Tribal Mobilization,Fragmented Groups,and Public Goods Provision in Jordan

Abstract

Diversity has been blamed for poor public goods provision in a number of different contexts. It is associated with reduced spending on services, meager rates of tax collection, and poor policies. I argue in this article, however, that in semidemocratic or authoritarian countries, where political parties are weak, diversity can be an important source of electoral competition, leading to better services. In diverse communities where multiple identity groups are politically mobilized, candidates are forced to seek the support of voters outside of their group, who are more likely to vote based on qualifications than on group affiliation, resulting in better public officials who provide superior services. Moreover, I find internal group fragmentation to be important in understanding the impact of heterogeneity on public goods provision: Candidates in areas where only one identity group is politically mobilized but where that group is politically fragmented will also seek votes from other groups within the community similar to candidates in locations with “multigroup” mobilization, leading to improved public goods provision. These arguments are confirmed through the analysis of tribal mobilization and public goods across Jordan’s municipalities.

Keywords

Middle East non-democratic regimes subnational politics corruption and patronage

Introduction

Diversity has been blamed for a number of social and political ills. It is said to be responsible for poor economic policies in Africa leading to its “growth tragedy” (Easterly & Levine, 1997). Racially and ethnically diverse communities have been associated with reduced spending on service items and lower rates of tax collection (Alesina, Baqir, & Easterly, 1999). These same communities have been found to suffer from the poor provision of public goods whether it is fewer paved roads, lack of school facilities, or poor maintenance of infrastructure (Bardhan, 2007; Kimenyi, 2006; Miguel & Gugerty, 2005).

At the same time, recent research has shown that diversity is not consistently associated with poor service provision. An analysis of tribes in Yemen found tribal diversity to be accompanied by higher levels of educational resources allocated by the central government. Areas with more tribes also possessed more teachers and classrooms per capita (Egel, 2012). Polat (2012) noted a positive relationship between ethnic heterogeneity and public goods provision in Jordan. A number of studies have also highlighted a neutral or positive influence of diversity on public goods within the United States (Boustan, Ferreira, Winkler, & Zolt, 2010; Hopkins, 2011; Putnam, 2007; Rugh & Trounstine, 2013) as well as outside of it (Gerring, Thacker, Lu, & Huang, 2015; Gibson & Hoffman, 2013; Glennerster, Miguel, & Rothenberg, 2013; Miguel, 2004; Singh, 2010). Alongside these works are further studies that qualify the conditions under which diversity impedes services. Baldwin and Huber (2010) found that only when groups differed socioeconomically were public goods affected. However, Waring (2011) and Waring and Bell (2013) note that hierarchical stratification need to be present for heterogeneity to have a negative impact. Gisselquist (2014) concludes that the nature of the relationship depends on the type of public good with ethnic diversity having no impact or a positive influence on some goods.

Like these recent studies and those included in this special issue, I argue that diversity does not necessarily impair public goods provision; rather, in settings where political institutions encourage candidates to compete for the votes of those outside of their identity group, heterogeneity is positively associated with services. Unlike previous studies, however, I focus on a new mechanism—that of political competition. Where political parties are weak, diversity is an important source of electoral competition, urging candidates to appeal to voters outside of their group, who are not bound by group affiliation but support the candidates they believe will provide the best services. I also argue that in homogeneous settings where only one group is mobilized that if the dominant group is electorally fragmented then candidates will similarly be forced to seek the support of those outside of their identity group to win elections. In other words, I find good public goods provision in locations with “multigroup” mobilization because of high levels of cross-group voting but that in communities with “monogroup” mobilization, the quality of services to be dependent upon the level of political fragmentation. Locations with monogroup mobilization but where the mobilized group is politically fragmented provide good services whereas locations with monogroup mobilization where the group is cohesive do not.

This article makes several contributions to the study of heterogeneity and public goods provision. Empirically, I use an original dataset that measures diversity by counting the number of tribes offering candidates in recent municipal elections in Jordan. Unlike many studies of diversity and public goods provision, I enumerate the number of groups that matter politically rather than utilizing a demographic measure that captures absolute levels of diversity without distinguishing between groups that are or are not politically active. Theoretically, I demonstrate that multigroup mobilization can actually have a positive impact on public goods provision and offer a new mechanism based on political competition for understanding this phenomenon. Finally, this study emphasizes that when considering the relationship between group mobilization and public goods provision, internal “heterogeneity” can be as important as external heterogeneity.

Identity Group Mobilization, Fragmentation, and Public Goods Provision

In one of the first pieces to highlight the negative impacts of diversity, Easterly and Levine (1997) find that ethnic heterogeneity hinders growth via poor economic policies. They explain that ethnic groups have differing preferences, making it difficult to compromise and coordinate government policy. As a result, officials pursue policies that benefit their own group but not the collective. A number of other studies have confirmed this negative relationship. Heterogeneous communities have been associated with lower expenditures on education, roads, sewers, and trash collection (Alesina et al., 1999), decreased access to public goods (Banerjee, Iyer, & Somanathan, 2005; Banerjee & Somanathan, 2007), and less cooperation in the production of these goods (Bardhan, 2000; Vigdor, 2004). Diverse communities also fare worse in raising school funds and maintaining school facilities (Addison & Rahman, 2001; Miguel & Gugerty, 2005).

Despite what might seem to be emerging consensus, there are reasons why we should be wary about whether heterogeneity is associated with poor services. Although many different types of diversity (i.e., religious, linguistic, cultural) have been studied, the bulk of the literature focuses on ethnic diversity, with most scholars (i.e., Alesina et al., 1999; Easterly & Levine, 1997; Miguel & Gugerty, 2005) using demography to measure diversity and to calculate the ethnolinguistic fractionalization (ELF) index. One weakness of this index, however, is that it does not differentiate between reinforcing and cross-cutting cleavages (Dunning & Harrison, 2010; Gubler & Selway, 2012; Selway, 2011a, 2011b, 2015; Stewart, 2002). ELF by itself is an insufficient measure because it does not take into account differences between groups such as the distribution of income (Baldwin & Huber, 2010; Lupu & Pontusson, 2011; Østby, 2008), their geographic distribution (Matuszeski & Schneider, 2006), or the distribution of power across groups (Cederman & Girardin, 2007). Nor does it account for the size of ethnic groups relative to one another as captured by a polarization index (Montalvo & Reynal-Querol, 2005). Finally, as Birnir (2007) and Selway (2015) demonstrate, ELF ignores political institutions which influence whether ethnic groups have a stabilizing or destabilizing effect on democratic politics and whether multiethnic societies underprovide public goods.

A second weakness of the ELF index is the use of demography as the foundation for capturing diversity. Demography by itself is a poor measure as it fails to distinguish between ethnic structure, defined as the “distribution of descent based attributes that all individuals possess” with ethnic practice or the “act of using one or more identities embedded in this structure to guide behavior” (Chandra & Wilkinson, 2008, p. 523; Lieberman & Singh, 2012). A listing of ethnic groups and their memberships by itself does not tell us which ethnic identities are politically activate and ignores the possibility that other cleavages—like language, religion, or class—might be more salient (Posner, 2005).

In the tradition of these critiques, I argue in this article that it is not diversity per se that matters but whether political institutions encourage groups to compete for the votes of members outside of their specific identity group. This argument is similar to that proposed by Selway (2015) and Gibson and Hoffman (2013) in explaining the positive relationship between diversity and public goods provision, but while they focus on the formation of broad-based coalitions across ethnicities, the mechanism outlined here is political competition. In the theory described here, I posit that political competition improves public goods provision in communities and describe two sources of competition: the first is the result of the mobilization of multiple identity groups (“multigroup” mobilization), while the second occurs in communities where only one group is mobilized (“monogroup” mobilization) but where that group is fragmented into branches, clans, or other subgroups. The argument and its mechanisms are outlined in the arrow diagram and explained in detail below. In this article, I explore only the relationship between group mobilization, fragmentation, and service provision.

