Enhancing the diffusion of Information About Agricultural Technology (Figures)

Figures Figure 1: Timeline of the experimental design Baseline survey (N=990) Rice variety adop;on census, all farmers in each village (N=6,511) Survey with random subset of nonadopters: measure informa;on diffusion (N=1,387) Survey all demonstrators: measure characteris;cs (N≈676) Plan;ng May 2014 June 2014 Seeds distributed via 3 channels Sep/Oct 2014 Nov 2014 Farmer field days in 50 villages Feb/Mar 2015 May 2015 Jul/Aug 2015 Door-to-door sales to measure revealed demand Notes: The figure shows the timing of the activities that were carried out as part of the experiment. Planting for each season occurs in June and harvesting generally occurs in late November to December. 0 .1 Share of villages .2 .3 .4 Figure 2: Distribution of village-level adoption rates separately for field day and non-field day villages 0 .2 .4 .6 Village adoption rate Field day .8 1 No Field day Notes: The figure shows the distribution of the village-level adoption rate for field day and non-field day villages separately. The distributions are based on the adoption data for the approximately 15 farmers per village that received door-to-door sales. 0 .05 .1 .15 .2 Adoption .25 .3 .35 .4 .45 .5 .55 Figure 3: Adoption rates separately for field day attendees and non-attendees Control Treat, Non-attend Treat, attend Notes: The figure shows the raw adoption rates for farmers in control villages, farmers in treatment (field day) villages that did not attend the field days, and farmers in treatment villages that attended the field days. The black heavy lines are 95 percent confidence intervals, where standard errors are clustered at the village level. Tables Table 1: Summary statistics and balance tests Means No Field Day 0.781 Field Day 0.750 p-value 0.467 Mud walls 0.579 0.562 0.515 Thatched roof 0.333 0.340 0.925 Number rooms in house 2.044 2.193 0.128 Years education 5.998 5.970 0.903 Area cultivated in wet season (acres) 2.310 2.094 0.625 Private tubewells in house 0.353 0.331 0.927 Number of cows owned 1.805 1.862 0.627 Swarna user 0.526 0.610 0.088 Has SHG member 0.635 0.649 0.567 Owns mobile phone 0.792 0.837 0.119 BPL card holder 0.598 0.633 0.331 NREGS job card holder 0.671 0.642 0.445 Scheduled Caste or Tribe 0.312 0.337 0.760 Owns television 0.529 0.543 0.679 Owns motorbike 0.175 0.196 0.372 Access to electricity Owns refrigerator 0.042 0.061 0.155 The table shows summary statistics and balance tests using the sample of 1,387 farmers that were not demonstrators during year one. Column 1 displays the mean value of each characteristic in the 50 villages without farmer field days. Column 2 shows mean values in the field day villages. Column 3 shows the p-values for tests of equality, where standard errors are clustered at the village level and each regression includes block fixed effects (randomization strata). Table 2: Effects of field days on knowledge (1) (2) (3) (4) (5) (6) Ever heard Number Difference Maximum Most suitable Length of of farmers with survival land growing Swarna-Sub1 talked to Swarna when flooded type cycle Field day 0.060∗ 0.116∗ -0.037 0.133∗∗∗ 0.057 0.070∗∗ (0.031) (0.065) (0.042) (0.045) (0.038) (0.034) Mean in control villages 0.794 0.572 0.431 0.243 0.725 0.819 Number of Observations 1385 1369 1387 1387 1387 1387 R squared 0.071 0.025 0.109 0.133 0.081 0.127 Data are for 1,387 households that were surveyed in between harvesting during the first season and planting for the second season. The dependent variables are as follows: Column 1 is an indicator for whether the respondent has ever heard of Swarna-Sub1 prior to the interview. Column 2 is the number of other farmers in the village talked to about Swarna-Sub1. Column 3 is an indicator for selecting both flood tolerance and husk color as the main differences between Swarna and Swarna-Sub1. Column 4 is an