Agricultural bias and the electoral politics of democratic sovereign default Cameron Ballard-Rosa Department of Political Science, University of North Carolina, Chapel Hill International Political Economy Society, 2014
Previous work on debt default. Macroeconomic theories: Eaton & Gersovitz (1981), Bulow & Rogoff (1989), Reinhart, Rogoff & Savastano (2003), Bandiera, Cuaresma & Vincelette (2010)
Previous work on debt default. Macroeconomic theories: Eaton & Gersovitz (1981), Bulow & Rogoff (1989), Reinhart, Rogoff & Savastano (2003), Bandiera, Cuaresma & Vincelette (2010) Reputational theories: Tomz (2007)
Previous work on debt default. Macroeconomic theories: Eaton & Gersovitz (1981), Bulow & Rogoff (1989), Reinhart, Rogoff & Savastano (2003), Bandiera, Cuaresma & Vincelette (2010) Reputational theories: Tomz (2007) Institutional theories: Kraay & Nehru (2006), Van Rijckeghem & Weder (2009), Saiegh (2009), Kohlscheen (2010)
Previous work on debt default. Macroeconomic theories: Eaton & Gersovitz (1981), Bulow & Rogoff (1989), Reinhart, Rogoff & Savastano (2003), Bandiera, Cuaresma & Vincelette (2010) Reputational theories: Tomz (2007) Institutional theories: Kraay & Nehru (2006), Van Rijckeghem & Weder (2009), Saiegh (2009), Kohlscheen (2010) Coalitional/geographic theories: Ballard-Rosa (2014)
Previous work on debt default. Macroeconomic theories: Eaton & Gersovitz (1981), Bulow & Rogoff (1989), Reinhart, Rogoff & Savastano (2003), Bandiera, Cuaresma & Vincelette (2010) Reputational theories: Tomz (2007) Institutional theories: Kraay & Nehru (2006), Van Rijckeghem & Weder (2009), Saiegh (2009), Kohlscheen (2010) Coalitional/geographic theories: Ballard-Rosa (2014) Given that the choice between default and repayment is partly political, how do the preferences of voters and interests groups matter, and what role do political institutions play in determining whether sovereigns repay? We suspect these questions will become major foci in the literature." Tomz & Wright (2013)
Previous work on debt default. Macroeconomic theories: Eaton & Gersovitz (1981), Bulow & Rogoff (1989), Reinhart, Rogoff & Savastano (2003), Bandiera, Cuaresma & Vincelette (2010) Reputational theories: Tomz (2007) Institutional theories: Kraay & Nehru (2006), Van Rijckeghem & Weder (2009), Saiegh (2009), Kohlscheen (2010) Coalitional/geographic theories: Ballard-Rosa (2014) Given that the choice between default and repayment is partly political, how do the preferences of voters and interests groups matter, and what role do political institutions play in determining whether sovereigns repay? We suspect these questions will become major foci in the literature." Tomz & Wright (2013)
My argument During fiscal crises, politicians are unwilling to make changes that affect groups that keep them in power.
My argument During fiscal crises, politicians are unwilling to make changes that affect groups that keep them in power. In democracies:
My argument During fiscal crises, politicians are unwilling to make changes that affect groups that keep them in power. In democracies: Politically powerful groups are those with disproportionate electoral influence.
My argument During fiscal crises, politicians are unwilling to make changes that affect groups that keep them in power. In democracies: Politically powerful groups are those with disproportionate electoral influence. Rural electoral advantages > rural bias > agricultural price supports.
My argument During fiscal crises, politicians are unwilling to make changes that affect groups that keep them in power. In democracies: Politically powerful groups are those with disproportionate electoral influence. Rural electoral advantages > rural bias > agricultural price supports. Agricultural support policies can become a large fiscal burden on the state.
My argument During fiscal crises, politicians are unwilling to make changes that affect groups that keep them in power. In democracies: Politically powerful groups are those with disproportionate electoral influence. Rural electoral advantages > rural bias > agricultural price supports. Agricultural support policies can become a large fiscal burden on the state. Democratic rulers will be unwilling to adjust these policies if the alternative is heavy electoral losses.
Rural-biased default hypotheses Must exist rural electoral advantages.
Rural-biased default hypotheses Must exist rural electoral advantages. H1: More rural democracies will be more likely to default on their sovereign debt.
Rural-biased default hypotheses Must exist rural electoral advantages. H1: More rural democracies will be more likely to default on their sovereign debt. Farm subsidy program must be costly this depends greatly on food trade status.
