Um So welcome, everybody. Um, um, my name is Dion Filmer, I'm director of the World Bank's Research Department. I'm really Excited to host this policy research talk today. Some of you may know these talks provide us an opportunity to present work coming out of the research department, uh, sharing with each other, sharing outside of the department, and also sharing with people outside the World Bank. I just want to welcome all our online audience, both on Webex and YouTube. For those on Webex, we're doing things a little differently this time. Um, you'll all be flipped over into the panelist mode. So that hopefully when we get to the Q&A you can all come in alive. Um, so if you don't, uh, since you'll be a panelist mode, you might want to make sure your camera's off if you don't want your camera to appear and I'd ask you to mute as well. Um, today's talk is going to focus on public utility policies in low and middle income countries. Uh, my colleague Hannan Jacoby, who's a lead economist in the sustainability and infrastructure team in the, in the department, research department, will discuss how behavioral economics can provide insights into the patterns of consumer behaviors. Which can then help shape public policies related to electricity and water provision. Um, we're grateful to have Richard Demena today as discussant. Richard is the chief economist of the Sustainable Development Practice here at the bank. He was previously a senior advisor in the water practice and also previously a lead economist in each of the Africa, South Asia, and LAC regions of the bank. So he's been around the institution. Prior to that, he was, uh, to joining the bank, Richard was an economics professor at the University of Adelaide. With that, I'll turn it over to Hannan, who talked for approximately 45 minutes, after which we'll hear from Richard for about 10 to 15 minutes. Uh, we'll conclude the session with Q&A from the audience. And as I mentioned, if you have a question, Jeez, please, uh, just put your name in the chat and I can call on you. Uh, if you're following on YouTube, uh, please submit your question in the chat, and that'll get relayed to me. Uh, with that, OK, over to you, Hannan. OK, thank you. Uh. Can you see my Slides. Yeah, slides. All right, very good. I hope you can hear me OK, um. All right, uh, so. I want to thank Dion for that introduction and uh I thank Richard in advance for his comments and uh thank you all for tuning in today. Um, so governments, uh, all over the world are spending considerable sums, uh, providing basic utility services such as water and electricity. So naturally, the provision and the pricing of these services are major policy concerns. And um sort of the policy questions that arise can be organized into four bins. Uh, there's questions of cost recovery, ah. How much to charge for the service versus say charging for connection to the utility. Um, there's, there are issues of expanding access to the service. Um, say, to underserved regions, ah, and, and the question of course is, is the willingness to pay, ah, for the service high enough to justify, ah, expansion? And if not, how much, uh, uh, of a subsidy will be required, ah, to the utility, ah, to, ah, bring it out to a particular area. Um, then we have questions of affordability, um, you know, how should the so-called social tariff be set? Um, you know, that would be, say, the, the lowest, uh, uh, price, the lowest tariff in the, in the, in the tariff structure. Uh, how should we set that? Um, Ah, what are the, ah, welfare and distributional impacts of subsidy reform? Of course, that's a big issue, um, ah, in the, ah, in energy, um. And, uh, a related point is, uh, it was also coming to the fore more in recent years, uh, particularly is conservation and efficiency goals. Ah, so how do consumers respond to conservation incentives such as peak load pricing, um, and what is the role of providing usage information, um, ah, in, in enhancing our conservation efforts. So that, that's an issue we'll, we'll talk about today. Um, so Broadly, ah, how does, um, neoclassical, ah, economics, and by that I mean sort of your, your econ 101, ah, basic price theory. How, how does it approach these policy questions? Um, and so there's a toolkit, um, Um, that, that is in wide use, um, ah, that we, uh, that we use to, to, to look at these questions. And it consists really of 22 tools, two graphs you might say. Um, the first is, uh, looks at the extensive margin, which is the, the margin of connection to the utility. And we have, um, an extent extensive margin demand curve. Um, you know, as the number of connections, ah, ah, increases, the willingness to pay, ah, goes down. Ah, so, you know, as you, as you connect, um, say more and more, ah, you know, more and more remote areas, the willingness to pay for the service, ah, may go down. So that's the idea of a, of a, of a, of a downward sloping extensive margin demand curve. Um, and then there's a, an average cost of providing a connection, which is also declining in the number of connections. Now, the reason for that is that utilities are characterized by, by high fixed cost, uh, you know, to build the, the plant and equipment and then, and the distribution network and so on. And so there's a following average cost of connecting our customers. So, so you have this kind of picture. Um, and in this case, ah, you know, the way it's pictured, um, this is not a viable utility because the average cost is everywhere above the willingness to pay. So a simple cost-benefit analysis indicates this is not a viable utility. Now, this picture is literally, um, ah, was estimated, the two figures, the two curves in this, in this diagram were estimated in a recent Article by Lee et al. in looking, looking at electrification, rural electrification in, in Kenya. So, uh, this, this was, uh, you know, they use this, this tool um to, to answer the question, you know, should, should we, uh, should we expand electrification in rural, rural Kenya. Um, and they came to the conclusion that, that it was, it was not viable to do so. Now, the second tool in the box, uh, is, ah, looks at the intensive margin of usage, that is, ah, say usage, ah, within a billing cycle, ah, for, for a consumer. Ah, as, as a function of price. So, ah, the typical price schedule will have a low price, ah, for the service up to a particular, ah, threshold of usage beyond which the price will increase. So this is the increasing block tariff, IBT and I'll, I'll be, I'll come back to this again and again throughout the talk. Um, so, um, Ah, there will be consumers who are, have low demand, who are, let's say poor, ah, and they will, ah, choose to consume on the low price block, uh, here. Um, and, ah, they'll face the low price and there'll be, ah, ah, high-demand consumers, let's say the rich, who will face, ah, who will be on the high price block. ah. And what you can see is that the increasing block tariff is a, is a, is a form of price discrimination. It, it allows the utility to efficiently raise revenue by charging higher prices to high demand consumers. Um, and, but this basic, ah, tool, this basic diagram, you might say, ah, is, is really underlies, ah, a lot of, uh, uh, you know, policy analysis, um, you know, how, how to, ah, to get at the impacts of, of, of, um, price reform, tariff reform in utilities. And it's been used in a recent article by Sean McCrea, for example, to look at electric, electric, ah, electricity demand in, in, in Colombia. OK, so that's the neoclassical toolbox. What is behavioral economics about? Well, um, it's, it's largely about inattention, um, or lack of attention to, ah, to the true price, ah, to one's true ability, ah, inattention to the future, and inattention to future circumstances by anchoring on present circumstances, so-called projection bias. And this is all from, from Chabet's recent survey. Um, so we're going to focus on these, these two, types of inattention in, in, in this talk. Um, ah, you certainly may have heard of nudges, ah, as popularized by Taylor and, ah, Sunstein's book. Um, so nudges are really about getting consumers to pay attention, ah, in large measure. Um, and nudges consist of things like changing the default option. You know, say on a 401k, you know, change, change the default contribution option. Um, uh, social heuristics, for example, um, uh, showing um Um, how your peers, uh, how much they're spending on electricity, uh, how much electricity they're using, uh, showing that on the electric bills, that's so-called social heuristics. Um, so nudges, ah, ah, as described by Taylor and Sunstein, ah, ah, constitutes sort of a normative program of improving, ah, decision making on the part of consumers. So consumers are, are quote unquote irrational, nudges move them toward rational, ah, more rational decision making. And in fact, we have a unit in, in the bank that, that focuses on, on, on finding such nudges. Um, but the point I want to make here is, ah, the critical point is that nudges are often not costless. Um, ah, to be sure, um, ah, changing the default option on a 401k may be a trivial, ah, trivial, uh, have trivial cost, but there are other kinds of nudges, like providing individually tailored information, ah, that are not costless and should be subjected to cost-benefit analysis like any other intervention that we talked about. So can we apply the insights of behavioral economics to utility policy questions? Um, and I'm gonna answer that in the affirmative, of course, but before I get to the, the answer, I want to, ah, ah, take a detour, uh, a short detour, talk about, um, methodology, um, and, and we're gonna be, I'm gonna be discussing, ah, a couple of experiments, randomized experiments, ah, involving utilities, partnering with utilities. Um. Now, why are utilities uh amenable to, to these kinds of experiments? Um, So they have a fixed customer base. So you, you, you, you're really looking, you're, you're, you're really looking purely at the intensive margin in these kinds of experiments, uh, which is good. And you are, um, you are looking at