• Keine Ergebnisse gefunden

Comparison Treatment

106

FIGURE 19 DISTRIBUTION OF IMPROVED WATER SCORES AT BASELINE

To quantify this effect, we estimate the impact of the WASH interventions on the probability of having a tap connection and for the water score described above. Table 49 presents the results for the probability of having a household tap and Table 50, the improved water scores show significant positive effects, implying that households moved from lower scores towards higher scores and that households also entered the maximum score category of having a tap connection in their house. We estimate that treatment households experienced an increase in the average score of 0.61 units to 1.05 units; 20 percent of the average score.

In Figure 20, we explore heterogeneity across baseline water source to identify if the households that used worse water sources are the ones that are connecting to the new projects. The estimates show that the increase in access to private taps was driven by households that used public taps, public wells, and trucks.

Households living in treatment segments and using public taps at baseline were 66 percentage points more likely to have a private tap after the WASH interventions when compared to similar households in the comparison group. Consistent with these results, we find larger impacts on the improved water score for the households using natural sources, poliducts, and wells at baseline.

0 5 10 15 20 25 30 35 40 45 50

0 Unimproved 1 Natural Sources 2 Poliducts, Wells 3 Public Taps 4 Backyard Tap,

Neighbors 5 In house Tap

Distribution of Improved Water Scores at Baseline

Comparison Percent Treatment Percent

107 TABLE 49 IMPROVE WATER SERVICES UPTAKE: PROBABILITY OF HAVING A HOUSEHOLD TAP

(1) (2) (3) (4) (5) (6) (7)

Inside Project Area # Post Period 0.35

[0.045]***

Std. errors are clustered at the census segment level.

All equations include year fixed effects. Equations are DID, in (1)-(3) the treatment is defined as living in a treatment assigned segment, (4)-(5) treatment is defined as living inside the project area within the matched pairs, (6)-(7) treatment is defined as having reported being beneficiary of the WASH project within the matched pairs. Equations (4)-(7) control for initial treatment assignment.

Equation (2) includes household fixed effects. Pair dummies indicated in the table are based in on nearest neighbor matching propensity score matching based on 2007 census segment data.

IV estimates in columns (5) and (7) use the census segment treatment assignment to instrument for indicators for being in a project area in 2012-2013 (5); and the households reporting being a beneficiary of the WASH projects from MCC in (7)

^IV estimates partial out the indicators for pairs to compute the std. errors of the coefficients of interest. We report the K-P rk Wald F statistic following the results in Stock and Yogo (2005)

* p<0.10, ** p<0.05, *** p<0.01

108

TABLE 50 IMPROVE WATER SERVICES UPTAKE:: INCREASES IN BEST WATER SOURCE SCORE

(1) (2) (3) (4) (5) (6) (7)

DID DID-FE DID-Pairs RF-Area-ITT

IV-Area-ITT^ RF-ATE IV-LATE^

Treatment # Post Period 0.61 0.6 0.61

[0.10]*** [0.10]*** [0.10]***

Inside Project Area # Post Period 0.86

[0.12]***

Inside Project Area-ITT in Post Period 0.88

[0.15]***

Std. errors are clustered at the census segment level.

All equations include year fixed effects. Equations are DID, in (1)-(3) the treatment is defined as living in a treatment assigned segment, (4)-(5) treatment is defined as living inside the project area within the matched pairs, (6)-(7) treatment is defined as having reported being beneficiary of the WASH project within the matched pairs. Equations (4)-(7) control for initial treatment assignment.

Equation (2) includes household fixed effects. Pair dummies indicated in the table are based in on nearest neighbor matching propensity score matching based on 2007 census segment data.

IV estimates in columns (5) and (7) use the census segment treatment assignment to instrument for indicators for being in a project area in 2012-2013 (5); and the households reporting being a beneficiary of the WASH projects from MCC in (7)

^IV estimates partial out the indicators for pairs to compute the std. errors of the coefficients of interest. We report the K-P rk Wald F statistic following the results in Stock and Yogo (2005)

* p<0.10, ** p<0.05, *** p<0.01

FIGURE 20 UPTAKE: PROBABILITY OF HAVING A TAP BY INITIAL WATER SOURCE

Treatment # Post Period # Rain, Natural Sources Treatment # Post Period # Poliducts Treatment # Post Period # Private Wells Treatment # Post Period # Public Taps & Wells Treatment # Post Period # Neighbors Treatment # Post Period # Trucks Treatment # Post Period # Backyard Tap Treatment # Post Period # In house Tap

Has a acces to HH tap connection

-.5 0 .5 1

DID Triple DID

109 Improvements in access to water services are also reflected as decreases in time to specific sources, such as public, neighbors’, and private taps. Table 51 shows the average traveling time across water sources for specific sources at baseline. The average travel time to all the sources available to the households is eight minutes each way. The sources that are the farthest are springs and other natural sources, which are around 12 minutes each way. While these are not extremely large times to collect water, households tend to make multiple trips per week to fetch water and store it. This, coupled with poorer households being larger and the distribution of travel times as seen in Figure 21, implies that there is room for the water projects to generate considerable time savings.

TABLE 51 AVERAGE TRAVEL TIMES TO WATER SOURCES AT BASELINE (MINUTES)

Mean Standard Error Observations

FIGURE 21 DISTRIBUTION OF TRAVEL TIMES TO SOURCES AT BASELINE

Table 52 presents the results for the average travel time for all sources that households used and for the travel time to the best source the households used. We did not detect effects on the average time to all sources; this

0.05.1.15Density

0 30 60 90 120

How many minutes does it take to walk to it? (one-way)

Public Tap, Control