# Two Sword Lengths: Losers’ Consent and Violence in National Legislatures

Yonsei University

# Paper

Paper available at SSRN

# Outline

• What is legislative violence and why should we care?

• What might predict violence in legislatures?

• Loser’s consent & consensual institutions

• Culture & values

• Hard rules against violence

• Empirical analysis

• Conclusions

# What is legislative violence?

Physical fights between legislators in national legislative chambers.

# Examples of Legislative Violence (1)

From The Guardian

# Examples of Legislative Violence (2)

From The Guardian

# Examples of Legislative Violence (3)

From The Guardian

# Where Does Legislative Violence Occur?

Legislative Violence (1981-Winter 2011)

Data collected by author.

• Legislative violence is not confined to one country or geographic/cultural region.

• But violence is also not randomly distributed.

# So . . .

Legislative violence is more common in:

• weaker democracies

# Why should we care?

## Violence & Losing

• Legislative violence often occurs when losers are very dissatisfied with policy change.

• Legislative Losers: members of the national legislator who are unable to at least block new legislation that they prefer less than the status quo using accepted parliamentary procedures.

• Legislative Winners: members of the national legislator who are able to at least block new legislation that they prefer less than the status quo using accepted parliamentary procedures. Similar to a veto player (Tsebelis 2002).

• Examples:

• Ukraine 2010 (Russian military bases)
• South Korea 2009 (media ownership laws)

# What might predict violence in legislatures?

• As far as I know, there is no academic research that has examined the causes of legislative violence.

• However, there is a considerable body of work that has examined why/why not losers consent (see Anderson and Guillory, 1997; Anderson and Tverdova, 2001; Anderson et al., 2005; Bernauer and Vatter, 2011; Blais and Gélineau, 2007).

• Also there is a related and well developed literature on how consensual democratic instituions can shape the quality of winning and losing (see Lijphart 1977; 1984; 1991; 1999; 2003; 2004).

• Another literature led by Inglehart and Welzel (see Inglehart, 1988; Inglehart and Welzel, 2005, 2010) views societal-level values, such as trust in others, as a reason why people abide by democratic rules.

• Finally, legislatures where members are granted legal immunity may be more prone to violence.

# Violence & Winner/Loser Gaps

Building on the losers’ consent and consensual institutions literatures, I argue that:

Legislative violence is more likely when the gaps between winners and losers are bigger.

# Violence & Winner/Loser Gaps

Building on the losers’ consent and consensual institutions literatures, I argue that:

Legislative violence is more likely when the gaps between winners and losers are bigger.

Bigger gaps make losing more painful and winners’ decisions more difficult to consent to.

# Winner/Loser Gaps (1)

Experience in the other position under the current rules of the game:

• Having experience as both a winner ($$W$$) and a loser ($$L$$) enables actors to learn that “pretenders to office can expect to reach it, losers can expect to come back” (Przeworski 1991, 36).

• Losers learn that the current rules of the game do not always shut them out from decision-making. They can gain over the long-term by following the rules.

• If, however, losers feel that they are shut out, especially over the long-term, they will be less inclined to follow the “rules of the game”.

• Large experience gaps are more common in new democracies.

Experience Gap

# Winner/Loser Gaps (2)

Representation

• If the gap between winners’ ratio of legislative seats to electoral votes is much higher than losers’, losers will likely view winners’ power as more illegitimate.

• A large representation gap is more commonly known as electoral disproportionality.

• High electoral disproportionality is more common with majoritarian electoral systems. It is less common with consensual proportional electoral systems.

# Winner/Loser Gaps (2)

Representation Gap

# Winner/Loser Gaps (3)

Preferences & Power

• When winners’ and losers’ preferences are very far apart, losers will be more dissatisfied with policy changes and be less inclined to follow the rules of the game.

• Preference gaps tend to be larger under majoritarian systems, such as single party parliamentary governments. Gaps are usually smaller under multi-party coalitions.
• However, large preference gaps might not be too bad for losers if winners aren’t able to make dramatic policy changes.

# Winner/Loser Gaps (3)

Preferences & Power

• Powerful winners are able to move policy close to their ideal points.

• Winners are more powerful when there are fewer veto players, as in majoritarian governmental systems like single party governments.
• However, if a winner is powerful, but their preferences are close to the losers’, losers might not be too upset with policy changes.

• So, losers are more likely to be very dissatisfied with outcomes when winners have large power and preference gaps.

# Winner/Loser Gaps (2)

Representation Gap

$$N = \mathrm{New\:Policy}$$
$$SQ = \mathrm{Status\:Quo\:Policy}$$

# Framework

Predicted Probability of Legislative Violence

 New Democracy Old Democracy Majoritarian Likely Unlikely South Korea United Kingdom Consensual Unlikely Unlikely South Africa Sweden

# Empirical Model

• We are interested in the probability that a given country-year $${i}$$ $$(\pi_{i})$$ will experience legislative violence $$(Y_{i} = 1)$$ or not $$(Y_{i} = 0)$$ in a given year.

• Violence between legislators is thankfully rare.

• In my sample of ~200 countries and country years ranging from 1981-2011 there were only 88 observed incidences, 72 when the sample was constricted to 1981-2009 due to data availability.
• So, I used a rare logistic regression (King and Zeng, 2001a,b).

• Implemented with the relogit command in the R package Zelig (Kosuke, King and Lau, 2008).

# Rare Logistic Regression (1)

• Normal logistic regression has both a stocastic component $Y_{i} \sim \mathrm{Bernoulli}(\pi_{i})$ and a systematic component $\pi_{i} = \frac{1}{1 + \mathrm{exp}(-\mathbf{x}_{i} \beta)}.$

• King and Zeng (2001) demonstrated that when the number of events compared to non-events is small, coefficient estimates $$\mathbf{\hat\beta}$$ are biased downward.

# Rare Logistic Regression (2)

• So, they argue that estimated coefficients need to be corrected $$(\mathbf{\tilde\beta} = \mathbf{\hat\beta} - \mathrm{Bias}(\hat\beta))$$ such that the rare events logistic regression systematic component is given by $\tilde\pi_{i} = \mathrm{Pr}(Y_{i} = 1 | \tilde\beta) = \frac{1}{1 + \mathrm{exp}(-\mathbf{x}_{i} \tilde\beta)}.$

• Finally, using $$\mathbf{\tilde\beta}$$ to estimate $$\tilde\pi_{i}$$ gives us a better point estimate, but we still underestimate the uncertainty surrounding $$\tilde\pi_{i}$$.

• King & Zeng recommend using simulations to estimate both the $$\tilde\pi_{i}$$ point estimate and the uncertainty surrounding it.

# Independent Variables

Variable Source
Age of Democracy Marshall and Jaggers (2009)
Disproportionality Gallagher (2012) &
Carey and Hix (2011)
Majority Beck et al. (2001)
Gov. Fractionalization Beck et al. (2001)
Trust World Values Survey
Association (2009)
Immunity Fish and Kroening (2009)

For the full list of variables included in the analysis see the Base Variable Summary table in the paper’s Appendix.

# Main Results: Culture & Hard Rules

• Little evidence that societal level culture or values have an affect on legislative violence.

• Giving legislators’ immunity from prosecution/arrest was also not found to be associated with violence.

# Conclusions

• Legislative violence can be a symptom of some legislators’ dissatisfaction with new democratic institutions.

• The findings here suggest that to solve this problem institutional designers should focus on reducing representation and power/preference gaps. This can include creating:

• electoral systems that produce highly proportional electoral outcomes.

• legislative procedures that encourage supermajority coalitions.