1 00:00:00,040 --> 00:00:02,460 The following content is provided under a Creative 2 00:00:02,460 --> 00:00:03,970 Commons license. 3 00:00:03,970 --> 00:00:06,910 Your support will help MIT OpenCourseWare continue to 4 00:00:06,910 --> 00:00:10,660 offer high quality educational resources for free. 5 00:00:10,660 --> 00:00:13,460 To make a donation or view additional materials from 6 00:00:13,460 --> 00:00:17,390 hundreds of MIT courses, visit MIT OpenCourseWare at 7 00:00:17,390 --> 00:00:18,640 ocw.mit.edu. 8 00:00:26,370 --> 00:00:32,330 PROFESSOR: So I think we have a bit of background in mind. 9 00:00:32,330 --> 00:00:37,570 So now I'm going to summarize this discussion, and start 10 00:00:37,570 --> 00:00:41,580 asking about how do we go about answering some of the 11 00:00:41,580 --> 00:00:45,450 questions that have come about? 12 00:00:45,450 --> 00:00:51,910 So at the end of last lecture, we were left with our S-shaped 13 00:00:51,910 --> 00:00:54,560 curve, and saying that whether or not there would be a 14 00:00:54,560 --> 00:00:58,190 poverty trap was really dependent on the shape of the 15 00:00:58,190 --> 00:01:01,910 curve that relates income today to income tomorrow. 16 00:01:01,910 --> 00:01:05,340 And we gave sent out examples where it could go one way or 17 00:01:05,340 --> 00:01:08,400 the other, or it could be that its S shape, it could be that 18 00:01:08,400 --> 00:01:11,160 it's increasing but inverted L-shape. 19 00:01:11,160 --> 00:01:14,220 And then one of you asked the question, 20 00:01:14,220 --> 00:01:16,860 well, how do we know? 21 00:01:16,860 --> 00:01:19,830 How do we even get about-- 22 00:01:19,830 --> 00:01:23,560 how do we get to know the shape of the curve? 23 00:01:23,560 --> 00:01:26,440 This is not a question that is going to be solved 24 00:01:26,440 --> 00:01:27,280 theoretically. 25 00:01:27,280 --> 00:01:30,310 The debate we just had a moment ago suggested it won't 26 00:01:30,310 --> 00:01:32,680 be solved theoretically. 27 00:01:32,680 --> 00:01:36,760 Because, for example, take the question of whether if I give 28 00:01:36,760 --> 00:01:40,930 you a bed net for free today, you're less likely to buy the 29 00:01:40,930 --> 00:01:42,190 next one tomorrow. 30 00:01:42,190 --> 00:01:45,350 It depends on whether the entitlement effect is more 31 00:01:45,350 --> 00:01:47,660 important than the loaning effect. 32 00:01:47,660 --> 00:01:48,720 And how will I know that? 33 00:01:48,720 --> 00:01:52,200 There is no way for me to know it without 34 00:01:52,200 --> 00:01:54,160 looking at the data. 35 00:01:54,160 --> 00:01:58,500 And the question is, what's a good way to look at the data? 36 00:01:58,500 --> 00:02:02,410 And today, we're going to look at one such way, which is a 37 00:02:02,410 --> 00:02:06,230 way which is increasingly often used by development 38 00:02:06,230 --> 00:02:09,370 economists, by policy makers, and which is the main approach 39 00:02:09,370 --> 00:02:13,290 that we are using in the Poverty Action Lab, which is 40 00:02:13,290 --> 00:02:16,390 the idea of on randomized evaluation. 41 00:02:16,390 --> 00:02:18,690 So we'll start with our malaria example. 42 00:02:18,690 --> 00:02:21,550 I'm going to use the malaria example, because we had the 43 00:02:21,550 --> 00:02:23,410 debate, so we know the question we're asking. 44 00:02:23,410 --> 00:02:25,580 I'm going to use that as our example. 45 00:02:25,580 --> 00:02:29,080 But, of course, the point I'm making now are important 46 00:02:29,080 --> 00:02:32,200 methodological points as well. 47 00:02:32,200 --> 00:02:35,560 So what are we talking about? 48 00:02:35,560 --> 00:02:38,230 We're talking about a pretty important thing. 49 00:02:38,230 --> 00:02:42,700 Almost 900,000 people die of malaria every year, give and 50 00:02:42,700 --> 00:02:46,370 take, most of them in Africa, most of them under five. 51 00:02:46,370 --> 00:02:48,240 That's the biggest cause of under-five 52 00:02:48,240 --> 00:02:50,780 mortality in the world. 53 00:02:50,780 --> 00:02:54,730 Incidentally, you want to take every single statistic with a 54 00:02:54,730 --> 00:02:58,080 big pinch of salt, because it's not that all the malaria 55 00:02:58,080 --> 00:02:59,750 cases are recorded-- 56 00:02:59,750 --> 00:03:05,000 a lot of people who are not really diagnosed of malaria 57 00:03:05,000 --> 00:03:07,040 and are told to be dead of malaria, so it could be a 58 00:03:07,040 --> 00:03:09,930 gross overestimate or a gross underestimate. 59 00:03:09,930 --> 00:03:12,042 But I think it's fair to say it's a lot of people, one way 60 00:03:12,042 --> 00:03:13,292 or the other. 61 00:03:15,210 --> 00:03:19,250 As we discussed, malaria is transmitted by mosquitoes who 62 00:03:19,250 --> 00:03:22,280 carry the parasite from an infected person to an 63 00:03:22,280 --> 00:03:25,810 uninfected person, mainly at dusk. 64 00:03:25,810 --> 00:03:31,130 So sleeping under a bed net, particularly a bed net which 65 00:03:31,130 --> 00:03:36,550 has insecticide woven into the fiber of the bed net is an 66 00:03:36,550 --> 00:03:39,490 effective way to prevent mosquito bites, and therefore 67 00:03:39,490 --> 00:03:43,150 to prevent the transmission of malaria. 68 00:03:45,870 --> 00:03:51,450 And one thing that no one said, which was important, is 69 00:03:51,450 --> 00:03:56,660 that there is two effects of sleeping under a bed net. 70 00:03:56,660 --> 00:03:59,410 If you already don't have malaria, you're less likely to 71 00:03:59,410 --> 00:04:03,600 catch it, because the mosquitoes can't bite you. 72 00:04:03,600 --> 00:04:06,690 So this is a direct benefit that benefits you. 73 00:04:06,690 --> 00:04:10,650 So when making the decision to buy or not buy a bed net when 74 00:04:10,650 --> 00:04:14,510 making the decision to use the bed net or not to use it, you 75 00:04:14,510 --> 00:04:18,920 are going to take this direct benefit into account, right? 76 00:04:18,920 --> 00:04:21,529 But then there is another benefit which is the 77 00:04:21,529 --> 00:04:24,810 externality, which is an indirect effect, which is, if 78 00:04:24,810 --> 00:04:27,880 you already have malaria, someone else is less likely to 79 00:04:27,880 --> 00:04:29,605 get it because they don't first bite you 80 00:04:29,605 --> 00:04:31,310 and then bite them. 81 00:04:31,310 --> 00:04:35,480 And this benefit, unless you're an incredibly 82 00:04:35,480 --> 00:04:38,520 socially-minded person, is a benefit you don't take into 83 00:04:38,520 --> 00:04:40,420 account when buying your bed net. 84 00:04:40,420 --> 00:04:41,300 Because you don't care. 85 00:04:41,300 --> 00:04:41,660 It's not you. 86 00:04:41,660 --> 00:04:42,830 It's someone else. 87 00:04:42,830 --> 00:04:48,810 But as society, we are trying to maybe adjust the individual 88 00:04:48,810 --> 00:04:52,350 behavior to the social benefit. 89 00:04:52,350 --> 00:04:55,180 There's one more thing which is important for us to keep in 90 00:04:55,180 --> 00:04:58,600 mind, which is this externality is not linear. 91 00:04:58,600 --> 00:05:00,830 So it could be that the externality is linear, which 92 00:05:00,830 --> 00:05:06,460 is one more person who sleeps under a bed net, say half a 93 00:05:06,460 --> 00:05:08,410 fewer percent is infected. 94 00:05:08,410 --> 00:05:10,870 So it could be like the more bed nets you have, the less 95 00:05:10,870 --> 00:05:14,010 malaria you have, but in a disproportional way. 96 00:05:14,010 --> 00:05:16,736 But it turns out that it's not proportional. 97 00:05:16,736 --> 00:05:22,050 The reason why it's not proportionally is because once 98 00:05:22,050 --> 00:05:25,290 about half the population sleeps under the bed net, you 99 00:05:25,290 --> 00:05:29,050 have something called hard immunity, which is that the 100 00:05:29,050 --> 00:05:33,280 mosquitoes don't have time to bite someone else before 101 00:05:33,280 --> 00:05:37,140 either the mosquito or the parasite dies. 102 00:05:37,140 --> 00:05:42,450 So if you manage to cover about half of your population, 103 00:05:42,450 --> 00:05:45,710 you don't even need to carry the other half. 104 00:05:45,710 --> 00:05:48,020 I mean, it's better, but they already 105 00:05:48,020 --> 00:05:50,300 benefit almost as much. 106 00:05:50,300 --> 00:05:55,320 So the externality benefits are like this, if you want. 107 00:05:55,320 --> 00:05:58,220 They're like, first they're linear, and then at some 108 00:05:58,220 --> 00:06:00,770 point, they have a big jump in the externality benefits. 109 00:06:00,770 --> 00:06:06,560 And then again, going from 80% to 85%, maybe less. 110 00:06:06,560 --> 00:06:13,230 So that means that this would need to be reminding you of 111 00:06:13,230 --> 00:06:16,920 the S-shaped thing, which is there will be a distinct 112 00:06:16,920 --> 00:06:23,100 advantage in terms of cost per malaria averted, the benefit 113 00:06:23,100 --> 00:06:28,380 of going from a coverage of 25% to 30% of the population 114 00:06:28,380 --> 00:06:31,720 are going to be lower than the benefit of going, say, from 115 00:06:31,720 --> 00:06:33,950 45% to 50%. 116 00:06:33,950 --> 00:06:36,540 Because when you go from 45% to 50%, you're in the zone 117 00:06:36,540 --> 00:06:40,360 where every extra bed net is really helping a huge deal. 118 00:06:40,360 --> 00:06:41,990 So that is going to be relevant when 119 00:06:41,990 --> 00:06:43,240 we do the cost benefit. 120 00:06:45,580 --> 00:06:46,650 So that's the case for bed nets. 121 00:06:46,650 --> 00:06:51,380 So now, if we were traditional economists, we would make a 122 00:06:51,380 --> 00:06:55,140 traditional economic argument, which is to say, there is an 123 00:06:55,140 --> 00:06:56,570 externality. 124 00:06:56,570 --> 00:06:59,920 Whenever there is an externality, we want to try to 125 00:06:59,920 --> 00:07:06,460 do use prices or taxes to align the private incentive 126 00:07:06,460 --> 00:07:08,330 with the public incentive. 127 00:07:08,330 --> 00:07:11,170 That's why, for example, we would like to have a 128 00:07:11,170 --> 00:07:14,440 congestion charge in London, or we have a congestion charge 129 00:07:14,440 --> 00:07:18,080 in London to make it expensive to drive your car in London, 130 00:07:18,080 --> 00:07:20,530 in the center of London, because when you drive your 131 00:07:20,530 --> 00:07:22,535 car in the center of London, that's a pain 132 00:07:22,535 --> 00:07:23,785 for everybody else. 133 00:07:29,840 --> 00:07:32,120 That could be one of the reasons why we have taxes on 134 00:07:32,120 --> 00:07:36,710 cigarettes, that smokers impose an externality on the 135 00:07:36,710 --> 00:07:40,900 rest of the people, here both because of secondhand smoking, 136 00:07:40,900 --> 00:07:43,110 and because they get sick a lot. 137 00:07:43,110 --> 00:07:45,250 So you would think that society has a right to impose 138 00:07:45,250 --> 00:07:50,750 a tax on them to try and discourage them from smoking. 139 00:07:50,750 --> 00:07:56,550 So by using prices and taxes is where you can try and align 140 00:07:56,550 --> 00:07:59,260 the private incentive to the social benefit. 141 00:07:59,260 --> 00:07:59,610 Yes? 142 00:07:59,610 --> 00:08:02,005 AUDIENCE: Can you do community incentives, like you tell the 143 00:08:02,005 --> 00:08:04,879 community if everyone uses their bed nets as bed nets, 144 00:08:04,879 --> 00:08:07,760 you'll give the community a certain number of fishing nets 145 00:08:07,760 --> 00:08:10,168 so that people will encourage or force other 146 00:08:10,168 --> 00:08:11,500 people to use bed nets? 147 00:08:11,500 --> 00:08:12,150 ESTHER DUFLO: Right. 