*** (.102)

Formosa -.430*** (.112)

.271*** (.074)

-.078*** (.025)

.140* (.076)

-.531*** (.126)

.240*** (.061)

.751*** (.184)

.058 (.041)

.380*** (.097)

Santiago del Estero

-.393*** (.116)

.143*** (.063)

-.082*** (.027)

.050 (.074)

-.394*** (.125)

.110** (.054)

.553*** (.168)

.051 (.043)

.270*** (.088)

Panel C: Littoral

Corrientes -.287** (.121)

.263*** (.096)

-.146*** (.027)

.187* (.103)

-.490*** (.155)

.217*** (.087)

.769*** (.240)

.049 (.054)

.320*** (.119)

Entre Rios -.421*** (.115)

.245*** (.077)

-.078*** (.025)

.110 (.077)

-.502*** (.130)

.196*** (.067)

.688*** (.190)

.045 (.042)

.356** (.099)

Misiones -.581*** .325*** -.078*** .183** -.706*** .276*** .841*** .063 .474***

32

(.117) (.077) (.026) (.077) (.137) (.065) (.191) (.043) (.100) Panel D: Cuyo

La Rioja -.590*** (.110)

.354*** (.083)

-.064*** (.022)

.218*** (.085)

-.737*** (.126)

.290*** (.073)

.793*** (.213)

.048 (.047)

.508*** (.106)

Mendoza -.406*** (.113)

237*** (.077)

-.073*** (.025)

.113 (.076)

-.500*** (.128)

.198*** (.067)

.686*** (.189)

.042 (.042)

.355*** (.098)

San Juan -.440*** (.121)

.280*** (.081)

-.074*** (.025)

.109 (.081)

-.560*** (.138)

.215*** (.071)

.795*** (.201)

.054 (.045)

.430*** (.104)

San Luis -.406*** (.113)

.237*** (.076)

-.083*** (.025)

.125* (.076)

-.530*** (.127)

.207*** (.066)

.698*** (.188)

.048 (.042)

.350*** (.098)

Panel E: Pampas

Buenos Aires -.521*** (.113)

.233*** (.077)

-.069*** (.025)

.134* (.076)

-.525*** (.129)

.206*** (.066)

.687*** (.189)

.054 (.043)

.347*** (.098)

Buenos Aires City

-.417*** (.113)

.240*** (.076)

-.084*** (.025)

.125* (.076)

-.528*** (.127)

.202*** (.066)

.680*** (.189)

.045 (.042)

.355*** (.098)

Córdoba -.321*** (.119)

.218*** (.083)

-.087*** (.027)

.144* (.084)

-.404*** (.130)

.206*** (.073)

.632*** (.205)

.048 (.046)

.334*** (.105)

La Pampa -.367*** (.113)

.237*** (.076)

-.081*** (.025)

.125* (.076)

-.512*** (.127)

.203*** (.066)

.671*** (.189)

.043 (.042)

.349*** (.098)

Santa Fe -.353*** (.116)

.226*** (.079)

-.090*** (.026)

.149** (.077)

-.475*** (.136)

.192*** (.068)

.657*** (.190)

.045 (.044)

.339*** (.101)

Panel F: Patagonia

Chubut -.368*** (.113)

.226*** (.076)

-.082*** (.025)

.125* (.076)

-.482*** (.128)

.194*** (.066)

.643*** (.189)

.027 (.042)

.335*** (.098)

Neuquén -.339*** (.113)

.233*** (.076)

-.086*** (.025)

.128* (.076)

-.450*** (.128)

.196*** (.066)

.647*** (.189)

.044 (.042)

.347*** (.098)

Rio Negro -.350*** (.113)

.233*** (.076)

-.079*** (.025)

.127* (.076)

-.458*** (.128)

.192*** (.066)

.659*** (.189)

.039 (.042)

.340*** (.098)

Santa Cruz -.317*** (.114)

.242*** (.076)

-.081*** (.025)

.127* (.076)

-.418*** (.128)

.202*** (.066)

.682*** (.189)

.046 (.042)

.358*** (.098)

Tierra del Fuego

-.312*** (.114)

.190** (.078)

-.085*** (.025)

.141* (.078)

-.461*** (.129)

.171*** (.067)

.526*** (.191)

.015 (.042)

.282*** (.100)

Notes: the table presents the effects of the distance from the jurisdiction of Audiencia Real on the set of development outcomes and human capital formation in regression discontinuity design setup across multiple spatial subsamples. Each Panel reports the estimates of the effect of the proximity to the jurisdiction on each outcome by splitting the within-province department-level observations off the full sample. Local first-order polynomials are used to construct the confidence interval for the treatment effect of the colonial law on long-run development and human

33

capital formation with the bias-corrected discontinuity estimator. Calonico-Cattaneo-Titiunik standard errors, computed from the conditional variance of the linear combinations of weighted linear least-squares estimators, are denoted in the parentheses.

34

4.2.2 Stability of RD Estimates under Alternative Window Selection and Local Randomization

Does the uncertainty about the de facto colonial Audiencia jurisdiction matter for RD

estimates? We tackle the robustness of our RD estimates by re-estimating the effect of the proximity to the jurisdiction of the Audiencia Real to test whether the local randomization assumption holds. Because the department-level jurisdiction under Audiencia Real is “as if random” inside the treatment sample, the distribution of covariates and outcomes between treatment and control samples should be roughly the same in the absence of the treatment. Following Rosenbaum (2007) and Calonico et. al. (2015, 2016b), we employ a balance test on the set of outcomes and covariates to select the window where local randomization assumption might fail. Using the selected window, we replicate the core RD setup without spatial sample split-offs at 0.5 decimated point. For each covariate in the set of independent variables, except for the log population density, we find substantial differences in the mean values across alternative cutoffs (p-value =0.000) with the recommended window [7.141, 7.199] to perform and estimate the parameters in the RD setup. The core RD setup is replicated between the lower and upper bound of the window to test whether “as if” randomization assumption holds.

Table 6 reports the results on the size and stability of the RD estimates with alternative window selection around the neighborhood of the existing colonial border cutoff. In column (1), the evidence clearly suggests the effect of the proximity to the jurisdiction of Audiencia Real on the illiteracy rate is stable. The size of the effect tends to rise slowly until the border cutoff (7.17) and exhibits a tendency to decrease slightly afterwards. The point estimates highlight a substantial drop in the illiteracy rate outside the Audiencia’s jurisdiction. On average, the departments outside the jurisdiction tend to have lower illiteracy rate between 31.4 percent and 41.6 percent, respectively. In a similar fashion, in column (2), the effect on the computer literacy is similar although the size of the effect is noticeably larger left of the existing cutoff, and tends to decrease by a narrow margin right of the cutoff. In quantitative terms, the treated departments appear to be between 17 percent and 25.2 percent less computer literate than the departments outside the Audiencia’s jurisdiction. In column (3), the size of the effect on landownership rate appears to be both stable and comparable across alternative window selection. The effect remains essentially unchanged near the existing window and suggests that landownership rate in the non-treated departments drops between 7.9 percent and 10.9 percent.

In column (4), the point estimates readily imply that the effect of proximity to the Audiencia Real on the access to public water network is susceptible to the alternative window selection. In particular, the parameter estimates indicate that there is a sizeable effect below the existing cutoff whereas the effect slowly disappears at higher cutoffs. On the contrary, column (5) does not indicate such effect disparities for the absence of water supply inside the house. The effect of the proximity to Audiencia Real appears to be stable around the neighborhood of the existing cutoffs. It suggests that the departments outside the jurisdiction of the court tend to have between 37 percent and 51 percent lower likelihood of absence of water supply inside the house. For the sanitary facilities and water discharge, the likelihood for the non-treated departments in column (6) jumps to the level between 14.7 percent and 24 percent, respectively. In column (7), the alternative window selection does not seem to affect the size and stability of the effect of the proximity to the court on fixed-telephone line. In column (8), no evidence is found to support the notion that the proximity to the former colonial jurisdiction of Audiencia can explain the cross-department differences in cell phone ownership. The effect does seem to change at any alternative window locus. The column (9) confirm the strong and robust relationship between computer ownership and the proximity to the Audiencia Real. The departments outside the Audiencia’s jurisdiction have between 26.4 percent and 39.8 percent higher computer ownership rates compared to the treatment sample. In Appendix A3, we present the distribution of the effect across different window cutoffs graphically.

