Party system erosion: Evidence from Peru

Abstract

Weakly institutionalized party systems are a defining feature of third-wave democracies. Yet, in some countries like Peru, party weakness is not a static equilibrium but rather part of a dynamic process of “party system erosion” in which weak parties become weaker over time as independents come to dominate subnational posts. As I argue, party system erosion is driven by a particular configuration of institutional factors—weak party brands, ease of ballot access, and limited partisan control over resource distribution. These institutional features increase the likelihood that experienced candidates will run as independents. When these candidates are elected, they obtain more intergovernmental discretionary transfers, which are used to improve performance and maintain clientelistic networks. This resource advantage of independent officials further weakens party brands and reduces experienced candidates’ incentives to run with parties in future elections. I test this theory using a data set of 80,000 subnational officials and a regression discontinuity design.

Keywords

intergovernmental transfers Latin America Peru political parties regression discontinuity design

Weakly institutionalized party systems characterize many new democracies in Latin America (Levitsky et al., 2016; Mainwaring, 1999; Mainwaring and Scully, 1995), Eastern Europe (Mainwaring, 1999; Moser, 1999), and Southeast Asia (Hicken and Kuhonta, 2014). Often, scholars compare these weak party systems with strong ones. However, weak party systems may also be differentiated from one another. The Peruvian party system, for example, is particularly weak, even in comparison with other weakly institutionalized party systems in the region. More importantly, Peru’s parties have seemingly become weaker over time. This article explores the reasons underlying this erosion of the Peruvian party system.

As I argue, in the Peruvian case, there exist institutions and structures that reinforce the weakness of already weak political parties, a phenomenon that I call “party system erosion.” Such a dynamic process emerges when parties fail to recruit experienced candidates to run for subnational office under their party labels. The recruitment of such candidates is essential for enhancing parties’ reputations and developing a loyal, grassroots base of partisan support for national elections (Holland, 2016; Katz and Mair, 1995; Levitsky, 2003b; Samuels, 2006). Systematically failing to recruit these candidates, who often outperform their inexperienced counterparts, only further erodes public perceptions of already weak party labels, making it more difficult for parties to recruit experienced candidates in the future. Because subnational officials play an important role in signaling parties’ governing competence and in developing a party base, the failure to win these positions threatens a further deinstitutionalization of party systems at the national level.

Once in office, these experienced officials have an advantage in securing revenue for their constituents. Because parties do not control the distribution of intergovernmental transfers, experience becomes an important predictor of intergovernmental transfer receipt, as experienced officials are better positioned to navigate the government bureaucracy and secure resources. Officials can use these resources to either broadly provide public goods to constituents who most need them or more strategically target resources to core/swing voters. By enabling clientelistic and programmatic targeting, expanded resources allow independent, experienced officials to outperform political parties in office and at the polls, thus further reducing experienced candidates’ incentives to run with parties.

While any party system could experience such erosion, the most prominent one can be observed in contemporary Peru. Using a descriptive analysis of two original data sets of over 80,000 candidates for subnational office, I show that candidates with previous experience holding elective office tend to run as independents or in ephemeral coalitions with other independent candidates (i.e. regional movements), while inexperienced candidates run more often with Peru’s weak nationally based political parties. Then, using a natural experiment, I examine whether mayors elected from regional movements receive more discretionary transfers from the central government than mayors elected from political parties. Discretionary transfers in Peru are allocated on a nonpartisan basis, and funding decisions are based on merit and linkages with bureaucrats in the Ministry of Economy and Finance (MEF). I find that regional movement mayors are, in fact, better at securing these transfers than their political party counterparts.

The process of party system erosion has important implications for theories of political parties and democracy, as parties are generally expected to play an important role in fostering accountability among elected officials at lower tiers of government (Aldrich, 1995: 4; Mainwaring and Scully, 1995: 24–27; Palombara and Weiner, 2015). Through good performance, subnational officials improve perceptions of the party brand, a “collective good” from which they can benefit in future elections, and raise their profile within the party, providing opportunities for advancement through the ranks (Besley and Case, 1995: 774; Carey, 1998: 102; Hagopian et al., 2009: 366–367; Spear and Alesina, 1988; Strom, 1990: 578). Fernando Tuesta (2014) argues that Peruvian parties, though weak, promote better performance among elected officials than regional movements do, writing, “A regional organization presents a candidate, wins an election, governs poorly, finishes the term, and nothing happens. The organization can disappear or later change its name…Political parties, on the other hand, have the goal of persisting.” However, as I show, the relative performance of party and regional movement mayors is a function of not only their ability to induce accountability but also their capacity to recruit experienced candidates. In Peru, parties may be more successful at holding their affiliated officials accountable, but they often underperform regional movements because their officials are relatively inexperienced.

Theory

A key contribution of this project is to highlight the conditions under which weak parties not only remain weak but also become weaker over time. To do this, I highlight the role of party system erosion in which the subnational electoral performance of political parties declines across election cycles. In a first stage, parties and candidates must be matched to one another. Candidates are the first movers and decide whether they will attempt to run with a party or as an independent.¹ I assume that ceteris paribus parties prefer candidates who have held an elective office in the past. Parties seek to win elections, but weak parties have limited information about a given candidate’s probability of electoral success. In this sense, experienced candidates are appealing. They have demonstrated an ability to win elections in the past, which may capture both their “valence” appeal (Motz, 2012) and their campaigning skill (Carson et al., 2007). Thus, experience is a strong signal of a candidate’s ability to win an election. Furthermore, while experienced candidates may be penalized for being “insiders,” this cost is expected to be offset by their advantages in resources (e.g. networks and knowledge) that increase their probability of success once in office (Caselli and Morelli, 2004).

