• Do elections \(\to\) ethnic conflict?

• Institutions Matter

  • Horowitz
  • Posner
  • Chandra

Horowitz suggests ethnic parties contesting elections can drive conflict:

1. Ethnic parties face strategic logic that pulls them toward extreme positions.
2. Like all parties, ethnic parties downplay internal divisions: avoid issues that might unify voters across ethnic groups, emphasize issues that divide ethnic groups
3. Elections become ethnic censuses: for ethnic parties, loss portrayed as losing group is FOREVER shut out of power. exclusion \(\to\) 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.

• ethnic party leaders should mobilize people along ethnicity (around election time)
• ethnic parties should deploy divisive rhetoric (esp. when elections are competitive)

As a result,

• people should intensify their identification with an ethnic group, have more antagonistic inter-group attitudes
• violence should be more likely (we’ll come back to this later)

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…

  • identify ethnically vs other identities?
  • perceive ethnic discrimination?
  • distrust other ethnic groups?

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…

  • identify ethnically vs other identities?: YES (Eifert et al 2010)
  • perceive ethnic discrimination? YES (Gadjanova 2021)
  • distrust other ethnic groups? YES (Gadjanova 2021)

Especially, when elections are more competitive.

Empirical evidence is mixed:

• Nellis (2022) looks at surveys of attitudes/trust of religious out-groups globally
• proximity to elections does NOT affect distrust/negative attitudes out-groups, even in competitive elections.

Horowitz (1991) identifies how the specific institutions shape conflict

Says we want electoral institutions that:

1. Promote party fractionalization: we want there to be multiple viable parties. Makes it harder for anyone party/group to win/govern on its own \(\to\) coalition building
2. Reward parties for trying to win votes across ethnic boundaries
3. Reduce the risk / costs of not winning

Institutions:

• sets of rules that dictate how people interact
• can be laws/constitutions, rules within organizations
• can be informal rules that people just know

They are the “rules of the game”:

• just like in boardgames, card games, changing rules alter the rational strategies of self-interested players.

Worst institutional design?

First past the post: single member districts, plurality winner (Canada, UK, US, etc.)

• encourage two party competition (opposite of fractionalization)
• no gains by appealing to members of other ethnic groups
• winners get more seats than votes (exacerbates exclusion of losers)

proportional representation: many variants, but, generally: multi-member constituencies, seats allocated in proportion to votes won

Pros:

• encourages party fractionalization \(\to\) coalitions
• avoids total exclusion of minority groups

Cons:

• does not create incentives for ethnic parties to seek votes from other ethnic groups
• possibly unstable coalitions
• a clearer ethnic “census”

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.

• Northern Ireland: in recent years, government is coalition of Sinn Fein and DUP: the ethnic extremes. Took until 2020 to complete negotiations over 2017 elections.

Alternative Vote: single member districts, rank order candidates

• drop candidate with fewest first preferences, allocate second preferences
• repeat until one candidate has majority

Pros:

• encourages fractionalization
• encourages parties to win second preferences from other ethnic groups…

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

• can win some offices at local level that have meaningful power

But do other kinds of institutions limit/encourage conflict?

In Malawi:

• Extensive hostility between the groups
• Express negative stereotypes of each other

In Zambia

• These groups get along amicably
• Often express that they are actually united, one people.

Even when comparing villages just across the border from each other

Why do the same ethnic groups behave differently just across the border?

• cannot be electoral rules, as they are the same in both places

• If elites (and possibly voters) are rational, they seek to mobilize ethnic categories that generate minimum winning coalitions(Posner 2005)
• Jurisdictional boundaries (within which laws made, policies set, patronage given) determine ethnic demography (relative size of membership within different ethnic categories)

Posner (2004):

Malawi: Much smaller country…

• Chewas and Tumbukas make up a large share of the population. Large enough to win elections.
• President mobilized Chewas as Chewas during elections, expanded membership rules to include more people as “Chewa”.

Zambia: Much larger country…

• Chewas and Tumbukas are small groups, not large enough to matter for national elections.
• Parties mobilize regional language groups: Chewas and Tumbukas mobilized together as “Easterners”

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.

• Always used FPTP, SMD… can’t be electoral rules
• Zambia hasn’t changed in size… can’t be administrative boundaries

Between 1972 and 1990, Zambia was a one-party state. Elections were held within the ruling party.

• National level office (President, party control of parliament) not decided in competitive elections
• Only choice of who local MP would be was competitive \(\to\) competition at the local level

Before 1972 and after 1990, Zambia had multi-party competition.

• Elections do choose national leadership (President and ruling party)
• Competition takes place at national level

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:

• language does not form minimum winning coalition \(\to\) tribe

With multi-party competition, national demography matters:

• tribal groups are too small to win, rational to mobilize by language/region

Indian Constitutions recognized and created special protections for previously “untouchable” caste groups

• new category of “Scheduled Castes” or SCs

Despite these protections, by 1980s, growing number of educated and upwardly mobiles SCs faced discrimination, inadequate employment.

Enter the Bahujan Samaj Party:

• Ethnic political party that sought SC votes and campaigned on policies to help out SC community

Uttar Pradesh:

• Large province, population divided along religious, linguistic, and caste lines
• Large SC population (21%), with ~20% literate by 1981.
• SMD, SPTP, multi-party elections.

Karnataka:

• Medium province, population divided along religious, linguistic, and caste lines
• Large SC population (16%), with ~20% literate by 1981.
• SMD, SPTP, multi-party elections.

Despite similarities…

• In Uttar Pradesh, BSP won SC votes, able to lead state assembly
• In Karnataka, BSP could not even recruit candidates to run.

Why does the same ethnic party fail to win in two places?

• cannot be electoral rules, administrative boundaries, party system

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:

  • have enough support to win
  • include ethnic group members in leadership positions

In Uttar Pradesh:

• Major non-ethnic parties (Congress) run by upper castes, would not permit SC leaders to take high ranking positions in the party
• BSP leadership mostly SC and could conceivably win elections.

In Karnataka:

• Major non-ethnic party (Congress) had open competition for leadership positions; SCs included in proportion to their population
• BSP could not offer a credible improvement for representing SC interests

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:

• Constitution/electoral rules
• Administrative boundaries
• Party systems
• Internal party rules

Change the strategic benefits/costs to pursuing ethnic extremism or compromise.

But institutions are not easily be changed.