WEBVTT 00:00:00.000 --> 00:00:02.420 align:middle line:90% 00:00:02.420 --> 00:00:03.388 align:middle line:90% [SQUEAKING] 00:00:03.388 --> 00:00:04.356 align:middle line:90% [RUSTLING] 00:00:04.356 --> 00:00:05.808 align:middle line:90% [CLICKING] 00:00:05.808 --> 00:00:12.120 align:middle line:90% 00:00:12.120 --> 00:00:15.050 align:middle line:84% ROBERT TOWNSEND: Greetings, everyone. 00:00:15.050 --> 00:00:17.120 align:middle line:84% So before I start sharing my screen, 00:00:17.120 --> 00:00:18.880 align:middle line:84% let me ask if there are any questions 00:00:18.880 --> 00:00:20.110 align:middle line:90% from the last lecture. 00:00:20.110 --> 00:00:23.270 align:middle line:90% 00:00:23.270 --> 00:00:25.420 align:middle line:90% OK. 00:00:25.420 --> 00:00:30.640 align:middle line:84% So, where are we in terms of class? 00:00:30.640 --> 00:00:34.180 align:middle line:90% We have a review session today. 00:00:34.180 --> 00:00:38.050 align:middle line:84% The exam is next Thursday, and it covers material 00:00:38.050 --> 00:00:40.060 align:middle line:90% from lectures 8 through 15. 00:00:40.060 --> 00:00:42.340 align:middle line:84% And I'm going to take questions on those, 00:00:42.340 --> 00:00:45.380 align:middle line:84% or scroll through them, or review, or something. 00:00:45.380 --> 00:00:47.660 align:middle line:84% So that's the main thing that's happening. 00:00:47.660 --> 00:00:52.570 align:middle line:84% So the overview of that is we did the theory, 00:00:52.570 --> 00:00:54.760 align:middle line:84% and then we started in risk-sharing, 00:00:54.760 --> 00:00:58.090 align:middle line:84% with two applications, including production. 00:00:58.090 --> 00:00:59.890 align:middle line:90% We went to financial modeling. 00:00:59.890 --> 00:01:05.200 align:middle line:84% Then we went to contracts to get at obstacles, like information 00:01:05.200 --> 00:01:06.880 align:middle line:90% and moral hazard problems. 00:01:06.880 --> 00:01:13.570 align:middle line:84% We did a theory lecture of sorts, links to Bitcoin, 00:01:13.570 --> 00:01:17.060 align:middle line:84% and ledgers, and then another application lecture. 00:01:17.060 --> 00:01:20.470 align:middle line:84% We went to international trade in the context of all raising 00:01:20.470 --> 00:01:24.580 align:middle line:84% equilibria, and we did two trade lectures, including 00:01:24.580 --> 00:01:26.060 align:middle line:90% the one we did last time. 00:01:26.060 --> 00:01:28.370 align:middle line:90% So that's the overview. 00:01:28.370 --> 00:01:32.800 align:middle line:84% Some words about the exam, again, even though in principle 00:01:32.800 --> 00:01:34.840 align:middle line:84% the thing is comprehensive, that's 00:01:34.840 --> 00:01:38.440 align:middle line:84% because it's building on material that had previously 00:01:38.440 --> 00:01:39.500 align:middle line:90% been covered. 00:01:39.500 --> 00:01:43.090 align:middle line:84% But the focus will be on the material from lectures 8 00:01:43.090 --> 00:01:44.290 align:middle line:90% through 15. 00:01:44.290 --> 00:01:50.290 align:middle line:84% And again, because it's limited to 80 minutes, 00:01:50.290 --> 00:01:53.140 align:middle line:84% there's no way I can ask a question about everything 00:01:53.140 --> 00:01:55.390 align:middle line:90% you might be anticipating. 00:01:55.390 --> 00:01:59.380 align:middle line:84% But hopefully, the questions themselves 00:01:59.380 --> 00:02:03.160 align:middle line:84% are reasonable, and fair, and not terribly surprising 00:02:03.160 --> 00:02:06.070 align:middle line:84% if you've been following the lectures, doing the p-sets, 00:02:06.070 --> 00:02:07.465 align:middle line:90% and keeping up to date. 00:02:07.465 --> 00:02:10.039 align:middle line:90% 00:02:10.039 --> 00:02:15.040 align:middle line:84% And you have all the review session material, as well. 00:02:15.040 --> 00:02:16.960 align:middle line:84% That's the best guide to the lectures, 00:02:16.960 --> 00:02:19.810 align:middle line:84% or there was review questions we do 00:02:19.810 --> 00:02:21.760 align:middle line:90% at the beginning of each class. 00:02:21.760 --> 00:02:23.990 align:middle line:84% Which brings me to the next thing. 00:02:23.990 --> 00:02:27.070 align:middle line:90% So here's the study guide. 00:02:27.070 --> 00:02:34.810 align:middle line:84% And let's start with what we did last time, focusing on the US. 00:02:34.810 --> 00:02:40.930 align:middle line:84% I talked about China shocks and accounting identities, which 00:02:40.930 --> 00:02:44.590 align:middle line:84% led up to this question using the lens of the accounting 00:02:44.590 --> 00:02:45.310 align:middle line:90% identities. 00:02:45.310 --> 00:02:47.980 align:middle line:84% What is the preliminary evidence for the impact 00:02:47.980 --> 00:02:52.110 align:middle line:84% of China shocks on the variables of the current account? 00:02:52.110 --> 00:02:55.240 align:middle line:90% 00:02:55.240 --> 00:02:58.870 align:middle line:84% Somehow, I'm in the mood to take volunteers, rather than call 00:02:58.870 --> 00:03:00.920 align:middle line:90% on you. 00:03:00.920 --> 00:03:02.290 align:middle line:90% Any volunteers for that? 00:03:02.290 --> 00:03:09.600 align:middle line:90% 00:03:09.600 --> 00:03:12.890 align:middle line:90% Quiet group today. 00:03:12.890 --> 00:03:14.600 align:middle line:84% Well, I will-- how about Charles? 00:03:14.600 --> 00:03:19.270 align:middle line:90% 00:03:19.270 --> 00:03:22.612 align:middle line:90% AUDIENCE: Oh, let's see. 00:03:22.612 --> 00:03:25.870 align:middle line:84% I think I haven't got around to watching your last lecture 00:03:25.870 --> 00:03:26.720 align:middle line:90% recording yet. 00:03:26.720 --> 00:03:28.960 align:middle line:90% So I'm not exactly sure. 00:03:28.960 --> 00:03:31.560 align:middle line:90% But oh, yeah. 00:03:31.560 --> 00:03:32.415 align:middle line:90% Sorry. 00:03:32.415 --> 00:03:33.720 align:middle line:90% Is it this one? 00:03:33.720 --> 00:03:35.980 align:middle line:84% ROBERT TOWNSEND: Yeah it's in and out, 00:03:35.980 --> 00:03:37.720 align:middle line:90% synchronous and asynchronous. 00:03:37.720 --> 00:03:40.600 align:middle line:84% How about is someone ready to volunteer? 00:03:40.600 --> 00:03:45.220 align:middle line:84% China shock financial accounts and what did we find? 00:03:45.220 --> 00:03:47.200 align:middle line:90% AUDIENCE: I can try. 00:03:47.200 --> 00:03:48.610 align:middle line:90% I'm also very unsure. 00:03:48.610 --> 00:03:53.740 align:middle line:84% But is this about looking at the current account, 00:03:53.740 --> 00:03:59.905 align:middle line:84% and how the trade balance would be negative. 00:03:59.905 --> 00:04:03.805 align:middle line:84% And then that's kind of giving us evidence about how-- 00:04:03.805 --> 00:04:06.812 align:middle line:90% 00:04:06.812 --> 00:04:13.060 align:middle line:84% the China shock was about how since low-skilled workers, 00:04:13.060 --> 00:04:15.550 align:middle line:84% we have less of that commodity and more 00:04:15.550 --> 00:04:16.589 align:middle line:90% of high-skilled workers. 00:04:16.589 --> 00:04:19.360 align:middle line:84% So then we're like low-skilled workers are 00:04:19.360 --> 00:04:20.949 align:middle line:90% going to be losing in trade. 00:04:20.949 --> 00:04:23.323 align:middle line:90% 00:04:23.323 --> 00:04:24.740 align:middle line:84% ROBERT TOWNSEND: That's also true. 00:04:24.740 --> 00:04:27.640 align:middle line:84% So I was in part looking for this thing 00:04:27.640 --> 00:04:29.540 align:middle line:84% about the current account deficit. 00:04:29.540 --> 00:04:33.970 align:middle line:84% And so, yes, it's interstate trade and states 00:04:33.970 --> 00:04:37.240 align:middle line:84% that are hit the worst by China shock, 00:04:37.240 --> 00:04:41.960 align:middle line:84% meaning that their local manufacturers had to compete 00:04:41.960 --> 00:04:43.370 align:middle line:90% with these Chinese imports. 00:04:43.370 --> 00:04:48.720 align:middle line:84% Those states export less to other states, 00:04:48.720 --> 00:04:53.290 align:middle line:84% and also to some extent, import more. 00:04:53.290 --> 00:04:56.320 align:middle line:84% Does anyone remember what else is on the left-hand side 00:04:56.320 --> 00:04:59.130 align:middle line:90% of the current account? 00:04:59.130 --> 00:05:02.700 align:middle line:84% Current account consists of the trade balance plus something. 00:05:02.700 --> 00:05:05.240 align:middle line:90% 00:05:05.240 --> 00:05:07.040 align:middle line:90% AUDIENCE: Net income? 00:05:07.040 --> 00:05:13.690 align:middle line:84% ROBERT TOWNSEND: Net income, yeah, so meaning gifts incoming 00:05:13.690 --> 00:05:18.190 align:middle line:84% or remittances for factor payments 00:05:18.190 --> 00:05:20.980 align:middle line:90% earned in other states. 00:05:20.980 --> 00:05:22.780 align:middle line:90% And does anyone remember? 00:05:22.780 --> 00:05:25.501 align:middle line:90% Did it go up or down? 00:05:25.501 --> 00:05:27.517 align:middle line:90% [INTERPOSING VOICES] 00:05:27.517 --> 00:05:28.350 align:middle line:90% AUDIENCE: I'm sorry. 00:05:28.350 --> 00:05:28.850 align:middle line:90% Go ahead. 00:05:28.850 --> 00:05:31.670 align:middle line:90% 00:05:31.670 --> 00:05:34.460 align:middle line:84% I was going to say that I think transfers were positive. 00:05:34.460 --> 00:05:36.540 align:middle line:90% ROBERT TOWNSEND: Yeah, good. 00:05:36.540 --> 00:05:37.040 align:middle line:90% OK. 00:05:37.040 --> 00:05:39.020 align:middle line:84% So this linked up to risk-sharing. 00:05:39.020 --> 00:05:43.140 align:middle line:84% And I'm not sure if I emphasized it last time. 00:05:43.140 --> 00:05:46.310 align:middle line:84% But given that it's quote, a "shock," a China 00:05:46.310 --> 00:05:49.160 align:middle line:84% shock, or at least an adverse trend, 00:05:49.160 --> 00:05:51.470 align:middle line:84% the question is, what arrangements 00:05:51.470 --> 00:05:52.610 align:middle line:90% would cover these shocks? 00:05:52.610 --> 00:05:55.360 align:middle line:90% 00:05:55.360 --> 00:05:57.370 align:middle line:84% Public transfers in principle could 00:05:57.370 --> 00:06:01.480 align:middle line:84% have done that through fiscal taxes and transfers. 00:06:01.480 --> 00:06:03.370 align:middle line:84% But what we found in the data was 00:06:03.370 --> 00:06:06.700 align:middle line:84% that it was the private transfers that seemed 00:06:06.700 --> 00:06:10.010 align:middle line:90% over a 10-year period to go up. 00:06:10.010 --> 00:06:14.130 align:middle line:84% So it definitely relates to all the risk-sharing stuff 00:06:14.130 --> 00:06:16.770 align:middle line:84% we did in the last few lectures and that we 00:06:16.770 --> 00:06:18.390 align:middle line:90% are about to review today. 00:06:18.390 --> 00:06:23.820 align:middle line:90% 00:06:23.820 --> 00:06:26.760 align:middle line:84% And what about the other side of the current account, 00:06:26.760 --> 00:06:28.860 align:middle line:84% financial stuff, anyone remember? 00:06:28.860 --> 00:06:38.630 align:middle line:90% 00:06:38.630 --> 00:06:40.490 align:middle line:84% Maybe these financial accounts are tricky. 00:06:40.490 --> 00:06:43.100 align:middle line:90% Well, we'll review it today. 00:06:43.100 --> 00:06:46.190 align:middle line:84% But we didn't find very much adjusting. 00:06:46.190 --> 00:06:52.900 align:middle line:84% On the right-hand side, we should see a given state, 00:06:52.900 --> 00:06:53.650 align:middle line:90% losing-- 00:06:53.650 --> 00:06:56.950 align:middle line:84% losing its claims on foreigners in this other states, 00:06:56.950 --> 00:06:59.800 align:middle line:84% in the sense of running a deficit. 00:06:59.800 --> 00:07:03.670 align:middle line:84% So they should be running down their savings accounts, 00:07:03.670 --> 00:07:04.460 align:middle line:90% and so on. 00:07:04.460 --> 00:07:08.180 align:middle line:84% But so far, we haven't found that in the data. 00:07:08.180 --> 00:07:13.010 align:middle line:84% But it's still good to know what was the financial adjustment 00:07:13.010 --> 00:07:17.030 align:middle line:84% on the right-hand side, because the accounts have to balance. 00:07:17.030 --> 00:07:18.960 align:middle line:90% Let's do one more. 00:07:18.960 --> 00:07:21.890 align:middle line:84% What in practice in the US is the difficulty 00:07:21.890 --> 00:07:26.180 align:middle line:84% in reconciling income flows with the change in the balance 00:07:26.180 --> 00:07:27.890 align:middle line:90% sheet? 00:07:27.890 --> 00:07:35.370 align:middle line:84% That is to say explain what accounting identity ought 00:07:35.370 --> 00:07:39.340 align:middle line:84% to hold, and what in contrast we actually 00:07:39.340 --> 00:07:45.950 align:middle line:84% have in integrated macro accounts and household surveys. 00:07:45.950 --> 00:07:49.250 align:middle line:84% AUDIENCE: I think I could take a shot at what identity might-- 00:07:49.250 --> 00:07:50.810 align:middle line:84% I think I know what the identity is. 00:07:50.810 --> 00:07:53.480 align:middle line:84% I'm not sure exactly how it might feel though. 00:07:53.480 --> 00:07:54.787 align:middle line:90% ROBERT TOWNSEND: Go ahead. 00:07:54.787 --> 00:07:56.870 align:middle line:84% AUDIENCE: Isn't like, the identity should probably 00:07:56.870 --> 00:08:00.800 align:middle line:84% be something like the expenditures 00:08:00.800 --> 00:08:04.820 align:middle line:84% by households which should be matched up 00:08:04.820 --> 00:08:09.620 align:middle line:84% by the incomes to whatever firm is selling the goods 00:08:09.620 --> 00:08:11.500 align:middle line:90% or something like that. 00:08:11.500 --> 00:08:13.270 align:middle line:84% Those two sides of the balance sheets 00:08:13.270 --> 00:08:18.920 align:middle line:84% should match up if it's within a country within the country. 00:08:18.920 --> 00:08:19.880 align:middle line:90% Right? 00:08:19.880 --> 00:08:22.593 align:middle line:84% So it's something like that, is it? 00:08:22.593 --> 00:08:24.260 align:middle line:84% ROBERT TOWNSEND: Well, it has to do more 00:08:24.260 --> 00:08:28.520 align:middle line:84% with what is the relationship between flows and the income 00:08:28.520 --> 00:08:31.235 align:middle line:84% statement, and assets in the balance sheet? 00:08:31.235 --> 00:08:38.947 align:middle line:90% 00:08:38.947 --> 00:08:39.780 align:middle line:90% AUDIENCE: All right. 00:08:39.780 --> 00:08:43.756 align:middle line:90% 00:08:43.756 --> 00:08:46.620 align:middle line:90% Yeah, I'm sorry. 