In deciding whether to run under a party label, I argue that experienced candidates consider both the present (i.e. winning the election) and the future (i.e. constraints and benefits once in office). A first evaluation concerns the extent to which a party controls a candidate’s access to the ballot. High legal requirements for registration of new political parties advantage the existing parties over newcomers. When party control over ballot access is relatively high, candidates will be more likely to join an existing political party rather than forming a new party or running as independents (Brancati, 2008; Stratmann, 2005). When control is relatively low, we should observe heterogeneity in the decision to affiliate with a national party depending on candidates’ prior experience. Because of their advantages in networks and/or reputation, experienced candidates should be able to overcome the relatively low barriers to entry. Less experienced candidates, on the other hand, may find these barriers more prohibitive and, therefore, still run with a party.

Second, the candidate considers the value of the party “brand,” which is a composite function of the party’s performance in office and consistency between message and action (Hagopian et al., 2009; Lupu, 2014; Mershon and Shvetsova, 2013; Snyder and Ting, 2002). The worse a party’s reputation, the less likely candidates will be to run with that party (Dargent and Muñoz, 2011: 49). While a strong brand may be an electoral asset, party brands can also be operationalized as a postelection cost to officials, who seek to avoid the organizational or ideological constraints imposed by affiliating with that party. Where there exists variation in the strength of national-level party brands (e.g. Brazil and India) or where major party brands are uniformly strong (e.g. the United States), candidates may be expected to continue running with political parties. However, if no party is able to offer a strong brand, candidates should be less likely to run with parties and more likely to run as independents.

Third, candidates consider the resources that political parties can offer (Ansolabehere and Snyder, 2006; Bardhan, 2002; Brollo and Nannicini, 2012; Dargent and Muñoz, 2011; Hale, 2005: 11–19), which are here conceptualized as financial resources but could also be generalized to include organizational resources. These resources could either be offered to candidates during the election or promised to them upon winning the election (i.e. targeting of intergovernmental resources). In situations where elections are concurrent, candidates for subnational office who run with national parties can also benefit from the investments of copartisan candidates for national office (Casas-Zamora and Zovatto, 2015: 2). When parties cannot offer significant campaign resources and do not control the distribution of intergovernmental transfers, the utility of running with a political party decreases significantly.

When parties are weak along all three of these dimensions, experienced candidates will consistently opt out of running with political parties, choosing instead the agency of running as independents. If parties are relatively strong on any of these dimensions, experienced candidates will continue to run with them as long as the marginal benefit to running with a party exceeds the marginal benefit they assign to agency. However, any weakness on the three aforementioned dimensions will lead at least some subset of experienced candidates to run as independents or in loosely structured coalitions with other independents. Yet parties rarely disappear completely as they continue to serve as an important vehicle for inexperienced candidates, who lack the resources and reputation to run as independents.

The nonpartisan distribution of intergovernmental transfers plays a particularly important role in encouraging party system erosion. When parties control the distribution of resources, they can induce candidates loyal to other parties to switch to their parties, perhaps overcoming a disadvantage in party branding (e.g. Novaes, 2017). When parties do not control these resources, experienced candidates, who have previously demonstrated an ability to win elections, receive little postelection benefit from running with a party. Further, because the distribution of resources is nonpartisan and more meritocratic, experienced officials should prove more adept at securing transfers from the central government, as they have experience navigating the bureaucracy and may leverage the networks they developed during previous terms in office.

With these additional resources, independent, experienced officials are better able to provide public goods to constituents. By comparison, the public views parties to be poor performers, which leads to a further deterioration of party brands and an even greater challenge of recruiting experienced candidates in future elections. Further, additional resources can enable officials to reinforce and expand their clientelistic networks and, thus, electoral apparatus through the selective distribution of goods, services, and jobs to loyal or swing voters.² The development and/or maintenance of this electoral machinery puts experienced, independent officials at an advantage when competing against the skeletal organizations of political parties.

Thus, already weak parties become weaker in a process of party system erosion. In the following sections, I test this theory, showing that weakness on the three aforementioned dimensions prevents parties from recruiting experienced candidates and that when party candidates are elected to subnational office, they underperform their nonpartisan counterparts. Parties are thus systematically weakened over time.

Distinguishing weak parties in Latin America

A number of countries in Latin America have weakly institutionalized parties, but only Peru is currently experiencing party system erosion. Other countries in the region have institutions that permit the persistence of weak parties but that simultaneously preempt party system erosion. In Brazil, for example, parties are notoriously weak with the notable exception of the Worker’s Party (PT) (Lupu, 2016). Nearly, 95% of survey respondents claim to have little or no trust in the country’s political parties (Latinobarómetro, 2016). Yet ballot access requirements are high, requiring candidates who are not registered with a traditional party to obtain signatures from 0.5% of all voters in the most recent election for Federal Deputies, which includes 0.1% of the voters in every state. Running as an independent is explicitly forbidden. Brazilian parties are also advantaged by their ability to dole out campaign resources and intergovernmental transfers to subnational copartisans (Brollo and Nannicini, 2012; Desposato and Scheiner, 2008; Novaes, 2017). The combination of these factors induces experienced candidates to run with parties instead of developing new parties or running as independents.³

As in Brazil, Colombian parties are also weak. Just over 15% of Colombian citizens express either some or a lot trust in political parties (Latinobarómetro, 2016).⁴ Parties are further hampered in their recruitment of candidates by a lack of control over intergovernmental discretionary resources, which have been mostly controlled by technocrats since the 1990s (Gonzalez et al., 2002: 13; O’Neill, 2006: 176). However, as in Brazil, a majority of subnational candidates and the vast majority of elected officials (>85%) belong to political parties. This continued partisanship may be attributable to concurrent subnational and national elections, which allow affiliated subnational candidates to benefit—indirectly and directly—from the national campaigns of political parties (Casas-Zamora and Zovatto, 2015: 2). Parties also exercise a moderate amount of control over access to the ballot. While independent candidates have been eligible to run for office in Colombia since 1994 (O’Neill, 2006: 188), accessing the ballot requires 20% of signatures from the voting-eligible population in the circumscription where the candidate wishes to run.

