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 --> 00:00:10,650 offer high quality educational resources for free. 5 00:00:10,650 --> 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 00:00:27,040 --> 00:00:29,192 PROFESSOR: While people trickle in-- 9 00:00:29,192 --> 00:00:33,160 This slide is my attempt to summarize what we said last 10 00:00:33,160 --> 00:00:36,850 time, or you said last time, mostly. 11 00:00:36,850 --> 00:00:42,500 So it's basically, the movie is making a case for 12 00:00:42,500 --> 00:00:43,740 microfinance. 13 00:00:43,740 --> 00:00:47,380 And the case is all these different points. 14 00:00:50,230 --> 00:00:53,110 They're building a case for a particular version of 15 00:00:53,110 --> 00:00:54,420 microfinance. 16 00:00:54,420 --> 00:00:57,340 And they're making this set of points. 17 00:00:57,340 --> 00:00:59,490 They start with the point that poor have 18 00:00:59,490 --> 00:01:00,635 poor access to credit. 19 00:01:00,635 --> 00:01:03,690 That they're reliable borrowers. 20 00:01:03,690 --> 00:01:07,850 That maybe group lending has something to do with the 21 00:01:07,850 --> 00:01:09,200 reliability. 22 00:01:09,200 --> 00:01:14,910 That microfinance can lift people out of poverty. 23 00:01:14,910 --> 00:01:17,375 The way it does it is through new businesses. 24 00:01:21,600 --> 00:01:24,960 And the fact that this is these new business generate 25 00:01:24,960 --> 00:01:28,220 increased earnings of women. 26 00:01:28,220 --> 00:01:34,260 Has the implication that there is then greater investment in 27 00:01:34,260 --> 00:01:35,510 human capital. 28 00:01:39,920 --> 00:01:44,720 And finally, there is a implied but also, I think, 29 00:01:44,720 --> 00:01:50,640 important claim that there's no conflict between the 30 00:01:50,640 --> 00:01:55,600 commercialization of microcredit and an emphasis on 31 00:01:55,600 --> 00:01:56,810 its social role. 32 00:01:56,810 --> 00:01:59,460 So those are the claims being made. 33 00:01:59,460 --> 00:02:04,720 So what I want to do today is mostly go through the thinking 34 00:02:04,720 --> 00:02:08,789 behind what's the argument behind each of these. 35 00:02:08,789 --> 00:02:12,330 And then try to look at the data a bit. 36 00:02:12,330 --> 00:02:16,050 So I'm going to now tell you a little bit about an evaluation 37 00:02:16,050 --> 00:02:23,420 we did of a program, of an organization called Spandana. 38 00:02:23,420 --> 00:02:24,890 It's a very traditional program. 39 00:02:24,890 --> 00:02:28,020 Their group liability, mostly weekly payment. 40 00:02:28,020 --> 00:02:32,240 Starting loan is about 10,000 rupees. 41 00:02:32,240 --> 00:02:36,820 Interest rate was about 12%, 24% per year, really. 42 00:02:36,820 --> 00:02:38,290 The way they calculate is funny. 43 00:02:38,290 --> 00:02:41,240 So it's 24% per year. 44 00:02:41,240 --> 00:02:44,980 It was a very large MFI already, but it had not 45 00:02:44,980 --> 00:02:46,330 entered the city called Hyderabad. 46 00:02:49,300 --> 00:02:53,960 So the deal we had with them was, they were going to enter 47 00:02:53,960 --> 00:03:01,670 the city of Hyderabad, but not everywhere. 48 00:03:01,670 --> 00:03:06,770 Out of 104 poor neighborhoods we selected, 49 00:03:06,770 --> 00:03:08,750 they could enter 52. 50 00:03:08,750 --> 00:03:12,310 And they'll keep the other 52 for 18 months untouched. 51 00:03:12,310 --> 00:03:15,330 That was the deal. 52 00:03:15,330 --> 00:03:16,825 So that's the Spandana group. 53 00:03:21,660 --> 00:03:27,240 So this data is from 7,200 people who we surveyed. 54 00:03:27,240 --> 00:03:29,350 7,200 households. 55 00:03:29,350 --> 00:03:34,910 These households were selected based on the fact that they 56 00:03:34,910 --> 00:03:36,360 will be eligible for a loan. 57 00:03:36,360 --> 00:03:39,500 It's not interesting to compare people-- 58 00:03:39,500 --> 00:03:41,020 you could have also have done something else. 59 00:03:41,020 --> 00:03:43,970 But it makes sense to select people. 60 00:03:43,970 --> 00:03:47,380 Now, we didn't select them just in the places which got 61 00:03:47,380 --> 00:03:48,330 microcredit. 62 00:03:48,330 --> 00:03:50,350 Because we wanted to compare the places which got 63 00:03:50,350 --> 00:03:51,590 microcredit and which didn't. 64 00:03:51,590 --> 00:03:54,300 So we had equal numbers of people. 65 00:03:54,300 --> 00:03:59,370 3,600 people with these characteristics where they got 66 00:03:59,370 --> 00:04:03,210 the loan, and 3,600 with these characteristics who 67 00:04:03,210 --> 00:04:04,390 didn't get the loan. 68 00:04:04,390 --> 00:04:08,030 Basic idea is to compare people who could have got a 69 00:04:08,030 --> 00:04:11,110 loan if there had been a microfinance organization in 70 00:04:11,110 --> 00:04:13,930 their neighborhood with people who did get a loan. 71 00:04:17,550 --> 00:04:20,920 These people are relatively poor. 72 00:04:20,920 --> 00:04:25,290 They have about a family of five monthly 73 00:04:25,290 --> 00:04:32,688 expenditure of $125. 74 00:04:32,688 --> 00:04:36,790 Most of their children are going to school, though. 75 00:04:36,790 --> 00:04:38,570 They do borrow. 76 00:04:38,570 --> 00:04:40,080 Many of them have loans. 77 00:04:40,080 --> 00:04:43,490 When they borrow, 69% of the households have loans. 78 00:04:43,490 --> 00:04:46,480 But a lot of the households that have loans from 79 00:04:46,480 --> 00:04:50,830 moneylenders pay about 4% per month. 80 00:04:50,830 --> 00:04:56,050 So Spandana charges 24% percent per year, which is 81 00:04:56,050 --> 00:04:57,610 less than 2% a month. 82 00:04:57,610 --> 00:04:59,600 Because that's simple interest. 83 00:04:59,600 --> 00:05:05,520 2% a month simple interest, which is substantially less 84 00:05:05,520 --> 00:05:07,610 than half of that. 85 00:05:07,610 --> 00:05:09,460 There's no MFI borrowing at the beginning. 