Multigroup Mobilization or Fragmentation of One Group → Heightened Political Competition → Better Elected Officials → Improved Service Provision

In countries where identity politics are relevant, identity groups, regardless of the type of cleavage (i.e., ethnic, religious, linguistic), field candidates or use parties as a cover for their own political activities. Chandra (2004, 2007) notes that in India, limited information and the importance of patronage encourage ethnic voting. In a number of Arab countries, such as Jordan, Kuwait, and Yemen, tribes play an active part in nominating candidates (Langston, 2005; Layne, 1994; Tetreault, 2000). Similarly, in Zambia, the main political cleavage is linguistic (Gibson & Hoffman, 2013).

When only one group (i.e., one tribe or one ethnic group) is present in a community, group leader(s) can control the number of candidates in elections. However, this is difficult when several groups (i.e., several tribes or several ethnic groups) participate in politics. In communities where multiple groups are mobilized, it is unlikely that one group will obey the directives of another unless they are all members of the same electoral alliance. Consequently, more candidates are likely to run in communities with multigroup mobilization, forcing candidates to appeal to a broader audience to win elections. In locations where multiple groups are mobilized, no single group is likely to have enough members to win elections outright.

Indeed previous studies have emphasized how social heterogeneity heightens political competition. Arguing that social cleavages are often the basis of political preferences, Powell (1982) found that social heterogeneity increases the effective number of parties. In a study of Louisiana state elections, Jones (1997) noted a positive association between racial heterogeneity and the effective number of candidates. Ordeshook and Shvetsova (1994), Neto and Cox (1997), and Taagepera (1999), however, assert that heterogeneity does not have an additive effect, but rather, its influence depends on district magnitude.¹ When district magnitude is one, heterogeneity has virtually no impact on the effective number of parties, but at high levels of district magnitude, diversity and the number of parties are positively correlated. Stoll (2013) confirms these findings by examining a variety of sources of heterogeneity such as sex, foreign birth, geographic location, and socioeconomic background on party formation. Although different sources of heterogeneity have varying effects on the effective number of parties, the relationship is always positive.

When there is a candidate from one’s own identity group, members are also likely to support that individual but for reasons other than her ability to provide public goods. First, the candidate’s ability to provide patronage is often paramount in authoritarian elections, with group members eager to secure a seat for their candidate so that they are accorded priority (Auyero, 2000; Blaydes, 2011; Chandra, 2007; Shehata, 2008; Valeri, 2009). Second, each identity group can easily mobilize its members because of existing networks (Besley, Coate, & Loury, 1993; Scarritt & Mozaffar, 1999). Habyarimana, Humphreys, Posner, and Weinstein (2007), in a series of experiments in Uganda, found that members of the same ethnic group located co-ethnics with greater ease than those outside of the group. Third, findability is important because if members do not support a candidate, then they can be sanctioned. Even with secret ballots, it is possible for candidates to determine how members voted by employing a variety of techniques (Brusco, Nazareno, & Stokes, 2004; Stokes, 2005).

These reasons suggest, therefore, that candidates who receive most of their support from their own group members are likely to be inferior in terms of providing public goods compared with those who garner votes from many groups. Members may be pressured to support the candidate from their own group, regardless of his or her qualifications. Non-group members, however, are at greater liberty to support whomever they choose. Political competition, therefore, produces better public goods because it leads to the election of more competent public officials.²

Previous scholars have also highlighted the ability of electoral competition to improve public goods. In a study of antipoverty programs in Mexico, Hiskey (2003) found that multiparty electoral competition was associated with better provision of water, sewage, and electricity. Furthermore, the programs themselves had a larger and positive impact on these services in locations with competition compared with locations where the dominant party, the Instutional Revolutionary Party (PRI), was certain to win. Similarly, Hecock (2006) noted heightened electoral competition led to greater education spending in Mexican states. Other studies have found competition influences the characteristics of officials or their behavior. In China, villages holding elections were more likely to result in the selection of officials who shared similar views to those of residents (Manion, 1996). The more free and fair elections, the more likely village officials were also to advocate on behalf of their residents (Birney, 2007).

The argument thus far assumes that political competition is low in locations where only one group chooses to participate in politics. But there are also circumstances when high levels of political competition are present in these communities. Locations where only one group is mobilized also have competitive elections if the group in question is politically fragmented. Fragmentation can be defined as the inability of a group to stick together and remain united in the pursuit of instrumental objectives and/or for the satisfaction of members’ affective needs (Carron, Brawley, & Widmeyer, 1998, p. 213). A fragmented group is one where members do not support one another or work together even though doing so would help them achieve an outcome beneficial to the entire group. Despite attempts to impose a decision upon the group, implementation is impossible because of the lack of commitment of members to act in unison (Beal, Cohen, Burke, & McLendon, 2003).

There are various ways that fragmentation manifests itself in society, but this article focuses on fragmentation during elections. During these periods, the group leadership might decide to nominate a certain number of candidates, but if fragmentation is high, then various subdivisions of the main group will also contest elections and ignore the orders of the leadership at large (Glennerster et al., 2013; Kotler-Berkowitz, 2001). Under these circumstances, candidates, even though they are members of the mobilized group, will need votes from members of unmobilized groups in the community and/or from members of other subdivisions within their own group. Multiple candidates representing the same group may also provide greater freedom to its members, who become more at liberty to vote for their preferred candidate.

We would, therefore, expect locations with monogroup mobilization but high levels of group fragmentation like locations with multigroup mobilization to provide good services as well. Due to high levels of political competition, locations with multigroup mobilization should be associated with good service provision regardless of whether fragmentation is high or low. On the contrary, locations with monogroup mobilization but with low levels of fragmentation should experience poor service provision. Figure 1 summarizes these expectations where “poor” or “good” refers to the quality of services.

Figure 1.

Tribal mobilization, fragmentation, and public goods provision.

We expect good service provision in areas of high mobilization and high fragmentation because many groups will contest elections and each group is likely to splinter into several electoral factions. In such a scenario, the incentive to seek votes from non-members is high, leading to better services. This same phenomenon is also likely in areas of high mobilization but low fragmentation because many groups will be competing against one another, even though they are electorally cohesive. We expect, therefore, good public goods provision in these locations too. Similarly, the desire to seek the votes of non-members and hence, good services will be present in areas with low mobilization and high fragmentation. While few or perhaps only one identity group will be mobilized, this group will subdivide into several subgroups that compete against one another during elections, providing the incentive once again to seek the votes of non-members, which in turn improves services. However, service provision should be poor in locations with low levels of fragmentation and mobilization because not only will there be few identity groups competing against one another but those groups that do compete will remain electorally cohesive, dampening the desire to seek cross-group voters and resulting in poor public goods provision.

These expectations lead to two situations where the interactive effect of identity group mobilization and fragmentation is predicted to improve public goods provision:

Hypothesis 1: When political fragmentation is low, increasing the number of politically mobilized groups exerts a positive impact on service provision. At high levels of fragmentation, increasing the number of mobilized groups does not impact the quality of services.

Hypothesis 2: When the number of mobilized groups is low, increasing levels of political fragmentation also improves service provision. But when the number of mobilized groups is high, increasing the fragmentation of individual groups has no effect.

Tribes in Jordan

To evaluate the hypotheses, I use data collected from Jordanian municipalities. Political competition among Jordanians occurs between tribes, which are “group[s] of people distinguished from other groups by notions of shared descent, whether real or imagined” (Alon, 2009, p. 8). Members of these tribes can be ethnic Jordanians (35% of the population), ethnic Palestinians (60%), or Jordanians of Armenian, Chechen, or Circassian descent (5%; Minority Rights Group International, n.d.). In the political arena, however, competition is between tribal groups rather than ethnic groups. In other words, ethnic Palestinian tribes and ethnic Jordanian tribes do not ally together as two ethnic blocs. Palestinian tribes also compete against other Palestinian tribes.

Jordan is a good case of the “accommodationist” model of nation-building described in Prerna Singh and Matthias vom Hau’s (2016) introduction to this special issue. When the British relinquished power to the Hashemite monarchy in 1946, King Abdullah needed to gain the allegiance of Jordanian tribes (Alon, 2009; Robins, 2004). Like other examples described by Prerna Singh and Matthias vom Hau, the Jordanian monarchy’s provision of public goods was used to encourage political loyalty and to ensure the survival of the newly established royal family. Abdullah and later heirs gradually centralized and assumed most of the services once administered by tribes, but because they continue to honor tribal identity and permit some level of local autonomy, public goods provision varies at the subnational level. Today Jordan is an example of a weak accommodationist state. Almost all services (i.e., health, education, security) are provided by the central government, but some autonomy is still available to tribes, who might prefer to negotiate their own settlements during inter-tribal conflicts or offer charitable services through their own civic organizations.

