WEBVTT 00:00:00.000 --> 00:00:03.832 [SQUEAKING][RUSTLING][CLICKING] 00:00:09.485 --> 00:00:11.110 FRANK SCHILBACH: All right, so welcome. 00:00:11.110 --> 00:00:14.410 This is lecture 21 of 14.13. 00:00:14.410 --> 00:00:16.660 We're going to talk about poverty, as seen 00:00:16.660 --> 00:00:20.740 through the lens of psychology. 00:00:20.740 --> 00:00:21.910 What are we going to do? 00:00:21.910 --> 00:00:23.950 I'm going to give you some broad overview. 00:00:23.950 --> 00:00:27.820 We're going to particularly talk about this work on scarcity 00:00:27.820 --> 00:00:31.720 by Mani et al., which is the idea that thinking-- 00:00:31.720 --> 00:00:35.020 our thoughts about poverty captures people's minds, which 00:00:35.020 --> 00:00:38.020 in turn leaves less capacity in people's minds 00:00:38.020 --> 00:00:39.280 to do other things. 00:00:39.280 --> 00:00:41.200 And then we're going to talk a little bit 00:00:41.200 --> 00:00:46.270 at the end, at least briefly, about other deprivations 00:00:46.270 --> 00:00:48.900 that poverty entails. 00:00:48.900 --> 00:00:51.070 Some of this is what I have done. 00:00:51.070 --> 00:00:55.040 Some of this work by others. 00:00:55.040 --> 00:00:58.250 So I want to start with a simple observation that 00:00:58.250 --> 00:01:02.690 is sort of at the heart of a lot of this literature. 00:01:02.690 --> 00:01:06.932 And the observation is that when you look around in the world 00:01:06.932 --> 00:01:08.390 and look for evidence, and when you 00:01:08.390 --> 00:01:11.570 work on poverty-related issues, you'll 00:01:11.570 --> 00:01:16.700 see or observe quite a few seemingly suboptimal behaviors 00:01:16.700 --> 00:01:18.020 among the poor. 00:01:18.020 --> 00:01:22.160 This would be things like lack of investment or taking 00:01:22.160 --> 00:01:24.800 advantage of high return investment opportunities. 00:01:24.800 --> 00:01:27.470 This would be like investment in fertilizer or health 00:01:27.470 --> 00:01:33.560 investments or machines or low savings, 00:01:33.560 --> 00:01:36.020 in particular some things like precautionary savings, 00:01:36.020 --> 00:01:39.350 like being ready for shocks, saving for the future, 00:01:39.350 --> 00:01:41.460 for retirement, and so on. 00:01:41.460 --> 00:01:44.330 At the same time, sort of high return 00:01:44.330 --> 00:01:47.000 or high interest rate credit, people 00:01:47.000 --> 00:01:49.490 take very high interest loans. 00:01:49.490 --> 00:01:52.920 In part-- what I've shown you earlier in the class, 00:01:52.920 --> 00:01:55.460 that's particularly prevalent among the poor. 00:01:55.460 --> 00:01:59.990 There's some evidence of worse performance of parenting 00:01:59.990 --> 00:02:02.870 among poor parents, including things like violence 00:02:02.870 --> 00:02:05.810 against children and so on. 00:02:05.810 --> 00:02:11.630 There is some evidence of lower productivity, punctuality, 00:02:11.630 --> 00:02:15.410 lower medical adherence for drugs 00:02:15.410 --> 00:02:17.750 that are sort of even very high stakes drugs that 00:02:17.750 --> 00:02:19.430 are lifesaving. 00:02:19.430 --> 00:02:23.810 There's some evidence of food and drug consumptions 00:02:23.810 --> 00:02:30.950 in terms of healthy eating or excessive drug 00:02:30.950 --> 00:02:31.940 use and the like. 00:02:31.940 --> 00:02:35.510 All of those things tend to be correlated one way or the other 00:02:35.510 --> 00:02:39.890 with different forms of income or poverty. 00:02:39.890 --> 00:02:44.480 So now you might say, well that's interesting, but now 00:02:44.480 --> 00:02:46.780 what explains these behaviors? 00:02:46.780 --> 00:02:47.960 How do we think about this? 00:02:47.960 --> 00:02:50.510 What [AUDIO OUT] plausible explanation kinds 00:02:50.510 --> 00:02:53.000 of autocorrelations. 00:02:53.000 --> 00:02:56.920 To be very clear, these are correlations, not 00:02:56.920 --> 00:02:58.438 causal effects. 00:02:58.438 --> 00:03:00.730 So education could explain quite a few of those things. 00:03:00.730 --> 00:03:02.910 It could explain potentially productivity 00:03:02.910 --> 00:03:06.170 and it could explain potentially parenting 00:03:06.170 --> 00:03:10.690 and it could explain some form of financial investments. 00:03:10.690 --> 00:03:13.300 It could also explain things like medical insurance. 00:03:13.300 --> 00:03:20.060 For example, if you look at in some settings where 00:03:20.060 --> 00:03:24.110 people have not sort of learned in school 00:03:24.110 --> 00:03:27.277 any basics of medicine or biology and so on and sort of 00:03:27.277 --> 00:03:29.360 and it's not obvious that you should take medicine 00:03:29.360 --> 00:03:31.580 from a traditional doctor versus from some healer 00:03:31.580 --> 00:03:33.913 in the village, because how would you know what's better 00:03:33.913 --> 00:03:34.628 and what's worse. 00:03:34.628 --> 00:03:36.420 And sort of presumably some of these things 00:03:36.420 --> 00:03:38.820 you would learn in school. 00:03:38.820 --> 00:03:41.210 So one broad explanatory factor is 00:03:41.210 --> 00:03:43.330 sort of like, [INAUDIBLE] variables if you want, 00:03:43.330 --> 00:03:46.860 and one of them is potentially education. 00:03:46.860 --> 00:03:49.710 It's not about omitted variable. 00:03:49.710 --> 00:03:54.900 This is about a treatment effect of poverty itself. 00:03:54.900 --> 00:03:56.890 And I'm going to be more specific about that. 00:03:56.890 --> 00:04:00.615 But essentially it's something about being in poverty itself 00:04:00.615 --> 00:04:01.740 affects people's behaviors. 00:04:01.740 --> 00:04:04.047 So it causes them-- 00:04:04.047 --> 00:04:05.880 some of these outcomes, some of these things 00:04:05.880 --> 00:04:06.810 that we've seen here. 00:04:06.810 --> 00:04:09.480 We'll talk about that in detail, but that's exactly right. 00:04:09.480 --> 00:04:13.260 In some ways, people would say, even for-- 00:04:13.260 --> 00:04:15.520 if you take credit for example, or for some things, 00:04:15.520 --> 00:04:17.712 you probably can't explain. 00:04:17.712 --> 00:04:19.170 But for high return investment, you 00:04:19.170 --> 00:04:20.670 would say, while there's uncertainty 00:04:20.670 --> 00:04:22.620 and there's a good chance that there might be losses 00:04:22.620 --> 00:04:24.210 at the end of the day, and if you're 00:04:24.210 --> 00:04:27.210 really close to subsistence, you might not 00:04:27.210 --> 00:04:31.290 want to invest in stocks, because if a bad shock happens, 00:04:31.290 --> 00:04:33.055 that's really bad. 00:04:33.055 --> 00:04:34.430 So I think some of that is right. 00:04:34.430 --> 00:04:37.625 And punctuality, for example, could also 00:04:37.625 --> 00:04:38.750 be in some way uncertainty. 00:04:38.750 --> 00:04:40.820 It's like you try to take a bus to work 00:04:40.820 --> 00:04:42.570 and the bus just comes every hour, 00:04:42.570 --> 00:04:44.570 and then sometimes it comes a half an hour early 00:04:44.570 --> 00:04:45.778 and sometimes it comes later. 00:04:45.778 --> 00:04:48.110 And it's really hard to be actually punctual. 00:04:48.110 --> 00:04:50.840 Could just be some people are not productive 00:04:50.840 --> 00:04:52.460 and are just bad at doing stuff. 00:04:52.460 --> 00:04:55.340 And therefore, they become poor and then 00:04:55.340 --> 00:04:57.555 they end up being poor because of that. 00:04:57.555 --> 00:04:59.180 So that's essentially, I think, that we 00:04:59.180 --> 00:05:00.472 covered pretty much everything. 00:05:00.472 --> 00:05:03.170 So one is it could be sort of environmental conditions, which 00:05:03.170 --> 00:05:06.230 is transportation, it's really hard to get to work on time, 00:05:06.230 --> 00:05:07.970 or have the community for two hours 00:05:07.970 --> 00:05:10.730 and then you're really tired and that makes you less productive. 00:05:10.730 --> 00:05:12.440 Things like predatory lending sort 00:05:12.440 --> 00:05:14.630 of tricks you into lending. 00:05:14.630 --> 00:05:17.240 There could some some things about institutional structures, 00:05:17.240 --> 00:05:20.900 which is to say, I told you for example at MIT, 00:05:20.900 --> 00:05:23.510 there's a default-- 00:05:23.510 --> 00:05:25.340 or at other places, there's defaults 00:05:25.340 --> 00:05:26.690 for people to save money. 00:05:26.690 --> 00:05:28.880 And there's a default to get health insurance. 00:05:28.880 --> 00:05:31.550 And there's defaults for all sorts of good behaviors, 00:05:31.550 --> 00:05:36.470 like for vaccinations and so on and so forth. 00:05:36.470 --> 00:05:38.850 We are sort of essentially pushed by society, 00:05:38.850 --> 00:05:40.820 by the environmental conditions that we 00:05:40.820 --> 00:05:44.032 face into doing the right thing or whatever good behaviors, 00:05:44.032 --> 00:05:45.240 whatever you want to call it. 00:05:45.240 --> 00:05:48.650 And so the poor might just not have those things. 00:05:48.650 --> 00:05:50.210 So for example, having a bank account 00:05:50.210 --> 00:05:55.558 might be like if you paid daily or you can buy by trip. 00:05:55.558 --> 00:05:57.350 The rickshaw drivers that I was showing you 00:05:57.350 --> 00:05:59.900 earlier in the semester, they paid trip by trip 00:05:59.900 --> 00:06:02.960 and then they'd pass by the liquor stores, 00:06:02.960 --> 00:06:05.738 essentially daily, and have a bunch of cash on hand. 00:06:05.738 --> 00:06:08.280 Well, that's much trickier to sort of resist sometimes, since 00:06:08.280 --> 00:06:11.180 if you were particularly also prone to physical pain, 00:06:11.180 --> 00:06:13.790 and then sort of to use alcohol as a way 00:06:13.790 --> 00:06:15.650 to deal with your physical pain. 00:06:15.650 --> 00:06:18.950 As opposed to somebody who paid monthly and half of the money 00:06:18.950 --> 00:06:22.250 goes into the rent anyway and then other expenses and so on, 00:06:22.250 --> 00:06:23.993 which is much easier to manage. 00:06:23.993 --> 00:06:25.910 So that's kind of broadly I would summarize it 00:06:25.910 --> 00:06:28.500 as environmental conditions. 00:06:28.500 --> 00:06:31.170 Then there's selection and omitted variables, 00:06:31.170 --> 00:06:33.480 which is, I think Joseph has mentioned education, 00:06:33.480 --> 00:06:35.170 which is a very good one. 00:06:35.170 --> 00:06:37.790 There's things like intelligence, 00:06:37.790 --> 00:06:40.160 talent, effort, mistakes, and so on. 00:06:40.160 --> 00:06:43.850 You might say, well, if you are more prone-- if you're 00:06:43.850 --> 00:06:46.560 less productive, for example, if you're prone to making mistakes 00:06:46.560 --> 00:06:48.860 and so on, well that sort of leads 00:06:48.860 --> 00:06:52.610 to worse outcomes and that makes you end up being poor. 00:06:52.610 --> 00:06:57.530 And that's really sort of in some sense, a right wing 00:06:57.530 --> 00:06:59.665 kind of view, that people are-- 00:06:59.665 --> 00:07:01.290 they should do whatever they want to do 00:07:01.290 --> 00:07:04.380 and if the end of being poor, it's sort of their fault, 00:07:04.380 --> 00:07:05.030 and so on. 00:07:05.030 --> 00:07:07.430 That's of course not what we're going to argue here. 00:07:07.430 --> 00:07:08.600 What we're interested in-- or what 00:07:08.600 --> 00:07:11.225 I'm trying to argue [AUDIO OUT] show you some evidence off this 00:07:11.225 --> 00:07:15.770 that there's a treatment effect of poverty itself potentially. 00:07:15.770 --> 00:07:20.420 Which is to say that being poor itself causes these behaviors. 00:07:20.420 --> 00:07:24.240 And again, I'll show you the evidence in two ways. 00:07:24.240 --> 00:07:28.190 One is by inducing thoughts about money, 00:07:28.190 --> 00:07:31.220 essentially thinking about can you pay school fees, 00:07:31.220 --> 00:07:33.860 can pay your rent, can you get groceries for your children, 00:07:33.860 --> 00:07:36.360 are your children are going to be hungry tonight, and so on. 00:07:36.360 --> 00:07:38.120 These are all kind of thoughts that invade 00:07:38.120 --> 00:07:40.610 people's minds involuntarily. 00:07:40.610 --> 00:07:44.150 And by doing so, they capture people's attention. 00:07:44.150 --> 00:07:47.090 And then since attention overall is limited, 00:07:47.090 --> 00:07:50.660 people have less attention available for other things. 00:07:50.660 --> 00:07:55.120 That's sort of one broad theory. 00:07:55.120 --> 00:08:00.520 And I'll focus on that mostly which 00:08:00.520 --> 00:08:03.040 people refer to it under the name of scarcity. 00:08:03.040 --> 00:08:06.250 And then there's a bunch of other aspects of poverty 00:08:06.250 --> 00:08:08.950 as well that might be competing with people's 00:08:08.950 --> 00:08:12.370 cognitive function, their attention, that tire them out. 00:08:12.370 --> 00:08:16.840 They're just not sleeping well, be in pain, being exposed 00:08:16.840 --> 00:08:21.250 to pollution, environmental pollution, noise pollution, 00:08:21.250 --> 00:08:23.630 and so on, and so forth. 00:08:23.630 --> 00:08:27.045 And so now what's the underlying idea here? 00:08:27.045 --> 00:08:28.920 Well again, the underlying idea is that there 00:08:28.920 --> 00:08:31.240 might be what's called-- 00:08:31.240 --> 00:08:33.605 people might want to call it behavioral poverty trap. 00:08:33.605 --> 00:08:34.980 And the broad line of argument is 00:08:34.980 --> 00:08:37.679 that poverty affects people's cognition, decision making, 00:08:37.679 --> 00:08:38.909 or productivity. 00:08:38.909 --> 00:08:41.370 These factors in turn then influence 00:08:41.370 --> 00:08:45.180 people's future poverty, because if you're not productive 00:08:45.180 --> 00:08:46.680 and you can't save money and so on, 00:08:46.680 --> 00:08:49.020 you're going to be more poor in the future. 00:08:49.020 --> 00:08:50.640 And if those things are going on, 00:08:50.640 --> 00:08:54.330 there is essentially some form of a vicious cycle going on, 00:08:54.330 --> 00:08:57.490 which makes it difficult, if not impossible, 00:08:57.490 --> 00:08:59.370 for a person to become rich. 00:08:59.370 --> 00:09:02.428 Now notice at the very least, that will lower mobility. 