Table 2: Ethnic Diversity and Local Public Goods (Kenya and Tanzania) Annual school spending/ pupil, USD

Transcription

1 Table 2: Ethnic Diversity and Local Public Goods (Kenya and Tanzania) Annual school spending/ pupil, USD desks/ pupil latrines/ pupil classrooms/ pupil Proportion wells with normal flow Explanatory variable (1) (2) (3) (4) (5) Local ethnic diversity (ELF) * Kenya Indicator -7.7 (5.9) Local ethnic diversity (ELF) 4.1 (5.6) Kenya indicator variable -4.1 (3.6) Socio-economic controls Average years of education 0.52 (0.55) Proportion formal sector employment (9.0) Proportion homes with iron roofs -1.9 (2.3) Proportion households grow cash -0.8 crops (2.2) Proportion households own cattle -2.6 (2.5) Proportion Catholic 1.9 (3.3) Socio-economic controls * Kenya Indicator ** (0.15) 0.08 (0.09) (0.15) (0.011) 0.30 * (0.17) (0.04) (0.05) (0.09) (0.013) ** (0.003) ** (0.012) *** (0.0004) ** (0.006) *** (0.002) (0.002) *** (0.003) (0.003) (0.012) ** (0.009) * (0.0007) (0.010) (0.004) (0.002) (0.009) 0.20 (0.31) (0.26) ** (0.037) (0.58) 0.12 (0.18) (0.18) (0.52) Yes Yes Yes Yes Yes R Root MSE Number of observations Ethnic diversity effect, Kenya -3.6 * (2.0) ** (0.12) (0.012) (0.010) (0.19) H 0 : β = 0, p-value (SUR) 0.02 ** ** 0.08 * ** Table 2 Notes: 1.) Huber robust standard errors in parentheses. Significantly different than zero at 90% (*), 95% (**), 99% (***) confidence. Regression disturbance terms are clustered at the zone level for Kenya, and at the ward level for Tanzania. The data contains 84 primary schools in Busia, Kenya, and for 66 villages in Meatu, Tanzania, for all outcomes. 2.) The hypothesis that the coefficient estimate on each term is equal to zero across the five outcomes in Table 2 is tested using SUR in the final column. 43

2 Table 5: Logistic Estimates of Participation in Hezbollah Dependent Variable is 1 if Individual is a Deceased Hezbollah Militant, and 0 Otherwise Standard errors shown in parentheses All of Lebanon: Heavily Shiite Regions: Unweighted Estimates Weighted Estimates Weighted Estimates (1) (2) (3) (4) (5) (6) Intercept (0.365) (0.391) (0.230) (0.255) (0.232) (0.261) Attended Secondary School or Higher (1=yes) (0.191) (0.193) (0.159) (0.164) (0.159) (0.167) Poverty (1=yes) (0.221) (0.223) (0.158) (0.162) (0.159) (0.166) Age (0.015) (0.015) (0.008) (0.008) (0.008) (0.008) Beirut (1=yes) (0.219) (0.209) (0.222) South Lebanon (1=yes) (0.232) (0.221) (0.221) Pseudo R-Square Sample Size 120, , , ,925 34,826 34,826 Notes: Sample pools together observations on 129 deceased Hezbollah fighters and the general Lebanese population from 1996 PHS. Weights used in columns (3) and (4) are the relative share of Hezbollah militants in the population to their share in the sample and relative share of PHS respondents in the sample to their share in the population. Weight is for Hezbollah sample and.093 for PHS sample.

3 Explanatory variable Table 2: Vegetation Deviations and Economic Growth (First-Stage and Reduced-Form) Dependent variable: Annual economic growth rate (t t-1) Civil conflict 25 deaths, PRIO Civil conflict 1000 deaths, PRIO OLS OLS OLS OLS OLS OLS (1) (2) (3) (4) (5) (6) Deviation from annual average vegetation, time t 0.31 *** (0.10) Deviation from annual average vegetation, time t *** (0.10) (Vegetation, time t) (Vegetation, time t-1) 0.30 *** (0.08) 0.31 *** (0.08) Democracy (Polity IV), time t (0.0006) Ethno-linguistic fractionalization (0.0113) Religious fractionalization (0.013) Oil exporting country *** Log(mountainous) (0.0020) Log (national population), time t (0.0022) 0.32 *** (0.08) *** Country fixed effects No No No Yes Yes Yes R Root MSE Number of observations Table 2 Notes: confidence. Regression disturbance terms are clustered at the country level. A year time trend is included in all specifications (coefficient estimates not reported). Regressions 1-4 are first-stage specifications; regressions 5 and 6 are the reduced-form specifications *** 18