35

Table 6: Stability of RD Estimates and Alternative Window Selection under Local Randomization Department-Level Human Capital

Formation Landholding

Concentration Household-Level Development Outcomes Household-

Level Human Capital

Outcomes

Department-Level Human

Capital Formation

(1) (2) (3) (4) (5) (6) (7) (8) (9) Illiteracy Rate Computer

Literacy Land Ownership Access to

Public Water Network

Absence of Water Supply

Inside the House

Sanitary Facilities and

Water Discharge

Fixed Telephone Line

Cell Phone Ownership

Computer Ownership

Selected Window

7.140 -.314*** (.116)

.248*** (.087)

-.109*** (.025)

.270*** (.095)

-.371*** (.135)

.245*** (.078)

.634*** (.210)

.036 (.046)

.372*** (.108)

7.145 -.324*** (.116)

.241*** (.082)

-.096*** (.024)

.222*** (.086)

-.384*** (.133)

.225*** (.071)

.589*** (.194)

.037 (.042)

.363*** (.103)

7.150 -.401*** (.114)

.251*** (.080)

-.083*** (.024)

.201** (.083)

-.432*** (.132)

.233*** (.069)

.612*** (.192)

.041 (.041)

.394*** (.101)

7.155 -.416*** (.114)

.254*** (.081)

-.083*** (.024)

.201** (.084)

-.453*** (.132)

.235*** (.071)

.621*** (.195)

.042 (.042)

.398*** (.103)

7.160 -.409*** (.115)

.246*** (.079)

-.084*** (.025)

.183** (.081)

-.477*** (.130)

.225*** (.069)

.669*** (.192)

.045 (.042)

.388*** (.101)

7.165 -.409*** (.114)

.252*** (.079)

-.079*** (.024)

.169** (.079)

-.483*** (.130)

.219*** (.068)

.680*** (.192)

.044 (.042)

.384*** (.101)

7.170 -.393*** (.114)

.242*** (.078)

-.085*** (.025)

.154** (.079)

-.492*** (.129)

.212*** (.068)

.685*** (.191)

.045 (.042)

.366*** (.100)

7.175 -.391*** (.113)

.241*** (.076)

-.081*** (.025)

.127* (.076)

-.498*** (.127)

.202*** (.066)

.680*** (.187)

.045 (.042)

.356*** (.097)

7.180 -.399*** (.112)

.240*** (.075)

-.083*** (.025)

.116 (.074)

-.506*** (.127)

.194*** (.065)

.705*** (.186)

.053 (.042)

.360*** (.097)

7.185 -.376*** (.109)

.224*** (.073)

-.086*** (.024)

.148** (.076)

-.511*** (.126)

.172*** (.062)

.664*** (.182)

.059 (.041)

.340*** (.094)

7.190 -.363*** (.108)

.198*** (.071)

-.084*** (.024)

.129* (.074)

-.485*** (.125)

.158*** (.061)

.606*** (.179)

.056 (.040)

.306*** (.092)

7.195 -.327*** (.108)

.170** (.069)

-.087*** (.023)

.108 (.070)

-.472*** (.126)

.147*** (.058)

.551*** (.172)

.056 (.039)

.264*** (.089)

7.199 -.336*** (.108)

.170** (.069)

-.087*** (.023)

.107 (.070)

-.484*** (.126)

.147*** (.059)

.554*** (.174)

.057 (.040)

.265*** (.090)

Notes: the table presents the effects of the distance from the jurisdiction of Audiencia Real on the set of development outcomes and human capital formation in regression discontinuity design under alternative window selection. Local first-order polynomials are used to construct the confidence interval for the treatment effect of the colonial law on long-run development and human capital formation with the bias-corrected discontinuity estimator. Calonico-Cattaneo-Titiunik standard errors, computed from the conditional variance of the linear combinations of weighted linear least-squares estimators, are denoted in the parentheses.

36

4.2.3 Robustness of RD Estimates Against Alternative Explanations

Lastly, we address the robustness of our RD estimates against the alternative explanations. A drop in the set of human capital and long-run development outcomes in the Audiencia jurisdiction might not be driven by the persistence of legal and judicial institutions inherited from the centuries of Audiencia’s rule but could hinge on two alternative channels that could simultaneously change at the border cutoff. First, the effect of Audiencia Real might reflect the change in the disease environment and climatic conditions. Acemoglu et. al. (2001) exploit the differences in the mortality rates to estimate the effect of institutions and economic performance, and suggest that European colonizers adopted very different colonization policies in different colonies with notable institutional differences. In places where the colonizers faced high mortality rates, settlement had been very unlikely, and the colonizers were substantially more likely to set up extractive institutional framework. In places where the Europeans confronted low mortality rates, they were more likely to survive the disease environment. In the low-disease environment, colonizers were significantly more likely to establish inclusive institutions for a broad cross-section of society. And second, Glaeser et. al. (2004) point out to the theoretical ambiguity regarding the use of variation in settler mortality rates to estimate the effect of institutions on long-run development. In particular, the effects of colonial settlement may operate across multiple channels such as human capital, political institutions or something else. The Europeans who settled in the New World also brought with themselves human capital while the settler mortality rates may not tell us which channel matters. Glaeser et. al. (2004) findings suggest human capital might be a more important variable predicting long-run development than political institutions. Since the treatment and control samples in our setup differ markedly in terms of climatic and geographic conditions, the effect of Audiencia Real on human capital formation and long-run development might be a reflection of the disease environment and other observables different from the operative mechanism of the Audiencia Real. In Table 7, we assess the set of human capital formation and long-run development outcomes against the alternative explanations emphasizing the geographic and climatic conditions such as mean annual temperature, tropics indicator variable, and rainfall precipitation. Such conditions partially reflect the disease environment espoused by settler mortality rates. This particular kind of assessment is a rough outcome comparison across treatment and control samples under alternative forcing variables in the assignment mechanism. The set of beta coefficients across specifications suggests that the proximity to Audiencia Real dominates the potential confounding effect of geographic and climatic conditions. One standard deviation increase in the distance from Audiencia Real is associated with .2 S.D point decrease in illiteracy rates, .5 S.D point increase in computer literacy, .5 S.D point decrease in the absence of water supply inside the house, .4 S.D point increase in the likelihood of “in-house” sanitary facilities, .4 S.D increase in the ownership of fixed telephone line, .3 S.D point increase in cell phone ownership rate and .5 S.D point increase in computer ownership rate. We find no evidence of temperature and rainfall variation exhibiting a dominant effect on the set of human capital and development outcomes. At best, temperature and rainfall variation are weakly correlated with the set of outcomes. The departments located within the tropical geographic zone, which comprises 2 percent of the full sample, are somewhat more likely to experience less favorable human capital and development outcomes which partially reflects the adverse effects of the disease environment on human capital formation and long-run development. In terms of the relative importance, the size of the beta coefficient on the distance from Audiencia Real is significantly larger than the size of the beta coefficient on tropics or temperature variables which further advocates both the persistent effect and the relative dominance of the colonial law and institutions in explaining the differential paths of long-run development and human capital formation across Argentina. In Appendix A4 and Appendix A5, the weakness of

