Provincial partisan patronage and national party system stability in Argentina,

DRAFT PLEASE DO NOT CITE Provincial partisan patronage and national party system stability in Argentina, 1983-2005 Gerardo Scherlis* - Department of Political Science - Leiden University gscherlis@fsw.leidenuniv.nl Prepared for delivery at the workshop on Political Parties and Patronage, ECPR Joint Sessions, Nicosia, April 2005 I. Argentine party system. Between change and continuity When Argentine democracy was restored in 1983, the same political parties that had dominated politics from 1945 until the 1976 military coup emerged as the major political players. The Unión Cívica Radical (UCR) and the Peronist Party (PJ) had generated deeply antagonistic loyalties, both of them claiming to represent the real national interest (McGuire, 1995). The Radicals had traditionally represented middle class citizens and were identified by its followers as champions of civil rights and democratic institutions, which they felt were threatened by the Peronists. Peronism, in turn, developed high levels of societal rootedness among the working classes and the poor in general, who identified Perón and the movement he founded with the cause of social justice. These strong political identities started to thaw in the mid-80s, and shares of voters responding to party loyalty became increasingly volatile during the 1990s (Manzetti, 1993; Palermo y Novaro, 1998; Cheresky and Pousadela, 2004). However, the PJ and the UCR undoubtedly dominated the Argentine two-party system from 1983 to 2001. In the last quarter of 2001 Argentina underwent a severe political crisis, marked by an upsurge of radical anti-party feelings. In the October legislative polls, almost fifty percent of the citizens opted for what the media called an anger vote 1 and two months later a massive civic * Supported by the Programme AlBan, the European Union Programme of High Level Scholarships for Latin America, scholarship No. E05D050050AR. I want to thank Imke Harbers, Maria Spirova and Petr Kopecky for their help and comments. 1 Nearly four million blank and null votes 24% of the vote were cast, more than what any one party obtained except for PJ, while there was some 24.5% of abstention. 1

rebellion against the entire political class led to the resignation of the president. Neighbourhood assemblies and, above all, movements of the unemployed [piqueteros] took to the squares and the streets with a unified slogan that called every politician to quit: Que se vayan todos! For several months Argentina teetered on the brink of anarchy (Levitsky, 2005). In the light of these events, many considered this crisis a turning point in the Argentine political system, including its party system. Observers predicted that the UCR would collapse in the wake of De la Rúa s government failure, while the PJ would survive but undergo a process of deep internal fragmentation with unpredictable consequences (Ansaldi, 2002; Sidicaro, 2005). The beginning of a second transition, in reference to a complete rearrangement of the party system, was announced (Torre, 2003). Furthermore, President Kirchner (PJ), elected in a highly atypical election which exposed the fragmentation of the party system, 2 soon came to advocate the idea that a new, more coherent party system, comprised of a centre-left and a centre-right bloc, should emerge, leaving behind the traditional confrontation between Peronists and Radicals. Nevertheless, as shown in table 2, both traditional parties though the UCR to a lesser extent - have managed to keep their institutional dominance, while no third party emerged as a potential electoral alternative (Malamud and De Luca, 2005). After the 2003 and 2005 legislative and governorship elections it is clear that the party system did not suffer a Peru- or Venezuelastyle party-system collapse. Furthermore, the traditional parties remain largely as the first and second blocs in both national chambers and in control of 22 of the 24 governorships. (Tables 1 and 2) The resilience of the national two-party system must be understood primarily as a function of the strong partisan stability registered in most subnational arenas. Certainly, political alternation and certain degree of volatility was much more common at the presidential level than 2 In April 23 rd 2003, five candidates run with reasonable expectations. The PJ presented three different contestants, so resolving its primaries in general elections but also defying the most extreme minimalist definitions of what a political party is. Two candidates from the peronist party reached the second round, but former president Menem, regarding the unavoidable defeat foreseen by the surveys, gave up a few days before the date of the polls. In turn, the UCR s presidential candidate got only 2,3% of votes, while two former UCR leaders run with new parties, obtaining the third and fifth places. (See Table 1). 2

