Abstract
Are decentralization reforms in dictatorial regimes mere window dressing or can they deliver longevity for dictators? Many scholars in political science have argued that seemingly democratic institutions established in dictatorships play a crucial role in increasing the survival potential of the authoritarian regime by providing a channel to the dictator to co-opt members of the opposition, identify their bases of support and recruit effective coalition members. While most of the existing analysis has focused on national elections and political parties, we use primary data and social network methodology to analyze the 2001 decentralization reforms established during General Musharraf’s military regime in Pakistan. We show that seemingly democratically empowering decentralized governance structures can also be used to distribute rents to political supporters through ostensibly legitimate channels. Therefore, this paper adds to the literature on institutions in dictatorships on two fronts in addition to providing crucial information about the political landscape in a strategically important country.
Keywords
Do decentralization reforms, when established in dictatorial regimes, provide a veneer of legitimacy to the dictator, make the regime more democratic, or play a role in the dictator’s longevity? Decentralization reforms are generally understood as bringing people closer to the government and providing citizens with avenues for greater and more effective participation in decision-making processes (Dauda, 2006; Fung, 2004). They are also meant to improve development outcomes by increasing accountability in governance processes (Mueller, 2000). However, when General Musharraf’s military regime implemented decentralization reforms in 2001 in Pakistan, two years after the military takeover of the government, various groups in the society reacted with disdain. Political parties believed that it was an attempt by the dictatorial regime to weaken political forces (Mir, 2000), and the Human Rights Commission of Pakistan commented that the purpose ‘seems to be to depoliticize governance and to earn a lease of life for the military government’ (Quoted in ICG, 2004:10).
A quick look at developing countries suggests that the phenomenon of instituting some form of decentralization reforms within dictatorial regimes is not unique to Pakistan. China, the Democratic Republic of Congo, Rwanda, and Guinea, to name just a few, have established various forms of decentralization structures within dictatorial regimes. Despite their widespread prevalence, the objectives and incentives of creating decentralized governments in non-democratic regimes have not been explored. The literature that ventures into the incentives behind decentralization reforms assumes that the underlying system of governance is democratic. Most of these explanations suggest that it is economic development and globalization that result in public sector expansion, income and population growth, as well as integration of markets, and encourage decentralization by allowing better exploitation of economies of scale (Arzaghi and Henderson, 2005; Oates, 1972). Even when scholars go beyond the efficiency criteria to explore the motivations of the countries to decentralize, they do not include a non-democratic system of governance as a significant factor in their analyses (Burki et al., 1999). 1 Similarly, while there is a vast amount of literature exploring the impact of decentralization reforms on service delivery and other development outcomes, 2 the impact of these reforms on the workings of dictatorial regimes has not been investigated.
Recently a burgeoning political science literature (as summarized in the next section) that aims to understand the causes and consequences of establishing democratic institutions in authoritarian regimes provides insights into the motivation of dictators to establish decentralization reforms. Scholars have long noted that many dictators operate with some form of seemingly democratic institutions. 3 Recently, they have argued that authoritarian leaders establish these institutions to provide longevity and stability to their dictatorial regimes. While apparently similar in form, the function these institutions perform in authoritarian regimes is different from the function they are expected to perform in democracies. Seemingly democratic institutions provide dictators with a channel for negotiation with various groups in society so that they can co-opt members of the opposition (Gandhi and Przeworski, 2006; Lust-Okar, 2005), identify their bases of support, and recruit effective coalition members (Boix and Svolik, 2013; Brownlee, 2007; Magaloni, 2008).
This paper builds upon and contributes to this literature. We analyze the role that seemingly democracy-enhancing decentralization reforms instituted in 2001 played in the survival and stability of Musharraf’s military regime in Pakistan. The analysis is performed in two steps. First, we examine the local governance structure and political landscape within which these reforms occurred. This analysis shows that a patronage-based political landscape, dominated by family ties, motivated Musharraf to institute decentralization reforms, which allowed him to co-opt powerful local elites and distribute rents to them, all through ostensibly legitimate means: the ballot box. Second, using data on inter-governmental fiscal flows, we provide empirical evidence to show that decentralized governance structures were indeed used to distribute rents to the co-opted politicians, who in return provided political support to Musharraf, thereby helping him to increase his survival potential.