37

the relationship between temperature and rainfall precipitation and the set of outcomes is further demonstrated using both linear RD techniques as well as non-parametric local weighing techniques. Table 7: Human Capital Formation, Long-Run Development and Multiple Forcing Variables

Department-Level Human Capital

Formation

Landholding

Concentration

Household-Level Development Outcomes Household-Level

Human Capital

Outcomes

Department-Level

Human Capital

Formation

Illiteracy Computer

Literacy

Land

Ownership

Access to Public Water

Network

Absence of Water Supply

Inside the House

Sanitary Facilities and Water Discharge

Fixed Telephone

Line

Cell Phone Ownership

Computer Ownership

Panel A: Beta Coefficients

Log Distance

from Audiencia

-.288** .523***

(.092)

-.053

(.047)

.086

(.091)

-.524***

(.283)

.401***

(.074)

.437***

(.109)

.309***

(.042)

.511***

(.109)

Temperature .213 -.274***

(.022)

.210

(.005)

-.066

(.018)

.205

(.060)

-.180

(.023)

-.125

(.034)

.186**

(.004)

-.238

(.024)

Log

Precipitation

-.209 .182***

(.069)

.013

(.019)

-.113

(.044)

-.167

(.178)

.079

(.035)

.258**

(.153)

.039

(.019)

.189

(.088)

Tropics .070** -.0003

(.051)

.001

(.090)

.018

(.092)

-.273

(.246)

-.138***

(.063)

-.114*

(.275)

-.379***

(.088)

-.010

(.113)

# Observations 526 526 526 526 525 526 526 526 526

R2 0.21 0.46 0.06 0.03 0.39 0.33 0.33 0.34

# Department

Clusters

521 521 521 521 521 521 521 521 521

# Province

Clusters

24 24 24 24 24 24 24 24 24

# Region

Clusters

11 11 11 11 11 11 11 11

# Latitude

Clusters

429 429 429 429 429 429 429 429

Notes: the table reports the standardized effects of the multiple forcing variables in RD framework on the set of human capital and long-run development outcomes. Standard errors are adjusted for serially correlated stochastic disturbances and heteroskedastic error variance distribution into the department-level province-level, region-level and latitude-level clusters using Cameron-Gelbach-Miller non-nested multi-way clustering scheme for finite-sample adjustment of the empirical distribution function and cluster-robust parameter inference to remove the inconsistencies arising from biased OLS covariance matrix estimator and serially correlated intra-cluster residuals. Multiway cluster-robust standard errors are denoted in the parentheses for each empirical specification. Asterisk denote statistically significant parameter estimates at 10% (*), 5% (**) and 1% (***), respectively.

In Table 8, the set of falsification tests of our core RD estimates from Table 4 and Table

5 is presented. In the falsification setup, the distance from Audiencia Real is replaced by the mean annual temperature level, rainfall precipitation and distance from the 23rd parallel covariates as a alternative assignment mechanism and a potential channel that might confound the effect of the colonial law on human capital formation and long-run development. The departments are assigned to the artificial treatment and control samples by setting the cutoff in each covariate based on its mean value in Audiencia and non-Audiencia jurisdiction. The evidence readily suggests that human capital formation and long-run development outcomes only weakly depend on the climatic and geographic conditions, and on the tropical disease environment. In Panel A, the evidence highlights the weakness of mean annual temperature in isolating the effect on human capital formation and long-run development. The point estimates fail to indicate a potential causal channel from climatic conditions to human capital formation and long-run development even when we allow for the quadratic and cubic non-linearities left and right of the cutoff threshold.

38

Table 8: Falsification Test of RD Estimates Department-Level Human Capital

Formation Landholding

Concentration Household-Level Development Outcomes Household-Level

Human Capital Outcomes

Department-Level Human

Capital Formation

(1) (2) (3) (4) (5) (6) (7) (8) (9)

Illiteracy Rate Computer Literacy

Land Ownership Access to Public Water Network

Absence of Water Supply

Inside the House

Sanitary Facilities and

Water Discharge

Fixed Telephone Line

Cell Phone Ownership

Computer Ownership

Panel A: Assignment Mechanism: Mean Annual Temperature Point Estimate

(1st-Order Polynomial) .313

(.241) -.091 (.115)

.004 (.044)

-.148 (.104)

.155 (.360)

.006 (.070)

-.196 (.244)

-.060 (.058)

-.106 (.147)

Point Estimate (2nd-Order

Polynomial)

-.458 (.403)

.071 (.166)

-.058 (.068)

-.059 (.114)

-.556 (.557)

.022 (.077)

.396 (.334)

.082 (.066)

.193 (.212)

Point Estimate (3rd-Order Polynomial)

-.462 (.428)

.123 (.188)

-.041 (.084)

-.156 (.128)

-.721 (.606)

.101 (.110)

.527 (.425)

.024 (.081)

.300 (.244)

Panel B: Assignment Mechanism: Log Precipitation Point Estimate

(1st-Order Polynomial) .165

(.283) .345

(.225) .007

(.027) .077

(.150) -.169 (.416)

.226 (.171)

.241 (.372)

-.025 (.057)

.348 (.274)

Point Estimate (2nd-Order

Polynomial)

.158 (.367)

.467* (.255)

-.050 (.043)

.073 (.170)

-.252 (.497)

.352* (.209)

.387 (.476)

-.016 (.080)

.487 (.318)

Point Estimate (3rd-Order Polynomial)

.235 (.387)

.453* (.255)

-.022 (.041)

.049 (.182)

-.265 (.573)

.366* (.213)

.392 (.467)

-.010 (.084)

.468 (.315)

Panel C: Assignment Mechanism: Log Distance from 23rd Parallel Point Estimate

(1st-Order Polynomial) .311

(.263) -.156** (.064)

.057** (.026)

.019 (.071)

.121 (.287)

.013 (.034)

-.431*** (.174)

-.002 (.043)

-.252*** (.089)

Point Estimate (2nd-Order

Polynomial)

.114 (.301)

-.081 (.075)

.063 (.029)

-.204 (.077)

.165 (.297)

.041 (.057)

-.330* (.195)

.034 (.054)

-.067 (.118)

Point Estimate (3rd-Order Polynomial)

-.700* (.410)

.121 (.098)

.027 (.035)

.0008 (.107)

-.252 (.383)

.065 (.066)

.544** (.280)

.075 (.063)

.151 (.142)

Notes: the table reports RD estimates of the temperature level, precipitation and distance from the tropical zone on human capital formation and long-run development. The cutoff value for each assignment mechanism is determined by the values on each side of the border of the former Audiencia jurisdiction. Local first-order polynomials are used to construct the confidence interval for the treatment effects of each alternative mechanism on long-run development and human capital formation with the robust discontinuity estimator. Calonico-Cattaneo-Titiunik standard errors, computed from the conditional variance of the linear combinations of weighted linear least-squares estimators, are denoted in the parentheses.

