The Impact of Rural Electrification on Income and Education (Tables)

Table 1. Descriptive statistics of outcomes and matching variables Household Household Whole with without Variables sample electricity electricity (1) (2) (3) PANEL A Economic Outcomes Annual farm income (log) 5.56 5.59 5.53 Annual non-farm income (log) 3.60 3.81 3.27 Education Outcomes (7-18 years old) Literacy 0.85 Years of schooling 3.18 Study time at home (minutes per day) 72.57 PANEL B Matching variables Household size Gender of head of the household (Male=1) Age of household head Whether household head is literate (yes=1) Total number of literates in the household Marital status of household head Access to tap water Amount of land (acres) House structure (brick=1) Whether owns livestock (yes=1) Religion of household head Total population of the village Distance from Dzongkhag (km) Observations Difference (2)-(3) (4) 0.06 0.54*** 0.88 3.48 0.80 2.64 0.08*** 0.84*** 75.75 65.32 10.43*** 4.36 4.42 4.33 0.09 71.21 49.74 0.69 49.47 0.73 49.72 -0.04** -0.25 0.25 0.21 0.27 -0.06*** 1.63 0.73 0.56 3.39 0.71 0.87 0.7 305.03 47.4 2098 1.45 0.71 0.55 3.76 0.63 0.86 0.65 295.36 52.29 1304 1.74 0.75 0.57 3.24 0.75 0.89 0.72 308.9 43.66 794 -0.29*** -0.04** -0.02 0.52 -0.12*** -0.03*** -0.07*** -13.54 8.63*** Notes: *, **, *** denote significance at the 10%, 5%, 1% level, respectively. Dzongkhag is the district headquarter. Table 2: Probit estimates of household's access to electricity Household characteristics Household human capital assets Household size Square of household size Gender of head of household Age of head of household Square of age of head of household Marital status of head of household Literacy status of head of household Total no of literates Religion (1= Buddhist) Household physical assets Main source of drinking water (1 = tap water) Type of house (1 = brick) Amount of agricultural land Square of amount of agricultural land Own cow (1 = yes) Own bull (1 = yes) Own horse (1 = yes) Own poultry (1 = yes) Village-level variables Distance to district headquarter Population of the village Wald chi-square p-value McFadden's Pseudo R-square No of observations (1) Standard error (2) -0.024 -0.005 0.126 -0.019*** 0.000*** 0.096 0.035 0.150*** 0.050 0.057 0.005 0.078 0.007 0.000 0.080 0.082 0.029 0.071 -0.060 0.332*** -0.112*** 0.002*** -0.075 0.022 -0.479*** -0.141 0.060 0.70 0.022 0.001 0.131 0.133 0.134 0.107 Coefficient -0.008*** 0.001 0.000* 0.000 171.11 0.000 0.0635 2040 Notes: Outcome variable is an indicator of households’ electrification status. *** p<0.01, ** p<0.05, * p<0.1 Table 3: Post-matching means of the variables Household Household characteristics with electricity (1) Household size 4.32 Square of household size 22.70 Gender of head of household 0.73 Age of head of household 49.64 Square of age of head of household 2745.2 Marital status of head of household 0.75 Literacy status of head of household 0.27 Total no of literates 1.72 Religion (1= Buddhist) 0.72 Access to tap water) 0.57 Type of house (1 = brick) 0.75 Amount of agricultural land 2.85 Square of amount of agricultural land 18.55 Own livestock (1 = yes) 0.87 Distance to district headquarter 43.71 Population of the village 308.01 Household without electricity (2) 4.29 22.84 0.72 50.25 2801.4 0.73 0.27 1.69 0.72 0.55 0.75 2.96 24.43 0.84 43.89 300.17 % Bias reduction 67.6 89.1 65.1 -142.7 11.3 64.9 93.5 89.6 99.3 -37.5 96.2 78.8 96.8 33.9 98.0 42.1 Notes: *, **, *** denote significance at the 10%, 5%, 