The existence of a patronage structure and its utilization to distribute rents is integral to the argument of this paper that the reforms helped Musharraf remain in power. A patronage-based political landscape made it easy for Musharraf to identify powerful potential supporters of his regime. Once the patronage structure dominated by family ties was formalized through decentralization reforms, Musharraf co-opted the local elite by distributing rents through decentralized structures, providing them with a veneer of legitimacy. In return for rents, local elites brought their political clout to support the dictatorial regime (see Waseem, 2006), which in turn helped strengthen Musharraf’s hold on power.
Scholars have noted that the limited number of cases available restricts our understanding of the politics of democratic institutions in dictatorial regimes (Gandhi and Lust-Okar, 2009). This paper adds a much-needed case. Furthermore, the examination of Pakistan’s case highlights a function of seemingly democratic institutions as an ostensibly legitimate channel to distribute rents that has not been discussed in the literature previously. As the literature review in the next section shows, scholars have identified co-optation, facilitating bargaining, power-sharing, and commitment enforcement as the functions that seemingly democratic institutions play in dictatorial regimes, but have not focused on rent distribution through these institutions. In addition, this paper is a valuable addition to the small literature on understanding local-level elections in authoritarian regimes, 4 while most scholarly attention in analyzing seemingly democratic institutions in authoritarian regimes has focused on elections at the national level. 5 The analysis presented in the paper relates to a specific context, but the findings provide lessons that can be applied toward a broader understanding of the workings of dictatorships. Finally, this paper contributes to the limited literature available on Pakistan’s political landscape.
Literature review: Seemingly democratic institutions in dictatorships
Theories in the existing literature that model a dictator’s behavior agree that mutual gain is possible as a result of exchanges between the dictator and citizens (Wintrobe, 1998), the elite and citizens (Acemoglu and Robinson, 2005), and the dictator and a select group of the elite (Gandhi and Przeworski, 2006; Mesquita et al., 2003; Schofield and Levinson, 2008; Svolik, 2009). The idea is that the dictator and citizens exchange political support for a decrease in repression. This exchange makes the life of a dictator less vulnerable to conspiracies, military coups, and violent rebellions and reduces the extent of repression on citizens. The mutual agreement also allows the dictator to increase his/her rents by relying on the support of ‘bankers to loan them money, peasants to produce food and scientists to do research’ (Gandhi and Przeworski, 2006: 6) and citizens (or specific groups of citizens) to receive rents from the dictator.
Researchers agree that these mutually beneficial exchanges rarely occur because promises are not enforceable. The absolute sovereignty of the dictator, the dictator’s unlimited discretion, by definition means that there cannot be any judges or other sources of power in the society that the autocrat cannot rule (Olson, 1993). In other words, dictators and citizens (or the elite ruling coalition) have a problem of credible commitment (North and Weingast, 1989) and of enforcing commitments (Przeworski et al., 2000). Scholars argue that seemingly democratic institutions help the dictator and his participants in the exchanges to solve these commitment problems, and in turn make the dictatorial rule more stable.
For example, Gandhi and Przeworski (2006) contend that dictators, by co-opting opposition forces into the legislature, can credibly commit them to confer some degree of participation in policymaking, as well as offer other perks and privileges in exchange for their political support. In other words, by creating the opposition’s stake in the system, the institutions can transform opponents into supporters (Wright, 2009). Lust-Okar (2005) suggests that political parties in authoritarian regimes play a similar role.