39

A similar pattern is indicated in Panel B where the log rainfall precipitation is set as the underlying assignment mechanism. The evidence clearly suggests that even when non-linearites in the outcomes’ distribution are allowed on both sides of the former colonial border, rainfall precipitation fails to explain the differential outcomes between former Audiencia and non-Audiencia jurisdiction. In Panel C, setting the distance from the tropical climatic zone at 23rd parallel unveils slightly persistent effects of the disease environment on the outcome levels across treatment and control samples. In particular, the departments within the 23rd parallel have 15.6 percent lower rate of computer literacy, 5.7 percent increase in the rate of landownership, 43.1 percent fewer fixed telephone lines, and 25.2 percent lower rate of computer ownership compared to the non-Audiencia departments. The replicated RD estimates indicate that the proximity to tropical climatic zone as a proxy for the disease environment cannot be effectively ruled out as an alternative channel distinctive from the persistent effect of Audiencia Real. Our results arguably indicate a relatively greater importance of the Audiencia Real as a source of colonial law and institutions compared to the disease environment and climatic conditions in explaining the differential human capital and development outcomes across Argentina. 5. Conclusion In this paper, we examine the persistent effects of colonial law on human capital formation and long-run development across 527 departments in Argentina using the survey data from 2010 Population and Housing Census (Censo Nacional de Población, Hogares y Viviendas). Our identification strategy exploits the distance from the seat of Audiencia Real that held jurisdiction over the areas of the Rio de la Plata governorate in the Spanish empire for more than two centuries, as a plausibly exogenous source of variation in long-run development and human capital formation. We employ the regression discontinuity design with quasi-local randomization techniques and assign the departments by the distance from the seat of Audiencia

Real into treatment and control samples with the normalized colonial border point at the cutoff. Using a geo-referenced spatial dataset, we reconstruct the department-level former colonial borders in the modern administrative division of Argentina in the geographic discontinuity framework (Keele and Titiunik 2014). Our results show that after more than two centuries of demise, the departments that fell under the jurisdiction of Audiencia Real tend to have substantially worse human capital formation and development outcomes down to the present day. Our results are established from the jurisdiction-level colonial border specification adjusted for modern geo-referenced boundaries identified from the geographic discontinuity. In our preferred specification, departments historically under the jurisdiction of Audiencia Real have 39.6 percent higher rate of illiteracy, 24 percent lower rate of computer literacy and 8 percentage points higher rate of landownership. The households in the departments outside the colonial Audiencia jurisdiction are 20 percentage points more likely to have sanitary facilities and water discharge, 12 percentage points less likely to lack water supply inside the house, 68 percentage points more likely to possess a fixed-telephone line, 4 percentage points more likely to own a cell phone, and 24 percentage points more likely to own a computer. The obtained estimates from regression discontinuity design are statistically significant at conventional 5% level. Our results are robust to a battery of specifications checks, hold across multiple subsamples, and pass a number of placebo checks. We show that the effect of the proximity to Audiencia Real does not appear to be driven by the alternative mechanisms such as the climate and disease environment, and the distance from the equator that might simultaneously confound the effect of the proximity to Audiencia Real on human capital formation and long-run development. Our approach conveys several advantages.

40

The households across the departments in Argentina now live under common institutional environment but have been historically exposed to different sets of legal, political and economic institutions that shaped long-term development paths. In particular, the areas in the south of Argentina were never integrated into the Spanish colonial empire, and never fell under the jurisdiction of Audiencia Real, and were never exposed to the sets of rent-extracting institutions that pervaded the colonized north and northwest of Argentina such as hacienda, repatrimento, debt servitude, office selling and lock-in of economic opportunities for the non-elites. The differences in the historical exposure to the legal institutions established by the Audiencia Real overlap with the influence and ability of the colonial court to enforce the institution in its jurisdiction. Prior work by Mirow (2004) indicates that the influence of Audiencia was very powerful in its close radius but tended to decline at greater distances. Our results show that the departments in the closest proximity to the seat of Audiencia Real tend to have markedly lower human capital formation and development outcomes that last down to the present day. The results from our jurisdiction-level border specification clearly suggest that the influence of proximity to Audiencia Real on the whole set of outcomes disappears in the departments not integrated in the Spanish colonial empire. We employ a series of placebo tests and falsification checks which support and corroborate our main findings. From a general perspective, our investigation unravels persistent effects of historically determined legal institutions on human capital formation and long-run development even after more than two centuries of common institutional environment. The formal institutional framework and its de facto enforcement can leave behind a legacy of bad institutional choices, and repeated attempts to seize de facto political power by the entrenched local elites when their de jure political power is undermined. Such institutional legacy of high transaction costs and insecure property rights is adverse to human capital formation as it sets to encourage rent seeking instead of productive economic activity and may have profound long-lasting implications for economic and social development that tend to persist even after such institutions are overhauled. Exploiting the variation in the volume and types of legal rules within audiencias across Latin America warrants further understanding, and as a potential source of institutional and economic development is a fruitful research area to pursue.

41

Refererences

Abad, L. A., Davies, E., & Van Zanden, J. L. (2012). Between conquest and independence: Real wages

and demographic change in Spanish America, 1530–1820. Explorations in Economic History 49(2), 149-166. Abad, L. A., & van Zanden, J. L. (2016). Growth under extractive institutions? Latin American per

capita GDP in colonial times. The Journal of Economic History 76(4), 1182-1215. Acemoglu, D., Cantoni, D., Johnson, S., Robinson J.A. (2011). The consequences of radical reform: The

French revolution. American Economic Review 101(7): 3286-3307. Acemoglu, D., & Dell, M. (2010). Productivity differences between and within countries. American

Economic Journal: Macroeconomics 2(1): 169-188. Acemoglu, D., Johnson, S., Robinson, J.A. (2001). The colonial origins of comparative development: An

empirical investigation. American Economic Review 91(5): 1369-1401. Acemoglu, D., Johnson, S., Robinson, J.A. (2002). Reversal of fortune: Geography and institutions in

the making of the modern world income distribution. Quarterly Journal of Economics 117(4): 1231-1294. Acemoglu, D., Johnson, S., & Robinson, J. (2005a). The rise of Europe: Atlantic trade, institutional

change, and economic growth. The American Economic Review 95(3): 546-579. Acemoglu, D., Johnson, S., Robinson, J. A., (2005b). Institutions as a Fundamental Cause of Long-Run

Growth. In Aghion, P, Durlauf, S. (eds.), Handbook of Economic Growth, Vol. 1A, 386-464. Acemoglu D. & Robinson, J.A. (2006). De Facto Political Power and Institutional Persistence. American

Economic Review, Papers and Proceedings 96(2): 325-330. Acemoglu, D., & Robinson, J. A. (2013). Why nations fail: The origins of power, prosperity, and

poverty. Crown Business. Acemoglu, D., Gallego, F. A., & Robinson, J. A. (2014). Institutions, human capital, and development.

Annual Review of Economics 6(1): 875-912. Alesina, A., & Fuchs-Schündeln, N. (2007). Good-bye Lenin (or not?): The effect of communism on

people's preferences. The American Economic Review 97(4): 1507-1528. Algan, Y., & Cahuc, P. (2010). Inherited trust and growth. The American Economic Review 100(5):

2060-2092. Andrien, K. J. (1982). The sale of fiscal offices and the decline of royal authority in the Viceroyalty of

Peru, 1633-1700. The Hispanic American Historical Review 62(1): 49-71. Andrien, K. J. (1984). Corruption, inefficiency, and imperial decline in the seventeenth-century

viceroyalty of Peru. The Americas, 1-20. Arellano, M. (1987). Computing Robust Standard Errors for Within‐groups Estimators. Oxford bulletin

of Economics and Statistics 49(4): 431-434. Arthur, W. B. (1989). Competing technologies, increasing returns, and lock-in by historical events. The

Economic Journal 99(394), 116-131. Bakewell, P. J. (1988). Silver and entrepreneurship in seventeenth-century Potosí: the life and times of

Antonio López de Quiroga. Univ of New Mexico Press. Banerjee, A., & Iyer, L. (2005). History, institutions, and economic performance: the legacy of colonial

land tenure systems in India. The American Economic Review 95(4): 1190-1213.