148 00:08:12,150 --> 00:08:16,860 So one way is the prices, but another way could be to have 149 00:08:16,860 --> 00:08:18,110 community incentives. 150 00:08:20,190 --> 00:08:23,670 Indonesia has a program where they are trying to give 151 00:08:23,670 --> 00:08:30,170 community incentive so if, say 90% percent of the community 152 00:08:30,170 --> 00:08:35,000 is immunized, and x percent to the community is in school, 153 00:08:35,000 --> 00:08:38,770 then they get some amount of money to build a road, or an 154 00:08:38,770 --> 00:08:41,140 irrigation project, or something like that. 155 00:08:41,140 --> 00:08:41,420 AUDIENCE: I'm just curious. 156 00:08:41,420 --> 00:08:44,930 How much is given in foreign aid to fight malaria? 157 00:08:44,930 --> 00:08:46,800 ESTHER DUFLO: That's an excellent-- how much is given 158 00:08:46,800 --> 00:08:47,870 in foreign aid to fight malaria. 159 00:08:47,870 --> 00:08:51,990 It's an excellent question, the answer to which I don't 160 00:08:51,990 --> 00:08:54,728 know off the top of my head, but it's findable. 161 00:08:54,728 --> 00:08:57,540 AUDIENCE: Because I was going to ask if the cost of 162 00:08:57,540 --> 00:09:01,112 providing a country with bed nets is significantly less 163 00:09:01,112 --> 00:09:04,840 than the cost of giving [INAUDIBLE] money to fight 164 00:09:04,840 --> 00:09:08,816 malaria, then I don't even see why it's as big of a 165 00:09:08,816 --> 00:09:10,307 debate as it is. 166 00:09:10,307 --> 00:09:12,626 Clearly, they should figure out what's the most effective 167 00:09:12,626 --> 00:09:16,768 way to distribute the bed nets, make sure [INAUDIBLE]. 168 00:09:16,768 --> 00:09:20,247 But the option of not giving them just wouldn't make as 169 00:09:20,247 --> 00:09:22,235 much sense, in terms of cost [INAUDIBLE]. 170 00:09:25,250 --> 00:09:26,120 ESTHER DUFLO: Oh, I see. 171 00:09:26,120 --> 00:09:28,240 So providing bed nets would be part of the 172 00:09:28,240 --> 00:09:29,490 budget of fighting malaria. 173 00:09:31,880 --> 00:09:34,450 So your question is that, if we're already spending so much 174 00:09:34,450 --> 00:09:37,380 money on medicine against malaria, we would really be 175 00:09:37,380 --> 00:09:39,440 willing to spend a lot of money on bed nets. 176 00:09:39,440 --> 00:09:40,660 You're exactly right. 177 00:09:40,660 --> 00:09:46,050 And I don't think the debate is that we want to try and 178 00:09:46,050 --> 00:09:47,660 save money on the bed nets. 179 00:09:47,660 --> 00:09:51,770 I think the debate is always about what's the best way to 180 00:09:51,770 --> 00:09:54,080 get the money out. 181 00:09:54,080 --> 00:09:57,420 So this would argue, so the fact that the private 182 00:09:57,420 --> 00:10:01,350 incentive to get your bed net is low, would argue for 183 00:10:01,350 --> 00:10:08,530 subsidizing the bed net, but only if the private incentives 184 00:10:08,530 --> 00:10:11,760 are not large enough that people are already not buying 185 00:10:11,760 --> 00:10:12,860 your bed net. 186 00:10:12,860 --> 00:10:16,330 So suppose, for example, that people hated being sick, 187 00:10:16,330 --> 00:10:19,700 which, presumably, they do. 188 00:10:19,700 --> 00:10:21,910 Then, they would get the bed net anyway. 189 00:10:21,910 --> 00:10:24,770 You would not need to subsidize the bed net. 190 00:10:24,770 --> 00:10:27,020 So there would not be an effect of the price, because 191 00:10:27,020 --> 00:10:30,430 the demand would be very high at any price. 192 00:10:30,430 --> 00:10:34,030 So the fact that there is an externality justifies 193 00:10:34,030 --> 00:10:38,120 subsidizing, to the extent that the private incentives 194 00:10:38,120 --> 00:10:44,890 are too small for people to want to get them themselves. 195 00:10:44,890 --> 00:10:50,060 Another traditional economics argument would be the learning 196 00:10:50,060 --> 00:10:53,960 argument that was made earlier, which is that trying 197 00:10:53,960 --> 00:10:58,300 your bed net once might give you the experience of it, and 198 00:10:58,300 --> 00:11:01,260 once you knew how it works, and you know how well it 199 00:11:01,260 --> 00:11:06,530 protects your children, you might be more likely to get 200 00:11:06,530 --> 00:11:07,950 the next one. 201 00:11:07,950 --> 00:11:10,970 And then there is an argument that mixes the two, which is 202 00:11:10,970 --> 00:11:14,820 another externality argument, which is, if try a bed net, 203 00:11:14,820 --> 00:11:17,520 and I realize that it protects my children and it's really 204 00:11:17,520 --> 00:11:21,030 not so unpleasant, then maybe my neighbors will also see my 205 00:11:21,030 --> 00:11:22,730 bed net, and say, oh, isn't this cool. 206 00:11:22,730 --> 00:11:24,350 We want also a bed net. 207 00:11:24,350 --> 00:11:27,060 So there is an externality in my trying, potentially. 208 00:11:27,060 --> 00:11:28,050 Yeah? 209 00:11:28,050 --> 00:11:30,772 AUDIENCE: But if you were charged for bed nets and the 210 00:11:30,772 --> 00:11:34,485 borrower that could, again, come in to have the bed nets 211 00:11:34,485 --> 00:11:35,735 [INAUDIBLE]? 212 00:11:42,910 --> 00:11:43,950 ESTHER DUFLO: Yes. 213 00:11:43,950 --> 00:11:49,200 So eventually, since there are those contingent benefits, 214 00:11:49,200 --> 00:11:52,470 someone who is very rational might feel, already more than 215 00:11:52,470 --> 00:11:54,620 50% bed nets in this community. 216 00:11:54,620 --> 00:11:56,780 I don't need to get the next one. 217 00:11:56,780 --> 00:12:00,520 So the private incentives diminish as the coverage 218 00:12:00,520 --> 00:12:05,260 increases, which makes it more necessary either to subsidize 219 00:12:05,260 --> 00:12:10,510 it even more, or to tax people if they don't get a bed net. 220 00:12:10,510 --> 00:12:14,090 So this is an argument that is made for immunization. 221 00:12:14,090 --> 00:12:19,360 So in this country, immunization is compulsory for 222 00:12:19,360 --> 00:12:20,990 a range of diseases. 223 00:12:20,990 --> 00:12:23,630 Except if you really don't want to do it because you have 224 00:12:23,630 --> 00:12:29,940 some religious or whatever reasons for not wanting to be 225 00:12:29,940 --> 00:12:33,080 immunized, then you can say, sorry, I refuse the 226 00:12:33,080 --> 00:12:34,890 immunization. 227 00:12:34,890 --> 00:12:44,600 And people worry that as diseases such as polio, or 228 00:12:44,600 --> 00:12:48,450 diphtheria, or pertussis, these types of diseases become 229 00:12:48,450 --> 00:12:51,980 less frequent, people don't have the fear of getting those 230 00:12:51,980 --> 00:12:54,750 diseases anymore, and they want to free ride on the rest 231 00:12:54,750 --> 00:12:55,800 of the community. 232 00:12:55,800 --> 00:13:00,780 And that's, of course, bad, because being 100% immunized 233 00:13:00,780 --> 00:13:03,440 ensures that the disease goes away. 234 00:13:03,440 --> 00:13:07,480 As soon as a few people are not immunized, 235 00:13:07,480 --> 00:13:08,930 it could come back. 236 00:13:08,930 --> 00:13:13,820 And so there is this tension between once almost everyone's 237 00:13:13,820 --> 00:13:15,670 immunized, your private incentive to do 238 00:13:15,670 --> 00:13:17,080 it are almost zero. 239 00:13:17,080 --> 00:13:23,180 So you might need powerful inducement to get to do it. 240 00:13:23,180 --> 00:13:26,080 So those are the traditional economic arguments. 241 00:13:26,080 --> 00:13:27,620 These are your traditional public finance-- 242 00:13:27,620 --> 00:13:29,960 let's use the prices. 243 00:13:29,960 --> 00:13:32,870 That indicates that we should subsidize the bed nets. 244 00:13:32,870 --> 00:13:34,390 And by the way, there is no reason to 245 00:13:34,390 --> 00:13:35,740 stop at zero in principle. 246 00:13:35,740 --> 00:13:39,480 We could pay people to use bed nets, for that matter, if this 247 00:13:39,480 --> 00:13:45,250 was the way to get the optimal coverage in the community. 248 00:13:45,250 --> 00:13:46,228 Yeah? 249 00:13:46,228 --> 00:13:49,406 AUDIENCE: So one consideration that comes to mind is if 250 00:13:49,406 --> 00:13:51,607 there's a segment of the population that is really, 251 00:13:51,607 --> 00:13:54,378 really poor and cannot afford to pay any amount for the bed 252 00:13:54,378 --> 00:13:56,110 nets, whether they're subsidized or not, those 253 00:13:56,110 --> 00:13:58,830 people are also the least likely to be able to pay for 254 00:13:58,830 --> 00:14:00,244 health care should they get malaria. 255 00:14:00,244 --> 00:14:02,940 So they would have the greatest risk for everyone 256 00:14:02,940 --> 00:14:04,235 else and for themselves. 257 00:14:04,235 --> 00:14:06,760 So wouldn't you want to try to provide free bed nets for that 258 00:14:06,760 --> 00:14:10,320 set of people, no matter what? 259 00:14:10,320 --> 00:14:13,450 ESTHER DUFLO: Right, so that's another consideration that you 260 00:14:13,450 --> 00:14:17,730 might want to include, a redistribution consideration, 261 00:14:17,730 --> 00:14:24,540 which you might feel that people who are particularly 262 00:14:24,540 --> 00:14:27,435 poor might have a very high private evaluation of the bed 263 00:14:27,435 --> 00:14:30,320 nets, but might be credit constrained and not be able to 264 00:14:30,320 --> 00:14:31,440 get the bed net. 265 00:14:31,440 --> 00:14:33,880 And those are the same people who will also not be able to 266 00:14:33,880 --> 00:14:35,530 cure themselves of malaria. 267 00:14:35,530 --> 00:14:39,420 So here, that gives you another argument for 268 00:14:39,420 --> 00:14:42,580 subsidizing bed nets, which is not an argument in terms of 269 00:14:42,580 --> 00:14:45,380 disease eradication, but an argument in terms of 270 00:14:45,380 --> 00:14:48,560 redistribution between people. 271 00:14:48,560 --> 00:14:50,400 So those are traditional arguments. 272 00:14:50,400 --> 00:14:55,450 They are taken, say, by Jeff Sachs, or by the WHO, and say, 273 00:14:55,450 --> 00:15:00,350 we should subsidize bed net to a large extent, or 274 00:15:00,350 --> 00:15:03,870 potentially, up to giving them away for free. 275 00:15:03,870 --> 00:15:08,210 And then we have the less conventional economics, which 276 00:15:08,210 --> 00:15:13,070 we discussed, which has the argument that goes in the 277 00:15:13,070 --> 00:15:15,630 opposite direction, or the psychological sunk cost 278 00:15:15,630 --> 00:15:19,410 effect, which is, you have to pay for something to really 279 00:15:19,410 --> 00:15:21,780 appreciate its value. 280 00:15:21,780 --> 00:15:24,350 There is a selection effect which is, if you give things 281 00:15:24,350 --> 00:15:27,650 away for free, even people who don't want it or who don't 282 00:15:27,650 --> 00:15:30,630 want it for its legitimate use are going to come and get it. 283 00:15:30,630 --> 00:15:33,950 They're going to waste a lot of bed nets. 284 00:15:33,950 --> 00:15:37,190 And finally, the entitlement effect, which counters the 285 00:15:37,190 --> 00:15:39,390 learning effect, which is, once you get something for 286 00:15:39,390 --> 00:15:44,160 free, then you expect to get it for free in the future. 