00:08:46.620 --> 00:08:48.900 align:middle line:90% ROBERT TOWNSEND: Anyone else? 00:08:48.900 --> 00:08:50.790 align:middle line:84% AUDIENCE: Sure, but basically I think it all 00:08:50.790 --> 00:08:52.410 align:middle line:84% should be able to match in a sense 00:08:52.410 --> 00:08:54.535 align:middle line:84% that we should be able to account for these dollars 00:08:54.535 --> 00:08:57.420 align:middle line:84% on both sides, in both the flows and in the assets. 00:08:57.420 --> 00:09:00.220 align:middle line:84% But because of the way that the data is structured, 00:09:00.220 --> 00:09:03.300 align:middle line:84% there's no basically central accounting place 00:09:03.300 --> 00:09:06.147 align:middle line:84% where all of these are accounted for on a regional level 00:09:06.147 --> 00:09:07.980 align:middle line:84% in a way that just like has all of the data. 00:09:07.980 --> 00:09:10.270 align:middle line:84% And so basically nothing matches up. 00:09:10.270 --> 00:09:14.365 align:middle line:84% But we had that table with errors in the-- 00:09:14.365 --> 00:09:15.240 align:middle line:90% I forget what it was. 00:09:15.240 --> 00:09:19.218 align:middle line:84% Like 20 plus or percent or something like that. 00:09:19.218 --> 00:09:20.760 align:middle line:84% ROBERT TOWNSEND: Yes, that's correct. 00:09:20.760 --> 00:09:22.260 align:middle line:90% Great answer. 00:09:22.260 --> 00:09:27.350 align:middle line:84% So the logic of it is you have flows in the income statement. 00:09:27.350 --> 00:09:31.460 align:middle line:84% You have revenues less expenses equal profits, 00:09:31.460 --> 00:09:34.850 align:middle line:90% or for households net income. 00:09:34.850 --> 00:09:39.890 align:middle line:84% And what they don't eat or pay out as dividends, 00:09:39.890 --> 00:09:41.360 align:middle line:90% they should be saving. 00:09:41.360 --> 00:09:46.800 align:middle line:84% So the changes in the balance sheet, some balance sheet items 00:09:46.800 --> 00:09:50.790 align:middle line:84% ought to go up if households or firms are saving. 00:09:50.790 --> 00:09:55.710 align:middle line:84% And in practice, since as was just said, 00:09:55.710 --> 00:09:57.900 align:middle line:84% the balance sheet and income statements 00:09:57.900 --> 00:10:01.440 align:middle line:84% are not collected at the same time, 00:10:01.440 --> 00:10:06.240 align:middle line:84% we see very big discrepancies between the two. 00:10:06.240 --> 00:10:09.830 align:middle line:84% So the rest of the class last time, almost up to the very end 00:10:09.830 --> 00:10:13.130 align:middle line:84% was about what this discrepancy means 00:10:13.130 --> 00:10:16.010 align:middle line:84% for attempts to measure inequality, 00:10:16.010 --> 00:10:18.500 align:middle line:84% or the impact of COVID, and so on. 00:10:18.500 --> 00:10:21.540 align:middle line:90% 00:10:21.540 --> 00:10:23.120 align:middle line:84% So I'll just say what I was trying 00:10:23.120 --> 00:10:29.210 align:middle line:84% to do in the class is not to get you to memorize 00:10:29.210 --> 00:10:33.440 align:middle line:84% financial account identities, but rather to think creatively 00:10:33.440 --> 00:10:37.700 align:middle line:84% about how to use financial account data, and so on. 00:10:37.700 --> 00:10:40.740 align:middle line:90% 00:10:40.740 --> 00:10:42.840 align:middle line:84% Which brought me to the very end, where 00:10:42.840 --> 00:10:46.630 align:middle line:84% I talked about trying to fill in those gaps, 00:10:46.630 --> 00:10:52.240 align:middle line:84% and gathering detailed level information from households 00:10:52.240 --> 00:10:54.010 align:middle line:90% in West Virginia, and so on. 00:10:54.010 --> 00:10:59.020 align:middle line:90% 00:10:59.020 --> 00:11:00.974 align:middle line:90% But I won't ask about that. 00:11:00.974 --> 00:11:05.860 align:middle line:84% Well, that may come up again today, actually. 00:11:05.860 --> 00:11:09.420 align:middle line:84% So let's go to the review from lecture 8 00:11:09.420 --> 00:11:13.350 align:middle line:90% on, so I will try to highlight. 00:11:13.350 --> 00:11:16.620 align:middle line:84% We did risk sharing in India, in the villages, 00:11:16.620 --> 00:11:21.900 align:middle line:84% and also in medieval England in terms of dividing up the land. 00:11:21.900 --> 00:11:23.910 align:middle line:84% I talked at the beginning of the lecture 00:11:23.910 --> 00:11:28.950 align:middle line:84% about the high-risk environment and the way households could 00:11:28.950 --> 00:11:31.710 align:middle line:84% respond to it, either by diversifying 00:11:31.710 --> 00:11:35.280 align:middle line:84% or by helping each other out with financial gifts 00:11:35.280 --> 00:11:39.310 align:middle line:84% and transfers, and then ways to test. 00:11:39.310 --> 00:11:43.200 align:middle line:84% So in India, we decided to test by looking 00:11:43.200 --> 00:11:48.450 align:middle line:84% at incomes and consumptions, and not looking at the mechanism. 00:11:48.450 --> 00:11:51.330 align:middle line:84% And in contrast, with medieval England, 00:11:51.330 --> 00:11:55.710 align:middle line:84% we focus more on whether or not the mechanism 00:11:55.710 --> 00:12:01.320 align:middle line:84% of dividing up land could achieve an optimal allocation. 00:12:01.320 --> 00:12:03.090 align:middle line:84% So then we went through characteristics 00:12:03.090 --> 00:12:04.800 align:middle line:90% of village economies. 00:12:04.800 --> 00:12:08.760 align:middle line:84% And the main thing here was the risk. 00:12:08.760 --> 00:12:11.970 align:middle line:84% In particular, in village India, we 00:12:11.970 --> 00:12:15.090 align:middle line:84% looked at the risk of harvest yields 00:12:15.090 --> 00:12:22.560 align:middle line:84% on various crops, of crops by soil type, of income by type. 00:12:22.560 --> 00:12:26.310 align:middle line:84% And we uniformly saw a very high risk 00:12:26.310 --> 00:12:31.260 align:middle line:84% on all these categories, as well as some diversification 00:12:31.260 --> 00:12:32.180 align:middle line:90% possibilities. 00:12:32.180 --> 00:12:37.680 align:middle line:84% So for income, not everyone has the same income sources, 00:12:37.680 --> 00:12:39.420 align:middle line:90% and there's high risk of each. 00:12:39.420 --> 00:12:43.750 align:middle line:84% But the covariances are sometimes relatively low. 00:12:43.750 --> 00:12:46.560 align:middle line:84% So the focus here was on the fact 00:12:46.560 --> 00:12:49.020 align:middle line:84% that households are not completely diversified. 00:12:49.020 --> 00:12:51.780 align:middle line:84% If all the households did the same thing 00:12:51.780 --> 00:12:54.720 align:middle line:84% if, they planted the same crops in the same soil, 00:12:54.720 --> 00:12:57.780 align:middle line:84% and have the same income sources then 00:12:57.780 --> 00:13:03.070 align:middle line:84% their initial outcomes would be essentially the same. 00:13:03.070 --> 00:13:06.630 align:middle line:84% And if that's true, then there's not much left to share. 00:13:06.630 --> 00:13:08.110 align:middle line:90% But we see the opposite. 00:13:08.110 --> 00:13:11.970 align:middle line:84% We see they're not diversified over all these various crops, 00:13:11.970 --> 00:13:13.600 align:middle line:90% and soils, and activities. 00:13:13.600 --> 00:13:16.350 align:middle line:84% So there's a lot of idiosyncratic risk. 00:13:16.350 --> 00:13:19.230 align:middle line:84% And that was what was clear from this picture of what 00:13:19.230 --> 00:13:22.890 align:middle line:84% we call the Rocky Mountains, where the ups and downs are not 00:13:22.890 --> 00:13:26.640 align:middle line:84% together, not going up together across households in a given 00:13:26.640 --> 00:13:28.540 align:middle line:90% year, and so on. 00:13:28.540 --> 00:13:31.500 align:middle line:84% Whereas, they do achieve a lot of smoothing. 00:13:31.500 --> 00:13:34.890 align:middle line:84% So something is happening in the villages 00:13:34.890 --> 00:13:40.170 align:middle line:84% that whatever mechanism it is, is allowing them to smooth. 00:13:40.170 --> 00:13:46.020 align:middle line:84% Now one question that's come up in Piazza are these profiles. 00:13:46.020 --> 00:13:49.470 align:middle line:84% I mean so this is a bit of a review, 00:13:49.470 --> 00:13:50.700 align:middle line:90% but kind of interesting. 00:13:50.700 --> 00:13:55.110 align:middle line:84% We see this income profile from low to high. 00:13:55.110 --> 00:13:57.480 align:middle line:84% Households with a lot of land are consistently 00:13:57.480 --> 00:14:00.300 align:middle line:84% getting higher income, despite the fluctuations. 00:14:00.300 --> 00:14:04.450 align:middle line:84% And there's a big compression here. 00:14:04.450 --> 00:14:06.630 align:middle line:84% So something is transferring income 00:14:06.630 --> 00:14:10.170 align:middle line:84% from the landed households to the landless. 00:14:10.170 --> 00:14:13.110 align:middle line:84% But a little bit of this profile does remain. 00:14:13.110 --> 00:14:14.707 align:middle line:90% So it's not equal treatment. 00:14:14.707 --> 00:14:16.290 align:middle line:84% It's not as though everyone is getting 00:14:16.290 --> 00:14:18.390 align:middle line:90% the same amount of everything. 00:14:18.390 --> 00:14:21.510 align:middle line:84% There is heterogeneity in the treatment. 00:14:21.510 --> 00:14:24.510 align:middle line:84% What that mechanism is, is not so clear. 00:14:24.510 --> 00:14:27.690 align:middle line:84% It's some kind of, at least over these 10 years, 00:14:27.690 --> 00:14:32.500 align:middle line:84% redistribution of income from the wealthy to the poor. 00:14:32.500 --> 00:14:34.810 align:middle line:90% But it doesn't work perfectly. 00:14:34.810 --> 00:14:37.790 align:middle line:84% It doesn't completely equalize things. 00:14:37.790 --> 00:14:40.210 align:middle line:84% So this lecture was featuring something 00:14:40.210 --> 00:14:42.460 align:middle line:84% we had done in the previous one, which 00:14:42.460 --> 00:14:45.970 align:middle line:84% is to determine the pareto optimal allocation of risk, 00:14:45.970 --> 00:14:50.350 align:middle line:84% we maximize lambda weighted sums of expected utilities, 00:14:50.350 --> 00:14:53.360 align:middle line:84% subject to the resource constraint. 00:14:53.360 --> 00:14:58.330 align:middle line:84% And this way of dealing with problems 00:14:58.330 --> 00:15:03.170 align:middle line:84% generalizes considerably and in the subsequent lectures. 00:15:03.170 --> 00:15:04.810 align:middle line:84% So when we solved this problem, we 00:15:04.810 --> 00:15:07.000 align:middle line:84% looked at the first order conditions. 00:15:07.000 --> 00:15:10.090 align:middle line:84% We could see that weighted marginal utilities should 00:15:10.090 --> 00:15:14.530 align:middle line:84% be equalized to a common Lagrange multiplier. 00:15:14.530 --> 00:15:16.360 align:middle line:84% And you can do the algebra, and actually 00:15:16.360 --> 00:15:20.590 align:middle line:84% solve for the consumption that's supposed to result, 00:15:20.590 --> 00:15:24.070 align:middle line:84% and get these kinds of closed form solutions 00:15:24.070 --> 00:15:29.110 align:middle line:84% having to do with consumption moving with the aggregate, 00:15:29.110 --> 00:15:34.420 align:middle line:84% and not moving at all with idiosyncratic shocks, 00:15:34.420 --> 00:15:37.605 align:middle line:90% if the risk sharing is optimal. 00:15:37.605 --> 00:15:38.980 align:middle line:84% So here is another version of it, 00:15:38.980 --> 00:15:42.850 align:middle line:84% individual consumption co-moves with aggregate consumption 00:15:42.850 --> 00:15:45.430 align:middle line:84% for all the households, but should not 00:15:45.430 --> 00:15:48.580 align:middle line:84% move with things like individual incomes 00:15:48.580 --> 00:15:50.840 align:middle line:84% once we control for the aggregate. 00:15:50.840 --> 00:15:57.040 align:middle line:84% So we tested this model to see whether these coefficients were 00:15:57.040 --> 00:16:00.040 align:middle line:90% 0 or not, and largely they are. 00:16:00.040 --> 00:16:02.980 align:middle line:84% I mean there is some residual risk that 00:16:02.980 --> 00:16:04.960 align:middle line:84% is being borne by the households, 00:16:04.960 --> 00:16:07.370 align:middle line:84% depending on the village and the source of income. 00:16:07.370 --> 00:16:10.253 align:middle line:84% But the benchmark was doing quite well. 00:16:10.253 --> 00:16:11.170 align:middle line:90% AUDIENCE: [INAUDIBLE]. 00:16:11.170 --> 00:16:12.550 align:middle line:84% ROBERT TOWNSEND: I would be delighted. 00:16:12.550 --> 00:16:13.360 align:middle line:90% AUDIENCE: Yeah. 00:16:13.360 --> 00:16:17.590 align:middle line:90% I mean here, we assume-- 00:16:17.590 --> 00:16:18.880 align:middle line:90% I see the sample. 00:16:18.880 --> 00:16:23.320 align:middle line:84% There was about 40 households in each villagers. 00:16:23.320 --> 00:16:26.890 align:middle line:84% But in this model, we assume that the households 00:16:26.890 --> 00:16:29.360 align:middle line:90% should be large amount. 00:16:29.360 --> 00:16:33.520 align:middle line:84% If not, then the idiosyncratic risk 00:16:33.520 --> 00:16:36.340 align:middle line:90% might not be shared, right? 00:16:36.340 --> 00:16:40.810 align:middle line:84% And for example, if there was just two households and two 00:16:40.810 --> 00:16:49.460 align:middle line:84% situations, then although they face the idiosyncratic risks, 00:16:49.460 --> 00:16:56.840 align:middle line:84% but there was like 0.25 probability both of them 00:16:56.840 --> 00:17:01.670 align:middle line:84% were facing negative shocks and/or both of them 00:17:01.670 --> 00:17:03.030 align:middle line:90% face the positive shock. 00:17:03.030 --> 00:17:05.160 align:middle line:90% So they cannot share the risks. 00:17:05.160 --> 00:17:13.160 align:middle line:84% So in this model, we assume the sample is large enough. 00:17:13.160 --> 00:17:19.819 align:middle line:84% That is why we think that the idiosyncratic shocks will not 00:17:19.819 --> 00:17:22.910 align:middle line:90% affect people's consumption? 00:17:22.910 --> 00:17:25.667 align:middle line:84% Yeah, each household consumption. 00:17:25.667 --> 00:17:26.500 align:middle line:90% ROBERT TOWNSEND: OK. 00:17:26.500 --> 00:17:30.100 align:middle line:90% So it's a great question. 00:17:30.100 --> 00:17:34.780 align:middle line:84% What can we test if we only have a subset of the households, 00:17:34.780 --> 00:17:38.060 align:middle line:90% and are we being misled? 00:17:38.060 --> 00:17:41.150 align:middle line:84% Well, actually you could think about it 00:17:41.150 --> 00:17:44.030 align:middle line:90% as testing for subsets. 00:17:44.030 --> 00:17:45.750 align:middle line:90% And the theory holds. 