42

Banerjee, A., Iyer, L., & Somanathan, R. (2005). History, social divisions, and public goods in rural

India. Journal of the European Economic Association 3(2‐3), 639-647. Baskes, J. (2000). Indians, merchants, and markets: a reinterpretation of the Repartimiento and Spanish-

Indian economic relations in colonial Oaxaca, 1750-1821. Stanford University Press. Becker, S. O., Boeckh, K., Hainz, C., & Woessmann, L. (2016). The empire is dead, long live the

empire! Long‐run persistence of trust and corruption in the bureaucracy. The Economic Journal 126(590): 40-74. Becker, S.O., Wößmann, L. (2009). Was Weber Wrong? A Human Capital Theory of Protestant

Economic History. Quarterly Journal of Economics 124(2): 531-569. Bella, I. S., Rementería, C. D., & de la Hera Pérez-Cuesta, A. (1992). Historia del derecho indiano.

Madrid: Fundación MAPFRE. Belloc, M., & Bowles, S. (2013). The persistence of inferior cultural-institutional conventions. The

American Economic Review 103(3): 93-98. Benton, L. (1999). Colonial law and cultural difference: jurisdictional politics and the formation of the

colonial state. Comparative Studies in Society and History 41(3): 563-588. Bertrand, M., Duflo, E., & Mullainathan, S. (2004). How much should we trust differences-in-

differences estimates? The Quarterly Journal of Economics 119(1): 249-275. Binger, B. R., & Hoffman, E. (1989). Institutional persistence and change: The question of efficiency.

Journal of Institutional and Theoretical Economics (JITE)/Zeitschrift für die gesamte Staatswissenschaft, 67-84. Boettke, P. J., Coyne, C. J., & Leeson, P. T. (2008). Institutional stickiness and the new development

economics. American Journal of Economics and Sociology 67(2), 331-358. Bruhn, M., & Gallego, F. A. (2012). Good, bad, and ugly colonial activities: do they matter for

economic development?. Review of Economics and Statistics 94(2): 433-461. Bulmer-Thomas, V. (2003). The economic history of Latin America since independence. Cambridge

University Press. Burkholder, M. A., & Chandler, D. S. (1977). From impotence to authority: the Spanish Crown and the

American audiencias, 1687-1808. University of Missouri Press. Cameron, A. C., Gelbach, J. B., & Miller, D. L. (2011). Robust inference with multiway clustering.

Journal of Business & Economic Statistics 29(2): 238-249. Campbell, L. G. (1972). A colonial establishment: Creole domination of the audiencia of Lima during

the late eighteenth century. The Hispanic American Historical Review 52(1), 1-25. Cameron, A. C., Trivedi, P.K. (2005). Microeconometrics: Methods and Applications. New York, NY:

Cambridge University Press. Cantoni, D. (2015). The economic effects of the Protestant Reformation: testing the Weber hypothesis in

the German lands. Journal of the European Economic Association 13(4): 561-598. Calonico, S., Cattaneo, M. D., & Titiunik, R. (2014a). Robust Nonparametric Confidence Intervals for

Regression‐Discontinuity Designs. Econometrica 82(6): 2295-2326. Calonico, S., Cattaneo, M. D., & Titiunik, R. (2014b). Robust data-driven inference in the regression-

discontinuity design. Stata Journal 14(4), 909-946.

43

Calonico, S., Cattaneo, M. D., & Titiunik, R. (2015). Optimal data-driven regression discontinuity plots. Journal of the American Statistical Association 110(512), 1753-1769.

Calonico, S., Cattaneo, M. D., Farrell, M. H., & Titiunik, R. (2016). Regression discontinuity designs

using covariates. Working Paper, University of Michigan. Cattaneo, M. D., Frandsen, B. R., & Titiunik, R. (2015). Randomization inference in the regression

discontinuity design: An application to party advantages in the US Senate. Journal of Causal Inference 3(1): 1-24. Cattaneo, M. D., Titiunik, R., & Vazquez-Bare, G. (2016a). Comparing inference approaches for rd

designs: A reexamination of the effect of head start on child mortality. Journal of Policy Analysis and Management, Forthcoming.

Cattaneo, M. D., Titiunik, R., Vazquez-Bare, G., & Keele, L. (2016b). Interpreting regression

discontinuity designs with multiple cutoffs. The Journal of Politics 78(4), 1229-1248. Cattaneo, M. D., Titiunik, R., & Vazquez-Bare, G. (2016c). Inference in regression discontinuity designs

under local randomization. Stata Journal 16(2): 331-367. Cebrián, A. M. (1976). Venta y beneficios de los corregimientos peruanos. Revista de Indias, 36, 143-

144. Cebrián, A. M. (1977). El corregidor de indios y la economía peruana del siglo XVIII:(los repartos

forzosos de mercancias). Madrid: Editorial CSIC-CSIC Press. Chanda, A., Cook, C.J., Putterman, L. (2014). Persistence of Fortune: Account for Population

Movements, There Was no Post-Columbian Reversal. American Economic Journal: Macroeconomics 6(3): 1-28. Clark, G., & Gray, R. (2014). Geography is not destiny: geography, institutions and literacy in England,

1837–63. Oxford Economic Papers 66(4), 1042-1069. Cleveland, W. S. (1979). Robust locally weighted regression and smoothing scatterplots. Journal of the

American Statistical Association 74(368): 829-836. Coatsworth, J. H. (2005). Structures, endowments, and institutions in the economic history of Latin

America. Latin American Research Review 40(3): 126-144. Coase, R. H. (1984). The new institutional economics. Zeitschrift für die gesamte

Staatswissenschaft/Journal of Institutional and Theoretical Economics, (H. 1), 229-231. Coate, S., & Morris, S. (1999). Policy persistence. The American Economic Review 89(5): 1327-1336. Coatsworth, J. H. (2008). Inequality, institutions and economic growth in Latin America. Journal of

Latin American Studies 40(03), 545-569. Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika 16(3), 297-

334. Cutter, C. R. (1986). The Protector de Indios in Colonial New Mexico, 1659-1821. University of New

Mexico Press. Cutter, C. R. (1998). The administration of law in colonial New Mexico. Journal of the Early Republic

18(1): 99-115. Cutter, C. (1999). The legal culture of Spanish America on the eve of independence. Zimmerman,

Judicial Institutions in Nineteenth Century Latin America, 9.

44

David, P.A. (1994). Why are the Institutions the Carriers of History? Path Dependence and the Evolution of Conventions, Organizations and Institutions. Structural Change and Economic Dynamics 5(2): 205-220.

Davis, P. (2006). Estimating Multi-Way Error Components Models with Unbalanced Panel Structures.

Journal of Econometrics 106(1): 67-95. de Velasco, J. A. M. (1983). Relación y documentos de gobierno del virrey del Perú, José A. Manso de

Velasco, conde de Superunda (1745-1761) (Vol. 10). Madrid: Editorial CSIC-CSIC Press. del Castillo, G. C. (1953). Reorganización de la Hacienda virreinal peruana en el siglo XVIII. Anuario

de historia del derecho español (23): 329-370. Dell, M. (2010). The persistent effects of Peru's mining mita. Econometrica 78(6), 1863-1903. Diamond, J. (1997). Guns, Germs, and Steel: The Fates of Human Societies. New York, NY: W.W.

Norton. Dory-Garduño, J. E. (2013). The Forging of Castilian Law: Land Disputes Before the Royal Audiencia

and the Transmission of a Legal Tradition (Doctoral dissertation). University of New Mexico. Eakin, M.C. (2007). The History of Latin America: Collision of Cultures. London: Palgrave Macmillan. Easterly, W. (2007). Inequality does cause underdevelopment: Insights from a new instrument. Journal

of Development Economics 84(2): 755-776. Easterly, W. & Levine, R. (2003). Tropics, Germs and Crops: How Endowments Influence Economic

Development. Journal of Monetary Economics 50(1): 3-39. Elliot, J. H. (1963). Imperial Spain, 1469-1716. New York: St. Martin's Press. Elliott, J. H. (1967). A provincial Aristocracy: the Catalán ruling class in the sixteenth and seventeenth

centuries. Homenaje a Jaime Vicens Vives. Engerman, S. L., & Sokoloff, K. L. (1997). Factor endowments, institutions, and differential paths of

growth among new world economies. In: Haber, S. (ed.). How Latin America Fell Behind (260-304), Stanford University Press.

