WEBVTT 00:00:00.000 --> 00:00:02.445 align:middle line:90% [SQUEAKING] 00:00:02.445 --> 00:00:04.401 align:middle line:90% [RUSTLING] 00:00:04.401 --> 00:00:06.357 align:middle line:90% [CLICKING] 00:00:06.357 --> 00:00:17.953 align:middle line:90% 00:00:17.953 --> 00:00:20.370 align:middle line:84% PROFESSOR: So we have a lot to talk about in the last week 00:00:20.370 --> 00:00:21.240 align:middle line:90% in terms of credit. 00:00:21.240 --> 00:00:23.530 align:middle line:84% I'm not quite done with the labor section, 00:00:23.530 --> 00:00:26.130 align:middle line:84% so what I'm going to do is just spend about 10 minutes 00:00:26.130 --> 00:00:27.990 align:middle line:84% just giving a couple of highlights 00:00:27.990 --> 00:00:29.130 align:middle line:90% of some of the last-- 00:00:29.130 --> 00:00:32.130 align:middle line:84% a little overview of some of the topics that are in the labor 00:00:32.130 --> 00:00:32.729 align:middle line:90% notes. 00:00:32.729 --> 00:00:34.410 align:middle line:84% You're welcome to review them in detail, 00:00:34.410 --> 00:00:38.320 align:middle line:84% and then I want to get into the credit stuff in detail. 00:00:38.320 --> 00:00:45.150 align:middle line:84% So there were two other topics that I just wanted to give you. 00:00:45.150 --> 00:00:48.150 align:middle line:84% The first is, there's been some additional attention-- 00:00:48.150 --> 00:00:50.010 align:middle line:84% and these are all what's happening in kind 00:00:50.010 --> 00:00:51.550 align:middle line:90% of the urban labor markets. 00:00:51.550 --> 00:00:53.280 align:middle line:84% So there's been some work recently 00:00:53.280 --> 00:00:55.800 align:middle line:84% on understanding matching frictions. 00:00:55.800 --> 00:00:59.550 align:middle line:84% And in particular, a big issue in labor markets 00:00:59.550 --> 00:01:02.813 align:middle line:84% is the fact that not all jobs are the same, 00:01:02.813 --> 00:01:04.230 align:middle line:84% not all people are the same, so we 00:01:04.230 --> 00:01:06.660 align:middle line:90% want to match people to jobs. 00:01:06.660 --> 00:01:08.850 align:middle line:84% A big issue that differentiates labor markets 00:01:08.850 --> 00:01:11.190 align:middle line:84% from other markets is kind of the rich heterogeneity, 00:01:11.190 --> 00:01:14.130 align:middle line:84% both in the employer side and on the other side. 00:01:14.130 --> 00:01:15.570 align:middle line:84% And so job search is a big issue. 00:01:15.570 --> 00:01:18.096 align:middle line:90% 00:01:18.096 --> 00:01:22.080 align:middle line:84% And this question is a big one in developed country labor 00:01:22.080 --> 00:01:23.580 align:middle line:84% markets, and I think people are just 00:01:23.580 --> 00:01:26.190 align:middle line:84% starting to understand how those issues may 00:01:26.190 --> 00:01:30.102 align:middle line:84% be different in developing country contexts. 00:01:30.102 --> 00:01:32.310 align:middle line:84% And one issue that people have started to think about 00:01:32.310 --> 00:01:35.530 align:middle line:84% is that part of how we help solve these problems 00:01:35.530 --> 00:01:40.980 align:middle line:84% is, we have various signals that help us signal in the market 00:01:40.980 --> 00:01:42.420 align:middle line:90% what our skills are. 00:01:42.420 --> 00:01:46.380 align:middle line:90% So you get a PhD from MIT. 00:01:46.380 --> 00:01:49.020 align:middle line:84% That is a signal that you can take the future employers 00:01:49.020 --> 00:01:51.570 align:middle line:84% and say, I know some economics, for example. 00:01:51.570 --> 00:01:54.850 align:middle line:84% At least we hope that's a useful signal. 00:01:54.850 --> 00:01:58.800 align:middle line:84% But if those signals are weak and the quality of the signals 00:01:58.800 --> 00:02:02.410 align:middle line:84% is not as good, there may be some challenges. 00:02:02.410 --> 00:02:07.270 align:middle line:84% So one paper is trying to understand these issues, 00:02:07.270 --> 00:02:09.699 align:middle line:84% looking in the South African labor market. 00:02:09.699 --> 00:02:14.590 align:middle line:84% And they do some interesting things on this. 00:02:14.590 --> 00:02:19.065 align:middle line:84% They look at giving people information about-- 00:02:19.065 --> 00:02:21.190 align:middle line:84% and they note that these skill assessments tell you 00:02:21.190 --> 00:02:22.330 align:middle line:90% multiple things. 00:02:22.330 --> 00:02:25.675 align:middle line:84% In the MIT example, if you get a PhD from MIT, 00:02:25.675 --> 00:02:27.550 align:middle line:84% that's a valuable signal in the labor market. 00:02:27.550 --> 00:02:29.497 align:middle line:84% It's also a signal to you that actually you're 00:02:29.497 --> 00:02:31.330 align:middle line:84% pretty good at economics because you managed 00:02:31.330 --> 00:02:32.950 align:middle line:90% to get a PhD in Economics. 00:02:32.950 --> 00:02:36.400 align:middle line:84% So those information, they provide information 00:02:36.400 --> 00:02:40.990 align:middle line:84% both to the job seeker and to potential employers. 00:02:40.990 --> 00:02:42.340 align:middle line:90% To get at this idea-- 00:02:42.340 --> 00:02:44.620 align:middle line:90% so not with PhDs, for example-- 00:02:44.620 --> 00:02:46.750 align:middle line:90% they basically help people-- 00:02:46.750 --> 00:02:49.930 align:middle line:84% they do a skill assessments of people. 00:02:49.930 --> 00:02:52.510 align:middle line:84% And they give them a certificate with a World Bank brand, 00:02:52.510 --> 00:02:55.840 align:middle line:84% like a credible brand saying, here's your skills assessment 00:02:55.840 --> 00:02:56.560 align:middle line:90% or whatever. 00:02:56.560 --> 00:02:58.180 align:middle line:84% Then they also do a second arm where 00:02:58.180 --> 00:03:00.490 align:middle line:84% they give the same information but kind of privately to job 00:03:00.490 --> 00:03:00.880 align:middle line:90% seekers. 00:03:00.880 --> 00:03:03.380 align:middle line:84% They show it to them, but they don't give them a certificate 00:03:03.380 --> 00:03:04.840 align:middle line:90% they can bring out to people. 00:03:04.840 --> 00:03:06.640 align:middle line:84% And they show, in order to decompose, 00:03:06.640 --> 00:03:10.805 align:middle line:84% what is the role of the job seekers from the alleviation 00:03:10.805 --> 00:03:12.430 align:middle line:84% to the job seekers from the information 00:03:12.430 --> 00:03:15.350 align:middle line:90% to potential employers. 00:03:15.350 --> 00:03:15.850 align:middle line:90% I'm sorry. 00:03:15.850 --> 00:03:17.890 align:middle line:90% This is about certificates. 00:03:17.890 --> 00:03:19.550 align:middle line:84% So they basically decompose this. 00:03:19.550 --> 00:03:21.842 align:middle line:84% So I'm not going to go through this in a lot of detail. 00:03:21.842 --> 00:03:23.980 align:middle line:84% I just wanted to highlight that this area of, how 00:03:23.980 --> 00:03:27.032 align:middle line:84% do we think about the job search in general, 00:03:27.032 --> 00:03:28.990 align:middle line:84% is an area people are starting to think through 00:03:28.990 --> 00:03:37.730 align:middle line:84% and this question about understanding 00:03:37.730 --> 00:03:40.010 align:middle line:84% the role of credible signals for employers. 00:03:40.010 --> 00:03:41.825 align:middle line:84% And that is one way to look at this. 00:03:41.825 --> 00:03:42.950 align:middle line:90% And it's a very nice paper. 00:03:42.950 --> 00:03:44.783 align:middle line:84% So I'm not going to go through it in detail, 00:03:44.783 --> 00:03:47.720 align:middle line:84% but I encourage you to look at it. 00:03:47.720 --> 00:03:49.945 align:middle line:84% The other thing people are starting to look at 00:03:49.945 --> 00:03:52.280 align:middle line:84% is job search and online platforms. 00:03:52.280 --> 00:03:57.710 align:middle line:84% So this has been a big move in the developed country labor 00:03:57.710 --> 00:03:58.210 align:middle line:90% markets. 00:03:58.210 --> 00:04:00.430 align:middle line:84% It's the world of these online job matching programs. 00:04:00.430 --> 00:04:03.700 align:middle line:84% Is that actually relevant in these lower middle-income 00:04:03.700 --> 00:04:04.220 align:middle line:90% countries? 00:04:04.220 --> 00:04:06.262 align:middle line:84% I think people are starting to think maybe it is, 00:04:06.262 --> 00:04:08.080 align:middle line:84% so that's something people are working on. 00:04:08.080 --> 00:04:09.610 align:middle line:84% Job training, there's a lot of work 00:04:09.610 --> 00:04:13.748 align:middle line:84% on this question of, how do we teach people jobs, et cetera. 00:04:13.748 --> 00:04:16.040 align:middle line:84% Is there a general purpose training, specific training? 00:04:16.040 --> 00:04:17.329 align:middle line:84% Again, I'm not going to go through this in detail, 00:04:17.329 --> 00:04:19.550 align:middle line:84% but I wanted to give you a pointer that some of this work 00:04:19.550 --> 00:04:19.880 align:middle line:90% exists. 00:04:19.880 --> 00:04:21.713 align:middle line:84% And if you're interested in these questions, 00:04:21.713 --> 00:04:23.750 align:middle line:90% you should have a look at it. 00:04:23.750 --> 00:04:26.750 align:middle line:84% The second issue in urban labor markets 00:04:26.750 --> 00:04:30.140 align:middle line:84% that I also want to highlight is the rule of working conditions. 00:04:30.140 --> 00:04:37.430 align:middle line:84% And this is an issue, I think, if we 00:04:37.430 --> 00:04:40.130 align:middle line:84% think both about the development of the Industrial Revolution 00:04:40.130 --> 00:04:43.550 align:middle line:84% and the area post the Industrial Revolution. 00:04:43.550 --> 00:04:48.740 align:middle line:84% In a lot of countries, like the US or the UK 00:04:48.740 --> 00:04:51.350 align:middle line:84% or other kind of richer countries today, 00:04:51.350 --> 00:04:54.290 align:middle line:84% or in a lot of lower middle-income countries today, 00:04:54.290 --> 00:04:57.860 align:middle line:84% how do you think about, what are working 00:04:57.860 --> 00:05:03.693 align:middle line:84% conditions like in a lot of these industrial factories? 00:05:03.693 --> 00:05:05.610 align:middle line:84% There still not perfect everywhere, of course, 00:05:05.610 --> 00:05:07.485 align:middle line:84% but there's a lot of regulation, for example, 00:05:07.485 --> 00:05:08.990 align:middle line:90% in more developed countries. 00:05:08.990 --> 00:05:11.690 align:middle line:84% It helps make you think about worker safety and things 00:05:11.690 --> 00:05:12.440 align:middle line:90% like that. 00:05:12.440 --> 00:05:14.510 align:middle line:84% If we have large informal sectors that 00:05:14.510 --> 00:05:17.480 align:middle line:84% aren't subject to those regulations, 00:05:17.480 --> 00:05:20.870 align:middle line:84% what's happening in these lower middle-income countries? 