287 00:15:44,160 --> 00:15:47,960 And it might even have contingent things, which is 288 00:15:47,960 --> 00:15:51,400 you're not going to want to get chlorine to purify your 289 00:15:51,400 --> 00:15:54,690 water, unless someone is giving it to you. 290 00:15:54,690 --> 00:15:59,380 You're not going to want to send your kids to school 291 00:15:59,380 --> 00:16:03,750 unless they get a school meal, other things like that, which 292 00:16:03,750 --> 00:16:06,210 find that you become dependent. 293 00:16:06,210 --> 00:16:06,710 Yeah? 294 00:16:06,710 --> 00:16:07,710 AUDIENCE: I have a question. 295 00:16:07,710 --> 00:16:11,710 Couldn't you argue that it's not sustainable, [INAUDIBLE] 296 00:16:11,710 --> 00:16:15,210 long-term policy and exit strategies, because if you're 297 00:16:15,210 --> 00:16:16,210 [INAUDIBLE] 298 00:16:16,210 --> 00:16:19,710 subsidizing and giving away bed nets, you might crowd out 299 00:16:19,710 --> 00:16:20,710 the private sector. 300 00:16:20,710 --> 00:16:25,460 And you won't [INAUDIBLE] your desired outcome, which is for 301 00:16:25,460 --> 00:16:26,710 people to be using bed nets. 302 00:16:29,210 --> 00:16:31,840 ESTHER DUFLO: So in the traditional public finance 303 00:16:31,840 --> 00:16:35,140 argument, why would you ever want to stop to 304 00:16:35,140 --> 00:16:37,154 subsidize the bed net? 305 00:16:37,154 --> 00:16:38,480 AUDIENCE: Want to not subsidize-- 306 00:16:38,480 --> 00:16:38,650 ESTHER DUFLO: Yeah. 307 00:16:38,650 --> 00:16:43,105 Why would you ever want to stop, given those arguments? 308 00:16:43,105 --> 00:16:44,570 AUDIENCE: To create a market? 309 00:16:44,570 --> 00:16:45,750 ESTHER DUFLO: Why would you? 310 00:16:45,750 --> 00:16:49,360 Take, for example, immunization in the US. 311 00:16:49,360 --> 00:16:51,830 Immunization is subsidized and compulsory. 312 00:16:51,830 --> 00:16:56,450 Is there any plan to make immunization not subsidized or 313 00:16:56,450 --> 00:16:57,225 compulsory? 314 00:16:57,225 --> 00:16:57,690 AUDIENCE: That's true. 315 00:16:57,690 --> 00:16:59,290 ESTHER DUFLO: There's no such plan. 316 00:16:59,290 --> 00:17:02,270 This kind of argument would tell you that you want to 317 00:17:02,270 --> 00:17:07,510 subsidize forever, because the private gains and the social 318 00:17:07,510 --> 00:17:10,970 gains are not aligned for a fundamental reason, which is 319 00:17:10,970 --> 00:17:13,670 my being protected by a bed net protects you as well, and 320 00:17:13,670 --> 00:17:15,660 that's never going to change. 321 00:17:15,660 --> 00:17:17,780 So in that sense, you're not doing a policy 322 00:17:17,780 --> 00:17:18,579 for the short run. 323 00:17:18,579 --> 00:17:21,450 You're doing this policy for the long run. 324 00:17:21,450 --> 00:17:25,420 Now, again, we had this discussion earlier, which is 325 00:17:25,420 --> 00:17:28,079 the questions of who provides the bed net, whether they can 326 00:17:28,079 --> 00:17:30,630 be provided locally instead of imported, are all very 327 00:17:30,630 --> 00:17:33,770 legitimate questions, but in a sense, unrelated to the price 328 00:17:33,770 --> 00:17:36,040 you're going to charge. 329 00:17:36,040 --> 00:17:38,360 So that's an excellent question, because it comes 330 00:17:38,360 --> 00:17:40,280 back all the time, which is to say, oh, but 331 00:17:40,280 --> 00:17:41,720 don't we want to exit? 332 00:17:41,720 --> 00:17:44,380 Just think, whenever you want to ask yourself this question 333 00:17:44,380 --> 00:17:46,680 about sustainability, ask yourself, why 334 00:17:46,680 --> 00:17:47,560 would I want to exit? 335 00:17:47,560 --> 00:17:48,520 Maybe I don't. 336 00:17:48,520 --> 00:17:53,890 Maybe it is actually socially efficient to subsidize 337 00:17:53,890 --> 00:17:59,180 something for eternity, because as Ben was saying, you 338 00:17:59,180 --> 00:18:06,020 save on the cost of curing people from malaria. 339 00:18:06,020 --> 00:18:11,100 So the argument that goes against giving people for free 340 00:18:11,100 --> 00:18:14,760 are these arguments that giving bed nets for free or 341 00:18:14,760 --> 00:18:18,240 even subsidizing them is counterproductive, in terms of 342 00:18:18,240 --> 00:18:19,940 distributing bed nets. 343 00:18:19,940 --> 00:18:22,860 It's counterproductive because of the sunk cost effect, 344 00:18:22,860 --> 00:18:25,140 because of the selection effect, or because this 345 00:18:25,140 --> 00:18:29,460 entitlement effect that then nobody is ever going to want 346 00:18:29,460 --> 00:18:34,004 to buy anything, either a bed net or something else. 347 00:18:34,004 --> 00:18:40,970 And these elements are very reasonable. 348 00:18:40,970 --> 00:18:43,300 Both sides of the arguments are, in principle, very 349 00:18:43,300 --> 00:18:44,480 reasonable. 350 00:18:44,480 --> 00:18:45,530 And they might apply here. 351 00:18:45,530 --> 00:18:47,250 We just don't know. 352 00:18:47,250 --> 00:18:50,140 It's a very heated policy debate. 353 00:18:50,140 --> 00:18:54,520 The laws that are involved in bed nets are nowhere near the 354 00:18:54,520 --> 00:18:56,900 laws that are involved in oil or anything, but it's still a 355 00:18:56,900 --> 00:18:59,640 fair amount of dollars, so people like those. 356 00:18:59,640 --> 00:19:02,270 So there is a lot of debate. 357 00:19:02,270 --> 00:19:06,650 So Jeff Sachs of the WHO relies on the convention 358 00:19:06,650 --> 00:19:08,330 economics argument to say you should give the 359 00:19:08,330 --> 00:19:10,080 bed nets for free. 360 00:19:10,080 --> 00:19:12,900 Then you have people like Bill Easterly or Dambisa Moyo, who 361 00:19:12,900 --> 00:19:18,820 are making the argument based on this entitlement effect, 362 00:19:18,820 --> 00:19:24,690 sunk cost effect, the development of market, to say 363 00:19:24,690 --> 00:19:26,540 we shouldn't give them for free. 364 00:19:26,540 --> 00:19:29,790 And both arguments tell you, in a sense, we know the 365 00:19:29,790 --> 00:19:33,960 economics, and the answer is obvious. 366 00:19:33,960 --> 00:19:36,360 There is no reason to find out. 367 00:19:36,360 --> 00:19:41,070 And I think there is value in both type of arguments that 368 00:19:41,070 --> 00:19:41,670 they are making. 369 00:19:41,670 --> 00:19:44,380 But there is no value in this last one. 370 00:19:44,380 --> 00:19:47,360 Because since there is value in both arguments, we don't 371 00:19:47,360 --> 00:19:49,010 know the answer. 372 00:19:49,010 --> 00:19:53,530 The answer depends on the balance of how they turn out 373 00:19:53,530 --> 00:19:55,505 in the specific application of bed nets. 374 00:19:58,070 --> 00:20:00,810 So the true question we want to ask is the extent to which 375 00:20:00,810 --> 00:20:03,770 we should subsidize bed nets. 376 00:20:03,770 --> 00:20:06,080 And for that, what do we need to know? 377 00:20:06,080 --> 00:20:09,480 Well, first we need to know whether or not bed nets are 378 00:20:09,480 --> 00:20:11,490 really price elastic. 379 00:20:11,490 --> 00:20:15,230 So are we losing a lot of coverage in terms of people 380 00:20:15,230 --> 00:20:20,380 acquiring bed nets, if we start selling them instead of 381 00:20:20,380 --> 00:20:22,810 giving them for free. 382 00:20:22,810 --> 00:20:24,620 Second question we need to ask is the 383 00:20:24,620 --> 00:20:27,120 immediate effect on use. 384 00:20:27,120 --> 00:20:31,087 So is it the case that people who pay for bed nets are more 385 00:20:31,087 --> 00:20:34,920 likely to use it than people who got it for free? 386 00:20:34,920 --> 00:20:37,570 Third question is the longer-term effect. 387 00:20:37,570 --> 00:20:39,610 Will it wreck markets? 388 00:20:39,610 --> 00:20:43,050 And for that, that's the entitlement effect. 389 00:20:43,050 --> 00:20:46,680 For people who get it for free, in the future, are they 390 00:20:46,680 --> 00:20:49,930 more or less likely to buy one? 391 00:20:49,930 --> 00:20:53,560 And for people who are around them in the community-- 392 00:20:53,560 --> 00:20:55,870 this is something we didn't discuss much, but if I'm 393 00:20:55,870 --> 00:20:58,410 around you in the community, and I see you having a free 394 00:20:58,410 --> 00:21:02,210 bed net, it could affect me in one of two ways. 395 00:21:02,210 --> 00:21:06,280 I could say, first of all, oh, she has a bed net. 396 00:21:06,280 --> 00:21:07,170 That's really nice. 397 00:21:07,170 --> 00:21:08,160 Her kids are doing well. 398 00:21:08,160 --> 00:21:10,160 I'm going to get one too. 399 00:21:10,160 --> 00:21:12,070 I could say, oh, she got a free bed net. 400 00:21:12,070 --> 00:21:13,510 Let me wait until I get-- 401 00:21:13,510 --> 00:21:15,730 I was going to buy a bed net, but now I'm going to wait 402 00:21:15,730 --> 00:21:17,550 until I get a free bed net. 403 00:21:17,550 --> 00:21:20,170 And so I wasn't going to sell a bed net, and now I'm going 404 00:21:20,170 --> 00:21:21,740 to have give it for free. 405 00:21:21,740 --> 00:21:24,190 So that's wasted money. 406 00:21:24,190 --> 00:21:27,380 So again, we don't know which way it cuts. 407 00:21:27,380 --> 00:21:30,190 We have to look in practice. 408 00:21:30,190 --> 00:21:32,650 And those questions are pretty well-defined questions. 409 00:21:32,650 --> 00:21:34,930 They are empirical questions we could principally 410 00:21:34,930 --> 00:21:37,280 bring an answer to. 411 00:21:37,280 --> 00:21:39,070 There are a lot of anecdotes. 412 00:21:39,070 --> 00:21:44,180 They are this one, which is bed nets used as fishing nets. 413 00:21:44,180 --> 00:21:47,280 That's a picture that was in the Malaria Journal, so it's 414 00:21:47,280 --> 00:21:51,470 all peer reviewed and all that. 415 00:21:51,470 --> 00:21:55,670 So this is the Easterly Dambisa Moyo argument, that 416 00:21:55,670 --> 00:21:59,300 you are going to see the used as a fishing net, or wedding 417 00:21:59,300 --> 00:22:02,020 veils, or what have you. 418 00:22:02,020 --> 00:22:05,880 And then this is the other argument, of the other 419 00:22:05,880 --> 00:22:09,750 anecdote of the bed net being nicely hanged in the room, and 420 00:22:09,750 --> 00:22:12,570 the child, the healthy child who slept in there 421 00:22:12,570 --> 00:22:13,500 under the bed net. 422 00:22:13,500 --> 00:22:15,910 So that's not too helpful. 423 00:22:15,910 --> 00:22:19,350 So we want to go further that the anecdote. 424 00:22:19,350 --> 00:22:22,690 There are many anecdotes, and they cut both ways. 425 00:22:28,700 --> 00:22:33,270 How do we go further than the anecdote? 426 00:22:33,270 --> 00:22:37,065 So the first thing we could do-- so suppose you 427 00:22:37,065 --> 00:22:38,680 were given this job. 428 00:22:38,680 --> 00:22:42,816 I tell you, find out for me whether people who get their 429 00:22:42,816 --> 00:22:45,470 bed nets for free are more likely to use it as fishing 430 00:22:45,470 --> 00:22:49,830 nets or more likely to use them as bed nets. 431 00:22:49,830 --> 00:22:53,670 And also, find out whether if I distribute the bed nets for 432 00:22:53,670 --> 00:22:57,230 free, I distribute many more of them or I distribute the 433 00:22:57,230 --> 00:22:58,100 same amount. 434 00:22:58,100 --> 00:23:03,280 What would your first response to what you 435 00:23:03,280 --> 00:23:04,530 should go out and do? 436 00:23:07,640 --> 00:23:08,130 AUDIENCE: Yeah? 437 00:23:08,130 --> 00:23:11,070 I would find two similar communities, and then 438 00:23:11,070 --> 00:23:12,540 [INAUDIBLE] 439 00:23:12,540 --> 00:23:16,950 give bed nets would be one, and maybe help someone start a 440 00:23:16,950 --> 00:23:20,070 small business of selling bed nets in the other. 