4 Table 3: Economic Growth and Civil Conflict Dependent variable: Civil conflict 25 deaths, PRIO Civil conflict 1000 deaths, PRIO Probit OLS OLS IV-2SLS IV-2SLS IV-2SLS (marginal) Explanatory variable (1) (2) (3) (4) (5) (6) Annual economic growth rate (t t-1) ** (0.28) Democracy (Polity IV), time t (0.006) Ethno-linguistic fractionalization 0.22 (0.26) Religious fractionalization (0.26) Oil exporting country (0.20) Log(mountainous) ** (0.043) Log (national population), time t (0.052) ** (0.28) (0.27) (0.19) * (0.040) (0.048) * (0.25) ** (0.98) (0.27) (0.20) * (0.039) 0.71 (0.49) ** (0.96) *** (0.78) Country fixed effects No No Yes No Yes Yes R Root MSE Number of observations Table 3 Notes: confidence. Regression disturbance terms are clustered at the country level. The IV for annual economic growth in (4) and (5) is {(Vegetation, time t) (Vegetation, time t-1)}. A year time trend is included in all specifications (coefficient estimates not reported). 19

5 Table 3: Extreme Rainfall and Village Calamities Dependent variable: Annual per capita consumption expenditures (USD) Famine Human disease epidemic OLS OLS OLS OLS OLS Explanatory variable (1) (2) (3) (4) (5) Extreme rainfall (drought or flood) ** (24.8) * (26.6) Human disease epidemic 4.4 (25.7) Drought * (21.3) Flood (48.4) Average years of education 1.7 (13.0) Proportion Sukuma ethnic group (63.5) Proportion households grow cash crops -2.7 (56.2) Households per village / Proportion practice traditional religions 17.2 (52.5) Women s community groups per 2116 household (2492) 1.8 (13.4) (64.8) -2.9 (56.3) (53.4) 2083 (2465) 0.0 (12.9) (65.3) 3.7 (56.2) (52.4) 2333 (2571) 0.47 *** 0.04 (0.05) (0.04) Geographic division fixed effects Yes Yes Yes No No Village fixed effects (67 villages) No No No Yes Yes R Root MSE Mean of dependent variable Number of observations Table 3 Notes: confidence. Observations are weighted by the number of households per village. Regression disturbance terms are clustered at the village level. Regression 1 only contains data for 2001, the only year in which a household consumption expenditure survey was conducted. In Regression 3, we cannot reject the hypothesis that the coefficient estimates on Drought and Flood are equal (p-value=0.50). 31

6 Table 4: Extreme Rainfall and Witch Murders Dependent variable: Witch murders OLS OLS OLS OLS OLS Explanatory variable (1) (2) (3) (4) (5) Extreme rainfall (drought or flood) ** ** (0.037) (0.059) Extreme rainfall, previous year Extreme rainfall, current year and previous year (0.080) ** Human disease epidemic (0.036) (0.038) Village fixed effects (67 villages) Yes No Yes Yes Yes Socioeconomic controls, and geographic division fixed effects No Yes No No No Year fixed effects (11 years) No No No No Yes R Root MSE Mean of dependent variable Number of observations Table 4 Notes: confidence. Observations are weighted by the number of households per village. Regression disturbance terms are clustered at the village level. Socioeconomic controls include Average years of education, Proportion Sukuma ethnic group, Proportion households grow cash crops, Households per village / 1000, Proportion practice traditional religions, and Women s community groups per household. 32

7 Table 5: Satellite Vegetation (NDVI) Data and Witch Murders Dependent variable: Witch Murders OLS OLS OLS OLS Explanatory variable (1) (2) (3) (4) Extreme rainfall (drought or flood) ** Deviation from average vegetation during rainy season > 0.08 Deviation from average vegetation during rainy season > 0.09 Deviation from average vegetation during rainy season > (0.034) * (0.037) Village fixed effects (67 villages) Yes Yes Yes Yes R Root MSE Mean of dependent variable Number of observations Table 5 Notes: confidence. Observations are weighted by the number of households per village. Regression disturbance terms are clustered at the level of the satellite image pixel, and there are a total of 51 such clusters. Regression 1 reproduces the results of Table 4, Regression 1. When it is included as the main explanatory variable, the coefficient estimate on Deviation from average vegetation during rainy season is 0.31 (standard error 0.35). 33

8 Dependent variable Table 6: Extreme Rainfall and Violent Crime Coefficient estimate on Extreme rainfall (drought or flood) Panel A: Witch Murders and Attacks 1) Witch murders ** 2) Witch murders per 1000 households * (0.094) 3) Witch murders and attacks * (0.082) 4) Witch murders and attacks per 1000 households (0.162) Panel B: Non-witch Murders 5) Non-witch murders (0.036) 6) Non-witch murders per 1000 households (0.08) Panel C: Total Murders 7) Total murders (0.068) 8) Total murders per 1000 households (0.124) R 2 Root MSE Table 6 Notes: confidence. Observations are weighted by the number of households per village. Regression disturbance terms are clustered at the village level. Village fixed effects are included in all specifications, which are analogous to Table 4, regression 1. All regressions have 736 observations. Each coefficient estimate is from a separate regression. 34