1% level, respectively. Matched samples are constructed using nearest neighbor with replacement and common support. Unmatched Matched Table 4: Absolute bias, pseudo-R2, LR p> 2 2 Pseudo-R LR χ χ2 Standardized bias (1) (2) (3) (4) 0.060 160.39 0.000 9.53 0.003 9.60 0.887 2.56 t-stat (3) 0.34 -0.17 0.78 -0.92 -0.87 0.71 0.23 0.50 1.44 1.02 -0.26 -0.77 -1.01 1.89 -0.14 0.73 Table 5: OLS-impacts on household income and children’s schooling Economic outcomes Educational outcomes Log nonStudy Log farm farm Years of time at income income Literacy schooling home Access to electricity Household controls Village controls (1) (2) (3) (4) (5) 0.204 (0.189) yes yes 0.495*** (0.179) yes yes 0.030* (0.015) yes yes 0.546*** (0.113) yes yes 10.17*** (03.478) yes yes Notes: Income is expressed in log form. Educational outcomes are for 7-18 years old children in the household. Economic outcomes and study time are at household level. Household control variables are household size, gender, religion, marital status and literacy of the household head, total number of adult literate members in the household, cultivable land area, access to potable water, housing structure, ownership of livestock; and village control variables are village population and distance to dzongkhag headquarter from the village. *, **, *** denote significance at the 10%, 5%, 1% level, respectively. Table 6: ATT effects of household electrification on income and schooling Matching methods Nearest Caliper Local linear neighbor (5) d=0.01 d=0.001 bw=0.1 bw=0.2 (1) (2) (3) (4) (5) Economic outcomes Log farm income Log non-farm income 0.225 (0.256) 0.221 (0.227) 0.119 (0.239) 0.296 (0.251) 0.261 (0.247) Kernel bw=0.1 (6) bw=0.2 (7) 0.163 (0.181) 0.073 (0.178) 0.680*** 0.736*** 0.689*** 0.763*** 0.758*** 0.672*** 0.612*** (0.211) (0.205) (0.179) (0.199) (0.193) (0.197) (0.179) Educational outcomes Literacy 0.029* 0.027* 0.016 0.027* 0.025* 0.038* 0.051*** (0.017) (0.014) (0.018) (0.017) (0.013) (0.015) (0.018) Years of schooling 0.589*** 0.602*** 0.553*** 0.606*** 0.595*** 0.657*** 0.720*** (0.133) (0.118) (0.142) (0.144) (0.115) (0.105) (0.126) Study time at home (minutes per day) 10.11*** 9.05** 12.05*** 11.48*** 11.07*** 9.91*** 9.35*** (4.82) (4.35) (4.43) (3.68) (3.78) (3.57) (3.19) Notes: Bootstrapped standard errors are shown in parenthesis. Kernel uses normal density. Nearest neighbor done with replacement with five neighbors. Educational outcomes are for 7-18 years old children in the household. Economic outcomes and study time are at household level. *, **, *** denote significance at the 10%, 5%, 1% level, respectively. Table 7. Rosenbaum bound sensitivity analysis test for hidden bias p-critical Gamma (Γ) Log (nonfarm income) Literacy Years of schooling Study time at home 1 < 0.0001 < 0.0001 < 0.0001 0.004 1.1 < 0.0001 < 0.0001 < 0.0001 0.05 1.2 < 0.0001 < 0.0001 0.0002 0.231 1.3 0.002 < 0.0001 0.010 0.539 1.4 0.044 < 0.0001 0.109 0.808 1.5 0.254 < 0.0001 0.406 0.944 Notes: Study time at home is in minutes. Gamma is the log odds of differential assignment due to unobserved factors. Rosenbaum bounds are reported only for the significant ATT estimates in Table 6. Figure 1. Distribution of propensity scores for electrified and non-electrified households Electrified .1 0 Fraction .2 Un-electrified 0 .2 .4 .6 .8 1 0 .2 .4 .6 .8 1 Pr(elec) Graphs by electrification status prior to matching Figure 2. Overlap and common support .2 .4 .6 Propensity Score Untreated Treated: Off support .8 Treated: On support 1