441 00:23:20,070 --> 00:23:22,370 ESTHER DUFLO: So you would take two similar communities, 442 00:23:22,370 --> 00:23:27,870 and in once case, give bed nets for free, and in one 443 00:23:27,870 --> 00:23:32,340 case, start to sell them, maybe by finding a small 444 00:23:32,340 --> 00:23:35,010 business, or maybe importing them to the village. 445 00:23:35,010 --> 00:23:37,320 What would be an even--- 446 00:23:37,320 --> 00:23:38,080 we'll go back to that. 447 00:23:38,080 --> 00:23:42,510 What would be an even easier, even more immediate way to do 448 00:23:42,510 --> 00:23:47,190 it that we could think of to start with, if we needed to do 449 00:23:47,190 --> 00:23:49,276 this quickly? 450 00:23:49,276 --> 00:23:49,748 Yes? 451 00:23:49,748 --> 00:23:50,456 AUDIENCE: Just look at the 452 00:23:50,456 --> 00:23:53,524 communities where it is already. 453 00:23:53,524 --> 00:23:53,996 ESTHER DUFLO: Right. 454 00:23:53,996 --> 00:23:56,270 You would look at a community where this is already in 455 00:23:56,270 --> 00:23:59,090 place, and you would say, well, in this community, the 456 00:23:59,090 --> 00:24:02,365 bed nets are given for free, and in this community, the bed 457 00:24:02,365 --> 00:24:04,070 nets are sold. 458 00:24:04,070 --> 00:24:04,590 Let me try my thing. 459 00:24:04,590 --> 00:24:05,780 Yeah. 460 00:24:05,780 --> 00:24:07,800 In this community, the bed nets are given for free. 461 00:24:07,800 --> 00:24:10,240 In this community, the bed nets are sold. 462 00:24:10,240 --> 00:24:13,960 First question, do I see more bed nets where the bed nets 463 00:24:13,960 --> 00:24:16,230 are not given for free? 464 00:24:16,230 --> 00:24:20,140 Second question, do I see the bed nets more likely to be 465 00:24:20,140 --> 00:24:22,790 used as fishing nets in those communities? 466 00:24:22,790 --> 00:24:27,150 So that's, I think, what most people would-- 467 00:24:27,150 --> 00:24:31,260 that would be your initial instinct. 468 00:24:31,260 --> 00:24:34,230 The problem you would have if you did that, as opposed to-- 469 00:24:34,230 --> 00:24:35,870 sorry, can you remind of your name? 470 00:24:35,870 --> 00:24:36,820 AUDIENCE: [? Sif. ?] 471 00:24:36,820 --> 00:24:36,960 ESTHER DUFLO: [? Sif. ?] 472 00:24:36,960 --> 00:24:39,160 The problem to do that, as opposed to do what [? Sif ?] 473 00:24:39,160 --> 00:24:46,740 suggested, is to know why some community happened to be the 474 00:24:46,740 --> 00:24:50,110 one that got them for free and some community happened not to 475 00:24:50,110 --> 00:24:52,270 have the free distribution. 476 00:24:52,270 --> 00:24:55,540 So let's think of one reason why there might be a free-- 477 00:24:55,540 --> 00:24:59,190 if you were to start a free bed net distribution program 478 00:24:59,190 --> 00:25:03,240 somewhere, where would you start it? 479 00:25:03,240 --> 00:25:05,730 You have a country and you want to start a free bed net 480 00:25:05,730 --> 00:25:07,330 distribution, what would you pick? 481 00:25:07,330 --> 00:25:07,610 Yes? 482 00:25:07,610 --> 00:25:08,951 AUDIENCE: If there was a particularly bad 483 00:25:08,951 --> 00:25:10,740 increase, or something. 484 00:25:10,740 --> 00:25:11,140 ESTHER DUFLO: Right. 485 00:25:11,140 --> 00:25:14,010 You would try to pick a place where there is a lot of 486 00:25:14,010 --> 00:25:17,030 malaria, presumably, or maybe a place where 487 00:25:17,030 --> 00:25:19,770 people are very poor. 488 00:25:19,770 --> 00:25:25,580 So let's work with the example where you would pick, to start 489 00:25:25,580 --> 00:25:28,080 your free bed net distribution, the region where 490 00:25:28,080 --> 00:25:29,790 there is a lot of malaria. 491 00:25:29,790 --> 00:25:31,040 It would make sense. 492 00:25:33,560 --> 00:25:36,720 So if that's the case that free bed nets are distributed 493 00:25:36,720 --> 00:25:41,400 where malaria is a huge problem, then in these places, 494 00:25:41,400 --> 00:25:45,660 even if they had to pay for the bed net, they might have 495 00:25:45,660 --> 00:25:48,780 purchased them more willingly than in places where malaria 496 00:25:48,780 --> 00:25:51,490 is not such a big problem anyway. 497 00:25:51,490 --> 00:25:54,180 Because the private benefits of having a bed net is higher 498 00:25:54,180 --> 00:25:56,110 in those [INAUDIBLE]. 499 00:25:56,110 --> 00:25:59,810 So this is what we call the counter-factual, which is what 500 00:25:59,810 --> 00:26:03,190 would have happened in the place where the bed net got 501 00:26:03,190 --> 00:26:08,280 distributed for free if they had actually been sold. 502 00:26:08,280 --> 00:26:11,010 So let's look at it in the form of a simple graph. 503 00:26:11,010 --> 00:26:13,100 So suppose we have two-- 504 00:26:13,100 --> 00:26:15,800 we are going to do the argument for purchase, and the 505 00:26:15,800 --> 00:26:19,500 same kind of discussion we could have for use. 506 00:26:19,500 --> 00:26:21,970 So suppose we have two regions, a high-malaria region 507 00:26:21,970 --> 00:26:24,040 and a low-malaria region. 508 00:26:24,040 --> 00:26:31,000 And suppose that God told us this is how many people would 509 00:26:31,000 --> 00:26:34,750 buy a bed net, the fraction of people who would buy a bed net 510 00:26:34,750 --> 00:26:39,380 in the high-malaria region and the low-malaria region, if the 511 00:26:39,380 --> 00:26:42,130 bed net was not subsidized. 512 00:26:42,130 --> 00:26:48,570 And this dot is higher, because since malaria is more 513 00:26:48,570 --> 00:26:51,620 of a problems, people are more willing to buy bed nets. 514 00:26:51,620 --> 00:26:53,100 OK? 515 00:26:53,100 --> 00:26:57,500 So that's what we would have if bed nets were expensive. 516 00:26:57,500 --> 00:27:03,550 And if bed nets were free, we would have those ones. 517 00:27:03,550 --> 00:27:06,310 So compared to the previous dots, they are higher. 518 00:27:06,310 --> 00:27:07,730 Why are they higher in both regions? 519 00:27:11,480 --> 00:27:12,550 AUDIENCE: It's because they're free. 520 00:27:12,550 --> 00:27:14,270 ESTHER DUFLO: Because they're free, so you're more likely to 521 00:27:14,270 --> 00:27:16,285 get something free than to have to pay for it. 522 00:27:18,910 --> 00:27:21,230 But they are still higher in the high-malaria region, 523 00:27:21,230 --> 00:27:24,270 because why would you get a bed net in a region where 524 00:27:24,270 --> 00:27:25,430 malaria is not a problem? 525 00:27:25,430 --> 00:27:26,880 You really don't need it. 526 00:27:26,880 --> 00:27:29,080 If you don't have a fishing boat, you don't need a fishing 527 00:27:29,080 --> 00:27:30,260 net either. 528 00:27:30,260 --> 00:27:33,695 So they are still higher in a high-malaria region then in a 529 00:27:33,695 --> 00:27:36,220 low-malaria region when bed nets are free. 530 00:27:36,220 --> 00:27:40,100 So this is what God would have given you. 531 00:27:40,100 --> 00:27:43,690 But sadly, God didn't give you that. 532 00:27:43,690 --> 00:27:49,680 God gave you either this point or that point. 533 00:27:49,680 --> 00:27:54,330 What you observe is the purchase of bed nets in the 534 00:27:54,330 --> 00:27:58,790 area where they are given for free, and the purchase of bed 535 00:27:58,790 --> 00:28:01,690 nets in the area where they are expensive. 536 00:28:01,690 --> 00:28:03,982 And those regions are different because one is a 537 00:28:03,982 --> 00:28:07,550 high-malaria region and one is a low-malaria region. 538 00:28:07,550 --> 00:28:10,400 So what you observe is the demand for bed nets when they 539 00:28:10,400 --> 00:28:14,215 are free in a high-malaria region, and the demand for bed 540 00:28:14,215 --> 00:28:17,590 nets when they're expense in the low-malaria region. 541 00:28:17,590 --> 00:28:20,590 And what you don't see is those guys. 542 00:28:20,590 --> 00:28:23,630 What you see is this and this. 543 00:28:23,630 --> 00:28:27,440 What is the true effect of price on purchase in the 544 00:28:27,440 --> 00:28:28,390 high-malaria region? 545 00:28:28,390 --> 00:28:30,392 Where do I find it on the graph? 546 00:28:33,490 --> 00:28:34,880 That's the vertical thing. 547 00:28:34,880 --> 00:28:39,120 That's the true effect of free versus expensive in the 548 00:28:39,120 --> 00:28:41,030 high-malaria region. 549 00:28:41,030 --> 00:28:45,090 And this is the true effect of free versus expensive in the 550 00:28:45,090 --> 00:28:47,790 low-malaria region. 551 00:28:47,790 --> 00:28:49,050 So this is the true effect. 552 00:28:49,050 --> 00:28:50,740 That's what I'm looking for. 553 00:28:50,740 --> 00:28:54,110 Sadly, I don't have this, and I don't have this. 554 00:28:54,110 --> 00:28:56,080 I'm not in very good shape. 555 00:28:56,080 --> 00:28:58,670 What I do have is this and this. 556 00:28:58,670 --> 00:29:04,480 So if I just did to comparison that we started with, we would 557 00:29:04,480 --> 00:29:08,450 compare the high-malaria region purchase when it's free 558 00:29:08,450 --> 00:29:11,510 to the low-malaria purchase region when it's expensive. 559 00:29:11,510 --> 00:29:12,250 Take the difference. 560 00:29:12,250 --> 00:29:15,070 That would be our effect. 561 00:29:15,070 --> 00:29:17,780 So is that the true effect? 562 00:29:17,780 --> 00:29:22,250 Is it biased upwards, or is it biased downwards? 563 00:29:22,250 --> 00:29:25,336 Is that too big, too small, or the right size? 564 00:29:25,336 --> 00:29:26,150 AUDIENCE: Too big. 565 00:29:26,150 --> 00:29:27,090 ESTHER DUFLO: Too big. 566 00:29:27,090 --> 00:29:29,890 And we can see it here, which we can re-put. 567 00:29:29,890 --> 00:29:34,512 We can say, well, suppose we had what God would have given 568 00:29:34,512 --> 00:29:36,990 all the numbers. 569 00:29:36,990 --> 00:29:40,170 In the low-malaria region, this is the effect of price 570 00:29:40,170 --> 00:29:41,790 and purchase. 571 00:29:41,790 --> 00:29:44,030 And the rest is the best. 572 00:29:44,030 --> 00:29:47,120 In the high-malaria region, this is the true effect of 573 00:29:47,120 --> 00:29:51,940 price and purchase, and the rest is the bias. 574 00:29:51,940 --> 00:29:56,420 So if we knew this and this, we could calculate the effect, 575 00:29:56,420 --> 00:29:59,230 and we calculate the bias that we would get if we compare 576 00:29:59,230 --> 00:30:00,890 this to this. 577 00:30:00,890 --> 00:30:04,370 But suddenly, we don't have those. 578 00:30:04,370 --> 00:30:07,000 So what this is saying, in other words, is that when we 579 00:30:07,000 --> 00:30:13,670 compare purchase when it's free to purchase when it's 580 00:30:13,670 --> 00:30:20,980 expensive, in the high- and low-malaria region, we combine 581 00:30:20,980 --> 00:30:21,730 two effects. 582 00:30:21,730 --> 00:30:25,220 We combine the effect that even at the same price, the 583 00:30:25,220 --> 00:30:27,260 purchase would have been different. 584 00:30:27,260 --> 00:30:29,580 So the counter-factual would have been different-- 585 00:30:29,580 --> 00:30:33,450 those are this versus this, and this versus this-- 586 00:30:33,450 --> 00:30:36,430 and the fact that there is actually an effect. 587 00:30:36,430 --> 00:30:41,450 So that's why it's not a very good answer for the question 588 00:30:41,450 --> 00:30:42,770 we're asking. 