indicator for knowledge that Swarna-Sub1 can survive up to 2 weeks when the field is flooded during the vegetative stage of the growing season. Column 5 is an indicator for knowledge that Swarna-Sub1 is most appropriate for medium land where flash flooding occurs. Column 6 is an indicator for knowledge that the duration (time from planting to harvest) of Swarna-Sub1 is approximately 140 days. All regressions include block fixed effects. Standard errors that are clustered at the village level are in parentheses. Asterisks indicate that coefficient is statistically significant at the 1% ∗∗∗ , 5% ∗∗ , and 10% ∗ level Table 3: Effects of field days on demand revealed with door-to-door sales Field day HH Controls (1) Buy 0.122∗∗ (0.048) (2) Buy 5 KG 0.086∗∗ (0.043) (3) Buy 10 KG 0.036 (0.032) (4) Buy 0.121∗∗ (0.047) (5) Buy 5 KG 0.083∗∗ (0.042) (6) Buy 10 KG 0.038 (0.032) No No No Yes Yes Yes Strata FE Yes Yes Yes Yes Yes Yes Mean in control villages 0.297 0.147 0.150 0.297 0.147 0.150 Number of Observations 1384 1384 1384 1384 1384 1384 R squared 0.042 0.028 0.012 0.062 0.043 0.028 The dependent variable in columns 1 and 4 is an indicator for whether the farmer purchased Swarna-Sub1 when given a sales offer. The dependent variable in columns 2 and 5 is an indicator for purchase of 1 seed package (5 kg). The dependent variable in columns 3 and 6 is an indicator for purchase of 2 seed packages (10 kg). Household controls are indicator for ST or SC, indicator for BPL card, indicator for NREGS job card, cultivated area, indicator for thatched roof, indicator for mud walls, education of the farmer, age of the farmer, indicator for SHG membership, indicator for private tubewell ownership, indicator for cultivating Swarna, and elevation of the household. Standard errors that are clustered at the village level are in parentheses. Asterisks indicate that coefficient is statistically significant at the 1% ∗∗∗ , 5% ∗∗ , and 10% ∗ levels. Table 4: Differential effects of field days as functions of caste and poverty status Field day Field day * ST or SC (1) Buy 0.083 (0.050) (2) Buy 5 KG 0.046 (0.048) (3) Buy 10 KG 0.036 (0.039) 0.118 (0.079) 0.114∗ (0.065) 0.004 (0.055) Field day * BPL card HH Controls Yes Yes Yes (4) Buy 0.073 (0.062) (5) Buy 5 KG 0.022 (0.057) (6) Buy 10 KG 0.051 (0.039) 0.079 (0.059) 0.101∗ (0.055) -0.022 (0.044) Yes Yes Yes Strata FE Yes Yes Yes Yes Yes Yes Mean in non-field day villages 0.297 0.147 0.150 0.297 0.147 0.150 Number of Observations 1384 1384 1384 1384 1384 1384 R squared 0.066 0.047 0.028 0.064 0.047 0.028 The dependent variable in columns 1 and 4 is an indicator for whether the farmer purchased Swarna-Sub1 when given a sales offer. The dependent variable in columns 2 and 5 is an indicator for purchase of 1 seed package (5 kg). The dependent variable in columns 3 and 6 is an indicator for purchase of 2 seed packages (10 kg). Household controls are indicator for ST or SC, indicator for BPL card, indicator for NREGS job card, cultivated area, indicator for thatched roof, indicator for mud walls, education of the farmer, age of the farmer, indicator for SHG membership, indicator for private tubewell ownership, indicator for cultivating Swarna, and elevation of the household. Standard errors that are clustered at the village level are in parentheses. Asterisks indicate that coefficient is statistically significant at the 1% ∗∗∗ , 5% ∗∗ , and 10% ∗ levels. Table 5: Characteristics of demonstrators Coefficients and SE: (1) Ward Mean 0.305∗∗∗ (0.040) (2) Village meeting -0.096∗ (0.054) (3) SHG meeting -0.188∗∗∗ (0.054) (4) p-value (2)-(3) 0.074 HH has elected panchayat representative 0.114∗∗∗ (0.023) -0.064∗∗ (0.026) -0.082∗∗∗ (0.026) 0.319 HH has member of SHG 0.476∗∗∗ (0.071) 0.056 (0.092) 0.508∗∗∗ (0.071) 0.000 Family or close friend with an SHG president 0.138∗∗∗ (0.033) -0.030 (0.041) 0.064 (0.050) 0.039 Self reported village