Sokoloff, K. L., & Engerman, S. L. (2000). History lessons: Institutions, factors endowments, and paths

of development in the new world. The Journal of Economic Perspectives 14(3): 217-232. Eicker, F. (1967). Limit theorems for regressions with unequal and dependent errors. In Proceedings of

the fifth Berkeley Symposium on Mathematical Statistics and Probability (Vol. 1, No. 1, pp. 59-82). Engerman, S. L., & Sokoloff, K. L. (2005). The evolution of suffrage institutions in the new world. The

Journal of Economic History 65(04): 891-921. Engerman, S. L., & Sokoloff, K. L. (2012). Economic development in the Americas since 1500:

endowments and institutions. Cambridge University Press. Fisher, L. E. (1926). Viceregal administration in the Spanish-American colonies (Vol. 15). University of

California Press. Fukuyama, F. (2015). Why is democracy performing so poorly?. Journal of Democracy 26(1), 11-20. Galor, O., Moav, O., & Vollrath, D. (2009). Inequality in landownership, the emergence of human-

capital promoting institutions, and the great divergence. The Review of economic Studies 76(1): 143-179.

45

Glaeser, E. L., La Porta, R., Lopez-de-Silanes, F., & Shleifer, A. (2004). Do institutions cause growth? Journal of Economic Growth 9(3), 271-303.

Golte, J. (1980). Repartos y rebeliones. Tupac Amaru y las contradicciones de la economía colonial.

Lima: Instituto de Estudios Peruanos. Gorodnichenko, Y., Roland, G. (2011). Which Dimensions of Culture Matter for Long-Run Growth?

American Economic Review 101(3): 492-498. Grafe, R., & Irigoin, M. A. (2006). The Spanish Empire and its legacy: fiscal redistribution and political

conflict in colonial and post-colonial Spanish America. Journal of Global History 1(02): 241-267. Grafe, R., & Irigoin, A. (2012). A stakeholder empire: the political economy of Spanish imperial rule in

America1. The Economic History Review 65(2): 609-651. Greif, A. (1989). Reputation and Coalitions in Medieval Trade: Evidence on the Maghribi Traders.

Journal of Economic History 49(4): 857-882. Guardado, J. (2016). Office-Selling, Corruption and Long-Term Development in Peru. Georgetown

University, Mimeo. Guiso, L., Sapienza, P., Zingales, L. (2006). Does Culture Affect Economic Outcomes? Journal of

Economic Perspectives 20(2): 23-48. Hall, P. A., & Soskice, D. (Eds.). (2001). Varieties of capitalism: The institutional foundations of

comparative advantage. OUP Oxford. Hanke, L., & Mendoza, L. (1965). Historia de la villa imperial de Potosi. Edicion de Lewis Hanke y

Gunnar Mendoza. Providence, Brown University Press. Haber, S. (2013). Political institutions and economic growth in Latin America: essays in policy, history,

and political economy. Hoover Institution Press. Hall, R.E., Jones, C.I. (1999). Why Do Some Countries Produce So Much More Output Per Worker

Than Others? The Quarterly Journal of Economics 114(1): 83-116. Hansen, C. B. (2007). Asymptotic properties of a robust variance matrix estimator for panel data when T

is large. Journal of Econometrics 141(2): 597-620. Haring, C. H. (1952). The Spanish Empire in America. Oxford University Press. Helland, E., & Klick, J. (2011). Legal origins and empirical credibility. In: Faure, M, Smits, J. (eds.)

Does law matter (99-114), Antwerpen: Intersentia. Hodgson, G.M. (2006). What Are Institutions? Journal of Economic Issues 40(1): 1-25. Huber, P.J. (1967). The Behavior of Maximum Likelihood Estimates under Non-Standard Conditions.

Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability 1, 223-231. Huntington, S. (1968). Political Order in Changing Societies. Yale University Press, New Haven. Irigoin, A., & Grafe, R. (2008). Bargaining for absolutism: a Spanish path to nation-state and Empire

Building. Hispanic American Historical Review 88(2), 173-209. Islam, R., & Montenegro, C. E. (2002). What determines the quality of institutions? Mimeo Jha, S. (2013). Trade, institutions, and ethnic tolerance: Evidence from South Asia. American Political

Science Review 107(04): 806-832.

46

Kahle, L. G. (1951). The Spanish Colonial Judiciary. The Southwestern Social Science Quarterly, 26-37. Keele, L.J., Titiunik, R. (2014). Geographic boundaries as regression discontinuities. Political Analysis

23(1): 127-155. Kezdi, G. (2004). Robust Standard Errors Estimation in Fixed-Effects Models. Hungarian Statistical

Review, Special Number 9, 95-116. Klick, J. (2010). The Perils of Empirical Work on Institutions: Comment. Journal of Institutional and

Theoretical Economics (JITE)/Zeitschrift für die gesamte Staatswissenschaft 166(1): 166-170. Landes, D.S. (1998). The Wealth and Poverty of Nations: Why Some Are So Rich and Some So Poor.

New York, NY: W.W. Norton & Company. Liang, K. Y., & Zeger, S. L. (1986). Longitudinal data analysis using generalized linear models.

Biometrika, 13-22. Mariscal, E., & Sokoloff, K. L. (2000). Schooling, Suffrage, and the Persistence of Inequality in the

Americas, 1800-1945. In: Haber, S. (ed.) Political Institutions and Economic Growth in Latin America: Essays in Policy, History, and Political Economy (159-217), Stanford University Press.

Maseland, R. (2013). Parasitical Cultures? The Cultural Origins of Institutions and Development.

Journal of Economic Growth 18(2): 109-136. Mitchener, K. J., & McLean, I. W. (2003). The productivity of US states since 1880. Journal of

Economic Growth 8(1): 73-114. Pfeffermann, D., & Nathan, G. (1981). Regression analysis of data from a cluster sample. Journal of the

American Statistical Association 76(375), 681-689. Porta, R. L., Lopez-de-Silanes, F., Shleifer, A., & Vishny, R. W. (1998). Law and finance. Journal of

Political Economy 106(6): 1113-1155. Lange, M., Mahoney, J., & Vom Hau, M. (2006). Colonialism and Development: A Comparative

Analysis of Spanish and British Colonies. American Journal of Sociology, 111(5): 1412-1462. Lockhart, J., & Schwartz, S. B. (1983). Early Latin America: A History of Colonial Spanish America

and Brazil (Vol. 46). Cambridge University Press. Lynch, J. (1992). The institutional framework of colonial Spanish America. Journal of Latin American

Studies 24(S1): 69-81. Mansilla, R.E. (1985). Las reformas de Carlos III y la reestructuración de la Hacienda americana. Quinto

Centenario, 8, 61. Milanovic, B., Lindert, P. H., & Williamson, J. G. (2011). Pre‐industrial inequality. The Economic

Journal 121(551): 255-272. Miller, J. M. (2003). A typology of legal transplants: using sociology, legal history and argentine

examples to explain the transplant process. The American Journal of Comparative Law 51(4): 839-885. Mirow, M. C. (2004). Latin American law: A history of private law and institutions in Spanish America.

University of Texas Press. Mirow, M. C. (2014). The Age of Constitutions in the Americas. Law and History Review 32(02): 229-

235.

47

Mokyr, J. (1990). The Levers of Riches: Technological Creativity and Economic Progress. New York: Oxford University Press.