at the provincial executives from 1983 to 1999, but the gap between these two tiers became more striking since 2001. (Table 3) This stability of the provincial party systems has been attributed mainly to the characteristics of the Argentine institutional design, particularly to its electoral system (Calvo and Escolar, 2005; Malamud and De Luca, 2005). In brief, it is argued that the three electoral arenas in which national positions are elected president, senators, and deputies- present strong majoritarian biases. These pose an effective barrier to the emergence and rise of third parties thus consolidating the dominance of the established PJ and UCR. In addition, according to the federal system of government, every one of the 23 governors of the provinces and the chief of government of the Federal Capital are directly elected by their constituencies, enhancing even more the concentration of the vote in the provinces. Finally, malapportionment in both legislative chambers benefits PJ and UCR who dominate small-sized overrepresented provinces. I find these arguments convincing, but still insufficient. If they explain why many provinces present obstacles to third parties emergence they do not account for significant disparities across provinces with similar institutional features. In this paper I attempt to make a contribution by proposing a complementary explanation which turns the attention to the issue of provincial party-patronage. Patronage is frequently underlined as a major institution of the informally institutionalized Argentine democracy (O Donnell, 1997) and it has been suggested that the provincial electoral machines are one of the reasons for the persistence of main parties (Levitsky and Murillo, 2003). However, the development of more comprehensive studies on the mechanisms that enable provincial political systems to avoid the fragmentation that took place at the presidential tier remains a major pending task. As an attempt in this direction, I suggest that party patronage is a main component of party clientelist networks at the provincial level, which in turn plays a key role in maintaining provincial party stability in a number of constituencies. More precisely, the main hypothesis of this paper is that provinces with higher levels of patronage present lower levels of political alternation and more closed and stable party systems. From the observation of the presidential tier it seems reasonable to conclude that the 3

Argentine party system is changing from a relatively closed and predictable two-party system to a more open and unpredictable one. Nonetheless, this apparent change is softened by the closed nature and the stability of many subnational party systems. I argue that these features, often manifested by predominant party systems and absence of alternation rest, in a significant number of districts, on the electoral impact of party patronage. This paper proceeds as follows. First, it highlights the factors that make the provinces not only the locus of Argentine party politics but also of party patronage. Second, it shows how patronage combines with clientelism to generate an electoral impact. In this part the study propose an original proxy for assessing the electoral impact of party patronage in Argentine provinces. Thirdly, the main hypothesises are empirically tested. The final section is devoted to a discussion of some inferences drawn from the data. 2. Decentralized Party Patronage 2. a. Decentralized Public Administration Argentine public administration is highly decentralized. Around fifty percent of the total national expenditure are spent at subnational level and provinces concentrate more than seventy percent of total public employment of the country. While the size of the central government public sector fell dramatically during the early 90s due to the process of market-oriented economic reform, the levels of provincial public employment remained stable and started to rise again after the 2001-2002 s crisis (See table 4). (Table 4) Provincial administrations revenues, however, finance less than a quarter of their expenditures whereas the rest is provided by a set of transfer mechanisms from the central government. 3 In recent years, a number of studies have shown how these transfers of funds are subjected to discretionary criteria, usually linked to the stance adopted by provincial officers with regard to the central governments bills and policies. Scholars agree that these trade-offs benefit small and overrepresented provinces, which manage to keep high levels of public employment by granting political support to national governments in exchange for side payments 3 The major of these mechanisms is the Federal Tax-Sharing Agreement, by which provinces delegate tax authority to the national government, which in turn devolves the collected funds to the provinces. 4

in the form of funds and subsidies (Remmer and Wibbels,2000; Jones et al. 2000; Gordin 2005). Being small-sized and politically overrepresented in both national chambers, these provinces have been called low-maintenance constituencies, in reference to the comparatively inexpensive payments that the national executive power has to make to obtain their legislators support (Gibson and Calvo, 2000). The fact is that the shrinkage of national public sector reduced the scope of opportunities for patronage at national level but reinforced its relative share in the provinces where the absence of civil service regulations - often mentioned as a reference to national public administration - is even more noticeable (Calvo and Murillo, 2004). In any case, and regardless the significant variances across them, the rate of provincial public employment in all Argentine provinces is comparatively very high (Benton, 2003). At the same time, provincial governors, as well as the chief of government of the Federal Capital, exercise substantial autonomy, enjoying considerable and discretionary powers over the administration of budgets and public policies (Eaton, 2004). Actually, even when the president is still the axis of the Argentine political system he coexists with powerful governors, who have turned into major political players (Spiller and Tomassi, 2002; Benton, 2003). The significance of governors, in comparison with other senior officials, is well illustrated by the fact that in the last two decades holding a governorship became the necessary stepping stone to the presidency. 4 2. b. Provincially-centred parties It is well-established that the Argentine party system is provincially centred (Calvo and Murillo, 2004; Jones and Hwang, 2005). Federalism stimulates parties to organize and compete at the sub-national level, orienting their activity to the control of local political institutions (Jones and Mainwaring, 2003). 5 The process of political and economic decentralization implemented during the 1980s and early 1990s, transfered functions and funds to subnational administrations and reinforced the role of provincial political institutions even more (Eaton, 2004). In a context of disaffection with politics and ideological vacuum, decentralization favoured the territorialization of political parties and, more in general, of Argentine politics. With regard to 4 From 1989 onwards, not only every elected president held at the moment of his election a governor position, but also excepting for one single case in 1995, the runner-up. 5 It must be noted that Argentina has a significant number of parties that only compete in one province, the so called provincial parties, some of which are dominant or the main opposition in their districts. 5