589 00:30:42,770 --> 00:30:45,710 This we did for only two points, but we would have the 590 00:30:45,710 --> 00:30:48,710 same thing if we had many price points. 591 00:30:48,710 --> 00:30:51,370 So suppose this is the observed 592 00:30:51,370 --> 00:30:53,260 demand at various prices-- 593 00:30:53,260 --> 00:30:55,580 when it's free, when it's 10 shilling, when it's 20 594 00:30:55,580 --> 00:30:58,310 shilling, when it's 30 shilling. 595 00:30:58,310 --> 00:30:59,800 That's the observed demand. 596 00:30:59,800 --> 00:31:02,550 So someone could say, well, I'm going to trace a line 597 00:31:02,550 --> 00:31:06,930 here, and that's going to give you my elasticity of purchase 598 00:31:06,930 --> 00:31:09,250 with respect to the price of the good. 599 00:31:12,710 --> 00:31:17,840 Now, let's say if God had given us the observed demand, 600 00:31:17,840 --> 00:31:21,375 the demand we would observe in the region with the free bed 601 00:31:21,375 --> 00:31:24,610 net if the bed net were not free, that's the demand we 602 00:31:24,610 --> 00:31:30,470 would observe in this highest-malaria region. 603 00:31:30,470 --> 00:31:34,690 It also decreases with price, but less 604 00:31:34,690 --> 00:31:36,930 than we see in reality. 605 00:31:36,930 --> 00:31:42,660 And why does it decrease less than we see in the data, the 606 00:31:42,660 --> 00:31:43,910 yellow dots? 607 00:31:47,660 --> 00:31:52,780 So the yellow dots are what we would see in 608 00:31:52,780 --> 00:31:54,030 the high-price region-- 609 00:31:56,820 --> 00:32:00,252 so in the free region, what we would see in the zero price 610 00:32:00,252 --> 00:32:04,740 region, if the price were in fact 10, 20, 30, 40. 611 00:32:04,740 --> 00:32:08,590 So why is it flatter than the observed curve? 612 00:32:12,006 --> 00:32:12,494 Yes? 613 00:32:12,494 --> 00:32:16,410 AUDIENCE: Because it was free, [INAUDIBLE]. 614 00:32:16,410 --> 00:32:18,700 ESTHER DUFLO: So this is what is free? 615 00:32:18,700 --> 00:32:21,680 This is a bit more. 616 00:32:21,680 --> 00:32:23,400 This is the demand when it's free. 617 00:32:23,400 --> 00:32:24,840 It's very high. 618 00:32:24,840 --> 00:32:27,410 This is in the next region, where it costs 10 shilling. 619 00:32:27,410 --> 00:32:29,090 It's a bit lower. 620 00:32:29,090 --> 00:32:34,660 This point is telling us, this is how many bed nets the 621 00:32:34,660 --> 00:32:39,610 people in this region would purchase if the price was, in 622 00:32:39,610 --> 00:32:42,880 fact, 10 shilling instead of being free. 623 00:32:42,880 --> 00:32:46,120 So this and this and this, we don't observe. 624 00:32:46,120 --> 00:32:48,580 But we just say, let's say I'm telling you this is God told 625 00:32:48,580 --> 00:32:50,390 you what it would be. 626 00:32:50,390 --> 00:32:54,300 The reason why it is higher is because the region where it is 627 00:32:54,300 --> 00:32:58,750 free is a region where people need the malaria nets a lot. 628 00:32:58,750 --> 00:33:01,000 So even if they had pay 10 shilling, they 629 00:33:01,000 --> 00:33:02,880 would still buy them. 630 00:33:02,880 --> 00:33:04,950 And even if they had pay 20 shilling, they 631 00:33:04,950 --> 00:33:06,740 would still buy them. 632 00:33:06,740 --> 00:33:09,550 And even if they had to pay 30 shilling, they would still buy 633 00:33:09,550 --> 00:33:11,700 them, although less, because of course they are sensitive 634 00:33:11,700 --> 00:33:13,490 to the price. 635 00:33:13,490 --> 00:33:17,190 But the difference between the yellow point and the red point 636 00:33:17,190 --> 00:33:21,000 is the yellow point is just the effect of the price, 637 00:33:21,000 --> 00:33:25,120 whereas the red point also has the effect of the fact that 638 00:33:25,120 --> 00:33:28,330 you're more likely to have to pay for bed nets in places 639 00:33:28,330 --> 00:33:32,770 where the malaria is not such a big problem. 640 00:33:32,770 --> 00:33:37,970 So the red dot, which is what we observe in reality, 641 00:33:37,970 --> 00:33:41,360 combines the effect of the price and the fact that the 642 00:33:41,360 --> 00:33:45,090 regions where people have to pay are different. 643 00:33:45,090 --> 00:33:45,580 Yeah, Ben? 644 00:33:45,580 --> 00:33:47,750 AUDIENCE: The x-axis is price? 645 00:33:47,750 --> 00:33:48,790 ESTHER DUFLO: The x-axis is price. 646 00:33:48,790 --> 00:33:49,540 Yeah, sorry. 647 00:33:49,540 --> 00:33:50,810 I should have-- 648 00:33:50,810 --> 00:33:54,462 it's like this, 0, 10, 20, 30. 649 00:33:54,462 --> 00:33:57,382 AUDIENCE: It's still not obvious to me why if it's 650 00:33:57,382 --> 00:34:00,975 free, if you go from zero to, let's say, a million, why the 651 00:34:00,975 --> 00:34:05,725 demand for free bed nets would change [INAUDIBLE]? 652 00:34:05,725 --> 00:34:09,444 ESTHER DUFLO: No, it's not the demand for free bed nets. 653 00:34:09,444 --> 00:34:12,739 This is always going to be the demand at this price-- 654 00:34:12,739 --> 00:34:14,120 0, 10, 20. 655 00:34:14,120 --> 00:34:15,940 So it's still declining, you see? 656 00:34:15,940 --> 00:34:17,670 AUDIENCE: No, I'm saying for free bed nets. 657 00:34:17,670 --> 00:34:17,916 ESTHER DUFLO: Oh. 658 00:34:17,916 --> 00:34:18,530 For free, it's pretty high. 659 00:34:18,530 --> 00:34:21,407 AUDIENCE: Why does it change at all, if you go 660 00:34:21,407 --> 00:34:22,719 from price 0 to 30-- 661 00:34:22,719 --> 00:34:24,530 ESTHER DUFLO: Oh, you're wondering about this one. 662 00:34:27,040 --> 00:34:28,370 You're wondering about this point. 663 00:34:28,370 --> 00:34:32,429 You're wondering, in the low-malaria region, if I give 664 00:34:32,429 --> 00:34:34,744 you a bed net for free, why don't you get it? 665 00:34:34,744 --> 00:34:35,966 AUDIENCE: Well, no. 666 00:34:35,966 --> 00:34:39,017 Because the x-axis is high instances of malaria, low 667 00:34:39,017 --> 00:34:41,090 instances of malaria. 668 00:34:41,090 --> 00:34:42,610 ESTHER DUFLO: Yeah. 669 00:34:42,610 --> 00:34:43,514 AUDIENCE: You switched the price. 670 00:34:43,514 --> 00:34:47,516 I'm confused why changing how much a free bed nets costs 671 00:34:47,516 --> 00:34:51,260 would decrease the amount of purchases for free bed nets. 672 00:34:51,260 --> 00:34:53,370 ESTHER DUFLO: So what is a little tricky is that there 673 00:34:53,370 --> 00:34:55,166 are two things in the axis. 674 00:34:55,166 --> 00:34:58,900 There is the fact that this is a high-malaria region, and the 675 00:34:58,900 --> 00:35:00,950 fact that it's a low-malaria region. 676 00:35:00,950 --> 00:35:03,500 And it turns out-- but this is what is very important-- 677 00:35:03,500 --> 00:35:06,190 it turns out that the price is related to the 678 00:35:06,190 --> 00:35:08,020 incidence of malaria. 679 00:35:08,020 --> 00:35:12,260 So here, for example, in the two-region case, I have given 680 00:35:12,260 --> 00:35:15,220 the bed net for free in the region where 681 00:35:15,220 --> 00:35:16,410 the malaria was high. 682 00:35:16,410 --> 00:35:19,420 And the reason is because if I'm a good policy maker and I 683 00:35:19,420 --> 00:35:21,700 don't have that much money, that's what I will do. 684 00:35:21,700 --> 00:35:24,350 I will first give the bed net for free in the region where 685 00:35:24,350 --> 00:35:25,980 malaria is high. 686 00:35:25,980 --> 00:35:28,680 And I will ask people to pay for the bed net in the region 687 00:35:28,680 --> 00:35:30,910 where malaria is low. 688 00:35:30,910 --> 00:35:35,220 So what I observe in reality is the demand for free bed 689 00:35:35,220 --> 00:35:38,360 nets in the high-malaria region and the demand for 690 00:35:38,360 --> 00:35:42,180 expensive bed nets in the low-malaria region. 691 00:35:42,180 --> 00:35:45,910 So if I compare the difference between the two, 692 00:35:45,910 --> 00:35:47,510 and if I call this-- 693 00:35:47,510 --> 00:35:49,560 I'm combining two effects. 694 00:35:49,560 --> 00:35:53,700 I'm combining the effect of the price, which is, given the 695 00:35:53,700 --> 00:35:58,005 same region, people are more likely to get a bed net for 696 00:35:58,005 --> 00:36:04,790 free than to pay for it, and the fact that people, given 697 00:36:04,790 --> 00:36:08,670 that bed nets are more useful in a region where there is a 698 00:36:08,670 --> 00:36:11,090 lot of malaria than in a region where 699 00:36:11,090 --> 00:36:12,670 there is little malaria. 700 00:36:12,670 --> 00:36:16,580 People are more likely to get bed nets at any price-- 701 00:36:16,580 --> 00:36:20,240 the red dots are for the same price-- 702 00:36:20,240 --> 00:36:24,010 at any price, people will be more likely to get bed nets in 703 00:36:24,010 --> 00:36:27,110 the high-malaria region than in the low-malaria region. 704 00:36:27,110 --> 00:36:31,340 So if you compare within a line the yellow to the red 705 00:36:31,340 --> 00:36:34,080 dot, you get the effect of the price. 706 00:36:34,080 --> 00:36:39,000 If you compare across columns within the same color, you get 707 00:36:39,000 --> 00:36:43,240 the effect of being in high- versus low-malaria region. 708 00:36:43,240 --> 00:36:45,870 And you see they tend to go in the same direction, which is 709 00:36:45,870 --> 00:36:49,180 what creates your bias. 710 00:36:49,180 --> 00:36:52,820 All right, do you get the 711 00:36:52,820 --> 00:36:55,150 two-region, two-prices example? 712 00:36:55,150 --> 00:36:58,640 If you get the two-region, two-prices example, then this 713 00:36:58,640 --> 00:37:01,510 one is the same, but with four prices. 714 00:37:01,510 --> 00:37:03,200 Now we have four regions-- 715 00:37:03,200 --> 00:37:06,330 a region where we give them for free, a region where we 716 00:37:06,330 --> 00:37:09,000 give them for 10 shilling, a region where we give them for 717 00:37:09,000 --> 00:37:12,720 20 shilling, a region where we give them for 30 shilling. 718 00:37:12,720 --> 00:37:16,400 And this is what I observe in reality, a very steep slope. 719 00:37:16,400 --> 00:37:20,060 At 30, almost no one buys it. 720 00:37:20,060 --> 00:37:22,330 But this is due, again, to two things. 721 00:37:22,330 --> 00:37:24,060 One is that that's expensive. 722 00:37:24,060 --> 00:37:28,470 So even if malaria was a real problem, people would be less 723 00:37:28,470 --> 00:37:30,540 likely to buy them for 30. 724 00:37:30,540 --> 00:37:35,470 And, in addition, those regions where people had to 725 00:37:35,470 --> 00:37:38,660 pay a lot of money for malaria is precisely because malaria 726 00:37:38,660 --> 00:37:41,310 is not such a big problem in those regions. 727 00:37:41,310 --> 00:37:45,280 So in order to decompose those two effects, what I'm doing is 728 00:37:45,280 --> 00:37:49,080 that I'm playing God and I'm telling you, this is what the 729 00:37:49,080 --> 00:37:52,860 demand at various prices would have been in this first 730 00:37:52,860 --> 00:37:57,090 region, in the zero region, in the high-malaria region. 731 00:37:57,090 --> 00:38:00,390 So this I don't see in reality. 732 00:38:00,390 --> 00:38:02,880 In the world, I don't see that, but this is what it 733 00:38:02,880 --> 00:38:04,630 would have been. 734 00:38:04,630 --> 00:38:10,045 And it's a little flatter than this one because 735 00:38:10,045 --> 00:38:11,310 this is like the bias. 736 00:38:11,310 --> 00:38:19,400 This is saying, these are places which don't really need 737 00:38:19,400 --> 00:38:23,640 malaria, and this is the effect of prices. 