Peru provides a unique case in which parties are weak on all three dimensions—resource distribution, ballot access, and party reputation. Discretionary intergovernmental transfers are no longer allocated along party lines as described in Schady (2000) but rather by technocrats at the MEF, who allocate primarily based on the merit of proposals with no apparent partisan interference. Campaign finance regulations make it difficult for parties to accumulate significant resources that can be distributed to copartisan candidates. Furthermore, differences in term lengths for subnational and national offices make it so that elections are rarely concurrent and, thus, subnational officials cannot free ride off the investment of copartisan candidates for national office. Peruvian parties are also hampered by the low cost of registering a new political organization and gaining access to the ballot. Registration requires that candidates obtain signatures from only 5% of a municipality’s voting population in the previous national elections. Subnational officials can also join an existing regional movement or “coalition of independents” (Zavaleta, 2014). These relatively low barriers to entry eliminate a key advantage that national parties in Brazil and, to a lesser but significant extent, in Colombia have in recruiting experienced candidates. Finally, as in Brazil and Colombia, Peruvian parties are poorly perceived with just over 15% of respondents expressing some or a lot of trust in political parties (Latinobarómetro, 2016).

There also exists a cost to running with a party. Interviews conducted by the author suggest that although parties are weak, they impose a constraint on the decision-making autonomy of local officials. A high-ranking bureaucrat in the Peruvian state of Junin argued that candidates often run as quasi-independent candidates because doing so allows for greater agency. “Mayors get calls from [copartisan officials in] Lima or Huancayo (the capital of Junin), telling them to do this or that,” he said. “They have freedom if they don’t join a party. Author interview, April 2017.” Thus, in Peru, there exist few benefits and a clear cost to running with a party. Inexperienced candidates, on the other hand, lack the networks, the reputation, and possibly the financial resources of those with experience and may choose to run with traditional political parties because of easier ballot access.

Party system erosion in Peru

The weakness of Peru’s party system can be linked to a number of factors. Prior to the 1980s, the Peruvian party system was weakly institutionalized but coherent (Levitsky and Cameron, 2003). The 1980s brought about crises that the extant party system found itself incapable of adequately managing. Economic shocks and an increasingly violent insurgent movement resulted in parties attacking each other instead of cooperating against common threats (Dietz and Myers, 2007). These crises, along with a string of corruption scandals, led voters to abandon traditional parties in favor of newly emergent, populist, and personalistic “movements” (Morgan, 2011; Seawright, 2012). In the 1986 local elections, Peru’s four major parties combined to account for more than 90% of the total vote share. By 1993, these parties combined to account for only a third of the total vote share (Levitsky and Cameron, 2003).

President Alberto Fujimori (1990–2000) both embodied and propagated this new political landscape. Fujimori strategically placed himself as an enemy of the extant Peruvian party system and relied on populist tactics to gain support. Partly because of his perceived governing success and partly because of the centrist parties’ governing failures in the 1980s, the public increasingly accepted this new personalistic brand of politics (Levitsky and Cameron, 2003).

In place of national parties, candidates for subnational office began to form “regional movements,” which have best been described as “coalitions of independents” and are essentially very loose confederations of political free agents (Zavaleta, 2014). These new political organizations have limited ideological or programmatic coherence and frequently do not endure past one election. The chief advisor to a mayor in the state of Arequipa could not even recall the name of the regional movement to which the mayor belonged, recalling only that “it is the one with the arbolito,” or little tree. In an interview conducted by the author in the state of Junín, a mayor had to think for several moments before recalling—partially incorrectly—the name of the movement under which he had been elected. Another mayor in Junín, who had run with a regional movement, forcefully asserted, “I’m an independent,” when asked about her party affiliation.

The weakness of Peruvian parties has only increased over time, resulting in party system erosion. Table 1 highlights the recent explosion of regional movements in Peru using an original data set of all candidates for local office (mayors and councillors) for every election since 2002. Between 2006 and 2014, there is a notable expansion in the number of candidates running with locally based movements. There is also a marked expansion in the number of winners who belong to regional movements, an increase that is not solely explained by more regional movement candidates. In 2014, despite fielding only 57% of candidates, regional movement candidates won nearly 68% of local offices. In fact, regional movements have outperformed political party candidates in every election since 2006. Such a decline has coincided with a general deterioration in public confidence in political parties. The number of Peruvians expressing “no trust” in political parties rose from 30% in 2001 to 50% in 2013 (Latinobarómetro, 2001–2013). Such a decline in public faith in parties may reflect a weakening of party brands in the country.

Table 1.

Peruvian movement and party candidates for local office (2002–2014).^a

	2002	2006	2010	2014
Movement candidates	35,919	27,667	38,290	41,763
Percentage of total candidates	43.2	39.2	52.8	56.9
Party candidates	47,322	42,823	34,162	31,657
Percentage of total candidates	56.8	60.8	47.2	43.1
Movement winners	4893	4842	6461	6938
Percentage of total winners	48.1	47.7	64	67.7
Party winners	5277	5304	3631	3310
Percentage of total winners	51.9	52.3	32.3	32.3
Percentage of movement (t − 1) to party (t)	—	52.3	34.7	29.4
Percentage of party (t − 1) to movement (t)	—	36.3	52.1	50.3
Percentage of movement A (t − 1) to movement B (t)	—	96.4	88	89

^a District councillors and mayors.

Perhaps perceiving this advantage, a number of candidates have begun to switch from running with political parties to running with regional movements or as independents. In Table 1, candidates are matched across elections using their full names (first, second, and last) and the region in which they ran. Among candidates who ran in 2002 and 2006, switching was more common from a movement to a party than vice versa. By 2010, this trend had reversed, with far more candidates switching from national parties to regional movements.

A final observation from Table 1 concerns the volatility of belonging to a regional movement. Of candidates who ran in multiple elections, only about 10% ran with the same movement from election to election. Far more typical was for candidates to switch from one movement to another. Movements are thus characterized by a high degree of switching, suggesting that officials do not have strong ties to the movements with which they run.

