1 Political Skill and the Democratic Politics of Investment Protection Erica Owen University of Minnesota November 13, 2009
Research Question 2 Low levels of FDI restrictions in developed democracies are paired with increasing public opposition to FDI in the same set of countries Concern about backlash against globalization Skilled and unskilled labor are affected differently by inward FDI Groups in the electorate with more political skills have more influence over investment policy
The Argument 3 1. Distributional consequences of FDI produce cleavages Proportion of labor force that is skilled vs. unskilled Endowment of factors determines comparative advantage Unskilled workers at a comparative disadvantage in developed economies 2. Political skills determine which groups are influential Information Collective action
Departures from Previous Work 4 1. Labor is partitioned into skilled and unskilled workers 2. Take into account that all voters are not equally informed about how FDI inflows affect them 3. Distinguish between economic and political skills
Hypotheses 5 H1: In a comparison of countries, those with a greater proportion of the labor force that has a university education are likely to have lower levels of investment protection than those with a lower proportion. H2: Countries with higher newspaper circulation are likely to have higher levels of investment protection than those with lower newspaper circulation. H3: Countries with higher levels of union membership will be more protectionist than those with lower levels of union membership.
6 Measuring Investment Protection Using Gravity Model Estimate a de facto measure of investment protection De facto measure captures other instruments of protection beyond explicit restrictions (e.g. enforcement, industrial policy, public opinion) Gravity model equation: ( ) FDIijt ln GDP it = α it + β 1 ln(gdp it + GDP jt ) + β 2 ln(similar Market Size ijt ) + β 3 ln( GDPPC it GDPPC jt ) + β 4 ln(distance ij ) + β 5 Contiguous + β 6 Language + β 7 EU + β 8 ln(polity differential) + ɛ ijt
Independent Variables and Empirical Method 7 Data: 21 developed democracies from 1985-2006 Measures of political skill: percent of labor force with higher education, newspaper circulation (per 1,000 people), percent of labor force in a union Other independent variables: unemployment, dummy variables for proportional representation and left party government Control variables: trade openness, government spending as share of GDP, growth of GDP per capita Estimation: Feasible generalized least squares (FGLS) to account for estimation uncertainty in DV
FGLS Results 8 Model 1 Model 1 Proportional representation -3.47*** -3.66*** Left government -1.79*** -1.50*** % Labor force with higher education -0.09*** -0.07*** Newspaper circulation 0.01** 0.02*** Unemployment -0.10 0.48*** Union membership 0.01 0.03*** Newspaper circulation*unemployment -0.002*** Trade openness t 1-0.03*** -0.03*** Government share of GDP t 1-0.20*** -0.18*** Growth GDP per capita t 1-0.12-0.14 Constant 23.1 18.38 N 141 141 Adjusted R 2 0.44 0.51 ***p < 0.01, **p < 0.05, *p < 0.1
Conclusions 9 Open black box about the politics of investment protection Support for hypotheses about role of political skills in shaping the level of investment protection Impact of domestic political institutions: PR systems are less protectionist than majoritarian ones with respect to FDI De facto measure of investment protection Future research: examine how different types of investment affect distributional consequences, collect data to test theory of political skill at industry level
10 Gravity Model ( ) FDIijt ln GDP it = α it + β 1 ln(gdp it + GDP jt ) + β 2 ln(similar Market Size ijt ) + β 3 ln( GDPPC it GDPPC jt ) + β 4 ln(distance ij ) + β 5 Contiguous + β 6 Language + β 7 EU + β 8 ln(polity differential) + ɛ ijt % Reduction in FDI it = 100 ((ˆα benchmark ˆα it )/ŷ benchmark )
Estimation of Gravity Model 11 FE host FE host & FE host & source source & LDV Size 4.81*** 7.18*** 5.73*** GDPpc it GDPpc jt 0.06*** 0.42*** 0.26*** Similarity in size 2.46*** 3.69*** 2.88*** Distance -2.38*** -2.18*** -1.25 Common language 3.12*** 2.70*** 1.56*** Contiguous -0.65*** -0.26-0.09 BIT -1.93-0.15*** 0.05 Polity it -Polity jt -4.31*** -0.26-0.48 FDI flows t 1 - - 0.42*** Constant -78.78 - - Adjusted R 2 0.61 0.61 0.67 N 18865 18865 16990
Levels of Investment Protection in 2005 12 De jure a De facto Level Rank Level Rank Australia 70 7 14.20 17 Austria 70 7 11.81 11 Belgium 90 1 10.26 8 Canada 50 20 15.30 22 Czech Republic 70 7 9.18 5 Denmark 70 7 9.65 7 Finland 70 7 12.56 14 France 70 7 12.05 12 Germany 90 1 14.29 18 Greece 50 20 14.31 19 Hungary 70 7 8.51 2 Ireland 90 1 8.98 4 Italy 70 7 14.74 20 Japan 50 20 18.92 24 Netherlands 90 1 11.71 10 New Zealand 90 1 9.56 6 Norway 50 20 14.02 16 Poland 50 20 8.56 3 Portugal 70 7 10.79 9 Slovak Republic 70 7 8.15 1 Spain 70 7 12.15 13 Switzerland 70 7 12.60 15 UK 90 1 14.95 21 US 70 7 18.45 23 a Investment component of Index of Economic Freedom
Robustness of Measure 13 Primary Primary Primary Base+ Tobit No model +FE +LDV LDV+FE zero-flow Primary model 1 Primary+FE 0.95 1 Primary+LDV 0.86 0.79 1 Primary+LDV+FE 0.88 0.96 0.85 1 Tobit 0.99 0.91 0.85 0.84 1 No zero-flow 0.74 0.84 0.60 0.86 0.67 1 2000 1.00 0.93 0.82 0.86 0.99 0.73 1990 0.99 0.94 0.93 0.89 0.99 0.77 US -0.16-0.45 0.10-0.36-0.29 0.44 France 0.99 0.98 0.85 0.79 0.97 0.59 Demand-side 0.89 0.89 0.76 0.86 0.86 0.78
Feasible Generalized Least Squares 14 Know the variance for estimates of the DV: ω 2 Error from the 2nd stage equals ω 2 plus unexplained variation, σ 2 WLS refits regression using (1/ω 2 ) Want to refit the regression based on the total error: 1/(ω 2 + σ 2 )