86 00:05:12,310 --> 00:05:16,480 Lots of them run businesses, 31%. 87 00:05:16,480 --> 00:05:17,640 That's a number to note. 88 00:05:17,640 --> 00:05:20,030 Because you take the rich country average. 89 00:05:20,030 --> 00:05:25,590 The rich country average is something like 12%. 90 00:05:25,590 --> 00:05:29,700 9% run more than one business. 91 00:05:29,700 --> 00:05:32,510 But these businesses are extremely basic. 92 00:05:32,510 --> 00:05:35,590 Like, I have a shop. 93 00:05:35,590 --> 00:05:38,640 In the shop, I have bottles. 94 00:05:38,640 --> 00:05:44,210 In the big bottles with cookies, matchboxes, 95 00:05:44,210 --> 00:05:48,370 cigarettes, and candy in them. 96 00:05:48,370 --> 00:05:49,790 That's my shop. 97 00:05:49,790 --> 00:05:51,480 I sell that. 98 00:05:51,480 --> 00:05:57,330 Or I have a little push cart and I sell vegetables on it. 99 00:05:57,330 --> 00:05:59,860 So these businesses and not high skilled, they have very 100 00:05:59,860 --> 00:06:02,600 little capital, and they have no employees. 101 00:06:02,600 --> 00:06:07,990 Their profits are about $75 a month, so very small 102 00:06:07,990 --> 00:06:10,420 businesses. 103 00:06:10,420 --> 00:06:11,670 Like those. 104 00:06:13,730 --> 00:06:16,220 When we asked them in the beginning, why do you want to 105 00:06:16,220 --> 00:06:17,330 get a loan? 106 00:06:17,330 --> 00:06:20,150 A lot of them said what you had already anticipated. 107 00:06:20,150 --> 00:06:23,450 You might want to get a loan, not just to start a business, 108 00:06:23,450 --> 00:06:27,020 but also to repay a loan. 109 00:06:27,020 --> 00:06:28,870 To do something for your household. 110 00:06:28,870 --> 00:06:35,100 To buy a fridge or a TV. 111 00:06:35,100 --> 00:06:37,280 So they were quite candid about the fact that they 112 00:06:37,280 --> 00:06:40,480 didn't necessarily want to start a business. 113 00:06:40,480 --> 00:06:47,390 So, skip that. 114 00:06:47,390 --> 00:06:50,810 So before we get into the results, it's worth thinking a 115 00:06:50,810 --> 00:06:53,580 little bit about what we should expect to see. 116 00:06:56,440 --> 00:07:02,670 And in particular, should we expect people on average to be 117 00:07:02,670 --> 00:07:05,010 consuming more when they get a lot? 118 00:07:11,640 --> 00:07:12,890 Yes, no? 119 00:07:16,610 --> 00:07:19,040 Answers, answers. 120 00:07:19,040 --> 00:07:19,980 There, yes. 121 00:07:19,980 --> 00:07:21,070 Go ahead, explain. 122 00:07:21,070 --> 00:07:24,500 AUDIENCE: If they intend to start businesses [INAUDIBLE], 123 00:07:24,500 --> 00:07:27,930 then you might expect that they would tighten their 124 00:07:27,930 --> 00:07:31,050 belts, or consumption in the first two weeks of running it. 125 00:07:31,050 --> 00:07:33,771 Therefore, their consumption would go down. 126 00:07:33,771 --> 00:07:35,027 PROFESSOR: You might think their consumption would go 127 00:07:35,027 --> 00:07:37,970 down as a result of starting a business, because right now 128 00:07:37,970 --> 00:07:39,290 I'm starting a business. 129 00:07:39,290 --> 00:07:41,840 I don't want to start it on the smallest possible scale. 130 00:07:41,840 --> 00:07:44,160 I might want to buy some inputs for my 131 00:07:44,160 --> 00:07:45,780 business, expand it. 132 00:07:45,780 --> 00:07:48,790 So when I start a business, I might actually want to cut 133 00:07:48,790 --> 00:07:53,550 back on consumption for a while before I go back to. 134 00:07:53,550 --> 00:07:55,130 Over time, it should go back up. 135 00:07:55,130 --> 00:07:57,600 But at the beginning, I might very well see a fall in 136 00:07:57,600 --> 00:07:58,240 consumption. 137 00:07:58,240 --> 00:08:02,470 So that's an important thing to keep in mind. 138 00:08:02,470 --> 00:08:07,470 That if people are starting businesses, they may 139 00:08:07,470 --> 00:08:08,690 actually cut back. 140 00:08:08,690 --> 00:08:12,290 So you may not see a big net increase. 141 00:08:12,290 --> 00:08:13,340 You'll see an increase. 142 00:08:13,340 --> 00:08:15,930 If they start the business and business is successful, then 143 00:08:15,930 --> 00:08:17,900 two years later you'll see an increase. 144 00:08:17,900 --> 00:08:19,150 But not right away. 145 00:08:21,200 --> 00:08:24,600 So here's the first problem we had in this evaluation, which 146 00:08:24,600 --> 00:08:27,540 is that despite the fact that Spandana said they would not 147 00:08:27,540 --> 00:08:31,390 enter these neighborhoods that were put into control, the 148 00:08:31,390 --> 00:08:32,559 loan officers did. 149 00:08:32,559 --> 00:08:32,890 Why? 150 00:08:32,890 --> 00:08:36,330 Because they thought, well, that looks like a nice 151 00:08:36,330 --> 00:08:36,680 neighborhood. 152 00:08:36,680 --> 00:08:39,720 We could lend more to people and make more money. 153 00:08:39,720 --> 00:08:40,960 So they went in. 154 00:08:40,960 --> 00:08:44,720 And we had to kind of fight a lot to stop them. 155 00:08:44,720 --> 00:08:49,700 And so in the end, the difference in Spandana loans 156 00:08:49,700 --> 00:08:52,490 was only 13%. 157 00:08:52,490 --> 00:08:54,490 It should have been bigger. 158 00:08:58,160 --> 00:09:01,290 Now, because that happens-- 159 00:09:01,290 --> 00:09:04,325 so some of the control neighborhoods got the loans-- 160 00:09:07,366 --> 00:09:11,702 we're not going to be using the actual fact of who got a 161 00:09:11,702 --> 00:09:15,440 loan and who didn't get a loan. 162 00:09:15,440 --> 00:09:18,970 We're going to use the intention to treat. 163 00:09:18,970 --> 00:09:22,320 We'll compare places that we thought should not have a loan 164 00:09:22,320 --> 00:09:25,210 with places which we thought should have a loan. 165 00:09:25,210 --> 00:09:27,980 That's something that is determined by randomizing. 166 00:09:27,980 --> 00:09:30,055 Whereas who actually got a loan was not random. 