00:05:20.870 --> 00:05:23.120 align:middle line:84% I mean, to me, this is a really interesting and really 00:05:23.120 --> 00:05:27.693 align:middle line:84% important issue and one that I feel is not particularly 00:05:27.693 --> 00:05:29.360 align:middle line:84% studied to the extent it probably should 00:05:29.360 --> 00:05:30.610 align:middle line:90% be in the economic literature. 00:05:30.610 --> 00:05:34.030 align:middle line:84% So I think it's a really important area. 00:05:34.030 --> 00:05:37.770 align:middle line:84% So as I mentioned, there were huge uproars 00:05:37.770 --> 00:05:39.870 align:middle line:84% about this kind of 100 years ago in the US, 00:05:39.870 --> 00:05:44.800 align:middle line:84% for example, about some of these workplace tragedies. 00:05:44.800 --> 00:05:47.982 align:middle line:84% There have been similar tragic incidents. 00:05:47.982 --> 00:05:50.190 align:middle line:84% There were a couple of very famous ones in Bangladesh 00:05:50.190 --> 00:05:51.430 align:middle line:90% about 10 years ago. 00:05:51.430 --> 00:05:54.960 align:middle line:84% There was this collapse of this factory that had many textile 00:05:54.960 --> 00:05:56.430 align:middle line:90% shops, hundreds-- 00:05:56.430 --> 00:05:59.050 align:middle line:84% or actually, over 1,000 people were killed. 00:05:59.050 --> 00:06:01.838 align:middle line:84% So how do we think about some of these kind of workplace issues? 00:06:01.838 --> 00:06:03.630 align:middle line:84% Again, particularly when you have a large-- 00:06:03.630 --> 00:06:07.260 align:middle line:84% and why is this kind of a development issue? 00:06:07.260 --> 00:06:10.230 align:middle line:84% It's because we think that regulation actually 00:06:10.230 --> 00:06:14.970 align:middle line:84% is what helps solve some of these problems in more 00:06:14.970 --> 00:06:16.290 align:middle line:90% developed countries. 00:06:16.290 --> 00:06:18.420 align:middle line:84% But with these large informal sectors, 00:06:18.420 --> 00:06:22.770 align:middle line:84% the regulatory issues, maybe not so. 00:06:22.770 --> 00:06:25.513 align:middle line:84% Regulatory solutions may not work so well. 00:06:25.513 --> 00:06:27.930 align:middle line:84% So I just think I wanted to flag that as something I think 00:06:27.930 --> 00:06:29.250 align:middle line:90% as a really important area. 00:06:29.250 --> 00:06:32.580 align:middle line:84% There was one paper recently by Laura Boudreau, 00:06:32.580 --> 00:06:34.860 align:middle line:84% which I thought was a very nice one about looking 00:06:34.860 --> 00:06:43.700 align:middle line:84% at these worker management safety committees, 00:06:43.700 --> 00:06:45.920 align:middle line:84% and actually randomizing the creation of these 00:06:45.920 --> 00:06:50.002 align:middle line:84% into Bangladeshi factories and looking at some issues here. 00:06:50.002 --> 00:06:51.710 align:middle line:84% More generally, I think, what is the role 00:06:51.710 --> 00:06:52.910 align:middle line:90% of multinationals actually? 00:06:52.910 --> 00:06:55.840 align:middle line:84% And are they playing a role in disciplining the working 00:06:55.840 --> 00:06:58.340 align:middle line:84% conditions in the factories in lower middle-income countries 00:06:58.340 --> 00:07:00.007 align:middle line:84% that are producing their stuff, I think, 00:07:00.007 --> 00:07:01.807 align:middle line:90% is a really interesting area. 00:07:01.807 --> 00:07:03.890 align:middle line:84% Again, I'm not going to go through this in detail, 00:07:03.890 --> 00:07:05.960 align:middle line:90% given the limited time I have. 00:07:05.960 --> 00:07:08.660 align:middle line:84% But I kind wanted to just flag this paper 00:07:08.660 --> 00:07:13.880 align:middle line:84% if you're interested in it and just more generally 00:07:13.880 --> 00:07:17.240 align:middle line:84% say that I think that this issue about multinational-- issue 00:07:17.240 --> 00:07:18.980 align:middle line:84% of working conditions in general, 00:07:18.980 --> 00:07:21.170 align:middle line:84% and multinationals in particular, and what role they 00:07:21.170 --> 00:07:23.060 align:middle line:84% play is an important one and one that's 00:07:23.060 --> 00:07:25.710 align:middle line:90% a good area for future research. 00:07:25.710 --> 00:07:27.690 align:middle line:84% OK, so that's all I'm going to say, 00:07:27.690 --> 00:07:29.460 align:middle line:84% and then I'm going to switch gears. 00:07:29.460 --> 00:07:31.740 align:middle line:84% Do people have any other comments or questions 00:07:31.740 --> 00:07:33.300 align:middle line:84% before we switch to the last topic? 00:07:33.300 --> 00:07:36.140 align:middle line:90% 00:07:36.140 --> 00:07:36.962 align:middle line:90% OK. 00:07:36.962 --> 00:07:39.420 align:middle line:84% So sorry I didn't get to go through those papers in detail, 00:07:39.420 --> 00:07:42.847 align:middle line:84% but there's a lot to talk about in credit too. 00:07:42.847 --> 00:07:44.180 align:middle line:90% I want to make sure I get to it. 00:07:44.180 --> 00:07:52.930 align:middle line:90% 00:07:52.930 --> 00:08:00.730 align:middle line:84% OK, so credit is obviously a hugely important issue. 00:08:00.730 --> 00:08:04.330 align:middle line:84% If you think about what we've talked about, 00:08:04.330 --> 00:08:07.000 align:middle line:84% if you think about the overall production function, 00:08:07.000 --> 00:08:07.600 align:middle line:90% there's like-- 00:08:07.600 --> 00:08:12.123 align:middle line:90% 00:08:12.123 --> 00:08:14.540 align:middle line:84% we think about a production function that looks like this. 00:08:14.540 --> 00:08:16.760 align:middle line:84% We have labor, which I just talked 00:08:16.760 --> 00:08:24.790 align:middle line:84% about, human capital, which Ezra talked about 00:08:24.790 --> 00:08:27.070 align:middle line:90% in extensive detail. 00:08:27.070 --> 00:08:28.810 align:middle line:84% There are some other components of this. 00:08:28.810 --> 00:08:35.669 align:middle line:84% One is productivity, and the other is capital. 00:08:35.669 --> 00:08:38.669 align:middle line:90% 00:08:38.669 --> 00:08:42.809 align:middle line:84% So understanding what are the challenges 00:08:42.809 --> 00:08:45.540 align:middle line:84% and overall production in lower middle-income countries 00:08:45.540 --> 00:08:48.570 align:middle line:84% compared to the developed world, you want to understand all four 00:08:48.570 --> 00:08:49.890 align:middle line:90% of these components. 00:08:49.890 --> 00:08:51.420 align:middle line:84% So we've talked a lot about labor. 00:08:51.420 --> 00:08:54.390 align:middle line:84% We've talked a lot about human capital. 00:08:54.390 --> 00:08:56.880 align:middle line:84% We're going to talk next semester much more 00:08:56.880 --> 00:08:57.963 align:middle line:90% about productivity. 00:08:57.963 --> 00:08:59.880 align:middle line:84% There's going to be a lot of discussion about, 00:08:59.880 --> 00:09:01.578 align:middle line:84% how do we think about productivity, 00:09:01.578 --> 00:09:03.870 align:middle line:84% and how do we think about heterogeneity in productivity 00:09:03.870 --> 00:09:04.860 align:middle line:90% across firms. 00:09:04.860 --> 00:09:07.110 align:middle line:84% And the final piece is, we should think about capital. 00:09:07.110 --> 00:09:09.210 align:middle line:90% 00:09:09.210 --> 00:09:11.760 align:middle line:84% And we're going to split our discussion of capital 00:09:11.760 --> 00:09:15.720 align:middle line:84% across what I'm going to talk about today and next time 00:09:15.720 --> 00:09:19.180 align:middle line:84% and what we're going to talk about in 14.772 next semester. 00:09:19.180 --> 00:09:23.440 align:middle line:84% And in particular, what we're going to talk about 00:09:23.440 --> 00:09:26.080 align:middle line:90% is credit constraints. 00:09:26.080 --> 00:09:30.730 align:middle line:84% So normally, we're maximizing this production function, 00:09:30.730 --> 00:09:32.920 align:middle line:90% you know, minus-- 00:09:32.920 --> 00:09:36.850 align:middle line:84% if you were maximizing something like this, 00:09:36.850 --> 00:09:42.310 align:middle line:84% over l, h, and k minus whatever-- 00:09:42.310 --> 00:09:44.440 align:middle line:90% wL minus whatever. 00:09:44.440 --> 00:09:49.820 align:middle line:84% I'll call it wh of H. You would choose the amount of capital, 00:09:49.820 --> 00:09:51.700 align:middle line:90% such that a prime-- 00:09:51.700 --> 00:09:55.810 align:middle line:84% sorry-- AF prime sub k is equal to r. 00:09:55.810 --> 00:09:57.213 align:middle line:90% You would just maximize. 00:09:57.213 --> 00:09:58.630 align:middle line:84% Choose your level of capital, such 00:09:58.630 --> 00:10:00.700 align:middle line:84% that the marginal return to the capital 00:10:00.700 --> 00:10:03.010 align:middle line:84% was equal to the interest rate or the rate of return 00:10:03.010 --> 00:10:06.865 align:middle line:84% to capital, which we think of as the interest rate. 00:10:06.865 --> 00:10:08.490 align:middle line:84% So what we're going to talk about today 00:10:08.490 --> 00:10:11.200 align:middle line:84% are reasons why we may not be able to do that. 00:10:11.200 --> 00:10:13.770 align:middle line:84% And in particular, we want to understand 00:10:13.770 --> 00:10:17.730 align:middle line:84% how does the market for capital or the market for credit work, 00:10:17.730 --> 00:10:21.330 align:middle line:84% and are there reasons why firms might not be 00:10:21.330 --> 00:10:23.577 align:middle line:90% able to get this much capital. 00:10:23.577 --> 00:10:25.410 align:middle line:84% Firms, businesses, anybody who needs capital 00:10:25.410 --> 00:10:27.202 align:middle line:84% might not be able to get this much capital. 00:10:27.202 --> 00:10:28.960 align:middle line:84% It might be limited in their capital. 00:10:28.960 --> 00:10:31.758 align:middle line:84% So today, I want to talk about the basic theory of credit 00:10:31.758 --> 00:10:34.300 align:middle line:84% constraints, and that's going to basically take me all today. 00:10:34.300 --> 00:10:37.650 align:middle line:84% So how do we think about why might people not be 00:10:37.650 --> 00:10:39.812 align:middle line:90% able to get this much capital? 