The Impact of Candidates’ Political Connections on Voters' Expressed Support in Pakistan

Researchers:
Location:
Sargodha District, Punjab Province, Pakistan
Sample:
2,969 registered voters
Timeline:
2015 - 2015
Partners:

Previous research on what affects vote choice has focused on candidates’ qualifications, policy positions, and performance in office. However, voters might also care about candidates’ connections with influential, higher-level politicians—particularly in settings where other information is hard to come by. Researchers evaluated the impact of randomly varied information about candidates’ political connections on voter’s beliefs and self-expressed support in the context of a 2015 local election in Sargodha, a district in rural Pakistan. Providing information on candidates’ connectedness increased voter’s expressed support for more well-connected politicians. In contrast, information on past public spending increased voter satisfaction with the government’s performance, but did not affect expressed support for incumbent candidates.

Policy issue

Voters in low- and middle-income contexts may lack information about the qualifications and policy positions of political candidates, which can result in the election of low-quality candidates and politicians who do not address citizens’ needs. Past literature has shown that providing information on candidates’ qualifications, policy positions, and performance in office can affect voter turnout and for whom people vote1.  

In some contexts, where such information is hard to come by, voters might also care about candidates’ levels of political connectedness—a factor that has been understudied in rigorous evaluations. This is particularly the case in contexts where political and personal networks often determine a politician’s ability to provide services for their constituents. In such cases, voters might not be able to obtain information about a politician’s past performance, but they may be able to predict their future performance if they know politicians’ levels of connectedness. 

Does providing voters information about candidates’ connectedness affect voters’ self-expressed electoral support? 
 

Context of the evaluation

This study took place in the context of the 2015 local government elections in the rural Sargodha district in Punjab, Pakistan, the most populous province in the country. With a history of periodic military governments and subsequent civilian governments, this election brought in the first local government in the province after a gap of 10 years and saw a considerable turnout of 61 percent.  Approximately 50% of voters in focus groups in this study were undecided about whom to vote for weeks before the election. 

Local government in Punjab is organized around a two-tier system: a lower tier (called the union council) and the higher tier (called the district council). While each of these tiers is responsible for performing different local functions, a wide range of citizen services (such as road maintenance, sewage, irrigation, electricity, education, health, and social welfare programs) remain centralized at the provincial and federal levels. 

In this setting, local candidates who have strong connections with higher-level politicians inspire greater confidence among voters—a factor that may influence their choices in local elections. In Sargodha, participants in focus groups for this study demonstrated high knowledge of a local candidate’s connections, and nearly three-fourths of focus groups declared political connections as an extremely important reason to support a local candidate. In addition, a majority of focus group members expressed a deep concern for services that went beyond the jurisdiction of local union councils and nearly 50 percent revealed that they would be willing to support an opposition or independent candidate in the election if they were well-connected and willing to use their connections to deliver services locally. 

Candidates contesting the 2015 local elections made an explicit attempt to display their connectedness with the Punjab Muslim League Nawaz (PML-N), the incumbent party at the federal and provincial levels. Campaign materials and posters featured pictures of the Prime Minister and other high-level party members and in some cases, even highlighted their connections with higher-level politicians across parties. 

Female voter casts her vote at a polling station in Pakistan
Photo: A M Syed | Shutterstock.com

Details of the intervention

Researchers conducted a randomized evaluation to test the impact of providing randomly varied information about candidates’ political connections on voters’ expressed support for candidates in the 2015 local government election in Sargodha, Pakistan.

Researchers compared how voters responded to information about candidate connectedness against how voters responded to information about candidates’ past performance. To do so, researchers held one-on-one interviews with 2,969 registered voters across 84 union councils. They measured voters’ electoral preferences and beliefs before and after randomly exposing them to one of three types of information during the interview: 

  • Candidate connectedness: Voters in this group received information on three primary kinds of connections of candidates in the election: personal or familial, political, and their connections with local police and courts. 
  • Past performance: Voters in this group received information on an incumbent candidate’s past performance, measured by reported spending in that union council as compared to spending in other councils in Sargodha over the five years preceding the election (2011-2015).
  • Local election information (comparison group): A third group of voters, which served as a comparison to the two other interventions, received “neutral” information regarding the local government election in a particular union council (e.g. the number of villages in the union council etc.). 