Moulton, B.R. (1986) Random Group Effects and the Precision of Regression Estimates. Journal of

Econometrics 32(3): 385-397. Moulton, B.R. (1990). An Illustration of the Pitfall in Estimating the Effects of Aggregate Variables on

Micro Units. Review of Economics and Statistics 72(2): 334-338. Mukerjee, A. (2008). La negociación de un compromiso: la mita de las minas de plata de San Agustín de

Huantajaya, Tarapacá, Perú (1756-1766). Bulletin de l'Institut français d'études andines, 37 (1): 217-225. North, D. C. (1989). Institutions and economic growth: An historical introduction. World Development

17, 1319-1332. North, D. C. (1990). Institutions, institutional change and economic performance. Cambridge University

Press. North, D. С. (1991). Institutions. The Journal of Economic Perspectives 5(1): 97-112. North, D. C., & Weingast, B. R. (1989). Constitutions and commitment: the evolution of institutions

governing public choice in seventeenth-century England. The Journal of Economic History 49(04): 803-832. Nunn, N. (2008). The long-term effects of Africa's slave trades. The Quarterly Journal of Economics

123(1): 139-176. Nunn, N. (2009). The importance of history for economic development. Annual Review of Economics

1(1), 65-92. Ostrom, E. (2005). Understanding Institutional Diversity. Princeton, NJ: Princeton University Press. Parry, J. H. (1948). The Audiencia of New Galicia in the Sixteenth Century: A Study in Spanish

Colonial Government. CUP Archive. Pepper, J. V. (2002). Robust inferences from random clustered samples: an application using data from

the panel study of income dynamics. Economics Letters 75(3): 341-345. Putnam, R. (1993). Making Democracy Work: Civic Traditions in Modern Italy. Princeton University

Press. Putterman, L. & Weil, D.N. (2010). Post-1500 Population Flows and the Long-Run Determinants of

Economic Growth and Inequality. Quarterly Journal of Economics 125(4): 1627-1682. de la Escosura, L. P. (2009). Lost Decades? Economic Performance in Post-Independence Latin

America. Journal of Latin American Studies 41(02), 279-307. Rajan, R. G., & Zingales, L. (2006). The persistence of underdevelopment: Institutions, human capital or

constituencies. Availble at Social Science Research Network (SSRN). Remacha, J. R. (1994). Traces of the Spanish Legal System in New Mexico. New Mexico Historical

Review 69(3): 281-295. Rodríguez, A. D. (1994). Manual de historia del derecho indiano. Universidad Nacional Autónoma de

México. Roland, G. (2004). Understanding Institutional Change: Fast-Moving and Slow-Moving Institutions.

Studies in Comparative International Development 38(4): 109-131.

48

Romero, F. M. (1978). El" beneficio" de oficios públicos con jurisdicción en Indias. Notas sobre sus orígenes. Anuario de estudios americanos 35, 1-67.

Rosenbaum, P. R. (2007). Interference between units in randomized experiments. Journal of the

American Statistical Association 102(477), 191-200. Gallup, J. L., Sachs, J. D., & Mellinger, A. D. (1999). Geography and Economic Development.

International Regional Science Review 22(2), 179-232. Salazar, O. R. (1983). Anotaciones al funcionamiento de la Real Hacienda en el Nuevo Reino de

Granada. Siglo XVIII. Anuario colombiano de historia social y de la cultura (11): 71-88. Schäfer, H. B., & Wulf, A. J. (2014). Jurists, Clerics, and Merchants: The Rise of Learned Law in

Medieval Europe and its Impact on Economic Growth. Journal of Empirical Legal Studies 11(2), 266-300. Soberanes Fernández, J. L. (1980). Los tribunales de la Nueva España. Antología, México, UNAM. Spolaore, E., & Wacziarg, R. (2013). How deep are the roots of economic development?. Journal of

Economic Literature 51(2), 325-369. Summerhill, W. (2010). Colonial institutions, slavery, inequality, and development: Evidence from São

Paulo, Brazil. Availible at Social Science Research Network (SSRN). Tabellini, G. (2010). Culture and Institutions: Economic Development in the Regions of Europe. Journal

of the European Economic Association 8(4): 677-716. Tapia, Á. S. (2009). ¿ Corrupción o necesidad?: la venta de cargos de gobierno americanos bajo Carlos

II (1674-1700) (Vol. 13). Madrid: Editorial CSIC-CSIC Press. Taylor, M. M. (2006). Beyond judicial reform: courts as political actors in Latin America. Latin

American Research Review 41(2): 269-280. Van Zanden, J.L., Buringh, E., Bosker, M. (2012). The Rise and Decline of European Parliaments, 1188-

1789. The Economic History Review 65(3): 835-861. Veliz, C. (2014). The centralist tradition of Latin America. Princeton University Press. Villena, G. L. (1976). La memorable crisis monetaria de mediados del siglo XVII y sus repercusiones en

el virreinato del Perú. Anuario de Estudios Americanos 33, 579-639. White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for

heteroskedasticity. Econometrica 48(4): 817-838. White, H. (1984). Asymptotic Theory for Econometricians. University of San Diego Press. Williamson, O., & Masten, S. (1995). Transaction cost economics. Edward Elgar Publishing. Williamson, J. G. (1999). Real wages, inequality and globalization in Latin America before 1940.

Revista de Historia Económica/Journal of Iberian and Latin American Economic History (Second Series), 17(S1), 101-142.

Williamson, O. E. (2000). The new institutional economics: taking stock, looking ahead. Journal of

Economic Literature 38(3), 595-613. Wooldridge, J. M. (2003). Cluster-sample methods in applied econometrics. The American Economic

Review 93(2), 133-138.

49

Young, H. P. (2001). Individual strategy and social structure: An evolutionary theory of institutions. Princeton University Press.

50

Appendix A1: Proximity to the Jurisdiction of the Audiencia Real, Human Capital Formation, and Long-Run Development – Linear Regression

(a) Illiteracy Rate (b) Computer Literacy (c) Land Ownership

(d) Access to Public Water Network (e) Sanitary Facilities and Water Discharge (f) Absence of Water Supply

(g) Fixed Telephone Line (h) Cell Phone Ownership (i) Computer Ownership

-2-1

01

23

Lo

g Illi

tera

cy R

ate

-3000 -2000 -1000 0 1000Normalized Distance to Colonial Border

22

.53

3.5

44.5

Lo

g C

om

pu

ter

Litera

cy R

ate

-3000 -2000 -1000 0 1000Normalized Distance to Colonial Border

3.5

44.5

Lo

g L

and

Ow

ners

hip

-3000 -2000 -1000 0 1000Normalized Distance to Colonial Border

02

46

Access to P

ublic

Wa

ter

Ne

twork

-3000 -2000 -1000 0 1000Normalized Distance from the Colonial Border

23

45

Lo

g S

an

ita

ry F

acili

tie

s a

nd

Wa

ter

Dis

ch

arg

e

-3000 -2000 -1000 0 1000Normalized Distance to Colonial Border

-20

24

6Lo

g H

ouse

ho

lds w

ithou

t W

ate

r S

upp

ly in

sid

e the

Hou

se

-3000 -2000 -1000 0 1000Normalized Distance to Colonial Border

51

-20

24

6L

og F

ixed

Tele

pho

ne

Lin

e

-3000 -2000 -1000 0 1000Normalized Distance from Colonial Border

2.5

33.5

44

.5Lo

g C

ell

Ph

on

e O

wn

ers

hip

-3000 -2000 -1000 0 1000Normalized Distance to Colonial Border

12

34

5Lo

g C

om

pu

ter

Ow

ners

hip

-3000 -2000 -1000 0 1000Normalized Distance to Colonial Border

52

Appendix A2: Proximity to the Jurisdiction of the Audiencia Real, Human Capital Formation, and Long-Run Development – Local Kernel-Weighted Non-Parametric Regression