00:09:02.428 --> 00:09:04.720 Sometimes if you're poor and you want to become richer, 00:09:04.720 --> 00:09:06.637 you have got to deal with a bunch of bad stuff 00:09:06.637 --> 00:09:10.260 that happens to you every day. 00:09:10.260 --> 00:09:13.340 But [INAUDIBLE] mobility is low, [INAUDIBLE] might be actually 00:09:13.340 --> 00:09:13.840 a trap. 00:09:13.840 --> 00:09:17.190 It's just next to or actually impossible to become richer, 00:09:17.190 --> 00:09:21.690 because all these forces are too strong to overcome for people. 00:09:21.690 --> 00:09:24.610 So today I'm going to talk about two broad lines of research. 00:09:24.610 --> 00:09:27.480 I'm going to focus mostly on the first one, which 00:09:27.480 --> 00:09:30.763 is monetary concerns capture people's cognitive function. 00:09:30.763 --> 00:09:33.180 I'm going to talk a little bit about other poverty-induced 00:09:33.180 --> 00:09:35.460 deprivation of behaviors which might have sort 00:09:35.460 --> 00:09:39.180 of similar effects overall. 00:09:39.180 --> 00:09:43.780 So now I want to do a simple test with you. 00:09:43.780 --> 00:09:47.970 So what I'd like you to do is to-- 00:09:47.970 --> 00:09:49.530 sorry, one second. 00:09:49.530 --> 00:09:52.750 I'm trying to-- give me a sec. 00:09:52.750 --> 00:09:54.418 I'd like you to-- 00:09:54.418 --> 00:09:55.210 hold on one second. 00:09:55.210 --> 00:09:56.224 One second. 00:09:59.900 --> 00:10:00.400 All right. 00:10:00.400 --> 00:10:00.710 Ready? 00:10:00.710 --> 00:10:01.850 Don't write anything down. 00:10:01.850 --> 00:10:05.660 Just try to remember it and then try to recall it in a bit. 00:10:05.660 --> 00:10:07.230 I'm going to start reading. 00:10:07.230 --> 00:10:17.880 Bed, rest, dream, doze, snore, slumber, blanket, snooze, wake, 00:10:17.880 --> 00:10:20.565 nap, awake, tired. 00:10:29.380 --> 00:10:32.950 OK, now please write down as many words as you remember. 00:11:08.853 --> 00:11:09.520 10 more seconds. 00:11:23.430 --> 00:11:24.830 All right, so done. 00:11:24.830 --> 00:11:27.948 So usually I would have people raise their hands, which 00:11:27.948 --> 00:11:28.740 is a little tricky. 00:11:28.740 --> 00:11:30.750 But I'm going to just read you the list of words 00:11:30.750 --> 00:11:31.708 and you can nod to me-- 00:11:31.708 --> 00:11:33.060 I can see some of you, at least. 00:11:33.060 --> 00:11:34.518 You can nod and tell me whether you 00:11:34.518 --> 00:11:40.082 remember the words that I read. 00:11:40.082 --> 00:11:41.040 OK, I'm going to start. 00:11:41.040 --> 00:11:47.160 Bed, rest-- pretty good-- 00:11:47.160 --> 00:11:57.460 dream, doze, snore-- no? 00:11:57.460 --> 00:12:15.700 Slumber, sleep, blanket, snooze, wake, nap, awake, tired. 00:12:19.010 --> 00:12:20.840 So I know this is making everybody sleepy 00:12:20.840 --> 00:12:22.420 and want to take a nap. 00:12:22.420 --> 00:12:23.350 Me, too. 00:12:23.350 --> 00:12:27.650 So here's the list of words that I asked you to remember. 00:12:27.650 --> 00:12:30.003 Now one thing that's kind of nice 00:12:30.003 --> 00:12:33.680 [INAUDIBLE] most people remembered quite a few things. 00:12:33.680 --> 00:12:35.180 The majority, for example, remembers 00:12:35.180 --> 00:12:37.320 bed and so on and so forth. 00:12:37.320 --> 00:12:40.100 Now what's quite interesting about these words is 00:12:40.100 --> 00:12:41.930 that while I remember-- 00:12:41.930 --> 00:12:43.550 I put all these names up here, which 00:12:43.550 --> 00:12:46.550 is-- these are the words that I actually read. 00:12:46.550 --> 00:12:48.830 Now one word that's actually missing in this list 00:12:48.830 --> 00:12:51.350 here is sleep. 00:12:51.350 --> 00:12:53.240 And so you sort of-- 00:12:53.240 --> 00:12:56.780 I snuck this in here on-- 00:12:56.780 --> 00:12:58.700 this is why I had to go back. 00:12:58.700 --> 00:13:00.920 It turns out I actually didn't read the word sleep, 00:13:00.920 --> 00:13:03.410 but quite a few of you were nodding happily 00:13:03.410 --> 00:13:05.090 when I asked you about sleep. 00:13:05.090 --> 00:13:09.960 And what's the reason why you remembered sleep? 00:13:09.960 --> 00:13:13.180 So all of these words is like-- 00:13:13.180 --> 00:13:15.640 they all around sleep. 00:13:15.640 --> 00:13:20.360 So it's really like sleep should have been there. 00:13:20.360 --> 00:13:21.860 It's kind of the type of thing where 00:13:21.860 --> 00:13:25.250 people would ask you, have you been to the parity last week, 00:13:25.250 --> 00:13:27.950 where people just assume-- where essentially all of the friends 00:13:27.950 --> 00:13:29.750 were there-- people just assume that you were there, 00:13:29.750 --> 00:13:31.610 as well, because you're supposed to be there. 00:13:31.610 --> 00:13:33.320 It's like sleep is supposed to be in this list, 00:13:33.320 --> 00:13:34.445 but we didn't put it there. 00:13:34.445 --> 00:13:36.800 And this is what is called false memory tests. 00:13:36.800 --> 00:13:39.258 And there's quite a bit of work on actually false memories. 00:13:39.258 --> 00:13:41.420 And people's minds essentially make up things. 00:13:41.420 --> 00:13:43.130 In particular, stuff that's sort of-- 00:13:43.130 --> 00:13:46.670 or where you have all situations that you think 00:13:46.670 --> 00:13:47.960 that should be the case. 00:13:47.960 --> 00:13:52.730 You can do the same with man and other lists and so on. 00:13:52.730 --> 00:13:56.330 And now psychologists have been doing studies 00:13:56.330 --> 00:13:57.710 for the rich and the poor. 00:13:57.710 --> 00:13:59.000 And so here's two lists. 00:13:59.000 --> 00:14:01.470 One is about sort of money-related stuff. 00:14:01.470 --> 00:14:05.660 And one is about sort of just anything that's related to, 00:14:05.660 --> 00:14:09.133 I guess, man or gender in some ways. 00:14:09.133 --> 00:14:10.550 And they looked at how many people 00:14:10.550 --> 00:14:12.160 have certain false memory. 00:14:12.160 --> 00:14:13.910 And what these studies show is essentially 00:14:13.910 --> 00:14:16.880 that when you take this list for a man 00:14:16.880 --> 00:14:21.590 and you do this with high income and low income people, 00:14:21.590 --> 00:14:27.890 the fraction who remember a man wrongly is about 15% to 20%. 00:14:27.890 --> 00:14:31.730 The fraction among the poor who remember money 00:14:31.730 --> 00:14:33.800 is similarly high. 00:14:33.800 --> 00:14:36.590 Among the rich, that's not the case. 00:14:36.590 --> 00:14:37.340 Now why is that? 00:14:37.340 --> 00:14:40.400 Well, it seems to be that for the poor, when you read lists 00:14:40.400 --> 00:14:42.890 like groceries, expense, bill, gas, et 00:14:42.890 --> 00:14:44.830 cetera, what they think about is essentially, 00:14:44.830 --> 00:14:46.080 how I'm going to pay for this. 00:14:46.080 --> 00:14:48.020 And they think about money. 00:14:48.020 --> 00:14:50.330 When for the rich, 00:14:50.330 --> 00:14:52.368 Instead, when you say groceries or whatever, 00:14:52.368 --> 00:14:54.410 they think about what I'm going to buy for dinner 00:14:54.410 --> 00:14:56.285 and what are the nice things I'm going to do. 00:14:56.285 --> 00:14:58.502 And for them, this is really not top of mind. 00:14:58.502 --> 00:15:00.710 So this is essentially at least one piece of evidence 00:15:00.710 --> 00:15:05.390 that money captures the minds of the poor moreso 00:15:05.390 --> 00:15:08.400 than for the rich. 00:15:08.400 --> 00:15:13.060 Now, a second study is a study about the doctor's visit. 00:15:13.060 --> 00:15:17.460 So this is a study where the participants were 00:15:17.460 --> 00:15:19.410 asked about the following situation, which 00:15:19.410 --> 00:15:22.260 is imagine that you have been feeling sick lately 00:15:22.260 --> 00:15:24.640 and finally decided to go see a doctor about it. 00:15:24.640 --> 00:15:27.090 The doctor explains that you have a serious condition that 00:15:27.090 --> 00:15:28.770 requires immediate attention. 00:15:28.770 --> 00:15:31.530 The good news, however, is you are virtually guaranteed 00:15:31.530 --> 00:15:32.940 to make a full recovery. 00:15:32.940 --> 00:15:34.710 The doctor writes several prescriptions. 00:15:34.710 --> 00:15:38.000 You will also need to make several appointments. 00:15:38.000 --> 00:15:41.040 What would be on your mind or how would you feel 00:15:41.040 --> 00:15:42.750 about when you hear this news? 00:15:42.750 --> 00:15:45.930 What are the three things that you think about it? 00:15:45.930 --> 00:15:47.813 Or you can think about for a second what 00:15:47.813 --> 00:15:49.980 is it that you would feel, and what are you thinking 00:15:49.980 --> 00:15:52.080 about when you see this. 00:15:52.080 --> 00:15:53.970 And what's particularly interesting 00:15:53.970 --> 00:15:56.890 is how when you look at people's responses 00:15:56.890 --> 00:16:01.440 some people say wife, son, co-worker. 00:16:01.440 --> 00:16:06.330 Some people are scared, afraid, worried, relief, hope, joy, 00:16:06.330 --> 00:16:11.850 so like emotions or just of relationships and other people. 00:16:11.850 --> 00:16:13.860 Not exactly sure what the co-workers doing there 00:16:13.860 --> 00:16:19.380 but anyway, but what's particularly interesting 00:16:19.380 --> 00:16:22.360 is the fraction of people who say costs. 00:16:22.360 --> 00:16:27.540 And as you can imagine, when you do this by low and high income 00:16:27.540 --> 00:16:34.050 when you think about medical conditions for the rich, 00:16:34.050 --> 00:16:37.470 this is much more about emotions about their health, 00:16:37.470 --> 00:16:40.810 about their relationships, their loved ones, and so on. 00:16:40.810 --> 00:16:44.310 And when the poor think about health or health expenditures 00:16:44.310 --> 00:16:46.350 or health issues, they think very much 00:16:46.350 --> 00:16:48.630 about health expenditures, and that is essentially 00:16:48.630 --> 00:16:51.070 what's top of mind for them. 00:16:51.070 --> 00:16:53.700 And so again, when you split the sample by low and high income 00:16:53.700 --> 00:16:58.080 people, the poor have much higher proportions of money 00:16:58.080 --> 00:17:01.890 really good thoughts as reported by these kinds of questions 00:17:01.890 --> 00:17:03.240 compared to the rich. 00:17:03.240 --> 00:17:06.651 The third study-- the last thing I'm going to show you 00:17:06.651 --> 00:17:08.109 is the study that I showed you very 00:17:08.109 --> 00:17:11.560 much at the beginning of the class, which was a study about 00:17:11.560 --> 00:17:14.990 imagine you are about to purchase an iPad for $500. 00:17:14.990 --> 00:17:18.670 The salesman says you can get the same iPad, exact same iPad 00:17:18.670 --> 00:17:20.260 for $15 off. 00:17:20.260 --> 00:17:23.800 Would you walk for 30 minutes in total for $15. 00:17:23.800 --> 00:17:25.273 Would you go for the other store? 00:17:25.273 --> 00:17:27.190 And similarly, you'd have to do the same thing 00:17:27.190 --> 00:17:28.760 for an iPad case. 00:17:28.760 --> 00:17:31.000 And so the result that we found in this class 00:17:31.000 --> 00:17:32.560 and like the other studies have found 00:17:32.560 --> 00:17:34.900 is that people are way more likely to report 00:17:34.900 --> 00:17:38.740 that they would walk for the iPad than for the iPad itself. 00:17:38.740 --> 00:17:41.320 And the explanation, of course, is 00:17:41.320 --> 00:17:45.340 that people evaluate things in relative terms, right. 00:17:45.340 --> 00:17:46.840 So if you think about the iPad case, 00:17:46.840 --> 00:17:49.870 $15 is really high relative to like 00:17:49.870 --> 00:17:54.700 $30 in the iPad cases of value, while for the iPad itself, 00:17:54.700 --> 00:17:58.960 $15 compared to $500 is not very much. 00:17:58.960 --> 00:18:01.420 Now that's some form of money illusion 00:18:01.420 --> 00:18:03.820 or some form of community confusion or the like, 00:18:03.820 --> 00:18:05.440 because at the end of the day you 00:18:05.440 --> 00:18:08.920 should just be willing to walk for 30 minutes for either $15 00:18:08.920 --> 00:18:09.610 or not. 00:18:09.610 --> 00:18:11.860 That's not going to change depending on whether you 00:18:11.860 --> 00:18:15.200 buy an iPad or an iPad case. 00:18:15.200 --> 00:18:17.740 Now many studies have found the kind of patterns 00:18:17.740 --> 00:18:21.130 including like in this class, but once you 00:18:21.130 --> 00:18:23.020 do that with the poor, they're much less 00:18:23.020 --> 00:18:26.400 likely if at all to engage in this kind of behavior, which 00:18:26.400 --> 00:18:28.380 seems to suggest that the poor, in fact, 00:18:28.380 --> 00:18:30.870 are less prone to this, kind of, labeling or, sort of, 00:18:30.870 --> 00:18:34.710 like monetary sort of comparisions. 00:18:34.710 --> 00:18:36.572 For them essentially $1 is $1 and they, 00:18:36.572 --> 00:18:38.030 sort of, understand that, like, $15 00:18:38.030 --> 00:18:42.660 is valuable regardless of what is surrounding comparisons 00:18:42.660 --> 00:18:43.857 like. 00:18:43.857 --> 00:18:45.940 And so if I have to summarize what I've just shown 00:18:45.940 --> 00:18:49.330 you very quickly is that thoughts about money 00:18:49.330 --> 00:18:53.530 tend to capture or take much more space in the poor's minds 00:18:53.530 --> 00:18:56.200 instead of the first two studies that I showed you, 00:18:56.200 --> 00:18:58.060 and second, the extra focus on money 00:18:58.060 --> 00:19:01.120 has actually some positive consequences in some sense. 00:19:01.120 --> 00:19:04.090 In some sense, you can say here this is an irrational decision 00:19:04.090 --> 00:19:07.840 to say, yes, that you will go for the iPad case 00:19:07.840 --> 00:19:12.230 compared to the iPad or some of this optimization going on. 00:19:12.230 --> 00:19:14.590 So the poor are actually doing better or the less prone 00:19:14.590 --> 00:19:18.280 to these kinds of errors compared to the rich. 00:19:18.280 --> 00:19:20.440 I think that's a good thing for like potentially 00:19:20.440 --> 00:19:24.020 better monetary decisions here. 00:19:24.020 --> 00:19:26.330 So there's some positive consequences 00:19:26.330 --> 00:19:29.680 for impacts based on that. 00:19:29.680 --> 00:19:32.440 now, and this is where we're going to go, is, 00:19:32.440 --> 00:19:35.050 well, are there also some negative effects potentially? 