INNOVATIONS FOR POVERTY ACTION S RAINWATER STORAGE DEVICE EVALUATION. for RELIEF INTERNATIONAL BASELINE SURVEY REPORT

Transcription

1 INNOVATIONS FOR POVERTY ACTION S RAINWATER STORAGE DEVICE EVALUATION for RELIEF INTERNATIONAL BASELINE SURVEY REPORT January 20, 2010

2 Summary Between October 20, 2010 and December 1, 2010, IPA conducted a baseline survey to set the stage for the implementation of RI s Rainwater Storage Device project in Kamwenge District, Uganda. During that period IPA visited 40 households in each of 81 sample villages and conducted a thorough survey with respondents in their homes. The baseline survey serves three primary purposes: 1. Establish the individual households that will be tracked for the duration of the intervention and study, 2. Establish a starting point, or a baseline, with which to compare against follow-up data for the impact assessment, and 3. After randomly assigning households to treatment groups, if there are any differences in the characteristics between control and treatment groups, this baseline data will allow us to control for these differences. This report is designed to share the preliminary results of the Baseline Survey. IPA would like to make clear that this report focuses primarily on questions and variables that will be used to inform the study as implementation moves forward. The survey produced over 1,300 unique variables for each of the 3,240 respondents and their complete analysis is a long process. We hope and believe that this report will answer any and all questions RI might deem important or pressing but are certainly happy to address anything else of interest in the coming days or weeks. IPA Baseline Survey Report Page 2

3 Section 1. Household Characteristics Household Composition: The mean household size is 5.71 people (median 6) with a composition of: Figure 1 As is illustrated in Figure 1. the mean household has 3.23 children under the age of 18. Educational Background: Household Composition Men (age) Mean Median Adult 17 < Sec School Age 12 < x < Pri School Age 5 < x < Child < Women Adult 17 < Sec School Age 12 < x < Pri School Age 5 < x < Child < The educational attainment of adults in the study area is relatively low. Only 25.2% of household heads report having completed primary school and far fewer have attained a secondary school degree (7.8%). However, self reported vernacular literacy rates reflect a more optimistic level of education: Figure 2 Literacy Rates Men (age) Mean Adult 17 < 81.4% Sec School Age 12 < x < % Pri School Age 5 < x < % Women Adult 17 < 59.0% Sec School Age 12 < x < % Pri School Age 5 < x < % The broad numbers in Figure 2. would also suggest that the gender gap in access to basic education appears to have diminished considerably in the younger generation. IPA Baseline Survey Report Page 3

4 School attendance rates, one particular outcome of interest, is shown in Figure 3.: Figure 3 School Attendence Men (age) Mean Sec School Age 12 < x < % Pri School Age 5 < x < % Women Sec School Age 12 < x < % Pri School Age 5 < x < % One development objective of the RSD intervention is to ease the burden of water collection, a job we think of as usually conducted by women and children. It might plausibly be the case that children are forced to skip school in order to collect water for the household. We will be monitoring this observation at the conclusion of the project to determine whether the adoption of a RSD leads to improved school attendance rates. Section 2. Employment & Income The overwhelming majority of adults in Kamwenge work as subsistent farmers. That is also to say, most adults (18 and over) work. An outline of the labor force in Kamwenge is shown in Figure 4. : Figure 4 Labor Force Basics Working men 81.3% women 85.8% Work as Subsistence Farmers (conditional on working) men 66.0% women 91.2% For working adults, those not farming are dispersed throughout various small industries with the most popular being Small Business Owner (3.4%), Teacher (2.2%), and Taxi Driver (1.0%). Other professions with a scattering of workers include: Animal Husbandry, Vocational work, and Police / Army officer (< 1% each). The dominant presence of subsistence farmers is also reflected in the type of compensation reported by participants. Of people who report working, only 20.1% of men and 4.6% of women are paid a wage in cash. The rest receive their income through the profit derived from their own farm or small business. This reality is reflected in the statistics regarding cash savings and cash income, illustrated below: IPA Baseline Survey Report Page 4