intra-party politics, since ideological motivations lost preponderance it turned inconceivable that internal tendencies arose out of programmatic differences as it was the case to some extent in the past. By contrast, the predominance of material resources encouraged the development of factions centred around the management of financial resources, which appeared to be attainable at the provincial level. Although both PJ and UCR nominally remain national parties, their structures increasingly resemble that of loose and heterogeneous leagues of provincial organizations (Levitsky and Murillo, 2003). Fragmented competition and lack of homogeneous and cohesive parties have become rampant in Argentina. The level of fragmentation is such that lately major parties have contested elections under different labels in many provinces, forming contradictory alliances in different districts. It is clear, at the same time, that the provincial branches of the parties are the locus of Argentine politicians careers. Given that national legislators are nominated and elected by subnational branches, they remain beholden to provincial leaderships. In the overwhelming majority of the cases, successful politicians are those who follow the cursus honorum within a provincial section and manage to dominate and strengthen it (Jones et al, 2002). As I try to demonstrate below, in the dominance of these subnational organizations in which the core of Argentine politics takes place, patronage plays an essential role, constituting the main fount for obtaining and conserving leadership. 3. Party patronage, clientelism and votes 3.a. Disentangling concepts The argument I present in this paper alludes to the linkage between three concepts often used as if interchangeable, namely: patronage, clientelism, and pork barrel. Even when the three of them may be conceived as a means of securing electoral support for a party, a faction or a particular contestant through particularistic exchanges, they do differ in three major aspects: the content of the supply, the actors involved, and the suppliers main purpose. In mass democracies, party clientelism entails the provision of a wide variety of goods and/or services. In general terms, it is a way of wooing people s support to a party, faction or contestant through satisfying some of their daily needs. The clients are usually poor and even marginalized people and the providers are politicians linked to the state apparatus. Regardless of 6

the undeniable complexity of these relationships, which usually include elements of identity and loyalty (Auyero, 2000; Piattoni, 2001), in simple and general terms it may be said that the suppliers primary goal is obtaining of clients votes. Pork barrel, in turn, refers to the implementation of particularistic policies, as the provision of funds or the enactment of a law which benefit particular constituencies, in order to obtain their political support. The actors involved are usually elected officials on the one hand and their constituencies as a collective subject- on the other. However, as the Argentine case clearly illustrate, pork barrel simultaneously supposes trade offs between sub national officials and central government, the former demanding transfers of funds and others side payments in exchange for the latter s political support (Gibson and Calvo, 2000). The concept of party patronage, finally, is much more restricted, as it connotes the allocation of appointments and jobs in public and semi-public positions by a political party or faction, which acts as a collective patron. Party patronage has been usually referred to as a tool for satisfying party organizational needs rather than an electoral strategy (Panebianco, 1988; Müller, 1989; Blondel, 2002; Rybar, 2006, van Biezen and Kopecky, 2007). In contemporary cartel parties, devoid of appeal to voluntary activism, embeddness in the state apparatus allows politicians who hold office to use patronage for the recruitment of supporters (Van Biezen, 2004). In this sense, as Kopecky (2006) underlines, clientelism is a much more penetrating and diversified phenomenon, but which requires the previous existence of patronage, since without an ability to control appointments within the state institutions, political parties would not be in the position to distribute selective benefits. In accordance with this literature, patronage functions in Argentina as a resource for building party organizations. However, these party organizations built and feed by patronage are primarily electoral machines which, in turn, make use of the resources extracted from patronage to develop the clientelist networks, crucial for winning elections. In reference to Argentine municipalities, a recent study sharply differentiated between clientelism and patronage to stress that only the former is effective in buying votes (Brusco et al, 2005). Nonetheless, it remains true that clientelist practices are unthinkable without patronage. If appointments do not entail by themselves necessarily an electoral impact, they allow electoral machines to work. As Jones (2004) puts it in reference to party primaries in Argentina, patronage positions are particularly important for maintaining the support of second and third tier party leaders, who in turn possess 7

the ability to mobilize voters. Hence I assume that patronage has an electoral impact as well and thus consider it is worthwhile to assess whether and when this impact is relevant. 3. b. Patronage and electoral machines During the 80s and 90s party primaries spread as a mechanism for candidates selection in Argentine political parties. Likewise, primaries were increasingly held to vote uppermost parties positions (Mustapic, 2002; De Luca et al., 2002). This implied the impossibility of climbing up within the parties structures without the ability to win a primary. It must be noted that when democracy was restored, in 1983, thousands of activists joined party organizations, which had, during the so-called democratic spring, huge popular support. However, disenchantment with politics and disaffection with political parties came soon, and politicians and parties were increasingly perceived as untrustworthy. 6 As ideologically-motivated activists stopped being active, party structures turned into semi-state agencies absorbed by, and increasingly dependent on the state s resources for their functioning. Access to state positions thus became needed for organizational survival. At the same time, the extension of both intra and inter party competition from 1983 onwards boosted the search for a new kind of resources, the electoral clienteles. Given the dearth of an ideologically-motivated membership, internal competition turned into clientelist-based struggles among factions and primaries became clashes between electoral machines (Palermo, 1986; Levitsky, 2003). As De Luca et al. (2006) argues, the core task in a primary consists in the mobilization of clienteles to the polls, which entails significant spending in several kinds of goods and services, from the salaries of the neighbourhood brokers (punteros), hired by the provincial or municipal governments, to the monies for buses, taxis and drivers to transport voters to the ballot rooms. In the primaries there are not ideological issues involved, and contestants backgrounds do not matter either. The victory is solely determined by the amount of resources at disposal and the efficiency in the use of them. Consequently, patronage is the cornerstone on which politicians fund their leadership and upon which depend their careers. (Jones et al, 2002). 7 6 While political parties were positively evaluated in 1984 by 84 percent of the population, they had fell to 63 percent in 1988, to become the least trustful institution along the nineties, with only 4 percent of the interviewed manifesting some confidence in 2002 (Catterberg, 1991; Adrogué y Armesto, 2001; Latinoabrómetro Report, 2002). 7 It is worth noting that the networks of brokers stretching, particularly over ever-increasing poor sectors, are not the result of a systematic plan deliberately promoted by the national party for external competition; rather they are an 8