167 00:09:33,260 --> 00:09:35,480 When the people went into control neighborhoods, they 168 00:09:35,480 --> 00:09:39,110 went into the best, in the most attractive ones. 169 00:09:39,110 --> 00:09:40,090 So we can't do that. 170 00:09:40,090 --> 00:09:45,060 We need to compare ones which actually got the loan. 171 00:09:45,060 --> 00:09:49,670 And even worse than that, where Spandana didn't go, 172 00:09:49,670 --> 00:09:51,520 other MFIs went. 173 00:09:51,520 --> 00:09:53,570 Because they wanted to fill up the market. 174 00:09:53,570 --> 00:09:56,085 So this difference is small. 175 00:09:56,085 --> 00:09:57,880 So when you see-- 176 00:09:57,880 --> 00:09:58,910 why is that relevant? 177 00:09:58,910 --> 00:10:01,180 Well, when you see the difference, I'm going to show 178 00:10:01,180 --> 00:10:06,320 you the average difference between places that were 179 00:10:06,320 --> 00:10:10,840 intended to get microfinance with places that were not 180 00:10:10,840 --> 00:10:13,070 intended to get microfinance. 181 00:10:13,070 --> 00:10:16,020 Notice that that difference is only 8%. 182 00:10:16,020 --> 00:10:19,290 The places that were intended to get microfinance were only 183 00:10:19,290 --> 00:10:20,060 8% percent. 184 00:10:20,060 --> 00:10:23,700 People there are only 8% more likely to have a loan than 185 00:10:23,700 --> 00:10:29,350 places that were intended not to get microfinance. 186 00:10:29,350 --> 00:10:35,980 So when I look at the average of those places, what will 187 00:10:35,980 --> 00:10:37,475 happen to the magnitude of the effect? 188 00:10:40,370 --> 00:10:41,570 It will be very small. 189 00:10:41,570 --> 00:10:44,640 Because only 8% difference-- 190 00:10:44,640 --> 00:10:48,210 in either, both places, a lot of people got loans. 191 00:10:48,210 --> 00:10:50,030 So the difference was small. 192 00:10:50,030 --> 00:10:53,420 So to get the real sense of the magnitude of the effect, I 193 00:10:53,420 --> 00:10:55,100 need to multiply by-- 194 00:10:55,100 --> 00:10:59,150 since it's 8% of the people generated this difference, to 195 00:10:59,150 --> 00:11:03,830 get the full effect of this, what should I do? 196 00:11:03,830 --> 00:11:06,860 To get a sense of the magnitude, what 197 00:11:06,860 --> 00:11:09,240 would I need to do? 198 00:11:09,240 --> 00:11:16,761 If 8% generated an effect of 5, what would the effect have 199 00:11:16,761 --> 00:11:22,725 been if the difference was 100%? 200 00:11:22,725 --> 00:11:26,220 AUDIENCE: You multiply by 100, divide by 8. 201 00:11:26,220 --> 00:11:28,630 PROFESSOR: You'd multiply by 100, divide by 12. 202 00:11:28,630 --> 00:11:31,227 You would multiply by 12. 203 00:11:31,227 --> 00:11:34,054 AUDIENCE: Isn't it the case that you could look up what is 204 00:11:34,054 --> 00:11:35,219 the average-- 205 00:11:35,219 --> 00:11:38,213 if there is no microfinance, then it's 100% of the people 206 00:11:38,213 --> 00:11:40,708 that [INAUDIBLE]. 207 00:11:43,720 --> 00:11:45,290 PROFESSOR: So you'll do something that. 208 00:11:45,290 --> 00:11:47,502 So when you look at the numbers, they look small. 209 00:11:47,502 --> 00:11:49,440 But you have to realize that that's a 210 00:11:49,440 --> 00:11:52,390 difference generated by-- 211 00:11:52,390 --> 00:11:55,530 the difference is generated by a very small 212 00:11:55,530 --> 00:11:57,120 difference in take up. 213 00:11:57,120 --> 00:12:00,280 So these results are look much smaller than they really are. 214 00:12:05,220 --> 00:12:08,360 So here's what happened in the business space. 215 00:12:12,060 --> 00:12:15,210 There was some increase in new businesses, which was 216 00:12:15,210 --> 00:12:16,570 significant. 217 00:12:16,570 --> 00:12:19,140 Some increase in profits, which was not significant. 218 00:12:19,140 --> 00:12:22,870 Some increase in revenues, which was not significant. 219 00:12:22,870 --> 00:12:28,040 But the new businesses was significant. 220 00:12:28,040 --> 00:12:29,780 So think of it this way. 221 00:12:29,780 --> 00:12:31,830 This is the number of businesses set up 222 00:12:31,830 --> 00:12:33,450 in 1 and 1/2 years. 223 00:12:33,450 --> 00:12:36,850 So basically, 31% already had a business. 224 00:12:36,850 --> 00:12:40,020 Another 7% set up businesses in the treatment. 225 00:12:40,020 --> 00:12:42,910 And 5.3% set up business in the control. 226 00:12:42,910 --> 00:12:48,770 So that's a 30% increase in the number of new businesses. 227 00:12:48,770 --> 00:12:50,290 So it's not small. 228 00:12:50,290 --> 00:12:53,070 It looks small numbers, but it's in fact 229 00:12:53,070 --> 00:12:56,940 a significant increase. 230 00:12:56,940 --> 00:13:00,430 In any case, people don't start that many new business. 231 00:13:00,430 --> 00:13:02,920 So is there was some effect. 232 00:13:05,560 --> 00:13:07,010 Consumption. 233 00:13:07,010 --> 00:13:11,610 Blue is always control, yellow is treatment. 234 00:13:11,610 --> 00:13:14,510 So consumption didn't really go up significantly. 235 00:13:14,510 --> 00:13:17,320 But we discussed why that might happen. 236 00:13:17,320 --> 00:13:20,020 You do see more durable purchases, more business 237 00:13:20,020 --> 00:13:20,950 durable purchases. 238 00:13:20,950 --> 00:13:22,470 And I won't say what temptation 239 00:13:22,470 --> 00:13:24,070 goods are right now. 240 00:13:24,070 --> 00:13:26,300 So less temptation goods. 241 00:13:26,300 --> 00:13:29,930 But you do see more business. 242 00:13:29,930 --> 00:13:32,160 Here's what didn't happen. 243 00:13:32,160 --> 00:13:35,500 Despite the gender nature of spending, 244 00:13:35,500 --> 00:13:37,490 apparently nothing changed. 245 00:13:37,490 --> 00:13:41,230 So schooling is hard to change, because 246 00:13:41,230 --> 00:13:42,550 everybody goes to school. 247 00:13:42,550 --> 00:13:44,480 So let's say that's difficult to change. 