00:10:39.812 --> 00:10:42.270 align:middle line:84% And I'll talk about moral hazard reasons, adverse selection 00:10:42.270 --> 00:10:46.650 align:middle line:84% reasons, and monitoring reasons, and we'll think 00:10:46.650 --> 00:10:49.110 align:middle line:90% about what does that all mean. 00:10:49.110 --> 00:10:53.730 align:middle line:84% And then I'm going to talk on Wednesday about the role 00:10:53.730 --> 00:10:58.110 align:middle line:84% of microfinance and how we think about microfinance as one way 00:10:58.110 --> 00:11:01.020 align:middle line:84% of solving these kind of credit-- or not solving, 00:11:01.020 --> 00:11:04.470 align:middle line:84% but ameliorating some of these credit constraints for really 00:11:04.470 --> 00:11:05.670 align:middle line:90% poor-- 00:11:05.670 --> 00:11:10.830 align:middle line:84% not poor-- but typically poor, very small micro enterprises. 00:11:10.830 --> 00:11:12.870 align:middle line:84% And there's been an explosion of work 00:11:12.870 --> 00:11:16.740 align:middle line:84% understanding both whether or not microfinance does or does 00:11:16.740 --> 00:11:18.330 align:middle line:84% not ameliorate credit constraints, 00:11:18.330 --> 00:11:20.882 align:middle line:84% and understanding all the many different features that 00:11:20.882 --> 00:11:22.590 align:middle line:84% happen in microfinance, and understanding 00:11:22.590 --> 00:11:24.440 align:middle line:90% whether those are important. 00:11:24.440 --> 00:11:26.650 align:middle line:84% So that's what I'm going to do this week. 00:11:26.650 --> 00:11:31.270 align:middle line:84% What we're, then, going to think about is in 14.772, 00:11:31.270 --> 00:11:34.960 align:middle line:84% we'll think about the role of banks as intermediaries. 00:11:34.960 --> 00:11:36.100 align:middle line:90% What are banks doing? 00:11:36.100 --> 00:11:37.540 align:middle line:90% How are they allocating capital? 00:11:37.540 --> 00:11:39.645 align:middle line:84% What are they doing in credit constraints? 00:11:39.645 --> 00:11:41.020 align:middle line:84% And more generally, understanding 00:11:41.020 --> 00:11:44.230 align:middle line:84% a firm's demand for credit and the impact of credit on firms. 00:11:44.230 --> 00:11:46.780 align:middle line:84% You can think of it in some ways as estimating that Fk thing. 00:11:46.780 --> 00:11:49.900 align:middle line:84% What is the impact of additional capital on the firm sector? 00:11:49.900 --> 00:11:52.660 align:middle line:84% So I split this topic, either the stuff 00:11:52.660 --> 00:11:54.970 align:middle line:84% that we think of as a little more-- 00:11:54.970 --> 00:11:57.350 align:middle line:84% this is mostly about people and the basics 00:11:57.350 --> 00:11:58.600 align:middle line:90% of understanding these models. 00:11:58.600 --> 00:11:59.380 align:middle line:90% I'm going to do now. 00:11:59.380 --> 00:12:01.797 align:middle line:84% The stuff about really getting into how this affect firms, 00:12:01.797 --> 00:12:04.140 align:middle line:90% we'll do next semester. 00:12:04.140 --> 00:12:07.030 align:middle line:90% That's the plan. 00:12:07.030 --> 00:12:12.275 align:middle line:84% OK, so here's the neoclassical model of the capital market. 00:12:12.275 --> 00:12:14.150 align:middle line:84% Imagine that everyone faces the same interest 00:12:14.150 --> 00:12:17.505 align:middle line:90% rate, adjusted for risk. 00:12:17.505 --> 00:12:18.880 align:middle line:84% Now, what that means, by the way, 00:12:18.880 --> 00:12:23.230 align:middle line:84% is if there's a d percent risk of default, then-- 00:12:23.230 --> 00:12:25.770 align:middle line:90% 00:12:25.770 --> 00:12:27.600 align:middle line:84% so we're going to work with what's 00:12:27.600 --> 00:12:30.420 align:middle line:84% called the gross interest rate, which is how much-- 00:12:30.420 --> 00:12:33.780 align:middle line:84% so you repay the gross, which is often more convenient. 00:12:33.780 --> 00:12:35.760 align:middle line:84% The interest rate we think of as like-- 00:12:35.760 --> 00:12:39.090 align:middle line:84% so imagine I have a 3% interest rate, so I would 00:12:39.090 --> 00:12:41.340 align:middle line:90% have to pay 3% of interest. 00:12:41.340 --> 00:12:44.250 align:middle line:84% The gross interest rate is like 1.03. 00:12:44.250 --> 00:12:46.650 align:middle line:84% It's the total amount you have to pay back, 00:12:46.650 --> 00:12:49.830 align:middle line:84% and that just turns out to be convenient for the math 00:12:49.830 --> 00:12:50.940 align:middle line:90% sometimes. 00:12:50.940 --> 00:12:57.200 align:middle line:84% So imagine I default. And different people 00:12:57.200 --> 00:13:00.050 align:middle line:90% differ in their default risks. 00:13:00.050 --> 00:13:02.480 align:middle line:84% Then what is the bank going to get back? 00:13:02.480 --> 00:13:05.250 align:middle line:84% Well, the bank's going to get back-- well, with probability 00:13:05.250 --> 00:13:07.040 align:middle line:90% d, it's going to get nothing. 00:13:07.040 --> 00:13:09.290 align:middle line:84% With probability 1 minus d, it's going 00:13:09.290 --> 00:13:12.260 align:middle line:90% to get the gross interest rate. 00:13:12.260 --> 00:13:15.415 align:middle line:84% So in a basic kind of market, if we 00:13:15.415 --> 00:13:16.790 align:middle line:84% have different chances of default 00:13:16.790 --> 00:13:20.750 align:middle line:84% and that chance is exogenous or whatever and observable, 00:13:20.750 --> 00:13:24.230 align:middle line:84% then I'll just be happy to offer you all different loans, 00:13:24.230 --> 00:13:27.260 align:middle line:84% and the gross interest rate I'm going to charge for your loan 00:13:27.260 --> 00:13:29.820 align:middle line:90% is 1 minus d times r. 00:13:29.820 --> 00:13:38.500 align:middle line:84% So I'm going to get back my r in equilibrium. 00:13:38.500 --> 00:13:39.000 align:middle line:90% OK. 00:13:39.000 --> 00:13:43.180 align:middle line:90% 00:13:43.180 --> 00:13:45.040 align:middle line:84% And basically, that's kind of what you get. 00:13:45.040 --> 00:13:48.452 align:middle line:90% 00:13:48.452 --> 00:13:50.910 align:middle line:84% That's what you get, and then the expected marginal product 00:13:50.910 --> 00:13:52.785 align:middle line:84% of capital should be equal to 1 minus d times 00:13:52.785 --> 00:13:55.960 align:middle line:84% r because that's the interest that you have to pay. 00:13:55.960 --> 00:13:57.780 align:middle line:90% Fine. 00:13:57.780 --> 00:14:00.550 align:middle line:84% So here are some stylized facts, and I'm not 00:14:00.550 --> 00:14:02.300 align:middle line:84% going to go into the detail between these, 00:14:02.300 --> 00:14:10.280 align:middle line:84% but there's some work that suggests these are true. 00:14:10.280 --> 00:14:13.778 align:middle line:84% So we suggest that maybe that simple neoclassical model is 00:14:13.778 --> 00:14:15.070 align:middle line:90% not everything that's going on. 00:14:15.070 --> 00:14:18.972 align:middle line:90% 00:14:18.972 --> 00:14:21.180 align:middle line:84% So then, by the way, one of the implications of this, 00:14:21.180 --> 00:14:23.910 align:middle line:84% by the way, is that what's happening to depositors 00:14:23.910 --> 00:14:28.320 align:middle line:84% and what's happening to the people who are getting 00:14:28.320 --> 00:14:32.910 align:middle line:90% the capital should be small. 00:14:32.910 --> 00:14:34.950 align:middle line:84% Imagine there's free entry into banks. 00:14:34.950 --> 00:14:38.760 align:middle line:84% Then basically, that gap will be kind of competed to zero 00:14:38.760 --> 00:14:40.980 align:middle line:84% because depositors get the highest interest rate, 00:14:40.980 --> 00:14:41.970 align:middle line:84% borrowers will get the lowest interest rate. 00:14:41.970 --> 00:14:43.730 align:middle line:90% That gap will be very small. 00:14:43.730 --> 00:14:44.580 align:middle line:90% So what do we see? 00:14:44.580 --> 00:14:46.747 align:middle line:84% So we see, first of all, there's a big gap sometimes 00:14:46.747 --> 00:14:49.960 align:middle line:84% between the deposit rates and loan rates. 00:14:49.960 --> 00:14:51.710 align:middle line:84% It seems like there's a lot of money going 00:14:51.710 --> 00:14:53.660 align:middle line:90% in that intermediary sector. 00:14:53.660 --> 00:14:55.160 align:middle line:90% What's that about? 00:14:55.160 --> 00:14:57.050 align:middle line:90% Talk about that. 00:14:57.050 --> 00:14:59.670 align:middle line:84% A lot of variability in interest rate 00:14:59.670 --> 00:15:01.670 align:middle line:84% within the same subeconomy, and that's true even 00:15:01.670 --> 00:15:04.180 align:middle line:84% with very low levels of default. So we're 00:15:04.180 --> 00:15:10.750 align:middle line:84% going to see heterogeneity in interest rates that 00:15:10.750 --> 00:15:14.485 align:middle line:84% seems to be maybe something else going on beyond just defaults. 00:15:14.485 --> 00:15:16.360 align:middle line:84% So we'll see if there's something about that, 00:15:16.360 --> 00:15:18.658 align:middle line:84% even though there is ex-ante competition in markets. 00:15:18.658 --> 00:15:20.200 align:middle line:84% And another thing we're going to find 00:15:20.200 --> 00:15:24.010 align:middle line:84% is that richer people seem to be able to borrow more and pay 00:15:24.010 --> 00:15:27.490 align:middle line:84% lower rates of interest than poorer people. 00:15:27.490 --> 00:15:29.620 align:middle line:84% And more generally, it seems like people 00:15:29.620 --> 00:15:31.720 align:middle line:84% who borrow more are paying lower interest rates, 00:15:31.720 --> 00:15:33.020 align:middle line:84% and that seems a little weird, actually, 00:15:33.020 --> 00:15:34.240 align:middle line:90% when you think about it. 00:15:34.240 --> 00:15:38.740 align:middle line:84% You might have thought that maybe the more you borrow, 00:15:38.740 --> 00:15:41.380 align:middle line:84% the riskier you are, and maybe the default rate should go up. 00:15:41.380 --> 00:15:42.430 align:middle line:84% But actually, it doesn't seem that way. 00:15:42.430 --> 00:15:44.638 align:middle line:84% It seems like there's actually a positive correlation 00:15:44.638 --> 00:15:45.930 align:middle line:90% between the amount you borrow-- 00:15:45.930 --> 00:15:48.430 align:middle line:84% sorry-- a negative correlation between the amount you borrow 00:15:48.430 --> 00:15:49.680 align:middle line:90% and the interest rate. 00:15:49.680 --> 00:15:51.780 align:middle line:84% So that also seems kind of interesting. 00:15:51.780 --> 00:15:53.690 align:middle line:84% So it seems like there's something 00:15:53.690 --> 00:15:56.380 align:middle line:84% a little more complicated going on in the credit markets. 00:15:56.380 --> 00:15:59.470 align:middle line:90% And so what might that be? 00:15:59.470 --> 00:16:01.800 align:middle line:84% So here's a model, a simple model, 00:16:01.800 --> 00:16:04.605 align:middle line:84% that of moral hazard in investment choices. 00:16:04.605 --> 00:16:07.120 align:middle line:90% 00:16:07.120 --> 00:16:12.700 align:middle line:84% So imagine that there is a menu of possible investments 00:16:12.700 --> 00:16:16.380 align:middle line:90% that I as a firm can choose. 