leader 0.257∗∗∗ (0.033) -0.023 (0.044) -0.028 (0.054) 0.922 Cooperative member 0.210∗∗∗ (0.051) -0.001 (0.068) 0.179∗∗ (0.088) 0.035 Scheduled Caste or Tribe 0.462∗∗∗ (0.088) -0.034 (0.111) -0.196∗ (0.104) 0.066 Education 5.652∗∗∗ (0.599) -0.293 (0.710) 0.092 (0.729) 0.497 Age 49.010∗∗∗ (0.992) 1.526 (1.344) -7.646∗∗∗ (1.337) 0.000 Area cultivated 2.232∗∗∗ (0.212) -0.096 (0.279) -0.305 (0.234) 0.312 Mud walls 0.533∗∗∗ (0.059) 0.053 (0.073) 0.041 (0.088) 0.881 BPL card holder 0.610∗∗∗ (0.062) 0.026 (0.078) 0.023 (0.081) 0.973 Cows owned 2.790∗∗∗ (0.234) -0.509∗ (0.296) -0.806∗∗∗ (0.296) 0.247 Sharecrops land 0.452∗∗∗ (0.067) -0.024 (0.079) -0.037 (0.088) 0.856 Ward member, family, or close friend 3.015∗∗∗ -0.033 -0.119 0.552 (0.086) (0.130) (0.136) The data are from the first survey with all 676 early adopters of Swarna-Sub1. Each row shows regression coefficients of the listed characteristic on indicators for village meeting and SHG meeting villages. The omitted category is the ward member villages and thus the coefficient reported in column 1 is the constant for each regression. Column 4 reports the p-value for the test of equality of the village meeting and SHG meeting villages. Standard errors that are clustered in the village level are in parentheses. Asterisks indicate that coefficient is statistically significant at the 1% ∗∗∗ , 5% ∗∗ , and 10% ∗ levels. Cognitive ability Table 6: Interaction effects between farmer field days and meetings (1) Buy 0.184∗∗ (0.070) (2) Buy 5 KG 0.139∗∗ (0.058) (3) Buy 10 KG 0.045 (0.044) (4) Buy 0.187∗∗∗ (0.067) (5) Buy 5 KG 0.142∗∗ (0.055) (6) Buy 10 KG 0.045 (0.043) Field day * SHG meeting -0.125 (0.108) -0.148 (0.100) 0.023 (0.071) -0.134 (0.102) -0.154 (0.096) 0.020 (0.070) Field day * Village meeting -0.066 (0.113) -0.020 (0.098) -0.047 (0.075) -0.072 (0.110) -0.032 (0.094) -0.041 (0.075) SHG meeting 0.073 (0.082) 0.082 (0.073) -0.009 (0.042) 0.078 (0.079) 0.092 (0.071) -0.014 (0.042) Village meeting 0.015 (0.078) 0.017 (0.055) -0.002 (0.058) 0.021 (0.074) 0.028 (0.053) -0.007 (0.057) No No No Yes Yes Yes Field day HH Controls Strata FE Yes Yes Yes Yes Yes Yes Mean in control villages 0.297 0.147 0.150 0.297 0.147 0.150 Mean in ward villages 0.357 0.185 0.172 0.357 0.185 0.172 Number of Observations 1384 1384 1384 1384 1384 1384 R squared 0.046 0.035 0.015 0.066 0.050 0.030 The dependent variable in columns 1 and 4 is an indicator for whether the farmer purchased Swarna-Sub1 when given a sales offer. The dependent variable in columns 2 and 5 is an indicator for purchase of 1 seed package (5 kg). The dependent variable in columns 3 and 6 is an indicator for purchase of 2 seed packages (10 kg). Household controls are indicator for ST or SC, indicator for BPL card, indicator for NREGS job card, cultivated area, indicator for thatched roof, indicator for mud walls, education of the farmer, age of the farmer, indicator for SHG membership, indicator for private tubewell ownership, indicator for cultivating Swarna, and elevation of the household. Standard errors that are clustered at the village level are in parentheses. Asterisks indicate that coefficient is statistically significant at the 1% ∗∗∗ , 5% ∗∗ , and 10% ∗ levels. Table 7: Pooled effects of identifying demonstrators with meetings on seed adoption Village or SHG meeting Field day HH Controls (1) Buy -0.005 (0.047) (2) Buy 5 KG 0.007 (0.041) (3) Buy 10 KG -0.012 (0.031) (4) Buy -0.004 (0.046) (5) Buy 5 KG 0.013 (0.040) (6) Buy 10 KG -0.016 (0.031) 0.123∗∗ (0.047) 0.086∗∗ (0.042) 0.037 (0.032) 0.121∗∗∗ (0.046) 0.083∗∗ (0.042) 0.039 (0.032) No No No Yes Yes Yes Strata FE Yes Yes Yes Yes Yes Yes Mean in non-field day villages 0.297 0.147 0.150 0.297 0.147 0.150 Mean in Ward villages 0.357 0.185 0.172 0.357 0.185 0.172 Number of Observations 1384 1384 1384 1384 1384 1384 R squared 0.043 0.028 0.013 0.062 0.043 0.029 The dependent variable in columns 