• Keine Ergebnisse gefunden

S USTAINABILITY OF I MPACTS AND H ETEROGENEITY BY E XPOSURE TO T REATMENT

While the panel survey was conducted from 2011 to 2013, the difference timing in which each project started functioning provides us with the opportunity to explore the sustainability of the infrastructure and the persistence of the impacts we observe relative to the date from which the project started functioning.

As described in the methodology section, we calculate the time between the finalization of the project (obtained from administrative data) and the day the household was surveyed. Using this information, we obtained the average duration between the finalization of the project and the survey activities for each treatment segment.

Comparison segments do not have a project finalization date, but this information is needed to make meaningful comparisons of before and after with the treatment group. The solution to this is to assigned to each comparison segment a placebo “project finalization date” that corresponds to finalization date of the treatment segment to which it was paired in the matching stage. Since for the comparison segments, this is an arbitrary date, one would expected that the difference between the survey date and this date would not influence the comparison group across years and would create a differential trend in the outcomes of the treatment group after the finalization date. We compute the difference between this date and the date of the survey for all observation, effectively pooling the panel and creating a time-to-event variable that is centered at zero, indicating the days that the WASH projects were finalized.48

The estimating procedure is similar to that used to explore heterogeneity across a categorical variable. The main difference is that the characteristic being interacted is a continuous variable (number of weeks between contract end date and the survey date) instead of a categorical variable and estimate a flexible function or prediction of the outcome based on the continuous variable.

The purpose of this section is to shed some light on the sustainability of the effects found by exploiting the variation in the time each community benefited from the WSS interventions. The impact of the WASH interventions in the following figures is reflected as a change in the slope of the treatment prediction line (solid line) after the zero point on the x-axis. The change in the slope around zero give shows the short-term impacts, as they reflect the effects in the weeks following the finalization of the WASH interventions. As we move to the right on the x-axis (over 100 weeks), the figures show if the effects observed in the short term are sustained as the WASH interventions have been functioning for longer time periods.

First we show the persistence of the impacts on the access to water and sanitation. Figure 45 shows the evolution of the improved sanitation rate in treatment and comparison segments. We found impacts of 3 percentage points in our main ITT specification and 5 percentage point among households that report being direct beneficiaries (ATT), and the figure shows that the impact as time passes (space between the lines) increases, suggesting that new household in treatment segments obtain improved sanitation facilities after the project implementation. In the case of access to tap water in Figure 46, we see that the 25 percentage points impact we found, reflects a persistent impact. Over 80 percent of households in treatment segments have access to tap water 100 weeks after the project end date compared to 60 percent in the comparison segments.

48 While the time-to-event variable should not affect the comparison group it is possible/likely that for comparison segments adjacent to the treated segment these would be affected. This follows from our previous discussion of spillovers in the methodology section. Here we focus on the ITT based on the segment treatment assignment.

Note that the prediction around the beginning and end tend to be noisier, as there are fewer data points at the extremes of the time-to-event variable.

174

FIGURE 45 SUSTAINABILITY OF IMPACTS: ACCESS TO IMPROVED SANITATION

FIGURE 46 SUSTAINABILITY OF IMPACTS: ACCESS TO IMPROVED WATER SYSTEMS

Given the effects on contamination at the point of consumption we detected before, the reliability of the new water system crucial. If the water systems reliability or the frequency of service is degraded with time or rationed, households will re-adopt behaviors, such as inadequate storage and treatment of drinking water, that opens more opportunities for contamination. Figure 47 and Figure 48 shows how the probability of having less than 7 days of service and the number of days per week of available service to explore the question of rationing and see the evolution of the water service in the treatment communities. The increase in the availability of service increased over the short term, up to week 50 after the finalization of the project.

0.2.4.6.81Has own sanitation service

-50 0 50 100

No. Weeks between Contract end and Survey

95% CI Comparison 95% CI Treatment

0.2.4.6.81Has a acces to HH tap connection

-50 0 50 100

No. Weeks between Contract end and Survey

95% CI Comparison 95% CI Treatment

175 However, the figures show that communities might have re-adjusted the days of service provided, as they become familiar with the costs of maintaining the system.

Indeed, anecdotally, in focus groups the water committee leaders that received the training to manage the water systems part of the WASH interventions, expressed that because of the increased costs on electricity for the water pumps of the systems their communities would adjust the days and hours when the service was available. Figure 48 suggest that in treatment segments there was an initial increase in the number of days of service was reverted by the end of the study period; this is consistent with the non-significant impact we found for this indicator. While we found a significant increase in the number of hours per week the service of 16.5 hours per week, Figure 49 shows that this increase in the treatment group served to catch up the treatment to group to the levels in the comparison group. Together these results suggest that treatment communities initially had more days and hours per day of water service and then adjusted back the number of days while increasing the hours each day and maintain a higher and predictable number of hours per week than before the WASH interventions.