Similarly, legislatures and political parties create a forum in which dictators and various groups can bargain with each other for mutually beneficial exchanges (Gandhi and Przeworski, 2006). The legislatures and political parties can serve as legitimate avenues in which ‘demands can be revealed and agreements can be hammered out’ (Gandhi and Przeworski, 2006: 14). The ‘legitimate’ bargaining process through these institutions also provides a mechanism for the dictator and opposition forces to share information about their preferences (Boix and Svolik, 2013).
Building on this literature, we analyze the role that local elections and decentralization reforms play in increasing the stability of authoritarian regimes, by examining the case of decentralization reforms instituted in Pakistan in 2001 by military ruler General Musharraf. Before delving into the details of the reforms, we briefly summarize Pakistan’s political history and describe the context that enabled and motivated the military dictator to co-opt local elites and distribute rents through a decentralization structure.
Context of decentralization reforms: History and motivations
Pakistan’s political landscape
Since its independence in August 1947, Pakistan has struggled with formulating a coherent governance structure. In that struggle, three military dictators have ruled Pakistan for about half of its life. Almost all scholars of Pakistan’s political history, including Jalal (1999), Alavi (1983), and Rizvi (2003) believe that the seeds of Pakistan’s political trouble were sown at the time of its independence. They argue that the major political party that represented the idea of Pakistan in the subcontinent’s efforts of independence, the Pakistan Muslim League (PML), failed to amass broad-based support from the Muslims of India. Lacking organizational machinery, the PML continued to rely on already established and tested patronage, caste and communal modes of mobilizing voters, as well as communication with the masses through landlords and local power brokers (Jalal, 1999; Ali, 1983).
Politics in Pakistan thus came to be based heavily on patronage: a trend that has been strengthened throughout the country’s history (Waseem, 2006; Wilder, 1999). A politician’s political clout depends largely on his/her influence in the locality, efforts to mobilize financial resources, and community ties rather than on ideology or policy positions. Thus, politicians can trade party loyalties without losing political support (Waseem, 2006).
Military regimes have not only strengthened this patronage-based structure of politics, but have also benefited from it. The dictator only has to buy off support of the local elites and politicians, who will then bring their own political clout to support his/her regime. Three dictators who came to power in Pakistan relied on the local elite to build their own political base and consolidate their regime. The first dictator, General Ayub, who led a coup in 1958, created a new faction of the Muslim League and called it the Convention Muslim League. The second dictator, General Zia, who came to power in 1977, created a new PML. Members of these parties held various ideological or policy positions; they only shared the fact that they carried local influence (and with that, the ability to provide political support) and a willingness to go along with the military regime for whatever patronage it could bring (Jones, 2003). Hence, both powerful local politicians and military dictators in Pakistan have leveraged each other’s power to sustain themselves.
Political parties have also not been willing to challenge the reliance of politics on patronage structures. As norms of patronage progressed, politics became increasingly localized (Ziring, 1988) and parochial identities became the ‘vehicles of mobilization’ (Hasan Askari Rizvi quoted by Ziring, 1988: 19). Consequently, as the worldview of the politicians came to be shaped by their local, tribal, or ethnic links, political parties lost any incentive to coalesce for national purpose or put aside their parochial concerns (Rizvi, 2003; Ziring, 1988). Power of the local politicians does not translate to citizens’ power, as citizen-politician relationships are also based on patronage. Political parties have remained little more than an agglomeration of local elites whose main objective is to obtain access to state patronage. When political forces and civil society have tried to dilute the impact of local elites on national politics, the patronage structure has shown remarkable tenacity (Jones, 2003).
History of decentralization reforms
All the non-representative governments in Pakistan so far have attempted to create local government structures to co-opt the powerful local elite; they have tried to formalize the patronage structure in different ways. General Ayub Khan, who first imposed martial law in Pakistan, instituted a system of Basic Democracies (BD). The BD system was a multi-tiered, interlocking, and interdependent system of institutions, at the base of which were the local bodies in rural areas, towns, and larger cities (Jones, 2003). The local bodies were composed of 10–15 members called Basic Democrats (BD), who were elected by a universal franchise and became the Electoral College for presidential elections. Local bodies also had severely restricted fiscal capacity, which prevented them from performing any significant function.