a homogeneous good like electricity or, or piped water, ah, the consumption of which is metered by the company, so you have a very, uh, precise measure of usage at a very high frequency, uh, typically. Um, so, Ah this, these, ah these conveniences were not lost on economists early on, even in as early as late 70s, ah there were ah ah experiments, randomized experiments done with electrical utilities to understand ah the response to time of use or, or ah so-called peak load pricing ah of electricity. And the idea was to understand, was to estimate ah how much consumers ah ah gained. As a result of, of this kind of, ah, you know, having these meters installed and, and, you know, was it, was it worth the cost of, of this technology. Ah, subsequently, um, and more recently, um, ah, economists have looked at real-time pricing which involves the smart grid technology. Ah, to, ah, you know, how, how that affects, ah, electricity use, um, ah, using experiments again. And, and more, more broadly, ah, in a, in a, a recent, uh, survey article, uh, or synthesis article in Science, um, Alcott and Millan Nathan have, ah, advocated for, for randomized experiments, ah, to look at energy pricing and efficiency interventions, particularly those with a behavioral, um, kind of twist to them. So with that, let me turn to the experiments that we did in Vietnam. Now when I say we, ah, all this work is joined with my colleague Tuano in the research department. Um, the electricity, ah, ah, experiment, um, asked the question, you know, what is, ah, what is the value of providing usage information in real-time, ah, when consumers are inattentive. And this is also joined with Lanfeng Lee. Um, and the work is still in progress as you'll, as you'll see. Ah, the piped water, ah, experiment asks how should Um, new utility services of which piped water in our, in our setting is relatively new. How should they be priced when consumers unwittingly form habits? And I underscore here the behavioral element of both of these uh uh studies. Um. Um, with those, with those terms, uh, as we'll see. So, ah, they're a combination of, uh, you know, sort of standard economics and some behavioral science. Um, so let's turn to the electricity experiment. And before I get into the weeds, I want to talk again about increasing block tariffs. So increasing block tariffs are um Ah, ubiquitous, um, uh, all over the world. And, ah, the reason for that is, you know, they help the utility recover their costs. They're, they're a form of progressive taxation. The rich are paying, uh, pay, pay more. Um, but, um, what isn't, ah, always recognized is that they can create a welfare distortion, uh, that is they can impose costs on consumers. When consumers face demand uncertainty. When the demand over the billing cycle for, you know, for the service is uncertain. The consumer is not sure. Ah, what price they're going to face. Um, so their, their day to day day to day decisions about how much electricity or water, uh, to use, um, is, uh, is based on the a marginal price that is uncertain. They're not sure about it. And that's going to create distortions as, as I'll discuss momentarily. Now these welfare distortions are compounded. When consumers in addition, are inattentive to their past usage, OK, and that implies, in other words, they're, they're even further misinformed about the marginal price. Um, they really can't, you know, they, they're not keeping track of, you know, where they are in the, in the increasing block tariff schedule at any point in time. That's the issue. Now, policy, uh, intervention. Can't really fix the uncertainty problem, uh, but it can potentially, um, ameliorate the, uh, inattention problem by providing information about past usage. And the question we're addressing here is whether providing such individually tailored information. Um, You know, which is costly, is, is worth it in terms of the, the, you know, the, the gain, the welfare gains. By eliminating this distortion, ah, it, it, it could potentially improve welfare, but is it, is it enough to justify the cost? So to answer this question, we, we did a, uh, uh, an experiment in uh uh urban Vietnam in the town of Ven Phuc, um, in the northern part of Vietnam. And, uh, You know, Vietnam has a, has a very fast growing residential power demand in, in urban areas and that's driven by the adoption of air conditioning. As you see this, this new air conditioning unit in, in, in Vinh Phuc is, is typical. Uh, so, um, households are adopting air conditioning. Air conditioning, you know, uses up a lot of, a lot of electricity. Um, so, ah, the, the reason we chose Vinh Phuc is because they had rolled out digital metering. Which allows us to, to implement our intervention of providing real-time usage information. Um, And um Uh, I should just say, and this is true both for the electricity, uh, work and also in, in for the piped water study later. I'll talk about later, bill payment. Default is extremely rare in Vietnam. So, so default on bills is not, is not an issue here. The experiment involved 1300 electricity customers, all of whom owned at least one air conditioning unit. In the interventions, there were two interventions that we, we randomized. The first was a price intervention where we, um, so the, the figure here, the, the green is, is the sort of the typical block tariff that, that, uh, households will, will face. Uh it's, it's actually the, the, the tariff schedule that That electricity customers will face. And we, we introduced a conservation bonus which effectively raised the price of the 2nd, uh, 2nd block price substantially. Um, and then we, ah, we had, uh, uh, uh, the other second intervention, which was cross-randomized with, with the conservation bonus was providing, uh, every couple of weeks, we provided a text message, um, giving, uh, the household it's, it's a usage for them, for that, for that billing cycle, it's cumulative usage and how close they were to reaching the 200 kilowatt-hour threshold, which got them into the higher price pool. We have 5 months of daily consumption data, including hot and cool months. Now, the role of air conditioning is critical here. Um, This is the, this is sort of the time series of daily consumption, you know, average across, across households and also of relative temperature, which, sorry, heat index, which is a combination of temperature and relative humidity. That's the green. So 80% of the day to day variation in electricity usage in our sample is explained by variation, day to day variation in the heat index, OK? So, air conditioning is really driving consumption patterns in this population. Um, Uh, uh, and, uh, demand uncertainty within the billing cycle is quite important. And so what I, what, what we have here is the coefficient of variation in daily usage within a billing within a billing cycle. So that's a measure of the uncertainty. In, in demand, and you see that particularly in the hot season. Uh, you have high, also high uncertainty. Now why is demand uncertainty important here? Let's go to the case of where there's no uncertainty, where households know exactly where their demand curve is, in fact, for the, for the billing month. So let's, let's look on the On the left side here. That's our, that's our, our picture right out of our toolbox. Um. And, um, and Uh, we have, you know, the high, high demand consumers on the high priced block, the low demand consumers on the low price block. Now let's consider uh uh there's a sort of a continuum of consumers with, with different demands. So there's a higher demand uh household, um, and their choice where their demand curve intersects the, the price line. We'll put them right exactly at the threshold. Um, for the, for the next, uh, price block. And if you increase demand a little bit more, so you have another household, they're also going to be consuming at the threshold, uh, as will this household. So there's actually a range of, of household demand that will put the household exactly at the threshold. And so what you will see in this scenario is a bunching that will occur at the threshold. A bunching of households will be consuming exactly at the their monthly consumption will be exactly at the At the threshold. And those of you who do public economics and work with notches and kinks and so on, and know exactly what I'm talking about. There's a bunch of, uh, with uncertain demand. The story is quite different. Uh, with uncertain demand, you're not sure exactly where your, where your demand is going to be, whether you're going to be facing the high price or the low price. So you have to form some expectations. So you're gonna make your decisions on a day to day basis based on your expected price, which will be somewhere in between the high and the low price, depending on how, how likely you think it is, your demand will be at the end of the day, will be above, will be at the, on the high price or the low price block. Now let's look at consumers again. There's a continuum of consumers. You have consumers with varying demands. And what will their usage be? Well, their usage will, will vary as, as the, where the expected price intersects their demand. Now, of course, the expected price will also vary across these consumers, but let's, I think we can safely ignore that for simplicity and The only point I'm trying to make is that there will be, there will be a very variation in usage across these consumers. They will not exhibit a bunching as in the previous example. OK. Uh, so that's one implication of uncertain demand, and the second implication which I, I, uh, I, I, uh talked about already is that there's going to be a distortion or cost associated with uncertain demand. So for the low, for the low demand household, the poor household, Their actual price is is is is lower than the expected price, so the price they end up facing. is the green, the green line is lower than the expected price, their expected price, and therefore they're going to underconsume, and there's a loss associated with underconsuming electricity. And by contrast, the The wealthy