1 and 4 is an indicator for whether the farmer purchased Swarna-Sub1 when given a sales offer. The dependent variable in columns 2 and 5 is an indicator for purchase of 1 seed package (5 kg). The dependent variable in columns 3 and 6 is an indicator for purchase of 2 seed packages (10 kg). Household controls are indicator for ST or SC, indicator for BPL card, indicator for NREGS job card, cultivated area, indicator for thatched roof, indicator for mud walls, education of the farmer, age of the farmer, indicator for SHG membership, indicator for private tubewell ownership, indicator for cultivating Swarna, and elevation of the household.. Standard errors that are clustered at the village level are in parentheses. Asterisks indicate that coefficient is statistically significant at the 1% ∗∗∗ , 5% ∗∗ , and 10% ∗ levels. Table 8: Effects of field days on adoption for the entire village, including demonstrators (1) -0.008 (0.028) (2) -0.003 (0.031) SHG meeting 0.016 (0.025) 0.040 (0.029) Field day 0.062∗∗∗ (0.022) 0.079∗∗ (0.033) Village meeting Field day * SHG meeting -0.044 (0.046) Field day * Village meeting -0.007 (0.056) Strata FE Yes Yes Mean in control villages 0.103 0.103 Mean in Ward villages 0.147 0.147 Number of Observations 6511 6511 R squared 0.054 0.055 The dependent variable in both columns is an indicator for whether the farmer adopted Swarna-Sub1 for the 2015 season. The data are from a census of varietal adoption that was carried out with all households in each village shortly after planting decisions were made for the 2015 season. Standard errors that are clustered at the village level are in parentheses. Asterisks indicate that coefficient is statistically significant at the 1% ∗∗∗ , 5% ∗∗ , and 10% ∗ levels. Table 9: Differential effects of farmer field days as a function of connectedness to first year adopters All villages (1) Adopters w/ same surname SHG villages Field day 0.130∗∗ (0.052) (2) Adopter fields within 250 M 0.111∗∗ (0.052) Interaction with Field day -0.001 (0.025) 0.008 (0.024) 0.009 (0.013) 0.147 (0.118) 0.204∗ (0.101) Level term 0.022 (0.018) 0.001 (0.017) -0.000 (0.010) 0.032 (0.083) -0.074 (0.082) Yes 0.303 1354 0.048 Yes 0.296 1332 0.044 Yes 0.296 1332 0.046 Yes 0.350 445 0.057 Yes 0.350 445 0.052 Strata FE Mean in control villages Number of Observations R squared (3) Adopter fields within 500 M 0.090 (0.061) (4) SHG member household -0.025 (0.116) (5) Friend / family of SHG president -0.023 (0.099) The dependent variable in all columns is an indicator for whether the farmer purchased Swarna-Sub1 when given a door-to-door sales offer. Columns 1-3 contain all observations and columns 4-5 only includes SHG villages. The second row reports the coefficient on the interaction between the field day indicator and the variable corresponding to the column title. The third row reports the coefficient on the variable corresponding to the column title. Standard errors that are clustered at the village level are in parentheses. Asterisks indicate that coefficient is statistically significant at the 1% ∗∗∗ , 5% ∗∗ , and 10% ∗ levels. Appendix - For Online Publication Table A1: Correlates of farmer field day attendance SHG meeting Village meeting (1) 0.015 (0.097) (2) 0.014 (0.094) 0.025 (0.089) 0.024 (0.088) ST or SC 0.056 (0.046) Age 0.001 (0.001) Educ. in years 0.011∗ (0.006) HH has Below Poverty Line Card -0.025 (0.044) Rice area -0.009 (0.014) Strata FE Yes Yes p-value Village=SHG 0.88 0.88 Mean in Ward villages 0.671 0.671 Number of Observations 724 722 R squared 0.130 0.142 The dependent variable in both regressions is an indicator for attending the farmer field day. The data are for the 50 villages where field days took place. Standard errors that are clustered at the village level are in parentheses. Asterisks indicate that coefficient is statistically significant at the 1% ∗∗∗ , 5% ∗∗ , and 10% ∗ levels. Table