Experience disparities between movements and parties

A central proposition of this article is that experienced candidates have increasingly run with regional movements. To test this claim, I collect original evidence on all candidates for these Peruvian municipal offices in 2014.⁵ I match the names from a database of candidates in 2014 to a database of winning candidates in the 2010, 2006, and 2002 municipal elections. Experience is thus defined as having been elected as either a councillor or a mayor in any of the preceding three elections.

As shown in Table 2, candidates with previous experience holding elective municipal-level posts are nearly twice as likely to run with a movement as they are to run with a political party. Approximately, two-thirds of candidates with previous experience run with regional movements as opposed to around 56% of candidates without such experience who run with regional movements. When subsetting to either only mayors or only councillors, the numbers remain virtually the same. The remaining one-third of experienced candidates who still run with a political party may do so because a certain party is strong in that municipality. Parties are generally weak in Peru, but some maintain strong bases of support, like Fuerza Popular and Alianza para el Progreso in Peru’s northern departments.

Table 2.

Municipal-level candidates’ experience in elective office by party type.^a

Office	Independent/Movement	Party
All officials
Previous experience: Yes	64.3% (3,117)	35.7% (1,728)
Previous experience: No	56.4% (38,640)	43.6% (29,926)
Mayors
Previous experience: Yes	64.9% (1,821)	35.1% (985)
Previous experience: No	57.1% (4,800)	42.9% (3,600)
Councillors
Previous experience: Yes	63.6% (1,296)	36.5% (743)
Previous experience: No	56.2% (33,840)	43.8% (26,326)

^a N listed in parentheses.

Thus, examining multiple levels of government, I have shown that experienced candidates have more often run with regional movements than with political parties. In the next section, I show that regional movement candidates, when elected, are more adept at obtaining intergovernmental transfers than their national party counterparts. Specifically, do winning candidates from regional movements, who tend to be more experienced, obtain more discretionary transfers for their districts than winning candidates from political parties?

Natural experimental evidence: Performance in office

In Peru, discretionary transfers are apolitical, administrated by technocrats at the MEF. Evidence from interviews with nonpartisan MEF officials, scholars of local politics in Peru, and local elected officials have revealed that obtaining transfers from the government requires a great deal of effort from mayors. Mayoral connections with MEF officials and a thorough planning of project proposals have a decisive effect on whether a municipality obtains transfers. Experience should help mayors navigate this process, allowing them to obtain more transfers for their municipalities. If experienced candidates are selecting into regional movements, as suggested above, we should observe that municipalities governed by movements receive more discretionary transfers than those governed by political parties. Given that officials would likely prefer to exert maximum effort right before an election, we would expect these dynamics to be particularly pronounced in election years.⁶ Interviews with mayors and mayoral candidates suggest that effort is highest in such years and fairly low in nonelection years. We would thus expect any election-year effect to attenuate in nonelection years.

Discretionary transfers in Peru

Transfers serve as the largest revenue source for municipal governments in Peru. In 2014, 84% of total income came in the form of grants and loans from other levels of government. Resources from the central government have been particularly notable, accounting for nearly half of municipal income (http://apps5.mineco.gob.pe/transparenciaingresos/mensual/). Bureaucrats decide transfer amounts during the annual budgetary process. In an initial phase, the national budget office (DNPP), a bureaucratic agency housed within the MEF, creates a budget plan. Nondiscretionary grants, such as the common fund (Foncomun) and mineral royalty (Canon) grants, are allocated based on a formula (Martinez-Vazquez and Sepulveda, 2011). Other central government funds have no clear distribution criteria.⁷ The decision of how to allocate these “ordinary resources” is a key point in which mayors can insert themselves into the MEF’s decision-making process. Through lobbying and strong project proposals, mayors can obtain access to these coveted resources, which account for more than a fifth of local governments’ revenue. Interviews with MEF officials suggest that Congress’s attempts to intervene in the process are almost always unsuccessful. Further, bureaucrats at the MEF have long time horizons, often holding the same position for decades. Thus, they are not beholden to a particular party, and given the enormous volatility of the Peruvian party system, the party that originally appointed them to their position may no longer hold seats in Congress (or even exist).

Design and estimation

I use the results of these largely bureaucratic decisions—the amount of discretionary transfers received—to test whether municipalities governed by regional movements in fact receive more ordinary resources (i.e. discretionary transfers) than those governed by political parties.⁸ Specifically, I examine the difference in 2014 and 2010 discretionary transfers received. The years 2010 and 2014 were election years and, thus, the former is pretreatment and the latter posttreatment. To obtain data on this outcome, I use data from line items in municipal government budgets (http://apps5.mineco.gob.pe/transferencias/gl/default.aspx).

I employ a close-race regression discontinuity design. In situations where elections are decided by a sufficiently close margin, winners and losers can be treated as if they were randomly assigned (Lee, 2008). This approach relies on the assumption that elections are fair and that there is a certain randomness to closely decided elections (Dunning, 2012: 77). I use near winners and near losers from the 2010 Peruvian municipal elections to construct my sample (https://www.web.onpe.gob.pe/modElecciones/elecciones/elecciones2010/resultados_erm/index.php). Mayors elected in this period served from 2011 to 2015. Treatment is assigned at the district level; treated districts are those where a regional movement won (and a party lost), and control are those where a regional movement lost (and a party won).

I primarily rely on the natural experiment interpretation of the regression discontinuity, employing a nonparametric difference-in-differences estimator, such that

{\hat{τ}}_{diff} = \frac{1}{m} \sum_{i \in R} [Y_{i,2014} (1) - Y_{i,2010} (1)] - \frac{1}{n - m} \sum_{i \in N} [Y_{i,2014} (0) - Y_{i,2010} (0)]

where ${\hat{τ}}_{diff}$ is the difference in 2014 and 2010 discretionary transfers, R is the random subset of treated units (regional movement-governed municipalities), S is the random subset of control units (party-governed municipalities), m is the number of treated units, and $n - m$ is the number of control units.