FIGURE 47 SUSTAINABILITY OF IMPACTS: RELIABILITY, PROBABILITY OF <7 DAYS OF SERVICE

.4.5.6.7.8.9HH tap is available <7

-50 0 50 100

No. Weeks connected between Contract and Survey

95% CI Comparison 95% CI Treatment

176

FIGURE 48 SUSTAINABILITY OF IMPACTS: RELIABILITY, DAYS OF SERVICE

FIGURE 49 SUSTAINABILITY OF IMPACTS: RELIABILITY, HOURS PER WEEK AVAILABILITY

Continuing with satisfaction of household and their perception of how probable it is to get sick from drinking water from the systems. Figure 50 shows the under 25 percent of households living in treatment segments expressed to be satisfied with their water system and that this proportion increases soon after the contract end date and remains stable after 50 weeks of finalizing the WASH interventions. In the case of the probability of getting sick, households in treatment segments that were connected to the improved system for a longer

4567No. days HH tap available

-50 0 50 100

No. Weeks between Contract end and Survey

95% CI Comparison 95% CI Treatment

-1000100200300Hrs per week HH tap is available

-50 0 50 100

No. Weeks between Contract end and Survey

95% CI Comparison 95% CI Treatment

177 period were more likely to respond that they were satisfied with the system. Figure 51 shows that households in treatment segments are much confident of the safety of the new systems and the proportion of household in treatment segments that perceived that is safe to drink the water directly from the system increases as they become more familiarized with the new system.

FIGURE 50 SUSTAINABILITY OF IMPACTS: SATISFACTION WITH WATER SYSTEMS

FIGURE 51 SUSTAINABILITY OF IMPACTS: PERCEPTION OF RISKS

Consistent with these results, Figure 52 show the effects of duration with service on the probability of using tap water as a principal source for drinking. The duration variable has no effect for the comparison group, as expected. Then, for the treatment group in the weeks after the finalization of the projects, proportion of households using a tap as a principal source increases rapidly, from 30 percent at baseline to 70 percent 100 weeks after the projects were finalized; consistent with the 28 percentage points increase we found.

0.25.5.751Prob. Very satisfied with Water Service

-50 0 50 100

No. Weeks between Contract end and Survey

95% CI Comparison 95% CI Treatment

0.25.5.751Prob. No chance of Getting Sick

-50 0 50 100

No. Weeks between Contract end and Survey

95% CI Comparison 95% CI Treatment

178

In the main impact estimations, we did not found and effect of the WASH interventions on total water consumption and this corroborated in Figure 53. The confidence bands are wide and the prediction line in the treatment and control groups hover around 2,000 liters per week per household. Figure 54 Show the rapid increase in the water consumption from metered taps. Here we note the parallel trends and the difference in levels before the projects were finalized. After the projects are finalized in the treatment segments, their consumption increases and quickly catches up with that of the comparison group.

FIGURE 52 SUSTAINABILITY OF IMPACTS: EFFECTIVE USE, WATER FOR DRINKING

FIGURE 53 SUSTAINABILITY OF IMPACTS: TOTAL WATER CONSUMPTION

0.2.4.6.8 Principal drinking water source is tap connection

-50 0 50 100

No. Weeks connected between Contract and Survey

95% CI Comparison 95% CI Treatment

02000400060008000 Total Water Consumption Lts./Week

-50 0 50 100

No. Weeks connected between Contract and Survey

95% CI Comparison 95% CI Treatment

179 FIGURE 54 SUSTAINABILITY OF IMPACTS: TAP WATER CONSUMPTION

To conclude this section on the sustainability of the WASH interventions and the observed impacts, we present the effects on expenses and time coping costs. Figure 55 shows the increase in the tap expense, and we observe that the expense increases after the implementation for the treatment; due to the new expense in the water bill. The trends in both groups are similar and the increase in expenses for the treatment group, from 3 USD just before the finalization of the project, to 7 USD around 100 weeks after implementation. This provides more evidence that the communities are billing households as the training during the intervention taught them, to make sure that they could cover the maintenance cost of the projects and ensure their economic sustainability.

As before, the time coping costs are reflected on the probability of carrying water and doing laundry outside the home and the time spent in these activities. We present the results on the probability of carrying in Figure 56 and of doing laundry in Figure 57. The change in the probability of a household having to carry water decreases from almost 40 percent to almost 10 percent; however, there was a decreasing trend in the pre-period in the treatment group, which would indicate a possible upward bias if we would a pre-post difference using this adjusted time variable and this comparison group as a reference. The results in Figure 57 for the probability of doing laundry outside the home are smaller but still large, with a decrease from around 30 percent of households doing laundry outside to under 10 percent 100 weeks after the projects were finalized;

consistent with the 13-percentage points impact estimate we found previously. Similar trends can be seen in Figure 58 for the time spent carrying water and in Figure 59 for the time spent doing laundry outside the house. The decrease in the time spent doing these activities imply that the WASH interventions, all but in the eliminated the time spent in these activities, particularly the time spent carrying water.

-100001000200030004000 Liters/Week from Bill from HH Tap

-50 0 50 100

No. Weeks connected between Contract and Survey

95% CI Comparison 95% CI Treatment

180

FIGURE 55 SUSTAINABILITY OF IMPACTS: HOUSEHOLD TAP EXPENSES

FIGURE 56 SUSTAINABILITY OF IMPACTS: COPING, PROBABILITY OF CARRYING WATER

0510 HH tap expense

-50 0 50 100

No. Weeks between Contract end and Survey

95% CI Comparison 95% CI Treatment

0.2.4.6.81 HH carries water

-50 0 50 100

No. Weeks between Contract end and Survey

95% CI Comparison 95% CI Treatment