household, the high demand household. will use a price that is too low compared to the price they actually are facing. And so they will overconsume. But again, there's a loss associated with each of these choices. They're using the wrong quote unquote, wrong price to make their decision, so there's a distortion. Now, ah, coming back to the first point, if you look at the, the actual data we have, um, the, uh, you know, uh, the monthly usage does not bunch at the 200 kilowatt-hour threshold. So there's no evidence of bunching at all, ah, suggesting that uncertainty is, is, uh, is probably, uh, important. Um. What about um attention or inattention? I talked about inattention. Uh, I haven't shown you anything about inattention. So, let me, um Provide some evidence that consumers are, are attentive to some degree. Um, and so, um, what I want to do is look at how I want to look at what happens to behavior. As a cumulative usage for the billing month approaches the 200 kilowatt-hour threshold where the, the new price, the higher price, uh, takes, takes hold. Um, so first, uh, I want to look at how daily usage responds to higher temperature, higher heat index, OK? Of course, um, as, as temperatures go up, um, uh, to maintain the comfort level of your home, you have to use more electricity. Um, Uh, but the issue is, Um, how you change that over, over the billing cycle. So if consumers are perfectly inattentive, they have no clue what their usage is. Right? Then the response of consumption to higher temperatures should be just a flat, a flat line. There should be no change as you get closer to the threshold. We find that's not the case. We find that the response is actually attenuated as you get closer to the threshold. And here's what, here's the idea, you know, as, um, as you get close to the threshold and you realize that Your, your price is, you know, is, is starting, the price of electricity is, is, is going to be higher as you use more. Um, uh, when, when there's an increase in temperature, uh, you may want to dial back the comfort level in your home. So, uh, whereas before, you were far away from the threshold, you didn't think it, it, the price effectively was lower. So you don't adjust, you just keep the same comfort level. So that kind of adjustment. attenuates the increase in power usage that would otherwise occur when temperatures go up. Ah, so that's, ah, some evidence of, ah, forward-looking behavior and, and, and, and attention. Ah, the second piece of evidence comes back to our conservation bonus. Ah, so of course, you know, the conservation bonus effectively raised the price of electricity. If people were, um, and therefore they should, you know, people will lower their, their consumption. If, ah, if they were, ah, if they were completely inattentive to how much, ah, to how their usage was, was changing over the billing cycle. They would there would simply be a constant response, a negative response to the conservation bonus. Um, but in fact we find that the response is getting larger. As your, your usage approaches um the, the threshold where that conservation bonus is going to bite. So that's a plausible evidence that people are attentive to, to their usage. Uh, so, um, um, there's some attentiveness, uh, but we don't really know how much. We need a quantitative model to, to sort of put this, to put some numbers on this. Um. Before I get to that mall, I want to talk about the 2nd. Um, intervention that we did in this experiment. So, we randomized um this text message program. Um, and what, uh, we find in a nutshell is that, uh, on the, in the text messages were given out on, on, say, on, on an afternoon typically. What we find is on the next full day, After the text message was received by a household, the consumption of that household electricity consumption falls by 3%. But only when um on that day, the, the household is close to the 200 kilowatt hour usage threshold. So this, this suggests that um uh the SMS is actually increasing their attention, at least for that, for that day. Um, now to Let me, uh, flesh out that result of, uh, uh, uh. A little bit. Um, so, um, So this is, this is the, this is the result here. This is the coefficient on, on, on getting a text message. Um, so this is uh when, when the text message is received uh on the, on the previous day and uh you are within 40 kilowatt hours of the 200. Kilowatt hour per month threshold, then you respond. Uh, if you're away from the threshold, you know, a household doesn't respond at all. OK. Uh, now, what, what we're doing is comparing two households on the same calendar day, exact same calendar day. One household has been randomly assigned to the text group and, and, and, and received a text on that day, on the previous day, and the other household has been randomly assigned as a control group with no text message. So that's what we're comparing. Um The, uh, you know, uh, just to, to, to, uh, to check whether this is not some sort of spurious result. We, we've also look at some, we do some placebo tests briefly. Um, uh, we ask Whether it matters if you are not whether you received a text tomorrow or the day after tomorrow. Of course, this should have no effect on your, uh, consumption behavior and it doesn't. Um. So, uh, it's, it's not just being random randomized into the text group, it's actually getting the text the previous day that matters. OK. Ah, and we can do some other, ah, ah, placebo tests that, oops, that, that indicate, um, this is a real effect of the, of the text of that information. Um, Now, the question is, uh, sure, SMS may increase your attention for that day, but what is, what is the benefit to the household of, of having their attention increased? Um, so, For that, again, we need a, a quantitative model or what's sometimes called a structural model. Um, and, um, so the way we, ah, the way we do that. Um, and this is where the behavioral economics comes in, and, and blends in with the, with the traditional economics is we assume that households Make their decisions based on their perceived usage of electricity over the cumulative usage. And the perceived usage is a weighted average of their actual usage and a default or a heuristic. Say for example, There, you know, the usage is proportional to the number of billing days that have elapsed. Um, so, um, the SMS and, and, you know, the, the SMSs, uh, that we sent, the text messages, um, Uh, increase the weight that the household puts on. Their actual usage. Um, it makes actual usage more salient to them, so they increase the weight on actual usage in their decision process. So we want to estimate the attention weight in this population. Um, and, ah, the way we do this is to note that the expected price, and we talked about expected price a couple of slides ago, we now want to make that expected price, um, uh, depend on their perceived usage. Of course, when perceived usage is very high, ah, then people have a higher expected price. But the expected price will vary. Over the billing cycle along with perceived usage. Um, and so this, our quantitative model then relates daily consumption choices to the expected price each day of the billing cycle. And from that, we can um We can, ah, we can get, uh, we've, we've done some preliminary ah estimation and we get the following kinds of results. Um, first of all, The attention weight that we estimate is about 0.6, which is ah very, ah, you know, it's far below 1, which would be perfect attention. So, so households pay attention but they don't, they're far from perfectly attentive to their usage. And we can now calculate from the model or, or, or simulate from the model the value to consumers, ah, and we do this for the, for the, uh, for the hot months, for, for the high usage months. Um, we can ask, what is the value of electricity to consumers for that month. On average, and we can subtract from that the actual expenditures, not the expenditures they not what they expect to pay based on their expected price, but what they actually are paying based on where they are in the price block during, during the course of the month. And uh we can take the difference and that's sort of the net value to the household of electricity for the month. And so, with imperfect attention, which is you know what, what we estimate, um, Households, um, Uh, have a value of about $7 per month for their electricity use in, in this context. But now the interesting thing we can do is we can say, uh, let's consider a counterfactual scenario where households are perfectly attentive. So somehow we make households perfectly attentive. We increase their attention weight to one. Now they're making better decisions, more informed decisions about their electricity on a day to day basis, and we can compute the value of this electricity to them net of its actual cost. We find it's, it's a little bit higher as, as you'd expect, but, but only a little bit, only about 10 cents a month. So the benefit of being perfectly attentive for the whole month is only about 10 cents a month. Again, these are preliminary results. I don't want to oversell them. Uh, so, um, Ah, uh, of course, the SMS that we sent, the intervention that we, we did in the experiment is only raising attention for perhaps a day. So the benefit of being attended for the whole month is only 10 cents. So the, you know, the cost of the, of the SMS it better be really low to justify it as a, as a policy option, um, at least based on our preliminary estimates. So Increasing block tariffs do create distortions. Part of the distortion is due to inattention. Policy can only ah affect ah the inattention um ah through, through information, um, but since the cost of inattention is small, the welfare gains from providing ah this usage information is also ah ah likely to be small. Um, so looking forward, um, so that's, ah, that summarizes the electricity experiment thus far. Um, and looking forward in this sort of metering space, there are, there are many research opportunities, um, out there, ah, involving new technologies for metering. Um, Jack and Smith have recently published, uh, an