To evaluate the plausibility of the as-if random assumption in the window around the cut point, I conduct balance and placebo tests using permutation inference and K-S tests (Figure 1). The covariates I examine include formula-based transfers, discretionary transfers, demographic factors, and development indicators. If baseline characteristics are not significantly different between treatment and control groups within a sufficiently small window—here defined as the mean-squared error (MSE) two-sided bandwidth—the as-if random assumption is more plausible. The balance and placebo tests do not indicate significant differences between the treatment and control groups. In the Online supplemental appendix, I show there are also no obvious differences in the geographic distribution of treated and control units.

Figure 1.

Balance and placebo tests (MSE bandwidth). MSE: mean-squared error.

The natural experimental approach provides easily interpretable results. However, I also use probability limits to determine a local average treatment effect at the cut point (Imbens and Lemieux, 2008) and estimate a local linear regression, such that

Y_{i} = α + β_{1} w_{i} + β_{2} v_{i} + β_{3} (v_{i} \times w_{i})

where $Y_{i}$ is the difference in 2010 and 2014 transfers received by municipality i, $w_{i}$ is an indicator of movement victory in municipality i, and $v_{i}$ is the movement margin of victory in municipality i.

With this approach, identification does not come from the as-if random assumption, but rather from the weaker assumption of continuity of potential outcomes across the cut point (Sekhon and Titiunik 2017). While this assumption cannot be directly tested—due to the fact that potential outcomes remain unobserved—we can test for indications of sorting at the cut point. McCrary’s (2008) sorting test and plot of the density of observations around the cut point provide no definitive evidence of sorting (Online supplemental appendix). I also plot selected covariates against the running variable to show that there exists no clear discontinuous jump in those covariates at the cut point (Figure 2).

Figure 2.

Balance and placebo tests: MSE bandwidth. (a) Ordinary resources (2007–2010). (b) Foncomun (2011–2014). (c) Canon (2011–2014). (d) Population (2007). (e) Electricity (2007). (f) Indoor plumbing (2007). MSE: mean-squared error.

In the analysis below, I employ data-driven bandwidth selectors. Specifically, I include MSE (Calonico et al., 2014) and coverage error rate (CER, Calonico et al., 2017) optimal bandwidth calculations. The MSE and CER bandwidth calculations allow bandwidth size to vary on either side of the cut point.⁹

Results

Figure 3 and Table 3 offer support for a central hypothesis of this article. For all six specifications, the results are in the expected positive direction and significant at conventional benchmarks. Using the natural experimental interpretation, movement-governed municipalities saw their funding increase between 2010 and 2014 by about S./800 per capita, or approximately US$250, more than party-governed municipalities saw their funding increase. The limit interpretation yields an even greater estimate of around S./1150 or US$355. While the results reported in Table 3 examine results for discrete bandwidth choices, the results do not seem to be sensitive to bandwidth selection (Figure 4), which suggests that these election-year spikes in funding for movement-governed municipalities are not limited to highly competitive districts.

Figure 3.

Discretionary transfers (per capita, Peruvian soles), difference-in-differences estimation (2014 minus 2010, MSE bandwidth). (a) Linear. (b) Quadratic. (c) Cubic. MSE: mean-squared error.

Table 3.

Regression discontinuity estimates of the effect of electing a movement on transfers (per capita) received, Peru, difference in differences (2014–2010).

	Bandwidth (BW)
	MSE	CER	Full
Difference in means^a	740.861***(239.934)	844.004***(282.22)	510.11***(188.93)
Local linear^b	1193.39**(555.08)	1099.62*(648.77)	1133.66***(318.52)
N (left)	206	172	298
N (right)	206	167	322
N	412	339	620
Left BW	0.09	0.07	−0.45
Right BW	0.11	0.08	0.83

Note: Robust standard errors in parentheses.

MSE: mean-squared error; CER: coverage error rate.

^a Two-tailed p-values assuming sharp null of no effect across units.

^b Two-tailed p-values.

*p < 0.1; **p < 0.05; ***p < 0.01.

Figure 4.

Bandwidth sensitivity, difference-in-differences (2014 minus 2010, per capita soles).

The tests presented above focus solely on the dynamics present in municipal election years. These tests are also performed for nonelection years, as politician effort is expected to generally be weaker in nonelection years, and thus, we would expect an attenuation of the treatment effect. Table 4 examines discretionary transfer receipt for 2014 as well as nonelection years.¹⁰ Consistent with these expectations, the only year for which transfers are significantly different in movement- and party-governed municipalities is the election year, 2014.

Table 4.

Regression discontinuity results by year (MSE bandwidth).

	Transfers received (per capita)
	2014	2013	2012	2011
2010 Movement mayor (conventional)	1126.73***	−737.76	−362.12	318.53
Standard error (HC robust)	(424.94)	(559.63)	(686.74)	(418.82)
Movement mayor (bias-corrected)	1168.63***	−984.34*	−371.09	422.37
Standard error (HC robust)	(424.94)	(559.63)	(686.74)	(418.82)
Movement mayor (bias-corrected)	1168.63**	−984.34	−371.09	422.37
Standard error (CCT robust)	(476.49)	(658.43)	(840.86)	(522.75)
Conventional bandwidth (left)	0.07	0.07	0.07	0.07
Conventional bandwidth (right)	0.13	0.08	0.1	0.11
Bias-corrected bandwidth (left)	0.12	0.12	0.11	0.12
Bias-corrected bandwidth (right)	0.24	0.17	0.16	0.17
N (left)	169	168	166	166
N (right)	223	155	191	197
N	392	323	357	363

MSE: mean-squared error; HC: Heteroskedasticity consistent; CCT: Calonico, Cattaneo, and Titiunik.

*p < 0.1; **p < 0.05; ***p < 0.01.

The findings thus suggest that municipalities governed by regional movements receive more financial resources from the central government than do municipalities governed by national parties. Given the descriptive data above, combined with knowledge of the budget allocation process, these findings can best be explained by a selection effect of experienced candidates into regional movements. The regression discontinuity design used here does not grant that observed and unobserved candidate-level factors, in expectation, will be balanced across treatment and control groups or continuous across the cut point. In the thought experiment, candidates are not randomly assigned to a party or regional movement. Rather, municipalities are randomly assigned to be governed by a party or regional movement and, thus, balance is only expected on the observed and unobserved pretreatment attributes of municipalities.