738 00:38:23,640 --> 00:38:27,020 The yellow one is a true effect of prices, and the 739 00:38:27,020 --> 00:38:29,830 vertical lines are the bias that are added by the fact 740 00:38:29,830 --> 00:38:33,670 that prices are expensive in places where a bed net is not 741 00:38:33,670 --> 00:38:36,734 very useful because malaria is not so much of a problem. 742 00:38:36,734 --> 00:38:37,228 Yeah? 743 00:38:37,228 --> 00:38:38,710 AUDIENCE: I have a question. 744 00:38:38,710 --> 00:38:42,906 Why is there not a bias based on your environment and how 745 00:38:42,906 --> 00:38:45,850 much malaria there is when they're for free? 746 00:38:45,850 --> 00:38:48,560 ESTHER DUFLO: Well, when they're for free, we observe 747 00:38:48,560 --> 00:38:51,110 the true demand at free. 748 00:38:51,110 --> 00:38:54,820 So that happens to be the same point, which is, I'm giving 749 00:38:54,820 --> 00:38:56,340 you what I observe. 750 00:38:56,340 --> 00:38:59,640 And the free region is the demand when it's free. 751 00:38:59,640 --> 00:39:03,320 What I would like to know, then, is what would the demand 752 00:39:03,320 --> 00:39:08,060 in the free region be if, in fact, it had been 10 shilling? 753 00:39:08,060 --> 00:39:10,093 So this is this counter-factual, which is, 754 00:39:10,093 --> 00:39:12,500 there is something we don't observe, which is precisely 755 00:39:12,500 --> 00:39:16,370 what we need to compare to the region where, 756 00:39:16,370 --> 00:39:19,450 in fact, it is free. 757 00:39:19,450 --> 00:39:21,020 Ask as many questions as you have. 758 00:39:21,020 --> 00:39:22,130 The more questions, the better. 759 00:39:22,130 --> 00:39:22,847 Yeah? 760 00:39:22,847 --> 00:39:25,829 AUDIENCE: I was wondering why [INAUDIBLE] 761 00:39:25,829 --> 00:39:29,308 more likely to go from zero to 10 shillings. 762 00:39:29,308 --> 00:39:31,793 Do you see that the [INAUDIBLE] decrease in the 763 00:39:31,793 --> 00:39:33,043 [INAUDIBLE] 764 00:39:43,300 --> 00:39:45,900 ESTHER DUFLO: So you have-- so what is your name? 765 00:39:45,900 --> 00:39:47,130 AUDIENCE: Katja. 766 00:39:47,130 --> 00:39:47,540 ESTHER DUFLO: Again? 767 00:39:47,540 --> 00:39:48,360 AUDIENCE: Katja. 768 00:39:48,360 --> 00:39:48,920 ESTHER DUFLO: Katja. 769 00:39:48,920 --> 00:39:53,225 Katja has a question, which is, she doesn't believe that 770 00:39:53,225 --> 00:39:55,250 the elasticity is the way I drew it. 771 00:39:55,250 --> 00:39:58,475 You think that it's more like steep and then flatter? 772 00:39:58,475 --> 00:39:58,850 AUDIENCE: Yeah. 773 00:39:58,850 --> 00:39:59,630 ESTHER DUFLO: Possibly. 774 00:39:59,630 --> 00:40:01,971 That's the way I made that up. 775 00:40:01,971 --> 00:40:05,570 I consider that zero and 10 is kind of similar, but that has 776 00:40:05,570 --> 00:40:06,780 no reason to be true. 777 00:40:06,780 --> 00:40:09,870 And substantially, you're absolutely right. 778 00:40:09,870 --> 00:40:13,500 It could be that, in fact, there is a big drop in demand. 779 00:40:13,500 --> 00:40:17,580 But then the red point would probably be even lower. 780 00:40:17,580 --> 00:40:19,690 Yeah? 781 00:40:19,690 --> 00:40:23,458 AUDIENCE: Why were the experiments conducted so that 782 00:40:23,458 --> 00:40:26,284 you only get the orange data points? 783 00:40:26,284 --> 00:40:29,300 Why did you not mix up different communities that 784 00:40:29,300 --> 00:40:31,071 have had malaria and [INAUDIBLE] malaria 785 00:40:31,071 --> 00:40:32,037 [? for each-- ?] 786 00:40:32,037 --> 00:40:33,490 ESTHER DUFLO: That's an excellent point. 787 00:40:33,490 --> 00:40:38,650 So this is not supposed to have come from an experiment. 788 00:40:38,650 --> 00:40:39,650 This is the real world. 789 00:40:39,650 --> 00:40:40,320 [INAUDIBLE] 790 00:40:40,320 --> 00:40:43,490 a real world outside of an experiment. 791 00:40:43,490 --> 00:40:45,520 And outside of an experiment in the real 792 00:40:45,520 --> 00:40:47,250 world, it is likely-- 793 00:40:47,250 --> 00:40:49,510 it doesn't have to be the case, but it's likely that 794 00:40:49,510 --> 00:40:53,080 policy maker will make the bed net available for free in 795 00:40:53,080 --> 00:40:56,530 high-malaria region, because it makes sense for them. 796 00:40:56,530 --> 00:41:00,500 But your question exactly brings to what the answer to 797 00:41:00,500 --> 00:41:02,320 the problem is. 798 00:41:02,320 --> 00:41:07,100 So we have a problem, which is that we observe the low prices 799 00:41:07,100 --> 00:41:10,870 in places where the need is high, and high prices in the 800 00:41:10,870 --> 00:41:12,710 regions where they need is low, so that 801 00:41:12,710 --> 00:41:14,670 leads to this bias. 802 00:41:14,670 --> 00:41:18,180 And the solution is exactly what you suggested, which is, 803 00:41:18,180 --> 00:41:21,680 let's have variation in prices in regions where 804 00:41:21,680 --> 00:41:23,090 the need is the same. 805 00:41:23,090 --> 00:41:24,584 Yes? 806 00:41:24,584 --> 00:41:27,572 AUDIENCE: So on the x-axis, are you exactly correlating 807 00:41:27,572 --> 00:41:32,440 the price of the bed nets with the incidence of malaria, or-- 808 00:41:32,440 --> 00:41:36,040 ESTHER DUFLO: So what I'm assuming, I'm telling you this 809 00:41:36,040 --> 00:41:37,110 doesn't have to be true. 810 00:41:37,110 --> 00:41:39,600 And the problem is we'll never know, because we won't observe 811 00:41:39,600 --> 00:41:41,060 both in the real world. 812 00:41:41,060 --> 00:41:44,910 But I'm telling you this happens to be like that in 813 00:41:44,910 --> 00:41:47,590 this particular fictional world that I constructed, 814 00:41:47,590 --> 00:41:50,177 where this was like that. 815 00:41:50,177 --> 00:41:52,278 AUDIENCE: But if you go along the x-axis to the right, you 816 00:41:52,278 --> 00:41:54,847 mean that bed nets get more expensive and malaria 817 00:41:54,847 --> 00:41:56,250 incidence goes up? 818 00:41:56,250 --> 00:41:57,600 ESTHER DUFLO: Goes down. 819 00:41:57,600 --> 00:41:58,610 AUDIENCE: I'm sorry, goes down? 820 00:41:58,610 --> 00:41:59,120 ESTHER DUFLO: Exactly. 821 00:41:59,120 --> 00:42:00,060 AUDIENCE: OK. 822 00:42:00,060 --> 00:42:01,310 ESTHER DUFLO: I constructed it that way. 823 00:42:04,300 --> 00:42:07,220 You could construct the opposite example, and you 824 00:42:07,220 --> 00:42:10,920 could make up a story where the richer regions have more 825 00:42:10,920 --> 00:42:14,470 political power, and they also have less malaria. 826 00:42:14,470 --> 00:42:17,780 So the richer region will get the free bed nets, and they 827 00:42:17,780 --> 00:42:19,560 will also have less malaria. 828 00:42:19,560 --> 00:42:23,550 In that case, where would the observed compared to the true 829 00:42:23,550 --> 00:42:27,580 elasticity, where would the red dot be, if, in fact, the 830 00:42:27,580 --> 00:42:28,890 correlation was the other way? 831 00:42:33,840 --> 00:42:34,830 AUDIENCE: They would go the other way? 832 00:42:34,830 --> 00:42:36,070 ESTHER DUFLO: They would go the other way. 833 00:42:36,070 --> 00:42:44,020 So the red dot would be above, because the bed nets are more 834 00:42:44,020 --> 00:42:46,660 expensive where they are more needed. 835 00:42:46,660 --> 00:42:50,140 And your question brings in an essential question, which is, 836 00:42:50,140 --> 00:42:52,880 sadly, we can't know. 837 00:42:52,880 --> 00:42:55,510 We can know some stuff, which is we can, for example, 838 00:42:55,510 --> 00:42:57,360 measure malaria prevalence. 839 00:42:57,360 --> 00:43:04,050 But we never observe the counter-factual. 840 00:43:04,050 --> 00:43:07,890 So either we observe the demand, where the price is 841 00:43:07,890 --> 00:43:14,010 high, in the places where it's high, or we observe the demand 842 00:43:14,010 --> 00:43:16,570 where the price is low in the places where it's low. 843 00:43:16,570 --> 00:43:20,950 And we never get to see the demand where the price is low 844 00:43:20,950 --> 00:43:22,730 in the places where it's high. 845 00:43:22,730 --> 00:43:24,840 That's the missing data point. 846 00:43:24,840 --> 00:43:27,270 That's why this is what we are missing. 847 00:43:27,270 --> 00:43:30,000 In other words, we don't see those yellow points. 848 00:43:30,000 --> 00:43:34,410 We don't see what the demand would have been in a zero 849 00:43:34,410 --> 00:43:38,440 price region if the price, in fact, had been 10, 20, 30. 850 00:43:38,440 --> 00:43:39,690 We just don't know. 851 00:43:43,980 --> 00:43:48,840 So the solution is what you suggested, which is to say, 852 00:43:48,840 --> 00:43:52,950 why don't we take one region and try to vary the prices in 853 00:43:52,950 --> 00:43:53,520 those regions? 854 00:43:53,520 --> 00:43:56,385 Or in fact, we could take two regions, a high-malaria region 855 00:43:56,385 --> 00:43:59,190 and a low-malaria region, and vary the prices 856 00:43:59,190 --> 00:44:01,080 in those two regions. 857 00:44:01,080 --> 00:44:04,070 And this is where the word experiment comes up. 858 00:44:04,070 --> 00:44:07,160 This is where the experiment is, which is the way you would 859 00:44:07,160 --> 00:44:11,910 do it to be completely sure that there is no difference 860 00:44:11,910 --> 00:44:15,690 between the places or the people which got the bed net 861 00:44:15,690 --> 00:44:18,780 for free, and the places or people which had to pay for 862 00:44:18,780 --> 00:44:23,385 it, is you would randomly assign the price of the bed 863 00:44:23,385 --> 00:44:27,470 net to different regions or two different people. 864 00:44:27,470 --> 00:44:32,200 So this is what is done, for example, in medicine when you 865 00:44:32,200 --> 00:44:34,100 want when you want to find out the 866 00:44:34,100 --> 00:44:37,010 effectiveness of a new drug. 867 00:44:37,010 --> 00:44:41,020 You don't give the new drug to people who need it the most. 868 00:44:41,020 --> 00:44:42,620 Otherwise, you would find that the drug 869 00:44:42,620 --> 00:44:44,810 killed a lot of people. 870 00:44:44,810 --> 00:44:48,800 You take some part of people who need the drug, and you 871 00:44:48,800 --> 00:44:51,680 randomly assign some people to take the drug and some people 872 00:44:51,680 --> 00:44:53,090 to get the placebo. 873 00:44:53,090 --> 00:44:55,810 You might also take a sample of people who need the drug 874 00:44:55,810 --> 00:45:00,020 less, and randomly assign the drug within them, to find out 875 00:45:00,020 --> 00:45:02,240 side effect if you misprescribe. 876 00:45:02,240 --> 00:45:05,530 But you wouldn't give the drug to people who need it, and not 877 00:45:05,530 --> 00:45:08,085 the drug to people who don't need it, otherwise all of the 878 00:45:08,085 --> 00:45:10,790 drugs would be sure to kill a lot of people. 879 00:45:10,790 --> 00:45:17,380 So most evaluation of social policy is trying to estimate 880 00:45:17,380 --> 00:45:23,030 the effect of policies by doing these kind of 881 00:45:23,030 --> 00:45:26,530 comparisons, this point to this point. 882 00:45:26,530 --> 00:45:30,215 And that creates this bias situation. 883 00:45:32,720 --> 00:45:36,130 And in fact, you can try and make more progress on this 884 00:45:36,130 --> 00:45:40,620 question by creating experiments where you fix the 885 00:45:40,620 --> 00:45:43,570 region, or you fix the environment, and you randomly 886 00:45:43,570 --> 00:45:45,000 assign the prices. 