00:19:35.050 --> 00:19:37.930 The reason being that if you have a certain amount 00:19:37.930 --> 00:19:40.600 of attention available overall, or cognitive resources 00:19:40.600 --> 00:19:43.240 available, if your brain is taken up 00:19:43.240 --> 00:19:47.680 by essentially these thoughts about money, 00:19:47.680 --> 00:19:50.380 then there will be less available resources 00:19:50.380 --> 00:19:53.310 for other stuff. 00:19:53.310 --> 00:19:55.398 Any questions on this so far? 00:20:04.860 --> 00:20:10.240 OK, so then let's talk about the scarcity work specifically. 00:20:10.240 --> 00:20:13.050 So scarcity is actually defined generally here 00:20:13.050 --> 00:20:15.660 as not having enough of something. 00:20:15.660 --> 00:20:19.150 This could be money, friends, hair, time, et cetera. 00:20:19.150 --> 00:20:21.420 It's very general, actually. 00:20:21.420 --> 00:20:23.010 And so the idea is just like if you 00:20:23.010 --> 00:20:25.280 have something that you think you don't have enough. 00:20:25.280 --> 00:20:27.780 Notice that the definition is pretty vague in the sense that 00:20:27.780 --> 00:20:30.773 like it's not just saying like having low income, 00:20:30.773 --> 00:20:32.190 but it's sort of thinking that you 00:20:32.190 --> 00:20:34.080 don't have enough income, which could be also 00:20:34.080 --> 00:20:36.290 like in relative terms and so on. 00:20:36.290 --> 00:20:40.160 And then the hypothesis is that scarcity captured 00:20:40.160 --> 00:20:43.670 people's mental cognitive capacity, their bandwidth, 00:20:43.670 --> 00:20:44.960 as they call it. 00:20:44.960 --> 00:20:47.645 This happens automatically and not intentionally. 00:20:47.645 --> 00:20:49.520 That is to say, that it's not people choosing 00:20:49.520 --> 00:20:52.860 to spend a lot of time on thinking about money, 00:20:52.860 --> 00:20:55.920 but rather, money captures their minds, 00:20:55.920 --> 00:20:59.383 and then they have trouble thinking about other things. 00:20:59.383 --> 00:21:00.800 Think about this as like, it would 00:21:00.800 --> 00:21:03.710 be like suppose you have like some fights 00:21:03.710 --> 00:21:06.440 with a good friend of yours, or a partner or the like, 00:21:06.440 --> 00:21:09.120 and then you have an important job interview. 00:21:09.120 --> 00:21:10.730 It's really not productive to think 00:21:10.730 --> 00:21:16.280 about the fight with your friend or family member 00:21:16.280 --> 00:21:17.545 or partner or the like. 00:21:17.545 --> 00:21:18.920 During the job interview, there's 00:21:18.920 --> 00:21:20.520 nothing productive about doing that. 00:21:20.520 --> 00:21:22.145 You should be focusing on the interview 00:21:22.145 --> 00:21:23.220 and try to do your best. 00:21:23.220 --> 00:21:25.520 But it might be that like essentially, these thoughts 00:21:25.520 --> 00:21:27.560 invade your minds and you sort of get really 00:21:27.560 --> 00:21:29.810 distracted unintentionally, and then 00:21:29.810 --> 00:21:32.030 do worse in those kinds of-- 00:21:32.030 --> 00:21:34.460 And that's the idea here similarly here 00:21:34.460 --> 00:21:38.500 as well, which is that essentially, there 00:21:38.500 --> 00:21:41.690 might be some benefit of figuring out monetary problems 00:21:41.690 --> 00:21:44.250 and figuring out how to pay the bills and so on and so forth. 00:21:44.250 --> 00:21:47.630 But overall, these benefits might 00:21:47.630 --> 00:21:53.030 be overwhelmed by taking over people's brain 00:21:53.030 --> 00:21:59.010 that leaves less resources for other things, the decisions 00:21:59.010 --> 00:22:02.800 or productivity that people have to do in life. 00:22:02.800 --> 00:22:05.950 And so in particular, poverty now, so this 00:22:05.950 --> 00:22:07.420 is a general introduction. 00:22:07.420 --> 00:22:10.840 Poverty specifically makes monetary concerns top of mind, 00:22:10.840 --> 00:22:13.690 so the poor might be acting more rational in some ways, 00:22:13.690 --> 00:22:15.880 in some sense, in financial choices, which 00:22:15.880 --> 00:22:18.100 is kind of what the evidence that I just showed you 00:22:18.100 --> 00:22:21.230 in there, several studies that try to make that point. 00:22:21.230 --> 00:22:25.030 But you know, scarcity might then deteriorate performance 00:22:25.030 --> 00:22:27.480 at other cognitive tasks. 00:22:27.480 --> 00:22:30.285 Now, here's sort of a PowerPoint version of that, 00:22:30.285 --> 00:22:31.660 which, you know, people may think 00:22:31.660 --> 00:22:32.743 about many things in life. 00:22:32.743 --> 00:22:34.270 They might think about rent. 00:22:34.270 --> 00:22:36.610 They might think about the kids doing well, 00:22:36.610 --> 00:22:41.977 eating healthy, utilities, car payments, and so on. 00:22:41.977 --> 00:22:43.060 But what does scarcity do? 00:22:43.060 --> 00:22:44.830 It magnifies, essentially, people's 00:22:44.830 --> 00:22:47.410 thoughts about rent, car payments and utility bills. 00:22:47.410 --> 00:22:50.350 These are all monetary things that people think about a lot, 00:22:50.350 --> 00:22:53.380 and thn it might sort of like drown out or like force 00:22:53.380 --> 00:22:56.380 out other thoughts about other things, about eating healthy 00:22:56.380 --> 00:22:59.890 or about taking care of the kids in other ways 00:22:59.890 --> 00:23:02.680 than paying the bills, OK. 00:23:02.680 --> 00:23:04.690 And so now, I'm going to show you 00:23:04.690 --> 00:23:07.260 evidence of two studies that sort of support this point. 00:23:07.260 --> 00:23:09.668 The first one, the paper is called 00:23:09.668 --> 00:23:11.210 "Poverty impedes cognitive function." 00:23:11.210 --> 00:23:12.860 That was the reading for today. 00:23:12.860 --> 00:23:14.907 The first one is called the mall study. 00:23:14.907 --> 00:23:16.240 So what does this mall study do? 00:23:16.240 --> 00:23:18.850 It's kind of like a lab experiment, 00:23:18.850 --> 00:23:21.700 but it's sort of in some sense like a field experiment. 00:23:21.700 --> 00:23:27.850 So it's a major Jersey mall that asks shoppers or stops shoppers 00:23:27.850 --> 00:23:31.240 with a median household income of roughly $70,000, 00:23:31.240 --> 00:23:35.110 so sort of richer than the average American, 00:23:35.110 --> 00:23:39.010 but with quite a bit of range of income overall. 00:23:39.010 --> 00:23:42.520 So there's like very rich people and fairly poor people. 00:23:42.520 --> 00:23:45.910 And then there's a sequence of three events, essentially, 00:23:45.910 --> 00:23:48.820 which is, people are asked hypothetical questions 00:23:48.820 --> 00:23:50.230 involving money. 00:23:50.230 --> 00:23:52.330 Some are hard and some are easy. 00:23:52.330 --> 00:23:54.670 And then people are asked to do cognitive tests, 00:23:54.670 --> 00:23:57.940 a non-verbal test that sort of measures people's 00:23:57.940 --> 00:24:00.340 fluid intelligence, which is Raven's Matrices, which 00:24:00.340 --> 00:24:01.765 you might be familiar with. 00:24:01.765 --> 00:24:03.730 They're sort of IQ type tests. 00:24:03.730 --> 00:24:05.440 And some form of cognitive control, 00:24:05.440 --> 00:24:06.940 which essentially is about like, are 00:24:06.940 --> 00:24:14.350 you able to overwrite automatic responses when 00:24:14.350 --> 00:24:15.970 doing cognitive tasks. 00:24:15.970 --> 00:24:19.180 And then number three, then, is like people that answer 00:24:19.180 --> 00:24:20.540 the hypothetical questions. 00:24:20.540 --> 00:24:21.850 So what does this look like? 00:24:21.850 --> 00:24:23.950 A hypothetical question is involving 00:24:23.950 --> 00:24:25.960 money, which is the following. 00:24:25.960 --> 00:24:28.810 Your car is having some trouble and requires x dollars 00:24:28.810 --> 00:24:29.770 to be fixed. 00:24:29.770 --> 00:24:32.710 You can pay in full, take a loan or take a chance 00:24:32.710 --> 00:24:34.880 and forgo the service at the moment. 00:24:34.880 --> 00:24:37.525 How would you go about making this decision? 00:24:37.525 --> 00:24:39.650 And so then, people ask this question and they say, 00:24:39.650 --> 00:24:41.442 well, you can think about this for a while. 00:24:41.442 --> 00:24:43.420 And afterwards, I'm going to ask you 00:24:43.420 --> 00:24:45.430 about the answer to that question. 00:24:45.430 --> 00:24:49.570 And then there's a hard version and an easy version. 00:24:49.570 --> 00:24:53.470 The idea is that the experiment wants to keep everything 00:24:53.470 --> 00:24:56.680 the same for like across treatments, 00:24:56.680 --> 00:25:00.460 and the idea is that $150 for the average American 00:25:00.460 --> 00:25:02.050 is not that much money, so it's not 00:25:02.050 --> 00:25:04.030 going to be a huge problem for most people, 00:25:04.030 --> 00:25:06.250 at least in normal times of their lives. 00:25:06.250 --> 00:25:09.670 But $1,500 is actually a lot of money for like a low income 00:25:09.670 --> 00:25:11.770 person in the US, but perhaps not 00:25:11.770 --> 00:25:14.530 that big of a deal for a rich person. 00:25:14.530 --> 00:25:17.152 Like for a rich person, the $1,500 or $150 are both fine. 00:25:17.152 --> 00:25:19.360 They're going to have some money in the bank account, 00:25:19.360 --> 00:25:21.110 and they're going to deal with it somehow. 00:25:21.110 --> 00:25:22.150 It's not a huge issue. 00:25:22.150 --> 00:25:27.580 For the poor person, $150 is not great, but not a huge problem 00:25:27.580 --> 00:25:28.750 or not a huge headache. 00:25:28.750 --> 00:25:31.120 But $1,500 surely is really bad. 00:25:31.120 --> 00:25:34.510 That's sort of the idea here, right. 00:25:34.510 --> 00:25:38.435 Now what the study that finds is that-- and let 00:25:38.435 --> 00:25:39.810 me walk you through these graphs. 00:25:39.810 --> 00:25:41.580 Let's look on the left side. 00:25:41.580 --> 00:25:43.650 On the right side is quite similar. 00:25:43.650 --> 00:25:46.950 You'll see here always four bars. 00:25:46.950 --> 00:25:52.440 The dark bars are how people are doing after the hard task, 00:25:52.440 --> 00:25:56.760 and the lighter bars after the easy task. 00:25:56.760 --> 00:25:59.940 Let's start with the light gray bars, 00:25:59.940 --> 00:26:01.703 comparing the poor versus the rich. 00:26:01.703 --> 00:26:03.370 And what you essentially see is the poor 00:26:03.370 --> 00:26:05.730 and the rich are doing pretty much the same 00:26:05.730 --> 00:26:08.563 after doing the easy task. 00:26:08.563 --> 00:26:10.230 But to be clear, what's poor versus rich 00:26:10.230 --> 00:26:13.150 is essentially being split by the median income. 00:26:13.150 --> 00:26:15.773 So if you're above the median, you're called rich. 00:26:15.773 --> 00:26:17.190 If you're below the median, you're 00:26:17.190 --> 00:26:19.830 called poor for the purpose of this study. 00:26:19.830 --> 00:26:24.460 Now, for the easy task here, this 00:26:24.460 --> 00:26:27.220 is like the rich do essentially as well or as badly 00:26:27.220 --> 00:26:28.330 as the poor do. 00:26:28.330 --> 00:26:35.200 They get about 40% correct in these Raven's puzzles, 00:26:35.200 --> 00:26:37.070 and the poor do essentially the same. 00:26:37.070 --> 00:26:39.640 So there's no difference after the easy task. 00:26:39.640 --> 00:26:41.530 Now, after the hard task, for the rich, 00:26:41.530 --> 00:26:42.040 essentially it's the same. 00:26:42.040 --> 00:26:43.750 If anything, they do a little bit better. 00:26:43.750 --> 00:26:46.930 The difference is not statistically significant. 00:26:46.930 --> 00:26:51.160 In contrast, you see like a clear drop in performance 00:26:51.160 --> 00:26:52.500 among the poor, OK. 00:26:52.500 --> 00:26:56.050 They do about like 12 percentage points worse. 00:26:56.050 --> 00:27:00.830 This is really like a huge difference in how well people-- 00:27:00.830 --> 00:27:02.320 so this is for Raven's Matrices. 00:27:02.320 --> 00:27:04.420 This is for what's called hearts and flowers 00:27:04.420 --> 00:27:06.460 for cognitive control, which is essentially 00:27:06.460 --> 00:27:10.590 how good are people at controlling their attention 00:27:10.590 --> 00:27:11.377 or resources. 00:27:11.377 --> 00:27:13.210 And what you see is essentially more or less 00:27:13.210 --> 00:27:15.860 the same pattern, that for the easy task, 00:27:15.860 --> 00:27:18.740 the rich and the poor do pretty much the same. 00:27:18.740 --> 00:27:20.530 The hard task compared to the easy task 00:27:20.530 --> 00:27:21.760 does nothing for the rich. 00:27:21.760 --> 00:27:23.440 This is actually, the performance 00:27:23.440 --> 00:27:25.210 is the same or slightly different 00:27:25.210 --> 00:27:27.610 but not statistically significantly so. 00:27:27.610 --> 00:27:30.490 For the poor, there is a large drop in performance 00:27:30.490 --> 00:27:32.450 compared to the rich. 00:27:32.450 --> 00:27:33.590 Any questions on that? 00:27:40.990 --> 00:27:45.790 OK, So then I'm going to show you the second study that's 00:27:45.790 --> 00:27:46.742 related to this. 00:27:46.742 --> 00:27:48.200 So one thing you might sort of say, 00:27:48.200 --> 00:27:51.780 well, we're going to show you, already, a study, so why 00:27:51.780 --> 00:27:53.890 do we need another study when we already did one. 00:27:53.890 --> 00:27:56.710 We showed you that like poverty and thinking about poverty 00:27:56.710 --> 00:28:02.770 affects people's cognition, so why do we need more evidence. . 00:28:02.770 --> 00:28:05.600 So effort is an interesting issue that-- 00:28:05.600 --> 00:28:08.080 so there's, in fact, this study has-- 00:28:08.080 --> 00:28:10.960 and I took this out of the slides because I had too many. 00:28:10.960 --> 00:28:13.180 There's an incentivized version of this task. 00:28:13.180 --> 00:28:15.130 And the authors do find that, even 00:28:15.130 --> 00:28:18.670 once you incentivize these tasks, 00:28:18.670 --> 00:28:20.320 we find the same patterns. 