This is the sixth of the six videos on information treatments as a policy intervention. In this video, I will discuss an example of a case for information treatment from our topology. This is the case in which households are provided with periodic information over an extended period, and the agency has the explicit objective of changing household's behavior in a particular way. To illustrate this type of information treatment and its effectiveness in changing households behavior in one specific location. I will summarize a new paper by Marcela Jaime and Fredrik Carlsson. The paper is entitled, Social Norms and Information Diffusion in Water-saving Programs: Evidence from a Randomized Field Experiment in Colombia. To illustrate this type of information treatment and its effectiveness in changing households' behavior in one specific location, I will summarize a new paper by Marcela Jaime and Fredrik Carlsson. The paper is entitled Social Norms and Information Diffusion in Water-saving Programs: Evidence from a Randomized Field Experiment in Colombia. The policy intervention investigated was to provide information in a household's water bill about the amount of water used by the household's neighbors, and it compare this to the household's own water use. This information was updated monthly and included in household's water bills for one year. The goal of the information treatment was to promote water conservation. That is, to reduce households' water use. The experiment was conducted in Jericho, a town with a population of about 13,000 in Central Northwestern Colombia. The water utility in Jericho provides piped water services to 1,857 households in the urban portion of Jericho. To give you a visual image of the study location, let me show you a few pictures of the town. This is a picture of Jericho, looking down into the valley from the side of a nearby mountain. Lovely place to do research, don't you think? Here is a picture of the town center. This picture shows some of the housing along a street in Jericho. These houses are typical of those owned by the researcher sample households. Before I describe Dr. Jaime and Professor Carlsson's research study in more detail, I want to say a few things about water conservation as a goal of behavioral change in the wash sector. The average monthly household water use in Jericho is about 15 cubic meters per month. The average tariff paid by customers in this town is low, about $0.22 per cubic meter. Water utility officials are selling water below the cost of service, and they feel households are using too much water. One policy option would be to increase tariffs to cost recovery levels. This will send a better signal to households about the cost of providing piped water services. If prices were higher, household water use would shift closer to the socially optimal level. This improved pricing policy will thus to conserve water, but as we have seen, raising water tariffs is often seeing as politically infeasible. So an information treatment that results in water conservation would be welfare enhancing because household water use is higher than socially optimal. Water conservation also reduces utility's cost and this is financially beneficial to the water utility. Thus in Jericho, if the information treatment reduces water use, it may increase economic benefits and reduce financial losses. This is the second best solution, compared to increasing tariffs, because it does not send the proper signal to the utility, to increase its production efficiency. But water conservation is not always welfare enhancing, in some locations, the benefits of increased household water use may be greater than the cost of service. In such a situation, an information treatment then encouraged water conservation what actually cause economic losses, by driving water use further way from socially optimal level. The research design of this study falls closely a design used by Paul Feraro and his co-authors in their research information treatments in Atlanta, Georgia. If you took our 2014 MOOC, you will recall that we discussed Professor Feraro research in the materials on household water demand. Similar to Professor Feraro, Dr. Jaime and Professor Carlson collaborated closely with a local water utility in Jericho, which provided the research team with access to customers' billing records. The sample frame for this experiment included all of the local utilities, 1857 residential customers with working meters. In total, 1548 households were contacted, and 1311 households decided to participate in the survey. The treatment group consisted of 656 households that received the information on social norms. 655 households did not receive the information treatment, and served as controls. A baseline survey was conducted in December 2012, to collect socio-economic and demographic information from sample households. Households that were interviewed in this baseline survey were randomly allocated to either the treatment or control group. The information treatment was provided to the treatment households from January through December, 2013. An ex-post survey was conducted in April, 2014. This ex-post survey updated the socioeconomic information in the baseline survey, and also collected information on household social networks. Next, I want to describe the information treatment in more detail. Households in the treatment group received a personalized report about their water use, and how it compared to their neighbors' water use. They also received a message that encouraged them to conserve water. This is an example of the information received in graphical form. The top figure shows each household, its own water use, the water use of efficient neighbors, and the average water use of all neighbors. These social norm reports were delivered to treatment households each month for one year, starting in January 2013. The basic idea behind this