article looking at prepaid, uh, metering in South Africa. Where there's high default and they, they find that it reduces default on electricity bills, ah, which, um, which is not a problem, as I mentioned in our context, ah, but it's certainly a problem in many, many countries. Um, and, ah, with the rollout of smart grid and real-time pricing. Um, where also you could have, um, meters in the home showing usage over the billing cycle. Um, you know, these have, ah, been looked at in, in the US and developed country context, but a lot more work, ah, ah, is needed to understand the implications of these technologies in developing countries. So with that, let me turn to the um To the pricing, uh, water pricing experiment. So, uh, the The big picture here. Is access and cost recovery. Um, Uh OK. Ah, as we discussed at the very, uh, at the very outset of the talk, um, you know, the decision to provide utility services, uh, to an area, say, ah, you know, is based on, traditionally is based on a cost-benefit analysis that compares willingness to pay with the cost of connection. So willingness to pay is a critical component of, um, uh, uh, of assessing whether, uh, access should be provided. Um. Similarly, ah, ah, willingness to pay is a critical element in, in, ah, in thinking about, um, optimal pricing. In particularly the optimal two-part tariff. So optimal two-part tariff involves, um, you know, what price do you charge for the service versus uh the the fee for connecting to the utility. OK. And there's a trade-off between, between these, these two prices, the, the two parts of the two-part tariff that involve a trade-off. Um, and Auerbach and Pleo many years ago showed that you want to charge a price above marginal cost. So marginal cost of say electricity generation and distribution. You want to charge a higher price than that to consumers and you use the, the proceeds to reduce their connection fee to allow poorer households or low, low willingness to pay households to, to connect to the service. Um, so that's, that's the basic, ah, trade-off, ah, with, with a two-part tariff. Now, so willingness to pay again is, is a critical component. Now, what if, ah, ah, household preferences are not fixed over time? In other words, what if households, um, ah, it takes some time for households to get used to using the, the service, ah, or, or they, they, they have to grow to, to like it and value it. In other words, what happens if there's, if there's habit formation? Well, ah, then the question of when to, ah, assess willingness to pay becomes, uh, salient. Ah, do you use the willingness to pay elicited before habits are formed or the willingness to pay. For the service after habits are formed. Uh, and so that's what, uh, that's the issue we want to address in, in the, in the piped water experiment. Uh, the setting now is, uh, rural Vietnam. Uh, a relatively prosperous commune in, in the Red River Delta. Um, now, prior to, um, having piped water, uh, available, um, households, um, generally used, uh, rainwater that they collected in tanks, and they use the the water for drinking and cooking after boiling it and also for showers. So piped water, uh, uh, you know, uh, Provides really convenience and uh pressure, higher pressure for things like showers and, and, and uh uh and washing and so on. Um, but it has not fully displaced rainwater in our setting. So households still have these concrete tanks above their, above their homes to collect rainwater. Now, we partnered with a private water utility ah that, that acted like a, a private company but was subject to government rate regulation. And they provided us with cost information, which we think is reasonably accurate, so we're gonna use that a little later. Um, but the, the key point is that this utility only came into being, um, only started operating, I should say, um, uh, 3 or 4 years before we started the experiment and households were, were being connected, uh, throughout this period. So, um, this is a relatively new, uh, thing in, in, in, uh, this commune, piped water. Um, so our experiment involved 1500 water customers, um, followed over about a three-year period. Um, And we divided them into three groups that ran randomly, and each group got a series of price discounts on their piped water, um, starting with a 50% discount, then a 75% discount on the price, and then They reverted back to no discount. Um, there was always a group, um, at any point in time that had no discount at all, so it was like a control group, and these discounts lasted 6 months. Um, every time there was a, a, a, a, a new discount was applied, ah, the households were informed in a, in a couple of different ways beforehand. Ah, to make the, um, uh, the price change salient to them. And of course, we have monthly metered consumption data. Um, before I get into, to the results, um, ah, there's a good evidence of habit formation in utility services, but it comes from, ah, basically electricity, ah, studies. Um, there, there are, um, Several well-identified, very carefully done studies showing persistent responses to price changes and conservation programs. So for example, after a conservation program has expired and, and, you know, the incentives have gone back to normal, households continue to conserve energy, um, and, and to behave as though, you know, they, they've changed their behavior, apparently. Ah, so there's good evidence of this in, in, in developed and developing countries. Um, so what about piped water customers in, in our, in our commune? Did they, um, Do they form habits? Well, the way to think about this is to take our intensive margin demand, demand picture, um, and look at two consecutive months of consumption. Um, in the, in the first month, um, in the lower panel, we're going to give a subsidy to consumers and they're increasing, of course, they moved down their demand curve and increase their water usage. Um, now, what happens is, as, uh, they form habits, um, and this makes, ah, this shifts the demand out in the subsequent month. So the current month now, the demand is shifted out, which means that at any price, uh, at any current price, Uh, consumption usage is, is, has gone up. OK. Do we find evidence that this is actually what's happening in our experiment? And the answer is yes. Um, if you look at, uh, monthly, ah, usage, it's, um, it's negatively associated with the price, ah, in the current month, contemporaneous price. Um, and it's also negatively associated with the previous month's price. So that's kind of the reverse of this diagram as, you know, price goes up in, in the previous month, consumption goes down in the previous months, there's less habits being formed and lower demand. Um, now, Are consumers aware of this process of habit formation? Um, and this is, this is sort of where the behavioral economics enters the picture. So do consumers internalize The future effect of today's piped water consumption. So, you know, when, just to make it concrete, when the When, when the household members is taking their shower with a high pressure, do they think about how, you know, having a longer shower will increase the, um, you know, the desire to take long showers in the, in the future, ah, or do they form their habits unwillingly, naively assuming unchanging future preferences. So, um, You know, intuitively, you know, our intuition tells us that it's implausible. The scenario where you're taking a shower in order to, a longer shower in order to, ah, you know, build habits for, you know, to, to, to enjoy showers more in the future seems highly implausible. And indeed, um, the lack of awareness of, of habit formation is consistent with evidence of projection bias. Um. Ah, in various, ah, ah, ah, ah, studies, ah, and in particular, ah, ah, in recent experiments which have, ah, ah, been very carefully designed and show, ah, ah, a strong evidence that, ah, consumers lack sophistication in, in their habit formation. And therefore, ah, ah, are basically unaware that this process is, is happening to them. Ah, ah, and it's also consistent with projection bias, ah, idea is consistent with Our own data, ah, our own experimental data. Ah, so let's, let's just take it as a given that, that households, ah, exhibit, uh, this projection bias. They're not aware of their, of their habit formation. What are the implications? The first implication is that there's a long-run demand curve ah that incorporates habit formation. And that's distinct from the short-run intensive margin demand curve, which fixes habits at, at, as constant. Um, so again, let's go back to our, uh, uh, diagram from the previous, uh, the previous diagram. Uh, we're now going to consider a permanent reduction in the price of water. Um, of course, uh, that will increase usage in previous months, uh, which will build habits, increase demand, um, shift demand out, I should say. And, um, and if you, uh, now connect the dots, what you get is the long-run demand curve. Um, so, the purple line is, is longer demand curve, um, and you can see that there's a greater response to the price on account of habit formation. So, the long-run price elasticity, uh, the demand elasticity is greater in the long run than the short run. And our estimates, which again are based on a quantitative model, um, our estimates suggest that the long-run price elasticity of demand is almost 3 times greater than the short-run elasticity. Now, both elasticities are still relatively small, but the long-run elasticity is considerably greater than short. So, uh, this diagram. Is, is, uh, is, is being reflected in our, in our results, our estimations. Um, now, why is that important? We need to know the long-run demand elasticity to talk about optimal pricing policy. That's what the uh utility faces, the long-run demand curve. Ah, the second implication of projection bias is, uh, has to do with willingness to pay, which, as I said, is a, is a critical, uh, uh, issue for policy. The willingness to pay, uh, you can think about an ex-ante willingness to pay. And by ex ante, I mean prior to connection to the