00:28:20.320 --> 00:28:24.050 But even there, it's a bit sort of like a task that 00:28:24.050 --> 00:28:26.360 is not a task that people do in their real life. 00:28:26.360 --> 00:28:29.720 And in a way, which is actually not what the harvest study is 00:28:29.720 --> 00:28:33.950 addressing, you might wonder. 00:28:33.950 --> 00:28:37.040 Does this translate into real world choices and decisions 00:28:37.040 --> 00:28:40.340 that actually have consequence for people's behaviors? 00:28:40.340 --> 00:28:42.680 To be clear, these effects here, these differences 00:28:42.680 --> 00:28:44.487 in cognitive performances, are very large. 00:28:44.487 --> 00:28:46.820 This is like the order of magnitude something like 10 IQ 00:28:46.820 --> 00:28:49.220 points, which is really huge. 00:28:49.220 --> 00:28:52.340 So you might sort of argue that we think that IQ, et cetera, 00:28:52.340 --> 00:28:53.400 matters a lot. 00:28:53.400 --> 00:28:57.240 So in some sense, that's what Mullainathan and Shafir 00:28:57.240 --> 00:28:57.897 would argue. 00:28:57.897 --> 00:28:58.980 But, no, I agree with you. 00:28:58.980 --> 00:29:00.855 In some sense, we would like to see does this 00:29:00.855 --> 00:29:03.330 translate into real world choices 00:29:03.330 --> 00:29:10.850 that we see in people's everyday lives. 00:29:10.850 --> 00:29:13.630 Now, I think we talked about a few of those things. 00:29:13.630 --> 00:29:15.590 So one is external validity concerns, 00:29:15.590 --> 00:29:19.400 which is to say, look, we did this study in the mall, which 00:29:19.400 --> 00:29:21.380 is a little bit of a funny thing to do. 00:29:21.380 --> 00:29:23.980 And you hear about the scenario, which is kind of weird 00:29:23.980 --> 00:29:25.130 and maybe tricks you out. 00:29:25.130 --> 00:29:26.755 And maybe the poor are more freaked out 00:29:26.755 --> 00:29:28.760 than the rich for some reason. 00:29:28.760 --> 00:29:32.060 But really, if you think about paying back 00:29:32.060 --> 00:29:36.410 your debt or any sort of school fees or just not having money, 00:29:36.410 --> 00:29:38.510 that's a repeated exposure of something 00:29:38.510 --> 00:29:39.873 that know in your real life. 00:29:39.873 --> 00:29:41.790 For example, not having harvest versus harvest 00:29:41.790 --> 00:29:46.650 is a thing that you experience every year or every twice 00:29:46.650 --> 00:29:47.398 a year even. 00:29:47.398 --> 00:29:48.940 And so you think some of these things 00:29:48.940 --> 00:29:51.520 might sort of go away with repeat exposure. 00:29:51.520 --> 00:29:54.360 And so we kind of would like to have a real world setting where 00:29:54.360 --> 00:29:57.540 we have actually real world natural variation 00:29:57.540 --> 00:30:00.270 or sort of experimental variation potentially as 00:30:00.270 --> 00:30:04.320 opposed to the somewhat weird feature in some mall somebody 00:30:04.320 --> 00:30:07.740 talks you up and asks you some questions, which people might 00:30:07.740 --> 00:30:10.590 find funny and might not be representative for what people 00:30:10.590 --> 00:30:12.330 do with their lives. 00:30:12.330 --> 00:30:14.080 And then the second part is kind of how do 00:30:14.080 --> 00:30:15.930 we interpret magnitudes, right? 00:30:15.930 --> 00:30:21.100 So I showed up and said, look, these effects are really large. 00:30:21.100 --> 00:30:23.860 After the hard task, the poor do worse than the rich. 00:30:23.860 --> 00:30:25.570 But we have no idea how we should 00:30:25.570 --> 00:30:28.960 scale this in terms of what does it actually mean 00:30:28.960 --> 00:30:32.020 for actual effects of poverty. 00:30:32.020 --> 00:30:35.090 Suppose I give somebody $1,000 or $10,000, 00:30:35.090 --> 00:30:37.220 what does it do to people's performance? 00:30:37.220 --> 00:30:38.890 It essentially, what this shows, is 00:30:38.890 --> 00:30:40.840 that the poor are more sort of frazzled 00:30:40.840 --> 00:30:42.155 by these financial questions. 00:30:42.155 --> 00:30:43.780 But it gives you essentially-- and that 00:30:43.780 --> 00:30:45.697 seems to be important because the magnitude is 00:30:45.697 --> 00:30:46.977 large and these effects. 00:30:46.977 --> 00:30:49.060 But there's no way to actually scale these effects 00:30:49.060 --> 00:30:51.970 in some sense in thinking about what is this priming 00:30:51.970 --> 00:30:54.700 intervention actually correspond to in the real world in terms 00:30:54.700 --> 00:30:57.203 of dollars and how rich or poor people are. 00:30:57.203 --> 00:30:58.870 So that's another sort of reason why you 00:30:58.870 --> 00:31:02.320 might want to do another study. 00:31:02.320 --> 00:31:02.920 OK. 00:31:02.920 --> 00:31:04.570 So what is the other study? 00:31:04.570 --> 00:31:06.240 As Lauren was already saying, it's 00:31:06.240 --> 00:31:09.490 a study with 464 sugarcane farmers 00:31:09.490 --> 00:31:12.460 in 54 villages in Tamil Nadu. 00:31:12.460 --> 00:31:14.080 These are small scale farmers that 00:31:14.080 --> 00:31:19.000 earn at least the majority of 60% of income from sugarcane. 00:31:19.000 --> 00:31:21.940 So these are mostly sugarcane farmers. 00:31:21.940 --> 00:31:25.360 And crucially for the study, the harvest 00:31:25.360 --> 00:31:27.220 is staggered over a three to five 00:31:27.220 --> 00:31:30.280 month period set by the sugar mills due to capacity 00:31:30.280 --> 00:31:31.250 constraints. 00:31:31.250 --> 00:31:34.742 So it's not some people harvest earlier than others 00:31:34.742 --> 00:31:36.700 because they decide to do so maybe because they 00:31:36.700 --> 00:31:38.020 need more cash. 00:31:38.020 --> 00:31:40.240 But it's rather because of capacity constraints. 00:31:40.240 --> 00:31:45.850 The sugar mill is essentially deciding who to harvests early 00:31:45.850 --> 00:31:47.440 and who harvests late. 00:31:47.440 --> 00:31:49.210 And so, now, what this study explored 00:31:49.210 --> 00:31:54.190 is that the sugarcane harvest cycle takes about 18 months. 00:31:54.190 --> 00:31:55.870 It's extremely long. 00:31:55.870 --> 00:32:03.220 And as a result, farmers tend to be relatively poor pre-harvest. 00:32:03.220 --> 00:32:07.840 Think of that this is like grad school and undergrad stipends. 00:32:07.840 --> 00:32:09.460 If you have that money for the month 00:32:09.460 --> 00:32:12.580 or the semester or the like, the beginning of the semester 00:32:12.580 --> 00:32:14.040 is always amazing-- 00:32:14.040 --> 00:32:16.093 or month, for that matter, is always amazing. 00:32:16.093 --> 00:32:17.010 You have lots of cash. 00:32:17.010 --> 00:32:19.270 And you can do lots of nice things. 00:32:19.270 --> 00:32:22.990 And then towards the end of the month or semester, 00:32:22.990 --> 00:32:26.290 people tend to run out of cash, at least 00:32:26.290 --> 00:32:30.920 when I was a grad student or student in general. 00:32:30.920 --> 00:32:33.740 Similar things are true for food stamps 00:32:33.740 --> 00:32:35.158 or other sort of cash programs. 00:32:35.158 --> 00:32:36.950 There's quite a bit of evidence that shows, 00:32:36.950 --> 00:32:38.492 at the beginning of the month, people 00:32:38.492 --> 00:32:41.060 are doing a lot better than at the end, which 00:32:41.060 --> 00:32:44.450 seems to suggest that people have trouble smoothing 00:32:44.450 --> 00:32:47.960 their income over time, which presumably 00:32:47.960 --> 00:32:51.380 has to do with some form of self-control or other problems. 00:32:51.380 --> 00:32:53.210 But so for the purpose of this study, 00:32:53.210 --> 00:32:55.910 farmers are relatively poor pre-harvest. 00:32:55.910 --> 00:32:58.160 And that now allows the authors to look 00:32:58.160 --> 00:33:00.740 at within person comparisons of people, 00:33:00.740 --> 00:33:03.800 sometimes when they're poor and sometimes when they're rich, 00:33:03.800 --> 00:33:06.470 at looking at the cognitive performance, 00:33:06.470 --> 00:33:09.920 by looking at people pre and post harvest. 00:33:09.920 --> 00:33:12.775 Now, first, you want to kind of see some evidence 00:33:12.775 --> 00:33:14.650 that farmers' financial situation is actually 00:33:14.650 --> 00:33:16.180 worse pre-harvest. 00:33:16.180 --> 00:33:18.280 And so there's quite a few-- 00:33:18.280 --> 00:33:21.220 and you can focus on the first column here-- 00:33:21.220 --> 00:33:22.600 pieces of evidence. 00:33:22.600 --> 00:33:25.920 People are more likely to have their belongings pawned. 00:33:25.920 --> 00:33:29.350 They're more likely to have loans outstanding. 00:33:29.350 --> 00:33:32.020 They also have more loans outstanding. 00:33:32.020 --> 00:33:37.120 They're also more likely to say yes to the question or higher 00:33:37.120 --> 00:33:40.750 scores to the question about their ability 00:33:40.750 --> 00:33:49.120 to cope with ordinary bills in the past 15 days post-harvest. 00:33:49.120 --> 00:33:51.580 And so all of these things sort of are saying, essentially, 00:33:51.580 --> 00:33:55.642 before harvest, farmers are in a financially tricky situation. 00:33:55.642 --> 00:33:57.850 And after harvest, that gets better because they get, 00:33:57.850 --> 00:33:59.950 essentially, a bunch of cash from this harvest. 00:33:59.950 --> 00:34:01.600 Because, essentially, it's a cash crop. 00:34:01.600 --> 00:34:04.017 You sell your sugarcane, and then you get a bunch of money 00:34:04.017 --> 00:34:07.170 that you can use over the next 18 months. 00:34:07.170 --> 00:34:11.190 Now, then the authors find, again, using Raven's matrices 00:34:11.190 --> 00:34:13.050 and some measures of cognitive control, 00:34:13.050 --> 00:34:16.409 that people are doing a lot worse pre-harvest compared 00:34:16.409 --> 00:34:17.550 to post-harvest. 00:34:17.550 --> 00:34:20.580 In terms of Raven's matrices, this is accurate. 00:34:20.580 --> 00:34:22.560 Essentially, people do better than getting 00:34:22.560 --> 00:34:28.050 more Raven's matrices correct compared to the control group. 00:34:28.050 --> 00:34:31.380 They also are doing better in terms of cognitive control. 00:34:31.380 --> 00:34:33.460 Post-harvest response times go down, 00:34:33.460 --> 00:34:34.770 which is a good thing here. 00:34:34.770 --> 00:34:38.350 And post-harvest, also, people's errors 00:34:38.350 --> 00:34:39.550 go down in these measures. 00:34:39.550 --> 00:34:41.480 And there's a lot of detail related to that. 00:34:41.480 --> 00:34:45.600 But in some sense, that doesn't matter much for you right now. 00:34:45.600 --> 00:34:48.120 People, essentially, are doing better post-harvest 00:34:48.120 --> 00:34:51.150 when they have more cash. 00:34:51.150 --> 00:34:54.770 Now, what confounds can you think about? 00:34:54.770 --> 00:34:57.060 What are potential problems with this result? 00:35:01.060 --> 00:35:04.250 Actually, this is consistent with poverty 00:35:04.250 --> 00:35:05.838 impeding cognitive function, but there 00:35:05.838 --> 00:35:07.130 could be other things going on. 00:35:07.130 --> 00:35:09.710 And what are these other things? 00:35:09.710 --> 00:35:12.140 I think there's some good evidence-- 00:35:12.140 --> 00:35:15.380 or one good thing is that it would 00:35:15.380 --> 00:35:18.410 have to be a very specific shock that has to do with harvest. 00:35:18.410 --> 00:35:21.080 Because, essentially, the harvest period 00:35:21.080 --> 00:35:23.250 was staggered over several months. 00:35:23.250 --> 00:35:25.850 So it's not like, at some point, the weather 00:35:25.850 --> 00:35:27.410 got worse or the weather got better. 00:35:27.410 --> 00:35:29.810 Because some people actually harvest earlier than others, 00:35:29.810 --> 00:35:31.260 and there's a pre-post comparison. 00:35:31.260 --> 00:35:33.860 But there could be other stuff, time trends or something, 00:35:33.860 --> 00:35:36.630 going on at least in some ways. 00:35:36.630 --> 00:35:39.892 In a way, I think a confound or, in a way, 00:35:39.892 --> 00:35:41.600 it's also just a different interpretation 00:35:41.600 --> 00:35:44.090 perhaps in some ways in the sense of saying, 00:35:44.090 --> 00:35:47.550 well, it's not just worries about the level of income. 00:35:47.550 --> 00:35:51.090 But it's also worries about, essentially, risk or anxiety, 00:35:51.090 --> 00:35:51.590 essentially. 00:35:51.590 --> 00:35:52.965 And so that's slightly different, 00:35:52.965 --> 00:35:55.190 but the anxiety could also sort of capture 00:35:55.190 --> 00:35:58.440 people's minds in some ways. 00:35:58.440 --> 00:36:00.440 Of course, then you would want to see something, 00:36:00.440 --> 00:36:02.390 like maybe that's worse for the rich 00:36:02.390 --> 00:36:04.620 compared to the poor or things like that. 00:36:04.620 --> 00:36:05.840 But that's exactly right. 00:36:05.840 --> 00:36:08.690 So there could be essentially labor costs 00:36:08.690 --> 00:36:10.700 or any sort of other distractions related 00:36:10.700 --> 00:36:14.628 to the harvest, which-- 00:36:14.628 --> 00:36:16.920 this is kind of related to what [INAUDIBLE] was saying, 00:36:16.920 --> 00:36:18.740 which is the harvest itself might 00:36:18.740 --> 00:36:22.438 have these kinds of effects. 00:36:22.438 --> 00:36:24.230 OK, I think we mentioned most of the things 00:36:24.230 --> 00:36:27.440 that I have on my list. 00:36:27.440 --> 00:36:29.750 So there's first one is calorie consumption, which 00:36:29.750 --> 00:36:32.630 is sort of the typical thing that development economists can 00:36:32.630 --> 00:36:36.860 study a lot, which is to say, poor nutrition 00:36:36.860 --> 00:36:39.200 might make people do worse. 00:36:39.200 --> 00:36:40.390 It's bad for attention. 