In sum, the control of the provincial section of the party requires a well-developed electoral machine, for which resources obtained from patronage are indispensable. 8 As it was the case in many American cities in the early 20 century (Scott, 1969), in Argentine provinces patronage constitutes the basis for building the machines which, in turn, extend the clientelistbased networks for gaining votes. 9 It seems logical to infer, accordingly, that higher rates of politically appointed jobs should lead to broader and more influential political machines. 3. c. Party Patronage: Public employment and Party membership Certainly, data on patronage is always soft. As Geddes stresses, in the real world, there is no way to measure amounts of patronage or how much influence on the vote it has. (1994: 105). That is the reason why, when scholars try to assess patronage impact, proxy measures are necessarily employed. In this paper, following some of the most recent approaches on the subject (Gibson and Calvo, 2000; Remmer and Wibbels, 2000; Calvo and Murillo, 2004), I make use of provincial public employment as a proxy for party patronage. Definitely, almost every provincial job in Argentine provinces might be counted as patronage, insofar as there are no civil service laws preventing political appointments. However, since this study aims at assessing the influence of party patronage in keeping provincial party systems stable, it is not solely concerned with the number of jobs discretionally appointed by the ruling party but, more specifically, with the electoral weight of those voters encouraged to support the ruling party as a payment for getting or keeping their jobs. For that purpose, public employment figures by themselves would seem to be too rough. For instance, it is natural that Patagonian provinces, due to their hard geographic conditions and sparse population, show higher rates of public employment than metropolitan districts, where population is much more concentrated, this not necessarily entailing a partisan bias. Furthermore, higher rates of public employment could as well result from more developed state run services, as schools or hospitals, in a country in which provinces exhibit large disparities in those fields. Therefore, I attempt here to refine the proxy by using a control unintended by-product of a relatively autonomous dynamics, associated with internal factions competing for power in the party (Auyero, 2000). 8 State financing, however meaningful for newcomers and small parties very first organizational needs, is by far less significant for established and major parties, and definitely scarce for the building of the machine required for primaries (De Luca 2004; Jones and Hwang, 2005). 9 The scope for clientelist practices was enlarged in the context of the neoliberal model implemented in the 90s. With the state s abandonment of certain essential functions, palliative welfare aid became a method of political struggle among the internal factions of parties (Brusco et al, 2004). 9

variable, i.e. ruling party membership. In this section, resuming the argument exposed in previous paragraphs, I explain why and how I do so. Provincial electoral machines, arisen from internal competition, remain at disposal and are utilized for general elections as well. However, given that the main goal in the development of the machines refers to the local internal competition within the party, it is not far-fetched to expect that those citizens involved are registered as party members, as a sine qua non for getting the handouts. 10 In every provincial section, power tends to be concentrated in the hands of a single boss or a reduced group. In the cases of provincial ruling parties, the governor is the natural party leader, sustained by his unequal access to material resources. His ability to allocate jobs is crucial for the construction and the control of party electoral machines (Jones, 2004, De Luca, 2004), providing him and his group with a position that guarantees a large advantage over any potential intra-party competitor. Nonetheless, provincial party leaderships strive to enlarge and strengthen their supporting machines. They have two major reasons to do so. First and foremost, they need to discourage the emergence of any potential challenger. Even when the governor is seen as the unquestionable party local boss, he is well aware that potential challengers, despite their initial disadvantaged position, might emerge at any moment (De Luca et al, 2006). These potential challengers can be mayors of large cities and national legislators, allied to other provincial leadership or even to the president, when the latter envisages the governor as a threat for his own leadership. In fact, when candidacies and party s positions are not submitted to primaries it is because the leadership appears to be too powerful to be challenged. 11 In second place, just as the number of affiliate members one broker can mobilize provides the parameter for determining his power vis-à-vis his peers, the size of the political machine, measured by party membership, determines the position of the local boss within the national party. That is because the number of delegates every province sends to national conventions, and thereby its presence in party national boards, depends on the amount of its 10 Not affiliated voters, who have been increasingly allowed to take part in candidate selection primaries, have never represented a significant share of the turnout. In 2005, when all the electorate was allowed to take part in party primaries, only 3.2 percent voted. Simply put, clients are useful for the brokers only if they join the party. 11 However, it is to be expected that the electoral machines swell more when the internal competence actually takes place. So happened in the context of a hard internal dispute in the Peronist Party in the Province of Buenos Aires in 2004, when 700.000 new members were reported. See Clarín newspaper, 12-10-2004: La Interna del Peronismo. El PJ bonaerense dice que sumó 700 mil afiliados en un solo mes http://www.clarin.com/diario/2004/10/12/ elpais/p- 00601.htm. 10