utility. There's a, there's hypothetically, there's a demand curve you can think about before any habits have been formed to the area, uh, the area under that demand is, is the ex ante willingness to pay. Now, this is a hypothetical, but you, you can certainly elicit the willingness to pay uh before anybody has connected to a utility. That's, that's a a a completely feasible thing to do and that's exactly what the Lee et al. uh study, which I, I cited at the very beginning, that's exactly what they do in rural Kenya. They elicit the, the ex ante willingness to pay for an electricity connection. Um, so, the point I want to make here is that the, uh, long-run demand is, it looks like this. And the ex post willingness to pay, the ex post willingness to pay which Accounts for habit formation. Is by by necessity greater than the ex ante willingness to pay. OK. And the idea here, this is an implication of projection bias. And why is that? Because if I, if I could anticipate my habit formation, Perfectly, um, then when you elicited my ex ante willingness to pay, I would just, I would just tell you it's, it's, it's the expos willingness to pay. Because I understand that I'm going to be accumulating habits as I use, uh, as I use the service. Um, so, um, Uh, the, the idea then is that projection bias creates a wedge between the, your ex-post and your ex-ante willingness to pay. And in fact, we can estimate the magnitude of this wedge and it's, it's quite large. We find that the ex-post willingness to pay for piped water is about more than 3 times greater than the uh ex ante. Ex-post is greater than the ex ante by a factor of 3 or more. So, and again, the decision to connect is based on ex ante willingness to pay. So let's now put the Hanan, can I just ask you, uh, you're kind of running out of time. So if you have, if you can wrap up in as few minutes as you, as possible, yeah, I, I, I have really two slides left, right, um, so, um, putting these, uh, ideas into practice, we can come back to the question of, uh, the two-part tariff. Again, the question is, Ah, how do you set prices, particularly, ah, ah, the price and the connection fee to, ah, most efficiently recover, ah, your fixed costs as a utility when consumers form habits. And the basic implication of, of habit formation and, and projection bias is that you want to defer the consumer's payments until they've formed habits. Um, So in other words, you want to tax consumers' future selves. Um, to, uh, in order to subsidize their present selves. Their present selves are the ones who are making the connection decision. The future selves are the ones who have habits already formed. That's the idea. Now, the way, you know, there are two ways to do this. The first is to simply charge a markup above marginal cost. That is you charge a price for the service above its marginal cost. And you use the proceeds to lower the connection fee. So, um, so that's one way. But a markup has a downside which is it creates, uh, um, it's like a tax, it creates a distortion. So you want to, uh, ideally like to eliminate that distortion and you can do so. By charging instead of a zero markup and charging a monthly fee, a recurrent fee. To use the service. A bit like a, a mobile phone plan where you, you have unlimited usage. Now, of course, with mobile phones, ah, the marginal cost is basically zero, so you can, unlimited usage like a 00 price. But the idea is, is to have zero markup and, and a monthly fee. And if you do that in our setting, um, We can show that the welfare gain, given the cost of the utility, that given the fixed cost that our water utility needs to recover. Um, the going to a recurrent fee with zero markup, uh, is, is, is, uh, is yielding a 12% welfare gain over a markup and a connection fee, uh, pricing model. Ah, so with that, let me, the main takeaway I think from this. Analysis is that behavioral economics tells us it matters when willingness to pay is elicited. If there were no projection bias, um, in, in a, in a, in a setting where habits are being formed, it wouldn't make any difference when you elicited willingness to pay. But if there's projection bias, which behavioral economics suggests is important, Um, then, ah, it makes a big difference potentially. Cause if you do a cost-benefit analysis based on ex ante willingness to pay, you may reject a socially desirable project, like as in this, uh, diagram here where the average cost is strictly above the, the demand. Ah, but if in fact ex post willingness to pay were, were much greater, ah, the project could be actually a desirable one from, from society's point of view. And you're, you're, you're not doing that. So, so that's a cost to, to society. Um, So let me conclude then, ah, giving you a couple of examples of behavioral economics enriching our econ 10 1 toolkit and providing insights that allow better policymaking. Um, and I would suggest a formula for, for future progress is, is a, is a combination of economic theory, behavioral science, and experimental methods. With that, I thank you for your attention. Thanks, Annan. Um, uh, let's turn it over to Richard. Um, just as a reminder, um, if you'd like to ask a question, please just put your name in the chat and I can call on you when the time comes. Uh, thank you. Richard, over to you. Thank you so much, Hernan. Uh, thank you, Dion. Uh, firstly, let me say that I'm certainly honored to do, to be asked to comment on this. This is first rate work of immense relevance to what we do in the bank, and of course the quality is beyond reproach. Little confession, um, when Hannan first asked me whether will I comment on this work, I was really enthusiastic because every time you get to read something that Hannan does, you learn a tremendous amount that's new. Then I suddenly learned it was about behavioral economics that I know nothing about. So, in advance, my apologies if what I say is less relevant than it ought to be, but much of what we've heard is standard neoclassical economics. So hopefully, I have something. Um, that might be somewhat relevant to share, uh, to, to suggest. So, let me start with a bunch of sort of small little comments in terms of both papers, what we've heard, and perhaps some issues that might be worth exploring later on as, uh, the team go on to work on this. So, um, sorry, let me get this full screen. OK. On the first paper that we had, I think one of the notable aspects is that, of course, you've managed to get experiments, which is very rare to do in these sorts of circumstances, especially getting utilities to agree either to share data or do experiments, tremendous work. So kudos to the team for actually having got there. Now in terms of IBT's increasing block tariff, Uh, the key finding, if, uh, if I may oversimplify and summarize it thus, is that consumers don't respond much to the higher price bracket, and they don't respond even when information is actually provided. Therefore, the welfare benefit that you discover is reasonably low. It goes from $7 to $7.10 a month, which by any standards does seem to be quite low. So the question then must be, why is it that you actually find that? And they seem to me to be two possibilities. One possibility might be that the level of prices is so low that consumers don't really care. So, you know, an issue that I didn't see you exploring and perhaps you might want to explore is what proportion of income the electricity bill comprises for the average consumer in the, in, in the higher range. And the lower range and how might this compare to say prices and proportional prices in other countries, because clearly you'd get this result of prices for electricity of the utility prices are so low that consumers simply don't regard it as important. So when you send them an SMS message, it's really not important enough for them to worry about this. In other words, the benefits don't outweigh the cognitive costs of either looking at the message or of changing behavior. There's another possible explanation, and we do get a kind of a partial sense of whether this is or isn't at play, is that it may well be when you're talking about the two blocks, the differential between the two blocks is too low. If the differential between the two blocks is too low, then consumers don't really care very much if I go from 200 kilowatts from one block to the other. And one way of trying to, I guess, unearth these differences, and we did see a couple of graphs on this, is to try to infer what the elasticity of usage with respect to heat is across the two blocks, and we did get a graph, but we got it in terms of eyeballing. It would be good to see if there are statistical differences there. Another important thing perhaps to look at is whether there are any discernible statistical differences in the consumers who are on the lower block relative to those that are in the upper block. Clearly, you'll get differences in consumption levels, but will there, are there any other differences that you can work out? Are there differences in income? Are there differences in size of households? And it seems to me that unless we know why we are seeing this behavior, it's going to be somewhat hard to dispense with policy advice, um, that we can be confident about. A general finding though, I, I certainly believe, and it's a finding that, that, that, that you read about in many different circumstances and certainly in terms of a lot of the water utilities, there's been work done by Dale Whittington for a number of organizations where he finds the same thing when it comes to water pricing and in IBTs. Generally, consumers seem to respond to some kind of an average price and not a marginal price. And one of the explanations that's, that's given is rather than this behavioral economics one, is that consumers respond to average prices because they understand that, and that you can do household budgeting quite well when dealing with actual average prices. Rather than having to worry about these marginal prices that entail cognitive problems that are difficult to know when you are or are not at a threshold. Another