00:16:16.380 --> 00:16:22.850 align:middle line:84% And they each differ in their probability of success, p. 00:16:22.850 --> 00:16:26.120 align:middle line:84% And investment p is going to yield some return, r of p, 00:16:26.120 --> 00:16:29.300 align:middle line:84% with probability p and return zero, 00:16:29.300 --> 00:16:30.875 align:middle line:90% with probability 1 minus p. 00:16:30.875 --> 00:16:34.560 align:middle line:90% 00:16:34.560 --> 00:16:40.002 align:middle line:84% So my expected return is going to be e of p, 00:16:40.002 --> 00:16:41.710 align:middle line:84% and that's going to be our-- that's going 00:16:41.710 --> 00:16:44.710 align:middle line:84% to be the probability of getting the positive return 00:16:44.710 --> 00:16:46.420 align:middle line:90% times the return, if you get it. 00:16:46.420 --> 00:16:49.720 align:middle line:90% That's my expected return. 00:16:49.720 --> 00:16:53.280 align:middle line:84% We're going to assume that r prime is negative, 00:16:53.280 --> 00:16:59.010 align:middle line:90% that the product with lower-- 00:16:59.010 --> 00:17:01.980 align:middle line:84% the product with higher probabilities of success 00:17:01.980 --> 00:17:04.810 align:middle line:84% are going to return lower amounts as they succeed. 00:17:04.810 --> 00:17:07.410 align:middle line:84% And I think that's a pretty general assumption 00:17:07.410 --> 00:17:09.839 align:middle line:84% because, otherwise, there would be dominated projects. 00:17:09.839 --> 00:17:11.839 align:middle line:84% So we can choose the best project for everybody, 00:17:11.839 --> 00:17:15.359 align:middle line:90% and then it would happen. 00:17:15.359 --> 00:17:17.280 align:middle line:90% Otherwise, not very interesting. 00:17:17.280 --> 00:17:20.490 align:middle line:84% And we're going to assume that that's going to yield. 00:17:20.490 --> 00:17:24.030 align:middle line:84% And then we'll assume that this e function is concave, 00:17:24.030 --> 00:17:26.700 align:middle line:84% so there'll be an interior level of the optimal project 00:17:26.700 --> 00:17:30.000 align:middle line:90% for maximizing expected returns. 00:17:30.000 --> 00:17:31.170 align:middle line:90% OK. 00:17:31.170 --> 00:17:33.250 align:middle line:84% And we're going to note by p star 00:17:33.250 --> 00:17:37.200 align:middle line:84% the maximizing project, the efficient project. 00:17:37.200 --> 00:17:40.350 align:middle line:84% That's the one that maximizes the efficient returns. 00:17:40.350 --> 00:17:44.550 align:middle line:84% So that's just maximize. p star is just the max of this thing. 00:17:44.550 --> 00:17:49.620 align:middle line:84% OK, so we would like people to choose p star in some sense. 00:17:49.620 --> 00:17:56.370 align:middle line:84% In addition to the project choice, the p, 00:17:56.370 --> 00:17:59.190 align:middle line:84% there's a return to scale parameter, 00:17:59.190 --> 00:18:01.050 align:middle line:90% which is given by f of k. 00:18:01.050 --> 00:18:03.157 align:middle line:84% And maybe that's going to be diminishing 00:18:03.157 --> 00:18:04.365 align:middle line:90% returns of scale or whatever. 00:18:04.365 --> 00:18:07.240 align:middle line:90% 00:18:07.240 --> 00:18:13.350 align:middle line:84% So your choice is, I'm going to invest capital k in project p. 00:18:13.350 --> 00:18:15.540 align:middle line:84% So I'm going to make two choices, a k and a p, 00:18:15.540 --> 00:18:18.660 align:middle line:84% and my return is going to be the-- the p is going 00:18:18.660 --> 00:18:20.790 align:middle line:84% to choose the expected p, and then I'm 00:18:20.790 --> 00:18:22.560 align:middle line:84% going to scale it up by this f of k. 00:18:22.560 --> 00:18:25.150 align:middle line:90% 00:18:25.150 --> 00:18:26.470 align:middle line:90% OK, is that clear? 00:18:26.470 --> 00:18:28.900 align:middle line:90% So I could go into making-- 00:18:28.900 --> 00:18:31.180 align:middle line:84% I don't know-- belts or t-shirts, that's p. 00:18:31.180 --> 00:18:33.520 align:middle line:84% And then I can choose how many of them I want to do. 00:18:33.520 --> 00:18:34.420 align:middle line:90% That's f of k. 00:18:34.420 --> 00:18:36.980 align:middle line:90% 00:18:36.980 --> 00:18:37.480 align:middle line:90% Clear? 00:18:37.480 --> 00:18:40.530 align:middle line:90% 00:18:40.530 --> 00:18:47.600 align:middle line:84% Finally, suppose that the investor has some wealth, w. 00:18:47.600 --> 00:18:50.885 align:middle line:84% So if the investor wants to borrow-- 00:18:50.885 --> 00:18:55.430 align:middle line:84% wants to invest k, she has to borrow k minus w. 00:18:55.430 --> 00:18:57.080 align:middle line:90% So that's the credit. 00:18:57.080 --> 00:18:57.950 align:middle line:90% That's credit. 00:18:57.950 --> 00:19:01.070 align:middle line:84% It's how much they have to borrow. 00:19:01.070 --> 00:19:03.220 align:middle line:90% The gross industry is r. 00:19:03.220 --> 00:19:05.150 align:middle line:84% And we'll have limited liability, 00:19:05.150 --> 00:19:09.310 align:middle line:84% which is, if you can't pay, what can they do? 00:19:09.310 --> 00:19:11.110 align:middle line:90% There's some collateral. 00:19:11.110 --> 00:19:13.450 align:middle line:84% Another way to think about is, you put up your wealth 00:19:13.450 --> 00:19:16.300 align:middle line:84% as collateral, they can take your wealth but nothing else. 00:19:16.300 --> 00:19:19.510 align:middle line:90% 00:19:19.510 --> 00:19:21.850 align:middle line:84% So I'll just tell you, that is important 00:19:21.850 --> 00:19:27.910 align:middle line:84% because you can see why this is going to matter. 00:19:27.910 --> 00:19:34.050 align:middle line:84% So if I am investing k, and my wealth, w, is approximately 00:19:34.050 --> 00:19:36.030 align:middle line:84% equal to k so I'm only borrowing a little bit, 00:19:36.030 --> 00:19:40.620 align:middle line:84% then if my project kind of fails, I'm going to lose a lot. 00:19:40.620 --> 00:19:42.630 align:middle line:84% They're going to take almost my entire wealth. 00:19:42.630 --> 00:19:47.370 align:middle line:84% Whereas if my wealth is, say, zero, then if my product fails, 00:19:47.370 --> 00:19:49.560 align:middle line:84% I get nothing in that state so I'm only 00:19:49.560 --> 00:19:51.540 align:middle line:90% interested in the upside. 00:19:51.540 --> 00:19:55.860 align:middle line:84% So the amount of wealth, the amount of collateral I have, 00:19:55.860 --> 00:20:02.160 align:middle line:84% is determining what share of the downside 00:20:02.160 --> 00:20:06.660 align:middle line:84% of this project I have to bear versus the bank has to bear. 00:20:06.660 --> 00:20:10.890 align:middle line:84% And this idea is, you can immediately 00:20:10.890 --> 00:20:15.480 align:middle line:84% see why it's going to be nice for banks 00:20:15.480 --> 00:20:18.240 align:middle line:84% to want you to have some collateral. 00:20:18.240 --> 00:20:19.800 align:middle line:84% And particularly why they're going 00:20:19.800 --> 00:20:20.940 align:middle line:84% to want you to put some of your own money 00:20:20.940 --> 00:20:22.892 align:middle line:84% into projects is because they want make sure 00:20:22.892 --> 00:20:24.850 align:middle line:84% that you're bearing some of the downside risks, 00:20:24.850 --> 00:20:26.308 align:middle line:84% that you're not taking-- otherwise, 00:20:26.308 --> 00:20:29.280 align:middle line:84% you take super risky projects because heads, you win, tails, 00:20:29.280 --> 00:20:30.662 align:middle line:90% they lose. 00:20:30.662 --> 00:20:32.370 align:middle line:84% That's not a very good deal for the bank. 00:20:32.370 --> 00:20:34.880 align:middle line:90% 00:20:34.880 --> 00:20:37.708 align:middle line:84% And so lots of loans have this feature. 00:20:37.708 --> 00:20:39.500 align:middle line:84% So for example, if you want to buy a house, 00:20:39.500 --> 00:20:42.590 align:middle line:84% you need to put 20% down on the house or something. 00:20:42.590 --> 00:20:45.950 align:middle line:84% That's basically saying, they want the loan 00:20:45.950 --> 00:20:48.020 align:middle line:84% to be no more than 80% of the total 00:20:48.020 --> 00:20:50.420 align:middle line:84% because they want you to have some stake in what 00:20:50.420 --> 00:20:51.680 align:middle line:90% happens if things go wrong. 00:20:51.680 --> 00:20:56.170 align:middle line:90% 00:20:56.170 --> 00:20:59.400 align:middle line:84% There are more complicated situations in mortgage markets 00:20:59.400 --> 00:21:03.192 align:middle line:84% actually because in addition, often when the loan goes bad, 00:21:03.192 --> 00:21:04.150 align:middle line:90% it starts from the top. 00:21:04.150 --> 00:21:09.750 align:middle line:84% And so if you have 20% down, the first 20% is your loss, 00:21:09.750 --> 00:21:11.548 align:middle line:84% and the bank only loses it more than that. 00:21:11.548 --> 00:21:13.590 align:middle line:84% So actually, the ordering of it is important too, 00:21:13.590 --> 00:21:16.730 align:middle line:84% but we're not going to get into that in this model. 00:21:16.730 --> 00:21:18.960 align:middle line:90% OK, is this clear so far? 00:21:18.960 --> 00:21:27.860 align:middle line:84% OK, so suppose that we can contract over k the amount 00:21:27.860 --> 00:21:31.930 align:middle line:84% of capital, but we cannot contract over the project 00:21:31.930 --> 00:21:32.895 align:middle line:90% choice, p. 00:21:32.895 --> 00:21:37.760 align:middle line:90% 00:21:37.760 --> 00:21:43.640 align:middle line:84% So k is going to be fixed, but then the borrower is going to-- 00:21:43.640 --> 00:21:47.420 align:middle line:84% for a given k, once we agree on k, 00:21:47.420 --> 00:21:48.920 align:middle line:84% the borrower gets to say, OK, now 00:21:48.920 --> 00:21:50.930 align:middle line:84% I know what my k is, I'm going to choose my p. 00:21:50.930 --> 00:21:53.932 align:middle line:84% And they can't contract over that. 00:21:53.932 --> 00:21:55.890 align:middle line:84% So in particular, what are they going to solve? 00:21:55.890 --> 00:21:59.460 align:middle line:84% Well, with probability, p, they're 00:21:59.460 --> 00:22:02.620 align:middle line:84% going to get f of k times the return. 00:22:02.620 --> 00:22:05.220 align:middle line:84% So the f of k times r of p-- that's how good the product 00:22:05.220 --> 00:22:05.880 align:middle line:90% is-- 00:22:05.880 --> 00:22:09.030 align:middle line:90% times the scale. 00:22:09.030 --> 00:22:11.250 align:middle line:84% And then they have to pay the interest 00:22:11.250 --> 00:22:12.420 align:middle line:90% on the amount they borrowed. 