In 1971, a few days after assuming the powers of president (later Pakistan’s first elected prime minister), Zulfikar Ali Bhutto abandoned the local government system after declaring that ‘it has bred nothing but nepotism and corruption, a system that reduced democracy to a farce’ (Rizvi, 1980: 228). Although two Local Governance Ordinances (LGOs) were introduced in 1972 to institute a different system of local government, they were not implemented. The 1973 Constitution of Pakistan failed to allocate local government any recognition as the formal third tier of government (Cheema et al., 2005).
The second round of decentralization reforms was initiated in 1979, after General Zia took over the government and imposed martial law through the LGO. Cheema and Mohmand (2003) conclude that the 1979 LGO was not significantly different from the 1959 BDO system in terms of functions and financial powers available to the local governments. Three local body elections were held under the 1979 LGO.
The role of local governments was again marginalized after democracy returned to the country in 1988. National and provincial assemblies assumed some functions that had been assigned to local government previously under the 1979 LGO, and were eventually dissolved. Various reasons were presented for their dissolution, among which mismanagement and corruption of local government were the most dominant (Zaidi, 2005).
Musharraf’s regime and 2001 local governance reforms
In 1999, against this backdrop, martial law was instituted and General Musharraf, then Chief of Army Staff, assumed the specially created office of chief executive. In the immediate wake of the military takeover, the constitution was suspended and parliament dissolved. Political parties were disqualified and some politicians were banned.
Empowerment of local government became a priority in Musharraf’s agenda from the beginning. He touted local governance reforms as an attempt to institute ‘real’ democracy rather than ‘sham’ democracy in the country (Waseem, 2006). The local government system, introduced in August 2001 through another Local Government Ordinance, was more comprehensive and significantly more ambitious than the previous reforms in its objective to transfer meaningful power to grassroots organizations.
The constitution was reinstated after the first round of local elections was held under the 2001 LGO. A referendum that confirmed Musharraf’s position as president of the country for the next 5 years was also held prior to the reinstatement of the constitution. Following his predecessors, Musharraf had created a new ‘wing’ of the PML, the PML-Q group or the Q League, with himself as its leader. Many previous members of the then major political parties joined the new party. When parliamentary elections were held in October 2002, the PML-Q won the majority of the seats in national and provincial assemblies. A prime minister was appointed; however, most powers had been transferred to the office of president through the Legal Framework Order, ratified by parliament in 2003 (Aslam, 2011).
2001 local governance reforms: Instruments of co-optation and rent distribution
Under the 2001 LGO, Pakistan is a federation divided into four provinces, namely Punjab, Sind, Baluchistan, and Khyber Pakhtunkhwah, plus a federally administered tribal area and the Islamabad Capital Territory. Local governments form the third tier of the governance structure and are further divided into three tiers, namely district, municipal administration, and union council from the highest to the lowest tier. Union council is further divided into villages. Each tier has its own council and a council head or mayor and deputy mayor (Cheema et al., 2005).
The reforms transferred key provincial functions to district and municipal levels. The district mayor is the executive head of the district administration. In this section, we aim to highlight the structural features that we argue allowed Musharraf to make use of patronage culture to co-opt locally powerful elites.
Elections on a non-party basis
Local elections under the 2001 LGO were held on a non-party basis after all political parties and some politicians were banned. Elections allow authoritarian regimes to identify their most valuable colluders (Blaydes, 2010; Brownlee, 2007); the politicians who win local elections have shown that they have significant political clout in their locality and can garner political support for the non-democratic regime. A political landscape based on patronage provides another advantage: since politicians’ power depends on their personal clout rather than their party affiliation, they can easily shift alliances or ‘reset’ their political affiliation. The non-party basis of local elections under LGO 2001 especially facilitated these shifts.