248 00:13:44,480 --> 00:13:47,054 But nothing else changed either. 249 00:13:47,054 --> 00:13:50,150 Children were just as unhealthy. 250 00:13:50,150 --> 00:13:54,220 Women were no more likely to have decision power on 251 00:13:54,220 --> 00:13:56,110 spending in the household. 252 00:13:56,110 --> 00:13:58,690 And health expenditures didn't go up. 253 00:13:58,690 --> 00:14:01,850 So if you think that women spend differently, we didn't 254 00:14:01,850 --> 00:14:03,460 find any evidence of it. 255 00:14:03,460 --> 00:14:09,120 So the gender effect that we had discussed doesn't seem to 256 00:14:09,120 --> 00:14:11,620 be in this data. 257 00:14:11,620 --> 00:14:13,360 The next thing we did is go back to that 258 00:14:13,360 --> 00:14:14,350 story we were telling. 259 00:14:14,350 --> 00:14:17,110 Me were telling a story where some people would invest, 260 00:14:17,110 --> 00:14:19,370 start a new business, and they would cut their consumption. 261 00:14:22,640 --> 00:14:23,600 Others will not start their business. 262 00:14:23,600 --> 00:14:26,820 And now they have no reason to cut their consumption. 263 00:14:26,820 --> 00:14:30,320 So imagine that the two kinds of people, one kind of people 264 00:14:30,320 --> 00:14:31,910 are going to start a new business and 265 00:14:31,910 --> 00:14:34,690 the other kind wont. 266 00:14:34,690 --> 00:14:37,110 So what we want to do is we want to predict who those 267 00:14:37,110 --> 00:14:38,360 people are. 268 00:14:44,145 --> 00:14:48,600 If we went to the treatment areas and we looked at who 269 00:14:48,600 --> 00:14:49,980 started a new business. 270 00:14:49,980 --> 00:14:52,720 And we went to the control area and we looked at who 271 00:14:52,720 --> 00:14:55,250 started a new business, and we compared them, would that be a 272 00:14:55,250 --> 00:14:56,500 valid comparison? 273 00:15:09,154 --> 00:15:11,120 Let me repeat the question. 274 00:15:11,120 --> 00:15:15,020 Suppose I went to the treatment areas-- 275 00:15:15,020 --> 00:15:16,352 the treatment area and the control area 276 00:15:16,352 --> 00:15:18,050 were randomly chosen. 277 00:15:18,050 --> 00:15:21,800 I went to the treatment areas, I found people there who had 278 00:15:21,800 --> 00:15:23,380 started a business. 279 00:15:23,380 --> 00:15:26,140 I went to the control areas, and I found people who had 280 00:15:26,140 --> 00:15:28,930 already newly started a business. 281 00:15:28,930 --> 00:15:30,110 And I compared them. 282 00:15:30,110 --> 00:15:31,485 Would that be a valid comparison? 283 00:15:35,366 --> 00:15:37,358 AUDIENCE: May I ask a question? 284 00:15:37,358 --> 00:15:39,848 In the first example, when you went to the treatment areas 285 00:15:39,848 --> 00:15:42,587 did you study the people newly started a business, or people 286 00:15:42,587 --> 00:15:43,350 that had a business? 287 00:15:43,350 --> 00:15:45,710 PROFESSOR: So when I say newly started a business, these are 288 00:15:45,710 --> 00:15:49,585 people who started a business in the 18 months when we were 289 00:15:49,585 --> 00:15:50,850 observing these areas. 290 00:15:50,850 --> 00:15:53,300 AUDIENCE: No, it wouldn't be valid [INAUDIBLE]. 291 00:15:53,300 --> 00:15:53,790 PROFESSOR: Why? 292 00:15:53,790 --> 00:15:55,260 AUDIENCE: And the reason why is because, while you are 293 00:15:55,260 --> 00:16:00,650 comparing new businesses started, in the treatment 294 00:16:00,650 --> 00:16:04,325 area, without more precise information, it's hard to 295 00:16:04,325 --> 00:16:07,510 understand who started the business due to the fact that 296 00:16:07,510 --> 00:16:11,070 they got the loan versus just due to the fact that they were 297 00:16:11,070 --> 00:16:12,040 starting it anyway. 298 00:16:12,040 --> 00:16:14,955 And yeah, you could compare a control area. 299 00:16:14,955 --> 00:16:17,137 But even your control versus your treatment area, it's not 300 00:16:17,137 --> 00:16:18,672 necessarily indicative that they had the same sort of 301 00:16:18,672 --> 00:16:20,290 basis, [INAUDIBLE] 302 00:16:20,290 --> 00:16:23,530 difference or differences of the treatment area before and 303 00:16:23,530 --> 00:16:25,390 after they got the loans. 304 00:16:25,390 --> 00:16:28,190 PROFESSOR: You were nodding your head. 305 00:16:28,190 --> 00:16:30,452 Do you have a different answer than that? 306 00:16:30,452 --> 00:16:32,210 AUDIENCE: [INAUDIBLE]. 307 00:16:32,210 --> 00:16:37,580 PROFESSOR: So basically, I think that's not quite right. 308 00:16:37,580 --> 00:16:38,850 I think you don't ever want to do it. 309 00:16:38,850 --> 00:16:41,070 The reason you never want to do it is 310 00:16:41,070 --> 00:16:43,640 because you don't actually-- 311 00:16:43,640 --> 00:16:47,500 think of people who start a business when I can't offer 312 00:16:47,500 --> 00:16:49,210 them a loan. 313 00:16:49,210 --> 00:16:53,080 Versus people who start a business when no loans were 314 00:16:53,080 --> 00:16:56,100 available or less loans were available. 315 00:16:56,100 --> 00:16:58,410 They're different kinds of people. 316 00:16:58,410 --> 00:17:01,800 When a loan is easy to get, I might start a business. 317 00:17:01,800 --> 00:17:04,040 But if loans are difficult to get, I'm lazy, I 318 00:17:04,040 --> 00:17:05,310 won't start a business. 319 00:17:05,310 --> 00:17:07,819 Whereas somebody who starts a loan when loans are not easily 320 00:17:07,819 --> 00:17:10,270 available is probably much more committed 321 00:17:10,270 --> 00:17:11,440 to starting a business. 322 00:17:11,440 --> 00:17:13,339 Is much one incentivized. 323 00:17:13,339 --> 00:17:15,020 So you don't want to compare them. 324 00:17:15,020 --> 00:17:19,180 So what we had to do, we had to think of what could predict 325 00:17:19,180 --> 00:17:20,700 who would start a business? 