00:22:12.420 --> 00:22:17.890 align:middle line:90% 00:22:17.890 --> 00:22:21.460 align:middle line:84% With probability 1 minus p, the bank 00:22:21.460 --> 00:22:24.260 align:middle line:84% takes their wealth and nothing else. 00:22:24.260 --> 00:22:28.180 align:middle line:84% So another way, you can also equivalently rewrite 00:22:28.180 --> 00:22:31.930 align:middle line:84% this as saying, they paid the gross interest rate on k 00:22:31.930 --> 00:22:36.760 align:middle line:84% in all periods, and then they lose their wealth 00:22:36.760 --> 00:22:38.830 align:middle line:90% in this other period. 00:22:38.830 --> 00:22:42.490 align:middle line:90% 00:22:42.490 --> 00:22:52.440 align:middle line:84% So the first order condition is that the f of k times e prime-- 00:22:52.440 --> 00:22:55.380 align:middle line:84% e prime is the derivative of the expected value with respect 00:22:55.380 --> 00:22:56.160 align:middle line:90% to p-- 00:22:56.160 --> 00:23:01.680 align:middle line:84% has to be equal to the effective r times k minus w. 00:23:01.680 --> 00:23:04.800 align:middle line:90% This part doesn't matter. 00:23:04.800 --> 00:23:08.550 align:middle line:84% And so we're just going to get-- so p times r, 00:23:08.550 --> 00:23:11.050 align:middle line:90% that's e, so the derivative. 00:23:11.050 --> 00:23:14.085 align:middle line:84% So p times r was e, the expected return, so it's just f prime-- 00:23:14.085 --> 00:23:18.390 align:middle line:84% I'm sorry-- f of k times e prime of p is equal to-- 00:23:18.390 --> 00:23:20.400 align:middle line:84% there's a p in here, so that's just 00:23:20.400 --> 00:23:23.660 align:middle line:90% equal to r times k minus w. 00:23:23.660 --> 00:23:28.080 align:middle line:84% OK, so that's the first order condition. 00:23:28.080 --> 00:23:37.930 align:middle line:84% So recall that the optimal p had e prime of p equal to zero. 00:23:37.930 --> 00:23:40.870 align:middle line:84% That was a socially efficient level of p. 00:23:40.870 --> 00:23:48.620 align:middle line:84% So what you can see is that whenever 00:23:48.620 --> 00:23:54.320 align:middle line:84% w is less than k, whenever you're borrowing something-- 00:23:54.320 --> 00:23:56.010 align:middle line:84% so if w is equal to k, by the way, 00:23:56.010 --> 00:23:57.385 align:middle line:84% if it's all your own capital, you 00:23:57.385 --> 00:24:00.000 align:middle line:84% pick the socially efficient level of a socially efficient 00:24:00.000 --> 00:24:00.500 align:middle line:90% project. 00:24:00.500 --> 00:24:01.000 align:middle line:90% Why? 00:24:01.000 --> 00:24:02.630 align:middle line:84% Because the cost and the benefits 00:24:02.630 --> 00:24:04.740 align:middle line:90% you totally internalize. 00:24:04.740 --> 00:24:06.960 align:middle line:84% Whenever we have a wedge where you're 00:24:06.960 --> 00:24:09.810 align:middle line:84% gaining more of the upside than downside, 00:24:09.810 --> 00:24:13.570 align:middle line:84% that's going to lead you to take riskier projects. 00:24:13.570 --> 00:24:16.540 align:middle line:90% And how do we see that? 00:24:16.540 --> 00:24:19.000 align:middle line:84% We see that basically that's because e 00:24:19.000 --> 00:24:22.675 align:middle line:90% prime is greater than zero. 00:24:22.675 --> 00:24:27.040 align:middle line:90% 00:24:27.040 --> 00:24:32.800 align:middle line:84% Because in order to make this thing positive, the-- 00:24:32.800 --> 00:24:34.718 align:middle line:90% oh, no, sorry. 00:24:34.718 --> 00:24:36.010 align:middle line:90% You're going to take more risk. 00:24:36.010 --> 00:24:36.510 align:middle line:90% Sorry. 00:24:36.510 --> 00:24:38.230 align:middle line:84% Because e double prime is less than zero, 00:24:38.230 --> 00:24:40.870 align:middle line:84% we're going to take a little bit more 00:24:40.870 --> 00:24:42.927 align:middle line:84% risk in order to make this thing positive. 00:24:42.927 --> 00:24:45.910 align:middle line:90% 00:24:45.910 --> 00:24:49.510 align:middle line:84% So the optimal project is going to be 00:24:49.510 --> 00:24:54.850 align:middle line:84% a little bit riskier in order to make e prime positive. 00:24:54.850 --> 00:24:57.010 align:middle line:84% And that's what I was intuitively saying before. 00:24:57.010 --> 00:24:58.750 align:middle line:84% Whenever you're not perfect, whenever 00:24:58.750 --> 00:25:01.120 align:middle line:84% you're getting more of the upside than the downside, 00:25:01.120 --> 00:25:03.490 align:middle line:84% that's going to lead you to take too much risk 00:25:03.490 --> 00:25:06.760 align:middle line:90% than is socially efficient. 00:25:06.760 --> 00:25:09.420 align:middle line:90% Is that clear? 00:25:09.420 --> 00:25:11.222 align:middle line:90% OK. 00:25:11.222 --> 00:25:12.430 align:middle line:90% That's basically what I said. 00:25:12.430 --> 00:25:15.030 align:middle line:90% 00:25:15.030 --> 00:25:15.980 align:middle line:90% OK, clear so far? 00:25:15.980 --> 00:25:17.370 align:middle line:90% Yeah? 00:25:17.370 --> 00:25:19.170 align:middle line:90% All right. 00:25:19.170 --> 00:25:23.810 align:middle line:84% So this already gives us some comparative statics. 00:25:23.810 --> 00:25:26.620 align:middle line:84% So this is the condition that I just derived on the last slide, 00:25:26.620 --> 00:25:33.890 align:middle line:84% that f of k times e prime of p was equal to r k minus w. 00:25:33.890 --> 00:25:40.370 align:middle line:84% So we can just divide through by a-- 00:25:40.370 --> 00:25:42.680 align:middle line:84% if we can pull out a k and make this 1 minus 00:25:42.680 --> 00:25:44.995 align:middle line:84% w over k, divide through by f of k, 00:25:44.995 --> 00:25:47.120 align:middle line:84% and we get the second expression here. e prime of p 00:25:47.120 --> 00:25:51.140 align:middle line:84% is equal to k over f of k times r times 1 minus w over k. 00:25:51.140 --> 00:25:53.930 align:middle line:84% So why is writing things in that way helpful? 00:25:53.930 --> 00:25:56.030 align:middle line:84% It's because we can immediately start to see 00:25:56.030 --> 00:25:59.520 align:middle line:90% a few comparative statics. 00:25:59.520 --> 00:26:04.100 align:middle line:84% So the first one is, we can see how does p prime depend 00:26:04.100 --> 00:26:05.030 align:middle line:90% on interest rate. 00:26:05.030 --> 00:26:06.110 align:middle line:90% That's our p hat. 00:26:06.110 --> 00:26:08.390 align:middle line:84% How does the equilibrium choice of project 00:26:08.390 --> 00:26:10.740 align:middle line:90% depend on the interest rate? 00:26:10.740 --> 00:26:14.270 align:middle line:84% So when interest rate goes up, e prime goes up, 00:26:14.270 --> 00:26:16.370 align:middle line:90% therefore p goes down. 00:26:16.370 --> 00:26:17.930 align:middle line:84% So when the interest rate is higher, 00:26:17.930 --> 00:26:21.250 align:middle line:84% we're going to be choosing riskier projects, again, 00:26:21.250 --> 00:26:22.750 align:middle line:84% assuming that we're in a world where 00:26:22.750 --> 00:26:24.400 align:middle line:90% you're borrowing something. 00:26:24.400 --> 00:26:27.508 align:middle line:84% Obviously, if w is equal to k, then we're 00:26:27.508 --> 00:26:29.050 align:middle line:84% just at the socially efficient level. 00:26:29.050 --> 00:26:31.210 align:middle line:84% But assuming w is less than k, we 00:26:31.210 --> 00:26:33.010 align:middle line:84% have that higher interest rates are going 00:26:33.010 --> 00:26:36.730 align:middle line:90% to lead to riskier projects. 00:26:36.730 --> 00:26:37.300 align:middle line:90% Clear? 00:26:37.300 --> 00:26:39.060 align:middle line:84% By the way, stop me if I'm going too fast. 00:26:39.060 --> 00:26:41.230 align:middle line:84% Sometimes when you go through a model on the slides, 00:26:41.230 --> 00:26:42.260 align:middle line:90% it's too fast. 00:26:42.260 --> 00:26:45.010 align:middle line:84% But if anything, just slow this down. 00:26:45.010 --> 00:26:47.297 align:middle line:84% STUDENT: Higher interest is called higher r? 00:26:47.297 --> 00:26:48.130 align:middle line:90% PROFESSOR: Higher r. 00:26:48.130 --> 00:26:50.950 align:middle line:84% STUDENT: And higher p is higher risk? 00:26:50.950 --> 00:26:52.790 align:middle line:84% PROFESSOR: No, lower p is higher risk. 00:26:52.790 --> 00:26:56.020 align:middle line:84% So higher r means higher e prime. 00:26:56.020 --> 00:26:59.540 align:middle line:84% But we have e is concave, so we're 00:26:59.540 --> 00:27:07.456 align:middle line:84% going to go we're going to go down to get a positive. 00:27:07.456 --> 00:27:08.820 align:middle line:90% STUDENT: [INAUDIBLE]. 00:27:08.820 --> 00:27:11.075 align:middle line:84% PROFESSOR: Yeah, it's the same thing 00:27:11.075 --> 00:27:14.763 align:middle line:84% because f prime is concave, and the higher marginal product 00:27:14.763 --> 00:27:15.930 align:middle line:90% means you have less capital. 00:27:15.930 --> 00:27:16.430 align:middle line:90% Same idea. 00:27:16.430 --> 00:27:20.108 align:middle line:84% Basically, we have higher levels of e prime. 00:27:20.108 --> 00:27:21.900 align:middle line:84% We're going to get higher levels of e prime 00:27:21.900 --> 00:27:26.105 align:middle line:84% with lower levels of p because e prime was concave. 00:27:26.105 --> 00:27:27.950 align:middle line:90% Sorry, e was concave. 00:27:27.950 --> 00:27:31.700 align:middle line:84% Similarly, we can say, how does p depend on the returns. 00:27:31.700 --> 00:27:35.480 align:middle line:90% Returns is f of k. 00:27:35.480 --> 00:27:40.490 align:middle line:84% F of k divided by k is, what is the return on capital right. 00:27:40.490 --> 00:27:45.080 align:middle line:84% Like you invest k, and you get back f of k, 00:27:45.080 --> 00:27:49.560 align:middle line:84% so this is like the average return on capital. 00:27:49.560 --> 00:27:52.093 align:middle line:84% This is going to be increasing for the same reason 00:27:52.093 --> 00:27:53.260 align:middle line:90% because this is 1 over that. 00:27:53.260 --> 00:27:55.830 align:middle line:90% So it all goes backwards. 00:27:55.830 --> 00:28:02.550 align:middle line:84% So higher returns leads you to choose higher probability 00:28:02.550 --> 00:28:04.060 align:middle line:90% projects. 00:28:04.060 --> 00:28:06.135 align:middle line:84% And this one, I think, is-- these two 00:28:06.135 --> 00:28:08.260 align:middle line:84% are certainly going to be the most relevant for us. 00:28:08.260 --> 00:28:12.070 align:middle line:84% The third one is, how does p hat depend on k over w, 00:28:12.070 --> 00:28:15.234 align:middle line:84% where k over w is leverage. k over w is-- 00:28:15.234 --> 00:28:19.000 align:middle line:90% obviously, this is 1 over k. 00:28:19.000 --> 00:28:22.922 align:middle line:84% How does it depend on what fraction you're borrowing? 00:28:22.922 --> 00:28:24.880 align:middle line:84% Well, we know that it's going to be decreasing. 