issue that people seem to worry about when talking about IBTs, and again you have the data to be able to answer this question, is the distributional implication. Whether it's water or electricity, what we know is that poorer households tend to be more numerous. There are more people living together in a single household. Now there's more people, no matter what you do, consumption levels in the aggregate, if not per capita, will be higher. So you're more likely to cross that threshold and therefore pay more. So it could be as contrary to what policymakers think. is that IBTs may end up being regressive rather than progressive. And it is alleged that the, the evidence is not terribly strong, but it, it is alleged that this is certainly what you tend to see in terms of water. So it'd be great to see if you actually find this also in terms of the Vietnam example. So all in all, I think we are converging to the point. That IBTs may not really be as beneficial as is so widely claimed, even though they're widely used both for reasons of equity and it is argued also for reasons of efficiency, but they may not be curbing demand in the ways that we might be wanting them to. Let me then very quickly move to make a couple of quite obvious points when it comes to the second paper that you've done in terms of habit formation and demand. And for our, for all our listeners, let me oversimplify the result again and simplify it in the following ways. So what you find is that water usage in this particular context is like an experienced good, therefore, it is habit forming. The more you use the good, the more your demand is likely to increase, obviously up to a certain limit. And therefore, if you face an experience good, perhaps one way of pricing is to eliminate the entry fee or to reduce it. And then to subsidize that in in in subsequent periods with some kind of marginal cost pricing and also to have a low price in earlier periods so you can begin to build demand on the intensive margin and perhaps in later periods, you might be able to increase the price if that is feasible. And this really seems to me to be quite realistic and an accurate description of the case that you have found. But of course, as a matter of curiosity, it might also be interesting to ponder why it is that very often when it comes to what seem to be experienced goods, we also observe the reverse, that is, we find a very, very high entry fee and sometimes even a zero marginal cost for users. Many of you will be familiar with or might have read or seen a reference to this famous paper. I think it came out in the QJE sometime in the 70s, all the Mickey Mouse Monopoly written by, by Walter Oy. And here he was worrying about why is it that Disney World and all of these entertainment places, by the way, I don't know if it's true. I've never been to any of them. Apparently, they charge you a very high entry fee to get into Disney World. And zero cost at zero marginal cost for any of the rides. So the rides are all free. And the question then becomes, as clearly the rides are an experience good. You have one ride, you have tremendous amounts of fun, you have the 2nd, the 3rd, and the 4th, and surely you should be charging the marginal cost rather than an entry fee. We observe something similar, though not identical, when it comes to Amazon Prime. And here you can check on the internet. What you do is you pay a lump sum fee to get in, and there is, it is claimed 100 million goods that will be shipped to you within 48 hours, all for free. So why is it that we have this $150 or whatever it is that we actually pay as an entry fee, but nothing there thereafter? And some fancy gyms also have this sort of thing where you have a very high entry fee, but zero marginal costs, zero monthly payment thereafter. And of course, there are some phone plans as well that go one way and some the other way. You know, this is just purely conjecture on my part as to why we see this in some cases and not others. It may well be, as Walter Roy has conjectured, for example, that what you want is to lower the marginal cost. If you lower the marginal cost of usage, then you maximize consumer surplus and you can extract more entry fee, particularly when you've got heterogeneity of customers. That might be one reason that you have it. Another reason that you might, might, might have this different kind of pricing, of course, might occur when you have problems of congestion, signaling snob value or something or other. So, in other words, what I'm trying to say over here is to make for a more complete paper. It would be nice to See when you would have one sort of pricing of the kind that you are recommending for Vietnam, but yet in much of the real world, we see different types of pricing and why we're actually observing that even though we're talking about experienced goods in both places. And just, just in terms of a small little warning before I end off, um, in the paper that's shared with the PowerPoint. There's talk about uh pipe water as being a merit good, it's been wonderful. Perhaps in some ways it is, but in other ways, it is not because there is an environmental burden. And many of you will know that are on this call, I see the sectors that you're from, is that very often the largest consumer of energy in a city tends to be the water utility. So the energy burden is tremendous, and of course, we want to minimize water usage for environmental reasons. So unless there are water quality benefits, I wouldn't be over relying on the merit good argument. Then I saw at the end that you do cover this notion of willingness to pay, the experts willingness to pay does cover costs, and my apologies, I did not pick that up earlier. But that, let me end, stop sharing, and it's over to you, Dion and Hannan, and again, congratulations on a great piece of work. Thanks, Richard, for those insightful comments and, and, and, you know, since you did ask a couple of pointed questions, maybe before I turn to the audience, and again, please just put your name in the chat and I can call on you. Hanan, do you want to have a crack at sort of um responding to some of those issues that, that Richard raised? You're muted, Hannan. All right. Uh, I'll just respond. Can you hear me now? Yes, I can hear you. Yeah, sorry. Um, so just 33, points, uh, uh, respond to the, uh, Issue about the, yeah, the, the percentage of the share of electricity in the household budget. I don't have the exact uh uh figure, but uh I think, uh, I think in this setting, it's, it's non-trivial. Um, uh, I would say, uh, certainly greater than the share of water expenditures in, in our rural sample, which is, is much smaller. Um, so, electricity is more, is more expensive and um And so it, it, it is a, a non-trivial fraction, uh, although I don't have the exact number in front of me. Um, on the question, the issue of um using average uh price, uh, to make decisions, um, uh, I mean, I, uh, uh, uh, uh, this has been, this has been argued, uh, that, that consumers base their uh their, their electricity choices on, uh, Average prices from past bills. Um, in effect, our, our, our, our conceptual framework, um, is, is really, is really an average price framework because as I, as I showed you in that, in that graph, people are basing their decisions on expected price. Um, now, although expected price is a forward-looking concept, um, It is an average of prices. So, what we're providing, I think in this study is, is, is kind of a, a microfoundation if you will, of, uh, of the idea that households base their choices on some average price. Um, it's just, it's not an average from past bills, it's a forward-looking average. Um, So, I, I think, uh, uh, we're sympathetic to the, to the idea that consumers are behaving in this way. And that's in fact, part of, you know, part of our story. Um, and the third point about the Mickey Mouse monopoly story, um, And by the way, I, uh, Walter Oy was a, was a former colleague of mine, so I, I know him quite well. Um, no, this is different. Our, our, our, our setting is different than the Mickey Mouse Monopoly story because, um, uh, the, the utility, um, is facing a budget constraint, um, or, or, or let's say the government or social planner who's deciding on the optimal price structure faces a budget constraint. Um, they have to cover, they have to cover fixed costs. And that changes the story a lot. I mean, yes, if you, if you're just a standard profit maximizing monopoly, ah, you would, you would want to set the price to marginal cost and extract all the surplus with, with the connection fee. But that's, ah, that, uh, um, but once you get to have a budget constraint, you have to meet, um, then the, there's a trade-off between setting a connection fee too high and setting a price. Uh, so, um, so it's a very, it's a very different setting theoretically. That's all I wanted to say. Thank you. Thank you for your comments, Richard. OK, thanks, Hannan. And, and Richard, maybe towards the end of the session, we'll come back to you and maybe you can pick up even on some of the questions from the audience and maybe if you wanted to respond to anything that Hannan said, that'd be great. Um, so we have 4 people who, who have kind of raised their hand at the moment. Uh, let me just go in order, and if you don't mind sort of unmuting yourself and asking a question and putting on your camera if you feel comfortable doing so. Uh, Luce, are you there? Luz Carrao, are you there? He said, he says his mic is not working. OK. I will then ask the question on behalf of this, um, which is, do you think, uh, you can create some habits with this SMS? For example, the first month they are paying attention for one day after and maybe after 3 months, they, they understand better and the fact is last longer, uh, lasting longer, longer lasting. Um, OK. Uh, the second question came from, uh, Bertha. Are you there? OK. It seems like we're unable to. Hear people. That's unfortunate. Um, OK, so let me just ask the question then on behalf of Bertha. Sorry about this, we'll fix this next time. Uh, please, how