The case of Jordan also exemplifies the central claim of this special issue: that both contemporary diversity and contemporary public goods provision are shaped by historical nation-building strategies. Because of the accommodationist strategy adopted by early Hashemite monarchs, individual tribal identities continue to be salient in Jordan. Likewise, the decision to centralize most services while permitting some degree of tribal autonomy was the result of historical policy to create a sense of national identity while preserving strong tribal allegiances. Furthermore, like other articles in this special issue, this article embodies the revisionist agenda outlined by Prerna Singh and Matthias vom Hau’s (2016) introduction by challenging the supposed consensus around the negative relationship between diversity and public goods provision.

Jordan is also an excellent setting for this research because politics at the subnational level is almost exclusively tribal. As there are few parties that offer candidates in local elections, this permits us to focus exclusively on the role of group loyalties, without having to deal with confounding factors such as party membership.³ This is also why a subnational focus is appropriate here—Parties do play a greater role in Jordanian national politics. Furthermore, a subnational focus permits us to concentrate on the level of governance where political competition for leadership, the proposed mechanism for understanding public goods provision in this article, actually matters. Many public goods at the national level are under the authority of the king and his ministers, none of whom are elected. In addition, tribal members are more closely monitored and sanctioned in local settings, where the proximity of relations is closer and the population of voters smaller.

The use of the single nontransferable vote system (SNTV) combined with multi-member electoral districts in Jordan also introduces variation in fragmentation among tribes. At the municipal level, mayors are elected via the majoritarian system, but council members are elected via SNTV where the top n vote getters fill the seats in the council. Sixteen percent of Jordanian municipalities are comprised of one electoral district, while 84% are divided into multiple districts, with varying numbers of council seats per district. The use of SNTV in Jordan, where 90% of the municipalities are comprised of at least one multi-member electoral district, therefore, introduces variation in tribal fragmentation, a phenomenon that would not be observed in single-member districts where it makes sense for tribes to nominate only one candidate. Large tribes like large parties are prone to over-nomination and tribal fragmentation under SNTV. While a small party rarely over-nominates, “a large party with two incumbents in a district in which it had a chance at three might find it difficult to convince its incumbents to take the risk of diluting the party vote too far” (Cox, 1996, pp. 741-742). SNTV also encourages candidates to cultivate a personal vote to distinguish themselves from competitors even within their same party (Carey & Shugart, 1995). Finally, it should be noted Jordan also has a women’s quota where 20% of all council seats are allocated to females. These seats are awarded to the n women who win the greatest percentage of votes in their electoral district but do not win competitively, where n is the number of quota seats.

Finally, unlike other studies where measures of mobilization are based on approximate measures, the unique data used in this article permit exact identification of which groups have mobilized and the degree to which electoral fragmentation has occurred. As described below, candidate lists from the 2007 municipal elections were utilized to identify mobilized groups and the number of candidates they offered to gauge their level of fragmentation.

Data and Methods

To evaluate the relationship between tribal mobilization, fragmentation, and service provision, I use data collected across 93 municipalities in Jordan. At the time of data collection, there were 93 municipalities, but now the number has expanded to 100. I use an original dataset that combines measures of public goods provision with a new operationalization of diversity: the number of tribes offering candidates in the 2007 municipal elections. To create this dataset, more than 2300 candidates were categorized by tribe through discussion with locals and consultation of reference books.

In Jordan, the municipality is responsible for a total of 26 tasks, many of which affect the daily lives of its residents. These responsibilities include issuing business licenses, approval of planning applications, street paving and maintenance, street lighting, street cleaning, garbage collection, health and sanitary inspections, monitoring zoning violations, landscaping of public areas, and public building projects. The central government also works collaboratively with municipalities on a number of responsibilities such as disaster relief, fire prevention and safety, and the creation and maintenance of sewage systems. As noted earlier, the electoral system used for municipal council elections is SNTV.

Because I argue that the positive impact of tribal mobilization on public goods provision is dependent upon tribal fragmentation (Hypothesis 1) and that the positive impact of tribal fragmentation on public goods provision is also dependent upon the level of tribal mobilization (Hypothesis 2), I use an interactive model to evaluate my expectations where the main independent variables are tribal mobilization, fragmentation, and their interaction: Tribal Mobilization × Fragmentation. I outline below the measurement of these variables as well as measures of the dependent and control variables.

Dependent Variable

The dependent variable, service provision, is measured by two outcomes: (a) the log of the percentage of municipal revenues that are self-collected (revenues) and (b) a count of the quantity of heavy equipment owned by the municipality such as bulldozers, steamrollers, and air compressors used to carry out construction or maintenance tasks (equipment). Data for revenues and equipment are for the 2009-2010 fiscal year and were collected from the Cities and Villages Development Bank and the Ministry of Municipal Affairs in Jordan, respectively. Descriptive statistics for this and other variables are available in Table 1. Both tasks are solely under the jurisdiction of municipalities and are neither the responsibility of the central government nor other local authorities. Moreover, each measure captures an important aspect of service provision.

Table 1.

Summary Statistics.

Variable	n	M	SD	Minimum	Maximum
Revenues	93	−1.38	0.64	−3.70	−0.21
Equipment	93	23.88	36.62	5	275
Tribal mobilization	93	8.17	8.47	1	51
Fragmentation	93	1.74	0.89	1	8.5
Population	93	31,471	63,315	1,427	414,760
Area	93	10,494	10,912	768	68,076
Poverty rate	93	18.45	14.54	1.7	73.6
Government grant	93	13.29	0.71	11.66	15.49
District magnitude	93	8.97	3.20	6	29
Electoral district	93	3.80	2.94	1	23
Palestinian population	93	0.22	0.41	0	1
Category 1	93	0.12	0.32	0	1
Category 2	93	0.52	0.50	0	1
Category 3	93	0.27	0.45	0	1

Note. This table contains the summary statistics of the independent, dependent, and control variables used in regression estimations in this article. Data sources: Jordan’s Ministry of Municipal Affairs, the Local Governance Development Program, and assorted tribal dictionaries.

Municipalities receive a grant from the central government, but they are also responsible for collecting their own taxes. However, many fail to collect all of the fees and taxes owed. Across 93 municipalities, the lowest percentage of revenues that are self-collected by the municipality is 3% while the highest is 81%. The average municipality collects 30% of its own revenue, and over three-quarters of municipality were in debt in 2009. As Clark (2012) notes, favoritism to family members and friends and the interference of members of parliament, who lobby on behalf of some residents, impedes this collection process.

Could it be that residents are taxed heavily in municipalities with multitribal mobilization, which is why more revenues are collected? Under this scenario, the municipality would have an improved financial situation, but this would not be better for residents. In municipalities with multitribal mobilization, it is not that residents find it difficult to evade taxes but that they are less likely to have personal links with municipal officials or staff. In locations with monotribal mobilization, many residents are likely to be members of the mobilized tribe—If several tribes had large memberships within the municipality, then multiple tribes rather than a single one would have politically mobilized. Therefore, many residents in monotribal communities have links with someone within the municipality; and while municipal officials may feel social obligations to waive or postpone the fees of members of their tribe, they are better able to avoid this temptation when there is no familial tie.

Self-collected revenues are indicative of good public goods provision for several reasons. First of all, because of recent strict control by the Ministry of Municipal Affairs over hiring decisions, it is unlikely for revenues to be spent on non-service items such as employing additional superfluous staff. Second, a high percentage of revenues that are self-collected indicates low levels of patronage within the municipality and potentially greater spending on services. If patronage levels were high, self-collected revenues are likely to be low because residents are able to avoid payment. Finally, a number of other studies examining the relationship between diversity and public goods provision have also used the generation of tax revenues as an indicator of service provision. Baldwin and Huber (2010) examine tax revenues as a percentage of gross domestic product (GDP); Alesina et al. (1999), the amount of local government taxes collected per capita; Miguel and Gugerty (2005), the total amount of locally collected school funding per pupil; and La Porta, Lopez-de-Silanes, Shleifer, and Vishny (1999), a measure of local tax compliance. Therefore, demonstrating that multitribal mobilization improves or at least does not dampen tax collection would be a valuable theoretical contribution.

