(When) Does Democratization Reduce State Repression?

30

40

50

60

70

−15 −10 −5 0 5 10 15 20Event time relative to base year

Aver

age

Rep

ress

ion

leve

l

Average of democratized countres Average of GSCG

7 richest countries

← 15 years before democratization 20 years after democratization →

←le

ss re

pres

sion

mor

e re

pres

sion

→

Synthetic Control Group

Observed average repression(richer 1/3 of countries)

← 15 years before democratization 20 years after democratization →

←le

ss re

pres

sion

m

ore

repr

essi

on →

30

40

50

60

70

−15 −10 −5 0 5 10 15 20Event time relative to base year

Aver

age

Rep

ress

ion

leve

l

Average of democratized countres Average of GSCG

14 poorest countries

Synthetic Control Group

Observed average repression(poorer 2/3 of countries)

Problem• Repression comprises acts of violations of physical integrity rights by state-sponsored actors• Over the last century, state repression has caused six times as many fatalities as all international

wars and civil conflicts taken together• The literature agrees that democracy is associated with lower repression• However, the effect of a regime change towards democracy is unclear, because regime changes

can go in hand with civil unrest.

Theory• We assume that the welfare of individuals is influenced by both economic and political factors• The objective of the government is to remain in power. To this aim, it can use repression.• In principle, democracy can affect government repression in two directions:

Motivation Opportunity

⤷ Democratic government represents wider spectre of societal groups

⤷ Better representation of preferences⤷ Less aggregate dissatisfaction⤷ Fewer violent opponents

⤷ Under democracy, opposition has more freedom to organize than under autocracy

⤷ Violent opponents more effective under democracy

⤷ Government faces greater threatà Democratizations reduce repression à Democratizations increase repression

• Using a mathematical model, we predict that income acts as a moderating variable• Civil unrest is more closely linked to economic factors in democracies than in autocracies• Hence, higher wealth of the population requires democratic governments to use less repression

Hypothesis: Democratizations are more likely to reduce repression in countries with high income levels and more likely to increase repression in low-income countries.

Data

Jonathan Old, Department of Economics and GRI

Research in collaboration with Martin Rößler and Patrick Zwerschke, TU Dresden, Germany

Conclusions

• We use ”Human Rights Scores” from Fariss (2014) to measure repression• We restrict our analysis to rapid and stable full democratizations. To define a democratization event,

we use the Polity Scores, an index for democracy. A country experiences a democratization event if:(1) It reaches the highest or second-highest value on the democracy index(2) This happens after a substantial improvement in the democracy index within a short time period(3) It has not been democratic for 10 years before and stays democratic for 5 years after the event

• Common control variables are included: GDP per capita, population, civil war, youth bulges, and trade

← le

ss re

pres

sion

mor

e re

pres

sion

→

← 10 years before democratization 10 years after democratization →

With interactions

Highest income per capita

Lowest income per capita

Vertical line marks the base year, two years

before democratization“Traditional” technique: Event study• Run regressions using observations around

democratization event• To test our hypothesis: Include interaction term

between event-time-dummies and income• Rich and poor countries are similar before event• After event, repression only declines in rich countries

and slightly increases in poor countries

• Our theoretical model predicts that democratizations reduce repression only in countries withsufficiently high income. This reduction is immediate and persistent.

• Both traditional and modern econometric techniques provide evidence for this hypothesis• This indicates that promoting democracy may, in some contexts, be detrimental to human rights

“Modern” technique (Xu 2017):Generalized synthetic control group• Divide sample in control and treatment group• Construct “synthetic” (averaged) control group similar to treatment group until the democratization• The difference between treatment and control group after the event is the effect estimate• After event, repression relative to the control group declines in rich countries and does not change

in poorer countries

Reversed Human Rights Scores, data from Fariss (2014)

least repressive

most repressive

Repression in 2013

XPolity Scores, data from Marshall and Gurr (2016), adjusted according to Vreeland (2008)

least democratic

most democratic

Democracy in 2013

The maps suggest a moderate relationship between the level of democracy and repression in 2013.

Very similar trend before democratization

Results

References• Vreeland, JR (2008): The effect of political regime on civil war: Unpacking anocracy. Journal of Conflict Resolution 52(3), 401–425.

• Xu, Y (2017): Generalized synthetic control method: Causal inference with interactive fixed effects models. Pol. Analysis 25(1), 57–76.

• Fariss, CJ (2014): Respect for human rights has improved over time: Modeling the changing standard of accountability.

American Political Science Review 108(2), 297– 318.

• Marshall, MG and TR Gurr (2016): Polity IV Project, Political regime characteristics and transitions, 1800-2015