affiliated members. Furthermore, the provincial boss image as a leader within his own party is reinforced when party membership in the district he controls remains high and surges up. It is not by coincidence that of the three Peronist presidential candidates in 2003 all were the bosses of sections with the highest rates of party membership. These reasons make it understandable why party membership has never stopped to rise in Argentine political parties even when, as noted above, parties are discredited as untrustworthy institutions. While in 2000 only five percent of Argentine citizens admitted to be part of a political organization, in the same year the number of party members reached 8.3 million, surpassing one third of the total voters. 12 The linkage between patronage practices and increments in ruling party membership has been underlined in some other cases of democratized countries (Sotiropoulos, 1996; Rybar, 2006). In the context of recent Argentine political studies, however, party membership has remained a black hole, most scholars dismissing the data as not reliable. Certainly, parties registers of members are inflated, not reflecting the real dimension of their organizations. Once a citizen signed the form, he remains a member, without any duty to be fulfilled, unless he/she decides to cancel the membership. 13 Undoubtedly, many people joined the parties between 1982 and 1984, and stayed so regardless of their virtually non existent linkage to his/her party. These noticeable weak points notwithstanding, party membership s figures are not meaningless. Given the functioning of political machines exposed above I consider them useful for checking the extent of party patronage. Remarkably, as Table 5 exposes, both PJ and UCR tend to increase their membership much more while in power than when they are in the opposition. 14 (Table 5) As opinion polls display time after time and as it is evident for any observer of contemporary Argentine politics, no large amounts of citizens joining these parties from 1987 once the democratic spring was definitely over- might be explained but for the intra-party dynamics described above. Thus, it is my contention that the figures of party membership are 12 The case suggests a very cautious appliance of the variable extension of party membership as an indicator of democratic participation, recently proposed by Diamond and Morlino (2005). 13 Which entails much more complicated proceedings than those required to join the party. See Clarín newspaper, 12.10.2004: La difícil tarea de desafiliarse de un partido [The hard work of leaving party membership]. 14 Exceptions, as will become clear later, are confined to districts with low levels of patronage. 11

linked to the enlargement of provincial political machines, which are mainly fuelled by party patronage. In consequence, I suggest that the linkage between public employment and ruling party membership, taking into account the processes of member recruitment exposed above, offers a tighter though still very rough- approach to the electoral reach of patronage. Therefore I will use provincial public employment as a proxy for patronage, but I will control this variable through the level of ruling party membership. Concretely, I will consider as patronage-oriented those provinces that, having high levels of provincial public employment, exhibit at the same time high levels of ruling party membership. (See Apendix) In the next section I proceed to test the main contentions of this paper: a) That higher levels of party patronage understood as high provincial public employment alongside high ruling party membership -- should go together with lower levels of political alternation; and b) That higher level of party patronage should lead to more closed and stable party systems. 4. Provincial public employment and party competitiveness in Argentine provinces A. Patronage and political alternation Argentine provinces elect their governors every four years. Since 1983, most provinces have had six governorship elections. 15 Nine of the twenty four provinces have been governed by the same ruling party with no interruptions, 16 six had only one change in the provincial executive 17, two provinces experienced two alternations 18 and, finally, in seven provinces there were three replacement of the ruler party. 19 Noticeably, continuity at the provincial level is much more frequent than alternation. Nevertheless, provinces show important variations. According to the data presented in Table 6, a strong correlation exists between high levels of provincial public employment and low levels of alternation. Among the nine provinces that had no alternation, six have high levels of public employment and two medium-high. In contrast, alternation is higher among provinces with medium-low and low provincial public employment. Nonetheless, following these parameters there are three deviant cases. Tierra del 15 A federal intervention led to one more election in Santiago del Estero. Tierra del Fuego and Ciudad de Buenos Aires gained political autonomy in 1990 and 1994, thus having four and three elections respectively. 16 La Rioja, Santa Cruz, Formosa, La Pampa, Jujuy, Río Negro, Neuquén, San Luis and Santa Fe 17 Buenos Aires, Misiones, Córdoba, Catamarca, Ciudad de Buenos Aires, Santiago del Estero. 18 Corrientes and Salta. 19 Mendoza, Tierra del Fuego, Chaco, Tucumán, Chubut, San Juan, Entre Ríos. 12