A2: Summary statistics and balance tests Means Ward Member 0.780 SHG Meeting 0.725 Village Meeting 0.788 Joint p-value 0.379 Mud walls 0.558 0.596 0.558 0.689 Thatched roof 0.326 0.386 0.300 0.197 Number rooms in house 2.098 2.164 2.107 0.846 Years education 5.905 6.220 5.836 0.698 Area cultivated in wet season (acres) 1.874 2.652 2.093 0.363 Private tubewells in house 0.367 0.339 0.317 0.663 Number of cows owned 1.811 1.832 1.862 0.976 Swarna user 0.604 0.516 0.589 0.341 Has SHG member 0.670 0.652 0.604 0.411 Owns mobile phone 0.849 0.810 0.786 0.174 BPL card holder 0.556 0.648 0.648 0.060 NREGS job card holder 0.616 0.683 0.672 0.401 Scheduled Caste or Tribe 0.357 0.219 0.396 0.022 Owns television 0.552 0.491 0.565 0.353 Owns motorbike 0.185 0.172 0.201 0.685 Access to electricity Owns refrigerator 0.046 0.047 0.063 0.489 The table shows summary statistics and balance tests using the sample of 1,387 farmers that were not demonstrators during year one. Column 1 displays the mean value in villages where ward members (local politicians) were used to determine demonstrators. Columns 2 and 3 show means in villages where demonstrators were selected by SHG meetings and village meetings, respectively. Column 4 shows the p-values for equality of means in all three arms. Standard errors are clustered at the village level and each regression includes block fixed effects (randomization strata). Table A3: Heterogeneous effect of field days on knowledge by caste Field day Field day * ST or SC (1) Ever heard of Swarna-Sub1 0.064∗ (0.036) (2) Number farmers talked to 0.070 (0.079) (3) Difference with Swarna -0.035 (0.048) (4) Maximum survival when flooded 0.179∗∗∗ (0.048) (5) Most suitable land type 0.048 (0.042) (6) Length of growing cycle 0.087∗∗ (0.037) -0.015 (0.048) 0.145 (0.104) -0.008 (0.075) -0.139∗ (0.073) 0.028 (0.065) -0.049 (0.064) 0.038 -0.122∗ 0.016 0.031 -0.044 -0.036 (0.039) (0.066) (0.055) (0.060) (0.052) (0.052) Mean in control villages 0.794 0.572 0.431 0.243 0.725 0.819 Number of Observations 1385 1369 1387 1387 1387 1387 R squared 0.073 0.027 0.109 0.140 0.082 0.135 Data are for 1,387 households that were surveyed in between harvesting during the first season and planting for the second season. The dependent variables are as follows: Column 1 is an indicator for whether the respondent has ever heard of Swarna-Sub1 prior to the interview. Column 2 is the number of other farmers in the village talked to about Swarna-Sub1. Column 3 is an indicator for selecting both flood tolerance and husk color as the main differences between Swarna and Swarna-Sub1. Column 4 is an indicator for knowledge that Swarna-Sub1 can survive up to 2 weeks when the field is flooded during the vegetative stage of the growing season. Column 5 is an indicator for knowledge that Swarna-Sub1 is most appropriate for medium land where flash flooding occurs. Column 6 is an indicator for knowledge that the duration (time from planting to harvest) of Swarna-Sub1 is approximately 140 days. All regressions include block fixed effects. Standard errors that are clustered at the village level are in parentheses. Asterisks indicate that coefficient is statistically significant at the 1% ∗∗∗ , 5% ∗∗ , and 10% ∗ level ST or SC Table A4: Heterogeneous effect of field days on knowledge by BPL status Field day Field day * BPL Card (1) Ever heard of Swarna-Sub1 0.015 (0.029) (2) Number farmers talked to 0.049 (0.085) (3) Difference with Swarna -0.027 (0.056) (4) Maximum survival when flooded 0.218∗∗∗ (0.053) (5) Most suitable land type 0.056 (0.050) (6) Length of growing cycle 0.093∗∗ (0.045) 0.076 (0.046) 0.115 (0.110) -0.014 (0.062) -0.140∗∗ (0.060) -0.001 (0.059) -0.042 (0.039) -0.088∗∗ -0.150∗∗ -0.070 0.068 -0.020 0.009 (0.034) (0.059) (0.043) (0.043) (0.044) (0.029) Mean in control villages 0.794 0.572 0.433 0.244 0.728 0.821 Number of Observations 1385 1369 1385 1385 1385 