Bypassing the provincial level
At the time of the local elections, no provincial or national elected governments existed. The focus of the governance reform was the devolution of provincial powers to the local level, not the devolution of federal powers to provinces (Cheema et al., 2005). The decentralization reforms, therefore, created direct administrative and political links between the central and the local government, bypassing the provincial level. Even when the provincial assemblies were reconstituted in 2003, 2 years after the first local elections, the district mayor continued to be linked directly to the central government (Cheema et al., 2005; Waseem, 2006).
Powerful, indirectly elected district mayor post, and discretionary transfers
One of the most significant aspects of the 2001 LGO that has made it prone to be used as a co-optation mechanism is the indirect election of the most powerful actor in the local government (World Bank 2009): the district mayor. Members of the district council make up the electoral college for the district mayor, when members of the district council themselves are indirectly elected by members of the union councils that fall within that particular district. The union council is the only directly elected tier of the local government system.
In this system, the preferences of citizens are reflected in policy-making only to the extent that district mayors are willing to listen to the union councilor; since the re-election of district mayors depends on the satisfaction of union councilors with the district mayor, it is hoped that union councilors would be able to discipline the district mayor. However, in a political landscape dominated by patronage politics, indirect election of the head of the local government unit opens up the possibility of the local elite capturing these offices. Hasanain (2008), while analyzing the impact of the 2001 LGO on accountability relationships, observes this to be the case. District mayors, he claims, have greater independent political standing than union councilors. Union councilors seek the approval of district mayors rather than the other way around; the councilors must toe the district mayors’ line to build their political future.
Therefore, by creating a powerful office of district mayor that was likely to be captured by the local elite, and creating a direct link between the central government and that office, the military regime was able to achieve two goals. It created a position through which it could directly control local government and created a mechanism through which it could co-opt the local elite.
Discretionary transfers to district mayors
The role that decentralization reforms played in the co-optation of locally powerful politicians becomes even more significant when we consider the rampancy of discretionary fiscal transfer to the districts. While district councils have the power to levy certain taxes, they depend largely on inter-governmental transfers, a large portion of which is earmarked for personnel costs (Cheema et al., 2005). As a result, a large number of funds are disbursed to local government through discretionary transfers. Hasanain (2008) estimates that in 2006–2007, discretionary transfers to local governments in Punjab totaled Rs. 14.5 billion, which is larger than the local government’s allocation through rule-based inter-governmental transfers.
Discretionary transfers provide a mechanism for the military regime to distribute rents to the politicians. Even within democratic structures, discretionary transfers can become a political decision (Lockwood, 2002), but when practiced in a dictatorial regime, they can become an effective channel of rent distribution through ostensibly legitimate means. Rent distribution is a tried and tested method of co-optation. 6 In exchange for access to rents, elites are expected to provide political support to the regime, bolstering its legitimacy and increasing its potential for survival.
Summing up the discussion
A deeper look at the structure of the 2001 LGO shows that local government elections and structures made it easier for the dictator to co-opt locally powerful politicians. Patronage culture, on the one hand, motivated Musharraf to co-opt local elites to build a support base. Decentralization reforms, on the other hand, formalized the patronage structure to make co-optation possible. Through these reforms, Musharraf was able to amass political support and build a coalition that helped him consolidate his rule. Though decentralization reforms were only one instrument in Musharraf’s survival strategy, they were the first tools he applied.
The question, however, remains: could such co-optation and rent-distribution occur without the 2001 LGO? If so, what was the significance of these reforms? Other scholars have also pondered these questions, and two answers have come out of this literature. First, institutionalized co-optation results in more durable ruling coalitions that succeed under less favorable circumstances than co-optation without institutions. Institutionalization provides both the dictator and his supporters with a better ability to verify each other’s commitments over long periods (Magaloni, 2008; Boix and Svolik, 2013). Second, dictators seek ostensibly legitimate means for co-optation and rent distribution (Gandhi and Przeworski, 2006). The desire to seek legitimacy may stem from reducing the risk of popular revolution or from the desire to not be labelled as an autocratic regime by the international community. Dictators in Pakistan have always sought to attain political legitimacy; they care about how Pakistani citizens and the international community see them (Alavi, 1983; Waseem, 2006).