00:28:24.880 --> 00:28:28.960 align:middle line:84% Again, we know that when k is equal to w, 00:28:28.960 --> 00:28:30.440 align:middle line:90% we're at the optimal level. 00:28:30.440 --> 00:28:31.990 align:middle line:84% And as we start to increase k, we're 00:28:31.990 --> 00:28:33.415 align:middle line:90% going to be choosing risk here. 00:28:33.415 --> 00:28:35.290 align:middle line:84% Holding w at face, we're going to be choosing 00:28:35.290 --> 00:28:36.498 align:middle line:90% riskier and riskier projects. 00:28:36.498 --> 00:28:39.080 align:middle line:90% 00:28:39.080 --> 00:28:40.220 align:middle line:90% Clear? 00:28:40.220 --> 00:28:43.950 align:middle line:84% STUDENT: What's the intuition for the second point? 00:28:43.950 --> 00:28:46.640 align:middle line:84% PROFESSOR: So I think it's a similar one, that 00:28:46.640 --> 00:28:49.500 align:middle line:90% basically the-- 00:28:49.500 --> 00:28:51.188 align:middle line:90% 00:28:51.188 --> 00:28:53.730 align:middle line:84% actually, I don't have a good intuition for the second point. 00:28:53.730 --> 00:28:55.397 align:middle line:84% Does anyone else want to give it a shot? 00:28:55.397 --> 00:28:56.590 align:middle line:90% Sorry, I take it back. 00:28:56.590 --> 00:28:57.590 align:middle line:90% I don't have a good one. 00:28:57.590 --> 00:28:58.730 align:middle line:84% It's a little hard to think about because you're 00:28:58.730 --> 00:28:59.855 align:middle line:90% holding w over k fixed. 00:28:59.855 --> 00:29:05.423 align:middle line:90% 00:29:05.423 --> 00:29:06.090 align:middle line:90% Hold on one sec. 00:29:06.090 --> 00:29:08.007 align:middle line:84% Let me see if I can give you a good intuition. 00:29:08.007 --> 00:29:12.940 align:middle line:90% 00:29:12.940 --> 00:29:13.440 align:middle line:90% I'm sorry. 00:29:13.440 --> 00:29:14.857 align:middle line:84% No, I don't have a good intuition. 00:29:14.857 --> 00:29:15.830 align:middle line:90% Let me think about it. 00:29:15.830 --> 00:29:16.830 align:middle line:90% Anyone else want to try? 00:29:16.830 --> 00:29:20.274 align:middle line:90% 00:29:20.274 --> 00:29:22.540 align:middle line:84% Ed, do you have a good intuition for it? 00:29:22.540 --> 00:29:23.040 align:middle line:90% No? 00:29:23.040 --> 00:29:24.332 align:middle line:90% Sorry, I should think about it. 00:29:24.332 --> 00:29:26.130 align:middle line:84% Sorry I don't have a good answer for it. 00:29:26.130 --> 00:29:26.630 align:middle line:90% OK. 00:29:26.630 --> 00:29:28.340 align:middle line:84% But luckily, I'm going to mostly talk 00:29:28.340 --> 00:29:29.715 align:middle line:84% about r and the other ones, which 00:29:29.715 --> 00:29:31.370 align:middle line:90% I do have a good intuition for. 00:29:31.370 --> 00:29:32.630 align:middle line:90% Yeah, Hazel? 00:29:32.630 --> 00:29:35.275 align:middle line:84% STUDENT: So we're assuming that the borrower is risk neutral, 00:29:35.275 --> 00:29:37.860 align:middle line:84% but if we assume risk aversion, would we-- 00:29:37.860 --> 00:29:39.585 align:middle line:84% so like taking too much risk-- so 00:29:39.585 --> 00:29:41.257 align:middle line:84% if we assume that they were risk averse, 00:29:41.257 --> 00:29:42.840 align:middle line:84% could we get rid of that in the model? 00:29:42.840 --> 00:29:47.520 align:middle line:90% 00:29:47.520 --> 00:29:49.470 align:middle line:84% PROFESSOR: If we assume they're risk averse, 00:29:49.470 --> 00:29:54.240 align:middle line:84% it's going to lead them to take less risk. 00:29:54.240 --> 00:29:54.740 align:middle line:90% Hold on. 00:29:54.740 --> 00:29:55.130 align:middle line:90% Let me think. 00:29:55.130 --> 00:29:55.630 align:middle line:90% What? 00:29:55.630 --> 00:29:56.138 align:middle line:90% Sorry. 00:29:56.138 --> 00:29:57.408 align:middle line:90% STUDENT: Sorry. 00:29:57.408 --> 00:29:58.700 align:middle line:90% PROFESSOR: No, go ahead, Hazel. 00:29:58.700 --> 00:30:00.690 align:middle line:84% STUDENT: Because they were taking too much risk before, 00:30:00.690 --> 00:30:00.890 align:middle line:90% right? 00:30:00.890 --> 00:30:02.200 align:middle line:84% And then if you introduce risk aversion, 00:30:02.200 --> 00:30:03.480 align:middle line:90% they might [INAUDIBLE]. 00:30:03.480 --> 00:30:04.267 align:middle line:90% PROFESSOR: Yes. 00:30:04.267 --> 00:30:05.350 align:middle line:90% That clearly sounds right. 00:30:05.350 --> 00:30:08.812 align:middle line:90% 00:30:08.812 --> 00:30:10.770 align:middle line:84% I think it's still going to be the case they're 00:30:10.770 --> 00:30:11.395 align:middle line:90% going to take-- 00:30:11.395 --> 00:30:19.240 align:middle line:90% 00:30:19.240 --> 00:30:19.960 align:middle line:90% so yes. 00:30:19.960 --> 00:30:21.050 align:middle line:84% For sure, if they're risk averse, 00:30:21.050 --> 00:30:21.790 align:middle line:90% they're going to take less risk. 00:30:21.790 --> 00:30:23.350 align:middle line:84% That's going to shift the whole thing in the direction of less 00:30:23.350 --> 00:30:24.520 align:middle line:90% riskiness. 00:30:24.520 --> 00:30:28.030 align:middle line:84% And you could get that maybe the optimal thing is to get e star, 00:30:28.030 --> 00:30:29.860 align:middle line:84% but e star may no longer be-- the same p 00:30:29.860 --> 00:30:31.270 align:middle line:84% star may no longer be optimal because it's 00:30:31.270 --> 00:30:32.510 align:middle line:90% no longer optimal for them. 00:30:32.510 --> 00:30:35.010 align:middle line:84% So I think it's still going to be the case that if you think 00:30:35.010 --> 00:30:37.870 align:middle line:84% about socially optimal including the risk 00:30:37.870 --> 00:30:39.590 align:middle line:84% aversion, like the riskiness to them, 00:30:39.590 --> 00:30:40.600 align:middle line:84% it's still going be the case they're going to take 00:30:40.600 --> 00:30:41.892 align:middle line:90% too much risk relative to that. 00:30:41.892 --> 00:30:46.230 align:middle line:90% 00:30:46.230 --> 00:30:47.240 align:middle line:90% But yes, you're right. 00:30:47.240 --> 00:30:48.590 align:middle line:84% Certainly, if we add risk aversion, 00:30:48.590 --> 00:30:49.790 align:middle line:84% they're going to take less risk, and that's going 00:30:49.790 --> 00:30:50.957 align:middle line:90% to shift the whole thing up. 00:30:50.957 --> 00:30:57.200 align:middle line:90% 00:30:57.200 --> 00:30:59.900 align:middle line:84% STUDENT: Intuition is just to care about f times e. 00:30:59.900 --> 00:31:01.458 align:middle line:90% And if f goes up-- 00:31:01.458 --> 00:31:02.750 align:middle line:90% PROFESSOR: You care about what? 00:31:02.750 --> 00:31:04.322 align:middle line:84% STUDENT: You care about f times e 00:31:04.322 --> 00:31:06.070 align:middle line:84% to [INAUDIBLE] objective function. 00:31:06.070 --> 00:31:12.070 align:middle line:84% If f goes up, then you're more invested in the return 00:31:12.070 --> 00:31:14.230 align:middle line:90% once your repayment [INAUDIBLE]. 00:31:14.230 --> 00:31:16.150 align:middle line:90% PROFESSOR: Oh, I see. 00:31:16.150 --> 00:31:17.237 align:middle line:90% This relative to that? 00:31:17.237 --> 00:31:17.820 align:middle line:90% STUDENT: Yeah. 00:31:17.820 --> 00:31:19.045 align:middle line:90% PROFESSOR: OK, there you go. 00:31:19.045 --> 00:31:20.170 align:middle line:90% Yeah, I think you're right. 00:31:20.170 --> 00:31:23.110 align:middle line:90% 00:31:23.110 --> 00:31:23.880 align:middle line:90% OK, thanks, Ed. 00:31:23.880 --> 00:31:28.162 align:middle line:90% 00:31:28.162 --> 00:31:29.620 align:middle line:84% The other thing I want to note here 00:31:29.620 --> 00:31:30.580 align:middle line:84% that comes out of this model, which 00:31:30.580 --> 00:31:32.710 align:middle line:84% is out of many of these models, is there's 00:31:32.710 --> 00:31:35.440 align:middle line:90% a positive probability-- 00:31:35.440 --> 00:31:37.990 align:middle line:84% there's a positive correlation between the default 00:31:37.990 --> 00:31:41.620 align:middle line:84% probability, which is 1 minus p, and the interest rate. 00:31:41.620 --> 00:31:43.930 align:middle line:84% So in general, we had higher interest rates 00:31:43.930 --> 00:31:46.840 align:middle line:84% lead to riskier projects, and we get that in this model. 00:31:46.840 --> 00:31:48.340 align:middle line:84% That was this prediction over here. 00:31:48.340 --> 00:31:52.840 align:middle line:84% And that positive correlation between the interest rate 00:31:52.840 --> 00:31:57.910 align:middle line:84% and the riskiness is a classic prediction 00:31:57.910 --> 00:32:01.060 align:middle line:84% of many of these kinds of asymmetric information models, 00:32:01.060 --> 00:32:05.250 align:middle line:84% including more hazard models like this one. 00:32:05.250 --> 00:32:06.000 align:middle line:90% Clear so far? 00:32:06.000 --> 00:32:08.810 align:middle line:90% 00:32:08.810 --> 00:32:10.460 align:middle line:84% So the final thing that we need to do 00:32:10.460 --> 00:32:18.990 align:middle line:84% is, we need to add an exogenous supply of funds for the lender. 00:32:18.990 --> 00:32:21.980 align:middle line:84% So the lender can go out to the capital markets-- 00:32:21.980 --> 00:32:23.760 align:middle line:84% call it Wall Street or whatever-- 00:32:23.760 --> 00:32:28.270 align:middle line:84% and they can borrow at rate, rho. 00:32:28.270 --> 00:32:34.420 align:middle line:84% So they're only getting repaid, of course, with probability, p. 00:32:34.420 --> 00:32:37.810 align:middle line:84% So their equilibrium condition is 00:32:37.810 --> 00:32:40.540 align:middle line:84% that their cost of funds, rho, has 00:32:40.540 --> 00:32:43.660 align:middle line:84% to be equal to their expected return. 00:32:43.660 --> 00:32:45.130 align:middle line:90% With probably p, they get r. 00:32:45.130 --> 00:32:48.430 align:middle line:84% With probability 1 minus p, they don't. 00:32:48.430 --> 00:32:52.120 align:middle line:84% And so that's going to determine their return. 00:32:52.120 --> 00:32:54.178 align:middle line:90% 00:32:54.178 --> 00:32:56.220 align:middle line:84% So what is it going to look like for equilibrium? 00:32:56.220 --> 00:32:59.175 align:middle line:84% So there's two different things going on. 00:32:59.175 --> 00:33:00.800 align:middle line:84% The first would be this market clearing 00:33:00.800 --> 00:33:03.950 align:middle line:84% thing, which is that rho is equal to p times r. 00:33:03.950 --> 00:33:05.860 align:middle line:90% So this is r, and this is p. 00:33:05.860 --> 00:33:10.240 align:middle line:84% And then we also derive from the previous model 00:33:10.240 --> 00:33:12.700 align:middle line:84% this moral hazard kind of equilibrium, which also 00:33:12.700 --> 00:33:16.160 align:middle line:90% gives you p as a function of r. 00:33:16.160 --> 00:33:18.427 align:middle line:84% So as I've drawn here, there's a single equilibrium. 