326 00:17:20,700 --> 00:17:22,599 Then what we could do is we could split-- 327 00:17:22,599 --> 00:17:27,329 we could look for the kind of people would start a business 328 00:17:27,329 --> 00:17:28,359 in treatment. 329 00:17:28,359 --> 00:17:30,950 And look for the same kind of people in control. 330 00:17:30,950 --> 00:17:34,215 So we could have a predictive model of who 331 00:17:34,215 --> 00:17:35,190 would start a business. 332 00:17:35,190 --> 00:17:38,100 We could compare the things that predict whether you start 333 00:17:38,100 --> 00:17:40,000 a business or not. 334 00:17:40,000 --> 00:17:42,950 So we want to divide the population based on fixed 335 00:17:42,950 --> 00:17:44,490 characteristics. 336 00:17:44,490 --> 00:17:47,030 And the fixed characteristic that worked very well is, if 337 00:17:47,030 --> 00:17:51,510 you own land and if you have a, typically a wife, a woman 338 00:17:51,510 --> 00:17:55,520 in the household who is literate and doesn't work. 339 00:17:55,520 --> 00:18:01,330 So if you have some assets, there's a woman in the 340 00:18:01,330 --> 00:18:03,670 household who is currently not working. 341 00:18:03,670 --> 00:18:06,400 But she has skills, skills she's literate. 342 00:18:06,400 --> 00:18:07,960 She's actually a prime candidate 343 00:18:07,960 --> 00:18:09,110 for starting a business. 344 00:18:09,110 --> 00:18:12,780 So it turns out that if you look for families with these 345 00:18:12,780 --> 00:18:15,160 characteristics in treatment and control, there were equal 346 00:18:15,160 --> 00:18:17,560 numbers of families in treatment and control. 347 00:18:17,560 --> 00:18:18,590 Not accidentally. 348 00:18:18,590 --> 00:18:19,550 That's how you'd expect. 349 00:18:19,550 --> 00:18:21,140 You'd expect randomly chosen. 350 00:18:21,140 --> 00:18:22,710 So they are about the same. 351 00:18:22,710 --> 00:18:27,270 So you take these kinds of families, which are available 352 00:18:27,270 --> 00:18:28,520 in treatment and control. 353 00:18:28,520 --> 00:18:30,173 They're identical-- yeah. 354 00:18:30,173 --> 00:18:31,159 AUDIENCE: [INAUDIBLE] 355 00:18:31,159 --> 00:18:32,409 instrumental variable? 356 00:18:34,800 --> 00:18:36,340 PROFESSOR: This is not an instrumental variable. 357 00:18:36,340 --> 00:18:40,330 It's just predicting who are going to start a business. 358 00:18:40,330 --> 00:18:42,350 Then what I'm going to do is I'm going to create-- 359 00:18:45,030 --> 00:18:47,000 this is like baseline characteristics. 360 00:18:47,000 --> 00:18:48,450 They're just different people. 361 00:18:48,450 --> 00:18:50,330 I can estimate the treatment effect for 362 00:18:50,330 --> 00:18:52,130 any subset of people. 363 00:18:52,130 --> 00:18:55,560 I could say what's the effect on tall people? 364 00:18:55,560 --> 00:18:59,140 This is just take a subset of the population who are with 365 00:18:59,140 --> 00:19:03,450 these characteristics that predict a business. 366 00:19:03,450 --> 00:19:06,580 Those people's consumption, does it do behave differently 367 00:19:06,580 --> 00:19:08,260 from the consumption of everyone else? 368 00:19:08,260 --> 00:19:10,010 That's the question I'm asking. 369 00:19:10,010 --> 00:19:12,910 If these are people whose characteristics predict that 370 00:19:12,910 --> 00:19:16,070 they will start a business, do they behave differently from 371 00:19:16,070 --> 00:19:18,690 everybody else whose characteristics predict that 372 00:19:18,690 --> 00:19:20,050 they won't start a business? 373 00:19:20,050 --> 00:19:21,860 So there are three groups I'm going to look at. 374 00:19:21,860 --> 00:19:24,770 People who's characteristics predict they will start a 375 00:19:24,770 --> 00:19:28,390 business, people who already have a business, and people 376 00:19:28,390 --> 00:19:31,680 neither have a business nor will start a business. 377 00:19:31,680 --> 00:19:32,850 So that's what I'm trying to. 378 00:19:32,850 --> 00:19:37,370 And if you do it, you do see the difference. 379 00:19:37,370 --> 00:19:42,000 In the treatment, the people who are who we predict should 380 00:19:42,000 --> 00:19:45,130 start a business are much more likely to start a business. 381 00:19:45,130 --> 00:19:47,380 The people who predict shouldn't start a businesses 382 00:19:47,380 --> 00:19:49,900 are actually slightly less likely to start a business. 383 00:19:49,900 --> 00:19:51,794 So that's sort of what you'd expect. 384 00:19:55,520 --> 00:19:59,410 And now you start to see much bigger differences. 385 00:19:59,410 --> 00:20:02,920 So the people who start a business and people who 386 00:20:02,920 --> 00:20:07,450 already had a business, expand their businesses. 387 00:20:07,450 --> 00:20:11,060 Or they buy durables. 388 00:20:11,060 --> 00:20:13,960 The people who didn't have a business and don't are not 389 00:20:13,960 --> 00:20:16,180 planning a business-- so durables can include, for 390 00:20:16,180 --> 00:20:19,520 example, business assets. 391 00:20:19,520 --> 00:20:20,900 They can repair the house where they're 392 00:20:20,900 --> 00:20:22,140 going to have a shop. 393 00:20:22,140 --> 00:20:23,890 So all those things are durable. 394 00:20:23,890 --> 00:20:27,280 So they seem to do stuff that could be 395 00:20:27,280 --> 00:20:29,280 related to the business. 396 00:20:29,280 --> 00:20:31,690 The people who we predict should not start a business, 397 00:20:31,690 --> 00:20:32,890 don't start a business. 398 00:20:32,890 --> 00:20:35,170 And they don't make any durable expenditures. 399 00:20:35,170 --> 00:20:41,600 They increase non durable expenditures a lot. 400 00:20:41,600 --> 00:20:46,110 So the bottom group increases non durable expenditures. 401 00:20:46,110 --> 00:20:51,430 That's like spending on food, and things like that. 402 00:20:51,430 --> 00:20:54,270 The middle group, basically nothing changes. 403 00:20:54,270 --> 00:20:57,400 And the top group cuts non durables. 