Intervention(s)
The proposed research will take place in Livingstone, Zambia, a city of around 140,000 located near Victoria Falls. The research partner, the Southern Water and Sewerage Company (SWSC), serves around 16,000 residential customers with metered household water connections. Like many urban services around the world, water is billed on an increasing block tariff, with three price tiers (kinks at 6 and 20 cubic meters). Analysis of historical billing data shared by SWSC indicates that consumption spans all three price tiers and that consumption varies from month to month at the customer level. Customers receive monthly bills based on consumption on each of the tiers. Our pilot phase confirms that customers have a low level of awareness of their water price, their consumption and how their bill is calculated. This lack of awareness undermines the price incentive that the increasing tariff is intended to deliver. This lack of awareness also creates a research opportunity. Providing selected households with accurate information about their marginal prices serves as a shock to their price beliefs. Piloting of how to frame price information suggests that describing price as the cost of running the tap for a fixed period of time is most salient to households. Analysis will rely both on administrative records from SWSC and on original data collected through surveys conducted by IPA.
Study sample
We will study the water consumption of households in Livingstone, all of which have been on metered water accounts for at least one year, have a currently well-functioning meter and are cohabiting or married. Households will be selected using existing billing records and a pre-screening visit. Based on the pilots from June-September 2014, roughly 35% of the households visited for screening met these criteria and were eligible for inclusion in our sample. Among eligible households, the modal age group for husbands and wives in the pilot sample was 31-40 years; the modal educational attainment was some college (post-secondary diploma or certificate) for men and Grade 12 (secondary schooling) for women.
Data collection and price variation
A single survey visit will be used to collect household-specific information and to provide salient price information to a sub-sample of households. The data collection and interventions consists of five parts:
1) Beliefs about water prices: Both the male and female heads of household will be asked about their beliefs about water prices. They will be asked both to state the marginal price (described in terms of "running a tap for normal household use”) and their confidence around that estimate. We refined elicitation of price beliefs during piloting. SWSC measures water usage in cubic meters, which is a unit of measurement that is hard for most users to translate into regular usage decisions. We piloted a number of different ways to elicit price beliefs, before settling on a framing that was straightforward for respondents. We ask, “Suppose you wanted to save K. 10 on your monthly bill. Then by how many minutes would your household as a whole have to reduce the use of the tap each day? Think about running the tap at a normal pressure, as you would for typical daily activities.” While the correct answer for most households in the middle pricing tier is 13 minutes, the average belief from the pilots was about 23 minutes, with responses ranging from 2 minutes to 60 minutes.
2) Household survey: Male and female household heads will be interviewed together and separated for a subset of questions and incentivized games. Together, they will be asked a series of questions about household demographics, income and water usage. Questions about income and water consuming durables will help characterize one dimension of heterogeneity in price sensitivity across households. Separately, they will be asked about the water usage decisions of each member of the household including themselves. Comparing answers about own and other's water usage across spouses will reveal how well informed household members are about the usage decisions made by others. In addition to standard survey measures, the male and female household heads will be asked to play two simple incentivized games that will be used to measure cooperation within the household. A measure of altruism toward the environment will serve as a control. These measures will be used to characterize the second dimension of heterogeneity. The survey and incentivized games were refined during piloting. In a modified dictator game, played between spouses, 25% of the husbands and 15% of wives sent their full endowment to their spouse, knowing that it would be doubled if that game were selected for implementation. On average, husbands sent 65% of their endowment to their wives and wives sent 45% of their endowment to their husbands. Controlling for gender, we find that the share of the endowment sent to the spouse is strongly correlated with other measures of household coordination such as whether the respondent says they regularly make plans with their spouse and whether they find it difficult to hide income from their spouse.
3) Price information: Following completion of the household survey, respondents will be given information about the true price of water that they face. Using the average across recent months, individuals will be told their expected marginal price in the coming month, again using easy to understand language (“running the tap”). To ensure the salience of the intervention, households will also be provided with written materials describing their marginal water price. 25% of the survey sample will receive this price information intervention.
4) SWSC credibility: Based on our piloting of the price-information intervention, we found that people doubt the accuracy of their bills, and sometimes think that the water company charges them for more water than they actually use. This is partly justified, because their analog water meters are difficult to read. In addition, the combination of water outages due to infrastructure constraints and annual increases in water prices has led people to think poorly of the water company. People's skepticism about their billing and the potential to actually save money by lowering their water consumption would limit the effectiveness of our information intervention. To overcome this, we have added a second treatment wherein a subset of the households in the treatment and control groups for the information treatment receive additional information about the water company. Specifically, roughly 50% of the overall study sample will be informed that the water company is committed to honest billing practices, and that they try to make the water bills as accurate as possible by reading the meters monthly and using the meter readings to generate the bills. We explain that even during months where the meter is inaccessible or unreadable, the bill is based on an average of previous months' usage, and that future bills are adjusted for any over- or under- charges once they are able to get a reading again. This is information households may not have had previously, and could lead them to take the price information we provide more seriously.
5) Lottery: We are adding a third treatment based on the pilot exercise and due to the additional grant funding we received from the IGC, as well as from the Urban Services Initiative at J-PAL. The treatment is a raffle for successful water conservation outcomes by participating households. We will inform a randomly selected subset of the sample that lowering their monthly water usage will qualify them to participate in the raffle. These households will also be given the price information. Prizes of 300K (about 50USD) will be given to one in every twenty households that make the required reduction each month. The raffle provides households with a clear incentive to reduce water use and allows us to directly observe which households are successful at lowering consumption and which are not. For a third of the households told about the raffle; only the husband will be informed about the raffle; for the other third, only the wife will be informed; and for the rest of the households, both spouses will be informed. We will randomly assign whether the target respondent(s) in each household will be the husband, the wife, or both spouses. Households will be automatically entered in the raffle if they reduce their water usage during the following billing period to a specific target. The surveyors will also show them where on the bill they can find the water usage for that month, so that the respondents can keep track of their usage and know if they have met their target or not. Households can continue to try and qualify for the raffle in future months until we end the program. For the vast majority of households, we believe this will be an interesting challenge that will help them save money and better understand their water price and billing. It will also assist the research by helping us to understand the role of water costs in the amount the household consumes. For example, if the lottery is effective, we can conclude that water prices affect consumption decisions. If the effectiveness varies depending on whether the couple both learn about the lottery versus if one individual in the couple learns about the lottery, then we can conclude that the importance of price depends on who knows about it and who benefits from cutbacks.