887 00:45:45,000 --> 00:45:47,740 The randomization ensures that there is nothing 888 00:45:47,740 --> 00:45:51,450 systematically different between the people who got the 889 00:45:51,450 --> 00:45:54,130 free bed net and people who didn't get it. 890 00:45:54,130 --> 00:45:56,970 In other words, there is no-- 891 00:45:56,970 --> 00:46:01,790 if we go back to this graph, for example, the places that 892 00:46:01,790 --> 00:46:04,900 got the free bed net and the places that had to pay for the 893 00:46:04,900 --> 00:46:08,170 bed net, if they had gotten it for free, they would be at the 894 00:46:08,170 --> 00:46:09,460 same level. 895 00:46:09,460 --> 00:46:11,300 And if they had pay for it, they would 896 00:46:11,300 --> 00:46:12,530 be at the same level. 897 00:46:12,530 --> 00:46:14,990 There is now no difference between the region who paid 898 00:46:14,990 --> 00:46:19,430 for the net and the region who didn't pay for the net. 899 00:46:19,430 --> 00:46:25,170 One point before going into the results of what Cohen, and 900 00:46:25,170 --> 00:46:31,170 then Pascaline Dupas found for the bed net is that of course, 901 00:46:31,170 --> 00:46:34,260 this is not true if it's just one village and another 902 00:46:34,260 --> 00:46:36,080 village, like Steve suggested. 903 00:46:36,080 --> 00:46:39,310 Why is it, if we took only one village, and in one case we 904 00:46:39,310 --> 00:46:42,640 make free, one case we make not free-- we can flip a coin, 905 00:46:42,640 --> 00:46:45,560 and then we compare-- 906 00:46:45,560 --> 00:46:47,200 why is that problematic? 907 00:46:47,200 --> 00:46:47,900 Two villages-- 908 00:46:47,900 --> 00:46:51,516 why is two villages not enough? 909 00:46:51,516 --> 00:46:53,262 AUDIENCE: Because you can't make sure that everything is 910 00:46:53,262 --> 00:46:56,007 the same [INAUDIBLE] 911 00:46:56,007 --> 00:46:59,999 like one village might have more [INAUDIBLE] 912 00:46:59,999 --> 00:47:03,492 actually care about each other, or one village might 913 00:47:03,492 --> 00:47:07,458 end up having a really rainy season, so all the mosquitoes 914 00:47:07,458 --> 00:47:09,930 happen to swamp around it. 915 00:47:09,930 --> 00:47:10,540 ESTHER DUFLO: Exactly. 916 00:47:10,540 --> 00:47:13,500 Two villages, even if there is nothing systematically 917 00:47:13,500 --> 00:47:16,510 different, because you've thrown a coin, so you didn't 918 00:47:16,510 --> 00:47:19,730 choose one village rather than the other for their 919 00:47:19,730 --> 00:47:22,670 characteristic to give the bed net for free or to ask people 920 00:47:22,670 --> 00:47:24,090 to pay for them. 921 00:47:24,090 --> 00:47:25,860 Even in that case, there are things that are different 922 00:47:25,860 --> 00:47:29,170 across villages that you can't fully control. 923 00:47:29,170 --> 00:47:32,220 So you need the samples to be large enough such that all the 924 00:47:32,220 --> 00:47:35,230 other differences between people or between regions 925 00:47:35,230 --> 00:47:39,670 average out, so that even though you do not observe what 926 00:47:39,670 --> 00:47:43,910 the demand for bed net would have been in the free region 927 00:47:43,910 --> 00:47:46,460 when actually you ask people to pay for it, you would not 928 00:47:46,460 --> 00:47:47,340 observe that. 929 00:47:47,340 --> 00:47:51,060 But you can be confident that what the demand you observe in 930 00:47:51,060 --> 00:47:56,010 the other region where it was free gives you the right 931 00:47:56,010 --> 00:47:57,090 counter-factual. 932 00:47:57,090 --> 00:47:59,580 And for that, the sample needs to be large enough so that all 933 00:47:59,580 --> 00:48:02,940 the noise and the idiosyncrasies get swamped. 934 00:48:05,740 --> 00:48:10,180 So let's look at two sets of experiments. 935 00:48:10,180 --> 00:48:16,210 Both were Pascaline Dupas, who is a professor at UCLA, was is 936 00:48:16,210 --> 00:48:17,900 involved in both of them. 937 00:48:17,900 --> 00:48:20,670 The first one was conducted with Jessica Cohen. 938 00:48:23,860 --> 00:48:25,330 In and around this place, which is 939 00:48:25,330 --> 00:48:27,500 called Busia in Kenya. 940 00:48:27,500 --> 00:48:30,560 This hotel, you can see that the beginning of the hotel is 941 00:48:30,560 --> 00:48:36,440 called "Blue York," where I spent a lot of time. 942 00:48:36,440 --> 00:48:39,190 And so what did they do? 943 00:48:39,190 --> 00:48:43,640 They randomly chose clinics, prenatal clinics, they gave 944 00:48:43,640 --> 00:48:47,070 them nets to give to people who come for their prenatal 945 00:48:47,070 --> 00:48:51,770 exam, either at zero or 20 shilling or 30 946 00:48:51,770 --> 00:48:53,240 shilling or 40 shilling. 947 00:48:53,240 --> 00:48:56,840 So the unit of randomization is the clinic. 948 00:48:56,840 --> 00:49:02,150 And then they compare how many bed nets they sold, and 949 00:49:02,150 --> 00:49:04,950 whether the bed net got used, and therefore the effective 950 00:49:04,950 --> 00:49:07,570 coverage in those clinics. 951 00:49:07,570 --> 00:49:09,560 And this is what they find. 952 00:49:09,560 --> 00:49:12,180 So the little vertical line here is 953 00:49:12,180 --> 00:49:13,840 the confidence interval. 954 00:49:13,840 --> 00:49:15,800 So it's taking into account all the noise that 955 00:49:15,800 --> 00:49:18,010 exists in the data. 956 00:49:18,010 --> 00:49:21,910 And so what is the conclusion from this graph, at free, 10 957 00:49:21,910 --> 00:49:23,390 shilling, 20 shilling, 40 shilling? 958 00:49:28,570 --> 00:49:34,225 This is the number of nets sold per week, I think. 959 00:49:34,225 --> 00:49:37,195 AUDIENCE: Does that indicate if the customers actually use 960 00:49:37,195 --> 00:49:38,190 the bed net? 961 00:49:38,190 --> 00:49:39,410 ESTHER DUFLO: No. 962 00:49:39,410 --> 00:49:44,840 So far, this is the first look at, which is just how many bed 963 00:49:44,840 --> 00:49:47,906 nets were given out. 964 00:49:47,906 --> 00:49:48,358 Yeah? 965 00:49:48,358 --> 00:49:49,262 AUDIENCE: [INAUDIBLE] 966 00:49:49,262 --> 00:49:51,070 the demand curves downward. 967 00:49:51,070 --> 00:49:52,740 ESTHER DUFLO: The demand curves downward, and it's 968 00:49:52,740 --> 00:49:54,420 quite steep, right? 969 00:49:54,420 --> 00:49:56,840 Basically, at 40 shilling, which is still heavily 970 00:49:56,840 --> 00:50:02,086 subsidized, because the full price of a bed net is more 971 00:50:02,086 --> 00:50:07,680 like 80 shilling, you distribute only about 10 bed 972 00:50:07,680 --> 00:50:11,310 nets per week, versus 40 per week, when you 973 00:50:11,310 --> 00:50:13,170 give them for free. 974 00:50:13,170 --> 00:50:16,580 So it goes down quite steeply. 975 00:50:16,580 --> 00:50:19,470 And I remember, because they were randomly selected, we 976 00:50:19,470 --> 00:50:23,300 know that it is due to the price, and not due to 977 00:50:23,300 --> 00:50:24,970 differences across those regions. 978 00:50:24,970 --> 00:50:25,463 Yeah? 979 00:50:25,463 --> 00:50:25,956 AUDIENCE: Yeah. 980 00:50:25,956 --> 00:50:29,407 That's funny, because the error bars are pretty big. 981 00:50:29,407 --> 00:50:31,379 Like, if you if you looked at the lower error, versus the 982 00:50:31,379 --> 00:50:33,850 higher error, it's [INAUDIBLE]. 983 00:50:33,850 --> 00:50:39,240 ESTHER DUFLO: So you take this one compared to this one. 984 00:50:39,240 --> 00:50:41,340 That's the confidence interval. 985 00:50:41,340 --> 00:50:45,150 So we are 95% confident that the truth is somewhere in 986 00:50:45,150 --> 00:50:46,610 between there. 987 00:50:46,610 --> 00:50:49,400 That the 95 confidence interval, [INAUDIBLE]. 988 00:50:52,410 --> 00:50:53,590 Right? 989 00:50:53,590 --> 00:50:55,535 So you can know that this is systematically 990 00:50:55,535 --> 00:50:56,785 different from this. 991 00:50:59,626 --> 00:51:02,500 You're right that this and this, the 992 00:51:02,500 --> 00:51:03,590 confidence interval overlaps. 993 00:51:03,590 --> 00:51:06,620 And in fact, the numbers are quite similar. 994 00:51:06,620 --> 00:51:09,140 This is why I don't answer someone who was asking why 995 00:51:09,140 --> 00:51:11,770 didn't I draw the elasticity that way I'd drawn it that 996 00:51:11,770 --> 00:51:14,740 way, because I already knew the answer to the question. 997 00:51:14,740 --> 00:51:17,290 I was ahead of the game, slightly. 998 00:51:17,290 --> 00:51:19,370 So zero to 10 is not such a big difference. 999 00:51:19,370 --> 00:51:21,500 Maybe 10 is not that expensive. 1000 00:51:21,500 --> 00:51:25,315 But 40 is definitely different from free. 1001 00:51:25,315 --> 00:51:25,970 You're right. 1002 00:51:25,970 --> 00:51:28,060 Certainly we would like those bars to be smaller. 1003 00:51:28,060 --> 00:51:30,142 If we had more observations, they would be. 1004 00:51:30,142 --> 00:51:31,910 They would be even more precise. 1005 00:51:31,910 --> 00:51:32,552 Yes? 1006 00:51:32,552 --> 00:51:35,996 AUDIENCE: Were the clinics far apart from each other? 1007 00:51:35,996 --> 00:51:39,276 It could be, say, I'd go to the clinic where I'd have to 1008 00:51:39,276 --> 00:51:40,916 pay 20 shillings [INAUDIBLE] 1009 00:51:44,370 --> 00:51:44,590 ESTHER DUFLO: Right. 1010 00:51:44,590 --> 00:51:45,710 That's a very good point. 1011 00:51:45,710 --> 00:51:48,150 The people shift from one clinic to the other. 1012 00:51:48,150 --> 00:51:50,020 So they couldn't really do that, because the clinics were 1013 00:51:50,020 --> 00:51:51,200 quite far away. 1014 00:51:51,200 --> 00:51:53,610 On the other hand, what they did find is a lot more people 1015 00:51:53,610 --> 00:51:57,430 coming to the clinic, presumably to get the bed net, 1016 00:51:57,430 --> 00:52:00,680 which is a good thing, because it cost subsidizes other 1017 00:52:00,680 --> 00:52:02,810 services that you get in the clinic, in 1018 00:52:02,810 --> 00:52:06,930 particular a test for HIV. 1019 00:52:06,930 --> 00:52:10,350 And if the mother is HIV positive, then you can give 1020 00:52:10,350 --> 00:52:13,400 them Nevirapine just before the delivery to prevent the 1021 00:52:13,400 --> 00:52:15,380 transmission to the [INAUDIBLE]. 1022 00:52:15,380 --> 00:52:20,990 So that's another story, which is a reason why you might want 1023 00:52:20,990 --> 00:52:23,240 to distribute bed nets if people like them. 1024 00:52:23,240 --> 00:52:26,130 Now, this was your question, which is, but are people are 1025 00:52:26,130 --> 00:52:28,610 using them? 1026 00:52:28,610 --> 00:52:31,680 And so far, the usage, you can ask people what they tell you. 1027 00:52:31,680 --> 00:52:33,085 But they could lie. 1028 00:52:33,085 --> 00:52:35,900 Or you can try and go to their home and see whether you see 1029 00:52:35,900 --> 00:52:37,050 the bed net. 1030 00:52:37,050 --> 00:52:39,440 And here, what's the conclusion? 1031 00:52:39,440 --> 00:52:42,670 So this is conditional, having gotten a bed net. 1032 00:52:42,670 --> 00:52:45,150 So what they did that they took the register. 1033 00:52:45,150 --> 00:52:48,340 They saw who got a bed net, and they went to their home 1034 00:52:48,340 --> 00:52:49,940 and said, what are you doing with the bed net? 1035 00:52:49,940 --> 00:52:53,510 Is it a fishing net or are you using it for a bed net? 1036 00:52:53,510 --> 00:52:55,382 And what do they find? 1037 00:52:55,382 --> 00:52:58,090 AUDIENCE: What's the y-axis? 1038 00:52:58,090 --> 00:53:01,750 ESTHER DUFLO: So it's percentage divided by 10. 