00:36:40.390 --> 00:36:43.860 We talked a little bit about shopping on a hungry stomach, 00:36:43.860 --> 00:36:49.020 but also think about doing tests or exams on an empty stomach 00:36:49.020 --> 00:36:50.760 is probably not a good idea. 00:36:50.760 --> 00:36:54.740 The author's have some evidence that the food expenditures are 00:36:54.740 --> 00:36:59.240 pretty similar pre and post high harvest from some other sample. 00:36:59.240 --> 00:37:01.430 Overall, malnutrition doesn't seem 00:37:01.430 --> 00:37:05.240 to be a huge issue in this area. 00:37:05.240 --> 00:37:06.950 Having said that, I think it's also 00:37:06.950 --> 00:37:10.255 reasonable to think that pre and post harvest you should have, 00:37:10.255 --> 00:37:11.130 actually, more money. 00:37:11.130 --> 00:37:13.297 And then you've got to spend the money on something. 00:37:13.297 --> 00:37:15.260 And presumably, some of that would go on food. 00:37:15.260 --> 00:37:19.070 So the authors argue that nutrition is not a big issue. 00:37:19.070 --> 00:37:21.500 Then there's sort of two broad issues, which 00:37:21.500 --> 00:37:25.580 I think was Joseph and Katherine's comments, which 00:37:25.580 --> 00:37:27.800 is about the uncertainty of the harvest size 00:37:27.800 --> 00:37:30.920 and some issues related to labor effort. 00:37:30.920 --> 00:37:34.490 And so, here, the authors have some evidence 00:37:34.490 --> 00:37:37.537 from post-harvest pre-payment sample. 00:37:37.537 --> 00:37:38.370 What does that mean? 00:37:38.370 --> 00:37:40.670 This is a sample that these are people 00:37:40.670 --> 00:37:48.350 who have harvested already, so they have already 00:37:48.350 --> 00:37:49.680 exerted the labor effort. 00:37:49.680 --> 00:37:55.190 And they also know already their approximate payment. 00:37:55.190 --> 00:37:59.000 And-- sorry. 00:37:59.000 --> 00:38:00.740 I was getting confused by this. 00:38:00.740 --> 00:38:04.370 So these are people who have already done their harvest. 00:38:04.370 --> 00:38:06.110 They have done all the work already. 00:38:06.110 --> 00:38:09.050 They know already what the harvest size will be like. 00:38:09.050 --> 00:38:12.260 But those people, essentially, are still 00:38:12.260 --> 00:38:14.190 not doing better compared to the people who 00:38:14.190 --> 00:38:15.710 have received their cash. 00:38:15.710 --> 00:38:18.530 So that essentially is to say, even 00:38:18.530 --> 00:38:22.170 once you have done all the effort and so on, 00:38:22.170 --> 00:38:27.020 that doesn't seem to help in a sense of you're fully doing, 00:38:27.020 --> 00:38:28.890 you're not exhausted from the work already. 00:38:28.890 --> 00:38:30.557 You already have all your harvest there. 00:38:30.557 --> 00:38:32.280 You already know how big your harvest is, 00:38:32.280 --> 00:38:35.723 but you haven't really received the money yet. 00:38:35.723 --> 00:38:37.640 And you know already what the harvest size is. 00:38:37.640 --> 00:38:40.067 It can't really be about anxiety of the harvest size. 00:38:40.067 --> 00:38:42.150 Yet if you compare those people to the people that 00:38:42.150 --> 00:38:45.000 received the money, you saw few differences 00:38:45.000 --> 00:38:48.960 between those people and the people 00:38:48.960 --> 00:38:51.150 who have done already the harvest. 00:38:51.150 --> 00:38:56.220 And so that argues that essentially it's 00:38:56.220 --> 00:39:00.390 not about anxiety, per se, about the harvest size, 00:39:00.390 --> 00:39:02.730 but rather about just having money 00:39:02.730 --> 00:39:05.640 in your pocket versus not. 00:39:05.640 --> 00:39:08.130 Let me see what the-- 00:39:08.130 --> 00:39:09.994 is this the correct graph? 00:39:15.778 --> 00:39:16.750 I see. 00:39:16.750 --> 00:39:18.940 So this is a little bit misleadingly labeled. 00:39:18.940 --> 00:39:21.040 What this is supposed to say is this 00:39:21.040 --> 00:39:23.020 is post-harvest and pre-harvest. 00:39:23.020 --> 00:39:25.390 But what they mean by post-harvest and pre-harvest, 00:39:25.390 --> 00:39:27.640 post-harvest is meant by you have 00:39:27.640 --> 00:39:29.630 received the cash versus not. 00:39:29.630 --> 00:39:32.440 But both of these are actually both post-harvest in the sense 00:39:32.440 --> 00:39:36.850 that they have done already the work. 00:39:36.850 --> 00:39:38.440 Anyway, sorry, that's a bit confusing. 00:39:38.440 --> 00:39:41.403 But essentially, what they're trying to say 00:39:41.403 --> 00:39:42.820 is essentially what really matters 00:39:42.820 --> 00:39:46.930 is not the work around the harvest, per se, 00:39:46.930 --> 00:39:48.610 or the uncertainty around the harvest. 00:39:48.610 --> 00:39:52.540 It's really about receiving the cash versus not. 00:39:52.540 --> 00:39:56.180 And then there's another problem, 00:39:56.180 --> 00:39:58.060 which you haven't really brought up 00:39:58.060 --> 00:40:00.950 which is the issue of learning effects, right? 00:40:00.950 --> 00:40:03.990 When people do the task several times, 00:40:03.990 --> 00:40:05.490 presumably when you do a test twice, 00:40:05.490 --> 00:40:08.350 you're going to do better. 00:40:08.350 --> 00:40:11.470 But to address this issue, they also have what's 00:40:11.470 --> 00:40:13.000 called a hold-out sample that takes 00:40:13.000 --> 00:40:16.040 the cognitive tests for the first time post-harvest. 00:40:16.040 --> 00:40:19.450 So what I showed you before was proposed for both people 00:40:19.450 --> 00:40:20.800 that have both of these tests. 00:40:20.800 --> 00:40:23.080 There are some samples that's only done 00:40:23.080 --> 00:40:25.360 the task once post-harvest. 00:40:25.360 --> 00:40:27.370 And they don't perform worse than people 00:40:27.370 --> 00:40:30.460 who do the task for the second time, which seems to say 00:40:30.460 --> 00:40:32.218 it's not really about learning. 00:40:32.218 --> 00:40:33.760 Having said that, that is essentially 00:40:33.760 --> 00:40:36.220 only a small subsample. 00:40:36.220 --> 00:40:38.980 Comparison really seems mostly underpowered. 00:40:38.980 --> 00:40:42.176 And there are some concerns, surely, about still effects. 00:40:44.930 --> 00:40:46.130 Any questions on these? 00:40:59.630 --> 00:41:00.130 OK. 00:41:06.180 --> 00:41:08.100 So this is a study from a few years ago. 00:41:08.100 --> 00:41:13.170 So then what kind of follow up is there that sort 00:41:13.170 --> 00:41:14.670 of considers further evidence? 00:41:14.670 --> 00:41:17.405 So there's one very nice study by Carvalho et al. 00:41:17.405 --> 00:41:21.390 in 2015 that looks at the same issue in the US. 00:41:21.390 --> 00:41:24.750 And they looked at cognition and decision making of US 00:41:24.750 --> 00:41:27.870 households around paydays. 00:41:27.870 --> 00:41:30.540 And essentially, they find no cognitive effects whatsoever. 00:41:30.540 --> 00:41:32.940 They also find no effects of decision 00:41:32.940 --> 00:41:36.070 making as one would have expected. 00:41:36.070 --> 00:41:39.870 It raises the question about what's going on here. 00:41:39.870 --> 00:41:41.380 And [INAUDIBLE] what's going on. 00:41:41.380 --> 00:41:43.380 So it could be that this is an external validity 00:41:43.380 --> 00:41:44.280 issue in some sense. 00:41:44.280 --> 00:41:46.860 In a sense, really what is needed here 00:41:46.860 --> 00:41:48.840 is really extreme poverty and really 00:41:48.840 --> 00:41:52.500 large differences in people's income or the amount of money 00:41:52.500 --> 00:41:53.810 that they have available. 00:41:53.810 --> 00:41:56.670 And when you look at US households, at least 00:41:56.670 --> 00:41:59.130 the households that were considered in the Carvalho 00:41:59.130 --> 00:42:00.690 study about payday, well, there's 00:42:00.690 --> 00:42:02.520 differences in how much money they have, 00:42:02.520 --> 00:42:03.570 cash on hand they have. 00:42:03.570 --> 00:42:06.080 The difference is actually not that huge. 00:42:06.080 --> 00:42:09.300 And so that might just be not sufficiently [INAUDIBLE] 00:42:09.300 --> 00:42:11.400 around those paydays to be able to generate 00:42:11.400 --> 00:42:13.560 those type of effects. 00:42:13.560 --> 00:42:17.830 It could also be that these effects don't really 00:42:17.830 --> 00:42:22.910 replicate in a sense of it's a replicability issue, which 00:42:22.910 --> 00:42:24.888 could be a concern as well. 00:42:24.888 --> 00:42:26.930 There's a nice study-- and this gets to the issue 00:42:26.930 --> 00:42:30.020 that I think Joseph mentioned, which 00:42:30.020 --> 00:42:35.570 is a study by Lichand and Mani looking at drought insurance. 00:42:35.570 --> 00:42:39.440 And what they essentially argue is that, 00:42:39.440 --> 00:42:41.660 when you [? provide ?] [? people with ?] insurance-- 00:42:41.660 --> 00:42:45.200 so usually insurance, or people would argue, is useful. 00:42:45.200 --> 00:42:47.582 If a bad state happens in the world, 00:42:47.582 --> 00:42:49.790 you can essentially get-- might get a payment, right? 00:42:49.790 --> 00:42:52.880 If I get some health shock or if my harvest is bad or the like, 00:42:52.880 --> 00:42:54.940 I get some money from the insurance company. 00:42:54.940 --> 00:42:56.210 And that is good. 00:42:56.210 --> 00:42:58.730 Because in the bad state, the marginal utility 00:42:58.730 --> 00:43:00.740 is really high of receiving money. 00:43:00.740 --> 00:43:04.280 And so if I am willing to pay for insurance right now, 00:43:04.280 --> 00:43:08.660 in the future receiving-- so I'm willing to pay for insurance 00:43:08.660 --> 00:43:09.160 now. 00:43:09.160 --> 00:43:12.070 Because in the future, in a bad state of the world, 00:43:12.070 --> 00:43:15.400 I value that money more than in a good state of the world. 00:43:15.400 --> 00:43:19.260 So in a good state [INAUDIBLE] willing to pay for insurance. 00:43:19.260 --> 00:43:21.810 Now, what Lichand and Mani are instead arguing is, 00:43:21.810 --> 00:43:23.400 well, there's an additional value 00:43:23.400 --> 00:43:24.900 of insurance, which essentially it's 00:43:24.900 --> 00:43:29.160 helping reduce their anxiety and worries about money ex ante 00:43:29.160 --> 00:43:30.330 in all states of the world. 00:43:30.330 --> 00:43:33.670 That is to say, if I'm really worried about bad harvest 00:43:33.670 --> 00:43:35.910 shocks or other stuff, health stocks happening 00:43:35.910 --> 00:43:38.503 in the future, that might really frazzle me and really 00:43:38.503 --> 00:43:40.920 sort of worry you all the time, which in turn might affect 00:43:40.920 --> 00:43:42.960 people's cognitive performance. 00:43:42.960 --> 00:43:44.790 Well, that's in all states of the world. 00:43:44.790 --> 00:43:46.860 Even when you're doing well right now, 00:43:46.860 --> 00:43:48.480 even if nothing bad ever happens, 00:43:48.480 --> 00:43:50.925 you're going to do worse today because you're 00:43:50.925 --> 00:43:52.050 worried about these things. 00:43:52.050 --> 00:43:54.450 And sort of providing insurance, therefore, 00:43:54.450 --> 00:43:57.240 might help people better even in good states of the world 00:43:57.240 --> 00:43:59.070 or in all states of the world in addition 00:43:59.070 --> 00:44:01.800 to sort of having the insurance value of helping people 00:44:01.800 --> 00:44:04.200 in that state of the world. 00:44:04.200 --> 00:44:06.228 Well, they have some evidence of it. 00:44:06.228 --> 00:44:07.770 There is some mixed evidence overall. 00:44:07.770 --> 00:44:09.900 But I think the idea of insurance 00:44:09.900 --> 00:44:13.290 being helpful for the poor, for anybody, not just 00:44:13.290 --> 00:44:14.910 because it helps in bad states, but it 00:44:14.910 --> 00:44:16.830 helps people worry about bad states, 00:44:16.830 --> 00:44:18.390 is a very nice one that hopefully 00:44:18.390 --> 00:44:21.080 will be explored further. 00:44:21.080 --> 00:44:25.038 Now, then another question that I think Jose was bringing up 00:44:25.038 --> 00:44:27.080 is to say, well, there are these cognitive tasks. 00:44:27.080 --> 00:44:28.580 Well, I showed you so far that there 00:44:28.580 --> 00:44:31.280 are these tasks and Raven's matrices and cognitive control. 00:44:31.280 --> 00:44:34.490 And sort of, arguably, IQ tests are important then and helpful. 00:44:34.490 --> 00:44:37.850 And we think that's helpful and maps to something 00:44:37.850 --> 00:44:39.182 useful for the world. 00:44:39.182 --> 00:44:40.640 But it's a little bit hard to tell. 00:44:40.640 --> 00:44:42.050 Does this actually matter for income? 00:44:42.050 --> 00:44:43.820 Is this really important or not important? 00:44:43.820 --> 00:44:46.650 And how much does it-- 00:44:46.650 --> 00:44:52.970 so in a paper with [INAUDIBLE] we 00:44:52.970 --> 00:44:54.500 looked at this in a field setting 00:44:54.500 --> 00:45:00.230 where the workers in Orissa and rural areas in India. 00:45:00.230 --> 00:45:03.140 And what we did there is we hired people 00:45:03.140 --> 00:45:07.910 for about two weeks to do a simple labor cast where they 00:45:07.910 --> 00:45:13.010 were doing leaf plates, so essentially putting together 00:45:13.010 --> 00:45:18.320 leafs into plates, which are sort of used in many settings, 00:45:18.320 --> 00:45:22.160 by stitching them together using little wooden sticks. 00:45:22.160 --> 00:45:24.770 And so people are hired for two weeks. 00:45:24.770 --> 00:45:28.280 And then for the sample, we varied. 00:45:28.280 --> 00:45:31.580 Some people were paid earlier, and some people 00:45:31.580 --> 00:45:32.760 were paid later. 00:45:32.760 --> 00:45:34.940 So everybody was paid exactly the same, 00:45:34.940 --> 00:45:38.120 conditional on showing up and conditional on how much they 00:45:38.120 --> 00:45:38.630 produced. 00:45:38.630 --> 00:45:41.172 So their piece rates, how much they were paid, were the same. 00:45:41.172 --> 00:45:43.552 And their attendance paid for showing up. 00:45:43.552 --> 00:45:46.010 The only difference was that some people were paid earlier, 00:45:46.010 --> 00:45:49.290 and some people were paid later. 00:45:49.290 --> 00:45:54.090 So that allows us then look at how do people, the same people, 00:45:54.090 --> 00:45:56.510 how would they perform when they're paid already 00:45:56.510 --> 00:46:00.380 versus not, compared to another group that is essentially 00:46:00.380 --> 00:46:02.340 only paid later. 