(a) Illiteracy Rate (b) Computer Literacy (c) Land Ownership

(d) Access to Public Water Network (e) Sanitary Facilities and Water Discharge (f) Absence of Water Supply

(g) Fixed Telephone Line (h) Cell Phone Ownership (i) Computer Ownership

-2-1

01

23

Lo

g Illi

tera

cy R

ate

-3000 -2000 -1000 0 1000Normalized Distance from Colonial Border

22.5

33

.54

4.5

Lo

g C

om

pu

ter

Litera

cy

-2000 -1000 0 1000 2000Normalized Distance from Colonial Border

3.5

44.5

Lo

g L

and

Ow

ners

hip

-3000 -2000 -1000 0 1000Normalized Distance from Colonial Border

02

46

Lo

g A

ccess t

o P

ub

lic W

ate

r N

etw

ork

-2000 -1000 0 1000 2000Normalized Distance from Colonial Border

22.5

33.5

44.5

Lo

g S

an

itary

Fa

cili

tie

s a

nd W

ate

r D

isch

arg

e

-3000 -2000 -1000 0 1000Normalized Distance to Colonial Border

-20

24

6Lo

g A

bse

nce

of W

ate

r S

up

ply

Insid

e the

Ho

use

-3000 -2000 -1000 0 1000Normalized Distance from Colonial Border

53

-20

24

6Lo

g F

ixed

Tele

pho

ne

Lin

e

-3000 -2000 -1000 0 1000Normalized Distance from Colonial Border

2.5

33

.54

4.5

Lo

g C

ell

Ph

on

e O

wn

ers

hip

-3000 -2000 -1000 0 1000Normalized Distance from Colonial Border

12

34

5Lo

g C

om

pute

r O

wn

ers

hip

-3000 -2000 -1000 0 1000

Normalized Distance from Colonial Border

54

Appendix A3: RD Effect Distribtion under Alternative Window Selection

-.8

-.6

-.4

-.2

0.2

Effect S

ize a

t th

e C

uto

ff

Log Distance from Colonial Audencia (km)

(a) Illiteracy Rate

0.2

.4.6

Effect S

ize a

t th

e C

uto

ff

Log Distance from Colonial Audiencia (km)

(b) Computer Literacy

-.1

0.1

.2E

ffect S

ize a

t th

e C

uto

ff

Log Distance from Colonial Audiencia (km)

(c) Land Ownership Rate

-.6

-.4

-.2

0.2

Effect S

ize a

t th

e C

uto

ff

Log Distance from Colonial Audencia (km)

(d) Access to Public Water Network

0.2

.4.6

Effect S

ize a

t th

e C

uto

ff

Log Distance from Colonial Audiencia (km)

(e) Households with Sanitary Facilities and Water Discharge Inside the House

.51

1.5

Effe

ct

Siz

e a

t th

e C

uto

ff

Log Distance from Colonial Audiencia (km)

(f) Households with Fixed Telephone Line

55

.3

.4.5

.6.7

.8E

ffe

ct S

ize a

t th

e C

uto

ff

Log Distance from Colonial Audiencia (km)

(g) Cell Phone Ownership

0.2

.4.6

.8E

ffect

Siz

e a

t th

e C

uto

ff

Log Distance from Colonial Audiencia

(h) Computer Ownership

56

Appendix A4: Temperature Variation, Human Capital Formation and Long-Run Development (a) Illiteracy Rate

Linear RD Estimate Locally Weighted Smooth RD Estimate

(b) Computer Literacy

Linear RD Estimate Locally Weighted Smooth RD Estimate

(c) Land Ownership

01

23

5 10 15 20 25Mean Annual Temperature (°C)

01

23

5 10 15 20 25Mean Annual Temperature (°C)

22.5

33.5

44.5

5 10 15 20 25Mean Annual Temperature (°C)

22.5

33.5

44.5

5 10 15 20 25Mean Annual Temperature (°C)

57

Linear RD Estimate Locally Weighted Smooth RD Estimate

(d) Access to Public Water Network

Linear RD Estimate Locally Weighted Smooth RD Estimate

(e) Sanitary Facilities and Water Discharge

Linear RD Estimate Locally Weighted Smooth RD Estimate

3.5

44.5

5 10 15 20 25Mean Annual Temperature (°C)

3.5

44.5

5 10 15 20 25Mean Annual Temperature (°C)

02

46

5 10 15 20 25Mean Annual Temperature (°C)

02

46

5 10 15 20 25Mean Annual Temperature (°C)

58

(f) Absence of Water Supply

Linear RD Estimate Locally Weighted Smooth RD Estimate

(g) Fixed Telephone Line

Linear RD Estimate Locally Weighted Smooth RD Estimate

23

45

5 10 15 20 25Mean Annual Temperature (°C)

22.5

33.5

44.5

5 10 15 20 25Mean Annual Temperature (°C)

-20

24

6

5 10 15 20 25Mean Annual Temperature (°C)

-20

24

6

5 10 15 20 25Mean Annual Temperature (°C)

59

(h) Cell Phone Ownership

Linear RD Estimate Locally Weighted Smooth RD Estimate

(i) Computer Ownership

Linear RD Estimate Locally Weighted Smooth RD Estimate

-20

24

6

5 10 15 20 25Mean Annual Temperature (°C)

-20

24

6

5 10 15 20 25Mean Annual Temperature (°C)

2.5

33.5

44.5

5 10 15 20 25Mean Annual Temperature (°C)

2.5

33.5

44.5

5 10 15 20 25Mean Annual Temperature (°C)

60

12

34

5

5 10 15 20 25Mean Annual Temperature (°C)

12

34

5

5 10 15 20 25Mean Annual Temperature (°C)

61

Appendix A5: Rainfall Variation, Human Capital Formation and Long-Run Development (a) Illiteracy Rate

Linear RD Estimate Locally Weighted Smooth RD Estimate

(b) Computer Literacy

Linear RD Estimate Locally Weighted Smooth RD Estimate

01

23

4 5 6 7 8Log Precipitation

01

23

4 5 6 7 8Log Precipitation

62

(c) Land Ownership

Linear RD Estimate Locally Weighted Smooth RD Estimate

(d) Access to Public Water Network

Linear RD Estimate Locally Weighted Smooth RD Estimate

22.5

33.5

44.5

4 5 6 7 8Log Precipitation

22.5

33.5

44.5

4 5 6 7 8Log Precipitation

3.5

44.5

4 5 6 7 8Log Precipitation

3.5

44.5

4 5 6 7 8Log Precipitation

63

(e) Sanitary Facilities and Water Discharge

Linear RD Estimate Locally Weighted Smooth RD Estimate

(f) Absence of Water Supply Linear RD Estimate Locally Weighted Smooth RD Estimate

02

46

4 5 6 7 8Log Precipitation

02

46

4 5 6 7 8Log Precipitation

22.5

33

.54

4.5

4 5 6 7 8Log Precipitation

22

.53

3.5

44.5

4 5 6 7 8Log Precipitation

64

(g) Fixed Telephone Line

Linear RD Estimate Locally Weighted Smooth RD Estimate

(h) Cell Phone Ownership

Linear RD Estimate Locally Weighted Smooth RD Estimate

-20

24

6

4 5 6 7 8Log Precipitation

-20

24

6

4 5 6 7 8Log Precipitation

-20

24

6

4 5 6 7 8Log Precipitation

-20

24

64 5 6 7 8

Log Precipitation

65

(i) Computer Ownership

Linear RD Estimate Locally Weighted Smooth RD Estimate

2.5

33.5

44.5

4 5 6 7 8Log Precipitation

2.5

33.5

44.5

4 5 6 7 8Log Precipitation

12

34

5

4 5 6 7 8Log Precipitation

12

34

54 5 6 7 8

Log Precipitation

66