404 00:20:57,400 --> 00:20:59,130 They're eating less. 405 00:20:59,130 --> 00:21:01,210 These people are spending less on food, 406 00:21:01,210 --> 00:21:03,040 they're buying more durables. 407 00:21:03,040 --> 00:21:06,250 The durables are what they want for the future. 408 00:21:06,250 --> 00:21:08,710 So they seem to be behaving in very different ways. 409 00:21:14,570 --> 00:21:17,510 So finally, this thing of temptation goods. 410 00:21:17,510 --> 00:21:18,880 What are temptation goods? 411 00:21:18,880 --> 00:21:20,000 Well we asked people. 412 00:21:20,000 --> 00:21:24,560 This is a beetle leaf cellar. 413 00:21:24,560 --> 00:21:28,790 So people in South Asia and Southeast Asia chew these 414 00:21:28,790 --> 00:21:31,460 leaves and these nuts, which are called beetle nuts and 415 00:21:31,460 --> 00:21:32,950 beetle leaves. 416 00:21:32,950 --> 00:21:35,960 They have slightly narcotic effects, mild narcotic 417 00:21:35,960 --> 00:21:40,000 effects, and a lot of people chew them. 418 00:21:40,000 --> 00:21:42,910 And this guy's a vendor. 419 00:21:42,910 --> 00:21:44,730 He's a small business owner. 420 00:21:44,730 --> 00:21:51,400 So we asked people in the baseline to tell us, what are 421 00:21:51,400 --> 00:21:54,110 things you want to give up on spending? 422 00:21:57,010 --> 00:22:00,500 So we asked them to give us a list of things they said they 423 00:22:00,500 --> 00:22:02,570 want to give up. 424 00:22:02,570 --> 00:22:04,850 And that was basically very simple. 425 00:22:04,850 --> 00:22:12,180 They want to give up on tea, coffee, snacks, beetle nuts, 426 00:22:12,180 --> 00:22:15,130 cigarettes, and alcohol. 427 00:22:15,130 --> 00:22:16,460 That's all they wanted to. 428 00:22:16,460 --> 00:22:19,310 These were the things they wanted to give up. 429 00:22:19,310 --> 00:22:21,590 Now we go back two years later, and 430 00:22:21,590 --> 00:22:22,760 we look at the data. 431 00:22:22,760 --> 00:22:26,000 And we classify spending on these things 432 00:22:26,000 --> 00:22:26,910 versus everything else. 433 00:22:26,910 --> 00:22:28,790 These things we called "temptation," because they're 434 00:22:28,790 --> 00:22:30,910 things that they say they don't want to do, but they end 435 00:22:30,910 --> 00:22:33,140 up spending on. 436 00:22:33,140 --> 00:22:36,150 And what you see quite dramatically is that that's 437 00:22:36,150 --> 00:22:38,700 where you see a big difference. 438 00:22:38,700 --> 00:22:43,790 The people who are starting a business cut back massively in 439 00:22:43,790 --> 00:22:47,730 their spending on temptation goods. 440 00:22:47,730 --> 00:22:54,410 The people who will never start a business seem to be 441 00:22:54,410 --> 00:22:55,480 having a party. 442 00:22:55,480 --> 00:22:59,160 They go out and they start spending more on those things. 443 00:22:59,160 --> 00:23:02,740 So basically what's going on for many of these people is 444 00:23:02,740 --> 00:23:06,000 they're turning their temptation expenditures, 445 00:23:06,000 --> 00:23:09,890 they're using that money to pay the loan payment. 446 00:23:09,890 --> 00:23:13,150 And then with the money, they're buying something else. 447 00:23:13,150 --> 00:23:16,540 So when they get the money, they either buy a durable or a 448 00:23:16,540 --> 00:23:17,490 business durable. 449 00:23:17,490 --> 00:23:22,620 They expand their consumption, their durable ownership, and 450 00:23:22,620 --> 00:23:24,610 then they pay down by cutting back on 451 00:23:24,610 --> 00:23:26,570 their temptation goods. 452 00:23:26,570 --> 00:23:29,540 So they're drinking less tea and they're owning more TVs, 453 00:23:29,540 --> 00:23:30,300 if you like. 454 00:23:30,300 --> 00:23:31,055 Yeah. 455 00:23:31,055 --> 00:23:32,305 AUDIENCE: [INAUDIBLE] 456 00:23:37,360 --> 00:23:40,280 less on temptation goods than those who were taking loans? 457 00:23:40,280 --> 00:23:41,860 PROFESSOR: Because these people 458 00:23:41,860 --> 00:23:43,350 are starting a business. 459 00:23:43,350 --> 00:23:44,430 They need the money. 460 00:23:44,430 --> 00:23:47,748 AUDIENCE: No, but in the control group also they are 461 00:23:47,748 --> 00:23:50,120 starting new businesses? 462 00:23:50,120 --> 00:23:50,840 PROFESSOR: No, they're much less likely. 463 00:23:50,840 --> 00:23:53,340 Remember, these people have high 464 00:23:53,340 --> 00:23:55,080 propensity to start a business. 465 00:23:55,080 --> 00:23:58,230 Nevertheless, in control, they're much less likely to. 466 00:23:58,230 --> 00:24:01,130 They're only half as likely to start a business in control as 467 00:24:01,130 --> 00:24:03,790 in treatment. 468 00:24:03,790 --> 00:24:07,940 So these people are much more likely to be people who 469 00:24:07,940 --> 00:24:09,215 started businesses in treatment. 470 00:24:14,980 --> 00:24:17,820 OK, so that's sort of the-- 471 00:24:17,820 --> 00:24:21,000 now, this is a story which is-- 472 00:24:21,000 --> 00:24:22,700 it depends on the-- 473 00:24:22,700 --> 00:24:25,770 So when we published these results, the reaction from 474 00:24:25,770 --> 00:24:30,275 microfinance industry was very negative, largely. 475 00:24:30,275 --> 00:24:32,220 They thought these results were showing 476 00:24:32,220 --> 00:24:33,560 that they had failed. 477 00:24:33,560 --> 00:24:36,720 And it really went ballistic. 478 00:24:36,720 --> 00:24:40,860 There was lots and lots of blogs which explained how we 479 00:24:40,860 --> 00:24:43,560 don't understand how to analyze microcredit. 480 00:24:48,270 --> 00:24:53,640 On the World Bank's website there where dismissive 481 00:24:53,640 --> 00:24:55,280 statements, and all that. 482 00:24:55,280 --> 00:24:58,670 So there were lots of these things basically claiming that 483 00:24:58,670 --> 00:25:02,530 microcredit is too subtle to be analyzed by us. 484 00:25:02,530 --> 00:25:04,960 So why was the reaction? 