00:46:02.340 --> 00:46:04.160 And this is a period in the lean period 00:46:04.160 --> 00:46:06.710 when people are extremely financially constrained. 00:46:06.710 --> 00:46:08.450 So people are really short on cash. 00:46:08.450 --> 00:46:10.640 And you can look, essentially, really 00:46:10.640 --> 00:46:14.650 at the impact of transfers or cash 00:46:14.650 --> 00:46:16.890 constraints in this setting. 00:46:16.890 --> 00:46:19.100 What we find is a relatively clear evidence 00:46:19.100 --> 00:46:23.510 that paying people early makes people a lot more productive. 00:46:23.510 --> 00:46:25.400 And moreover, these effects are the largest 00:46:25.400 --> 00:46:26.600 for the poorest workers. 00:46:26.600 --> 00:46:29.060 So essentially, when you sort of divide the sample in two, 00:46:29.060 --> 00:46:31.940 along the people who are richer and poorer to start with, 00:46:31.940 --> 00:46:35.240 people who are particularly poor have the largest impacts 00:46:35.240 --> 00:46:36.148 in the sample. 00:46:36.148 --> 00:46:37.940 And this have sort of pretty large effects. 00:46:37.940 --> 00:46:40.800 It's about 5% to 10% of the productivity, 00:46:40.800 --> 00:46:44.030 which is doesn't sound that large in absolute terms. 00:46:44.030 --> 00:46:46.070 But in relative terms or in terms 00:46:46.070 --> 00:46:48.200 of when you double people's wages, 00:46:48.200 --> 00:46:53.940 people's earnings go up only as much or not much more. 00:46:53.940 --> 00:46:55.820 So really increasing people's productivity 00:46:55.820 --> 00:46:57.800 is very, very hard in these settings. 00:46:57.800 --> 00:47:00.110 And relatively minor interventions, 00:47:00.110 --> 00:47:03.740 such as the one that we did, had pretty large effects 00:47:03.740 --> 00:47:07.250 in the real world setting for people for whom this is really 00:47:07.250 --> 00:47:11.340 their real jobs in their real life for about two weeks. 00:47:11.340 --> 00:47:14.190 So suppose you hired somebody for a year. 00:47:14.190 --> 00:47:19.530 Should you pay them daily, weekly, monthly, or what? 00:47:19.530 --> 00:47:22.220 And so what I'm arguing or what our evidence shows, 00:47:22.220 --> 00:47:25.370 that when you hire people for two weeks, paying people twice, 00:47:25.370 --> 00:47:28.460 as in paying people once after a week and then the rest 00:47:28.460 --> 00:47:31.010 at the end, makes workers more productive 00:47:31.010 --> 00:47:34.308 than paying workers only at the very end. 00:47:34.308 --> 00:47:36.350 And assuming, you know, depending on the interest 00:47:36.350 --> 00:47:40.680 rate-- but that seems to be a thing the company should do. 00:47:40.680 --> 00:47:43.440 But of course, if you hire people for a year, 00:47:43.440 --> 00:47:48.295 then paying people weekly versus monthly or biweekly, 00:47:48.295 --> 00:47:49.920 the comparison becomes quite different. 00:47:49.920 --> 00:47:51.837 Because essentially, the control group, that's 00:47:51.837 --> 00:47:53.028 paid after two weeks. 00:47:53.028 --> 00:47:54.570 Now, after two weeks, of course, they 00:47:54.570 --> 00:47:57.690 have a lot more money because they haven't spent it 00:47:57.690 --> 00:48:00.810 compared to the treatment group who's paid earlier. 00:48:00.810 --> 00:48:06.160 And I think there, essentially, we can't really speak to that. 00:48:06.160 --> 00:48:11.110 In some sense, our design is not really set up to do this. 00:48:11.110 --> 00:48:15.570 You would have to do at least another week or two 00:48:15.570 --> 00:48:17.760 to measure how is the control group now doing 00:48:17.760 --> 00:48:20.610 compared to the treatment group or do it repeatedly. 00:48:20.610 --> 00:48:23.915 And there might also be some effects on over time, 00:48:23.915 --> 00:48:25.290 repeatedly, some of these effects 00:48:25.290 --> 00:48:27.320 might be going away because workers learn 00:48:27.320 --> 00:48:29.400 and so on and so forth. 00:48:29.400 --> 00:48:33.060 I think of this evidence much more as a proof of concept 00:48:33.060 --> 00:48:38.520 that, A, providing cash or alleviating cash constraints 00:48:38.520 --> 00:48:41.580 among workers, either through some low interest rate 00:48:41.580 --> 00:48:45.660 loans or some early payments or some form of UBI or cash 00:48:45.660 --> 00:48:48.510 transfers, unconditional or conditional cash transfers, 00:48:48.510 --> 00:48:50.910 essentially improving workers financial situation 00:48:50.910 --> 00:48:54.410 in some ways, can be productive for firms. 00:48:54.410 --> 00:48:56.430 Now, how the firm then actually does that 00:48:56.430 --> 00:48:57.690 we don't necessarily know. 00:48:57.690 --> 00:48:59.340 And this is sort of too small, short, 00:48:59.340 --> 00:49:01.308 of an experiment to show that. 00:49:01.308 --> 00:49:03.600 But it's sort of proof of concept that, in a real world 00:49:03.600 --> 00:49:05.830 setting with very high stakes for workers-- 00:49:05.830 --> 00:49:09.790 this is their real earnings-- there seem to be these effects. 00:49:09.790 --> 00:49:11.377 So for firms, it might be, well, do 00:49:11.377 --> 00:49:13.500 they have to, in fact, try to improve 00:49:13.500 --> 00:49:16.770 their financial situation [INAUDIBLE].. 00:49:16.770 --> 00:49:19.890 Or for governments, it might be valuable to think 00:49:19.890 --> 00:49:23.157 about cash transfers as a way to do-- so usually, 00:49:23.157 --> 00:49:24.990 when you think about cash transfers-- so let 00:49:24.990 --> 00:49:27.010 me actually just talk about this for a second. 00:49:27.010 --> 00:49:30.540 And then we can talk about other confounds of the study. 00:49:30.540 --> 00:49:33.450 So usually, when you think about the public debate about cash 00:49:33.450 --> 00:49:36.840 transfers and about sort of welfare programs for the poor, 00:49:36.840 --> 00:49:39.065 often sort of the right-wing response is, 00:49:39.065 --> 00:49:40.690 well, aren't the poor going to be lazy? 00:49:40.690 --> 00:49:41.940 And they're not going to work. 00:49:41.940 --> 00:49:44.698 And then what about effort and so on and so forth? 00:49:44.698 --> 00:49:45.990 And they get used to the money. 00:49:45.990 --> 00:49:48.990 And then don't work hard enough eventually. 00:49:48.990 --> 00:49:51.510 Well, the evidence that we have here in our paper 00:49:51.510 --> 00:49:53.790 seems to say, in fact, the opposite. 00:49:53.790 --> 00:49:56.922 That is to say workers, when you alleviate 00:49:56.922 --> 00:49:59.130 their financial constraints, when you give them cash, 00:49:59.130 --> 00:50:00.570 they become more productive. 00:50:00.570 --> 00:50:03.850 And they work harder than otherwise. 00:50:03.850 --> 00:50:06.150 Granted, we don't have any labor supply response. 00:50:06.150 --> 00:50:08.650 What I'm showing you here is essentially productivity, which 00:50:08.650 --> 00:50:11.760 is how hard a worker's working or how much work a worker is 00:50:11.760 --> 00:50:15.450 producing per hour worked. 00:50:15.450 --> 00:50:19.088 But I think similar things could be true for cash transfers. 00:50:19.088 --> 00:50:20.880 And there's some quite interesting evidence 00:50:20.880 --> 00:50:23.460 on cash transfer programs. 00:50:23.460 --> 00:50:27.390 So both cash transfer and, in particular, 00:50:27.390 --> 00:50:29.070 what's called ultra-poor programs, which 00:50:29.070 --> 00:50:34.380 are sort of these multi-faceted asset programs, 00:50:34.380 --> 00:50:38.250 leads to large and sustained increases in asset, savings, 00:50:38.250 --> 00:50:39.690 consumptions, et cetera. 00:50:39.690 --> 00:50:44.190 That is to say, once you provide people with financial resources 00:50:44.190 --> 00:50:46.690 when they're very poor, not only do they better 00:50:46.690 --> 00:50:47.760 in the short run. 00:50:47.760 --> 00:50:49.800 But in the case of the ultra-poor programs, 00:50:49.800 --> 00:50:52.030 in the long run three to five years later, 00:50:52.030 --> 00:50:55.530 people are doing better in terms of having more assets, 00:50:55.530 --> 00:50:58.876 saving more, consuming more, and so on. 00:50:58.876 --> 00:51:04.300 And that seems to say that there might be an issue going on 00:51:04.300 --> 00:51:06.640 and some form of a poverty trap going on. 00:51:06.640 --> 00:51:08.890 And somewhere of the underlying potential channels 00:51:08.890 --> 00:51:11.510 might be things like scarcity, stress, mental health, and so 00:51:11.510 --> 00:51:12.010 on. 00:51:12.010 --> 00:51:14.635 We're going to talk a little bit about mental health and stress 00:51:14.635 --> 00:51:15.500 in a second. 00:51:15.500 --> 00:51:18.430 But essentially, if there are these types of effects going on 00:51:18.430 --> 00:51:22.810 for the government or any other sort of public policy, 00:51:22.810 --> 00:51:28.090 these kinds of programs, conditional cash transfers, 00:51:28.090 --> 00:51:31.240 in addition to having the advantage of improving people's 00:51:31.240 --> 00:51:37.170 well-being and having the high marginal utility of money when 00:51:37.170 --> 00:51:39.440 they're very poor, it might also, in addition, 00:51:39.440 --> 00:51:42.050 make people more productive as opposed 00:51:42.050 --> 00:51:45.378 to a working less or the like. 00:51:45.378 --> 00:51:46.920 So I think those two things, I think, 00:51:46.920 --> 00:51:51.270 are sort of the takeaways. 00:51:51.270 --> 00:51:53.730 If one wanted to understand optimal pay structures, 00:51:53.730 --> 00:51:56.940 one would have to do longer run types of experiments, 00:51:56.940 --> 00:52:00.280 as Natalie pointed out. 00:52:00.280 --> 00:52:04.080 So essentially, the way we set this up is we 00:52:04.080 --> 00:52:05.730 essentially pay people-- 00:52:05.730 --> 00:52:06.230 sorry. 00:52:06.230 --> 00:52:09.060 This is my Slack making noise-- 00:52:09.060 --> 00:52:14.990 pay people to-- sorry, [INAUDIBLE]---- 00:52:14.990 --> 00:52:16.250 people through the employer. 00:52:16.250 --> 00:52:18.500 Essentially, we said, OK, the employer is paying, now, 00:52:18.500 --> 00:52:20.390 earlier versus later. 00:52:20.390 --> 00:52:25.100 And naturally, then the worker might just [INAUDIBLE] happier 00:52:25.100 --> 00:52:28.350 or doing better financially, but also like the employer better 00:52:28.350 --> 00:52:30.097 and saying, look, you just paid me early. 00:52:30.097 --> 00:52:31.680 I now trust you more that you're going 00:52:31.680 --> 00:52:32.940 to be paying me in the future. 00:52:32.940 --> 00:52:34.470 Maybe that's some form of gift giving 00:52:34.470 --> 00:52:37.053 or reciprocity, again, sort of happy about things, [INAUDIBLE] 00:52:37.053 --> 00:52:40.240 work harder to make the employer happy. 00:52:40.240 --> 00:52:43.340 There are several things to get at this. 00:52:43.340 --> 00:52:45.490 And the most important part, perhaps, you also 00:52:45.490 --> 00:52:50.500 have that we announced essentially the structure 00:52:50.500 --> 00:52:52.990 a few days before the actual payment is made. 00:52:52.990 --> 00:52:56.080 That is to say, on day five, we announce and say, on the 8th, 00:52:56.080 --> 00:52:58.460 you're going to be paid versus we don't announce it. 00:52:58.460 --> 00:52:59.980 We say you're going to be paid only 00:52:59.980 --> 00:53:02.330 later at the end of the study. 00:53:02.330 --> 00:53:08.100 And so if you think this is about reciprocity or feelings 00:53:08.100 --> 00:53:11.550 towards the employer, you would think that then workers 00:53:11.550 --> 00:53:13.440 should then already be working harder 00:53:13.440 --> 00:53:16.560 once they hear about when they get these kind of announcement 00:53:16.560 --> 00:53:17.413 effects. 00:53:17.413 --> 00:53:19.080 And it seems to be the case that there's 00:53:19.080 --> 00:53:21.456 essentially no announcement effect whatsoever. 00:53:21.456 --> 00:53:26.190 Workers only become productive pretty much precisely 00:53:26.190 --> 00:53:29.110 once they receive their cash on hand, 00:53:29.110 --> 00:53:31.050 which is very much consistent with what 00:53:31.050 --> 00:53:35.010 I showed you earlier, the results from Mani et al., 00:53:35.010 --> 00:53:35.940 in their study. 00:53:35.940 --> 00:53:39.450 It seems to really be the case that knowing that somebody will 00:53:39.450 --> 00:53:44.040 be paid is really not helping workers or making workers 00:53:44.040 --> 00:53:47.722 more productive both in the form of their relationship 00:53:47.722 --> 00:53:49.680 or due to their relationship with the employer, 00:53:49.680 --> 00:53:59.297 but also due to other effects, including perhaps their worries 00:53:59.297 --> 00:54:01.380 or concerns that they might have about [INAUDIBLE] 00:54:01.380 --> 00:54:05.070 So it seems to be that similarly from the Mani et al. 00:54:05.070 --> 00:54:09.100 paper, when workers have done their harvest 00:54:09.100 --> 00:54:12.060 and you have the harvest in front of you 00:54:12.060 --> 00:54:15.330 and it's worth so many rupees, that 00:54:15.330 --> 00:54:17.977 doesn't seem to increase their cognitive scores. 00:54:17.977 --> 00:54:19.560 That's what I was showing you earlier. 00:54:19.560 --> 00:54:22.350 Really what it seems to be is you have to be actually paid. 00:54:22.350 --> 00:54:25.240 So once workers actually have the cash in hand, only 00:54:25.240 --> 00:54:27.505 then their cognitive function goes up. 00:54:27.505 --> 00:54:29.130 There seems to be something [INAUDIBLE] 00:54:29.130 --> 00:54:34.455 about this [INAUDIBLE] example I have for you is suppose 00:54:34.455 --> 00:54:35.580 you have to pay your bills. 00:54:35.580 --> 00:54:37.800 Suppose you don't have enough money. 00:54:37.800 --> 00:54:39.400 And then, on the first of the month, 00:54:39.400 --> 00:54:41.400 you're going to be paid your stipend or whatever 00:54:41.400 --> 00:54:42.460 as a student. 