Examining the experience level of candidates, we find that movement mayors have more experience than party mayors. Experience here is defined as a dummy variable, which takes on a value of “1” if mayors were previously elected mayor or councillor in 2002 or 2006 and a value of “0” otherwise. I subset to only those mayors who ran for elective office in 2002 or 2006, and within this group, I examine whether those who barely won with a movement in 2010 were more likely to have had prior experience holding elective office than those who barely won with a party in 2010. The analysis thus mirrors the prior regression discontinuity design, but in this iteration, mayors’ previous experience replaces discretionary transfers as the outcome variable. Table 5 suggests that movement mayors who barely won in 2010 were more likely to have had experience in elective office than party mayors who barely won.

Table 5.

Previous experience of movement versus party mayors.

	Previous experience (Mayor/councillor in 2002 or 2006)
	MSE bandwidth	CER bandwidth
Movement mayor (conventional)	0.331***	0.293**
Standard error (HC robust)	(0.126)	(0.144)
Movement mayor (bias-corrected)	0.357***	0.308**
Standard error (HC robust)	(0.126)	(0.144)
Movement mayor (bias-corrected)	0.357**	0.308**
Standard error (CCT robust)	(0.148)	(0.157)
Conventional bandwidth (left)	0.114	0.086
Conventional bandwidth (right)	0.059	0.045
Bias-corrected bandwidth (left)	0.173	0.173
Bias-corrected bandwidth (right)	0.110	0.110
N (left)	144	144
N (right)	144	144
N	288	288

MSE: mean-squared error; CER: coverage error rate.

*p < 0.1; **p < 0.05; ***p < 0.01.

Thus, experienced mayors tend to run with movements and ultimately obtain more transfers for their municipalities, but do these experienced mayors also get elected at higher rates? Descriptive evidence in Figure 5 suggests that they do. Among mayors elected in 2010, those with previous experience in elective office (defined in the same way as above) were more likely to be reelected in 2014 than those without such previous experience. The effects appear particularly pronounced for movement mayors. While not causally identified, these results suggest a gap in the electoral performance of experienced versus inexperienced incumbents.

Figure 5.

Reelection of 2010 mayors by experience (point estimates and 95% confidence intervals). (a) All (2010). (b) Movement mayors (2010). (c) Party mayors (2010).

Concluding remarks

The evidence presented above offers evidence of party system erosion in Peru. In an initial empirical section, I showed that party performance at the subnational level has declined sharply over four consecutive electoral cycles. I then proceeded to offer descriptive evidence that candidates with past experience holding elective office frequently opt out of running with political parties and instead run in loosely organized coalitions with other independent candidates. Finally, I showed that municipalities governed by regional movements receive more discretionary transfers than those governed by political parties. Regional movement mayors tend to have more previous experience in elective office, and they are also reelected at higher rates than their inexperienced, partisan counterparts.

Other explanations may account for the general decline in the electoral performance of political parties, but these theories do not offer a compelling explanation for the relatively unique rise of regional movements in Peru. If partisan identification is higher among older cohorts, then generational changes may result in a decline in the electoral performance of parties across time. However, this explanation fails to explain why the benefits of this decline accrue to Peru’s regional movements as opposed to new political parties. Parties may also be weakened by intense scrutiny from the national press, which could lessen their credibility and trustworthiness. However, if this explanation alone was sufficient, we might expect regional parties and independents to dominate local elections throughout Latin America, when, in reality, they do so in only Peru. While these dynamics are somewhat specific to the Peruvian context, they carry implications for other weak party systems.

In Peru, there exists a configuration of characteristics that leads to a particularly severe version of party system erosion—parties with weak labels, low barriers to entry for new political organizations, and limited partisan control over resource distribution. However, individual institutional reforms that affect each of these dimensions can reinforce or ameliorate existing weaknesses in already weak party systems. In Colombia, for example, the number of candidates running as independents expanded rapidly in the first decade of the 21st century (O’Neill, 2006: 188), facilitated by institutional reforms that reduced partisan control over discretionary transfers and that eased the process of registering new, local political organizations. In the Czech Republic, a 1994 institutional reform allowed independent candidates to form heterogeneous coalitions to run for office, which quickly became a popular choice for candidates in a country characterized by widespread “reactive anti-partisanship” (Linek, 2003 cited in Jüptner, 2008). By 2002, these “non-institutionalized independent associations” came to dominate rural politics (Jüptner, 2008: 36). And finally, in South Africa, a 2000 reform that reduced the number of signatures needed to register candidates for ward elections, along with declining public support for existing political parties, weakened the hold of political parties over subnational posts (Piombo and Nijzink, 2005: 77–78; Sethulego, 2017: 76). Policy reforms in each of these three countries have opened up a space for experienced candidates to run as independents; however, other tools still under the control of political parties—notably, electoral and/or postelection revenue disbursement—have prevented these countries from experiencing party system erosion. The framework presented above can thus help us to understand why these cases have followed a different path from Peru.

The framework can also help us understand cases that appear vulnerable to a process of party system erosion. In Argentina, for example, there exist a number of loosely organized, personalistic parties that are based at the provincial level (De Luca, 2012: 193–194; Eaton, 2004: 37–39; Lublin, 2014: 297–300). While different in name, these organizations often maintain clear and well-established linkages to national parties, even exercising a great deal of control over the nomination of national legislators (Jones et al., 2002; Lublin, 2014: 298–299). Yet the stability of this arrangement rests on a knife’s edge. The oft-predicted collapse of Argentina’s major parties would likely lead to a fragmented system of many provincial parties, perhaps resulting in party system erosion (Levitsky, 2003a).