00:33:18.427 --> 00:33:20.510 align:middle line:84% This is going to be the equilibrium in this market 00:33:20.510 --> 00:33:24.120 align:middle line:84% where we're going to have projects, p, down here 00:33:24.120 --> 00:33:27.290 align:middle line:90% and this interest rate. 00:33:27.290 --> 00:33:28.190 align:middle line:90% OK. 00:33:28.190 --> 00:33:32.130 align:middle line:90% But this thing is-- 00:33:32.130 --> 00:33:34.500 align:middle line:84% I mean, obviously this is not drawn exactly right-- 00:33:34.500 --> 00:33:38.430 align:middle line:84% but this function here, I guess this hyperbola, 00:33:38.430 --> 00:33:41.820 align:middle line:84% given by this rho equals pr, that's fixed. 00:33:41.820 --> 00:33:44.400 align:middle line:84% But the shape of this particular p hat of r 00:33:44.400 --> 00:33:46.710 align:middle line:84% is not obvious what it's going to be. 00:33:46.710 --> 00:33:49.740 align:middle line:84% It's going to depend on the underlying things that 00:33:49.740 --> 00:33:51.870 align:middle line:90% went into that previous model. 00:33:51.870 --> 00:33:54.780 align:middle line:84% In particular, it's going to depend on this e function, 00:33:54.780 --> 00:33:56.040 align:middle line:90% which could be-- 00:33:56.040 --> 00:33:58.600 align:middle line:84% we haven't specified what that's going to be. 00:33:58.600 --> 00:34:02.070 align:middle line:84% So we could draw it like this with a single equilibrium. 00:34:02.070 --> 00:34:06.375 align:middle line:84% It could look like this, which has-- 00:34:06.375 --> 00:34:08.250 align:middle line:84% I mean, we know that it has to be decreasing, 00:34:08.250 --> 00:34:11.489 align:middle line:84% but the shape of this thing is not clear. 00:34:11.489 --> 00:34:13.530 align:middle line:84% So if I've drawn it like this, there 00:34:13.530 --> 00:34:20.330 align:middle line:84% are three equilibria, where these are both, I think, 00:34:20.330 --> 00:34:23.406 align:middle line:90% the stable equilibria. 00:34:23.406 --> 00:34:25.239 align:middle line:84% And here, there could be two equi-- so here, 00:34:25.239 --> 00:34:26.620 align:middle line:84% there could be two equilibria in this credit market. 00:34:26.620 --> 00:34:28.787 align:middle line:84% We could have one where we have really high interest 00:34:28.787 --> 00:34:31.637 align:middle line:84% rates and very risky projects, or we 00:34:31.637 --> 00:34:34.179 align:middle line:84% can have very low interest rates and not very risky projects. 00:34:34.179 --> 00:34:37.472 align:middle line:84% Both of those could be equilibria, 00:34:37.472 --> 00:34:38.889 align:middle line:84% but it's a multiequilibrium model. 00:34:38.889 --> 00:34:42.628 align:middle line:90% 00:34:42.628 --> 00:34:44.420 align:middle line:84% Or you can maybe have a no lend equilibrium 00:34:44.420 --> 00:34:48.469 align:middle line:84% where, again, you have to draw both of these things going out 00:34:48.469 --> 00:34:49.580 align:middle line:90% like this. 00:34:49.580 --> 00:34:51.320 align:middle line:84% But it's not obvious these things are 00:34:51.320 --> 00:34:53.770 align:middle line:90% going to start to intersect. 00:34:53.770 --> 00:34:56.635 align:middle line:84% And so in this model, the credit market 00:34:56.635 --> 00:34:59.530 align:middle line:84% could collapse entirely if there is no intersection 00:34:59.530 --> 00:35:02.250 align:middle line:90% between these two points. 00:35:02.250 --> 00:35:02.750 align:middle line:90% OK. 00:35:02.750 --> 00:35:06.510 align:middle line:90% 00:35:06.510 --> 00:35:09.630 align:middle line:84% So this already gives us some important things going on. 00:35:09.630 --> 00:35:11.290 align:middle line:84% It can already be an explanation, 00:35:11.290 --> 00:35:13.950 align:middle line:84% for example, why you might have multiple different things going 00:35:13.950 --> 00:35:14.450 align:middle line:90% on. 00:35:14.450 --> 00:35:16.710 align:middle line:84% You could have really high interest rates and really 00:35:16.710 --> 00:35:19.410 align:middle line:90% terrible projects, for example. 00:35:19.410 --> 00:35:22.560 align:middle line:84% Or sometimes you can have credit markets that collapse entirely. 00:35:22.560 --> 00:35:26.235 align:middle line:84% And it just depends on the shape of that p hat of our function. 00:35:26.235 --> 00:35:28.770 align:middle line:90% 00:35:28.770 --> 00:35:30.860 align:middle line:90% OK. 00:35:30.860 --> 00:35:33.500 align:middle line:90% What's happening with capital? 00:35:33.500 --> 00:35:37.376 align:middle line:84% Well, you can think of us as having derived that optimal p 00:35:37.376 --> 00:35:43.160 align:middle line:84% hat function, which is a function of the returns, 00:35:43.160 --> 00:35:46.160 align:middle line:84% the leverage ratio, and the interest rate, which 00:35:46.160 --> 00:35:48.770 align:middle line:84% is in itself an endogenous function of k. 00:35:48.770 --> 00:35:52.090 align:middle line:90% 00:35:52.090 --> 00:35:58.300 align:middle line:84% So now, the bank and the lender know 00:35:58.300 --> 00:36:03.340 align:middle line:84% that conditional on a level k and an interest rate, 00:36:03.340 --> 00:36:06.010 align:middle line:84% they're going to pick project, p. 00:36:06.010 --> 00:36:07.510 align:middle line:84% They can't contract over p, but they 00:36:07.510 --> 00:36:09.640 align:middle line:90% know what that p function is. 00:36:09.640 --> 00:36:12.400 align:middle line:84% And so they can make a choice collectively 00:36:12.400 --> 00:36:15.850 align:middle line:84% like, how much capital would you like to borrow. 00:36:15.850 --> 00:36:17.110 align:middle line:90% So then I can solve. 00:36:17.110 --> 00:36:24.650 align:middle line:84% I can maximize over capital the return here, 00:36:24.650 --> 00:36:31.010 align:middle line:84% where I'm going to maximize the amount I 00:36:31.010 --> 00:36:34.220 align:middle line:84% get times the expected return minus the amount I 00:36:34.220 --> 00:36:35.802 align:middle line:90% have to pay back. 00:36:35.802 --> 00:36:37.760 align:middle line:84% And here, I've just substituted in for the fact 00:36:37.760 --> 00:36:41.190 align:middle line:84% that in equilibrium, I have to pay back rho. 00:36:41.190 --> 00:36:47.070 align:middle line:84% So with probability, p, I have to pay back r times-- 00:36:47.070 --> 00:36:49.230 align:middle line:84% I have to pay back r, so I know this is just pr, 00:36:49.230 --> 00:36:56.100 align:middle line:84% so I just substitute it for rho, the market amount. 00:36:56.100 --> 00:36:57.660 align:middle line:90% OK. 00:36:57.660 --> 00:36:58.160 align:middle line:90% What? 00:36:58.160 --> 00:36:58.820 align:middle line:90% Sorry. 00:36:58.820 --> 00:37:01.448 align:middle line:84% STUDENT: Is it still the same borrower making this decision, 00:37:01.448 --> 00:37:02.900 align:middle line:90% or is the bank deciding on it? 00:37:02.900 --> 00:37:05.943 align:middle line:84% PROFESSOR: This is contractable, so they can decide together. 00:37:05.943 --> 00:37:07.610 align:middle line:84% The borrower can decide it, but the bank 00:37:07.610 --> 00:37:10.870 align:middle line:84% will be happy to do this in equilibrium. 00:37:10.870 --> 00:37:13.330 align:middle line:84% The key thing we had to keep track of 00:37:13.330 --> 00:37:16.900 align:middle line:84% is, once they choose k together or whatever-- 00:37:16.900 --> 00:37:18.520 align:middle line:84% this is a contract-- then the borrower 00:37:18.520 --> 00:37:19.600 align:middle line:90% goes off and chooses p. 00:37:19.600 --> 00:37:23.410 align:middle line:84% So the key point is, we're choosing the level of k, 00:37:23.410 --> 00:37:27.640 align:middle line:84% taking the borrower's second period choice of p as that. 00:37:27.640 --> 00:37:31.520 align:middle line:90% That function is given. 00:37:31.520 --> 00:37:32.690 align:middle line:90% And that's what I just said. 00:37:32.690 --> 00:37:34.310 align:middle line:84% We're only maximizing this over k. 00:37:34.310 --> 00:37:37.370 align:middle line:84% We can't choose k and p jointly because we 00:37:37.370 --> 00:37:39.200 align:middle line:84% know that once we choose k, I'm going 00:37:39.200 --> 00:37:41.460 align:middle line:90% to go do whatever p I want. 00:37:41.460 --> 00:37:44.540 align:middle line:84% And we might do better if we commit to a level of k an p 00:37:44.540 --> 00:37:48.030 align:middle line:90% together, but we can't. 00:37:48.030 --> 00:37:50.240 align:middle line:90% OK. 00:37:50.240 --> 00:37:52.700 align:middle line:84% So that gives us a first order condition. 00:37:52.700 --> 00:37:54.200 align:middle line:84% So what's the first order condition? 00:37:54.200 --> 00:37:58.460 align:middle line:84% Well, it's just F prime of k times e of p 00:37:58.460 --> 00:38:04.130 align:middle line:84% plus the other term, F of k times e prime of p times 00:38:04.130 --> 00:38:09.280 align:middle line:90% p prime, so dp/dk. 00:38:09.280 --> 00:38:10.870 align:middle line:84% And all that has to be equal to rho. 00:38:10.870 --> 00:38:13.900 align:middle line:90% 00:38:13.900 --> 00:38:15.350 align:middle line:90% Clear? 00:38:15.350 --> 00:38:17.090 align:middle line:84% By the way, thanks to Ed for helping me 00:38:17.090 --> 00:38:19.110 align:middle line:84% clean up some my algebra on this slide, 00:38:19.110 --> 00:38:22.070 align:middle line:90% which I had some issues before. 00:38:22.070 --> 00:38:24.130 align:middle line:84% But I'm pretty sure this is correct. 00:38:24.130 --> 00:38:26.910 align:middle line:90% So that's what we just had. 00:38:26.910 --> 00:38:28.470 align:middle line:90% What would be the first best? 00:38:28.470 --> 00:38:31.200 align:middle line:90% 00:38:31.200 --> 00:38:36.930 align:middle line:84% The first best would be that the overall return, f prime 00:38:36.930 --> 00:38:39.600 align:middle line:84% of k times the expected return, should just 00:38:39.600 --> 00:38:42.100 align:middle line:84% be equal to the market interest rate. 00:38:42.100 --> 00:38:44.685 align:middle line:84% So that would be the optimal level of capital to invest. 00:38:44.685 --> 00:38:50.007 align:middle line:90% 00:38:50.007 --> 00:38:52.590 align:middle line:84% And we can see what's going to happen to the level of capital. 00:38:52.590 --> 00:38:57.110 align:middle line:84% So in particular, we can rewrite this equation as f prime e of p 00:38:57.110 --> 00:39:03.920 align:middle line:84% is equal to rho minus this term over here. 00:39:03.920 --> 00:39:05.300 align:middle line:90% OK. 00:39:05.300 --> 00:39:07.250 align:middle line:90% f of k is positive. 00:39:07.250 --> 00:39:09.050 align:middle line:90% e prime is positive. 