To provide credence to our claim that local governance structures were indeed used to co-opt politicians and to distribute rents, we empirically test whether discretionary fiscal transfers to the districts are linked to the district mayors’ political clout and the degree of his relationship with the dictator. This analysis is restricted to only one province of Pakistan (Punjab) because of the availability of data. Admittedly, the results of this analysis are not generalizable. However, since the focus of this study is to explore what happened in Punjab at one point in time, the inability to generalize does not affect our argument.
Mapping rent distribution through local governance structures
Specifically, we test two hypotheses: (1) the more political clout a district mayor has, the greater the fiscal transfers toward his district; and (2) the more connected a district mayor is to the military dictator, the greater the fiscal transfers to his district. Hypothesis 1 tests whether the military regime co-opted established politicians who are entrenched in the political landscape by distributing rents to them. Hypothesis 2, in addition to further reinforcing hypothesis 1, tests whether the military regime also created specific relationships to individual politicians in order to solicit their political support in return for rents.
To test these hypotheses, we use social network methodology: a method of analysis that seeks to understand the impact of social structures and relationships. At the core of this methodology is a social network depicted in a graph form that shows relationships among various entities. Nodes represent entities, and lines depict relationships among them. This methodology is appropriate for this analysis since the subject of our analysis is the relationship of politicians with each other and with the dictator. We use UCInet for the depiction and analysis of network data.
Mapping out the network
The first step in the methodology is to create a network. Data was collected for 31 of Punjab’s 33 districts. We start with district mayors elected in the 2005 local government elections in the Punjab province, and document their connections with other politicians. In the next step, we document the connections of these politicians to other politicians and to the military dictator. This is called the snowballing method of sampling. We categorize connections in three ways: (a) familial; (b) informal political connections measured by endorsement or support in the election campaign; and (c) formal political connections measured by affiliation with political parties. Familial networks include first relatives, including first cousins and first in-law relatives. Close personal friendships are also included in this network. We use an endorsement of one politician for the other during the election campaign of the 2005 local elections or the 2008 general elections as an indicator of an informal political relationship. A formal political relationship is the relationship between a politician and a political party. The network is built according to relationships observed in 2005. 7
Figures 1 and 2 depict familial and formal networks, respectively. Figure 3 combines all these networks. Figure 1 shows that the majority of the district mayors are related to the politicians at the local or the national level through family ties; they come from politically connected families. They are influential, central, and connected to other actors in the network. The network becomes even denser if we include informal and formal political ties (Figures 2 and 3).
Familial networks of district nazims.
Formal political networks of district nazims.
All networks.
Analyzing patterns of fiscal transfers
We wish to analyze the impact of two independent variables: political clout (hypothesis 1) and degree of connectedness to the dictator (hypothesis 2). The dependent variable is the fiscal flow to the district.
We assess the political clout of a district mayor by measuring how embedded he is in the network. The more embedded an actor is within a network, the greater his political power in terms of influencing the outcome of interaction among participants of the network (Knoke, 1990). Using the network we created, we can measure the embeddedness of each district mayor in different ways. Table 1 gives a list and definitions of the measures of embeddedness. The simplest measure of embeddedness or centrality is the number of pairs that each politician is a part of. We also measure the degree of connectivity of a politician’s immediate contacts, i.e. if the other politicians he is connected to are influential and central in their own right (Bonachich power), how many politicians he is connected to (reach central and eigenvector of geodesic distances) and how critical he is to other connections (nBroker).
Centrality and distance measures.
