Notes: All estimates are difference-in-differences estimates for coefficient κ_t for period t=1 (after the intervention). Standard errors are shown in parentheses. Statistical significance is denoted as follows: *** p<0.01; ** p<0.05; * p<0.1. The specification controls for time and household fixed effects, and includes additional time-varying controls for the number of refugees arriving in the sample area (a demand shock) and a household-specific wealth index yield very similar results (for alternative results omitting the controls and fixed effects, see Appendix 3). The subsample comparisons are as follows: WNP – Water Network Project treatment zones and matched control zones; WWNP – Wastewater Network Project treatment zones and matched control zones; Both – Water and Wastewater Network Project treatment zones and matched control zones.

¹ Measured on a four point scale (1 = excellent; 4 = poor)

² Water samples were only collected and analyzed in Zarqa

³ The regressions for this outcome do not control for the time-varying factors because we use the full utility database, rather than restricting to the survey sample.

Table 4. Summary of main impacts on small enterprise behaviors and outcomes