5 Figure 5 Household Cash Position (UGX) Cash currently on hand mean median total 83,426 10,000 < p95 28,525 5,000 Maximum saved in last 12 months total 473, ,000 < p95 213, ,000 When asked about their current financial situation, most households reported that they had no more than 10,000ugx in cash savings (~$4) (see Figure 5.). When asked about the maximum amount they had saved at any one point in the last 12 months these figures increased significantly which meshes well with what we already know about this population: that as subsistence and small-scale farmers their income stream is lumpy, coming mostly in two large pieces during the biannual harvest seasons. Figure 6. below shows self-reported income from the last seven days (weekly) and an average month.: Figure 6 Income (UGX) Weekly mean median men 12,304 0 women 4,152 0 Household 15,851 0 Monthly men 73,583 25,000 women 28,050 5,000 Household 87,606 25,000 Most households report having no cash income in the seven days prior to the survey. It should be noted that the survey was conducted at the height of the rainy season at which point most crops are not yet ready for harvesting. Also, in each of the above figures, note the large disparity between the mean and median observations. That the means are so much greater than the median values suggests that most people make far less than the average household. In fact, when we control for the top 5% of income earners, the figures above change significantly (< p95 in Figure: 5., and below in Figure 7.). IPA Baseline Survey Report Page 5

6 Figure 7 Income excluding top 5% (UGX) Weekly mean median men 5,081 0 women 1,424 0 Household 6,757 0 Monthly men 43,514 20,000 women 13,881 5,000 Household 48,018 20,000 We see in Figure 7. that the mean and median values begin to converge, but there is still a considerable proportion of the overall income of the community concentrated in a small minority. To be clear, to read Figure 7. here we would say that, excluding the top 5% of income earners, the mean household earned 6,757 (~$3) in income in the last seven days. It is important to note that this question gives us a snapshot in time and not a comprehensive understanding of a household s financial situation. In other sections we will look at household consumption and assets which may provide a more thorough understanding of a household s economic standing. Section 3. Assets & Consumption The following are some general statistics about the population under consideration: 76.7% of homes are constructed using mud and wattle walls, the remainder are either bricks or poured cement. 99.1% do not have electricity and 97.0% report lighting their homes with kerosene. 99.0% cook with wood. 95.7% report using a pit latrine and most do not share this with any other households (81.7%). Most households have lived on their property for over 15 years (median 16, mean 19.6). 89.6% report that they own their residency and 47.8% report owning more than one plot of land, most of which are in the same or a neighboring village (95.6%). The median plot size (primary property) is 2 acres (mean 6.8 acres). As is a now typical story in Sub-Saharan Africa, 47.8% of households have one or more working cell phone. IPA Baseline Survey Report Page 6

7 Figure 8. below illustrates the percentage of households that report each of these crops as one of their top three crops under cultivation. The most popular crops cultivated are, maize (74.4% of households grow), beans and peas (65.2%), and bananas (sweet & plantains, 61.7%). Figure 8 Savings & Social Safety Nets When asked, If you needed 50,000ugx ($22) for medical expenses, how would you most likely obtain this money?, the most common responses were: sell an asset (32.2%), borrow from a friend or family (22.7), take a loan (17.7%). 12.1% responded saying that they could not raise the money. The projected retail price of the RSD is 100,000ugx so this is one indicator that we will keep in mind when conducting an analysis of adoption rates: for instance, are people who report being more likely to borrow from friends also more likely adopt when compared to people who report being more inclined to take a loan? Only 11.4% of households have a savings account with a formal financial institution but among households that do, most have had the account for some time (mean 4.7 years). Most often it is only the head of household who has access to this account (65.3%). Far more households report having received a loan from a micro-finance institution (29.2%). Even more still report being members in a Rotating Savings and Credit Associates (ROSCAs, 67.1%). Also, 19.4% of households report being members of one or more Savings and Credit Cooperates (SACCOs, 19.4%), (membership in a ROSCA and a SACCO are not mutually exclusive). Consumption When asked about simple daily purchases such as food (meat & other foods) and drink (soda and alcohol), the mean household reported spending 7,966ugx in the previous seven days (median IPA Baseline Survey Report Page 7

