Table 2.
The effects of delivering maternal cash transfers on program performance
| (1) | (2) | (3) | (4) | (5) | (6) | |
|---|---|---|---|---|---|---|
| Woman included, but ineligible | Woman excluded but eligible | Woman received transfer | Tot transfers | Woman received full benefits (frequency and amount) | Stay longer (>25 months) | |
| GOV | −0.003 (0.007) | 0.043 (0.052) | −0.013 (0.019) | 1.823*(0.852) | 0.197** (0.087) | −0.146*** (0.042) |
| Wild P-value | 0.723 | 0.594 | 0.605 | 0.134 | 0.123 | 0.000 |
| Observations | 5110 | 857 | 5967 | 736 | 736 | 736 |
| Mean NGO | 0.02 | 0.24 | 0.12 | 20.07 | 0.78 | 0.15 |
| Clusters | 13 | 13 | 13 | 13 | 13 | 13 |
Notes: The table presents OLS estimates of the effects of delivering the maternal cash transfer program through government health workers (GOV) compared to an NGO on measures of performance. The sample includes the matched villages: 22 from GOV and 22 from NGO. The outcomes include: inclusion error, defined as whether an ineligible woman was enrolled in the program (sample includes all ineligible women) (1); exclusion error, defined as whether an eligible woman was excluded from the program (sample includes all eligible women) (2); whether an enrolled woman received at least one monthly cash transfer (sample includes all enrolled women) (3); total number of monthly cash transfers received since enrollment (4); whether an enrolled woman received the full amount of cash transfers since enrollment (5); whether an enrolled woman remained in the program beyond eligibility (6). Controls include (i) individual demographic controls, mother’s age and education, and head of household’s age and education for child-level analysis; mother’s age and education, and head of household’s age and education for mother-level analysis; (ii) village-level controls, including distance to large and small markets, main source of livelihood (agriculture, livestock, or casual labor), availability of government provided electricity, and participation in a concurrent WASH intervention. Standard errors are clustered at the cluster level. Wild bootstrapped (9999 reps) P- values are presented to take into account the small number of clusters. ***P <0.01, **P <0.05, *P <0.1.