Rural-biased default hypotheses Must exist rural electoral advantages. H1: More rural democracies will be more likely to default on their sovereign debt. Farm subsidy program must be costly this depends greatly on food trade status. H2: This effect will be most pronounced in food exporters.
Fiscal burden of farm support
Data & Estimation DV: default year (from R&R 2009) in democracies only.
Data & Estimation DV: default year (from R&R 2009) in democracies only. Main explanatory variable: rural population (as percent of total population)
Data & Estimation DV: default year (from R&R 2009) in democracies only. Main explanatory variable: rural population (as percent of total population) Economic controls (GDP per capita, GDP, debt/gdp, debt 2, inflation, trade, foreign reserves)
Data & Estimation DV: default year (from R&R 2009) in democracies only. Main explanatory variable: rural population (as percent of total population) Economic controls (GDP per capita, GDP, debt/gdp, debt 2, inflation, trade, foreign reserves) Demographic explanations (logged population, population density)
Data & Estimation DV: default year (from R&R 2009) in democracies only. Main explanatory variable: rural population (as percent of total population) Economic controls (GDP per capita, GDP, debt/gdp, debt 2, inflation, trade, foreign reserves) Demographic explanations (logged population, population density) Political controls (election, margin of victory, government fractionalization)
Data & Estimation DV: default year (from R&R 2009) in democracies only. Main explanatory variable: rural population (as percent of total population) Economic controls (GDP per capita, GDP, debt/gdp, debt 2, inflation, trade, foreign reserves) Demographic explanations (logged population, population density) Political controls (election, margin of victory, government fractionalization) Estimating model: default it = β 1 rural it 1 + γx it 1 + η i + θ t + ɛ it
Data & Estimation DV: default year (from R&R 2009) in democracies only. Main explanatory variable: rural population (as percent of total population) Economic controls (GDP per capita, GDP, debt/gdp, debt 2, inflation, trade, foreign reserves) Demographic explanations (logged population, population density) Political controls (election, margin of victory, government fractionalization) Estimating model: default it = β 1 rural it 1 + γx it 1 + η i + θ t + ɛ it All models include country and year fixed effects, and report multiple imputation estimates of coefficients with standard errors clustered at the country level.
Baseline specifications (1) (2) (3) (4) VARIABLES Bivar. Econ. Demog. Polit. Rural pop. (% total pop.) 0.015*** 0.013*** 0.015*** 0.015*** (0.005) (0.005) (0.005) (0.005) Economic controls No Yes Yes Yes Demographic controls No No Yes Yes Political controls No No No Yes Observations 1,621 1,621 1,621 1,621 Countries 55 55 55 55 *** p<0.01, ** p<0.05, * p<0.1
Food trade and democratic default (1) (2) (3) (4) (5) (6) VARIABLES Bivar. X Baseline X Polit. X Bivar. M Baseline M Polit. M Rural pop. 0.013*** 0.015*** 0.017*** -.0006163 0.005 0.005 (0.005) (0.006) (0.007) (0.002) (0.003) (0.004) Economic controls No Yes Yes No Yes Yes Political controls No No Yes No No Yes Observations 1,055 1,055 1,055 566 566 566 # countries 41 41 41 34 34 34 *** p<0.01, ** p<0.05, * p<0.1
Also in autocracies? (1) (2) VARIABLES Democ. Autoc. Rural pop. (% total pop.) 0.015*** -0.022*** (0.005) (0.006) Economic controls Yes Yes Demographic controls Yes Yes Observations 1,621 937 Countries 55 42 *** p<0.01, ** p<0.05, * p<0.1
Relative rate of assistance to agriculture
RRA and default in democ.
Next Steps Mechanism validation
Next Steps Mechanism validation Historical evidence
Next Steps Mechanism validation Historical evidence Macro-level correlates
Next Steps Mechanism validation Historical evidence Macro-level correlates Meso-level transmission
Next Steps Mechanism validation Historical evidence Macro-level correlates Meso-level transmission Micro-level interests
Thank you!
The Model Other output tables Multiple Imputation diagnostics Democratic cross-tabs
The Model Other output tables Multiple Imputation diagnostics Rural bias and electoral survival Why should we care about rural electoral interests?
The Model Other output tables Multiple Imputation diagnostics Rural bias and electoral survival Why should we care about rural electoral interests? Much formal work on electoral mobilization advantages has specifically highlighted agricultural producers as social groups with disproportional electoral influence (Dixit & Londgregan (1996), Persson & Tabellini (2000)).