00:17:45.750 --> 00:17:48.835 align:middle line:84% Of course, what is aggregate consumption for two 00:17:48.835 --> 00:17:50.960 align:middle line:84% households would be different from what it would be 00:17:50.960 --> 00:17:54.010 align:middle line:90% if you had data on all of them. 00:17:54.010 --> 00:17:57.185 align:middle line:84% But the theory still works for two, or three, or just 00:17:57.185 --> 00:17:57.685 align:middle line:90% a handful. 00:17:57.685 --> 00:18:00.370 align:middle line:90% 00:18:00.370 --> 00:18:02.500 align:middle line:84% If they are bearing risk optimally, 00:18:02.500 --> 00:18:05.815 align:middle line:84% then they should be bearing risk optimally in any subset. 00:18:05.815 --> 00:18:08.750 align:middle line:90% 00:18:08.750 --> 00:18:11.160 align:middle line:90% Let me put that a different way. 00:18:11.160 --> 00:18:16.680 align:middle line:84% The aggregate consumption within the group may be moving around. 00:18:16.680 --> 00:18:20.360 align:middle line:84% Individual consumptions in the group should move with that. 00:18:20.360 --> 00:18:22.530 align:middle line:84% The group consumption actually could 00:18:22.530 --> 00:18:24.870 align:middle line:84% look like an idiosyncratic shock if we 00:18:24.870 --> 00:18:27.450 align:middle line:90% considered the whole village. 00:18:27.450 --> 00:18:30.470 align:middle line:84% So yet more restrictions are implied 00:18:30.470 --> 00:18:35.520 align:middle line:84% when we and if we had data from the larger sample. 00:18:35.520 --> 00:18:38.890 align:middle line:84% But you can still test the theory 00:18:38.890 --> 00:18:41.440 align:middle line:84% with a small number of households. 00:18:41.440 --> 00:18:44.380 align:middle line:84% It's just that it's not as powerful. 00:18:44.380 --> 00:18:46.510 align:middle line:90% You could reject, however. 00:18:46.510 --> 00:18:47.630 align:middle line:90% Does that help? 00:18:47.630 --> 00:18:48.657 align:middle line:90% AUDIENCE: Yeah. 00:18:48.657 --> 00:18:49.490 align:middle line:90% ROBERT TOWNSEND: OK. 00:18:49.490 --> 00:18:50.490 align:middle line:90% It was a great question. 00:18:50.490 --> 00:18:54.760 align:middle line:90% 00:18:54.760 --> 00:18:58.870 align:middle line:84% So then the next thing we did was 00:18:58.870 --> 00:19:03.010 align:middle line:84% to take the same theory to different applications, back 00:19:03.010 --> 00:19:07.600 align:middle line:84% to medieval villages, and the landholding patterns which 00:19:07.600 --> 00:19:10.270 align:middle line:84% are spread out all over the different fields. 00:19:10.270 --> 00:19:13.000 align:middle line:84% And the idea is, does that make sense or not, 00:19:13.000 --> 00:19:17.440 align:middle line:84% and would dividing up the land in that way exactly 00:19:17.440 --> 00:19:21.020 align:middle line:84% achieve an optimal allocation of risk bearing or not. 00:19:21.020 --> 00:19:24.250 align:middle line:84% So we took the different types of land 00:19:24.250 --> 00:19:27.310 align:middle line:84% to have a different vector of yields. 00:19:27.310 --> 00:19:30.970 align:middle line:84% We talked about the endowment, so to speak, 00:19:30.970 --> 00:19:36.370 align:middle line:84% the yields of land type k across states epsilon. 00:19:36.370 --> 00:19:40.420 align:middle line:84% And then we have capital K types of land 00:19:40.420 --> 00:19:43.660 align:middle line:90% and capital S types of shocks. 00:19:43.660 --> 00:19:46.570 align:middle line:84% We can look at the covariance and the risk, just 00:19:46.570 --> 00:19:48.770 align:middle line:90% the way we did in India. 00:19:48.770 --> 00:19:51.490 align:middle line:84% So again, we get this picture, very, very 00:19:51.490 --> 00:19:54.880 align:middle line:84% risky with some covariances pretty low. 00:19:54.880 --> 00:19:57.580 align:middle line:84% And then I took you through a series of examples 00:19:57.580 --> 00:20:01.660 align:middle line:84% to see how close we could get to an optimal allocation of risk 00:20:01.660 --> 00:20:02.530 align:middle line:90% bearing. 00:20:02.530 --> 00:20:06.190 align:middle line:84% The first was household 1 is all of land type 1. 00:20:06.190 --> 00:20:08.620 align:middle line:84% Household 2 has all of land type 2. 00:20:08.620 --> 00:20:11.590 align:middle line:84% But these returns do not co-move, 00:20:11.590 --> 00:20:16.150 align:middle line:84% hence this would not constitute an optimal allocation 00:20:16.150 --> 00:20:17.140 align:middle line:90% of risk bearing. 00:20:17.140 --> 00:20:19.210 align:middle line:90% Household 1 would go 10, 6. 00:20:19.210 --> 00:20:21.790 align:middle line:90% Household 2 would go 2, 4. 00:20:21.790 --> 00:20:24.280 align:middle line:84% So although the aggregate is going down, 00:20:24.280 --> 00:20:29.980 align:middle line:84% I should say, from 12 to 10, household 2's consumption 00:20:29.980 --> 00:20:33.370 align:middle line:84% if they ate all their own crops would be going up. 00:20:33.370 --> 00:20:35.090 align:middle line:90% So that's a restriction. 00:20:35.090 --> 00:20:37.810 align:middle line:84% Then we considered a different way of doing it. 00:20:37.810 --> 00:20:40.030 align:middle line:90% Give those household shares. 00:20:40.030 --> 00:20:46.030 align:middle line:84% Household j say having a share alpha j 00:20:46.030 --> 00:20:48.200 align:middle line:90% of the two different crops. 00:20:48.200 --> 00:20:53.200 align:middle line:84% So now, this thing goes in the aggregate 12, 10, 7. 00:20:53.200 --> 00:20:56.200 align:middle line:84% Obviously whatever household 1's share 00:20:56.200 --> 00:21:00.160 align:middle line:84% is, if it's like one third, it's 1/3 of that aggregate. 00:21:00.160 --> 00:21:03.640 align:middle line:84% Household 2's shares would be 2/3 of that aggregate. 00:21:03.640 --> 00:21:06.010 align:middle line:84% So now they're each bearing a constant share 00:21:06.010 --> 00:21:08.320 align:middle line:84% of the aggregate and their consumptions 00:21:08.320 --> 00:21:11.320 align:middle line:90% are moving with the aggregate. 00:21:11.320 --> 00:21:14.000 align:middle line:90% So that might actually work. 00:21:14.000 --> 00:21:14.900 align:middle line:90% That's this. 00:21:14.900 --> 00:21:17.290 align:middle line:84% And then we looked at special utility functions, 00:21:17.290 --> 00:21:20.680 align:middle line:84% and started fussing about intercepts, and risk aversion, 00:21:20.680 --> 00:21:21.740 align:middle line:90% and so on. 00:21:21.740 --> 00:21:23.980 align:middle line:84% But then we come to the general idea 00:21:23.980 --> 00:21:29.590 align:middle line:84% that we have as many land types with independent return vectors 00:21:29.590 --> 00:21:31.790 align:middle line:90% as we have states of the world. 00:21:31.790 --> 00:21:35.960 align:middle line:84% And then we have a target optimal consumption allocation. 00:21:35.960 --> 00:21:39.010 align:middle line:84% So we're trying to achieve this allocation that 00:21:39.010 --> 00:21:42.790 align:middle line:84% came from maximizing a lambda weighted sums of utilities 00:21:42.790 --> 00:21:48.490 align:middle line:84% by dividing the different land types into shares. 00:21:48.490 --> 00:21:51.370 align:middle line:84% Except now we have even more power, 00:21:51.370 --> 00:21:54.580 align:middle line:84% because we do not require that household j have 00:21:54.580 --> 00:21:57.400 align:middle line:84% the same shares and all of the land types k. 00:21:57.400 --> 00:22:01.330 align:middle line:84% We have this alpha jk for household j in the land type k. 00:22:01.330 --> 00:22:03.010 align:middle line:90% And it's not perfect. 00:22:03.010 --> 00:22:06.040 align:middle line:84% But in general, since this vector of endowments 00:22:06.040 --> 00:22:11.170 align:middle line:84% is non-singular, you can solve for the consumptions. 00:22:11.170 --> 00:22:15.380 align:middle line:84% And one hopes that the shares will be positive in the end. 00:22:15.380 --> 00:22:17.920 align:middle line:84% So if there are as many types of land 00:22:17.920 --> 00:22:20.290 align:middle line:84% as there are states of the world, 00:22:20.290 --> 00:22:24.490 align:middle line:84% they have a shot at achieving the optimum by entirely 00:22:24.490 --> 00:22:27.190 align:middle line:84% dividing up the land, and not having 00:22:27.190 --> 00:22:31.340 align:middle line:84% to give gifts or transfers to one another after the fact. 00:22:31.340 --> 00:22:34.040 align:middle line:90% So then we come to-- 00:22:34.040 --> 00:22:36.790 align:middle line:84% we put in particular utility functions, 00:22:36.790 --> 00:22:42.010 align:middle line:84% and we assumed that we're going to solve 00:22:42.010 --> 00:22:46.570 align:middle line:84% this problem, maximizing lambda weighted sums of utilities, 00:22:46.570 --> 00:22:49.330 align:middle line:90% by the shares alpha jk. 00:22:49.330 --> 00:22:51.970 align:middle line:84% But we're going to assume that there 00:22:51.970 --> 00:22:56.120 align:middle line:84% are fewer land types than there are states of the world, 00:22:56.120 --> 00:23:01.720 align:middle line:84% namely only two types of land, and a larger 00:23:01.720 --> 00:23:04.190 align:middle line:90% number as states of the world. 00:23:04.190 --> 00:23:08.950 align:middle line:84% So now, we know we in general cannot achieve an optimum. 00:23:08.950 --> 00:23:12.310 align:middle line:84% But we were trying to describe this constrained optimal 00:23:12.310 --> 00:23:13.510 align:middle line:90% arrangement. 00:23:13.510 --> 00:23:17.050 align:middle line:84% I guess what I want to emphasize is despite the restrictions, 00:23:17.050 --> 00:23:19.180 align:middle line:84% we're still solving this programming problem. 00:23:19.180 --> 00:23:22.000 align:middle line:84% We're maximizing lambda weighted sums 00:23:22.000 --> 00:23:24.400 align:middle line:84% of utilities subject to constraints, 00:23:24.400 --> 00:23:26.410 align:middle line:84% namely the shares have to add up. 00:23:26.410 --> 00:23:29.320 align:middle line:84% So the method of achieving characterizing 00:23:29.320 --> 00:23:32.510 align:middle line:84% pareto optimal allocations is the same. 00:23:32.510 --> 00:23:35.860 align:middle line:84% And then we looked to see with those particular utility 00:23:35.860 --> 00:23:40.420 align:middle line:84% functions how far off we would be from achieving the optimum. 00:23:40.420 --> 00:23:43.270 align:middle line:84% And here you can actually see, as the aggregate endowment 00:23:43.270 --> 00:23:45.070 align:middle line:90% moves the adjustment. 00:23:45.070 --> 00:23:47.890 align:middle line:84% In other words, even if they did as much as they could 00:23:47.890 --> 00:23:51.910 align:middle line:84% by dividing up the two land types, it would not be perfect. 00:23:51.910 --> 00:23:55.330 align:middle line:84% And there should be incoming or outgoing transfers 00:23:55.330 --> 00:23:59.410 align:middle line:84% to try to make up the difference to get to the full risk sharing 00:23:59.410 --> 00:24:00.670 align:middle line:90% optimum. 00:24:00.670 --> 00:24:03.110 align:middle line:84% And that was the end of this lecture. 00:24:03.110 --> 00:24:05.200 align:middle line:84% Are there questions about what we did 00:24:05.200 --> 00:24:07.525 align:middle line:84% or what we were trying to accomplish in lecture 8? 00:24:07.525 --> 00:24:11.160 align:middle line:90% 00:24:11.160 --> 00:24:14.730 align:middle line:84% So then the second of the two application lectures 00:24:14.730 --> 00:24:17.850 align:middle line:84% included production in lecture 9. 00:24:17.850 --> 00:24:22.410 align:middle line:84% And the main difference between what we did just now 00:24:22.410 --> 00:24:25.740 align:middle line:84% and this lecture, is we actually have physical production, 00:24:25.740 --> 00:24:29.310 align:middle line:84% rather than treating output as an endowment. 00:24:29.310 --> 00:24:33.350 align:middle line:84% But we still solve this programming problem. 00:24:33.350 --> 00:24:36.580 align:middle line:84% So we're maximizing lambda weighted sums 00:24:36.580 --> 00:24:40.390 align:middle line:84% of expected utilities, just as before, 00:24:40.390 --> 00:24:44.950 align:middle line:84% subject to resource constraints, just as before. 00:24:44.950 --> 00:24:51.040 align:middle line:84% The choice variables now include the transfers, just as before, 00:24:51.040 --> 00:24:54.820 align:middle line:84% but we also are choosing which projects 00:24:54.820 --> 00:24:58.280 align:middle line:84% to fund and by how much capital investment. 00:24:58.280 --> 00:25:01.810 align:middle line:84% So the reason that transfers are the same as before, 00:25:01.810 --> 00:25:04.780 align:middle line:84% is because we have transfers or consumption. 00:25:04.780 --> 00:25:07.630 align:middle line:84% You can move consumption around by moving transfers. 00:25:07.630 --> 00:25:13.240 align:middle line:84% So it's arbitrary choice whether to solve for c or tau. 00:25:13.240 --> 00:25:15.010 align:middle line:84% The more interesting part, because it's 00:25:15.010 --> 00:25:17.860 align:middle line:84% new to this lecture, are these capital 00:25:17.860 --> 00:25:21.550 align:middle line:84% stocks, which has to do with deciding what 00:25:21.550 --> 00:25:25.510 align:middle line:84% to do with the current wealth, namely eat it or invested 00:25:25.510 --> 00:25:28.330 align:middle line:90% in projects for tomorrow. 00:25:28.330 --> 00:25:31.720 align:middle line:84% Here is the starting point, w, which 00:25:31.720 --> 00:25:34.180 align:middle line:84% is the trees and the fruit from the trees. 00:25:34.180 --> 00:25:38.380 align:middle line:84% And this was the ending point, what to do, either eat or save. 00:25:38.380 --> 00:25:40.720 align:middle line:84% Oh then, back to the accounting identities, 00:25:40.720 --> 00:25:44.800 align:middle line:84% you either eat something or save something, as in the flows. 00:25:44.800 --> 00:25:48.380 align:middle line:84% So we got two first order conditions out of this problem, 00:25:48.380 --> 00:25:51.790 align:middle line:84% one very familiar, lambda weighted marginal utilities 00:25:51.790 --> 00:25:53.540 align:middle line:90% are equated to each other. 00:25:53.540 --> 00:25:55.780 align:middle line:84% We also got a new equation having 00:25:55.780 --> 00:25:59.410 align:middle line:84% to do with the marginal utility of consumption today, 00:25:59.410 --> 00:26:03.170 align:middle line:84% and the marginal utility of consumption next period. 00:26:03.170 --> 00:26:06.590 align:middle line:84% And those typically will balance each other off 00:26:06.590 --> 00:26:08.510 align:middle line:90% for any project they're running. 00:26:08.510 --> 00:26:11.840 align:middle line:84% There was an issue that this thing could have been negative, 00:26:11.840 --> 00:26:16.342 align:middle line:84% in which case that project would not be implemented, 00:26:16.342 --> 00:26:18.800 align:middle line:84% because they would actually not want to put stuff in there. 