During the process of party system erosion, parties become weaker, but they do not disappear completely as they remain important tools for ambitious yet inexperienced candidates. In order to exit this trap, the surviving parties can enact—or reinforce—certain policies, which include, among others, higher requirements for ballot access, partisan control over discretionary transfers, concurrent elections, and permissive campaign finance laws. Each of these reforms would allow parties to recruit experienced candidates at higher rates and allow for better performance in local elections. Fielding strong candidates for these positions is a key way that parties can create a strong brand and cultivate a base of voters for national-level officials (Holland, 2016; Katz and Mair, 1995: 21; Levitsky, 2003a: 178). Such ties with the grassroots are essential for the creation of strong national parties, and the absence of such ties encourages a further deinstitutionalization—or erosion—of party systems (Levitsky et al., 2016: 11–12; Mainwaring and Scully, 1995: 5; Samuels, 2006).

Supplemental Material

Supplemental Material, sj-pdf-1-ppq-10.1177_1354068818793915 - Party system erosion: Evidence from Peru

Supplemental Material, sj-pdf-1-ppq-10.1177_1354068818793915 for Party system erosion: Evidence from Peru by Christopher L Carter in Party Politics

Footnotes

Author’s note

Research material including data and code can be accessed through author’s website upon publication.

Acknowledgements

The author thanks the editors, two anonymous reviewers, and seminar participants at the University of California, Berkeley; Stanford University; the 2016 Midwest Political Science Association annual meeting; and the 2017 European Political Science Association annual meeting. The author also thanks Natália Bueno, Anna Callis, Pradeep Chhibber, Danny Choi, Anirvan Chowdhury, Ana de la O, Thad Dunning, German Feierherd, Nikhar Gaikwad, Felix Hartmann, Kristen Kao, Ellen Lust, Aila Matanock, Paula Muñoz, Lucas Novaes, Mathias Poertner, Alison Post, Kenneth Scheve, Andrés Schipani, Renard Sexton, Martín Tanaka, Guadalupe Tuñón, and Mauricio Zavaleta for incredibly useful feedback on earlier drafts. For the anecdote about the “arbolito,” the author thanks Diego Cerna.

Declaration of Conflicting Interests

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

Funding

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

ORCID iD

Christopher L Carter

Supplemental Material

Supplemental material for this article is available online.

Notes

References

Aldrich

(1995) Why Parties? The Origin and Transformation of Political Parties in America. Chicago: University of Chicago Press.

Ansolabehere

Snyder

(2006). Party control of state government and the distribution of public expenditures. The Scandinavian Journal of Economics 108(4): 547–569.

Bardhan

(2002) Decentralization of governance and development. The Journal of Economic Perspectives 16(4): 185–205.

Besley

Case

(1995) Does electoral accountability affect economic policy choices? Evidence from gubernatorial term limits. The Quarterly Journal of Economics 110(3): 769–798.

Brancati

(2008) Winning alone: the electoral fate of independent candidates worldwide. The Journal of Politics 70(3): 648–662.

Brollo

Nannicini

(2012) Tying your enemy’s hands in close races: the politics of federal transfers in Brazil. American Political Science Review 106(04): 742–761.

Calonico

Cattaneo

Farrell

(2017) Coverage Error Optimal Confidence Intervals for Regression Discontinuity Designs . Working paper .

Calonico

Cattaneo

Titiunik

(2014) Robust nonparametric confidence intervals for regression-discontinuity designs. Econometrica 82(6): 2295–2326.

Carey

(1998) Term Limits and Legislative Representation. Cambridge: Cambridge University Press.

10.

Carson

Engstrom

Roberts

(2007) Candidate quality, the personal vote, and the incumbency advantage in congress. American Political Science Review 101(2): 289–301.

11.

Casas-Zamora

Zovatto

(2015) The Cost of Democracy: Campaign Finance Regulation in Latin America. Washington: Brookings Institution.

12.

Caselli

Morelli

(2004) Bad politicians. Journal of Public Economics 88(3-4): 759–782.

13.

Dargent

Muñoz

(2011) Democracy against parties? Party system deinstitutionalization in Colombia. Journal of Politics in Latin America 3(2): 43–71.

14.

De Luca

(2012) Political recruitment in Argentina. In: Peter

Scott

(eds.) Pathways to Power: Political Recruitment and Candidate Selection in Latin America. University Park: Penn State Press, pp. 189–218.

15.

Desposato

Scheiner

(2008) Governmental centralization and party affiliation: Legislator strategies in Brazil and Japan. The American Political Science Review 102(4): 509–524.

16.

Dietz

Myers

(2007) From thaw to deluge: party system collapse in Venezuela and Peru. Latin American Politics and Society 49(2): 59–86.

17.

Dunning

(2012) Natural Experiments in the Social Sciences: A Design-Based Approach. Cambridge: Cambridge University Press.

18.

Eaton

(2004) Politics Beyond the Capital: The Design of Subnational Institutions in South America. Palo Alto: Stanford University Press.

19.

Gonzalez

Rosenblatt

Webb

(2002) Stabilizing Intergovernmental Transfers in Latin America: A Complement to National/Subnational Fiscal Rules? Washington: World Bank.

20.

Hagopian

Gervasoni

Moraes

(2009) From patronage to program: the emergence of party-oriented legislators in Brazil. Comparative Political Studies 42(3): 360–391.

21.

Hale

(2005) Why Not Parties in Russia? Democracy, Federalism, and the State. Cambridge: Cambridge University Press.

22.

Hicken

Kuhonta

(2014) Party System Institutionalization in Asia: Democracies, Autocracies, and the Shadows of the Past. Cambridge: Cambridge University Press.

23.

Holland

(2016) Insurgent successor parties: Scaling down to build a party after war. In: Steven

James

Brandon

Jorge

(eds.) Challenges of Party-Building in Latin America. Cambridge: Cambridge University Press, pp. 273–304.

24.

Imbens

Lemieux

(2008) Regression discontinuity designs: a guide to practice. Journal of Econometrics 142(2): 615–635.

25.