The Model Other output tables Multiple Imputation diagnostics Rural bias and electoral survival Why should we care about rural electoral interests? Much formal work on electoral mobilization advantages has specifically highlighted agricultural producers as social groups with disproportional electoral influence (Dixit & Londgregan (1996), Persson & Tabellini (2000)). Recent empirical work has demonstrated a number of forms of rural especially agricultural biases in democratic policymaking (Varshney (1998), Stasavage (2005), Rodden (2010), Bates & Block (2011), Harding (2012), Golden & Min (2013)).
The Model Other output tables Multiple Imputation diagnostics Rural bias and electoral survival Why should we care about rural electoral interests? Much formal work on electoral mobilization advantages has specifically highlighted agricultural producers as social groups with disproportional electoral influence (Dixit & Londgregan (1996), Persson & Tabellini (2000)). Recent empirical work has demonstrated a number of forms of rural especially agricultural biases in democratic policymaking (Varshney (1998), Stasavage (2005), Rodden (2010), Bates & Block (2011), Harding (2012), Golden & Min (2013)). Work on politics of trade reform has suggested that agricultural price subsidies are one of the most intractable areas of trade liberalization (Davis (2003)).
The Model Other output tables Multiple Imputation diagnostics Countries in external sovereign debt default Source: Reinhart & Rogoff (2009)
The Model Other output tables Multiple Imputation diagnostics Actors I: Citizens 2 citizen groups (U and R), and 2 political parties (I and C). Rural citizens form proportion (1 α) of total population; each produces b units of food. Urban citizens form proportion α of total population; each produces x units of (non-food) good. Consumption utility for citizen i of type j: v ij (x, b) = x + ln(b) Citizens also have individual preferences for I winning office σ ij U[ 1 1 2ψ j, 2ψ j ] for j {R, U} as well as societal valence" shocks ɛ U[ 1 2η, 1 2η ]. Full utility for citizen i of type j: u ij (x, b) = v ij (x, b) + 1{G = I}(σ ij + ɛ)
The Model Other output tables Multiple Imputation diagnostics Actors II: Parties Utility for political party k from winning office: u k (δ) = χ + (1 δ)ρl Government can increase price of b by providing agricultural producer subsidy φ. Government budget constraint: γ C(φ) + (1 δ)d Probability that C wins the election: ν C (φ C φ I ) = η ψ (αψ Uln( π+φ I π+φ C ) + (1 α)ψ R [(φ C φ I ) b + ln( π+φ I π+φ C ]) + 1 2
The Model Other output tables Multiple Imputation diagnostics Timing of the Game 1. Nature draws a world price of food π F(.). 2. The Incumbent and Challenger simultaneously choose whether to default or not, and select a feasible producer price subsidy (δ k {0, 1}, φ k [φ, φ] for k {I, C}). 3. Nature draws values for ɛ and for all σ ij. 4. Citizens vote either for the Incumbent or the Challenger based on u ij (φ I ) versus u ij (φ C ). 5. The winning party assumes (or retains) control of the executive, and payoffs accrue.
The Model Other output tables Multiple Imputation diagnostics Level of rurality (1) (2) (3) (4) (5) (6) VARIABLES Bivar. R Baseline R Polit. R Bivar. U Baseline U Polit. U Rural pop. (%) 0.014*** 0.014** 0.026*** 0.008 0.006 0.009 (0.005) (0.006) (0.008) (0.008) (0.008) (0.01) Baseline controls No Yes Yes No Yes Yes Political controls No No Yes No No Yes Observations 733 733 733 888 888 888 # countries 31 31 31 34 34 34 *** p<0.01, ** p<0.05, * p<0.1
The Model Other output tables Multiple Imputation diagnostics Subsamples and placebos (1) (2) (3) (4) (5) (6) VARIABLES No OECD PR Major. Currency Inflation Banking Rural pop. (%) 0.026*** 0.016** 0.041*** -.0001809 -.0005694.0002611 Full controls Yes Yes Yes Yes Yes Yes Observations 775 1,032 344 1,621 1,621 1,621 # countries 33 44 16 55 55 55 *** p<0.01, ** p<0.05, * p<0.1
The Model Other output tables Multiple Imputation diagnostics Overimputation
The Model Other output tables Multiple Imputation diagnostics Overdispersion