00:26:18.800 --> 00:26:20.060 align:middle line:90% They'd want to take stuff out. 00:26:20.060 --> 00:26:21.680 align:middle line:90% But they hit zero. 00:26:21.680 --> 00:26:25.940 align:middle line:84% So the rest of it was to take expressions 00:26:25.940 --> 00:26:27.440 align:middle line:84% for these first order conditions, 00:26:27.440 --> 00:26:29.630 align:middle line:90% and start manipulating them. 00:26:29.630 --> 00:26:32.120 align:middle line:84% I'm not going to review that today. 00:26:32.120 --> 00:26:33.800 align:middle line:90% It's algebra. 00:26:33.800 --> 00:26:35.700 align:middle line:90% It's straightforward. 00:26:35.700 --> 00:26:37.970 align:middle line:84% We can even make some assumptions 00:26:37.970 --> 00:26:41.810 align:middle line:84% about linear technologies, subject to risk, 00:26:41.810 --> 00:26:45.050 align:middle line:90% and quadratic utility functions. 00:26:45.050 --> 00:26:49.130 align:middle line:84% And when we did that, we got two expressions directly 00:26:49.130 --> 00:26:52.220 align:middle line:84% from the theory that we could take to the data. 00:26:52.220 --> 00:26:56.270 align:middle line:84% One, we can measure from the data 00:26:56.270 --> 00:27:00.920 align:middle line:84% how household j returns are moving with the village 00:27:00.920 --> 00:27:02.990 align:middle line:84% aggregate return, this beta being 00:27:02.990 --> 00:27:07.010 align:middle line:84% a measure of the covariation of household j's return 00:27:07.010 --> 00:27:08.720 align:middle line:90% with the village average. 00:27:08.720 --> 00:27:14.480 align:middle line:84% Then having used the panel data to get these beta j's, we 00:27:14.480 --> 00:27:18.620 align:middle line:84% look in the cross section to see how the betas are related 00:27:18.620 --> 00:27:23.750 align:middle line:84% to each other, and in particular to the overall average 00:27:23.750 --> 00:27:25.530 align:middle line:90% or expected return. 00:27:25.530 --> 00:27:28.070 align:middle line:84% So the main empirical finding here 00:27:28.070 --> 00:27:31.760 align:middle line:84% was that returns have to be higher 00:27:31.760 --> 00:27:33.920 align:middle line:84% when the betas are higher, because 00:27:33.920 --> 00:27:38.060 align:middle line:84% those particular activities from household j co-move 00:27:38.060 --> 00:27:39.830 align:middle line:90% move a lot with the aggregate. 00:27:39.830 --> 00:27:45.170 align:middle line:84% So in effect, it's very risky, not so many diversification 00:27:45.170 --> 00:27:47.720 align:middle line:84% possibilities come from including household 00:27:47.720 --> 00:27:50.520 align:middle line:90% j in the village portfolio. 00:27:50.520 --> 00:27:53.600 align:middle line:84% And so if they're going to be running 00:27:53.600 --> 00:27:56.360 align:middle line:84% those projects of household j, there 00:27:56.360 --> 00:27:58.490 align:middle line:84% has to be a compensating differential, 00:27:58.490 --> 00:28:01.200 align:middle line:84% and that's the higher expected return. 00:28:01.200 --> 00:28:06.035 align:middle line:84% So then we started adopting this language of risk premia, which 00:28:06.035 --> 00:28:09.230 align:middle line:84% is the difference between the expected return 00:28:09.230 --> 00:28:12.000 align:middle line:90% and the risk-free rate. 00:28:12.000 --> 00:28:15.320 align:middle line:84% The risk premia has to compensate for the risk. 00:28:15.320 --> 00:28:17.510 align:middle line:84% So then we went to the Thai data, 00:28:17.510 --> 00:28:19.110 align:middle line:84% where we have everything we need. 00:28:19.110 --> 00:28:21.260 align:middle line:84% We have the income and the balance sheet. 00:28:21.260 --> 00:28:24.080 align:middle line:84% Oh yeah, so here are these accounts. 00:28:24.080 --> 00:28:26.330 align:middle line:84% This, I think, was the first time 00:28:26.330 --> 00:28:29.750 align:middle line:84% the accounts started to come up explicitly. 00:28:29.750 --> 00:28:34.370 align:middle line:84% We used the Thai monthly data over 13 years, 00:28:34.370 --> 00:28:37.940 align:middle line:84% and I featured net income to be profits, 00:28:37.940 --> 00:28:40.280 align:middle line:90% to be revenue minus costs. 00:28:40.280 --> 00:28:43.940 align:middle line:84% And then we looked at the assets to be 00:28:43.940 --> 00:28:46.490 align:middle line:84% assets of various kinds, typically 00:28:46.490 --> 00:28:48.830 align:middle line:90% physical tangible assets. 00:28:48.830 --> 00:28:51.270 align:middle line:84% So here you have the income statement and the balance 00:28:51.270 --> 00:28:51.770 align:middle line:90% sheet. 00:28:51.770 --> 00:28:54.510 align:middle line:84% I just didn't call it that at the time, 00:28:54.510 --> 00:28:56.310 align:middle line:90% but that's what we're using. 00:28:56.310 --> 00:28:58.640 align:middle line:84% So then we tested the model, and found 00:28:58.640 --> 00:29:01.190 align:middle line:84% that the betas, the expected returns, 00:29:01.190 --> 00:29:04.230 align:middle line:84% are indeed higher when the betas are higher, 00:29:04.230 --> 00:29:07.190 align:middle line:90% just like the model predicted. 00:29:07.190 --> 00:29:11.270 align:middle line:84% And here, we are back to idiosyncratic and aggregate 00:29:11.270 --> 00:29:13.820 align:middle line:84% risk, which was a theme I mentioned 00:29:13.820 --> 00:29:15.510 align:middle line:90% in the previous lecture. 00:29:15.510 --> 00:29:22.250 align:middle line:84% We can decompose the aggregate variation of household j, 00:29:22.250 --> 00:29:24.530 align:middle line:84% and explain the unexplained part. 00:29:24.530 --> 00:29:26.960 align:middle line:84% The explained part has to do with the covariation 00:29:26.960 --> 00:29:28.670 align:middle line:90% with the village average. 00:29:28.670 --> 00:29:32.630 align:middle line:84% And the unexplained part is just there were some idiosyncratic 00:29:32.630 --> 00:29:33.650 align:middle line:90% risk. 00:29:33.650 --> 00:29:36.920 align:middle line:84% And we decompose the risk premium 00:29:36.920 --> 00:29:40.640 align:middle line:84% as well into the part that's coming 00:29:40.640 --> 00:29:45.650 align:middle line:84% from the village as a whole, through the beta j, as well as 00:29:45.650 --> 00:29:50.600 align:middle line:84% the sigma square sigma j on that epsilon. 00:29:50.600 --> 00:29:52.950 align:middle line:84% And then got another decomposition. 00:29:52.950 --> 00:29:56.450 align:middle line:84% So it turned out that, as a quick review, 00:29:56.450 --> 00:30:00.470 align:middle line:84% idiosyncratic risk was the largest contributor 00:30:00.470 --> 00:30:03.260 align:middle line:84% to total risk, and aggregate village level 00:30:03.260 --> 00:30:04.940 align:middle line:90% risk was a small part. 00:30:04.940 --> 00:30:07.460 align:middle line:84% So there were a lot of idiosyncratic shocks 00:30:07.460 --> 00:30:09.650 align:middle line:84% in these villages, just like we imagined 00:30:09.650 --> 00:30:12.200 align:middle line:84% that there were in medieval England, 00:30:12.200 --> 00:30:17.120 align:middle line:84% and also that we documented there were in village India. 00:30:17.120 --> 00:30:19.100 align:middle line:84% But when you looked at the risk premia, 00:30:19.100 --> 00:30:23.900 align:middle line:84% it went the other way, namely most of the risk 00:30:23.900 --> 00:30:25.590 align:middle line:84% has to do with the aggregate risk. 00:30:25.590 --> 00:30:29.030 align:middle line:84% So that's consistent with them finding a way 00:30:29.030 --> 00:30:32.660 align:middle line:84% to diversify their idiosyncratic shocks. 00:30:32.660 --> 00:30:34.760 align:middle line:84% And there was evidence that I presented 00:30:34.760 --> 00:30:37.730 align:middle line:84% that they're doing that by gifts, and borrowing, 00:30:37.730 --> 00:30:38.930 align:middle line:90% and lending. 00:30:38.930 --> 00:30:40.760 align:middle line:90% And that's the mechanism. 00:30:40.760 --> 00:30:42.800 align:middle line:84% We started out talking about how we weren't 00:30:42.800 --> 00:30:44.570 align:middle line:90% going to test for mechanisms. 00:30:44.570 --> 00:30:47.310 align:middle line:84% Then we looked at ex-anti-division of land. 00:30:47.310 --> 00:30:49.700 align:middle line:84% Now we're looking at ex-post gifts 00:30:49.700 --> 00:30:54.710 align:middle line:84% and borrowing as a way of getting from their, quote, 00:30:54.710 --> 00:30:59.520 align:middle line:84% "production endowments" to their ultimate consumptions. 00:30:59.520 --> 00:31:04.330 align:middle line:84% And most of the shocks are covered, not all of them. 00:31:04.330 --> 00:31:06.660 align:middle line:84% So they're not completely following 00:31:06.660 --> 00:31:10.620 align:middle line:84% the full risk-sharing program, but they definitely 00:31:10.620 --> 00:31:13.960 align:middle line:90% diversified mostly 90% of it. 00:31:13.960 --> 00:31:16.770 align:middle line:84% So then we got to particular health shocks, 00:31:16.770 --> 00:31:21.450 align:middle line:84% and specific Illnesses, and I showed you 00:31:21.450 --> 00:31:26.070 align:middle line:84% that again those illness shocks, not production shocks, per se, 00:31:26.070 --> 00:31:30.130 align:middle line:84% also covered by gifts largely, although not entirely. 00:31:30.130 --> 00:31:32.430 align:middle line:84% And when they were not covered, they 00:31:32.430 --> 00:31:34.920 align:middle line:84% spill over into the labor network 00:31:34.920 --> 00:31:37.710 align:middle line:84% with less hiring of employed labor, 00:31:37.710 --> 00:31:41.430 align:middle line:84% and also less purchasing of intermediate goods 00:31:41.430 --> 00:31:43.270 align:middle line:90% from other households. 00:31:43.270 --> 00:31:45.850 align:middle line:90% So this was documenting that. 00:31:45.850 --> 00:31:50.250 align:middle line:84% So this is about kinship groups and risk 00:31:50.250 --> 00:31:54.030 align:middle line:84% sharing in village villages in Thailand. 00:31:54.030 --> 00:31:56.940 align:middle line:84% It also was about an illness as a shock. 00:31:56.940 --> 00:32:00.420 align:middle line:84% It was not about COVID, although it was about illness. 00:32:00.420 --> 00:32:03.300 align:middle line:84% The illness was idiosyncratic and not aggregate. 00:32:03.300 --> 00:32:06.330 align:middle line:84% The closest we could get in the Thai data to an aggregate 00:32:06.330 --> 00:32:10.590 align:middle line:84% was watching those shrimp ponds suffer catastrophe 00:32:10.590 --> 00:32:12.570 align:middle line:90% in one of the four provinces. 00:32:12.570 --> 00:32:15.060 align:middle line:84% And because it becomes an aggregate shock rather than 00:32:15.060 --> 00:32:18.540 align:middle line:84% idiosyncratic shock, it's the gifts didn't come in as much, 00:32:18.540 --> 00:32:20.370 align:middle line:84% and it's very harder to diversify. 00:32:20.370 --> 00:32:22.860 align:middle line:84% So that's a theme I keep mentioning now 00:32:22.860 --> 00:32:24.960 align:middle line:90% for the last 10 or 15 minutes. 00:32:24.960 --> 00:32:27.960 align:middle line:84% Idiosyncratic stuff, in principle, you can diversify. 00:32:27.960 --> 00:32:30.540 align:middle line:90% Aggregate stuff you cannot. 00:32:30.540 --> 00:32:33.330 align:middle line:84% But the other thing I want to draw your attention to 00:32:33.330 --> 00:32:35.070 align:middle line:90% is this network stuff. 00:32:35.070 --> 00:32:39.510 align:middle line:84% So the issue is not just in Thailand or India, 00:32:39.510 --> 00:32:45.030 align:middle line:84% for that matter, but do these networks exist in the US. 00:32:45.030 --> 00:32:47.340 align:middle line:90% And what's going on with COVID? 00:32:47.340 --> 00:32:50.520 align:middle line:84% So that inter-state risk-sharing stuff 00:32:50.520 --> 00:32:54.480 align:middle line:84% suggests that networks are not necessarily local. 00:32:54.480 --> 00:32:56.220 align:middle line:90% They could be global. 00:32:56.220 --> 00:32:59.160 align:middle line:84% But your relatives or other family members 00:32:59.160 --> 00:33:01.890 align:middle line:84% are engaged in risk sharing when you 00:33:01.890 --> 00:33:04.620 align:middle line:90% experience a shock back home. 00:33:04.620 --> 00:33:08.940 align:middle line:84% And then this project in West Virginia, well Michael, 00:33:08.940 --> 00:33:10.440 align:middle line:84% I hope you don't mind mentioning it. 00:33:10.440 --> 00:33:14.070 align:middle line:84% But Michael, thinking about networks, 00:33:14.070 --> 00:33:16.440 align:middle line:84% wondered about networks in the US 00:33:16.440 --> 00:33:19.410 align:middle line:84% and the advantage of collecting our own data, 00:33:19.410 --> 00:33:22.800 align:middle line:84% indeed would be to better measure the networks 00:33:22.800 --> 00:33:25.500 align:middle line:84% among relatives, or friends, and so 00:33:25.500 --> 00:33:29.460 align:middle line:84% on, in towns in rural areas in West Virginia. 00:33:29.460 --> 00:33:32.820 align:middle line:84% So everything that we've been doing featuring Thailand, 00:33:32.820 --> 00:33:35.070 align:middle line:84% because we have all that data, could 00:33:35.070 --> 00:33:38.220 align:middle line:84% be done in other contexts like the US, 00:33:38.220 --> 00:33:40.470 align:middle line:90% if we can add that data. 00:33:40.470 --> 00:33:42.390 align:middle line:84% And we've definitely decided to try 00:33:42.390 --> 00:33:45.180 align:middle line:84% to measure these networks as a suggestion 00:33:45.180 --> 00:33:47.350 align:middle line:90% that Michael had from last time. 00:33:47.350 --> 00:33:47.850 align:middle line:90% All right. 00:33:47.850 --> 00:33:55.540 align:middle line:84% So any questions on this lecture 9? 00:33:55.540 --> 00:33:58.360 align:middle line:90% AUDIENCE: I have a question. 00:33:58.360 --> 00:34:06.460 align:middle line:84% In the slides 19 data, when we're decomposing the risk, 00:34:06.460 --> 00:34:08.650 align:middle line:90% yeah, here. 00:34:08.650 --> 00:34:12.969 align:middle line:84% I remember before we have R [INAUDIBLE] 00:34:12.969 --> 00:34:17.510 align:middle line:84% that was represented the risk-free return. 00:34:17.510 --> 00:34:28.420 align:middle line:84% But why here, we just have beta and the idiosyncratic? 00:34:28.420 --> 00:34:32.620 align:middle line:84% That is because we both subtract the alpha 00:34:32.620 --> 00:34:35.679 align:middle line:84% from left-hand side and the right-hand side, 00:34:35.679 --> 00:34:41.380 align:middle line:84% or because the results we get before 00:34:41.380 --> 00:34:48.673 align:middle line:84% that alpha is not significant, so we don't care about it? 00:34:48.673 --> 00:34:50.590 align:middle line:84% ROBERT TOWNSEND: Yeah, another great question. 