In assessing the degree of connectedness to the military dictator, our main goal is to capture the ease and efficiency with which the district mayor can communicate with the dictator. We measure connectedness to three different entities: (a) the dictator himself; (b) individuals Chaudhry Pervez Elahi and Shujaat Hussain, who were the most ardent supporters of the military regime and were the gatekeepers to Musharraf (Aslam, 2011) (see Table 2) that shows the centrality of these individuals in the overall network); and (c) PML-Q, the political party that Musharraf created. Table 1 shows various measures of connectedness that we can calculate using social network methodology. Geodesic distance measures how efficient a path is, while the number of geodesic distances measures whether more than one efficient path is available for a district mayor to communicate with Musharraf and other power holders. Maximum flow calculates the total number of distances from the politician to the power holders.
Network characteristics and major actors.
To construct the fiscal flows dependent variable, we obtained data from the annual development plans of district governments. We only include the funds that are transferred from the central government to districts under special grants, and the funds that are transferred from provinces out of provincial retained funds. Transfers made through rule-based inter-governmental transfers are not included in this estimation, as we are only interested in discretionary transfers. Some of the funding sources include the President’s Program of Education, the Chief Minister Accelerated Program, the Social Action Program, and the Chief Minister Package. We aggregate all these fiscal sources into one category and use two indicators of fiscal flows: the number of total projects funded by discretionary transfers in each district, and the total amount of funds allocated to a district.
We use a simple ordinary least squares (OLS) regression to test whether embeddedness of the district mayor in the network or his degree of connectedness to the dictator impacts fiscal flows. The general form of the OLS equation is as follows:
where Y is the the magnitude of discretionary transfer and X is the measures of embeddedness and measures of connectedness to the dictator
Our primary purpose is the historical analysis of a particular social network. We have not included any control variables in the equation. Arguably, the backwardness index, population, and especially the political importance of the district for provincial politician of a district are important variables that can affect the fiscal flows to a district. Omitting variables in an OLS analysis makes inference difficult. However, the data included in the equation is a population in itself, as the constructed network includes all observed links of 31 out of 33 district mayors in Punjab (two districts were omitted due to a lack of data). The results obtained from the analysis are explanatory, limited to the particular network we are studying. Therefore, potential bias and inconsistency in the estimators caused by omission of potentially relevant variables does not affect our results.
Many tools of standard inferential statistics (for example, standard estimated errors and test statistics) do not apply directly to the network data, since the network data observations are not independent samplings from populations. The network characteristic of every individual in the network is dependent on the location of another actor in the network. Therefore, applying standard formulas for computing standard errors and inferential tests on network attributes can give misleading results.
In order to ameliorate these issues, we employ multiple strategies. First, we normalize the network measures to reduce the degree of dependence of one observation on other observations, which allows us to test robustly the presence or absence of relationships between the network and the independent variables. We also use node characteristics as variables rather than connections between the two nodes to lessen the degree of dependence. In addition, we estimate standard errors and significance using the random permutations method for constructing a sampling distribution of R2 and slope coefficients. 8 However, the point estimators are not efficient, which means that the results cannot be generated beyond the network that we are observing. For our study, this does not create a problem since our primary purpose is the historical analysis of a particular social network.
Results and discussion
Tables 3 and 4 show the coefficients of the fiscal flow variable when regressed on network variables. Only the coefficients that were significant at 10% level of significance are shown. The P-values are shown in parentheses.
Relationship between distance to power holders and discretionary funds.
Relationship between centrality in the network and discretionary fund.
The signs of the coefficients of fiscal flow indicators are as expected. In general, a smaller distance from the power holders and embeddedness within the political network of the district mayor increases the flow of discretionary resources to those districts. Both the distance and the centrality measures come out to be important determinants; politicians who are central to the network as well as the politicians who are located close to the main power holders in the network receive greater financial transfers. This could also mean that the politicians who are embedded in the network are also close to Musharraf and other power holders. Or it could mean that the closeness of these actors to Musharraf and other power holders makes them more central to the network, as Musharraf and the main power holders are themselves centrally connected. Either way, these results clearly show that in Punjab, the decentralization structure was used by the dictator to transfer rents to powerful politicians strongly embedded in the political network. Politicians who were affiliated with the central government through their relationship with Musharraf or because of their centrality in the network were systematically favored in the distribution of discretionary fiscal resources. 9 These local politicians, in return, provided political support to the regime.