The availability of heavy equipment is also critical for a multitude of essential municipal functions such as fixing streetlights, maintaining and paving roads, and collecting trash. Because of the high costs of new equipment, some municipalities lack sufficient machines and have to rent equipment or contract out services.⁴ One donor program working at the local level found that all municipalities required assistance with purchasing equipment. Renting equipment is not ideal as municipalities are unable to respond as quickly to the query (i.e., fixing a streetlight) and are at the whim of the timetable of other municipalities. Awarding contracts for services can also adversely affect the quality of services—Contractors for trash collection perform poorly compared with municipal trash collectors because they do not know the roads as well and are less adept at dealing with local merchants.⁵

Independent Variables

The independent variable, tribal mobilization (tribal mobilization), is measured by counting the number of tribes that offered candidates for the municipal council in the 2007 municipal elections. Unlike many studies, which account for absolute diversity, this measure takes into account only those groups that actually participate in politics. As Daniel Posner (2004) points out, “[t]o capture the contribution that a country’s ethnic heterogeneity makes to its policymaking process requires an index of fractionalization that reflects the groups that are actually doing the competing over policy” (2004, p. 853, emphasis original). If we posit that ethnic diversity exerts influence through its impact on government policies, then we must identify “the roster of actual participants in the competition over those policies” (p. 853).

Although similar to Posner’s measure, my indicator of tribal mobilization is better situated for this study for several reasons. Because Posner uses a number of resources across time to identify relevant groups, it becomes difficult to pinpoint which groups are relevant at any point in time. Second, the authors of those resources used by Posner did not always agree which groups were political players. Third, Posner’s criteria for politically relevant groups may be different from those of his sources leading to the inclusion of some inappropriate groups and the exclusion of some appropriate ones. However, my indicator of tribal mobilization uses the same criteria (i.e., participation in the 2007 municipal elections) and avoids the ambiguity present in Posner’s study.

To determine which tribe each candidate belonged to, I identified her family name and then located the corresponding entry in dictionaries of Jordanian and Palestinian tribes. A number of dictionaries are available, but because scholars can disagree about tribal structures and whether a tribe is truly a tribe or should be considered the branch of another tribe, I consistently used certain sources, the list of which is available upon request. For candidates for whom there was no relevant entry in the dictionaries, I asked locals knowledgeable about tribes (i.e., council members who ran alongside them in elections or residents who live in the municipality), called the municipality office to speak to a relevant employee, or asked the candidates themselves.

It should be noted that tribes are constantly evolving with new tribes being formed as well as established tribes merging together. Although this means that family surnames are not stable across time, anecdotal evidence suggests that the process of change is a slow one. For instance, the reference books used to classify tribes span 40 years, but the majority of tribes mentioned in these texts are the same.

Tribal Fragmentation

Ideally for a measure of tribal fragmentation, I would want to compare how many candidates were endorsed by each tribe with how many candidates actually ran from each tribe. As noted earlier, the electoral system used in Jordan is SNTV, and the vast majority of Jordanian municipalities have at least one multi-member electoral district. For tribes with low levels of fragmentation, the number of candidates running should be equivalent to the number of candidates endorsed by the tribe. But for tribes with high levels of fragmentation, we would expect the number of candidates to exceed the number who were endorsed.

Unfortunately, I do not know of the 2,300 candidates in the 2007 municipal elections, how many were actually endorsed across Jordan’s 93 municipalities. Therefore, I make a logical guess as to how many candidates the tribe “ought” to have run by examining the number of seats the tribe could have won. If a tribe could have won two seats, then I infer that tribe ought to have run only two candidates. If that tribe did indeed offer two candidates, then its tribal fragmentation is low, but if it offered four candidates, its level of tribal fragmentation is higher.

To ascertain the number of seats the tribe could have won, I examine the vote share garnered by candidates from that tribe as compared with the total number of votes in the electoral district. Using this vote share, I calculate how many seats on the council that tribe could logically have expected to win. The measure of tribal fragmentation (fragmentation) is, therefore, the ratio of the number of candidates the tribe actually ran to the number of candidates that it should have run. High values of tribal fragmentation mean that the tribe is electorally fragmented (the number of candidates representing the tribe and the number of seats that the tribe could have won differ dramatically), while low values indicate that the tribe is cohesive. Using the example above, the tribe that should have offered two candidates and did indeed do so has a fragmentation value of 1 while the tribe that should have offered two candidates but offered four has a fragmentation value of 2.

But how can we expect each tribe to know how many candidates it ought to run? Through repeated elections since 1955, tribes have gained a fairly good understanding of their own electoral position (National Democratic Institute, 1995). Small tribes know that they need to either coalesce together or to ally with a larger tribe to be successful. Larger tribes know that they have good chances of winning and almost always enter the race. Many tribes are aware of how many votes its members offer and the percentage of these votes their candidates are likely to garner. It is more difficult for the tribe to predict how many votes they will gain from members of other tribes. However, the candidate’s involvement in civic affairs and general status in the municipality give some indication. In the municipality of Hassa, for instance, even though the Hajaya tribe dominates the population demographically, the tribe coordinated both the number of candidates for mayoral and council positions.⁶ In the municipality of Madaba, however, the Shawabkeh tribe endorsed only one tribal candidate, but several others entered the race nonetheless.⁷

Control Variables

For each regression estimation, the following variables serve as controls.

Demographic controls

Municipalities with large populations (population) or large areas (area) should be associated with possessing a greater quantity of equipment as more garbage compressors, bulldozers, and so on, would be needed to serve such locations. A high population or a large area should also be associated with a greater percentage of revenues that are self-collected because in these municipalities there is less likelihood of residents knowing an employee within the municipality and therefore, using personal connections to receive a discount or postponement of their payment. Population is measured by the number of residents living in the municipality in 2004; area, in the unit of dunums ( $1 dunum = 1, 000 m^{2}$ ) in 2005.

Socioeconomic controls

A higher rate of poverty (poverty rate) among residents should be associated with a decrease in the quantity of equipment and the percentage of self-collected revenues. Poorer municipalities should be less able to afford equipment and residents less willing to pay fees and taxes. An indicator of financial assistance from the central government (government grant) is included as municipalities with more assistance from the Center can also spend more on public goods. The specific measure used is the log of the amount transferred by the central government to each municipality in 2008.

District controls

District magnitude (district magnitude) is likely to be positively associated with the tribal mobilization as more tribes are likely to offer candidates when there is a greater number of council seats available. The number of electoral districts within the municipality (electoral district) may also positively influence the number of politically active tribes. The political salience of a tribe will also be dependent on whether its members are divided across several districts or live primarily in one district.

Significant presence of Jordanians of Palestinian descent

Any tribe that offered a candidate in the 2007 municipal elections is included in the tribal mobilization and fragmentation measures, which means that all tribes of Palestinian descent that offered candidates are also included. Although ethnic Palestinians are included in both measures, their presence is likely to also exert an independent effect as municipalities with a significant presence of ethnic Palestinians may be associated with a higher percentage of revenues. Because of their high rate of participation in the Jordanian private sector, it is said that on average they are wealthier than ethnic Jordanians. To determine whether a municipality possesses a “significant” population of ethnic Palestinians, candidate lists were reviewed to identify candidates of Palestinian descent. Municipalities where candidates of Palestinian origin received at least 10% of the vote were initially coded as having a “significant” population (Palestinian population). Because members are likely to vote for their own tribe, the number of votes received by the candidate can be used as a rough measure of the size of her tribe. I use a fairly low threshold (10%) to denote the presence of ethnic Palestinians because Palestinians are less politically active and the percentage of votes received by Palestinian candidates is likely to under represent their actual numbers. This initial coding was then reviewed by locals knowledgeable on this subject and revised as necessary.

