ECONOMIC DEVELOPMENT & INTERNATIONAL POLITICS (Course

download report

Transcript ECONOMIC DEVELOPMENT & INTERNATIONAL POLITICS (Course

Where do domestic political
institutions come from?
READING ASSIGNMENTS:
Przeworski, Adam, Michael Alvarez, José Antonio Cheibub, and
Fernando Limongi. 2000. Democracy and Development: Political
Regimes and Economic Well-being in the World, 1950-1990. New
York: Cambridge University Press. CHAPTER 2
Boix, Carles and Susan C. Stokes. 2003. Endogenous Democratization.
World Politics 55 (4): 517-549
Gassebner, Martin, Michael J. Lamla, and James Raymond Vreeland.
2013. Extreme Bounds of Democracy. Journal of Conflict Resolution 57
(2): 171-197.
1
What causes democracy?
2
Different stories of democracy:
• Culturalist
• Economic (emergence vs. survival)
• Game theoretic: Credible promise/threat
& income distribution
• Regional Diffusion
3
Culturalist
• The Civic Culture (Almond & Verba 1963)
• Cross-national evidence (Inglehart 1988)
– Democracy associated with high levels of interpersonal
TRUST
• Seligson (2002) shows these correlations are spurious
– They disappear when PER CAPITA INCOME is
controlled for...
4
Democracy
% of pop. saying “most people can be trusted.”
5
$6,000
$5,000
$4,000
Democracy
$3,000
$2,000
$1,000
% of pop. saying “most people can be trusted.”
6
Modernization theory
• Dates back to Lipset (1959) – Correlation between
economic development & democracy
• Common interpretation: Development  Democracy
• But the DYNAMICS of regime transitions are ignored!
– Przeworksi & Limongi (1997) – Modernization: Theories
and Facts
– Przeworski, Alvarez, Cheibub & Limongi (2000) –
Democracy and Development
• (This study is concerned with economic growth –
drops oil rich countries – return to this with Boix and
Stokes!)
7
Think DYNAMICALLY
• Don’t just look at correlations
• Consider
– Onset
– Continuation
8
Basic Stata commands
• regress y x
• regress y x if ylag==0
• regress y x if ylag==1
9
Credible threat & income distribution
(Ross 2001, Rosendorff 2001, Boix 2003, Jensen and Wantchekon 2004,
Acemoglu & Robinson 2006)
• Democracy an elite-question:
– Costs of repression (autocracy)
– vs. Costs of income redistribution (democracy)
• Income distribution obviously matters (higher
income inequality makes repression more
attract)
• Asset specificity –
– oil can’t come with you (Middle East, Nigeria)
– education can (India, South Africa)
10
“That minority still controls the
police, the army, and the economy.
If we lose them, we cannot address
the other issues.”
11
Recall Pevehouse “Hands-tying” story
• How do IO’s get elites to acquiesce?
– The problem to solve:
• Business elites - fear expropriation under a populist leader
– How regional organizations solve the problem:
• Economic IO's may make credible the commitment to
preserve property rights
Temptation to
 expropriate
Populist
leader 
Regional
organizations!
12
Diffusion – an international story
• Pevehouse (2002) mechanism:
– Regional international organizations
• All (international) politics is local (Gleditsch 2002)
– Neighborhood effects
13
Evidence
14
15
A simple guide to
How to read basic “regression” results
1.
What is the analysis “explaining”?
Dependent variable, usually in the title of the table
2.
What is the unit of analysis?
How many observations… of what? (In IO studies, often “country-years”)
3.
What are the independent variables of interest?
Main independent variable(s), Control variables
4.
What is the effect of each independent (explanatory) variable?
Just ask: Is the “coefficient” positive/negative?
5.
Are the effects statistically significant?
a.
b.
c.
d.
Star-gazing *, **, ***
Is the standard error <1/2 the size of the coefficient?
OR: is the t-stat/z-stat >1.96?
OR: is the p-value<0.05?
16
17
Here’s a finding that’s statistically
significant in the first model, but NOT
ROBUST across the other models.
18
More meaningful…
19
Gassebner et al. 2013
20
How robust?
• There are many stories of democracy
• Empirical implications  many variables that purportedly
explain democracy
• Little consensus – different projects use different
specifications
• We apply EBA – 59 proposed factors (1.7 + 1.4 million
regressions)
• EBA is a high standard; variables that fail may matter
• Some variables, however, do survive. We suggest that
these variables may be the most important factors
determining democracy.
21
Data
• Przeworski et al. (2000) Democracy indicator:
– “Democracy is a system in which incumbents lose
elections.”
– Chief executive, legislature face “contested”
elections
– Ex ante uncertainty, Ex post irreversibility,
Repeatability
• Explanatory variables: In total we employ 59
previously suggested in the literature
• Central variable: GDP per capita (M vector)
– Measured in purchasing power parity 1995 US$
22
Results
• The most striking of our findings is that
MOST of the variables suggested in the
literature do NOT survive EBA.
23
Results – Emergence of Democracy
Transition from Autocracy to Democracy (robust variables)
• GDP/capita does not matter for emergence!
– GDP growth does matter but negative!!
• Past transitions (positive), OECD (positive),
Muslim share (negative), Fuel exports (negative)24
25
Results – Survival of Democracy
“Transition” from Democracy to Democracy (robust variables)
• GDP/capita matters (positive)
• Executive is a former military leader (negative)
• Neighboring democracies (positive)
• Past transitions (negative)
26
27
Final model – Do even these variables survive?
• Military leader is co-linear with past transitions
• GDP/capita, neighboring democracy, past transitions
survive, but “neighbor” marginal effect is not significant (the
baseline probability of a democratic survival is 98 percent)
28
Note on theory
• Other variables may indeed matter in wellspecified models.
• But there is great disagreement on theory, and
we choose not to take sides in this project.
• Methodologically neutral approach gives stark
results.
• Few robust determinants of Democracy.
• Policy-makers may care most about these…
29
Assignment
Propose a factor that influences democratization. Justify why this factor
matters with a logical (theoretical) argument. Test your theory using
data. You may use as your base-line model specification, the model of
Gassebner et al. (2013), a required reading listed below.
RECOMMENDED LENGTH: 500 WORDS + 1 page appendix
presenting statistical results.
Click here to download the Gassebner et al. (2013) Stata data file and
Stata do-file.
(Don't worry. The statistical work will be easier than you think! I'll help
you.)
30
Thank you
WE ARE GLOBAL GEORGETOWN!
31