8 4,300ugx). Note here that the median reported seven day consumption sum is roughly 40% of the median cash position reported in Figure 5. and almost equal to that same figure if we exclude the top 5% of cash holders. This is one example of why we look at a household s financial status from multiple viewpoints; if a household reports spending more than it earns and does not report going into debt then we can revise our estimates to reflect these realities. The more informative consumptive index is below. Figure 9. shows the spending patterns from the last 12 months on large items: Figure 9 12 Month Consumption (ugx) Consumptive Area mean median Education 305,819 46,000 Health 218,581 80,000 House 171,757 0 Bicycle 19,773 0 Property 61,874 0 Totals 785, ,500* * Total equal median total consumption. Not sum of above Again we see that most households spend well below the mean values (a small minority spend a disproportionate amount on each of these categories). That said, the mean sum spent in the last 12 months on these large purchases is equal to roughly $300usd. It also shows that most households can and do spend money on healthcare when they must. Another question of interest lies in who makes the purchasing decisions within a household. We hypothesize that a RSD will most likely ease the burden of water collection for women and children (who we assume do most of the collection, we will discuss this later) and this cohort might therefore place a greater value on this device than men. However, when asked, 60.8% of households report that the head of household makes many of the purchasing decisions alone. If in a given household others are involved in purchasing decisions, perhaps these households will be more likely to purchase a RSD. This is another question that we will be tracking once adoption rates become available. IPA Baseline Survey Report Page 8

9 Section 4. Water Sources, Water Collection & Rainwater Water Sources: Most households in the sample collect water from just one source. This primary source is on average estimated to be.52 miles from the household (median.5) with one trip taking an average of 38 minutes on foot (median 30). 91.2% of households collect their own water on foot. The distribution of primary water source types looks as follows: Figure % of households report that they can and do use their primary source year round. 93.4% report that they use this primary source for drinking; 71.7% report boiling the water before consumption. Often this primary source is communal, 91.7% of households share this source with more than five other households. Very few households pay for their water or its delivery (2.1% each). In addition, 41.2% of households report using more than one water source and these sources tend to be further away (.72 miles). If they do visit a second or even third source, fewer households will drink this water (77.6% & 79.1%, respectively). During the rainy season, 96.1% of households report collecting rainwater. Furthermore, 67.6% report that rainwater makes up over half of their total water used during the rainy season. However all but 4.2% of households report that even during the rainy seasons they continue to collect water from other sources. 90.4% use this rainwater for drinking, and in 70.0% of households rainwater constitutes over half of their drinking water when available. On the whole, households are less likely to boil rainwater before drinking (56.9%) than they do water from other sources (above). IPA Baseline Survey Report Page 9