did you obtain 2.7 to calculate the long-run elasticity and the 3.25 for the experts willingness to pay? I presume it's sort of elaborated in the papers, so maybe there's some technical detail to provide, but if you could kind of maybe give some intuition, that might be helpful. Um, And then Hector, I'll give you 1 2nd. If you can't unmute, I'll read your question. Uh, so essentially my question is with regard to the role of loss aversion. Of public utilities, especially with regard to the willingness to pay and essentially this difference between the ex ante willingness to pay and the Exposed willingness to pay and uh whether this would be something interesting to uh look in complement of what has already been done now in the work of uh um uh Hannan beer in the context of Vietnam. OK. And then last is Javier. Are you able to unmute? Yes, uh, hi, uh, thank you for the presentation. My question is, uh, regarding, uh, systems in which you have to prepay, uh, to consume, uh, and how the behavior of the consumer would be different. I assume they would be more, um, aware of what the price and how long the, the recharges last. Uh, I don't know if you found this type of systems, uh, uh, in your research. OK, Hannan, over to you. Just again, we, we might have a little bit more time, so if you feel like asking a question, please put your name in the, in the chat. OK. Uh, OK, um. Well, I, um, uh, uh, uh, the first question was, um, do people, um, I, I guess I could rephrase it to say, uh, um, might people respond to the text messages about their usage, um, differently, uh, as they become accustomed to receiving texts about their usage. Um, so there's sort of dynamic effects, um. Uh, and, uh, uh, we haven't looked at that, uh, yet. It's, it's, um, It's certainly possible. It's certainly possible. I'm not sure how Yeah, I'm not sure how we would work that in, but, um, but we, we can certainly, we can certainly have a look at that. Uh, the second. question was how did I get the 2.7 and the 3.25. Um, yeah, so, Um, So As I said, we had a quantitative or structural model of demand, um, that allowed for, uh, dynamics. In other words, it allowed for your past consumption to affect your current consumption, as well as, you know, the current price of past consumption affected your current consumption. And when we estimated that model and, and from, from those estimates, we could back out these numbers, this 2.7 and the 3.25 as along with the short-run elasticity. Um, so that, so that's in a nutshell how we got, how we got those numbers. Uh, all the details are laid out in the, in the paper which um Which I've referenced, uh, the, the policy research working paper that I referenced. Um The question about loss aversion. So, the idea, I, I, I think is that uh when you have loss aversion, and I try to elicit your willingness to pay, I, I, I'm gonna get a different answer if I ask you, uh, you know, before I give you the, the, the, whatever it is, um, you know, let's say it's a, it's a mug, you know, in these experiments. Uh, they, they give, they ask people, how much are you willing to pay for the mug before they give them the mug and they get an answer and then Then they give them the mug and then they, people start drinking in the mug and then they ask them, well, how much are you willing to pay uh to, um, to uh get rid of them, I mean, to, to give me the mug back. Uh, how much do I have to pay you that is. And there's a, there's a gap between those numbers typically, people, um, people are averse to losing something that they've possessed. Uh, so that's an issue with eliciting willingness to pay. Um, it's, uh, a somewhat, uh, it's, it's distinct from the issue we're talking about. We're not really talking about uh eliciting willingness to pay before and after connection and comparing. We're, we're, uh, what we're saying is that um Ah, there's a, theoretically or conceptually, there's, there's, there's a, there's a gap between expost and ex-ante willingness to pay that arises from habit formation. Um, So I think the loss aversion issue was more, if you wanted to go out and estimate willingness to pay ex post and ex ante, you would have to contend with loss aversion as a potential explanation for a gap, but that's not what we're doing in this, in this work. We're not eliciting willingness to pay directly in this manner. Uh, so I hope that answers that question. Um, and the question about prepaying, uh, I'm not really, uh, sure I follow the question. Uh, you know, the, the metering, the, the, the pricing here is all, you know, sort of pay as you go, uh, uh, you know, after, at the end of the billing cycle, uh, you pay the water company or the electricity company. Um, Now there have been, as I mentioned, there have been experiments, um, uh, not experiments, but there have been studies of a prepaid metering. Uh, so you pay in advance, you get a certain number of minutes, just like a, you know, phone card, and, you know, you get to use that many minutes. And if you want more minutes, you go and get some more. Um, and that's been shown to reduce bill default. Obviously, it's just going to reduce default because you're paying in advance. Um, But there, you know, aside from that, there hasn't, there hasn't been any, any work uh with this kind of uh metering. So, I, I'm not sure exactly uh if I answered that question, but that's all I can really tell you about prepaid metering. Thank you for those questions. OK, thanks. Nobody's put their name in the chat. So let me, just, let me ask something about that occurred to me early on in your talk, Hanan, which is, I was having trouble kind of disentangling and, and, and wondering how you separately identify the notion of Of, of, of, of certainty. Of expectations and certainty with attentiveness, because it seems like the two are kind of intricately linked. Uh, uh, in, in, I mean, and your experiment increased attentiveness, but surely when you're increasing attentiveness, you're also changing. The certainty or the expectation around the price. And I just wonder whether you, you actually do disentangle the two and maybe I just missed it, or, or whether you, you don't really care, but, but you're kind of getting the net effect ultimately. OK. Um, Well, uh, uh, let me describe, uh, what we do empirically, ah, just sort of in, in very simple terms. What we're doing is relating Uh, your, your daily consumption. To your past usage. I mean, I mean, think of what we're doing as a a very elaborate. Nonlinear, highly non-linear regression of current consumption on your past usage, OK? And the, the argument is, yeah, so, so how the um How your past usage affects your consumption and how that evolved, how that relationship evolves over time. He's telling us something about your attentiveness because, um, because it's, you know, how, how much do you use your, your past consumption information in your decisions. That's really what it's, it's teasing out. Uh, the uncertainty part, I mean, is, is, is in the model, so to speak. It's, it's in there, but I mean, we're not, we're not trying to, to, to get at that. But, but the point is that um the, the, the part about attention is being quote unquote, identified from the relationship between current consumption and past usage. I hope that It helps. I think so, but I, I, I know where to find you, so I can follow, right, right. I mean, I, I, I apologize, but these are conceptually quite difficult, um, uh, models to, to, to work with, uh, and intuition can be. Can be Difficult. OK. Fair enough. Um, so with that, um, nobody's put their name. Uh, Richard, I'll, I'll start wrapping up then. Um, Richard, I just wanted to give you a chance to, to come in if there was anything you wanted to add to the discussion before I turn it back to Hannan just to close. Just, just a small little point that I think is unresolvable. So I, I think your, I, I think your results, uh, Uh, absolutely plausible, right, especially on the water utility one that with the habit forming and that demand goes up. Um, and also this notion that you're therefore willingness to pay, pay will increase, but we have no way of eliciting that before we decide whether to build a utility or to build the pipes or not, I don't think. So, um, you know, I, I, I'm pretty sure some of our colleagues will sort of begin to worry about that. Yes, this, we believe that, but how, how can we actually infer to what extent will demand actually increase? I, I just thought I'd, I'd throw that in, you know, while the positive conclusion, I have absolutely no trouble with. Um, we don't seem to have a way of finding out what the future will behold. So let me end with that. That's, that's exactly, exactly my point. Uh, exactly my point. We, we have methods, sorry to, to break in here, but we have methods of, uh, uh, of, um, uh, eliciting willingness to pay before people have connected. I mean that, that was used in that Lee et al study, uh, to get at the willingness to pay for, for electricity in rural Kenya. Um, but the issue that you raise is that You know, we have no way of knowing what people's expect, you know, how people forecast their future demand, and perhaps, and if they are, uh, you know, if they're unaware of their habit formation, they're going to give a willingness to pay that differs from, uh, from what you would, you would find ex post after, after pitch. So, yes, uh. I agree. OK, Hannan, if you want, uh, in addition to that, if there's any last comment you just want to make, otherwise, we'll, we'll wrap it up here. I think Ryan has just given a message to everybody that every, all the materials will be online and he's given the link in the chat. Uh, Hannan, anything, any last? Thank you all for your attention. Appreciate it. Well, thanks again everyone for coming and with that, I'll close the session and, uh, we will not be having, uh, policy research talk in January. We're gonna pick up the series again in February, so look out for an invitation to that one kind of early February. And that, thanks everyone. Goodbye. Thank you. Thanks a lot. Thanks, Richard. Thanks.