00:34:50.590 --> 00:34:52.750 align:middle line:90% So there are two ways to do it. 00:34:52.750 --> 00:34:55.780 align:middle line:84% We should have been thinking about the rate of return 00:34:55.780 --> 00:34:58.600 align:middle line:90% minus the risk-free rate. 00:34:58.600 --> 00:35:01.420 align:middle line:84% But in fact, in the Thai data, we took the position 00:35:01.420 --> 00:35:05.350 align:middle line:90% that the risk-free rate was 0. 00:35:05.350 --> 00:35:09.260 align:middle line:84% So that's why we kind of dropped it. 00:35:09.260 --> 00:35:10.095 align:middle line:90% AUDIENCE: Thank you. 00:35:10.095 --> 00:35:11.720 align:middle line:84% ROBERT TOWNSEND: It's like having rice. 00:35:11.720 --> 00:35:14.000 align:middle line:90% You can put rice in storage. 00:35:14.000 --> 00:35:17.750 align:middle line:84% And you're going to get back more or less what you put in, 00:35:17.750 --> 00:35:18.860 align:middle line:90% et cetera. 00:35:18.860 --> 00:35:20.900 align:middle line:84% So that was the rationale for it. 00:35:20.900 --> 00:35:23.870 align:middle line:90% 00:35:23.870 --> 00:35:25.700 align:middle line:90% Other questions? 00:35:25.700 --> 00:35:26.870 align:middle line:90% OK. 00:35:26.870 --> 00:35:31.840 align:middle line:84% So that was all the stuff on risk sharing. 00:35:31.840 --> 00:35:36.490 align:middle line:84% Then in lecture 10, we started to introduce 00:35:36.490 --> 00:35:39.790 align:middle line:84% these ledgers, and financial accounts, 00:35:39.790 --> 00:35:42.400 align:middle line:90% and other measurement issues. 00:35:42.400 --> 00:35:45.340 align:middle line:84% Actually, there's a lot going on in this lecture, 00:35:45.340 --> 00:35:48.040 align:middle line:90% as I'll review with you. 00:35:48.040 --> 00:35:51.850 align:middle line:84% We did the wealth planner stuff, looking at the balance 00:35:51.850 --> 00:35:53.410 align:middle line:90% sheet and the income statement. 00:35:53.410 --> 00:35:55.810 align:middle line:84% Then I went to the other statement, 00:35:55.810 --> 00:35:57.640 align:middle line:90% the statement of cash flow. 00:35:57.640 --> 00:36:00.740 align:middle line:84% And we thought about Bitcoin as an example, 00:36:00.740 --> 00:36:03.820 align:middle line:84% having the balance sheet, and the cash flow in one place. 00:36:03.820 --> 00:36:07.330 align:middle line:84% And with Bitcoin, that motivated a review of distributed 00:36:07.330 --> 00:36:12.280 align:middle line:84% ledgers, and how it works with currency, physical currency, 00:36:12.280 --> 00:36:14.080 align:middle line:90% as opposed to the ledgers. 00:36:14.080 --> 00:36:16.660 align:middle line:84% And then we touched on Sweden and Kenya. 00:36:16.660 --> 00:36:18.790 align:middle line:84% So this lecture introduced Bitcoin. 00:36:18.790 --> 00:36:23.650 align:middle line:84% It also introduced distributed ledgers as financial accounts, 00:36:23.650 --> 00:36:26.320 align:middle line:90% and also value transfer. 00:36:26.320 --> 00:36:30.490 align:middle line:84% So we talked about this statement of income. 00:36:30.490 --> 00:36:34.310 align:middle line:84% By now, you may or not remember, it doesn't really matter. 00:36:34.310 --> 00:36:37.810 align:middle line:84% This household A appears twice subsequently. 00:36:37.810 --> 00:36:40.480 align:middle line:84% This was the first time we introduced it 00:36:40.480 --> 00:36:42.640 align:middle line:84% as an example of the income statement, 00:36:42.640 --> 00:36:46.460 align:middle line:84% and also household A's balance sheet. 00:36:46.460 --> 00:36:49.180 align:middle line:84% And then we started to talk about the wealth planner. 00:36:49.180 --> 00:36:52.180 align:middle line:84% So we implemented this in Thailand, 00:36:52.180 --> 00:36:55.990 align:middle line:84% maximize discount expected utility over the lifetime, 00:36:55.990 --> 00:36:59.590 align:middle line:84% even allowing for bequests to their kids, 00:36:59.590 --> 00:37:03.250 align:middle line:84% subject to the stocks and flows constraint, 00:37:03.250 --> 00:37:06.130 align:middle line:84% investing in risky and safe assets, 00:37:06.130 --> 00:37:10.510 align:middle line:84% and allowing for the risk sharing which you don't usually 00:37:10.510 --> 00:37:12.970 align:middle line:90% see in wealth planners. 00:37:12.970 --> 00:37:14.950 align:middle line:84% But we know we needed it for Thailand, 00:37:14.950 --> 00:37:18.310 align:middle line:84% because you've already seen how much risk sharing there is. 00:37:18.310 --> 00:37:20.080 align:middle line:84% So we want to smooth out the income 00:37:20.080 --> 00:37:23.890 align:middle line:84% profile consistent with the ins and outs of those gifts. 00:37:23.890 --> 00:37:27.760 align:middle line:84% And the residual is the household's exposure 00:37:27.760 --> 00:37:29.380 align:middle line:90% to their own income shocks. 00:37:29.380 --> 00:37:32.260 align:middle line:84% So we explained to those Thai households 00:37:32.260 --> 00:37:34.390 align:middle line:90% how the planner works. 00:37:34.390 --> 00:37:37.030 align:middle line:84% And again, to motivate this, think 00:37:37.030 --> 00:37:39.580 align:middle line:84% about going to the doorstep of a household living 00:37:39.580 --> 00:37:43.390 align:middle line:84% in West Virginia, having collected some of their data. 00:37:43.390 --> 00:37:46.000 align:middle line:84% Or better yet, what data do we want 00:37:46.000 --> 00:37:48.730 align:middle line:84% to collect in West Virginia consistent with running 00:37:48.730 --> 00:37:52.510 align:middle line:84% this wealth planner, so we can give them advice? 00:37:52.510 --> 00:37:55.820 align:middle line:84% The setting would be different, but the idea is the same. 00:37:55.820 --> 00:37:57.770 align:middle line:84% So with the wealth planner, we find out 00:37:57.770 --> 00:37:59.770 align:middle line:84% how much they're planning to give to their kids, 00:37:59.770 --> 00:38:00.670 align:middle line:90% if anything. 00:38:00.670 --> 00:38:03.910 align:middle line:84% We calibrated their income processes 00:38:03.910 --> 00:38:06.160 align:middle line:90% with trends and fluctuations. 00:38:06.160 --> 00:38:09.820 align:middle line:84% We looked at their committed consumption goals, 00:38:09.820 --> 00:38:13.420 align:middle line:84% and we're very careful about their initial wealth. 00:38:13.420 --> 00:38:15.790 align:middle line:84% Again, you come into the period with a certain amount 00:38:15.790 --> 00:38:19.240 align:middle line:84% of wealth, and you could consume it, or you could save it. 00:38:19.240 --> 00:38:21.370 align:middle line:84% You can save it and risky or safe assets. 00:38:21.370 --> 00:38:23.410 align:middle line:84% In principle, you could also borrow, 00:38:23.410 --> 00:38:26.570 align:middle line:84% and then that is going to determine wealth for next time. 00:38:26.570 --> 00:38:29.350 align:middle line:84% So I probably didn't emphasize it as much. 00:38:29.350 --> 00:38:31.420 align:middle line:84% But this is where the stocks and flows 00:38:31.420 --> 00:38:35.620 align:middle line:84% are coming together again early on, through this budget 00:38:35.620 --> 00:38:36.830 align:middle line:90% constraint. 00:38:36.830 --> 00:38:41.350 align:middle line:84% Initial wealth and consumption is subtracted off, 00:38:41.350 --> 00:38:45.040 align:middle line:84% and the remainder goes into risky and safe assets, which 00:38:45.040 --> 00:38:47.410 align:middle line:90% will determine wealth next time. 00:38:47.410 --> 00:38:49.810 align:middle line:84% And then we looked at actual flows 00:38:49.810 --> 00:38:53.230 align:middle line:84% in the data, the second household, 00:38:53.230 --> 00:38:57.310 align:middle line:84% the recommendation of the planner in terms of life cycle 00:38:57.310 --> 00:39:00.370 align:middle line:84% which is saving up enough when you're 00:39:00.370 --> 00:39:06.210 align:middle line:84% relatively young to be able to survive through older ages. 00:39:06.210 --> 00:39:08.350 align:middle line:84% Some of these households get in trouble. 00:39:08.350 --> 00:39:13.440 align:middle line:84% So we try to help this guy, giving advice and actually 00:39:13.440 --> 00:39:15.960 align:middle line:84% talking about the realism of the goals. 00:39:15.960 --> 00:39:19.260 align:middle line:84% And I made the point that this planner exists for the US. 00:39:19.260 --> 00:39:21.870 align:middle line:84% Although I don't know that it's particularly well designed 00:39:21.870 --> 00:39:24.120 align:middle line:84% for households in West Virginia, it 00:39:24.120 --> 00:39:28.050 align:middle line:84% does take into account details of the US tax code, 00:39:28.050 --> 00:39:32.310 align:middle line:84% and Social Security, and so on, but acting as if there's 00:39:32.310 --> 00:39:34.740 align:middle line:90% no networks, no risk sharing. 00:39:34.740 --> 00:39:35.760 align:middle line:90% Who knows? 00:39:35.760 --> 00:39:37.930 align:middle line:90% We'll eventually find out. 00:39:37.930 --> 00:39:42.000 align:middle line:84% Then I made this transition to Bitcoin as an account. 00:39:42.000 --> 00:39:45.150 align:middle line:84% So Bitcoin accounts have the ownership of the coin, 00:39:45.150 --> 00:39:47.890 align:middle line:84% and also a way to buy something else 00:39:47.890 --> 00:39:50.670 align:middle line:84% and transfer the coin to a third party, 00:39:50.670 --> 00:39:55.300 align:middle line:84% by sending the proposed transection to be validated. 00:39:55.300 --> 00:39:59.550 align:middle line:84% So the Bitcoin ledgers are a combination 00:39:59.550 --> 00:40:03.160 align:middle line:84% of the balance sheet and the statement of cash flow. 00:40:03.160 --> 00:40:06.060 align:middle line:84% And with the introduction of the idea of cash flow, 00:40:06.060 --> 00:40:08.460 align:middle line:84% I went back to the Thai financial accounts 00:40:08.460 --> 00:40:10.365 align:middle line:84% to look at that for the first time. 00:40:10.365 --> 00:40:12.490 align:middle line:84% We'd already looked at the income and balance sheet 00:40:12.490 --> 00:40:12.990 align:middle line:90% statement. 00:40:12.990 --> 00:40:15.480 align:middle line:84% Now we're looking at the cash flow statement. 00:40:15.480 --> 00:40:18.720 align:middle line:84% Cash flow from production, consumption, financing, 00:40:18.720 --> 00:40:20.850 align:middle line:90% and so on. 00:40:20.850 --> 00:40:24.360 align:middle line:84% So then the issue is, what do the ledgers do? 00:40:24.360 --> 00:40:29.180 align:middle line:84% Well, in Thailand, they're largely using paper currency. 00:40:29.180 --> 00:40:31.370 align:middle line:90% So they do not keep accounts. 00:40:31.370 --> 00:40:33.620 align:middle line:90% And there are no common ledgers. 00:40:33.620 --> 00:40:37.760 align:middle line:84% Bitcoin is all on an account, and there are common ledgers. 00:40:37.760 --> 00:40:41.030 align:middle line:84% That's why it's called a distributed common ledger. 00:40:41.030 --> 00:40:44.780 align:middle line:84% So we actually went to create ledgers 00:40:44.780 --> 00:40:46.820 align:middle line:84% that would be common in the Thai data, 00:40:46.820 --> 00:40:49.160 align:middle line:84% and discovered a lot of data discrepancies 00:40:49.160 --> 00:40:52.040 align:middle line:84% because we hadn't collected the data that way 00:40:52.040 --> 00:40:53.040 align:middle line:90% in the first place. 00:40:53.040 --> 00:40:55.100 align:middle line:84% This is also a reminder, if we're 00:40:55.100 --> 00:40:58.220 align:middle line:84% collecting data in West Virginia and we're 00:40:58.220 --> 00:41:01.040 align:middle line:84% getting the identity of the transacting partner 00:41:01.040 --> 00:41:04.040 align:middle line:84% when we interview that partner, those transactions 00:41:04.040 --> 00:41:05.270 align:middle line:90% should also show up. 00:41:05.270 --> 00:41:08.030 align:middle line:84% So ideally one would do that in real time 00:41:08.030 --> 00:41:12.050 align:middle line:84% almost, when memory is fresh, or for that 00:41:12.050 --> 00:41:14.540 align:middle line:84% matter their understanding of a transection 00:41:14.540 --> 00:41:17.240 align:middle line:84% with a commercial bank which is opposed to what 00:41:17.240 --> 00:41:18.350 align:middle line:90% commercial bank is doing. 00:41:18.350 --> 00:41:21.050 align:middle line:84% At least you think that, of course, 00:41:21.050 --> 00:41:22.640 align:middle line:90% the commercial banks are honest. 00:41:22.640 --> 00:41:26.600 align:middle line:84% Just remember Wells Fargo that actually created accounts 00:41:26.600 --> 00:41:28.970 align:middle line:84% for households they didn't know about, 00:41:28.970 --> 00:41:31.910 align:middle line:84% and then charged them money for it. 00:41:31.910 --> 00:41:36.170 align:middle line:84% So we then talked about the advantage of Bitcoin 00:41:36.170 --> 00:41:40.760 align:middle line:84% like distributed ledgers, namely data reconciliation. 00:41:40.760 --> 00:41:44.210 align:middle line:84% But in practice, of course, they're using paper currency. 00:41:44.210 --> 00:41:46.550 align:middle line:84% So with that opportunity, we looked 00:41:46.550 --> 00:41:50.490 align:middle line:84% at another problem of management, 00:41:50.490 --> 00:41:55.370 align:middle line:84% which is managing the currency, managing the cash flow, which 00:41:55.370 --> 00:41:58.340 align:middle line:84% we modeled and fit to the data, as having trends 00:41:58.340 --> 00:42:00.680 align:middle line:90% and stochastic fluctuations. 00:42:00.680 --> 00:42:03.740 align:middle line:84% We have the evolution of cash flow, 00:42:03.740 --> 00:42:07.610 align:middle line:84% adjusted taking into account the ups and downs with shocks 00:42:07.610 --> 00:42:09.770 align:middle line:84% of revenues and expenses, and adjusted 00:42:09.770 --> 00:42:12.350 align:middle line:90% by withdrawals and deposits. 00:42:12.350 --> 00:42:14.900 align:middle line:84% The two costs were going to the bank 00:42:14.900 --> 00:42:17.570 align:middle line:84% to make the physical adjustment, and the opportunity 00:42:17.570 --> 00:42:20.210 align:middle line:84% cost, foregone interest, of holding 00:42:20.210 --> 00:42:22.250 align:middle line:90% the currency in the house. 00:42:22.250 --> 00:42:25.685 align:middle line:84% Which led to another optimization problem. 00:42:25.685 --> 00:42:29.360 align:middle line:84% Now minimizing the expected costs, 00:42:29.360 --> 00:42:32.420 align:middle line:84% let me just say that our vision for West Virginia 00:42:32.420 --> 00:42:38.090 align:middle line:84% is to combine the high frequency cash flow with the longer term 00:42:38.090 --> 00:42:39.230 align:middle line:90% life cycle. 