Fuego, Chubut and San Juan, three provinces with high or medium high levels of public employment (4 th, 9 th and 12 th in national ranking) had, each of them, three alternations. Table 6. Provincial public employment and Alternation Public Employment Alternation None High (Above 55 employees out of 1.000 inhabitants) La Rioja Santa Cruz Formosa La Pampa Neuquén Jujuy Medium high (between 45 and 55) San Luis Río Negro Medium low (Between 35 and 45) Low (Below 35) Santa Fe Low (1) Catamarca Santiago Misiones Bs As Cordoba Capital Médium (2) Corrientes Salta High (3) Tierra del Fuego Chubut San Juan Chaco Entre Ríos Tucumán Mendoza Interestingly, when instead of provincial public employment we consider patronage as it was proposed here -- i.e. public employment plus party membership -- the correlation strengthens; there remain no deviant cases to the hypothesis. The three alluded provinces have, at the same time, very low levels of ruling party membership and thus should not be classified as patronage-oriented. 13

Patronage Alternation None Table 7. Patronage and alternation High Patronage Medium Patronage Low Patronage La Rioja Santa Cruz Formosa La Pampa Neuquén Jujuy San Luis Santa Fe Río Negro Low Catamarca Santiago del Estero Buenos Aires Córdoba Misiones Capital Medium Corrientes Salta High Entre Ríos Mendoza Tucumán Chubut San Juan Chaco Tierra del Fuego The province of Tierra del Fuego offers an extreme example of the combination of high public employment with low ruling party membership. As it is the smallest Argentine province, new parties there face the constraints imposed by the majoritarian biases of the institutional design. Tierra del Fuego was recognized as a province by the National Congress in 1990, thus it had its first democratically-elected government in 1991. The victor was then the local Movimiento Popular Fueguino (MPF), which won again in 1995. At the end of its second tenure, the province had the fourth highest rate of provincial public employment of the country. However, after eight years in government a very small share of provincial voters (in comparative terms) was affiliated with the ruling party (12.5 percent, 20 th in the national ranking). This, according to the parameters of this study, leads me to consider it as a non patronage-oriented case. In 1999 the MPF was defeated by PJ which, in turn, was displaced by UCR in 2003. In addition, the recently created ARI, a national centre-left party, set foot in the province, won the 2003 legislative elections and came out second in 2005. In brief, despite the institutional constraints and its levels of public employment, the province shows high rates of political alternation in the frame of an increasingly fragmented and volatile political scene. 14

The other two provinces with high rates of public employment and high alternation, Chubut and San Juan, also exhibit low levels of ruling party membership. Chubut has had, since 1983, high levels of provincial public employment, ranking permanently among the top-ten provinces. From then on, the radicals and the peronists have alternated in power in the province, both with extremely low levels of party membership. The UCR kept power for three tenures, between 1991 and 2003, but at least until 1999- it maintained the country s lowest rate of party membership. In 2003, UCR was defeated by PJ. San Juan, for its part, is also considered a low maintenance constituency, and, in fact, its level of public employment is among the top-half. Although both the provincial party Bloquista, which governed between 1983 and 1991, and PJ, which governed between 1991 and 1999, raised their levels of membership while in power, none of them did it in a comparatively significant extent, thus maintaining the province among the bottom-half in this regard. A thoroughly different picture is portrayed when analyzing provinces that have high or medium-high public employment rates and high or medium-high ruling party membership. La Rioja, La Pampa, Neuquén, Formosa, Santa Cruz, Jujuy and San Luis, have never had any party alternation, whereas Catamarca had it only once. Catamarca is particularly worthy of some remarks, given that unlike the others, permanently and undisputedly dominated by PJ - it reveals similarities in both major parties practices. From 1983 to 1990 PJ remained as the predominant party, unchallenged by any other. In 1990 a criminal scandal in which members of the governor s family were involved led to a federal intervention, which in turn paved the way for the triumph of UCR in 1991. From then onwards, UCR remained in power. Significantly, alongside their respective tenures at office, both parties significantly increased their memberships, whereas both decreased when they were in the opposition (see table 5). 20 B. Patronage and closed provincial party systems Regarding the structure of competition, Mair (1997) classifies party systems as closed, and thus predictable, and open, and thus more unpredictable. The pattern of alternation, the type of governing formula, and the range of parties gaining access to government are the key 20 Furthermore, observers of local politics have noted that UCR managed to retain power due to its resort to the same particularistic exchanges that peronists of Catamarca had used until 1990 (Gutiérrez, 1999). 15