Jones

Saiegh

Spiller

, et al. (2002) Amateur legislators—professional politicians: the consequences of party-centered electoral rules in a federal system. American Journal of Political Science 46(3): 656–669.

26.

Jüptner

(2008) Local lists in the Czech Republic. In: Marion

Everhard

(eds) Farewell to the Party Model?: Independent Local Lists in East and West European Countries. Berlin: Springer Science & Business Media, pp. 20–39.

27.

Katz

Mair

(1995) Changing models of party organization and party democracy: the emergence of the cartel party. Party Politics 1(1): 5–28.

28.

Latinobarómetro (2001-2013, 2016) Available at: www.latinobarometro.org (accessed 25 July 2017).

29.

Lee

(2008) Randomized experiments from non-random selection in US House elections. Journal of Econometrics 142(2): 675–697.

30.

Levitsky

(2003a) Argentina: From crisis to consolidation (and back). In: Jorge

Michael

(eds) Constructing Democratic Governance in Latin America. Baltimore: JHU Press, pp. 244–269.

31.

Levitsky

(2003b) Transforming Labor-Based Parties in Latin America: Argentine Peronism in Comparative Perspective. Cambridge: Cambridge University Press.

32.

Levitsky

Cameron

(2003) Democracy without parties? Political parties and regime change in Fujimori’s Peru. Latin American Politics and Society 45(3): 1–33.

33.

Levitsky

Loxton

Van Dyck

(2016) Introduction: Challenges of party-building in Latin America. In: Steven

James

Brandon

, and Dominguez

Jorge

(eds) Challenges of Party-Building in Latin America. Cambridge: Cambridge University Press, pp. 1–51.

34.

Lienert

(2010) Role of the Legislature in Budget Processes. Washington: International Monetary Fund. Google-Books-ID: t2028HU7pacC.

35.

Linek

(2003) Antistranické postoje české veřejnosti. In: Vladimra

and Andrea

(eds) Kongres Českých Politologů: Praha – Suchdol Česká společnost pro politické vědy: Prague. pp. 455–469.

36.

Lublin

(2014) Minority Rules: Electoral Systems, Decentralization, and Ethnoregional Party Success. Oxford: Oxford University Press.

37.

Lupu

(2014) Brand dilution and the breakdown of political parties in Latin America. World Politics 66(4): 561–602.

38.

Lupu

(2016) Building party brands in Argentina and Brazil. In: Steven

James

Brandon

Jorge

(eds.) Challenges of Party-Building in Latin America. Cambridge: Cambridge University Press, pp. 1–51.

39.

Mainwaring

(1999) Rethinking Party Systems in the Third Wave of Democratization: The Case of Brazil. Palo Alto, CA: Stanford University Press.

40.

Mainwaring

Scully

(1995) Building Democratic Institutions: Party Systems in Latin America. Palo Alto: Stanford University Press.

41.

Martinez-Vazquez

Sepulveda

(2011) Intergovernmental Transfers in Latin America: A Policy Reform Perspective. Atlanta, GA: Georgia State University.

42.

McCrary

(2008) Manipulation of the running variable in the regression discontinuity design: a density test. Journal of Econometrics 142(2): 698–714.

43.

Mershon

Shvetsova

(2013) The microfoundations of party system stability in legislatures. The Journal of Politics 75(4): 865–878.

44.

Morgan

(2011) Bankrupt Representation and Party System Collapse. University Park: Penn State Press.

45.

Moser

(1999) Electoral systems and the number of parties in postcommunist states. World Politics 51(3): 359–384.

46.

Motz

(2012) Who Emerges from Smoke-Filled Rooms? Political Parties and Candidate Selection. Working paper, University Library of Munich, Germany.

47.

Novaes

(2017) Disloyal brokers and weak parties. American Journal of Political Science 62(1): 84–98.

48.

O’Neill

(2006) Decentralized politics and political outcomes in the Andes. In: Scott

(ed) The Crisis of Democratic Representation in the Andes. Palo Alto, CA: Stanford University Press, pp. 171–203.

49.

Palombara

Weiner

(2015) Political Parties and Political Development. Princeton: Princeton University Press.

50.

Piombo

Nijzink

(2005) How are South Africans being represented? In: Piombo

Nijzink

(eds) Electoral Politics in South Africa: Assessing the First Democratic Decade. Berlin: Springer, pp. 64–86.

51.

Samuels

(2006) Sources of mass partisanship in Brazil. Latin American Politics and Society 48(2): 1–27.

52.

Schady

(2000) The political economy of expenditures by the Peruvian social fund (FONCODES), 1991-95. The American Political Science Review 94(2): 289–304.

53.

Seawright

(2012) Party-System Collapse: The Roots of Crisis in Peru and Venezuela. Palo Alto, CA: Stanford University Press.

54.

Sekhon

Titiunik

(2017) On interpreting the regression discontinuity design as a local experiment. In: Matias

Escanciano

(eds.) Regression Discontinuity Designs: Theory and Applications, Advances in Econometrics. Vol. 38, Bingley: Emerald, p. 7.

55.

Sethulego

(2017) Civil Strife against Local Governance: Dynamics of Community Protests in Contemporary South Africa. Leverkusen: Barbara Budrich.

56.

Snyder

Ting

(2002) An informational rationale for political parties. American Journal of Political Science 46(1): 90–110.

57.

Spear

Alesina

(1988) An overlapping generations model of electoral competition. Journal of Public Economics 37(3): 359–379.

58.

Stratmann

(2005) Ballot access restrictions and candidate entry in elections. European Journal of Political Economy 21(1): 59–71.

59.

Strom

(1990) A behavioral theory of competitive political parties. American Journal of Political Science 34(2): 565–598.

60.

Tuesta

(2014) Asfixia económica y desventaja electoral. Available at: http://larepublica.pe/columnistas/politika/asfixia-economica-y-desventaja-electoral-27-03-2014 (accessed 10 December 2016).

61.

Zavaleta

(2014) Coaliciones de independientes: Las reglas no escritas de la política electoral. Lima: Instituto de Estudios Peruanos.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.24 MB