00:42:39.230 --> 00:42:42.450 align:middle line:84% Because in practice they are intimately related. 00:42:42.450 --> 00:42:46.190 align:middle line:84% So we'll have a synthesis of those frameworks in mind. 00:42:46.190 --> 00:42:50.000 align:middle line:84% And in Thailand, they didn't manage the currency very well. 00:42:50.000 --> 00:42:52.970 align:middle line:84% In the rural areas, they're holding far too much of it. 00:42:52.970 --> 00:42:56.490 align:middle line:84% And in the urban areas, they actually did better. 00:42:56.490 --> 00:42:58.430 align:middle line:84% Of course, one advantage of getting 00:42:58.430 --> 00:43:01.580 align:middle line:84% them to switch from paper currency 00:43:01.580 --> 00:43:04.520 align:middle line:84% into electronic entries that is then in principle, 00:43:04.520 --> 00:43:09.740 align:middle line:84% it's easier to manage portfolios or to send reminders about cash 00:43:09.740 --> 00:43:15.350 align:middle line:84% management on apps on phones, if all the transactions 00:43:15.350 --> 00:43:16.430 align:middle line:90% are electronic. 00:43:16.430 --> 00:43:19.550 align:middle line:84% And I give you some examples of Yodlee 00:43:19.550 --> 00:43:25.830 align:middle line:84% and other apps that do this, that collect electronic data. 00:43:25.830 --> 00:43:29.720 align:middle line:84% So anyway, that led finally to, well, 00:43:29.720 --> 00:43:31.670 align:middle line:84% do you have to use paper currency? 00:43:31.670 --> 00:43:33.320 align:middle line:90% The answer being no. 00:43:33.320 --> 00:43:36.440 align:middle line:84% Sweden doesn't much anymore, although they're 00:43:36.440 --> 00:43:38.870 align:middle line:84% worried about it in the sense of worried 00:43:38.870 --> 00:43:42.500 align:middle line:84% that the private sector takes over all of the payments, 00:43:42.500 --> 00:43:46.310 align:middle line:84% and the Riksbank isn't needed anymore, the central bank. 00:43:46.310 --> 00:43:49.220 align:middle line:84% And the stuff in between, which was Kenya, 00:43:49.220 --> 00:43:54.830 align:middle line:84% so Kenya eMoney, digital money, is created by the cell phone 00:43:54.830 --> 00:43:57.770 align:middle line:84% company, allowing people to cash in and cash 00:43:57.770 --> 00:44:01.970 align:middle line:84% out, and electronically transfer the credits on the cell 00:44:01.970 --> 00:44:05.390 align:middle line:84% account, which turned out to help with guess what? 00:44:05.390 --> 00:44:06.050 align:middle line:90% Risk sharing. 00:44:06.050 --> 00:44:10.070 align:middle line:84% So you see a tie-in here back to those risk-sharing lectures. 00:44:10.070 --> 00:44:12.740 align:middle line:84% In this case, it's in the context of Kenya, 00:44:12.740 --> 00:44:17.210 align:middle line:84% and the fact that your kin back in the village 00:44:17.210 --> 00:44:20.060 align:middle line:84% have had a terrible harvest can be 00:44:20.060 --> 00:44:23.660 align:middle line:84% mitigated by having the migrant worker in Nairobi 00:44:23.660 --> 00:44:24.680 align:middle line:90% send more money. 00:44:24.680 --> 00:44:26.900 align:middle line:84% And evidently, that is exactly what 00:44:26.900 --> 00:44:31.580 align:middle line:84% happens, as opposed to having to give the Kenyan shillings, 00:44:31.580 --> 00:44:33.500 align:middle line:90% paper currency, to a friend. 00:44:33.500 --> 00:44:35.600 align:middle line:84% Put it in an envelope, hope for the best, 00:44:35.600 --> 00:44:41.270 align:middle line:84% or in a very costly way go back home and hand over the cash. 00:44:41.270 --> 00:44:44.090 align:middle line:84% So risk sharing on the ledgers actually helped. 00:44:44.090 --> 00:44:46.190 align:middle line:90% But these are private accounts. 00:44:46.190 --> 00:44:50.000 align:middle line:84% So they don't have what we typically 00:44:50.000 --> 00:44:52.130 align:middle line:84% refer to as distributed ledgers, where 00:44:52.130 --> 00:44:54.390 align:middle line:90% there's third-party validation. 00:44:54.390 --> 00:44:57.500 align:middle line:84% All of these transactions, cell phone companies accounts, 00:44:57.500 --> 00:45:00.290 align:middle line:90% are trusted to be accurate. 00:45:00.290 --> 00:45:02.450 align:middle line:84% Households can check them, when they 00:45:02.450 --> 00:45:05.150 align:middle line:84% happen to see if they get credited 00:45:05.150 --> 00:45:06.710 align:middle line:90% the way they're supposed to. 00:45:06.710 --> 00:45:11.030 align:middle line:84% But other than that, Safaricom could steal the money 00:45:11.030 --> 00:45:12.290 align:middle line:90% if they wanted to. 00:45:12.290 --> 00:45:13.820 align:middle line:90% They don't, evidently. 00:45:13.820 --> 00:45:16.010 align:middle line:84% They put it in the bank, but still, 00:45:16.010 --> 00:45:18.090 align:middle line:84% and the banks are subject to runs. 00:45:18.090 --> 00:45:20.280 align:middle line:84% So there's a lot of risk remaining, 00:45:20.280 --> 00:45:25.980 align:middle line:84% even though this digital currency is an innovation 00:45:25.980 --> 00:45:27.150 align:middle line:90% in the context of Kenya. 00:45:27.150 --> 00:45:29.640 align:middle line:84% It certainly would not be in the context of Sweden. 00:45:29.640 --> 00:45:32.380 align:middle line:84% They have their own interbank method. 00:45:32.380 --> 00:45:35.040 align:middle line:84% So are there questions about this lecture 10? 00:45:35.040 --> 00:45:38.570 align:middle line:90% 00:45:38.570 --> 00:45:40.530 align:middle line:90% Again, it's kind of a mouthful. 00:45:40.530 --> 00:45:43.860 align:middle line:84% This lecture does both the financial accounts, 00:45:43.860 --> 00:45:46.910 align:middle line:84% income statement, balance sheet, and cash flow, 00:45:46.910 --> 00:45:51.470 align:middle line:84% and is also looking at two problems namely cash management 00:45:51.470 --> 00:45:55.340 align:middle line:84% and lifecycle management, and finally merging 00:45:55.340 --> 00:45:58.820 align:middle line:84% into a discussion of what distributed ledgers, Bitcoin 00:45:58.820 --> 00:46:01.410 align:middle line:84% and distributed ledgers, do or don't do, 00:46:01.410 --> 00:46:03.680 align:middle line:90% and potential substitutes. 00:46:03.680 --> 00:46:08.390 align:middle line:84% How good or how bad is currency and Safaricom digital credits, 00:46:08.390 --> 00:46:10.820 align:middle line:90% relative to validated systems? 00:46:10.820 --> 00:46:12.980 align:middle line:84% Although we haven't really talked about validation. 00:46:12.980 --> 00:46:16.530 align:middle line:84% That happens I think in the next lecture. 00:46:16.530 --> 00:46:19.190 align:middle line:90% So that was lecture 10. 00:46:19.190 --> 00:46:23.900 align:middle line:84% Lecture 11 was about contracts and mechanism design. 00:46:23.900 --> 00:46:26.930 align:middle line:90% And I'll just give an overview. 00:46:26.930 --> 00:46:29.940 align:middle line:84% There were two parts to this lecture. 00:46:29.940 --> 00:46:32.060 align:middle line:84% The first part is figuring out what 00:46:32.060 --> 00:46:34.550 align:middle line:84% to do when there's private information. 00:46:34.550 --> 00:46:37.880 align:middle line:84% And the main conclusion was there is a way 00:46:37.880 --> 00:46:41.570 align:middle line:84% to write down additional constraints, called truth 00:46:41.570 --> 00:46:43.550 align:middle line:84% telling or incentive constraints, 00:46:43.550 --> 00:46:47.130 align:middle line:84% that capture the problem of private information. 00:46:47.130 --> 00:46:50.240 align:middle line:84% And we will be able to append those 00:46:50.240 --> 00:46:53.600 align:middle line:84% on top of the usual programming problems 00:46:53.600 --> 00:46:56.875 align:middle line:84% that maximize the lambda weighted sums of utilities. 00:46:56.875 --> 00:46:58.250 align:middle line:84% So we've got a tool, and the tool 00:46:58.250 --> 00:47:00.110 align:middle line:84% is getting better and better, because we 00:47:00.110 --> 00:47:01.890 align:middle line:90% can do more and more with it. 00:47:01.890 --> 00:47:03.860 align:middle line:84% So that was one of the main points of this. 00:47:03.860 --> 00:47:06.650 align:middle line:84% And the other was to think about mechanism 00:47:06.650 --> 00:47:11.300 align:middle line:84% design, private contracts, having to do with messages, 00:47:11.300 --> 00:47:14.210 align:middle line:84% and then back to the ledgers which 00:47:14.210 --> 00:47:16.940 align:middle line:84% have to do with the cryptography part of ledgers, 00:47:16.940 --> 00:47:20.180 align:middle line:84% and keeping messages secret, and whether or not they 00:47:20.180 --> 00:47:21.780 align:middle line:90% have to be validated. 00:47:21.780 --> 00:47:24.740 align:middle line:84% So the motivation for the first part 00:47:24.740 --> 00:47:27.830 align:middle line:84% was the fact that some of these villages 00:47:27.830 --> 00:47:31.700 align:middle line:84% lie far away from the central monastery. 00:47:31.700 --> 00:47:33.440 align:middle line:90% Some are very near. 00:47:33.440 --> 00:47:36.680 align:middle line:84% Some households work as serfs, slaves. 00:47:36.680 --> 00:47:38.910 align:middle line:90% Others far away pay rent. 00:47:38.910 --> 00:47:43.520 align:middle line:84% So we wanted to explain this kind of variation 00:47:43.520 --> 00:47:46.800 align:middle line:84% in the organizations that we see. 00:47:46.800 --> 00:47:50.960 align:middle line:84% So we introduced a simple two-agent model, where only one 00:47:50.960 --> 00:47:56.570 align:middle line:84% of the two, agent 1, had some privately observed endowment. 00:47:56.570 --> 00:47:59.420 align:middle line:84% Agent 2 was the villa or the monastery. 00:47:59.420 --> 00:48:03.170 align:middle line:84% And we wanted to know what would be an optimal resource 00:48:03.170 --> 00:48:06.140 align:middle line:84% allocation mechanism, granted that agent 1 has 00:48:06.140 --> 00:48:07.470 align:middle line:90% private information. 00:48:07.470 --> 00:48:13.130 align:middle line:84% So we did some derivation, had agent 1 sending messages, 00:48:13.130 --> 00:48:15.950 align:middle line:84% looked at what would be a maximizing message, 00:48:15.950 --> 00:48:18.860 align:middle line:84% derived a series of inequalities, 00:48:18.860 --> 00:48:21.920 align:middle line:84% and then changed the problems to sending messages 00:48:21.920 --> 00:48:25.010 align:middle line:84% about the underlying endowment, and ended up 00:48:25.010 --> 00:48:28.100 align:middle line:84% with something we call the truth telling constraint, which 00:48:28.100 --> 00:48:31.040 align:middle line:84% captures all of the problems associated 00:48:31.040 --> 00:48:34.170 align:middle line:84% with private information in this context. 00:48:34.170 --> 00:48:37.640 align:middle line:84% So then we're able to solve for optimal resource allocations 00:48:37.640 --> 00:48:42.290 align:middle line:84% just by adding on these kinds of incentive constraints, 00:48:42.290 --> 00:48:45.080 align:middle line:84% in addition to maximizing the lambda weighted sums 00:48:45.080 --> 00:48:46.250 align:middle line:90% of utilities. 00:48:46.250 --> 00:48:47.960 align:middle line:90% We did this in levels. 00:48:47.960 --> 00:48:51.140 align:middle line:84% Then we went to lotteries to make it more general. 00:48:51.140 --> 00:48:53.210 align:middle line:84% We had a good discussion at the beginning of one 00:48:53.210 --> 00:48:56.390 align:middle line:84% of the classes of the advantages of these lotteries, which 00:48:56.390 --> 00:49:00.440 align:middle line:84% we had talked about in the first one third of the class as well. 00:49:00.440 --> 00:49:04.250 align:middle line:84% Oh, and at some point, there was a starred article 00:49:04.250 --> 00:49:07.130 align:middle line:84% on the reading list, which was how to actually 00:49:07.130 --> 00:49:08.970 align:middle line:90% code these things up. 00:49:08.970 --> 00:49:11.060 align:middle line:84% It's more related to the next lecture. 00:49:11.060 --> 00:49:15.540 align:middle line:84% But anyway, trying not to be abstract about it. 00:49:15.540 --> 00:49:17.810 align:middle line:84% So then we went from one period to two periods. 00:49:17.810 --> 00:49:20.150 align:middle line:84% We solved the two-period problem, 00:49:20.150 --> 00:49:23.030 align:middle line:84% and had incentive constraints, truth telling 00:49:23.030 --> 00:49:24.920 align:middle line:90% constraints for both periods. 00:49:24.920 --> 00:49:27.890 align:middle line:84% In this case, what I was emphasizing 00:49:27.890 --> 00:49:30.470 align:middle line:84% both the dynamic program and how to work back 00:49:30.470 --> 00:49:33.500 align:middle line:84% from the second period to the first period sequentially, 00:49:33.500 --> 00:49:36.950 align:middle line:84% but also the fact that moving forward from the first 00:49:36.950 --> 00:49:40.370 align:middle line:84% to the second, the message is about underlying 00:49:40.370 --> 00:49:43.970 align:middle line:84% unobserved states would be key to implementing 00:49:43.970 --> 00:49:47.400 align:middle line:84% the optimal information constrained solution. 00:49:47.400 --> 00:49:50.120 align:middle line:84% And by emphasizing messages, I was already 00:49:50.120 --> 00:49:53.480 align:middle line:84% anticipating contracts, smart contracts, 00:49:53.480 --> 00:49:58.190 align:middle line:84% written on ledgers where messages replace transactions. 00:49:58.190 --> 00:50:01.250 align:middle line:84% And the optimal solution to that two-period problem 00:50:01.250 --> 00:50:04.550 align:middle line:84% was neither full insurance which wouldn't be feasible, 00:50:04.550 --> 00:50:08.060 align:middle line:84% nor limited borrowing and lending which was not optimal. 00:50:08.060 --> 00:50:10.910 align:middle line:84% And again, in class, we had a very good discussion 00:50:10.910 --> 00:50:14.390 align:middle line:84% about this hybrid borrowing and lending scheme. 00:50:14.390 --> 00:50:16.730 align:middle line:84% We did costly state verification. 00:50:16.730 --> 00:50:21.080 align:middle line:84% I won't dwell on that, except that over a range of shocks, 00:50:21.080 --> 00:50:24.410 align:middle line:84% you don't need to be verifying anything at all. 00:50:24.410 --> 00:50:26.690 align:middle line:84% But it's still effective, because there's the threat 00:50:26.690 --> 00:50:30.560 align:middle line:90% of an audit for low claims. 00:50:30.560 --> 00:50:35.720 align:middle line:84% And we went back and argued that costly state verification 00:50:35.720 --> 00:50:38.810 align:middle line:84% can help explain what we saw in the initial part 00:50:38.810 --> 00:50:39.720 align:middle line:90% of the lecture. 