1385 R squared 0.078 0.029 0.115 0.139 0.082 0.130 Data are for 1,387 households that were surveyed in between harvesting during the first season and planting for the second season. The dependent variables are as follows: Column 1 is an indicator for whether the respondent has ever heard of Swarna-Sub1 prior to the interview. Column 2 is the number of other farmers in the village talked to about Swarna-Sub1. Column 3 is an indicator for selecting both flood tolerance and husk color as the main differences between Swarna and Swarna-Sub1. Column 4 is an indicator for knowledge that Swarna-Sub1 can survive up to 2 weeks when the field is flooded during the vegetative stage of the growing season. Column 5 is an indicator for knowledge that Swarna-Sub1 is most appropriate for medium land where flash flooding occurs. Column 6 is an indicator for knowledge that the duration (time from planting to harvest) of Swarna-Sub1 is approximately 140 days. All regressions include block fixed effects. Standard errors that are clustered at the village level are in parentheses. Asterisks indicate that coefficient is statistically significant at the 1% ∗∗∗ , 5% ∗∗ , and 10% ∗ level BPL Card Table A5: Village-level production characteristics of demonstrators Village meeting SHG meeting (1) KG planted 0.112 (0.597) (2) Number adopters 1.862 (1.200) (3) More than 5 adopters 0.236∗∗ (0.115) (4) Share successful adopters -0.011 (0.085) (5) Acres transplanted 0.052 (0.126) -0.339 -0.498 0.104 -0.153 -0.197 (0.991) (0.936) (0.113) (0.094) (0.133) p-value Village=SHG 0.63 0.03 0.28 0.13 0.07 Mean in Ward member villages 24.31 6.56 0.22 0.77 0.86 Number of Observations 96 96 96 96 96 R squared 0.003 0.054 0.043 0.036 0.040 The data are from the first survey with all 676 demonstrators of Swarna-Sub1 and are collapsed to the village level. The dependent variables are the total amount of kilograms planted in the nursery bed in the village (column 1), the total number of farmers that received any seed during year 1 (column 2), an indicator variable if there were more than 5 recipients in the village (column 3), the number of recipients that did not lose the crop during nursery-stage flooding (column 4), and the total acres transplanted in the main field (column 5). Heteroskedasticity robust standard errors are in parentheses. Asterisks indicate that coefficient is statistically significant at the 1% ∗∗∗ , 5% ∗∗ , and 10% ∗ levels. Table A6: Heterogeneous effect of selecting demonstrators with SHG meetings SHG meeting SHG meeting*SHG household SHG household Village meeting (1) Ever heard of Swarna-Sub1 -0.015 (0.048) (2) Number farmers talked to -0.063 (0.098) (3) Difference with Swarna -0.103 (0.066) (4) Maximum survival when flooded -0.012 (0.073) (5) Most suitable land type -0.040 (0.062) (6) Length of growing cycle 0.019 (0.064) 0.121∗∗ (0.052) 0.158 (0.109) 0.170∗∗ (0.072) 0.106 (0.066) 0.035 (0.064) 0.016 (0.057) -0.126∗∗∗ (0.030) -0.105 (0.067) -0.153∗∗∗ (0.044) -0.076 (0.052) 0.018 (0.040) 0.031 (0.029) 0.044 -0.046 0.021 0.080 0.024 0.026 (0.040) (0.082) (0.057) (0.062) (0.044) (0.043) Mean in control villages 0.794 0.572 0.431 0.243 0.725 0.819 Number of Observations 1385 1369 1387 1387 1387 1387 R squared 0.088 0.025 0.123 0.123 0.080 0.121 Data are for 1,387 households that were surveyed in between harvesting during the first season and planting for the second season. The dependent variables are as follows: Column 1 is an indicator for whether the respondent has ever heard of Swarna-Sub1 prior to the interview. Column 2 is the number of other farmers in the village talked to about Swarna-Sub1. Column 3 is an indicator for selecting both flood tolerance and husk color as the main differences between Swarna and Swarna-Sub1. Column 4 is an indicator for knowledge that Swarna-Sub1 can survive up to 2 weeks when the field is flooded during the vegetative stage of the