indicators whose level of stability defines the type of party competition. Since 2001, when the apparent collapse of party system took place, provinces with higher levels of patronage have kept their party systems more closed and stable than their counterparts with low patronage levels. Certainly, the period since the 2001 crisis, which includes only one governorship election, is too short to allow a rigorous analysis of provincial party systems change from then on. However, some preliminary observations may be exposed, taking into account also national legislative elections. 21 The eight provinces identified as having the highest rates of patronage (high or medium-high in both items) keep the same closed structure of competition after 2001 than that they had had for the previous eighteen years. In six of them Peronists appear as unbeatable as ever, the Radicals remaining as the main opponent and reaping the corresponding institutional rewards, i.e. one senator and several deputies. In Neuquén, the local Movimiento Popular Neuquino (MPN) seems as invincible as it has been since 1983, the PJ being the runnerup. In Catamarca, the dominance of the Radicals has been, since they took power in 1991, threatened by Peronists, and it is still so. Exactly as it had been from 1983 to 2001, PJ and UCR (and the MPN in Neuquén), are currently the main political actors in these political systems whose structure of competition has not showed perceivable changes. The scene is much more complex when trying to analyze the provinces with the lowest levels of patronage (low or medium-low in both items). In two of them, Misiones and Buenos Aires, where PJ and UCR had disputed even elections, Radicals suffered an electoral collapse, leaving PJ as the unchallenged ruling party. In Córdoba, where UCR had been the predominant party from 1983 to 1999, PJ took its place. A new party, the local centre-left Frente Nuevo (New Front), became the runner up, so displacing UCR to a far off third place, and reasonably expecting to achieve the governorship in next elections. In Santa Fe, a third party, the Socialist, became the main opposition to the long-standing predominant PJ. Leading an alliance with a severely diminished UCR (which until mid-90s had been a real threat to PJ), Socialists defeated Peronists in 2005 legislative elections and are considered to be the most probable winners in the next governorship contest in 2007. The Federal Capital, finally, exhibits the most extreme case of party system change, being affected on the three cited factors (pattern of alternation, governing formula and the openness in access to government). The Radicals, who had governed 21 From 2001 up to now provinces passed through three national legislative electoral rounds. 16

from 1996 to 2000, vanished completely. The centre leftist Frepaso, who succeeded the Radicals in government, virtually disappeared in the wake of the 2001 s crisis, but the chief of government still managed to be re-elected in 2003 with the support of an heterogeneous coalition, including the explicit support of the recently elected President. New forces, such as the centre left ARI and centre right PRO, gained prominence, while PJ resurged in 2005 after several elections in which it had remained almost non existent. When considering the 2007 chief of government elections, a large number of personalities rather than traditional parties must be considered as the major players in this district. Interestingly, while until 2001 both PJ and UCR used to obtain similar shares of legislators in the whole country, or even larger shares in districts with low levels of patronage, an opposite tendency is visible since 2001. Third parties, most of them based on the popularity of a single personality but simultaneously identified with centre-left or centre-right programmes, managed to establish footholds in non patronage orientated districts. UCR and PJ, by contrast, as illustrated in table 8, keep a large predominance over highly patronage-oriented ones. The notorious difference between high and low patronage districts is also illustrated when examining electoral performances in 2003 and 2005 (Table 9). Successful third parties letting aside the long-standing local MPN in Neuquén- emerged only in non patronage districts, whereas PJ and UCR though the latter to a lesser extent- remain the two main parties in highly patronage-oriented constituencies. (Tables 8 and 9) Finally it must be noted that the relation between patronage and the presence of dominant parties can be proposed exactly in the opposite direction, i.e. that predominant party systems lead to higher levels of patronage. 22 However, in this case, disparities among the provinces levels of public employment have not significantly increased over time. Instead, those provinces that exhibited higher levels of public employment in 1987 and that developed strong political machines fast turned to be closed and with predominant party systems. Furthermore, most of the provinces with high rates of patronage exhibit at the same time significant rates of opposition parties membership, which says more about a shared way of doing party politics in some 22 That is the argument exposed by Grzymala Busse (2003) and O Dwyer (2004) for Eastern Europe countries. 17

districts rather than anything else. In any case, it is a self-reinforcing process. If patronage constitutes a key factor for electoral predominance and for keeping party systems closed, it is to be expected that the persistent control of the government in the absence of a strong opposition facilitates the usage of state resources with electoral purposes. Concluding Remarks: Patronage in Peronist and Radical persistence At the beginning of this paper I pointed out that it aimed to contribute to the understanding of the resilience of many sub-national party systems, and the consequent maintenance of the two-party system at national level as well. I argued that mainline explanations, focused on institutional design, are satisfactory but still insufficient, insofar as they do not account for disparities among provinces with similar institutional constraints. The notion of low maintenance constituencies presented by Gibson and Calvo proves to be also useful since it explains how some provinces manage to keep higher levels of public employment than others. However, nor this takes into account differences in the functioning of political systems in provinces with similar rates of employment. Low maintenance constituencies also performed differently, some showing more alternation than the others, some enabling the appearance of third parties while others did not. Another distinction between Argentine provinces is also common. On the one hand, the so called metropolitan provinces, those five districts where the bulk of Argentina s population and productive structure is located 23 ; on the other, the remaining nineteen districts, named as peripheral. If it is possible to win a presidential election just with the vote of metropolitan provinces (since they concentrate 70 percent of the population), governance at the national level requires legislative support coming from both areas. Due to the overrepresentation assigned to the sparsely populated peripheral provinces, 30 percent of the population holds 76.3 percent of seats in the Senate and 44.8 percent of seats in the Chamber of Deputies. While none of the metropolitan provinces was here classified as patronage-oriented, according to my findings eleven of the nineteen peripheral provinces were considered as highly or medium-highly patronage-oriented. These eleven provinces represent 16 percent of the national vote, but provide 45.8 per cent of the Senate seats and 25 per cent of the Chamber of Deputies. This implies that patronage, though not decisive for winning a national presidential 23 Province of Buenos Aires, the Federal Capital, Santa Fe, Córdoba and Mendoza. 18