Category of the municipality

The Ministry of Municipal Affairs categorizes municipalities according to their population and political importance (i.e., whether they are the capital of a governorate, district, or subdistrict). Currently, there are four categories of municipality with Category 1 municipalities having the highest populations and/or being the capital of a governorate. Municipalities in Categories 1 and 2 are likely to have a higher percentage of revenues that are self-collected and quantity of equipment because they have greater populations that are also likely to be wealthier. The category of the municipality is included as a series of binary variables (Category 1, Category 2, Category 3, Category 4) with the Category 4 variable left out of the regression as the base category.

Regression Estimation

Seemingly unrelated regression (SUR) was used to carry out analysis. SUR allows for the simultaneous estimation of a series of regressions (two regressions here, one for each of the service provision outcomes) rather than estimating each regression separately and permits standard errors in the different equations to be correlated (Tsai, 2007). Because municipalities only have limited resources, the widespread collection of fees and taxes may not be accompanied also by a large quantity of equipment. A municipality that places a great deal of resources in training staff to collect fees may not be able to spend additional resources to acquire equipment and vice versa. The advantage to SUR is that it permits joint hypothesis tests using information across both regressions (Miguel, 2004).

There are 93 municipalities in Jordan, but coefficient estimates are based on data for 90 of the observations. The three most populous municipalities have extraordinarily high levels of tribal mobilization and heavily influence results if included. Table 2 lists the mean values for tribal mobilization (independent variable) and the two measures of service provision (dependent variables) for the first 90 observations and also for the three outliers. As can be seen, the mean values for the three outliers differ quite remarkably from the remaining 90 municipalities. However, deleting these three outliers means throwing away information and leads to biased estimates of the error variance and all of its derivatives. To keep these outliers in the regression but to prevent them from exerting influence on coefficient estimates, I create three binary variables (Outlier 1, Outlier 2, Outlier 3) for each of these three observations.

Table 2.

Means of Independent and Dependent Variables for Outlier and Non-Outlier Observations.

	n = 90	n = 3	t value	p value
Tribal mobilization	7	43	42.3	.05%
% of revenues self-collected	29	56	8.9	1.2%
No. of municipal equipment	18	189	4.8	4.0%

Note. This table displays the mean levels of tribal mobilization, percentage of revenues that are self-collected, and the quantity of equipment for the three observations that are outliers as well as for the 90 observations that are not. The t values and p values are calculated for the difference in means between the 90 observations and the three outliers.

Regression Results

The first step is to examine Hypothesis 1 and whether tribal mobilization has a positive impact on public goods provision at low levels of tribal fragmentation. To do so, I estimate two regressions, one for equipment and the other for revenues that include tribal mobilization, fragmentation, their interaction, and all of the control variables (Model 1). The results are displayed in Table 3. From this table, we can see that tribal mobilization, fragmentation, and the interaction term are statistically significant. However, despite a significant relationship, we cannot directly use the substantive value or the signs of the coefficients on tribal mobilization, fragmentation, and their interaction term to understand their collective impact on public goods provision. To understand the relationship between these three factors, we must use Figures 2 and 3 to examine the marginal impact of tribal mobilization on public goods provision.

Table 3.

Effect of Tribal Mobilization and Fragmentation on Public Goods Provision.

Variable	Model 1		Model 2		Model 3
Variable	Revenues	Equipment	Revenues	Equipment	Revenues	Equipment
Tribal mobilization	0.086**	0.93**	0.09**	0.94**	0.09***	0.99***
Tribal mobilization	(0.04)	(0.39)	(0.04)	(0.39)	(0.04)	(0.037)
Fragmentation	0.30*	3.14**	0.31**	3.17**	0.30**	3.17**
Fragmentation	(0.16)	(1.58)	(0.15)	(1.57)	(0.14)	(1.56)
Mobilization × Fragmentation	−0.04*	−0.39*	−0.04**	−0.38*	−0.05**	−0.41*
Mobilization × Fragmentation	(0.02)	(0.22)	(0.021)	(0.22)	(0.02)	(0.21)
Ministry assistance			0.22***	0.75
Ministry assistance			(0.06)	(0.68)
Bedouin					−0.62***	−4.84***
Bedouin					(0.16)	(1.67)
Density					−0.06*	−0.41
Density					(0.04)	(0.40)
Population	0.00001*	0.0002***	0.00001	0.0002***	0.00002**	0.0003***
Population	(0.000007)	(0.00007)	(0.000006)	(0.00007)	(0.000007)	(0.00007)
Area	0.00001	0.00007	0.000009	0.00006	0.000003	0.00002
Area	(0.00001)	(0.0001)	(0.00001)	(0.00001)	(0.00001)	(0.0001)
Poverty rate	0.005	−0.002	0.005	−0.005	0.02***	0.08*
Poverty rate	(0.004)	(0.04)	(0.004)	(0.04)	(0.004)	(0.05)
Government grant	−0.48**	−0.42	−0.46**	−0.36	−0.68***	−2.02
Government grant	(0.21)	(21.5)	(0.21)	(2.13)	(0.21)	(2.16)
District magnitude	0.07*	0.42***	0.09**	1.61***	0.07*	1.53***
District magnitude	(0.04)	(0.42)	(0.04)	(0.42)	(0.04)	(0.40)
Electoral districts	−0.11***	1.33***	−0.12***	1.32***	−0.10**	1.49***
Electoral districts	(0.04)	(0.41)	(0.04)	(0.41)	(0.038)	(0.40)
Palestinian population	0.23	−0.90	0.091	−1.35	0.21	−0.96
Palestinian population	(0.14)	(1.50)	(0.15)	(1.59)	(0.14)	(1.49)
Category 1	0.49	8.81**	−0.18	5.96	0.46	8.25**
Category 1	(0.39)	(4.10)	(0.43)	(4.66)	(0.39)	(4.07)
Category 2	0.24	3.78	0.16	2.71	0.12	2.28
Category 2	(0.23)	(2.42)	(0.24)	(2.52)	(0.23)	(2.42)
Category 3	−0.02	1.47	−0.007	1.13	−0.08	0.66
Category 3	(0.20)	(2.12)	(0.21)	(2.19)	(0.20)	(2.11)
Outlier 1	−3.18*	86.48***	−3.56*	79.10***	−5.33***	72.41***
Outlier 1	(1.80)	(18.74)	(2.03)	(21.47)	(2.02)	(21.16)
Outlier 2	−1.32	−12.81	2.44	−26.04**	−3.12***	−28.17**
Outlier 2	(1.01)	(10.54)	(1.18)	(12.47)	(1.16)	(12.16)
Outlier 3	−2.44	91.32***	−3.78*	76.22***	−5.33**	69.13***
Outlier 3	(1.81)	(18.86)	(2.08)	(22.08)	(2.09)	(21.87)
N, municipalities	93	93	93	93	93	93

Note. This table displays the regression results for Models 1, 2, and 3.

p < .10. **p < .05. ***p < .01.

Figure 2.

The marginal effect of tribal mobilization on the percentage of revenues that are self-collected as tribal fragmentation changes.

Figure 3.

The marginal effect of tribal mobilization on the quantity of equipment as tribal fragmentation changes.

The solid lines in the middle of these two figures display the marginal impact of tribal mobilization on self-collected revenues and equipment at various levels of fragmentation (y axis). These solid lines showcase how the impact of tribal mobilization on equipment and revenues changes as tribal fragmentation changes. The impact of tribal mobilization on revenues and equipment is only statistically significant when the two dotted lines, which are the upper and lower bounds of the 95% confidence interval, are both either above 0 or below 0. When the upper and lower bounds are above 0 then tribal mobilization has a positive impact on public goods provision. Where the bounds are below 0, a negative impact is evident.

From Figures 2 and 3, we can see that where tribal fragmentation is low, tribal mobilization is associated with a greater quantity of heavy equipment and greater percentage of revenues that are self-collected. The effect of tribal mobilization is significant, however, only at low levels of fragmentation: when it is 1.1 or less for self-collected revenues and 1.5 or less for the quantity of equipment. In other words, tribal mobilization only has a positive impact on public goods provision when tribes nominate on average 1 or 1.5 candidates for every candidate they ought to nominate. At high levels of fragmentation, when the tribe nominates several candidates beyond what they should, tribal mobilization no longer has a statistically significant impact on services. Furthermore, the substantive impact of tribal mobilization on revenues and equipment decreases as fragmentation increases. These results are in accordance with the first hypothesis, where I had predicted that tribal mobilization would have a positive impact on public goods provision but only under circumstances where electoral competition is low (i.e., when fragmentation levels are low).