10 Water Collection Rainy Season Note: While we have gathered data on water collection during both the rainy season and the dry season, the report focuses primarily on the rainy season. Two reasons: first, the rainy season data comes from responses that required very little recall by the respondent. These questions were posed as, During the last seven days because the survey was conducted during the rainy season. Second, the RSD by design will primarily alleviate water collection by the household during the rainy season. Thus, this series of data not only more likely to be accurate, but more pertinent as well. That said, mostly for contrast, some numbers from the dry season are included as well. On average, each household has 2.7 people who participate in collecting water during the rainy season (median 2). In 30.4% of households this duty is shared by 4 or more household members. More women per household collect water than men (1.57 vs. 1.16) but the median household has one woman and one man collecting water during the rainy season. On average each household has 1.5 children who participate in water collection. At first glance this might suggest that men and women share the burden of water collection. However, in households headed by men, only 24.2% of those men collect water while in female headed households 49.6% of these women do. Not surprising, when we consider men and women other than the household heads we see less disparity in the share of labor between men and women. The average household collectively spends 11.5 hours (median 6) collecting water each week during the rainy season. When taking into consideration the number of people per household collecting water, this comes out to 4.2 hours per person (median 2.8). Women, conditional on participating in water collection, spends on average 4.3 hours collecting water (median 2.8). For men this same statistic comes out to 4.0 hours (median 2.3). During the dry seasons this numbers go up significantly. Households spend an average of 26.5 hours collecting water (median 16.6), which, when taking into account that more household members participate in water collection during the dry season comes to 8.6 hours per participant (median 6.1). As we can see the burden of collecting water more than doubles during the dry season. This additional burden appears to be shared at approximately the same rate as between men and women during the rainy season: again, conditional on participation in water collection women average 8.8 hours per week (median 6.5) while men average 7.8 hours (median 4.9). Section 5. Social Networks The social network component of the survey was designed with two distinct purposes. First, in the very near future IPA will be supplying RI with a list of Ambassadors for recruitment as one aspect of the marketing treatment. Second, and much later, in our analysis of adoption rates in each village we want to understand the impact of these social networks and how the positive or negative experience of one adopter might impact the propensity of friends versus strangers to adopt the same product. IPA Baseline Survey Report Page 10

11 For now we will focus on the indices that will lead to our selection of Ambassadors. We have elected to experiment with two different types of Ambassadors. The first is someone who is popular and well liked. The second is someone who is seen as being technical and from whom advice on a new technology might be seen as carrying some significance. We have also introduced a variable in the method of compensation, the details of which will be worked out with the RI team on the ground. The experimental design is illustrated in Figure 11.: Figure 11 Let us take a look at the data which is used to generate these indices. First we asked a series of open ended questioned intended to garner a list of people in each study village that either: A) Have the most friends, B) Are considered the most influential, or C) Are considered the most active in their communities. These people are not limited to those in the sample, they may come from any household in a given village. We asked each respondent to list their Top 5 in each of these categories before asking them about their knowledge of or relationship to any specific households. We then asked each respondent about how well they know each of the other 39 households in the sample for their village. To help gauge the level of knowing we included questions about the frequency of conversation and an indicator of trust. Finally, in order to save time and avoid wearing out our respondents, we asked each a series of detailed questions about a subset of ten households. These questions inquired about the level of respect the respondent had for each household and how often they sought out these village members for advice on farming, money and health issues. From these responses we have created two primary indices, one for each type of Ambassador we intend to recruit. The Friend type includes a measure of how frequently a name was listed in the Top 5 friends section, plus their ranking as known in their village along with their score on the respect question from the details module. The Techy index consists of measures of trust and a ranking of who people are most likely to go to for advice and specifically farming advice. IPA Baseline Survey Report Page 11

12 Conclusions One major aspect of the survey not described here is a very detailed account of each household s assets. We measured everything from the number of chickens and cattle to clocks and cooking pots. These responses provide us with insight into the relative welfare of one household versus another but are somewhat baseless in real terms. From these observations we will create asset indices that will allow us to analyze any differences in assets between control and treatment groups, or between specific cohorts of households between the baseline and the endline, but their presentation without comparison is largely meaningless. We also have results from a behavioral section that we hypothesize might predict adoption rates among respondents but again most of these questions simply rank households as the most or least risk averse, or having the characteristics of individuals with a relatively higher or lower discount rate, as the case may be. These and other insights will be forthcoming as analysis of the data continues and we begin to think about questions to include in the endline. Please feel free to contact us if there are any specific questions not addressed here or if clarification is needed on any of the material above. IPA Baseline Survey Report Page 12