00:50:39.720 --> 00:50:43.760 align:middle line:84% So here is where the transition happened to distributed ledgers 00:50:43.760 --> 00:50:44.810 align:middle line:90% again. 00:50:44.810 --> 00:50:48.080 align:middle line:84% And there was a lot of material here. 00:50:48.080 --> 00:50:53.030 align:middle line:84% I kind of argued that encryption was one third key element 00:50:53.030 --> 00:50:54.470 align:middle line:90% of distributed ledgers. 00:50:54.470 --> 00:50:56.520 align:middle line:90% We've seen it historically. 00:50:56.520 --> 00:50:59.930 align:middle line:84% We see it in contemporary applications. 00:50:59.930 --> 00:51:02.510 align:middle line:84% It really breaks down into encryption, 00:51:02.510 --> 00:51:04.400 align:middle line:84% of incoming and outgoing messages, 00:51:04.400 --> 00:51:08.390 align:middle line:84% of hashing of data and others so that it's uninterpretable 00:51:08.390 --> 00:51:12.380 align:middle line:84% unless one wants to prove that it was the original, which 00:51:12.380 --> 00:51:15.980 align:middle line:84% is possible to do in crypto graphic puzzles, which 00:51:15.980 --> 00:51:18.650 align:middle line:84% are hard, but not impossible to solve. 00:51:18.650 --> 00:51:22.710 align:middle line:84% Bitcoin is using various of those things, 00:51:22.710 --> 00:51:27.650 align:middle line:84% including a way of organizing the data through mining 00:51:27.650 --> 00:51:28.790 align:middle line:90% as the validation. 00:51:28.790 --> 00:51:33.140 align:middle line:84% And I emphasized that mining was not the only way 00:51:33.140 --> 00:51:37.130 align:middle line:84% to validate transactions on ledgers, if you don't trust 00:51:37.130 --> 00:51:40.100 align:middle line:90% Safaricom or some third party. 00:51:40.100 --> 00:51:43.080 align:middle line:84% And I also emphasized the encryption part, 00:51:43.080 --> 00:51:47.060 align:middle line:84% which Bitcoin doesn't use, homomorphic and that's 00:51:47.060 --> 00:51:48.260 align:middle line:90% not quite fair. 00:51:48.260 --> 00:51:50.090 align:middle line:84% The encryption part for the messages 00:51:50.090 --> 00:51:52.880 align:middle line:84% is related to homomorphic encryption. 00:51:52.880 --> 00:51:56.550 align:middle line:84% But Bitcoin doesn't use multiparty computation. 00:51:56.550 --> 00:52:00.080 align:middle line:84% So then we came back to the essence of the lecture, which 00:52:00.080 --> 00:52:03.350 align:middle line:84% was, what is the difference between validation 00:52:03.350 --> 00:52:07.700 align:middle line:84% on distributed ledgers and incentives and mechanism 00:52:07.700 --> 00:52:11.620 align:middle line:84% design, and the different philosophies of economics 00:52:11.620 --> 00:52:14.030 align:middle line:90% versus computer science? 00:52:14.030 --> 00:52:16.810 align:middle line:84% Computer science assumes that most nodes are 00:52:16.810 --> 00:52:22.720 align:middle line:84% honest and reliable, and designs algorithms on the assumption 00:52:22.720 --> 00:52:26.380 align:middle line:84% that the honest nodes will follow the algorithms, 00:52:26.380 --> 00:52:29.320 align:middle line:84% and just have to have enough total nodes 00:52:29.320 --> 00:52:32.560 align:middle line:84% relative to the faulty nodes in the population 00:52:32.560 --> 00:52:34.010 align:middle line:90% for those protocols to work. 00:52:34.010 --> 00:52:37.120 align:middle line:84% But if you take the mechanism design point of view, 00:52:37.120 --> 00:52:40.390 align:middle line:84% the protocols may not be incentive compatible. 00:52:40.390 --> 00:52:44.650 align:middle line:84% Economists assume everyone's very selfish and maximizing. 00:52:44.650 --> 00:52:47.470 align:middle line:84% So even the miners in Bitcoin have 00:52:47.470 --> 00:52:51.190 align:middle line:84% to have incentives to follow the prescribed protocol. 00:52:51.190 --> 00:52:53.560 align:middle line:84% And that point is true more generally. 00:52:53.560 --> 00:52:57.580 align:middle line:84% So that kind of led to the idea that why not separate the two 00:52:57.580 --> 00:53:02.500 align:middle line:84% things, write down the contract or mechanism design 00:53:02.500 --> 00:53:08.080 align:middle line:84% problem in code as in a smart contract, 00:53:08.080 --> 00:53:10.120 align:middle line:84% let the parties agree to it and make sure 00:53:10.120 --> 00:53:13.250 align:middle line:84% that it functions the way they think that it does, 00:53:13.250 --> 00:53:16.690 align:middle line:84% and then send messages to implement it? 00:53:16.690 --> 00:53:22.420 align:middle line:84% Those messages would be encrypted through HE and MPC. 00:53:22.420 --> 00:53:24.010 align:middle line:90% And that's a contribution. 00:53:24.010 --> 00:53:27.700 align:middle line:84% The mechanism design literature assumes we have a planner. 00:53:27.700 --> 00:53:29.440 align:middle line:90% I hate that language. 00:53:29.440 --> 00:53:33.430 align:middle line:84% And the planner is trusted to implement the mechanism. 00:53:33.430 --> 00:53:36.340 align:middle line:84% What encryption does nowadays is to get rid 00:53:36.340 --> 00:53:39.160 align:middle line:84% of trusted third parties and planners. 00:53:39.160 --> 00:53:41.920 align:middle line:84% Agents can keep secrets from each other 00:53:41.920 --> 00:53:45.370 align:middle line:84% and implement the optimal contract. 00:53:45.370 --> 00:53:47.210 align:middle line:90% The ledgers still play a role. 00:53:47.210 --> 00:53:49.600 align:middle line:84% I mean as in borrowing and lending, 00:53:49.600 --> 00:53:54.220 align:middle line:84% you can sequester the collateral and have it automatically 00:53:54.220 --> 00:53:59.700 align:middle line:84% transferred to the lender if the borrower were to default. 00:53:59.700 --> 00:54:02.830 align:middle line:84% So we're not claiming that we only need the cryptography. 00:54:02.830 --> 00:54:05.420 align:middle line:90% We also need the ledgers. 00:54:05.420 --> 00:54:07.190 align:middle line:84% On the other hand, you don't need 00:54:07.190 --> 00:54:10.670 align:middle line:84% to have all the line items of the code in the ledger being 00:54:10.670 --> 00:54:14.120 align:middle line:84% done, as with Ethereum, on a distributed ledger, 00:54:14.120 --> 00:54:16.470 align:middle line:84% because that's really, really costly. 00:54:16.470 --> 00:54:19.100 align:middle line:84% So again, there was a lot in this lecture. 00:54:19.100 --> 00:54:22.760 align:middle line:90% Questions about lecture 11? 00:54:22.760 --> 00:54:27.350 align:middle line:84% Lecture 12 was the application of contract theory. 00:54:27.350 --> 00:54:30.410 align:middle line:84% And we took it to the Thai villages, 00:54:30.410 --> 00:54:33.080 align:middle line:90% and looked at occupation choice. 00:54:33.080 --> 00:54:34.950 align:middle line:84% There's a couple of things going on here. 00:54:34.950 --> 00:54:37.760 align:middle line:84% One, we now know how to incorporate 00:54:37.760 --> 00:54:41.960 align:middle line:84% private information, and incentives, and collateral 00:54:41.960 --> 00:54:45.810 align:middle line:84% by just appending extra constraints onto the problem. 00:54:45.810 --> 00:54:51.350 align:middle line:84% So let me start skipping a little bit. 00:54:51.350 --> 00:54:54.170 align:middle line:84% Will the entrepreneur take appropriate action? 00:54:54.170 --> 00:54:55.130 align:middle line:90% Not necessarily. 00:54:55.130 --> 00:54:57.800 align:middle line:84% If it's not seen, then you have to induce it. 00:54:57.800 --> 00:55:00.710 align:middle line:84% Will the entrepreneur pay back the loan? 00:55:00.710 --> 00:55:02.990 align:middle line:84% If not, you have to have collateral, which 00:55:02.990 --> 00:55:06.860 align:middle line:90% is another kind of constraint. 00:55:06.860 --> 00:55:10.910 align:middle line:84% So here we have the household's utility in consumption, 00:55:10.910 --> 00:55:14.000 align:middle line:84% and a disutility of effort, and potentially 00:55:14.000 --> 00:55:16.460 align:middle line:84% risk neutral, potentially risk averse. 00:55:16.460 --> 00:55:21.060 align:middle line:84% And households vary by talent, wealth, and schooling. 00:55:21.060 --> 00:55:24.380 align:middle line:84% And we have they're running a firm with stochastic output 00:55:24.380 --> 00:55:27.320 align:middle line:84% as a function of effort and capital. 00:55:27.320 --> 00:55:29.510 align:middle line:90% And they can borrow. 00:55:29.510 --> 00:55:32.540 align:middle line:84% This is the slide I was kind of racing to get to. 00:55:32.540 --> 00:55:36.470 align:middle line:84% So here is the problem, which we posed 00:55:36.470 --> 00:55:40.070 align:middle line:84% as maximizing the utility of the borrower 00:55:40.070 --> 00:55:45.530 align:middle line:84% or entrepreneur, subject to the bank making zero profits. 00:55:45.530 --> 00:55:49.670 align:middle line:84% So it's worth emphasizing that in the previous lecture, 00:55:49.670 --> 00:55:53.210 align:middle line:84% as we've been maximizing lambda weighted sums of utilities 00:55:53.210 --> 00:55:55.880 align:middle line:84% subject to resource constraints, and incentive 00:55:55.880 --> 00:55:57.320 align:middle line:90% constraints, and so on. 00:55:57.320 --> 00:55:59.780 align:middle line:84% Well, here we made a shift from maximizing 00:55:59.780 --> 00:56:02.870 align:middle line:84% the utility of one party, subject to a fixed utility 00:56:02.870 --> 00:56:03.500 align:middle line:90% of the other. 00:56:03.500 --> 00:56:05.900 align:middle line:84% Or here, the utility of the entrepreneur 00:56:05.900 --> 00:56:09.920 align:middle line:84% subject to a fixed and zero profit level of the bank. 00:56:09.920 --> 00:56:12.050 align:middle line:84% One could have solved this problem the other way 00:56:12.050 --> 00:56:15.410 align:middle line:84% around, to maximize the profits of the bank subject 00:56:15.410 --> 00:56:20.040 align:middle line:84% to a fixed reservation utility on the part of the borrower. 00:56:20.040 --> 00:56:21.860 align:middle line:90% You should get the same answers. 00:56:21.860 --> 00:56:24.210 align:middle line:84% Of course, there are many, many solutions-- 00:56:24.210 --> 00:56:26.570 align:middle line:84% if you allow profits to be positive, 00:56:26.570 --> 00:56:31.190 align:middle line:84% if you allow some of the surplus to go to the bank, et cetera. 00:56:31.190 --> 00:56:34.220 align:middle line:84% In fact, the borrower could be pushed down to the point 00:56:34.220 --> 00:56:37.910 align:middle line:84% that they're almost indifferent to participating 00:56:37.910 --> 00:56:41.520 align:middle line:84% if the banks are getting the bulk of the surplus. 00:56:41.520 --> 00:56:45.110 align:middle line:84% But what I want to emphasize is it's all the same in the end. 00:56:45.110 --> 00:56:47.600 align:middle line:84% Maximizing lambda weighted sum of utilities 00:56:47.600 --> 00:56:50.630 align:middle line:84% is the same as maximizing the utility of one type, 00:56:50.630 --> 00:56:54.890 align:middle line:84% subject to fixed utility of the other, and then varying those. 00:56:54.890 --> 00:56:59.720 align:middle line:84% So the rest of it was just implementing the solution, 00:56:59.720 --> 00:57:04.430 align:middle line:84% limited liability, moral hazard, with separate equations. 00:57:04.430 --> 00:57:06.920 align:middle line:84% And then we got to this diagram where 00:57:06.920 --> 00:57:10.670 align:middle line:84% I said, if there were no constraints at all, that 00:57:10.670 --> 00:57:13.250 align:middle line:84% would be like the full information, full commitment 00:57:13.250 --> 00:57:18.260 align:middle line:84% solution, hit the bull's eye, it could be entirely moral hazard 00:57:18.260 --> 00:57:19.610 align:middle line:90% and not limited liability. 00:57:19.610 --> 00:57:23.630 align:middle line:84% It could be entirely limited liability and not moral hazard. 00:57:23.630 --> 00:57:26.630 align:middle line:84% Although in this case, if you had both constraints, 00:57:26.630 --> 00:57:29.960 align:middle line:84% you have two equations and two unknowns, the capital 00:57:29.960 --> 00:57:31.470 align:middle line:90% and the effort level. 00:57:31.470 --> 00:57:33.890 align:middle line:84% So when those two constraints are satisfied, 00:57:33.890 --> 00:57:36.230 align:middle line:84% if they're both in the program, you 00:57:36.230 --> 00:57:39.140 align:middle line:84% get a solution that has nothing to do with utility 00:57:39.140 --> 00:57:42.350 align:middle line:84% maximization, because there's only one point in the input 00:57:42.350 --> 00:57:46.150 align:middle line:84% space where both constraints are satisfied. 00:57:46.150 --> 00:57:48.865 align:middle line:84% But in general, if you have only one or the other, 00:57:48.865 --> 00:57:54.190 align:middle line:84% you have both max utility and then subject to constraints. 00:57:54.190 --> 00:57:57.460 align:middle line:84% The rest of this lecture was about implementation, 00:57:57.460 --> 00:58:00.730 align:middle line:84% namely solving linear programming problems 00:58:00.730 --> 00:58:04.840 align:middle line:84% and trying to pick parameters in such a way that 00:58:04.840 --> 00:58:08.620 align:middle line:84% maximize the probability that the model is true, 00:58:08.620 --> 00:58:11.140 align:middle line:90% given what we see in the data. 00:58:11.140 --> 00:58:15.010 align:middle line:84% Maximize the probability that the data we see in reality 00:58:15.010 --> 00:58:18.600 align:middle line:84% could have come from the model by choosing 00:58:18.600 --> 00:58:19.860 align:middle line:90% the parameters of the model. 00:58:19.860 --> 00:58:22.740 align:middle line:90% 00:58:22.740 --> 00:58:24.600 align:middle line:90% That was the application. 00:58:24.600 --> 00:58:27.390 align:middle line:84% And then we had the findings that the constraints 00:58:27.390 --> 00:58:29.190 align:middle line:90% varied by region. 00:58:29.190 --> 00:58:30.855 align:middle line:84% Are the questions about this lecture? 00:58:30.855 --> 00:58:34.830 align:middle line:90% 00:58:34.830 --> 00:58:38.390 align:middle line:90% So lecture 13 is also loaded. 00:58:38.390 --> 00:58:40.580 align:middle line:90% Although it's more recent. 00:58:40.580 --> 00:58:44.180 align:middle line:84% I'm somewhat comforted, just like the first review session, 00:58:44.180 --> 00:58:47.300 align:middle line:84% by the time we get to 13 and then 14, 15, 00:58:47.300 --> 00:58:49.160 align:middle line:90% you've already seen 15 tonight. 