The second step is to examine Hypothesis 2 and whether tribal fragmentation has a positive influence on public goods provision at low levels of tribal mobilization. Although the regression used to examine Hypothesis 2 is the same as that used for Hypothesis 1 (Model 1), we need to examine this time, the marginal impact of fragmentation on public goods provision, which is displayed in Figures 4 and 5. Once again, the solid lines here display this marginal impact—this time of fragmentation on public goods provision at varying levels of tribal mobilization—while the dotted lines are the upper and lower bounds of the 90% confidence interval. These figures show that fragmentation has a positive influence on both the quantity of heavy equipment and the percentage of revenues that are self-collected when levels of tribal mobilization are low: 4 or less for equipment and 3 or less than for self-collected revenues. This means that when the number of tribes mobilized is less than 3 or 4, increasing fragmentation results in greater quantity of equipment and the collection of revenues. This is in accordance with the predictions of the second hypothesis where I had anticipated fragmentation to positively influence public goods provision only when other sources of electoral competition (i.e., tribal mobilization) were low.

Figure 4.

The marginal effect of tribal fragmentation on the percentage of revenues that are self-collected as tribal mobilization changes.

Figure 5.

The marginal effect of tribal fragmentation on the quantity of equipment as tribal mobilization changes.

Figures 4 and 5 also show that at high levels of tribal mobilization “when the number of tribes mobilized is 14 or greater “ increasing fragmentation does not impact the quantity of heavy equipment, which is as predicted, but has a negative effect on the percentage of revenues that are self-collected, which is not in accordance with Hypothesis 2. This is a puzzling result as according to this hypothesis, at high levels of mobilization, increasing fragmentation should have no impact on public goods provision. What explains then these contradictory results?

One explanation for this finding is that municipalities with high levels of tribal mobilization and high levels of fragmentation are likely to be large sprawling urban metropolises where the city center has high levels of tribal mobilization but surrounding metropolitan areas are homogeneous with low levels of mobilization. Because council members in the metropolitan areas are likely to have won due to votes mainly from their own tribes, patronage levels are also likely to be high and collecting fees from residents difficult. Residents in metropolitan areas may also resent paying fees because they feel that bulk of the municipal budget is reserved for services in the city center than distributed to the areas where they live. After municipalities were conglomerated in 2001 residents of smaller municipalities that became attached to larger ones complained that their areas were neglected.⁸ But why does this geographical arrangement affect revenues but not the quantity of equipment? This may be because while dissatisfied residents can withhold financial contributions, they cannot control the number of equipment that the municipality acquires as this is a decision made by the municipal leadership and not by themselves.

Robustness Check

In a semidemocratic country like Jordan, the central government and the ability of those in power to play favorites among the municipalities—rather than tribal mobilization and fragmentation—may explain most of the variation in service provision. Perhaps some municipalities do better only because the Center has been more generous in providing assistance to those locations and not because municipalities with multitribal mobilization give candidates incentives to seek cross-ethnic votes. While I have already controlled for revenues from the central government in all regressions using the government grant variable, it is also possible that the Center has assisted the municipality in providing one of its services (i.e., paving or maintaining a road) so that additional funds remain for buying equipment or training employees to collect fees. In this scenario, the government grant variable by itself cannot capture the additional assistance provided by the Center.

I therefore re-estimate the regression and include a ministry assistance variable in Model 2. This variable was created by documenting every incidence of assistance from the Ministry of Municipal Affairs between 2002 and 2007 such as constructing a public marketplace for which the municipality can charge rental fees, paving roads within the municipality, or additional training for municipal employees. It also includes the participation of the municipality in an international donor program to improve local governance and participation. Table 3 displays regression results when this variable is included. Even with the inclusion of ministry assistance, tribal mobilization, tribal fragmentation, and the interaction variable do not change much in terms of the substantive value of coefficients or statistical significance.

Could it also be that modernity or historical legacy, rather than tribal fragmentation, is the real motivator of better public goods provision? For instance, it is said that Bedouin tribes are less fragmented than non-Bedouin tribes as the former used to travel out to isolated parts of the desert and therefore, group solidarity was particularly important for security, the sharing of food and resources, and for providing mutual company for one another (Lewis, 1987). Because Bedouin tribes also became sedentary much later than non-Bedouin tribes, this means that they have had less time to develop the skills needed for public goods provision (Jureidini & McLaurin, 1984). As Andreas Wimmer (2016) reminds us in this special issue, historical legacies of public goods provision (and lack thereof) continue to have contemporary impact.

Urban areas may also experience greater tribal fragmentation because residents there may be more highly educated than in rural areas and education may have led to more widespread rejection of tribal solidarity in favor of independent judgment⁹ (Department of Statistics [Jordan] & ORC Macro, 2003). The distribution of tribal members throughout various neighborhoods in urban areas may also hamper mobilization of members during elections and sanctioning them for noncompliant behavior.

To test this possibility, in Model 3 I include two additional variables in the regression estimation as indicators of modernity: the population density level (density) and whether the municipality contains a majority of residents of Bedouin descent (Bedouin). Density is measured by the following formula: $P_{i} / A_{i}$ , where P is the population of municipality i and A is its area. Municipalities where at least 60% of the votes were won by Bedouin candidates were considered to have residential populations that were mostly Bedouin in origin. Candidates were designated as Bedouin if they were from a tribe listed as Bedouin according to Jordanian electoral law.

The inclusion of the Bedouin and density variables does not change findings. While Bedouin does have a negative impact on both types of service provision, its presence in the estimation does not alter the relationship between tribal mobilization, fragmentation, and municipal service provision. In further robustness checks available upon request, I also use different cutoff points for designation as to whether the municipality had a majority of residents of Bedouin origin: (a) where at least 50% of the vote was won by Bedouin candidates, (b) where at least 70% of the vote was won by these candidates, and (c) the percentage of votes won by Bedouin candidates. As in previous estimations, the coefficients for tribal mobilization and fragmentation remain statistically significant.

In addition to the robustness checks discussed here, I also performed two additional checks available upon request: (a) using tobit instead of SUR to estimate the regressions (b) and using self-collected revenues as the dependent variable instead of the log of self-collected revenues. Tobit is recommended when the dependent variable is either bounded at a lower or upper limit. Here, the quantity of equipment is unlikely to exceed a certain value. The results do not change even after performing these tests.

Conclusion

This article makes several contributions to the study of diversity and public goods provision. First of all, it suggests that a negative relationship between heterogeneity and services is not as straightforward as some scholars suggest. Contrary to the conclusion of some who have proclaimed that there is now academic consensus regarding the negative relationship between diversity and public goods provision, this study demonstrates that we need to reconsider this proposition and in particular how we measure diversity. We should distinguish between group structure and practice and not use the former to make conclusions regarding the latter. Second, this study indicates that identity group mobilization can reduce group voting and motivate good governance in semidemocratic or authoritarian contexts where regime leaders have purposely weakened political parties. When other democratic mechanisms are absent, identity group diversity and intra-group fragmentation can provide “natural” motivators of good governance. The results of this article also suggest that elections even in electoral authoritarian or hybrid regimes can have a positive influence and are not meaningless activities (Levitsky & Way, 2002; Schedler, 2006).

Third, this article emphasizes that the relationship between diversity and public goods provision depends on an additional factor—the degree of fragmentation within the groups themselves. Scholars have previously argued that diversity impairs public goods provision because members of different identity groups possess diverging preferences, that group members can be punished by their leaders for failing to cooperate with other groups, and that members prefer to work with each other because of shared culture and preferences (Alesina et al., 1999; Easterly & Levine, 1997). However, if the group itself is politically or socially fragmented, then even homogeneous populations encounter these obstacles. We should not assume that members of the same identity group share similar preferences, are capable of punishing each other for renegade behavior, or desire even to collaborate with one another. Whereas previous studies have used these mechanisms to explain why heterogeneous settings underproduce public goods, this study highlights that members within groups can be as varied as those between groups.