Researchers collected original data on candidates’ connectedness by conducting interviews with key informants or people who were likely to have detailed firsthand information about the political affairs in the region (typically senior government officials and retired politicians). Researchers also calculated government spending at the union council level by aggregating the incumbent party’s expenditure on water filtration plans, water pipes, sanitation, drainage, and construction, compiled by the Water-Sanitation, Building, and Health Departments of the district of Sargodha.

Results and policy lessons

Providing information about candidates’ connectedness increased voters’ expressed support for better-connected politicians. By contrast, information on public spending increased voter satisfaction with the government’s performance, but did not affect expressed support for incumbent candidates. In a context where voters place a high value on local politicians’ connections, information on candidate connectedness was more likely to influence voters’ expressed support for a particular candidate than information on the incumbent government’s past performance. 

Voters’ knowledge about candidates’ connections: Despite already having a high degree of knowledge of candidate connections prior to the intervention, voters who received information about candidate connections updated their knowledge during the interview. At the start of the interview, voters could answer around 45 percent of surveyors’ questions about candidate connections on average. Voters who received additional information were 9 percentage points more likely to increase the number of questions they answered about candidate connections compared to voters in the comparison group at the end of the interview. Voters who received additional information were 16 percentage points more likely to answer these questions compared to the baseline; voters in the comparison group were only 7 percentage points more likely to do so. Furthermore, among those who answered questions about candidate connections, voters who received information were 9 percentage points more likely to have answered correctly. 

Voters’ satisfaction with government services: Between the beginning and end of the interview, voters who received information on public spending on local development programs by incumbents were 4.5 percentage points more likely than voters in the comparison group to increase their self-expressed satisfaction with local government services (voters were 8.7 percentage points likely to increase their satisfaction when receiving additional information, compared to 4.2 percentage points in the comparison group). These effects were driven by voters in high-performing union councils (that is, councils where the government had spent more money), who were 7.1 percentage points more likely to increase their self-expressed satisfaction with local government services. In contrast, in the case of the lowest-performing union councils, information on government spending reduced the likelihood of self-expressed satisfaction by 2 percent.

Voters’ expressed support: At the beginning of the interview, surveyors asked voters to indicate how much they supported a candidate using a 0 (strongly oppose) to 100 (strongly support) scale. Voters who received information about a local candidate’s connections to higher level politicians were 2 percentage points more likely than voters in the comparison group to increase their own support of a candidate along the 0-100 scale at the end of the interview. This change was driven in large part by information provided on more well-connected candidates; voters who received information about these candidates were 3 percentage points more likely to increase their expressed support. These effects are modest, which may be due to voters’ already-high levels of support for well-connected candidates at the start of the interview. 

In contrast, receiving information on public spending did not have an effect on voters’ expressed support for an incumbent candidate.

Electoral Outcomes: Researchers found suggestive evidence indicating that more connected candidates also performed better during the 2015 local government elections. Well-connected candidates earned approximately 560 more votes, had a 9.5 percent higher voter share, and were 26 percent more likely to win than their relatively less well-connected counterparts. On the other hand, there was no evidence indicating that government spending translated into better electoral outcomes.  

These results suggest that in the absence of concrete information on candidates’ qualifications and past performance, citizens look to other cues for estimating future local government performance–such as candidates’ connections to higher levels of power. This might have adverse consequences for democratic accountability, as in these contexts, candidates might not be motivated to meet voters’ needs if their performance is not considered by voters in future elections.

Liaqat, Asad, Michael Callen, Ali Cheema, Adnan Khan, Farooq Naseer, and Jacob N. Shapiro. Political Connections and Vote Choice: Evidence from Pakistan. Working Paper, Harvard University, 2019.

1.
Abdul Latif Jameel Poverty Action Lab (J-PAL). 2019. "The risks and rewards of voter information campaigns in low- and middle-income countries." J-PAL Policy Insights. Last modified March 2019. https://doi.org/10.31485/pi.2275.2019