00:54:42.460 --> 00:54:46.140 There's something distinctly different from having actually 00:54:46.140 --> 00:54:47.990 been paid the money on your bank account 00:54:47.990 --> 00:54:49.860 than having it in your possession 00:54:49.860 --> 00:54:52.313 that is yours or being sent a check or whatever, 00:54:52.313 --> 00:54:54.480 as opposed to knowing it's going to come in two days 00:54:54.480 --> 00:54:57.030 and you have your outstanding credit card bills 00:54:57.030 --> 00:54:59.300 even if you know that you can pay them. 00:54:59.300 --> 00:55:01.050 That's, in some sense, from introspection, 00:55:01.050 --> 00:55:03.427 the best example I have for you. 00:55:03.427 --> 00:55:05.010 But it seems to be-- and you're right. 00:55:05.010 --> 00:55:06.000 In some sense, there could have been 00:55:06.000 --> 00:55:08.000 some effect of just telling workers you're going 00:55:08.000 --> 00:55:09.742 to be paid early versus later. 00:55:09.742 --> 00:55:11.200 But it doesn't seem to be the case. 00:55:11.200 --> 00:55:13.590 This really seems to be the case of receiving 00:55:13.590 --> 00:55:15.300 the money, having cash in hand, and then 00:55:15.300 --> 00:55:17.010 being able to pay the money lender 00:55:17.010 --> 00:55:19.410 or pay essentially some person that you owe 00:55:19.410 --> 00:55:21.840 money, some money back, seems to be what's 00:55:21.840 --> 00:55:32.780 quite important for generating those kinds of effects OK 00:55:32.780 --> 00:55:35.393 so let's move on 00:55:35.393 --> 00:55:36.560 I already talked about this. 00:55:36.560 --> 00:55:37.935 Essentially, yeah, so there seems 00:55:37.935 --> 00:55:40.910 to be these large effects of ultra 00:55:40.910 --> 00:55:43.160 poor programs, which seem to say that there 00:55:43.160 --> 00:55:46.790 might be some underlying behavioral or other poverty 00:55:46.790 --> 00:55:48.045 trap going on. 00:55:48.045 --> 00:55:49.920 And sort of we don't have the answer to that, 00:55:49.920 --> 00:55:53.480 but one hypothesis is that some of the underlying psychological 00:55:53.480 --> 00:55:58.150 issues might be responsible for that. 00:55:58.150 --> 00:56:01.840 Now, what are other poverty induced deprivation? 00:56:01.840 --> 00:56:04.840 So poverty entails a bunch of different deprivations 00:56:04.840 --> 00:56:07.540 beyond money, including malnutrition, high levels 00:56:07.540 --> 00:56:10.040 of stress, sleep deprivation. 00:56:10.040 --> 00:56:12.620 People's sleeping conditions tend to be terrible. 00:56:12.620 --> 00:56:16.790 Noise pollution, the poor are more disproportionately exposed 00:56:16.790 --> 00:56:17.750 to noise. 00:56:17.750 --> 00:56:22.640 There's environmental pollution, in particular air, 00:56:22.640 --> 00:56:25.100 that the poor are much more exposed to. 00:56:25.100 --> 00:56:27.470 Heat, the poor often don't have access 00:56:27.470 --> 00:56:30.200 to heat and/or temperature more generally. 00:56:30.200 --> 00:56:34.440 The poor are more exposed to high temperature 00:56:34.440 --> 00:56:37.440 and don't have access to ACs often. 00:56:37.440 --> 00:56:40.330 There is also some form of stigma and exclusion, 00:56:40.330 --> 00:56:43.140 which might have certain cognitive and other effects. 00:56:43.140 --> 00:56:47.900 There's some evidence of disproportionate substance 00:56:47.900 --> 00:56:48.400 abuse. 00:56:48.400 --> 00:56:54.930 I've shown you some evidence of increased excessive drinking. 00:56:54.930 --> 00:56:57.900 And there's also quite a bit of evidence 00:56:57.900 --> 00:57:03.890 of increased mental illness, especially among the poor. 00:57:03.890 --> 00:57:06.530 So within any given location, the poor 00:57:06.530 --> 00:57:09.630 are disproportionately affected by mental illness, 00:57:09.630 --> 00:57:12.750 including things like depression and anxiety. 00:57:12.750 --> 00:57:15.500 But there's a bunch of research in other settings 00:57:15.500 --> 00:57:21.320 that shows that each of these factors affect either health 00:57:21.320 --> 00:57:23.090 and/or cognitive function. 00:57:23.090 --> 00:57:25.670 And so you might wonder how are these factors then affecting 00:57:25.670 --> 00:57:28.280 people's decision making and choices. 00:57:28.280 --> 00:57:30.170 And just to give you some example of this, 00:57:30.170 --> 00:57:33.000 this is some work that we have done in Chennai. 00:57:33.000 --> 00:57:35.610 So to give some sense of what this looks like, 00:57:35.610 --> 00:57:38.250 the poor essentially tend to have very, very challenging 00:57:38.250 --> 00:57:39.600 sleeping conditions. 00:57:39.600 --> 00:57:41.670 There's dust, noise, mosquitoes. 00:57:41.670 --> 00:57:42.930 People sleep in the same room. 00:57:42.930 --> 00:57:45.540 People sleep outside in the middle of the street. 00:57:45.540 --> 00:57:47.430 And it's hot. 00:57:47.430 --> 00:57:48.990 There's people sleeping on the floor. 00:57:48.990 --> 00:57:51.750 So essentially, any challenging sleeping conditions 00:57:51.750 --> 00:57:54.790 that you can think of you can find it among the urban poor. 00:57:54.790 --> 00:57:56.950 And so one question you might ask is, 00:57:56.950 --> 00:58:01.710 well, then does this lack of good sleep affect 00:58:01.710 --> 00:58:06.808 people's capacity to work and their decision making? 00:58:06.808 --> 00:58:08.350 And so in our study, what we do is we 00:58:08.350 --> 00:58:10.870 objectively measure sleep quantity and quality 00:58:10.870 --> 00:58:11.820 in Chennai. 00:58:11.820 --> 00:58:13.960 And we document severe sleep deprivation 00:58:13.960 --> 00:58:17.810 and alarmingly low quality of sleep in is setting. 00:58:17.810 --> 00:58:18.970 What do I mean by this? 00:58:18.970 --> 00:58:22.690 People sleep on average about 6.5 hours per night. 00:58:22.690 --> 00:58:25.390 And the quality of this sleep is low, 00:58:25.390 --> 00:58:30.370 which one measurement of that is their fraction of time 00:58:30.370 --> 00:58:34.180 that they sleep along the time that they spend in bed 00:58:34.180 --> 00:58:35.440 is about 70%. 00:58:35.440 --> 00:58:38.020 That's what's referred to as sleep efficiency. 00:58:38.020 --> 00:58:41.540 Sleep efficiency is way higher in rich countries. 00:58:41.540 --> 00:58:43.820 It's about 85% to 90%. 00:58:43.820 --> 00:58:49.560 So if the rich spend an hour in bed, they spend about 55 00:58:49.560 --> 00:58:51.420 or something minutes asleep. 00:58:51.420 --> 00:58:55.570 If the poor do that, it's about 40 something minutes. 00:58:55.570 --> 00:58:57.630 So essentially, there's lots of awakenings. 00:58:57.630 --> 00:58:59.670 And as a consequence, in our sample, 00:58:59.670 --> 00:59:03.900 people seem about 5 and 1/2 hours per night. 00:59:03.900 --> 00:59:06.420 Now, then what we do is we have two interventions 00:59:06.420 --> 00:59:08.550 to increase sleep and provide people 00:59:08.550 --> 00:59:12.780 information, sleep devices, which is mattress, ear 00:59:12.780 --> 00:59:14.310 plugs, eye shields, and so on. 00:59:14.310 --> 00:59:17.580 And we increased sleep by quite a bit for about three weeks. 00:59:17.580 --> 00:59:20.310 And in this experiment, we also have an intervention 00:59:20.310 --> 00:59:22.560 of a nap intervention that offers people naps 00:59:22.560 --> 00:59:28.470 at the office, which provides high quantity and high quality 00:59:28.470 --> 00:59:30.270 of sleep in that setting. 00:59:30.270 --> 00:59:32.970 Now, what we find then is, somewhat surprisingly, 00:59:32.970 --> 00:59:34.980 that actually increasing people's sleep 00:59:34.980 --> 00:59:35.880 does very little. 00:59:35.880 --> 00:59:39.120 So people sleep about 20 to 30 minutes more every night 00:59:39.120 --> 00:59:40.380 for three weeks. 00:59:40.380 --> 00:59:41.760 And we measure the labor supply. 00:59:41.760 --> 00:59:44.870 We hire them for about that time. 00:59:44.870 --> 00:59:48.050 People are not doing better in any way in terms of people 00:59:48.050 --> 00:59:48.950 not working more. 00:59:48.950 --> 00:59:50.370 People are not more productive. 00:59:50.370 --> 00:59:52.620 People are not doing better in cognitive tasks. 00:59:52.620 --> 00:59:54.670 They're not happier and so on and so forth. 00:59:54.670 --> 00:59:57.170 They also don't make different decisions. 00:59:57.170 --> 00:59:59.900 And if anything, you know, they decrease their labor supply. 00:59:59.900 --> 01:00:01.940 They actually work fewer hours. 01:00:01.940 --> 01:00:02.573 Why is that? 01:00:02.573 --> 01:00:04.490 Well, because people are spending so much more 01:00:04.490 --> 01:00:07.940 time in bed, they have now less time available during the day. 01:00:07.940 --> 01:00:12.960 And some of that time goes into working fewer hours. 01:00:12.960 --> 01:00:15.240 Now, in contrast, the nap treatment 01:00:15.240 --> 01:00:17.760 has pretty large effects on the productivity, attention, 01:00:17.760 --> 01:00:19.820 well-being, patience, and so on. 01:00:19.820 --> 01:00:23.100 So it's not sleep cannot affect people's outcomes in these 01:00:23.100 --> 01:00:23.940 settings. 01:00:23.940 --> 01:00:25.935 But rather, a night sleep, as it is 01:00:25.935 --> 01:00:29.280 does not really do very much, at least the type of night's sleep 01:00:29.280 --> 01:00:31.110 that people are getting. 01:00:31.110 --> 01:00:35.490 And so what we then provide some evidence of that what's 01:00:35.490 --> 01:00:37.320 really important here or the reason 01:00:37.320 --> 01:00:39.930 why people's sleep does not do very much 01:00:39.930 --> 01:00:44.530 in increasing their productivity or improving their outcomes 01:00:44.530 --> 01:00:47.470 is because the quality of sleep is so low. 01:00:47.470 --> 01:00:49.360 So let me sort of look at the sample 01:00:49.360 --> 01:00:52.150 and look at people who sleep better and worse at baseline 01:00:52.150 --> 01:00:53.770 at the beginning of the study. 01:00:53.770 --> 01:00:56.020 People who sleep better at the beginning of the study, 01:00:56.020 --> 01:00:58.330 they actually tend to have reasonably large effects 01:00:58.330 --> 01:00:59.740 on their productivity. 01:00:59.740 --> 01:01:02.380 In contrast, people who have bad sleep 01:01:02.380 --> 01:01:04.475 to start with, for them, essentially increasing 01:01:04.475 --> 01:01:08.510 their sleep does nothing and potentially makes things worse. 01:01:08.510 --> 01:01:09.430 So what did we learn? 01:01:09.430 --> 01:01:12.190 I do not want you to take away that sleeping does not 01:01:12.190 --> 01:01:13.110 do anything. 01:01:13.110 --> 01:01:15.670 MIT students are chronically undersleeping. 01:01:15.670 --> 01:01:19.630 And you could be probably sleeping more, all of you, 01:01:19.630 --> 01:01:21.130 or most of you. 01:01:21.130 --> 01:01:23.710 Essentially, what we learn from this 01:01:23.710 --> 01:01:26.950 is that, when people's sleep conditions are really terrible, 01:01:26.950 --> 01:01:32.400 well, then increasing one's sleep actually 01:01:32.400 --> 01:01:33.430 doesn't do very much. 01:01:33.430 --> 01:01:35.952 The return on sleeping is just very, very low. 01:01:35.952 --> 01:01:37.410 And so what we learn, in some ways, 01:01:37.410 --> 01:01:40.650 that the poor, at least in this setting, 01:01:40.650 --> 01:01:43.740 are actually doing as well as they could, in a sense of they 01:01:43.740 --> 01:01:44.907 do the best they can. 01:01:44.907 --> 01:01:47.490 And since sleep is so bad, is so terrible, people sleep really 01:01:47.490 --> 01:01:51.900 badly, increasing their sleep does not do very much. 01:01:51.900 --> 01:01:54.450 That's not to say, if you actually fix their sleeping 01:01:54.450 --> 01:01:56.940 conditions, if you put everybody into a nice hotel 01:01:56.940 --> 01:01:59.760 or a nice bedroom or give them a nice house 01:01:59.760 --> 01:02:01.740 and help them sleep better, that wouldn't 01:02:01.740 --> 01:02:03.810 have potentially large effects. 01:02:03.810 --> 01:02:06.668 But the study that we conducted was not doing that. 01:02:06.668 --> 01:02:08.460 The study that we conducted was essentially 01:02:08.460 --> 01:02:10.710 saying, take your sleep conditions as given. 01:02:10.710 --> 01:02:12.150 And look at what happens when you 01:02:12.150 --> 01:02:16.740 increase people's sleep duration for given sleep quality. 01:02:16.740 --> 01:02:18.810 And that doesn't do very much. 01:02:18.810 --> 01:02:21.420 In contrast, as I showed you or told you, 01:02:21.420 --> 01:02:25.020 naps seem to be quite effective in having a large number 01:02:25.020 --> 01:02:29.310 or range of effects, which seems to say that sleep can 01:02:29.310 --> 01:02:31.060 potentially matter quite a bit. 01:02:31.060 --> 01:02:34.380 But what really is important is try to figure out how could we 01:02:34.380 --> 01:02:36.930 perhaps increase people's sleep quality which might 01:02:36.930 --> 01:02:38.392 have then important effects. 01:02:38.392 --> 01:02:40.350 Sorry, that was a lot of information, more than 01:02:40.350 --> 01:02:42.060 I perhaps intended. 01:02:42.060 --> 01:02:43.110 Any questions on that? 01:02:58.050 --> 01:03:00.660 OK, so then the last piece, which 01:03:00.660 --> 01:03:04.770 I'm going to just sort of say a little bit about, 01:03:04.770 --> 01:03:09.750 is a paper that we just finished on the relationship 01:03:09.750 --> 01:03:13.290 between poverty and common mental disorders 01:03:13.290 --> 01:03:16.920 as they're referred to, which is anxiety and depression. 01:03:16.920 --> 01:03:19.650 And what we there sort of argue-- 01:03:19.650 --> 01:03:22.830 and it's a pretty short paper if you're interested. 01:03:22.830 --> 01:03:27.832 I'll, in fact, put it on my website tonight. 01:03:27.832 --> 01:03:29.790 We're looking at, essentially, the relationship 01:03:29.790 --> 01:03:33.030 between poverty and these two conditions. 01:03:33.030 --> 01:03:35.520 And what we do is, first, we show the causal effect 01:03:35.520 --> 01:03:37.017 in both directions. 01:03:37.017 --> 01:03:38.850 And so there's quite a bit of evidence that, 01:03:38.850 --> 01:03:42.402 essentially, being poor is bad for people's mental health. 