But perhaps what this study demonstrates is simply that locations where only one group is politically mobilized are not really homogeneous and that varying levels of tribal aggregation can be salient? In some respects, this is true: Certainly, across different types of Jordanian elections, the number of candidates that tribes offer varies according to what unit is politically salient. In parliamentary elections where the electoral district might comprise of several municipalities, large tribes tend to offer only one candidate per tribe, but in municipal competitions, the number of challengers per tribe greatly increases. But at the heart of the matter is not that tribes are capable of behaving strategically but whether they do or do not. What this study showcases is that even in areas where only one tribe is mobilized, some tribes coordinate competition between their candidates while others do not. Even within municipalities where only one tribe is mobilized, tribal fragmentation varies.

This study also emphasizes a new mechanism—electoral competition—and how the desire to win votes outside of one’s own group discourages clientelism and encourages inclusiveness. When the median voter is no longer an individual within one’s own group but one outside it, candidates are forced to accommodate the needs of non-members to win elections. This finding regarding competition also has implications for the design of electoral systems and the debate regarding whether proportional representation or majoritarian systems are superior for divided societies (Horowitz, 1991; Lijphart, 1977). Like Reilly (2002) and Selway and Templeman (2012), the results of this study suggest that majoritarian systems are favorable because they encourage candidates to reach out to non-group members and to craft policies that are moderate and inclusive.

This study has focused on the subnational level, and we should also consider whether findings here are generalizable to the national level. On the one hand, because national electoral districts are larger than subnational ones, we would expect greater cross-group voting, which suggests that good service provision will result. However, this is dependent upon whether the population within national electoral districts is segregated or not. If it is segregated, then members of parliament may represent the wishes of constituents within their specific identity groups and have few incentives to provide public goods rather than private goods. Furthermore, Gerring et al. (2015) notes that coordination and sanctioning dynamics may be more difficult at the national level where the number of groups is greater. At the local level, he notes that informal mechanisms for political participation are likely to be in place but that similar mechanisms may be absent at the national level. If this is true, then the elected official may be an individual that residents, especially those outside of the group of the official, have never met and whom they cannot sanction for poor behavior.

I hope that the results from this study will encourage scholars to seek a better understanding of the ability of heterogeneity to encourage cross-group voting and the significance of internal fragmentation and their collective impact on government services. Although many previous studies have highlighted the negative relationship between diversity and public goods provision, this work and others like it may serve as a reminder that there is not yet consensus regarding the direction and nature of this relationship.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biography

Eleanor Gao is a lecturer (assistant professor) in Middle East Politics at the University of Exeter (UK). Her dissertation, which won the 2013 American Political Science Association’s Best Dissertation in Urban Politics Award, investigated how tribal diversity and tribal cohesion impact municipal services, patronage, and electoral competition in Jordan.

References

Addison

Rahman

(2001). Why is so little spent on educating the poor? (Discussion Paper No. 2001/29). Helsinki, Finland: United Nations University, WIDER.

Alesina

Baqir

Easterly

(1999). Public goods and ethnic divisions. Quarterly Journal of Economics, 114, 1243-1283.

Alon

(2009). The making of Jordan: Tribes, colonialism, and the modern state. London, England: I.B. Tauris.

Auyero

(2000). The logic of clientelism in Argentina: An ethnographic account. American Political Science Review, 104, 644-662.

Baldwin

Huber

(2010). Economic versus cultural differences: Forms of ethnic diversity and public goods provision. Latin American Research Review, 35(3), 55-81.

Banerjee

Iyer

Somanathan

(2005). History, social divisions, and public goods in rural India. Journal of the European Economic Association, 32, 639-647.

Banerjee

Somanathan

(2007). The political economy of public goods: Some evidence from India. Journal of Development Economics, 82, 287-314.

Bardhan

(2000). Irrigation and cooperation: An empirical analysis of 48 irrigation communities in south India. Economic Development and Cultural Change, 48, 847-865.

Beal

D. J.

Cohen

R. R.

Burke

M. J.

McLendon

C. L.

(2003). Cohesion and performance in groups: A meta-analytic clarification of construct relations. Journal of Applied Psychology, 88, 989-1004.

10.

Besley

Coate

Loury

(1993). The economics of rotating savings and credit associations. The American Economic Review, 83, 792-810.

11.

Birney

(2007). The impact of local elections on the prospects for political reform in authoritarian countries: Theoretical perspectives from China. Unpublished manuscript, Yale University, New Haven, CT.

12.

Birnir

J. K.

(2007). Ethnicity and electoral politics. Cambridge, UK: Cambridge University Press.

13.

Blaydes

(2011). Elections and distributive politics in Mubarak’s Egypt. New York, NY: Cambridge University Press.

14.

Boustan

L. P.

Ferreira

Winkler

Zolt

(2010). Income inequality and local government in the United States, 1970-2000 (NBER working paper series). Cambridge, MA: National Bureau of Economic Research.

15.

Brusco

Nazareno

Stokes

(2004). Vote-buying in Argentina. Latin American Research Review, 39(2), 66-88.

16.

Carey

J. M.

Shugart

M. S.

(1995). Incentives to cultivate a personal vote: A rank ordering of electoral formulas. Electoral Studies, 14, 417-439.

17.

Carron

A. V.

Brawley

L. R.

Widmeyer

W. N.

(1998). Measurement of cohesion in sport and exercise. In Duda

J. L.

(Ed.), Advances in sport and exercise psychology measurement (pp. 213-226). Morgantown, WV: Fitness Information Technology.

18.

Cederman

Girardin

(2007). Beyond fractionalization: Mapping ethnicity onto nationalist insurgencies. American Political Science Review, 101, 173-185.

19.

Chandra

(2004). Why ethnic parties succeed: Patronage and ethnic headcounts in India. New York, NY: Cambridge University Press.

20.

Chandra

(2007). Counting heads. In Kitschelt

Wilkinson

(Eds.), Patronage and clientelism (pp. 84-109). New York, NY: Cambridge University Press.

21.

Chandra

Wilkinson

(2008). Measuring the effect of “ethnicity.” Comparative Political Studies, 41, 515-563.

22.

Clark

(2012). Municipalities go to market: Economic reform and political contestation in Jordan. Mediterranean Politics, 17, 358-375.

23.

Cox

G. W.

(1996). Is the single nontransferable vote super proportional? Evidence from Japan and Taiwan. American Journal of Political Science, 40, 740-755.

24.

Department of Statistics (Jordan) & ORC Macro. (2003). Jordan population and family health survey 2002: Key findings. Calverton, MD: ORC Macro.

25.

Dunning

Harrison

(2010). Cross-cutting cleavages and ethnic voting: An experimental study of cousinage in Mali. American Political Science Review, 104, 21-39.

26.

Easterly

Levine

(1997). Africa’s growth tragedy: Policies and ethnic divisions. Quarterly Journal of Economics, 112, 1203-1250.

27.

Egel

(2012). Tribal heterogeneity and the allocation of publicly provided goods: Evidence from Yemen. Journal of Development Economics, 101, 228-232.

28.

Gerring

Thacker

S. C.

Huang

(2015). Does diversity impair human development? A multilevel test of the diversity debit hypothesis. World Development, 66, 166-188.

29.

Gibson

C. C.

Hoffman

B. D.

(2013). Coalitions not conflicts: Ethnicity, political institutions, and expenditure in Africa. Comparative Politics, 45, 273-290.

30.

Gisselquist

R. M.

(2014). Ethnic divisions and public goods provision, revisited. Ethnic and Racial Studies, 37, 273-290.

31.

Glennerster

Miguel

Rothenberg

A. D.

(2013). Collective action in diverse Sierra Leone communities. Economic Journal, 123, 285-316.

32.

Gubler

J. R.

Selway

J. S.

(2012). Horizontal inequality, crosscutting cleavages, and civil war. Journal of Conflict Resolution, 56, 206-232.

33.

Habyarimana

Humphreys

Posner

D. N.

Weinstein

J. M.

(2007). Why does ethnic diversity undermine public goods provision? American Political Science Review, 101, 709-725.

34.

Hecock

D. R.

(2006). Electoral competition, globalization, and subnational education spending in Mexico, 1999-2004. American Journal of Political Science, 50, 950-961.

35.