election, is extremely relevant in terms of gaining institutional power, namely governorships and seats in both national chambers. Fragmentation and volatility impinge on the lowly patronage-oriented areas much more than on the others. This phenomena has made the presidential tier a more open and unpredictable one. Popular leaderships, loosely linked to any political organization, are able to gain large shares of votes there. Nonetheless, patronage oriented districts remain closed to new forces, thus ensuring the legislative predominance of the traditional major parties. A recent study labels the Argentine political system as wholly clientelist (De Luca et al, 2006). Paraphrasing Kitschelt (2000) I find it more appropriate to say that Argentine provinces exhibit two different predominant linkages between citizens and politicians. Briefly, in metropolitan districts, politics resemble Manin s (1997) Audience Democracy, politicians appealing to the public mostly through the media, and citizens selecting among personalities and issues much more than established political organizations. In telling contrast, in those provinces characterized as highly patronage-oriented the clientelist linkage and traditional parties predominate. Certainly, both types of linkages are present in every province. The functioning of the political machines described here are not circumscribed to particular districts. In fact, the most well known political machine (object of permanent criticism, media reports, and academic research) is the Peronist apparatus of the largest province of Buenos Aires. This study, however, is not concerned with the absolute amounts of political machines members but rather with its electoral reach. A big and well established machine does not suffice to win open elections by itself in big constituencies. In Córdoba, where more than 2.1 million people vote, Radicals developed a large apparatus, increasing its membership from 253.562 to 343.762 in four years (1995-1999), but just to lose provincial elections for the first time since 1983. Particularistic exchanges are definitely not confined to poor and underdeveloped areas (Piattoni, 2003; Kristinsson, 1996). Moreover, in advanced societies, patronage as the discretional allocation of appointments- becomes increasingly restricted to high officials and upper classes (Theobald, 1992). Nevertheless, it is noticeable that in societies in which poverty is rampant, political demands will be closer to the clientelist type than to the autonomous citizen s. As a recent study on the effectiveness of vote buying in Argentina shows, poor voters are more likely than the electorate as a whole to be involved in clientelist vote-seekers networks (Brusco 19

et al, 2004). The provinces that were here identified as patronage-oriented are within the socalled by O Donnell brown areas (1993), neofeudalized regions in which guarantees of democratic legality have almost no effectiveness. As O Donnell remarks, the parties operating there, even if they are nominally members of national parties, are no more than personalistic machines anxiously dependent on the prebends they can extract from the national and the local state agencies. Those parties and the local governments function on the basis of phenomena such as personalism, familism, prebendalism, clientelism, and the like. The mechanisms of partisan recruitment exposed above make it clear why party membership is higher in these poorer and traditionally acknowledged as less politicized provinces than in metropolitan areas. As table 10 shows, on average, in highly patronage-oriented provinces 11.2% of voters were provincial public employees and 24.8% were members of the ruling party by 1999. In contrast, in lowlypatronage oriented districts the figures were 5% and 14.2% respectively. (Table 10) Even when patronage is to be found at every province, it spreads in the form of diverse particularistic exchanges in some of them more than in others, thus having very different electoral impacts. Those in which the weight of patronage is heavier are those provinces less open to political competition. As Gordin suggests, the most important political source of patronage is the nature of political parties and more general party systems (2002). The nature of major Argentine parties, however, change substantially across Argentine provinces. Successful parties are those who manage to adapt themselves to changing situations and to divergent fields. In Argentina, it means that parties must have the ability to develop different kind of linkages with citizens, depending on the province they operate in. The Peronist Party have showed an enormous capacity to do it. It may function as a well-lubricated machine responding to a local boss who appears to be the provider of goods and services in the brown areas while, at the same time, portrays itself as a progressive centre-left modern party in the Federal Capital. Doing so, undoubtedly, the PJ surmounted the crisis. It controls national government, rules 15 of the 24 provinces, and counts with large legislative blocs that, glued by patronage and pork barrel, vote together backing the national executive in relevant issues. 20