01:03:42.402 --> 01:03:43.610 And these are causal studies. 01:03:43.610 --> 01:03:46.290 So for example, providing people cash transfers 01:03:46.290 --> 01:03:49.150 improves consistently their mental health. 01:03:49.150 --> 01:03:51.600 So there's a causal effective from not having money 01:03:51.600 --> 01:03:54.917 on people's mental health and anxiety and depression. 01:03:54.917 --> 01:03:56.625 At the same time, there's a causal effect 01:03:56.625 --> 01:04:01.590 of anxiety and depression on people's ability 01:04:01.590 --> 01:04:05.440 to earn income and, thus, their poverty. 01:04:05.440 --> 01:04:08.010 So for example, if you provide people with CBT 01:04:08.010 --> 01:04:11.760 or some other forms of psychotherapy, 01:04:11.760 --> 01:04:15.670 that reduces their depression. 01:04:15.670 --> 01:04:18.340 And then, in turn, it increases their labor supply, 01:04:18.340 --> 01:04:21.000 which in turn presumably increases their ability 01:04:21.000 --> 01:04:22.570 to earn income. 01:04:22.570 --> 01:04:25.200 So there's causal effects in both directions from poverty 01:04:25.200 --> 01:04:29.220 to common mental disorders and from common mental disorders 01:04:29.220 --> 01:04:30.960 to poverty. 01:04:30.960 --> 01:04:33.600 And then what the paper does in a lot more detail is then 01:04:33.600 --> 01:04:35.490 think about what are different mechanisms 01:04:35.490 --> 01:04:39.390 for this causal relationship and what exactly, what evidence, 01:04:39.390 --> 01:04:42.040 do they actually have of that. 01:04:42.040 --> 01:04:46.110 And so on the side from poverty and from economic conditions 01:04:46.110 --> 01:04:48.270 to mental health, there's a number 01:04:48.270 --> 01:04:51.510 of different issues that could be quite important ranging 01:04:51.510 --> 01:04:54.870 from stresses or just being exposed to lots of shocks 01:04:54.870 --> 01:04:57.630 and, therefore, worries and uncertainty 01:04:57.630 --> 01:04:58.823 to environmental conditions. 01:04:58.823 --> 01:05:00.240 In particular, there's quite a bit 01:05:00.240 --> 01:05:01.830 of emerging evidence of pollution 01:05:01.830 --> 01:05:03.800 being bad for people's mental health, 01:05:03.800 --> 01:05:08.650 so essentially pollution causing depression, for example. 01:05:08.650 --> 01:05:11.250 There's some potential for physical health, 01:05:11.250 --> 01:05:13.635 just being in bad health. 01:05:13.635 --> 01:05:16.350 It might be bad for your mental health in part 01:05:16.350 --> 01:05:21.130 because it causes physical pain and other illnesses. 01:05:21.130 --> 01:05:25.090 There's some evidence of early life conditions 01:05:25.090 --> 01:05:27.070 growing up in poverty tends to be 01:05:27.070 --> 01:05:28.550 bad for people's mental health. 01:05:28.550 --> 01:05:31.900 And there's some nice evidence of shocks 01:05:31.900 --> 01:05:37.630 to mothers' and fathers' income might cause mental illness much 01:05:37.630 --> 01:05:40.090 later in life for the children. 01:05:40.090 --> 01:05:42.970 There's quite a bit of evidence of trauma, violence, and crime, 01:05:42.970 --> 01:05:44.530 which the poor are disproportionately 01:05:44.530 --> 01:05:46.940 exposed to, affecting people's mental health. 01:05:46.940 --> 01:05:49.660 And finally, social status, shame, and isolation, 01:05:49.660 --> 01:05:53.950 if you sort of stigmatize and isolate from society, 01:05:53.950 --> 01:05:55.930 that might be bad for your mental health. 01:05:55.930 --> 01:05:57.910 Particular, sort of relative comparisons 01:05:57.910 --> 01:06:00.020 could be quite important. 01:06:00.020 --> 01:06:02.020 And then at the same time, there are 01:06:02.020 --> 01:06:03.640 effects in the other directions coming 01:06:03.640 --> 01:06:06.800 from depression and anxiety. 01:06:06.800 --> 01:06:10.420 So one profound way in which depression and anxiety affect 01:06:10.420 --> 01:06:14.320 people's behavior is cognitive function, the way people think. 01:06:14.320 --> 01:06:18.490 So in contrast to other things such as health, 01:06:18.490 --> 01:06:21.910 depression and anxiety affect the way people think. 01:06:21.910 --> 01:06:24.020 And then that might affect people's beliefs, 01:06:24.020 --> 01:06:25.810 their confidence, overconfidence, 01:06:25.810 --> 01:06:27.580 underconfidence, belief updating, 01:06:27.580 --> 01:06:30.220 all the stuff that we talked about in the class as well 01:06:30.220 --> 01:06:33.040 as their preferences, and time preferences, risk preferences, 01:06:33.040 --> 01:06:35.150 social preferences, and so on. 01:06:35.150 --> 01:06:38.290 It might affect their labor productivity, their labor 01:06:38.290 --> 01:06:41.260 supply, how much people work, and how productive they are. 01:06:41.260 --> 01:06:44.770 Of course, people might be very stigmatized and treated 01:06:44.770 --> 01:06:51.670 badly because of mental illness, which in addition sort of then 01:06:51.670 --> 01:06:55.030 makes people less likely to seek help potentially. 01:06:55.030 --> 01:06:56.783 Because if you're then stigmatized, 01:06:56.783 --> 01:06:58.700 you might not want to declare to your employer 01:06:58.700 --> 01:07:01.690 or to anybody else that you're depressed. 01:07:01.690 --> 01:07:05.710 People tend to have worse health conditions if they're depressed 01:07:05.710 --> 01:07:09.110 or they suffer from mental illness or other health 01:07:09.110 --> 01:07:12.390 conditions tend to get worse, health expenditures. 01:07:12.390 --> 01:07:14.970 There's also some evidence of depression 01:07:14.970 --> 01:07:17.220 affecting women's empowerment. 01:07:17.220 --> 01:07:20.500 I think there's two potential channels here. 01:07:20.500 --> 01:07:23.670 One is that women are, in fact, disproportionately affected 01:07:23.670 --> 01:07:26.710 by depression, anxiety, and so on. 01:07:26.710 --> 01:07:29.430 So the measured prevalence is about twice 01:07:29.430 --> 01:07:32.790 as high and sort of to the extent that now, 01:07:32.790 --> 01:07:37.160 if you improve people's mental health, 01:07:37.160 --> 01:07:40.053 women will disproportionately benefit from that. 01:07:40.053 --> 01:07:41.720 And then there's evidence that, once you 01:07:41.720 --> 01:07:44.630 do that, women have higher bargaining 01:07:44.630 --> 01:07:50.070 power in the household, more say about resources spent 01:07:50.070 --> 01:07:52.170 and so on and so forth. 01:07:52.170 --> 01:07:54.440 And then finally, potentially quite important 01:07:54.440 --> 01:07:56.480 is effects on young people. 01:07:56.480 --> 01:07:59.900 This ranges from primary and secondary school 01:07:59.900 --> 01:08:03.330 age to college students and so on, 01:08:03.330 --> 01:08:08.180 which is to say that a lot of mental health conditions 01:08:08.180 --> 01:08:09.920 appear during adolescence. 01:08:09.920 --> 01:08:12.800 And so those are often when very critical decisions are made 01:08:12.800 --> 01:08:14.142 when it comes to education. 01:08:14.142 --> 01:08:15.225 Do you drop out of school? 01:08:15.225 --> 01:08:18.229 Do you go to college and so on and so forth? 01:08:18.229 --> 01:08:21.050 It turns out there's actually quite not 01:08:21.050 --> 01:08:22.729 a lot of causal evidence on this, 01:08:22.729 --> 01:08:26.840 but it seems very plausible that this could be quite important. 01:08:26.840 --> 01:08:28.620 And sort of taking these things together, 01:08:28.620 --> 01:08:31.310 it seems to be that poverty, through various different 01:08:31.310 --> 01:08:33.950 channels, including scarcity, which I showed you 01:08:33.950 --> 01:08:35.870 at the beginning, but also including 01:08:35.870 --> 01:08:38.450 all these other conditions might affect profoundly 01:08:38.450 --> 01:08:41.180 the [INAUDIBLE] and how they make decisions 01:08:41.180 --> 01:08:43.490 and how productive they are at work, which then could 01:08:43.490 --> 01:08:46.520 reinforced poverty and make it harder 01:08:46.520 --> 01:08:49.401 for people to become richer. 01:08:49.401 --> 01:08:50.609 This is a lot of information. 01:08:50.609 --> 01:08:52.080 So any questions? 01:08:52.080 --> 01:08:54.859 I'm happy to discuss or answer any questions 01:08:54.859 --> 01:08:55.819 that you would have. 01:08:55.819 --> 01:08:58.430 I'm not expecting you to remember all of those things. 01:08:58.430 --> 01:09:00.350 I was just more elaborating for anybody 01:09:00.350 --> 01:09:01.740 who might be interested. 01:09:01.740 --> 01:09:05.683 You can also read all of those things in the paper 01:09:05.683 --> 01:09:07.800 that I'll add to the course website as well. 01:09:22.310 --> 01:09:25.520 All right, so on Wednesday, we're 01:09:25.520 --> 01:09:28.310 going to talk about happiness and mental health. 01:09:28.310 --> 01:09:30.859 Please read the paper by Kahneman and Deaton. 01:09:30.859 --> 01:09:32.760 Notice I just changed this in the syllabus. 01:09:32.760 --> 01:09:34.203 So if you have an old version of the syllabus, 01:09:34.203 --> 01:09:35.870 please make sure you read the new paper. 01:09:35.870 --> 01:09:39.250 I think it was very short, so it shouldn't be tricky to read. 01:09:39.250 --> 01:09:42.800 Essentially, it's looking at what kinds of things 01:09:42.800 --> 01:09:46.770 predict whether people are happy or not. 01:09:46.770 --> 01:09:50.500 And then on Monday, we're going to talk about policy 01:09:50.500 --> 01:09:53.740 with behavioral agents. 01:09:53.740 --> 01:09:56.380 Just as a reminder, we have a guest 01:09:56.380 --> 01:09:58.370 at the end of class next time. 01:09:58.370 --> 01:10:00.290 I hope the guests will come at 2:20. 01:10:00.290 --> 01:10:04.780 I'm hoping for a lama to visit us, 01:10:04.780 --> 01:10:06.100 but it could also be a goat. 01:10:06.100 --> 01:10:08.590 We'll see how that goes. 01:10:08.590 --> 01:10:10.963 Any questions or comments? 01:10:10.963 --> 01:10:12.630 When it comes to depression and anxiety, 01:10:12.630 --> 01:10:14.922 specifically I'm going to talk about it more generally. 01:10:14.922 --> 01:10:17.560 So one of the things I'm working on, for example, 01:10:17.560 --> 01:10:21.780 is looking at the impact of cognitive behavioral therapy 01:10:21.780 --> 01:10:25.410 or other psychotherapies to improve people's mental health 01:10:25.410 --> 01:10:28.770 and then look at the effects on people's beliefs, how 01:10:28.770 --> 01:10:31.350 confident are they in themselves, 01:10:31.350 --> 01:10:32.820 how they update their beliefs. 01:10:32.820 --> 01:10:36.000 And remember, we have this stuff from [INAUDIBLE] and others 01:10:36.000 --> 01:10:39.270 about asymmetric belief updating and is 01:10:39.270 --> 01:10:42.630 that affected, for example, by people's mental health. 01:10:42.630 --> 01:10:44.190 Moreover, we look at preferences, 01:10:44.190 --> 01:10:45.960 which is precisely time preferences, 01:10:45.960 --> 01:10:48.240 social preferences, risk preferences. 01:10:48.240 --> 01:10:52.110 And once you improve people's depression, 01:10:52.110 --> 01:10:57.240 does that affect people's time social and risk preference? 01:10:57.240 --> 01:11:00.500 More broadly, there's quite a few people trying to do 01:11:00.500 --> 01:11:03.110 studies-- and I'll run through this fairly quickly-- 01:11:03.110 --> 01:11:07.960 to look at these types of effects, which, for example, 01:11:07.960 --> 01:11:09.710 there's some work around malnutrition that 01:11:09.710 --> 01:11:12.980 shows, once you provide people with more calories, 01:11:12.980 --> 01:11:18.510 there's some indications of less discounting, which is kind 01:11:18.510 --> 01:11:22.680 of what you're alluding to. 01:11:22.680 --> 01:11:24.360 There's some work on the relationship 01:11:24.360 --> 01:11:25.655 between stress and discounting. 01:11:25.655 --> 01:11:27.780 There's actually not much evidence or less evidence 01:11:27.780 --> 01:11:29.160 than you would expect. 01:11:29.160 --> 01:11:31.500 For sleep deprivation, for example, we find, 01:11:31.500 --> 01:11:35.058 when people nap, they're less present biased compared 01:11:35.058 --> 01:11:36.225 to when they haven't napped. 01:11:39.530 --> 01:11:41.290 For example, for alcohol consumption, 01:11:41.290 --> 01:11:42.700 I do find that people save more. 01:11:42.700 --> 01:11:44.960 And it seems to be not driven by having more money. 01:11:44.960 --> 01:11:47.440 So again, that seems to be consistent with people's time 01:11:47.440 --> 01:11:50.830 preference in some way affected by these conditions. 01:11:50.830 --> 01:11:52.990 Overall, there's quite a bit of evidence 01:11:52.990 --> 01:11:55.820 so far on productivity. 01:11:55.820 --> 01:11:58.630 So papers essentially show that, when you're disproportionately 01:11:58.630 --> 01:12:02.980 exposed to heat, people are less productive. 01:12:02.980 --> 01:12:03.970 People learn less. 01:12:03.970 --> 01:12:06.590 When you're exposed to noise, people are less productive. 01:12:06.590 --> 01:12:09.370 When people have less nutrition, they're 01:12:09.370 --> 01:12:12.500 less productive and so on and so forth. 01:12:12.500 --> 01:12:16.000 There's less evidence, so far, on decision making and sort 01:12:16.000 --> 01:12:17.410 of preferences overall. 01:12:17.410 --> 01:12:19.377 That's what people are currently working on, 01:12:19.377 --> 01:12:20.710 but that's exactly the question. 01:12:20.710 --> 01:12:23.380 Can we sort of explained potentially some of this stuff 01:12:23.380 --> 01:12:26.000 that I was mentioning at the very beginning of the lecture? 01:12:26.000 --> 01:12:29.080 Can we explain potentially certain behavioral biases 01:12:29.080 --> 01:12:32.350 or certain deviations from the neoclassical model 01:12:32.350 --> 01:12:36.690 by alleviating or addressing some of the psychological 01:12:36.690 --> 01:12:37.510 [INAUDIBLE]? 01:12:37.510 --> 01:12:39.910 We don't have a lot of evidence, but we hope 01:12:39.910 --> 01:12:44.030 to have some more [INAUDIBLE]. 01:12:44.030 --> 01:12:47.050 But that's exactly what people are working on.