Hiskey

J. T.

(2003). Demand-based development and local electoral environments in Mexico. Comparative Politics, 36, 41-59.

36.

Hopkins

D. J.

(2011). The limited local impacts of ethnic and racial diversity. American Politics Research, 39, 344-379.

37.

Horowitz

D. L.

(1991). A democratic South Africa? Constitutional engineering in a divided society. Berkeley: University of California Press.

38.

Jones

M. P.

(1997). Racial heterogeneity and the effective number of candidates in majority runoff elections: Evidence from Louisiana. Electoral Studies, 16, 349-358.

39.

Jordan Center for Social Research. (2007). The 2007 municipal elections in Jordan: An exit poll conducted in six electoral districts located in six municipalities. Amman: Author.

40.

Jureidini

P. A.

McLaurin

R. D.

(1984). Jordan: The impact of social change and the role of tribes. New York, NY: Praeger.

41.

Kimenyi

(2006). Ethnicity, governance and the provision of public goods. Journal of African Economies, 15(Suppl. 1), 62-99.

42.

Kotler-Berkowitz

L. A.

(2001). Ethnicity and politics: Cohesion, division and British Jews. Political Studies, 49, 648-669.

43.

Langston

E. M.

(2005). The Islamist movement and tribal networks: Islamist party mobilization amongst the tribes of Jordan and Yemen (Doctoral dissertation). University of Kentucky, Lexington.

44.

La Porta

Lopez-de-Silanes

Shleifer

Vishny

(1999). The quality of government. Journal of Law, Economics, & Organization, 15, 222-279.

45.

Layne

(1994). Home and homeland: The dialogics of tribal and national identities in Jordan. Princeton, NJ: Princeton University Press.

46.

Levitsky

Way

L. A.

(2002). The rise of competitive authoritarianism. Journal of Democracy, 13, 51-65.

47.

Lewis

N. N.

(1987). Nomads and settlers in Syria and Jordan, 1800-1980. Cambridge, UK: Cambridge University Press.

48.

Lieberman

Singh

(2012). Conceptualizing and measuring ethnic politics: An institutional complement to demographic, behavioral, and cognitive approaches. Studies in Comparative International Development, 47, 255-286.

49.

Lijphart

(1977). Democracy in plural societies: A comparative exploration. New Haven, CT: Yale University Press.

50.

Lupu

Pontusson

(2011). The structure of inequality and the politics of redistribution. American Political Science Review, 105, 316-336.

51.

Manion

(1996). The electoral connection in the Chinese countryside. American Political Science Review, 90, 736-748.

52.

Matuszeski

Schneider

(2006). Patterns of ethnic group segregation and civil conflict. Unpublished manuscript, Harvard University, Cambridge, MA.

53.

Miguel

(2004). Tribe or nation? Nation building and public goods in Kenya versus Tanzania. World Politics, 56, 327-362.

54.

Miguel

Gugerty

M. K.

(2005). Ethnic diversity, social sanctions, and public goods in Kenya. Journal of Public Economics, 89, 2325-2368.

55.

Minority Rights Group International. (n.d.). World directory of minorities: Jordan. Retrieved from http://minorityrights.org/country/jordan/

56.

Montalvo

J. G.

Reynal-Querol

(2005). Ethnic diversity and economic development. Journal of Development Economics, 76, 293-323.

57.

National Democratic Institute. (1995). Democracy and local government elections in Jordan: 1995 municipal elections. Retrieved from http://www.ndi.org/node/13892

58.

Neto

O. A.

Cox

(1997). Electoral institutions, cleavage structures, and the number of parties. American Journal of Political Science, 41, 149-174.

59.

Ordeshook

P. C.

Shvetsova

O. V.

(1994). Ethnic heterogeneity, district magnitude, and the number of parties. American Journal of Political Science, 38, 100-123.

60.

Østby

(2008). Polarization, horizontal inequalities, and violent civil conflict. Journal of Peace Research, 45, 143-162.

61.

Polat

Ö.

(2012). How do ethnic fragmentation and ethnic civil society organizations relate to public good provision? Evidence from Jordan. Unpublished manuscript, Princeton University, NJ.

62.

Posner

D. N.

(2004). Measuring ethnic fractionalization in Africa. American Journal of Political Science, 48, 849-863.

63.

Posner

D. N.

(2005). Institutions and ethnic politics in Africa. New York, NY: Cambridge University Press.

64.

Powell

B. G.

(1982). Contemporary democracies: Participation, stability, and violence. Cambridge, MA: Harvard University Press.

65.

Putnam

R. D.

(2007). E Pluribus Unum: Diversity and community in the twenty-first century: The2006 Johan Skytte Prize Lecture. Scandinavian Political Studies, 30, 137-174.

66.

Reilly

(2002). Electoral systems for divided societies. Journal of Democracy, 13, 156-170.

67.

Robins

(2004). A history of Jordan. New York, NY: Cambridge University Press.

68.

Rugh

J. S.

Trounstine

(2013). The provision of local public goods in diverse communities: Analyzing municipal bond elections. Journal of Politics, 73, 1038-1050.

69.

Scarritt

Mozaffar

(1999). The specification of ethnic cleavages and ethnopolitical groups for the analysis of democratic competition in contemporary Africa. Nationalism and Ethnic Politics, 5, 82-117.

70.

Schedler

(Ed.). (2006). Electoral authoritarianism: The dynamics of unfree competition. London, England: Lynne Rienner.

71.

Selway

J. S.

(2011a). Cross-cuttingness, cleavage structures and civil war onset. British Journal of Political Science, 41, 111-131.

72.

Selway

J. S.

(2011b). The measurement of cross-cutting cleavages and other multidimensional cleavage structures. Political Analysis, 19, 48-65.

73.

Selway

J. S.

(2015). Coalitions of the wellbeing: How electoral rules and ethnic politics shape health policy in developing countries. New York, NY: Cambridge University Press.

74.

Selway

J. S.

Templeman

(2012). The myth of consociationalism? Conflict reduction in divided societies. Comparative Political Studies, 45, 1542-1571.

75.

Shehata

(2008). Inside Egyptian parliamentary campaigns. In Lust-Okar

Zerhouni

(Eds.), Political participation in the Middle East (pp. 95-120). Boulder, CO: Lynne Rienner.

76.

Singh

(2010). We-ness and welfare: A longitudinal analysis of social development in Kerala, India. World Development, 39, 282-293.

77.

Singh

(2016). Ethnicity in time: Politics, history, and the relationship between ethnic diversity and public goods provision. Comparative Political Studies, 49(10), 1303-1340.

78.

Stewart

(2002). Horizontal inequalities: A neglected dimension of development (CRISE working paper). Oxford, UK: Queen Elizabeth House, University of Oxford.

79.

Stokes

(2005). Perverse accountability: A formal model of machine politics with evidence from Argentina. American Political Science Review, 99, 315-325.

80.

Stoll

(2013). Changing societies, changing party systems. New York, NY: Cambridge University Press.

81.

Taagepera

(1999). The number of parties as a function of heterogeneity and electoral system. Comparative Political Studies, 32, 531-554.

82.

Tetreault

M. A.

(2000). Stories of democracy: Politics and society in contemporary Kuwait. New York, NY: Columbia University Press.

83.

Tsai

L. L.

(2007). Solidary groups, informal accountability, and local public goods provision in rural China. American Political Science Review, 101, 355-372.

84.

Valeri

(2009). Oman: Politics and society in the Qaboos state. London, England: C. Hurst.

85.

Vigdor

(2004). Community composition and collective action: Analyzing initial mail response to the 2000 census. Review of Economics and Statistics, 86, 303-312.

86.

Waring

T. M.

(2011). Ethnic forces in collective action: Diversity, dominance, and irrigation in Tamil Nadu. Ecology & Society, 16(4), 1-19.

87.

Waring

T. M.

Bell

A. V.

(2013). Ethnic dominance damages cooperation more than ethnic diversity: Results from multi-ethnic field experiments in India. Evolution & Human Behavior, 34, 398-404.

88.

Wimmer

(2016). Is diversity detrimental? Ethnic fractionalization, public goods provision, and the historical legacies of stateness. Comparative Political Studies, 49(11), 1407-1445.