showAllTranscripts

duration

PT1H23M23S

scene7File

worldbank/BehavioralEconomicsPublicUtilities

scene7Domain

https://worldbank.scene7.com/

scene7FileAvs

worldbank/BehavioralEconomicsPublicUtilities-AVS

title

BehavioralEconomicsPublicUtilities

description

BehavioralEconomicsPublicUtilities

showTimestampAndTranscript

yes

sec-spacing

section-none

sec-spacing-bottom

section-none

sec-swoosh

sec-swoosh-type16

style

default, transparent-white-components, heading-center, add-style, bg-primary-blue-70-gradient, bg-dynamic-height

col-xs-12

col-sm-12

col-md-3

col-lg-3

col-xs-12

col-sm-12

col-md-7

col-lg-7

add-style
lp-body-content

The provision and pricing of water and electricity services are perennial policy concerns in developing countries. In this talk delivered on December 8, 2020, World Bank economist Hanan Jacoby helped bridge the gap between the cognitive limitations of consumers flagged by behavioral economics and the policy questions faced by public utilities the world over. The talk drew on recent research in Vietnam, where unique experiments were undertaken in partnership with both a water and an electricity service provider. The electricity experiment allows a study of consumer inattention and the value of providing real-time usage information, while the piped water experiment addresses how new utility services should be priced when consumers unwittingly form habits.

lp-heading-top-medium

lp-heading-bottom-medium

col-xs-12

col-sm-12

col-md-2

col-lg-2

sec-spacing

section-none

sec-spacing-bottom

section-padding-between

style

default, video-transcript

everything

https://webapi.worldbank.org/aemsite/everything/search

true

{"facets":[],"orderby":"contentDate desc","select":"*","searchFields":"*","search":"*","filter":"contentType eq 'Video'","top":20,"skip":0}

See Rest of Series

yes

sec-spacing

section-none

sec-spacing-bottom

section-padding-between

style

default, bg-neutrals-15-gradient, remove-border, lp-mini-card-primary

The Implications of Behavioral Economics for Public Utility Policies

RELATED

See Rest of Series