485 00:25:04,960 --> 00:25:07,960 We actually didn't think this was so negative, in fact. 486 00:25:07,960 --> 00:25:09,750 We thought this was fine. 487 00:25:09,750 --> 00:25:10,930 New businesses went up. 488 00:25:10,930 --> 00:25:14,230 The people we thought would do better were doing better. 489 00:25:14,230 --> 00:25:15,540 Some people were doing worse. 490 00:25:15,540 --> 00:25:17,620 That's normal. 491 00:25:17,620 --> 00:25:19,600 If you give people a loan, not everybody is 492 00:25:19,600 --> 00:25:20,830 going to use it well. 493 00:25:20,830 --> 00:25:23,190 Some people are going to use it in ways that will get them 494 00:25:23,190 --> 00:25:23,970 into trouble. 495 00:25:23,970 --> 00:25:26,890 Others are going to use it very well and 496 00:25:26,890 --> 00:25:27,720 do well out of it. 497 00:25:27,720 --> 00:25:30,210 And you're always creating some risk when you create 498 00:25:30,210 --> 00:25:32,210 financial products. 499 00:25:32,210 --> 00:25:36,470 But people really hated these results, and so we 500 00:25:36,470 --> 00:25:37,830 got some push back. 501 00:25:37,830 --> 00:25:42,840 [INAUDIBLE] went away for a sad way. 502 00:25:42,840 --> 00:25:45,780 What happened next was that-- so this was 503 00:25:45,780 --> 00:25:49,040 going on through 2009. 504 00:25:49,040 --> 00:25:52,900 When the results came out, there was lots of push back. 505 00:25:52,900 --> 00:25:55,490 And then side-by-side with that, something 506 00:25:55,490 --> 00:25:56,470 else started happening. 507 00:25:56,470 --> 00:25:57,720 And that's sort of unfortunate. 508 00:26:00,250 --> 00:26:08,110 So the next thing that happened was that in 2009-- 509 00:26:08,110 --> 00:26:13,200 I guess both in 2009, once in an organization called 510 00:26:13,200 --> 00:26:20,322 Compartamos in Mexico had an IPO. 511 00:26:20,322 --> 00:26:24,680 You know what an IPO is? 512 00:26:24,680 --> 00:26:26,540 An initial public offering. 513 00:26:26,540 --> 00:26:28,530 They sold their stock. 514 00:26:28,530 --> 00:26:32,060 And they made a ton of money. 515 00:26:32,060 --> 00:26:35,420 And an organization called SKS in India, which we actually 516 00:26:35,420 --> 00:26:40,140 worked with on something else, not microcredit, then went 517 00:26:40,140 --> 00:26:44,920 ahead and they had another IPO. 518 00:26:44,920 --> 00:26:48,780 These IPOs raised a lot of money, like hundreds of 519 00:26:48,780 --> 00:26:50,660 millions of dollars. 520 00:26:50,660 --> 00:26:56,190 And the reaction to that was correctly, this means these 521 00:26:56,190 --> 00:26:57,645 organizations are very profitable. 522 00:27:01,070 --> 00:27:03,720 Suddenly, everybody got upset with them. 523 00:27:03,720 --> 00:27:06,715 Because they realized that if these companies can be sold 524 00:27:06,715 --> 00:27:11,840 for $300 million, then they must be making a lot of money. 525 00:27:11,840 --> 00:27:13,920 Now that's a bit unfair. 526 00:27:13,920 --> 00:27:16,750 Because they were making a lot of money, but they also had 527 00:27:16,750 --> 00:27:18,400 lots of clients. 528 00:27:18,400 --> 00:27:21,500 And if you have three million clients, then making $300 529 00:27:21,500 --> 00:27:26,610 million over the next 20 years, or something, is not so 530 00:27:26,610 --> 00:27:28,500 much after all. 531 00:27:28,500 --> 00:27:30,490 It's not that much money. 532 00:27:30,490 --> 00:27:35,310 If you think, I have 3 million clients, I'm making $300 533 00:27:35,310 --> 00:27:41,840 million, that's $100 per client over the next, let's 534 00:27:41,840 --> 00:27:42,860 say, ten years. 535 00:27:42,860 --> 00:27:48,740 So that's $10 per year per client, which is not nearly as 536 00:27:48,740 --> 00:27:52,320 profitable as it sounds when you say it's $300 million. 537 00:27:52,320 --> 00:27:53,860 Because these are big organizations, 538 00:27:53,860 --> 00:27:55,240 many of them are. 539 00:27:55,240 --> 00:28:00,590 So in any case, that got them into lots of trouble. 540 00:28:00,590 --> 00:28:03,000 So first thing that happened was that a lot of microfinance 541 00:28:03,000 --> 00:28:05,950 people started getting very upset. 542 00:28:05,950 --> 00:28:08,340 Mohammad Yunus, who started Grameen 543 00:28:08,340 --> 00:28:10,800 Bank, was really vehement. 544 00:28:10,800 --> 00:28:13,280 He called them the new userers. 545 00:28:13,280 --> 00:28:19,750 And basically said that this commercialization of 546 00:28:19,750 --> 00:28:23,310 microfinance goes against the spirit of microfinance. 547 00:28:23,310 --> 00:28:24,650 So that was already trouble. 548 00:28:24,650 --> 00:28:28,410 Because there was divide in the community. 549 00:28:28,410 --> 00:28:32,730 What happened after that was even-- 550 00:28:32,730 --> 00:28:33,980 so they said. 551 00:28:37,250 --> 00:28:38,050 We have a minute. 552 00:28:38,050 --> 00:28:44,700 So they said what we're trying to do, we're doing an IPO 553 00:28:44,700 --> 00:28:46,710 because this is the best way to raise capital. 554 00:28:46,710 --> 00:28:51,760 When we raise capital, we have a lot of equity, we can now go 555 00:28:51,760 --> 00:28:54,610 and borrow a lot based on our equity, and then 556 00:28:54,610 --> 00:28:56,020 we can expand lending. 557 00:28:56,020 --> 00:29:00,050 So instead of 10 million people, 100 million 558 00:29:00,050 --> 00:29:01,990 people can get loans. 559 00:29:01,990 --> 00:29:05,120 So they were making the case, like in the movie, that we 560 00:29:05,120 --> 00:29:07,150 really need to go in this direction. 561 00:29:07,150 --> 00:29:10,150 So this was a fight going on. 562 00:29:10,150 --> 00:29:12,900 Then what happened, this goes into the 563 00:29:12,900 --> 00:29:16,120 beginning of 2010 now. 564 00:29:16,120 --> 00:29:20,250 So there is two fights going on through 2009. 