Title page 1

Contents 1

Abstract 2

1. Introduction 4

2. Background 8

3. Experimental Design 10

3.1. Data 11

3.2. Empirical Strategy 12

4. Bounding consumer surplus 12

5. Results 16

5.1. Welfare 19

6. Conclusion 24

References 26

A. Appendix A - Supplemental tables and figures 35

B. Appendix B - Additional details for bounding consumer surplus 43

C. Appendix C - Variable demand for electricity 50

D. Appendix D - phone survey 54

E. Appendix E - all pre-registered results 66

Table 1. Pooled Average Treatment Effects from Incentives and Information 30

Table 2. Bounds on Elasticities and Consumer Surplus 31

Figure 1. The Monthly Reward for Different Consumer Types 32

Figure 2. Heterogeneous ATEs by Pre-Experimental Demand 33

Figure 3. Cross-Sectional Demand for PAYGo Solar 34

Table A1. Balance - Kenya 37

Table A2. Balance - Rwanda 38

Table A3. Pooled Average Treatment Effects from Incentives and Information, Excluding February 40

Table A4. Heterogeneous Average Treatment Effects by Access to Liquidity 41

Table A5. Bounds on Elasticities and Consumer Surplus using Constant Elasticity Demand 46

Table A6. Producer Surplus Calculation 49

Table A7. Pooled Average Treatment Effects from Incentives 67

Table A8. Disaggregated Average Treatment Effects from Incentives 68

Figure A1. Comparison of PAYGo Consumers to the Average Rural Household 35

Figure A2. Pre-Experimental Distribution of Demand among Sampled Consumers 36

Figure A3. Schedule of Incentives 36

Figure A4. Variance in Watt Hours Used Between Consumers 39

Figure A5. Distribution of the Proportion of Treated Consumers within Districts in Rwanda 40

Figure A6. Heterogeneity by the proportion of consumers treated in a district 42

Figure A7. Test for Intertemporal Substitution 52

Figure A8. Heterogeneous ATEs by Pre-Experimental SD (wH used) 53