00:58:49.160 --> 00:58:50.670 align:middle line:90% Today, we just reviewed it. 00:58:50.670 --> 00:58:53.420 align:middle line:84% So we're actually getting close to the end here. 00:58:53.420 --> 00:58:56.400 align:middle line:84% This lecture did two important things. 00:58:56.400 --> 00:58:59.510 align:middle line:84% One, the theory which was to define 00:58:59.510 --> 00:59:02.120 align:middle line:84% the notion of Walrasian equilibrium 00:59:02.120 --> 00:59:03.800 align:middle line:90% within and without transfers. 00:59:03.800 --> 00:59:05.720 align:middle line:90% And the second was to apply it. 00:59:05.720 --> 00:59:08.750 align:middle line:84% So we get general equilibrium solution 00:59:08.750 --> 00:59:12.470 align:middle line:84% in a small open economy with production. 00:59:12.470 --> 00:59:14.600 align:middle line:84% All of this slide had to do with, 00:59:14.600 --> 00:59:16.430 align:middle line:90% remember what an economy means. 00:59:16.430 --> 00:59:20.810 align:middle line:84% This is partly review of the first one third of the class-- 00:59:20.810 --> 00:59:23.210 align:middle line:84% numbers of households, numbers of firms, 00:59:23.210 --> 00:59:28.370 align:middle line:84% what consumers do, what firms do, aggregate resources. 00:59:28.370 --> 00:59:30.303 align:middle line:90% We went over private ownership. 00:59:30.303 --> 00:59:31.970 align:middle line:84% And again, we had a very good discussion 00:59:31.970 --> 00:59:34.160 align:middle line:84% at the beginning of the following class 00:59:34.160 --> 00:59:37.730 align:middle line:84% to review this slide in terms of individual endowments 00:59:37.730 --> 00:59:39.590 align:middle line:90% and ownership shares. 00:59:39.590 --> 00:59:42.230 align:middle line:84% And the key definition of Walrasian 00:59:42.230 --> 00:59:46.970 align:middle line:84% equilibrium with three items in it-- 00:59:46.970 --> 00:59:49.880 align:middle line:84% maximizing profits for firms, maximizing 00:59:49.880 --> 00:59:54.620 align:middle line:84% utility for households, and the feasibility of the allocations 00:59:54.620 --> 00:59:57.710 align:middle line:84% that result from those two optimization problems. 00:59:57.710 --> 01:00:01.070 align:middle line:84% The key being that prices are the things that 01:00:01.070 --> 01:00:03.350 align:middle line:90% allow the reconciliation. 01:00:03.350 --> 01:00:05.060 align:middle line:90% You can do partial equilibrium. 01:00:05.060 --> 01:00:05.870 align:middle line:90% We did. 01:00:05.870 --> 01:00:09.230 align:middle line:84% You can solve for profit maximization taking prices 01:00:09.230 --> 01:00:10.100 align:middle line:90% as given. 01:00:10.100 --> 01:00:12.260 align:middle line:84% You can solve for utility maximization, 01:00:12.260 --> 01:00:13.940 align:middle line:90% taking prices as given. 01:00:13.940 --> 01:00:16.490 align:middle line:84% Here in general equilibrium, we are only 01:00:16.490 --> 01:00:19.460 align:middle line:84% allowing certain prices where those decentralized 01:00:19.460 --> 01:00:22.070 align:middle line:84% optimization problems are consistent with each other 01:00:22.070 --> 01:00:26.840 align:middle line:84% in the sense that supply is equal to demand. 01:00:26.840 --> 01:00:28.790 align:middle line:84% Oh, and then this Edgeworth box, which 01:00:28.790 --> 01:00:33.320 align:middle line:84% is like economists' favorite environment, a sad statement, 01:00:33.320 --> 01:00:36.350 align:middle line:84% putting the two types of households on the axis 01:00:36.350 --> 01:00:39.560 align:middle line:84% and talking about what a competitive equilibrium would 01:00:39.560 --> 01:00:40.500 align:middle line:90% look like. 01:00:40.500 --> 01:00:44.210 align:middle line:84% And the price equilibrium with transfers and the welfare 01:00:44.210 --> 01:00:46.820 align:middle line:84% theorems, which so far I have only stated, 01:00:46.820 --> 01:00:49.130 align:middle line:84% although the last one third of the class 01:00:49.130 --> 01:00:52.790 align:middle line:84% next week will be focusing in part on these welfare 01:00:52.790 --> 01:00:54.480 align:middle line:90% theorems starting next week. 01:00:54.480 --> 01:00:57.620 align:middle line:84% So then we did production in the small open economy. 01:00:57.620 --> 01:01:00.750 align:middle line:84% And we again went through this algorithm pretty carefully. 01:01:00.750 --> 01:01:03.560 align:middle line:84% So I don't think I want to say too much about it, 01:01:03.560 --> 01:01:04.400 align:middle line:90% too, too much. 01:01:04.400 --> 01:01:08.150 align:middle line:84% But we simplify the problem down into firms 01:01:08.150 --> 01:01:12.080 align:middle line:84% hiring inputs and trying to find the factor prices that 01:01:12.080 --> 01:01:16.460 align:middle line:84% clear the input markets, taking out external prices of goods 01:01:16.460 --> 01:01:18.030 align:middle line:90% as given. 01:01:18.030 --> 01:01:20.330 align:middle line:84% So we had the firms maximization problem. 01:01:20.330 --> 01:01:23.210 align:middle line:84% We wrote down that the sum of the input demands 01:01:23.210 --> 01:01:26.900 align:middle line:84% equal the endowments, and reminded ourselves 01:01:26.900 --> 01:01:27.920 align:middle line:90% of Shepard's lemma. 01:01:27.920 --> 01:01:30.830 align:middle line:84% So we had a different statement of the input demands, 01:01:30.830 --> 01:01:35.360 align:middle line:84% which we could then solve, especially in the 2 by 2 model. 01:01:35.360 --> 01:01:39.350 align:middle line:84% Input demands are entirely a function of input price ratios. 01:01:39.350 --> 01:01:43.250 align:middle line:84% And with constant returns to scale, it's even easier. 01:01:43.250 --> 01:01:45.560 align:middle line:84% We can talk about cost minimization 01:01:45.560 --> 01:01:47.630 align:middle line:84% to achieve a unit of output, and then 01:01:47.630 --> 01:01:51.580 align:middle line:84% to scale up and down when we get to determining output. 01:01:51.580 --> 01:01:54.410 align:middle line:84% We talked about the intensity of goods, 01:01:54.410 --> 01:01:56.630 align:middle line:84% the factor intensity of the different goods 01:01:56.630 --> 01:01:58.040 align:middle line:90% with a definition. 01:01:58.040 --> 01:02:01.850 align:middle line:84% Here was the cost curves and their properties. 01:02:01.850 --> 01:02:04.460 align:middle line:84% Concave, we solve for the equilibrium factor 01:02:04.460 --> 01:02:08.690 align:middle line:84% prices, where both firms face common prices. 01:02:08.690 --> 01:02:11.420 align:middle line:84% Hence, there's only one point where 01:02:11.420 --> 01:02:13.250 align:middle line:90% that's true on this diagram. 01:02:13.250 --> 01:02:16.070 align:middle line:84% And then we went from that to production. 01:02:16.070 --> 01:02:18.350 align:middle line:84% From production, we were able to solve for the input 01:02:18.350 --> 01:02:20.120 align:middle line:90% quantities and the outputs. 01:02:20.120 --> 01:02:24.080 align:middle line:84% And then we got the external goods prices 01:02:24.080 --> 01:02:28.310 align:middle line:84% by considering the free trade equilibrium 01:02:28.310 --> 01:02:33.200 align:middle line:84% with two countries determining a price ratio moving from Autarky 01:02:33.200 --> 01:02:36.170 align:middle line:90% to international trade. 01:02:36.170 --> 01:02:38.150 align:middle line:90% So there were three theorems. 01:02:38.150 --> 01:02:40.740 align:middle line:84% We had this Heckscher-Ohlin theorem, 01:02:40.740 --> 01:02:43.040 align:middle line:84% which someone mentioned at the beginning of class 01:02:43.040 --> 01:02:46.040 align:middle line:84% today, that the returns to factors 01:02:46.040 --> 01:02:49.070 align:middle line:84% depend on whether the country is abundant in one 01:02:49.070 --> 01:02:51.140 align:middle line:90% factor versus another one. 01:02:51.140 --> 01:02:52.940 align:middle line:84% The Stopler-Samuelson, having to do 01:02:52.940 --> 01:02:55.790 align:middle line:84% with how factor prices are moving when 01:02:55.790 --> 01:03:00.400 align:middle line:90% we vary the price of outputs. 01:03:00.400 --> 01:03:03.430 align:middle line:90% And that's all. 01:03:03.430 --> 01:03:04.360 align:middle line:90% OK. 01:03:04.360 --> 01:03:06.055 align:middle line:84% Are there questions about this lecture? 01:03:06.055 --> 01:03:10.420 align:middle line:90% 01:03:10.420 --> 01:03:13.690 align:middle line:84% AUDIENCE: Here the HO theorem that it 01:03:13.690 --> 01:03:16.960 align:middle line:84% says that in the Autarky, the price 01:03:16.960 --> 01:03:21.760 align:middle line:84% of the capital intensive goods in the capital abundant country 01:03:21.760 --> 01:03:26.410 align:middle line:84% will be buy down, so here buy down means the relative price 01:03:26.410 --> 01:03:29.980 align:middle line:84% wars in this capital-abundant country 01:03:29.980 --> 01:03:34.360 align:middle line:84% was lower than another country, right? 01:03:34.360 --> 01:03:36.610 align:middle line:90% Not the absolute prices? 01:03:36.610 --> 01:03:43.400 align:middle line:90% 01:03:43.400 --> 01:03:46.680 align:middle line:84% ROBERT TOWNSEND: The good that is being produced, 01:03:46.680 --> 01:03:50.120 align:middle line:84% which is capital intensive, if produced 01:03:50.120 --> 01:03:55.250 align:middle line:84% in the capital abundant country, should be relatively cheap, 01:03:55.250 --> 01:03:59.850 align:middle line:84% because the other country is abundant in labor 01:03:59.850 --> 01:04:02.790 align:middle line:90% and scarce in capital. 01:04:02.790 --> 01:04:04.560 align:middle line:90% So-- 01:04:04.560 --> 01:04:07.800 align:middle line:84% AUDIENCE: Yeah my question is that these 01:04:07.800 --> 01:04:11.530 align:middle line:84% are cheap means that the relative price was cheaper. 01:04:11.530 --> 01:04:14.790 align:middle line:84% I mean in this capital-abundant country, 01:04:14.790 --> 01:04:18.840 align:middle line:84% the relative price of these capital intensive goods 01:04:18.840 --> 01:04:28.620 align:middle line:84% was lower than the same relative price of the capital 01:04:28.620 --> 01:04:31.170 align:middle line:84% intensive goods in another country, 01:04:31.170 --> 01:04:33.690 align:middle line:84% like labor-abundant country, right? 01:04:33.690 --> 01:04:35.670 align:middle line:90% ROBERT TOWNSEND: Yeah. 01:04:35.670 --> 01:04:36.720 align:middle line:90% When you look at this-- 01:04:36.720 --> 01:04:39.480 align:middle line:84% another great question, thank you. 01:04:39.480 --> 01:04:41.190 align:middle line:84% So when you look at this diagram, 01:04:41.190 --> 01:04:42.960 align:middle line:84% it's really only the relative price 01:04:42.960 --> 01:04:45.030 align:middle line:90% of the goods that matters. 01:04:45.030 --> 01:04:47.130 align:middle line:90% It's like p1 over p2. 01:04:47.130 --> 01:04:50.610 align:middle line:84% So there's a lot of things that are relative here 01:04:50.610 --> 01:04:54.120 align:middle line:84% that look like they're in Heckscher-Ohlin stated 01:04:54.120 --> 01:04:56.070 align:middle line:90% as absolute prices. 01:04:56.070 --> 01:05:00.360 align:middle line:84% But it's really only the relative price. 01:05:00.360 --> 01:05:01.590 align:middle line:90% AUDIENCE: I get it. 01:05:01.590 --> 01:05:04.314 align:middle line:90% ROBERT TOWNSEND: OK. 01:05:04.314 --> 01:05:10.920 align:middle line:84% All right, so that was lecture 13. 01:05:10.920 --> 01:05:15.510 align:middle line:84% And the other two lectures, 14, we 01:05:15.510 --> 01:05:18.810 align:middle line:84% took that trade model to Thailand 01:05:18.810 --> 01:05:22.080 align:middle line:84% and did some more with the financial accounts 01:05:22.080 --> 01:05:25.890 align:middle line:84% to get not just individual income and balance sheet 01:05:25.890 --> 01:05:29.250 align:middle line:84% statements, but to aggregate up those statements to get 01:05:29.250 --> 01:05:31.560 align:middle line:90% to village asset accounts. 01:05:31.560 --> 01:05:35.280 align:middle line:84% Village level savings accounts, flows across the villages, 01:05:35.280 --> 01:05:36.310 align:middle line:90% and so on. 01:05:36.310 --> 01:05:37.960 align:middle line:90% But my time is up. 01:05:37.960 --> 01:05:38.977 align:middle line:90% So I can't review it. 01:05:38.977 --> 01:05:40.560 align:middle line:84% Of course, the other thing we're doing 01:05:40.560 --> 01:05:46.110 align:middle line:84% is taking that 2 by 2 model to actual data with the exception 01:05:46.110 --> 01:05:48.990 align:middle line:84% that the model here actually did not 01:05:48.990 --> 01:05:51.240 align:middle line:90% have constant returns to scale. 01:05:51.240 --> 01:05:55.770 align:middle line:84% It had capital mobility, whereas the 2 by 2 did not. 01:05:55.770 --> 01:05:57.630 align:middle line:84% Factor prices would be equalized, 01:05:57.630 --> 01:05:59.640 align:middle line:90% even without factor flows. 01:05:59.640 --> 01:06:01.710 align:middle line:84% The thing that's mucking up the Thai data 01:06:01.710 --> 01:06:03.970 align:middle line:84% is that there's a borrowing constraint. 01:06:03.970 --> 01:06:06.780 align:middle line:84% So we have both real and financial flows, 01:06:06.780 --> 01:06:09.640 align:middle line:84% subject to borrowing constraints. 01:06:09.640 --> 01:06:12.480 align:middle line:84% And lecture 15 was what I reviewed 01:06:12.480 --> 01:06:14.500 align:middle line:90% at the beginning of class. 01:06:14.500 --> 01:06:16.620 align:middle line:90% Any questions about 14? 01:06:16.620 --> 01:06:19.490 align:middle line:90% 01:06:19.490 --> 01:06:23.240 align:middle line:84% OK, well I hope you find these reviews useful. 01:06:23.240 --> 01:06:25.550 align:middle line:84% I must say, you should all feel really good, 01:06:25.550 --> 01:06:29.000 align:middle line:84% because you're doing well in the class, 01:06:29.000 --> 01:06:31.880 align:middle line:84% and we are covering a lot of material. 01:06:31.880 --> 01:06:36.320 align:middle line:84% The constant themes here are theory, data, both, 01:06:36.320 --> 01:06:39.200 align:middle line:90% lots of tools, lots of concepts. 01:06:39.200 --> 01:06:43.370 align:middle line:84% By now, you're functioning like actual real economists, 01:06:43.370 --> 01:06:46.470 align:middle line:84% and the conversations on Piazza reflect that. 01:06:46.470 --> 01:06:50.210 align:middle line:84% So I'm really happy with how the class is going. 01:06:50.210 --> 01:06:53.270 align:middle line:84% And I hope you found this review to be useful, 01:06:53.270 --> 01:06:57.740 align:middle line:84% and to be a reminder of how much you have learned. 01:06:57.740 --> 01:06:58.280 align:middle line:90% All right. 01:06:58.280 --> 01:07:00.490 align:middle line:90% Thank you very much. 01:07:00.490 --> 01:07:08.000 align:middle line:90%