1039 00:53:01,750 --> 00:53:07,728 So it's 0%, 20%, 40%, 60%, 80%, 100%. 1040 00:53:07,728 --> 00:53:08,724 Ben? 1041 00:53:08,724 --> 00:53:10,716 AUDIENCE: I'm a little confused about the why. 1042 00:53:10,716 --> 00:53:15,198 So 10 is interesting to me, like why 50% of the people who 1043 00:53:15,198 --> 00:53:18,200 purchase the bed nets actually use the bed nets. 1044 00:53:18,200 --> 00:53:19,620 ESTHER DUFLO: So a bit more, it's more like 60%. 1045 00:53:22,160 --> 00:53:23,440 AUDIENCE: For 10? 1046 00:53:23,440 --> 00:53:25,270 ESTHER DUFLO: Yeah, for 10 shilling. 1047 00:53:25,270 --> 00:53:27,540 It's about 60%. 1048 00:53:27,540 --> 00:53:28,750 AUDIENCE: Can you help me understand 1049 00:53:28,750 --> 00:53:30,370 why that trend happens? 1050 00:53:30,370 --> 00:53:32,092 ESTHER DUFLO: Oh, why is it going down? 1051 00:53:32,092 --> 00:53:33,300 AUDIENCE: Right. 1052 00:53:33,300 --> 00:53:33,950 ESTHER DUFLO: It's not-- 1053 00:53:33,950 --> 00:53:36,350 I mean, going back to the question earlier, it's not 1054 00:53:36,350 --> 00:53:37,600 significantly different. 1055 00:53:39,890 --> 00:53:42,000 So it's kind of all similar. 1056 00:53:42,000 --> 00:53:46,030 So overall, about the trend, what do you have to say? 1057 00:53:49,105 --> 00:53:49,955 This is pretty similar. 1058 00:53:49,955 --> 00:53:51,030 It seems pretty flat. 1059 00:53:51,030 --> 00:53:54,870 So there is no very big evidence that how much of paid 1060 00:53:54,870 --> 00:53:56,936 for a bed net influenced whether or not you're going to 1061 00:53:56,936 --> 00:53:59,400 use it or not. 1062 00:53:59,400 --> 00:54:01,575 And it's not 100%. 1063 00:54:01,575 --> 00:54:04,970 It's more like towards 60% to 70%, so not everybody 1064 00:54:04,970 --> 00:54:06,760 uses the bed net. 1065 00:54:06,760 --> 00:54:11,580 But how much they pay for it doesn't affect whether or not 1066 00:54:11,580 --> 00:54:13,150 they use it or not. 1067 00:54:13,150 --> 00:54:16,020 Interestingly, when they went back six months to a year 1068 00:54:16,020 --> 00:54:18,700 later, more people are likely to use them. 1069 00:54:18,700 --> 00:54:22,380 So some people just took them because they were free, maybe, 1070 00:54:22,380 --> 00:54:25,660 or they were not very expensive, and decided to 1071 00:54:25,660 --> 00:54:28,030 store them for future usage, which is also 1072 00:54:28,030 --> 00:54:29,630 something you can do. 1073 00:54:29,630 --> 00:54:33,580 So people do not seem to not use the bed net because they 1074 00:54:33,580 --> 00:54:35,306 got them for free. 1075 00:54:35,306 --> 00:54:37,910 And the result is that if you look at the effective 1076 00:54:37,910 --> 00:54:41,290 coverage, the effective coverage is very strongly 1077 00:54:41,290 --> 00:54:44,190 declining with price. 1078 00:54:44,190 --> 00:54:48,630 So the conclusion from their study is that you should give 1079 00:54:48,630 --> 00:54:51,570 the bed nets for free, because you get many more bed nets 1080 00:54:51,570 --> 00:54:54,020 out, you're much more likely to go-- 1081 00:54:54,020 --> 00:54:55,600 if you get them for free, you're going to 1082 00:54:55,600 --> 00:54:58,350 be above 60% coverage. 1083 00:54:58,350 --> 00:55:03,730 And once you go above 60% coverage, you're more likely 1084 00:55:03,730 --> 00:55:06,000 to buy them. 1085 00:55:06,000 --> 00:55:07,770 But before you go, I want to give you the 1086 00:55:07,770 --> 00:55:11,680 epilogue, which is-- 1087 00:55:11,680 --> 00:55:14,390 so this is the reading I asked you to see, which is people-- 1088 00:55:14,390 --> 00:55:18,310 so this got posted on Daniel Roderick's blog, under the 1089 00:55:18,310 --> 00:55:22,180 title "Jeff Sachs Vindicated," which is to say, oh, yeah, we 1090 00:55:22,180 --> 00:55:23,140 should give bed nets for free. 1091 00:55:23,140 --> 00:55:25,120 This experiment showed that. 1092 00:55:25,120 --> 00:55:28,610 And then people complained, and said, this is only 1093 00:55:28,610 --> 00:55:33,010 pregnant women, and pregnant women all need the bed net, so 1094 00:55:33,010 --> 00:55:34,860 they are going to use it at any price. 1095 00:55:34,860 --> 00:55:37,990 The sunk cost effect won't be there for pregnant women. 1096 00:55:37,990 --> 00:55:41,690 And secondly, it's Kenya, where bed nets are still quite 1097 00:55:41,690 --> 00:55:43,060 well-known. 1098 00:55:43,060 --> 00:55:52,870 So what are people objecting? 1099 00:55:52,870 --> 00:55:55,740 People are objecting to saying, the experiment gives 1100 00:55:55,740 --> 00:55:57,460 you a lot of internal validity. 1101 00:55:57,460 --> 00:56:01,300 We know now that in the specific context of clinics in 1102 00:56:01,300 --> 00:56:09,050 western Kenya with pregnant mothers, there is no effect of 1103 00:56:09,050 --> 00:56:11,170 the price you pay on whether you use. 1104 00:56:11,170 --> 00:56:15,360 But maybe this doesn't apply to Uganda, Maybe it doesn't 1105 00:56:15,360 --> 00:56:18,490 affect non-pregnant women, et cetera, et cetera. 1106 00:56:18,490 --> 00:56:21,360 So this is a question of external validity. 1107 00:56:21,360 --> 00:56:22,610 So what's the solution? 1108 00:56:25,050 --> 00:56:27,440 When these kind of questions are raised, what's the 1109 00:56:27,440 --> 00:56:31,840 solution to make progress? 1110 00:56:31,840 --> 00:56:33,080 You've can't ignore these questions. 1111 00:56:33,080 --> 00:56:33,870 They are very good questions. 1112 00:56:33,870 --> 00:56:36,270 What's the solution to make progress? 1113 00:56:36,270 --> 00:56:36,560 Yes? 1114 00:56:36,560 --> 00:56:38,170 AUDIENCE: Try it in other areas. 1115 00:56:38,170 --> 00:56:40,130 ESTHER DUFLO: You want to try it in other areas. 1116 00:56:40,130 --> 00:56:42,630 So in fact, in bed nets, this got done. 1117 00:56:42,630 --> 00:56:45,740 So this got tried first in different countries, so in 1118 00:56:45,740 --> 00:56:47,550 Uganda, in Madagascar. 1119 00:56:47,550 --> 00:56:52,830 And secondly, it got tried in Kenya, but as I was going to 1120 00:56:52,830 --> 00:56:56,810 say, real people, not pregnant women, but people from the 1121 00:56:56,810 --> 00:56:58,740 market, men and women. 1122 00:56:58,740 --> 00:57:03,070 And here, vouchers got given at various prices to 1123 00:57:03,070 --> 00:57:09,490 individuals for them to pick them up in the shop. 1124 00:57:09,490 --> 00:57:13,110 So now this is an environment where have you would have 1125 00:57:13,110 --> 00:57:17,140 expected to be more likely to find a sunk cost effect. 1126 00:57:17,140 --> 00:57:20,080 Moreover, the experiment was also done to look at these 1127 00:57:20,080 --> 00:57:22,470 longer-term consequences. 1128 00:57:22,470 --> 00:57:26,640 So after six months, they went back again to the same people, 1129 00:57:26,640 --> 00:57:31,960 and tried to sell them a bed net for 100 shillings, so not 1130 00:57:31,960 --> 00:57:34,700 the lowest price, and not the highest price. 1131 00:57:34,700 --> 00:57:38,080 And so the question is, the people who got the first bed 1132 00:57:38,080 --> 00:57:42,290 net for free, are they more likely to buy the second one 1133 00:57:42,290 --> 00:57:44,670 for 100 shilling, or are they less likely to buy the second 1134 00:57:44,670 --> 00:57:46,990 one for 100 shilling? 1135 00:57:46,990 --> 00:57:50,160 So I'll run you to the result quickly. 1136 00:57:50,160 --> 00:57:57,930 So they find the same effect of price on purchase, that the 1137 00:57:57,930 --> 00:58:01,580 more you pay, the less you get, quite steeply. 1138 00:58:01,580 --> 00:58:04,430 They find the same lack of effect of price 1139 00:58:04,430 --> 00:58:05,800 on whether you use. 1140 00:58:05,800 --> 00:58:08,240 And so this is not pregnant woman anymore. 1141 00:58:08,240 --> 00:58:12,090 So this kind of generalizes a bit the result. 1142 00:58:12,090 --> 00:58:16,890 And this is the effect on future approaches, the 1143 00:58:16,890 --> 00:58:19,510 confidence bar, a little bit high. 1144 00:58:19,510 --> 00:58:23,910 But if anything, what you find is that people are more likely 1145 00:58:23,910 --> 00:58:29,195 to buy a second bed net when they got it for free or almost 1146 00:58:29,195 --> 00:58:33,030 free than when they were offered the 1147 00:58:33,030 --> 00:58:34,940 option to pay for one. 1148 00:58:34,940 --> 00:58:37,470 And this is despite the fact that at that time, they 1149 00:58:37,470 --> 00:58:39,180 already had a bed net. 1150 00:58:39,180 --> 00:58:42,770 So presumably, their need for a second bed net is less so. 1151 00:58:42,770 --> 00:58:46,460 So it seems that the learning effect dominates the 1152 00:58:46,460 --> 00:58:48,220 entitlement effect. 1153 00:58:48,220 --> 00:58:51,210 So if we go back to the wrecking market story, 1154 00:58:51,210 --> 00:58:55,100 sustainability question you could say, well actually, 1155 00:58:55,100 --> 00:58:58,580 giving a sweep of bed nets for free might be a good way to 1156 00:58:58,580 --> 00:58:59,930 create a market. 1157 00:58:59,930 --> 00:59:03,310 Because you create a demand among people who start 1158 00:59:03,310 --> 00:59:05,390 learning and say, actually, it's kind of nice, 1159 00:59:05,390 --> 00:59:07,490 that bed nets idea. 1160 00:59:07,490 --> 00:59:10,120 The last way in which you create a demand is by creating 1161 00:59:10,120 --> 00:59:13,170 this first user of a bed net. 1162 00:59:13,170 --> 00:59:16,490 So this is the last card that I'm going to show you. 1163 00:59:16,490 --> 00:59:20,260 This is the purchase of a bed net if you have to pay for it 1164 00:59:20,260 --> 00:59:24,100 if none of your neighbors got it for free. 1165 00:59:24,100 --> 00:59:29,250 And this is a purchase of a bed net, if you had to pay for 1166 00:59:29,250 --> 00:59:32,950 it, if all of your neighbors got it for free. 1167 00:59:32,950 --> 00:59:37,975 And that is randomly assigned, because depending on who your 1168 00:59:37,975 --> 00:59:40,050 neighbors are, where they are located, what kind of voucher 1169 00:59:40,050 --> 00:59:43,680 they get, this is just random whether they get this one. 1170 00:59:43,680 --> 00:59:47,920 And what you see is that if your neighbor got bed nets for 1171 00:59:47,920 --> 00:59:49,990 free, you're more likely to get one. 1172 00:59:49,990 --> 00:59:53,470 So there is actually a positive social effect. 1173 00:59:53,470 --> 00:59:56,830 So the reason why I wanted detail into this experiment is 1174 00:59:56,830 --> 01:00:01,380 that that's an experiment that shows us how we can use 1175 01:00:01,380 --> 01:00:04,630 experiments not only to evaluate the impact of a 1176 01:00:04,630 --> 01:00:07,910 policy, but the answer those questions that help us 1177 01:00:07,910 --> 01:00:13,530 understand how people decide whether the entitlement effect 1178 01:00:13,530 --> 01:00:16,580 is bigger or smaller than the learning effect, whether the 1179 01:00:16,580 --> 01:00:20,460 sunk cost fallacy exists, whether how the big the price 1180 01:00:20,460 --> 01:00:22,170 elasticities are, et cetera. 1181 01:00:22,170 --> 01:00:24,210 And we need to design the experiment to 1182 01:00:24,210 --> 01:00:25,720 answer all those questions. 1183 01:00:25,720 --> 01:00:28,820 And then we can start informing the policy debate. 1184 01:00:28,820 --> 01:00:30,070 Sorry for running over.