October 14, 2022
Do elections \(\to\) ethnic conflict?
Institutions Matter
Horowitz suggests ethnic parties contesting elections can drive conflict:
If ethnic parties likely to emerge (and reasons to expect this), then likely to generate conflict.
If this logic is true: should expect that proximity to elections, particularly competitive elections increases conflict.
As a result,
Eifert et al (2010); Gadjanova (2021)
look at respondents to Afro-Barometer (several rounds, random sample survey in many countries)
compare respondents in the same country when election is recent vs. distant in time
when elections are nearby, are people more/less likely to…
Eifert et al (2010); Gadjanova (2021)
look at respondents to Afro-Barometer (several rounds, random sample survey in many countries)
compare respondents in the same country when election is recent vs. distant in time
when elections are nearby, are people more/less likely to…
Especially, when elections are more competitive.
Empirical evidence is mixed:
Horowitz (1991) identifies how the specific institutions shape conflict
Says we want electoral institutions that:
Institutions:
They are the “rules of the game”:
Worst institutional design?
First past the post: single member districts, plurality winner (Canada, UK, US, etc.)
proportional representation: many variants, but, generally: multi-member constituencies, seats allocated in proportion to votes won
Pros:
Cons:
What is the problem with PR (according to Horowitz)?: difference between vote-pooling vs. seat-pooling:
vote pooling: political parties try to collect votes from multiple ethnic groups in advance of the election. This drives parties to propose moderate policies that are palatable to ethnic moderates in both groups.
seat pooling: political parties representing different ethnic groups form a ruling coalition after the election. May not be very stable.
Alternative Vote: single member districts, rank order candidates
Pros:
Why? Moderate ethnic parties can beat extremist ethnic parties with second preference votes from other ethnic group.
Federalism:
dispersing power from central government to regional/local governments can reduce the costs of losing
But do other kinds of institutions limit/encourage conflict?
Two ethnic groups divided by Zambia-Malawi border.
In Malawi:
In Zambia
Even when comparing villages just across the border from each other
Despite recognizing same cultural differences in both countries…
Posner (2004):
Malawi: Much smaller country…
Zambia: Much larger country…
In 1960s, Language/Region was politically salient ethnic cleavage
Between the 1970s and 1980s, tribal divisions were more important.
In the 1990s, language/region became the more salient divide.
Between 1972 and 1990, Zambia was a one-party state. Elections were held within the ruling party.
Before 1972 and after 1990, Zambia had multi-party competition.
Changing the focus of competition from national to local changes the relevant ethnic demography (Posner 2005)
With one-party rule, only demography of constituency matters:
With multi-party competition, national demography matters:
Indian Constitutions recognized and created special protections for previously “untouchable” caste groups
Despite these protections, by 1980s, growing number of educated and upwardly mobiles SCs faced discrimination, inadequate employment.
Enter the Bahujan Samaj Party:
Uttar Pradesh:
Karnataka:
Despite similarities…
Why does the same ethnic party fail to win in two places?
Chandra (2004)
Even if voters/elites want to distribute government resources to their group…
will support whatever party able to do this best, regardless of whether it is exclusively ethnic
parties that are best able to distribute to an ethnic group:
In Uttar Pradesh:
In Karnataka:
Key takeaway:
Party institutions in ethnic and non-ethnic parties, along with ethnic demography, determine whether ethnic parties can successfully compete.
Many formal/informal rules enable/constrain ethnic conflict:
Change the strategic benefits/costs to pursuing ethnic extremism or compromise.
But institutions are not easily be changed.