This analysis shows therefore that Pakistan’s political landscape, based on patronage and dense family ties, provided opportunities to Musharraf to co-opt powerful local politicians in return for their political support. A decentralized structure provided an institutionalized way of co-optation and of distributing rents in an ostensibly legitimate way. It also helped solve the commitment problem between the dictator and the political elite. These results are not only a testament to Musharraf’s clever strategy to amass political support through ostensibly legitimate means but also to the power of the local elite; both local elites and the military dictator feed off each other’s power.
The statistical results we obtained pertain to the specific social network we analyzed, i.e. the network of district mayors in Punjab, Pakistan. However, our findings have wider significance for understanding the role of seemingly democratic institutions in dictatorships. Our analysis illustrates that decentralization reforms in dictatorships can be used strategically by dictators to provide longevity to their regimes and are not always a step toward democratization. Our findings also demonstrate how seemingly democratic institutions can be used to distribute rents to political supporters of the dictatorial regime: a role hitherto ignored in the literature.
These phenomena should be studied in other cases to examine their generalizability. While we have used social network methodology to empirically test whether a dictator distributed rents in return for political support to politically powerful actors in Pakistan, other scholars may choose a different methodology to examine whether similar patterns exist in other dictatorial regimes.
Conclusion
Decentralization reforms are often understood as being consistent with democratization. However, in this paper, we show that this is not always the case; instead, dictators can use decentralization reforms to amass political support for their regime in a manner that can help lengthen their tenure. These findings complement recent literature studying democratic institutions in dictatorships that has suggested a much more sophisticated picture of dictatorships. Scholars have challenged the views that autocrats used coercion and repression to stabilize their regime, and that they used nominally democratic institutions as mere ‘window dressing.’ New literature suggests that these institutions are used as instruments of co-optation, channels of negotiation among political forces and tools for building political support. In addition to affirming the potential role of seemingly democratic institutions in the co-optation of politically powerful actors to support dictatorial regimes, we highlight a function of seemingly democratic institutions, an ostensibly legitimate channel to distribute rents, that has not been discussed in the existing literature.
Although the current analysis focused on one case, it provides significant lessons that can be applied toward understanding the workings of other dictatorial regimes. Our analysis suggests that decentralization reforms, when instituted within dictatorial regimes, should not be taken at face value as a step toward democracy, as these reforms can be used strategically by the dictator to bolster his regime. The role of decentralization reforms should be tested in other cases as well to examine whether the same phenomenon occurs in other contexts. Further research should aim to explore whether decentralization reforms in authoritarian regimes have performed similar roles as they did in Pakistan; as stated earlier, many testable cases of this phenomenon exist in which dictators have instituted decentralization reforms. Similarly, scholars should explore other cases in which seemingly democratic institutions, not just decentralization reforms, but also other institutions such as legislatures and multi-party systems, are used as ostensibly legitimate channels of rent distribution.
Notably, a recent study by Shirah (2015) has presented a counterargument. He argues that while seemingly democratic institutions such as elections may offer certain benefits to dictators, they also generate vulnerabilities for these regimes in the long run by providing opportunities for mass resistance. Could this vulnerability also apply to the 2001 decentralization reforms in Pakistan? Could the decentralization reforms that helped Musharraf amass political support at the time have played a role in his downfall in the long run? While such an analysis is beyond the scope of this paper, these are crucial questions that open up the debate to broader ones, such as whether and how seemingly democratic institutions established by dictators to stabilize their regimes can play a role in democratic transitions in the long run, and if so, in what circumstances.
Footnotes
Acknowledgements
I would like to thank Zahid Hasanain for compiling the data on fiscal inter-governmental transfers and for allowing me to use his analysis in my paper. I would also like to thank the anonymous reviewers for their constructive feedback.
Funding
The author received no financial support for the research, authorship, and/or publication of this article.