information treatment is that individuals will conform to the behavior of others. For example, if a household is informed that it is using twice as much water as it's neighbors, it will want to reduce it's water use to be closer to the social norm of water use in it's neighborhood. There are different explanations as to why an individual would want to conform the social norms. An individual may feel guilty using more water than neighbors, or they may seek social awards or acceptability by conforming to what others are doing. Another explanation is that an individual may not have complete information about the optimal quantity of water to use. Knowing what it's neighbors are doing may help a household determine the right amount of water use for its members. In this experiment, there are two ways that households may obtain information about neighbors water use. The first is from the information included in the treatment households water bills. This has turned to direct effect of the information treatment. The second is the treatment households may talk to neighbors about the information they receive. So, even if a household is not part of the treatment group, it may receive information from a neighbor that does. Also a treatment household that receives social known information in their water bill, may also receive information from a neighbor who wants to discuss the information they've heard about social norms in the neighborhood. This is termed an indirect or spillover effect. The research team tried to estimate both the direct and indirect effects of the social norm information treatment. Before I present the results of this randomized field experiment, let me tell you a little more about the socioeconomic and demographic characteristics of the sample households. Average household income was about US$257 US per month. Most households in Jericho had already been through the demographic transition. Average household size was 3.3 persons. The average house had about seven rooms, 21% of the sample households had a garden, however, household water use for households with and without a garden was almost the same. Now let's try the results of this randomized field experiment. The research team found that the provision of social norms information decreased water use in the treatment households by about 5 to 6% during the first year following the intervention. There was also a statistically significant but short term spillover effect. Interestingly, the research team concluded that the indirect spillover effects do not seem to result from the social and geographic proximity of household social networks. This figure shows the average direct monthly treatment effects over the 12 months of the experiment for the targeted households with working meters. The dash lines show the 95% confidence levels. This next figure shows how the combined direct and indirect effects of the information treatment decline over time. Still, at the end of the 12 month experiment, the combined treatment effect for all households is about 5%. Again, the dash lines are the 95% confidence levels. The resource team also analyzed the characteristics of the households most likely to reduce their water use in response to the social norms information. Households living in older dwellings reduced their water use by 6% more than households living in new dwellings. Interestingly, households that, prior to the campaign believed they were using less water than their neighbors, but in fact were using more, decreased their water use to a greater extent. 12% more than households who believed they were using more water than their neighbors. The researchers found that the effect of information treatments was only significant for households with high water use, and for high income households. Both groups decreased water use by about 10% compared to baseline. This last finding strikes me as particularly good news. One might be worried that the social norms information would pressure households with low water use to reduce their water use even further. Possibly below the social, the optimal level. But this does not seem to be the case. In conclusion, the size of the direct effect of the information treatment found in this experiment in Columbia, are very similar to the results from a growing number of such studies in the United States. Almost everyone is finding short run direct effects of about 5% or less. To put the results of this field experiment in perspective, in this location the social norms information treatment reduced the average water use of a treatment household from about 15 to 14 cubic meters per month. The cost savings to the utility in terms of current operations expenditures from an effect to this size would be minimal. This is because the majority of the operating expenses, such as labor, are relatively fixed in the short-run. However, the cost savings to the water utility could be significant if the information treatment resulted in a delay of capital investments and capacity expansion. This is because such investments are so capital intensive. Of course, the design and implementation of such information campaigns would cost money. It's not obvious a priority whether the benefits would exceed the costs. One would need to do the cost-benefit analysis. An important question in such a cost-benefit analysis is how long the effects of the information treatment would last, even if it was always included as part of the water bill. We don't have much informational long-run effects of such information treatments. It is possible that households will become less sensitive to social norms information over time. One final point, in another paper, the researchers investigate whether the social norm information treatment caused a spillover effect on household electricity use. They found that electricity use fell by 9% among treatment households that had low water use before the information treatment.