growing season. Column 5 is an indicator for knowledge that Swarna-Sub1 is most appropriate for medium land where flash flooding occurs. Column 6 is an indicator for knowledge that the duration (time from planting to harvest) of Swarna-Sub1 is approximately 140 days. All regressions include block fixed effects. Standard errors that are clustered at the village level are in parentheses. Asterisks indicate that coefficient is statistically significant at the 1% ∗∗∗ , 5% ∗∗ , and 10% ∗ level 0 Share of farmers .2 .4 .6 Figure A1: Distribution of surname connections for sample of non-adopting farmers 0 1 2 3 4 5 6 7 8 Early adopters with same surname 9 10 11 Notes: The figure uses the data for 1,387 non-adopting farmers to plot the histogram of the number of early adopters with the same surname. The average farmer in the sample shares a surname with one early adopter. .6 .4 Share of farmers .2 0 0 .2 Share of farmers .4 .6 Figure A2: Distribution of number of early adopter plots within 250 and 500 meters of the household 0 5 Early adopter fields w/in 250m 10 0 5 10 15 Early adopter fields w/in 500m 20 Notes: The figure uses the data for the sample of 1,340 non-adopting farmers to plot the histogram of the number of early adopter fields within 250 meters (the left panel) and 500 meters (the right panel) of the household. There are 47 households for which we did not have GIS coordinates. Figure A3: Heterogeneous effects of field days as function of distance to demonstrators 330 816 157 357 818 -.1 -.1 0 0 208 Effect of field day .1 .2 .3 .4 Effect of field day .1 .2 .3 .4 .5 B: Plots w/in 250 meters .5 A: Same surname 0 1 2+ Number demonstrators same surname 0 1 2+ Number demonstrator plots w/in 250 meters Effect of field day -.3 -.2 -.1 0 .1 .2 .3 .4 .5 C: Plots w/in 500 meters 496 756 80 0 1 2+ Number demonstrator plots w/in 500 meters Notes: Each figure shows the effect of farmer field days for three separate groups of farmers. The squares represent treatment effects and the bands are 95 percent confidence intervals. The number printed next to each estimate gives the number of observations in that regression. Panel A shows separate effects of field days for farmers that had 0, 1, and 2 or more demonstrators sharing their same surname. Panels B and C show effects by proximity between demonstrators’ plots and respondents’ households. Each regression includes block fixed effects and standard errors are clustered at the village level — just as in our main specifications. Table A7: Effects of farmer field days in SHG villages with household controls Field day (1) SHG villages -0.092 (0.119) (2) SHG villages -0.080 (0.095) (3) Other villages 0.144 (0.088) Field day * HH has SHG member 0.208∗ (0.120) 0.008 (0.091) HH has SHG member -0.009 (0.089) 0.064 (0.053) (4) Other villages 0.140∗ (0.074) Field day * Friend/family of SHG president 0.277∗∗ (0.106) 0.028 (0.083) Friend/family of SHG president -0.120 (0.098) 0.076 (0.059) Strata FE Yes Yes Yes Yes HH Controls Yes Yes Yes Yes Mean in control villages 0.350 0.350 0.273 0.273 Number of Observations 445 445 939 937 R squared 0.113 0.112 0.065 0.069 The dependent variable in all columns is an indicator for whether the farmer purchased Swarna-Sub1 when given a door-to-door sales offer. The data are for SHG villages only in columns 1 and 2 and for ward member and village meeting villages in columns 3 and 4. Household controls are indicator for ST or SC, indicator for BPL card, indicator for NREGS job card, cultivated area, indicator for thatched roof, indicator for mud walls, education of the farmer, age of the farmer, indicator for private tubewell ownership, indicator for cultivating Swarna, and elevation of the household. Standard errors that are clustered at the village level are in parentheses. Asterisks indicate that coefficient is statistically significant at the 1% ∗∗∗ , 5% ∗∗ , and 10% ∗ levels.