565 00:29:20,250 --> 00:29:22,360 One is basically against us. 566 00:29:22,360 --> 00:29:27,380 And another set of results by Dean [INAUDIBLE], who's at 567 00:29:27,380 --> 00:29:31,460 Yale, also showing small positive results. 568 00:29:31,460 --> 00:29:34,250 But microfinance companies didn't like the small positive 569 00:29:34,250 --> 00:29:39,430 results and they didn't like these IPOs. 570 00:29:39,430 --> 00:29:44,090 Then what happened is that a number of 571 00:29:44,090 --> 00:29:46,030 people committed suicide. 572 00:29:49,580 --> 00:29:57,340 A number of people in India, in one state of India, which 573 00:29:57,340 --> 00:30:01,860 is called Andhra Pradesh, committed suicide. 574 00:30:01,860 --> 00:30:04,830 Some of them claiming that it is because their loan burden 575 00:30:04,830 --> 00:30:05,710 was too high. 576 00:30:05,710 --> 00:30:08,740 It turned out that they had borrowed from six or seven 577 00:30:08,740 --> 00:30:10,620 different microfinance organizations. 578 00:30:10,620 --> 00:30:13,630 And they couldn't repay, and they committed suicide. 579 00:30:13,630 --> 00:30:19,800 And at that point, all these forces kind of lined up in 580 00:30:19,800 --> 00:30:23,130 exactly the worst possible way for the industry. 581 00:30:23,130 --> 00:30:25,240 First, there were these suicides, second there was a 582 00:30:25,240 --> 00:30:27,820 fight within the industry, and third there were these results 583 00:30:27,820 --> 00:30:31,470 showing that they hadn't done something so great. 584 00:30:31,470 --> 00:30:33,890 And the net result was that there was an attempt to 585 00:30:33,890 --> 00:30:38,630 basically, there was a political attempt, to shut 586 00:30:38,630 --> 00:30:39,910 down microcredit. 587 00:30:39,910 --> 00:30:42,100 And that's still continuing, in the sense. 588 00:30:42,100 --> 00:30:45,390 And what I want to end on is showing you a 589 00:30:45,390 --> 00:30:50,170 video, which is-- 590 00:30:50,170 --> 00:30:54,200 so BBC, as you saw a video that was very pro microcredit. 591 00:30:54,200 --> 00:30:59,570 I want to show you a video that's more recent and much 592 00:30:59,570 --> 00:31:00,610 less pro microcredit. 593 00:31:00,610 --> 00:31:01,860 Where is it? 594 00:31:10,720 --> 00:31:13,570 So you can see that you can have unfair views on 595 00:31:13,570 --> 00:31:15,600 any side you want. 596 00:31:15,600 --> 00:31:22,370 You can make, that's a hatchet job, if you've ever seen one. 597 00:31:22,370 --> 00:31:25,070 So main point being that I think that-- 598 00:31:25,070 --> 00:31:27,420 I only want to end on this one point. 599 00:31:27,420 --> 00:31:31,070 Which is that if you look at this study, it does exactly 600 00:31:31,070 --> 00:31:35,750 almost the same job of very similar. 601 00:31:35,750 --> 00:31:36,370 The [INAUDIBLE] 602 00:31:36,370 --> 00:31:37,470 strategy is very similar. 603 00:31:37,470 --> 00:31:40,930 I go to some person in some village, 604 00:31:40,930 --> 00:31:42,520 something bad happened. 605 00:31:42,520 --> 00:31:45,320 That means that this thing is bad. 606 00:31:45,320 --> 00:31:46,760 You just look for the people. 607 00:31:46,760 --> 00:31:49,380 And in some ways, I think this has been 608 00:31:49,380 --> 00:31:50,600 the industry's problem. 609 00:31:50,600 --> 00:31:53,910 And the reason why that guy got nailed so badly is 610 00:31:53,910 --> 00:31:57,760 because, instead of saying that this is something that's 611 00:31:57,760 --> 00:32:02,460 useful, and we should find out how useful it is, this started 612 00:32:02,460 --> 00:32:04,050 by over selling. 613 00:32:04,050 --> 00:32:06,120 And then they're easy target. 614 00:32:06,120 --> 00:32:07,820 Everybody now is-- 615 00:32:07,820 --> 00:32:11,450 it's like, if you oversell, people have high expectations. 616 00:32:11,450 --> 00:32:14,110 Then you don't deliver, they get very upset with you. 617 00:32:14,110 --> 00:32:15,950 And so that's what they're paying for now. 618 00:32:15,950 --> 00:32:19,670 And this is like exactly what you'd expect. 619 00:32:19,670 --> 00:32:22,980 It's the reason why we were trying to persuade them for at 620 00:32:22,980 --> 00:32:26,030 least seven, eight years that they should have a randomized 621 00:32:26,030 --> 00:32:28,115 evaluation. 622 00:32:28,115 --> 00:32:33,170 I remember going to an MFI and explaining that they should 623 00:32:33,170 --> 00:32:34,880 have a randomized evacuation. 624 00:32:34,880 --> 00:32:38,090 They should show that they have evidence of the impact. 625 00:32:38,090 --> 00:32:41,250 And the guy told me, he was an interesting character. 626 00:32:41,250 --> 00:32:46,000 He was a Canadian who had started an MFI in rural India. 627 00:32:46,000 --> 00:32:53,850 And he was clearly not doing this because he was making 628 00:32:53,850 --> 00:32:54,510 more money. 629 00:32:54,510 --> 00:32:56,880 He was dedicated to the cause. 630 00:32:56,880 --> 00:32:59,760 He was living in rural India, lending people 631 00:32:59,760 --> 00:33:02,450 who were buying cows. 632 00:33:02,450 --> 00:33:05,680 But he basically got so upset with the idea. 633 00:33:05,680 --> 00:33:12,000 He said, when you buy an apple, you don't look at the 634 00:33:12,000 --> 00:33:12,940 evaluation of that. 635 00:33:12,940 --> 00:33:15,610 People are buying credit, we are selling it. 636 00:33:15,610 --> 00:33:16,820 What's wrong with that? 637 00:33:16,820 --> 00:33:21,220 And so he just threw us out of his office, basically. 638 00:33:21,220 --> 00:33:22,440 But I think that's their problem. 639 00:33:22,440 --> 00:33:24,590 Their problem came from the beginning. 640 00:33:24,590 --> 00:33:29,180 They just pushed back on any possibility that you want to 641 00:33:29,180 --> 00:33:34,250 make a reasoned view of these things. 642 00:33:34,250 --> 00:33:36,880 And then, in some ways, that creates the problem. 643 00:33:36,880 --> 00:33:38,130 Yeah.