00:39:09.050 --> 00:39:12.290 align:middle line:84% We had that from one of our assumptions before. 00:39:12.290 --> 00:39:17.500 align:middle line:84% And we also just showed earlier that dp/dk is less than zero. 00:39:17.500 --> 00:39:20.220 align:middle line:90% 00:39:20.220 --> 00:39:24.480 align:middle line:84% OK, because of, say, for example, the leverage channel. 00:39:24.480 --> 00:39:28.190 align:middle line:90% 00:39:28.190 --> 00:39:33.490 align:middle line:84% So what this means is that if we have a concave function, 00:39:33.490 --> 00:39:35.508 align:middle line:84% a concave production function, then 00:39:35.508 --> 00:39:37.300 align:middle line:84% the level of investment we're going to have 00:39:37.300 --> 00:39:41.174 align:middle line:84% is less than k star, the optimal level of investment. 00:39:41.174 --> 00:39:46.770 align:middle line:90% 00:39:46.770 --> 00:39:48.690 align:middle line:84% Intuitively what's going on, what's going on 00:39:48.690 --> 00:39:52.740 align:middle line:84% is, the interest rate is kind of high, given the moral hazard, 00:39:52.740 --> 00:39:56.655 align:middle line:84% and there's going to be more of this with more capital holding 00:39:56.655 --> 00:39:58.038 align:middle line:90% this rate fixed. 00:39:58.038 --> 00:39:59.580 align:middle line:84% So the interest rate is already high. 00:39:59.580 --> 00:40:01.920 align:middle line:84% That's going to drive down the optimal level of k, 00:40:01.920 --> 00:40:03.900 align:middle line:84% and then the additional channel is only 00:40:03.900 --> 00:40:05.220 align:middle line:90% going to reinforce that. 00:40:05.220 --> 00:40:07.410 align:middle line:84% So the interest rate is high because we 00:40:07.410 --> 00:40:08.820 align:middle line:90% have the moral hazard problem. 00:40:08.820 --> 00:40:11.340 align:middle line:84% And then if the higher interest rate 00:40:11.340 --> 00:40:12.992 align:middle line:84% I choose a lower level of project, 00:40:12.992 --> 00:40:14.700 align:middle line:84% that makes the interest rate even higher, 00:40:14.700 --> 00:40:18.720 align:middle line:84% and we end up in a-- the equilibrium of that drives 00:40:18.720 --> 00:40:19.890 align:middle line:90% capital even lower. 00:40:19.890 --> 00:40:23.923 align:middle line:90% 00:40:23.923 --> 00:40:25.340 align:middle line:84% And so, again, that says, look, we 00:40:25.340 --> 00:40:27.380 align:middle line:84% prefer to be able to commit to this level of p that 00:40:27.380 --> 00:40:28.310 align:middle line:90% wouldn't have this problem. 00:40:28.310 --> 00:40:30.020 align:middle line:84% It would give us a lower level of interest rate 00:40:30.020 --> 00:40:31.910 align:middle line:84% and a higher level of capital and higher level return, 00:40:31.910 --> 00:40:33.077 align:middle line:90% but we can't commit to that. 00:40:33.077 --> 00:40:41.620 align:middle line:90% 00:40:41.620 --> 00:40:45.300 align:middle line:84% We can also show similarly that dk/dw, is greater than zero 00:40:45.300 --> 00:40:47.890 align:middle line:90% through a similar argument. 00:40:47.890 --> 00:40:49.660 align:middle line:84% Just stepping back, what have we shown? 00:40:49.660 --> 00:40:54.040 align:middle line:84% This is a moral hazard model where the choice of projects 00:40:54.040 --> 00:40:55.030 align:middle line:90% was uncertain-- 00:40:55.030 --> 00:40:59.020 align:middle line:90% sorry-- was uncontractable. 00:40:59.020 --> 00:41:01.240 align:middle line:84% And that plus limited liability led people 00:41:01.240 --> 00:41:02.680 align:middle line:90% to choose these risky projects. 00:41:02.680 --> 00:41:05.500 align:middle line:84% The choice of risky projects leads to these higher interest 00:41:05.500 --> 00:41:07.990 align:middle line:84% rates, which leads to lower levels of capital 00:41:07.990 --> 00:41:10.750 align:middle line:90% in equilibrium. 00:41:10.750 --> 00:41:13.450 align:middle line:84% STUDENT: In the real world, don't people 00:41:13.450 --> 00:41:15.338 align:middle line:84% often contract over p when they're like, oh, 00:41:15.338 --> 00:41:16.630 align:middle line:90% what do you want the money for? 00:41:16.630 --> 00:41:18.460 align:middle line:90% For this business idea. 00:41:18.460 --> 00:41:19.780 align:middle line:90% OK, no I don't like your idea. 00:41:19.780 --> 00:41:20.955 align:middle line:90% I won't give you the money. 00:41:20.955 --> 00:41:21.580 align:middle line:90% PROFESSOR: Yes. 00:41:21.580 --> 00:41:23.980 align:middle line:90% So I think you're exactly right. 00:41:23.980 --> 00:41:27.535 align:middle line:84% A whole lot of what banks are doing 00:41:27.535 --> 00:41:30.160 align:middle line:84% is trying to-- you can think of a lot of what banks are doing-- 00:41:30.160 --> 00:41:34.105 align:middle line:84% well, let me also talk about the adverse selection model too. 00:41:34.105 --> 00:41:36.425 align:middle line:90% 00:41:36.425 --> 00:41:38.050 align:middle line:84% It's not obvious to me how much they're 00:41:38.050 --> 00:41:40.240 align:middle line:84% doing they're screening about the adverse selection. 00:41:40.240 --> 00:41:42.400 align:middle line:84% They're definitely trying to figure out p 00:41:42.400 --> 00:41:43.630 align:middle line:90% and whether it's the ex-ante. 00:41:43.630 --> 00:41:45.047 align:middle line:84% Whether it's the moral hazard part 00:41:45.047 --> 00:41:48.100 align:middle line:84% or it's the adverse selection part is less clear to me, 00:41:48.100 --> 00:41:49.810 align:middle line:84% so we can come back to what they're 00:41:49.810 --> 00:41:50.510 align:middle line:90% doing in the real world. 00:41:50.510 --> 00:41:51.640 align:middle line:84% I think a lot of what they're doing looks more 00:41:51.640 --> 00:41:52.973 align:middle line:90% like the adverse selection part. 00:41:52.973 --> 00:41:56.080 align:middle line:84% So the selection part is kind of the ex-ante, which we'll 00:41:56.080 --> 00:41:58.000 align:middle line:90% talk about in the next model. 00:41:58.000 --> 00:42:00.222 align:middle line:84% Ex-ante is saying, hey, Tishana, what 00:42:00.222 --> 00:42:01.930 align:middle line:84% are you planning on doing with the money. 00:42:01.930 --> 00:42:03.100 align:middle line:90% What's your p? 00:42:03.100 --> 00:42:04.470 align:middle line:90% You can think of that as like-- 00:42:04.470 --> 00:42:06.970 align:middle line:84% that's ex-ante, trying to figure out, we have different p's. 00:42:06.970 --> 00:42:09.670 align:middle line:84% Your p is this, Aaron's p is this, and so on and so forth. 00:42:09.670 --> 00:42:11.170 align:middle line:84% We're trying to figure out who's got 00:42:11.170 --> 00:42:12.640 align:middle line:90% the good p's and lend to them. 00:42:12.640 --> 00:42:14.140 align:middle line:90% That's adverse selection. 00:42:14.140 --> 00:42:16.000 align:middle line:84% This model will basically say, I want 00:42:16.000 --> 00:42:17.350 align:middle line:84% you to come in here every month, and I want you 00:42:17.350 --> 00:42:18.558 align:middle line:90% to tell me what you're doing. 00:42:18.558 --> 00:42:20.800 align:middle line:84% Kind of that very, very regular monitoring. 00:42:20.800 --> 00:42:23.740 align:middle line:84% Banks also do that, but that's the closer to this one. 00:42:23.740 --> 00:42:25.250 align:middle line:90% So I think you're right. 00:42:25.250 --> 00:42:28.000 align:middle line:84% I think that you can think about the whole-- 00:42:28.000 --> 00:42:30.460 align:middle line:84% a lot of the relationship between banks 00:42:30.460 --> 00:42:33.520 align:middle line:84% and their clients as being about trying to combat these two 00:42:33.520 --> 00:42:34.922 align:middle line:90% basic issues. 00:42:34.922 --> 00:42:36.630 align:middle line:84% And if they're perfect at combating them, 00:42:36.630 --> 00:42:37.818 align:middle line:90% then this goes away. 00:42:37.818 --> 00:42:39.360 align:middle line:84% In reality, I think we probably think 00:42:39.360 --> 00:42:41.110 align:middle line:84% they can combat them to some extent, 00:42:41.110 --> 00:42:42.568 align:middle line:84% and this is capturing the residual. 00:42:42.568 --> 00:42:45.698 align:middle line:90% 00:42:45.698 --> 00:42:46.865 align:middle line:90% Other questions or comments? 00:42:46.865 --> 00:42:50.980 align:middle line:90% 00:42:50.980 --> 00:42:54.100 align:middle line:90% So question. 00:42:54.100 --> 00:42:57.387 align:middle line:84% Would you say there are credit constraints in this model? 00:42:57.387 --> 00:42:58.970 align:middle line:84% What do we mean by credit constraints? 00:42:58.970 --> 00:43:00.650 align:middle line:84% We're going to talk a lot about credit constraints. 00:43:00.650 --> 00:43:02.983 align:middle line:84% Do we think people are credit constrained in this model? 00:43:02.983 --> 00:43:05.703 align:middle line:90% 00:43:05.703 --> 00:43:07.870 align:middle line:84% It's kind of a little definitional, but I'm curious. 00:43:07.870 --> 00:43:08.450 align:middle line:90% What do you-- 00:43:08.450 --> 00:43:16.870 align:middle line:90% 00:43:16.870 --> 00:43:17.370 align:middle line:90% Yes? 00:43:17.370 --> 00:43:17.870 align:middle line:90% No? 00:43:17.870 --> 00:43:20.500 align:middle line:84% STUDENT: I would say, no, because they can 00:43:20.500 --> 00:43:21.862 align:middle line:90% borrow as much as they want. 00:43:21.862 --> 00:43:24.404 align:middle line:84% PROFESSOR: OK, no, because they can borrow as much they want. 00:43:24.404 --> 00:43:26.348 align:middle line:84% They're just choosing k as much as they want. 00:43:26.348 --> 00:43:28.140 align:middle line:84% What would be the argument to say that they 00:43:28.140 --> 00:43:29.761 align:middle line:90% are credit constrained? 00:43:29.761 --> 00:43:36.685 align:middle line:90% 00:43:36.685 --> 00:43:38.643 align:middle line:84% STUDENT: Because they're getting less than what 00:43:38.643 --> 00:43:39.570 align:middle line:90% the borrower gives. 00:43:39.570 --> 00:43:40.880 align:middle line:90% PROFESSOR: Yeah, exactly. 00:43:40.880 --> 00:43:41.880 align:middle line:90% So that's exactly right. 00:43:41.880 --> 00:43:43.338 align:middle line:84% So I just want to be clear on this. 00:43:43.338 --> 00:43:46.500 align:middle line:84% This is a model where I think you're exactly right, Cash, 00:43:46.500 --> 00:43:49.320 align:middle line:84% that basically this is a model where they're not constrained, 00:43:49.320 --> 00:43:53.070 align:middle line:84% in the sense that they could get more k if they wanted to. 00:43:53.070 --> 00:43:54.990 align:middle line:84% The optimal level of k for them to borrow 00:43:54.990 --> 00:43:57.330 align:middle line:84% is whether the interest rate will 00:43:57.330 --> 00:43:59.520 align:middle line:90% adjust as they borrow more k. 00:43:59.520 --> 00:44:02.880 align:middle line:90% 00:44:02.880 --> 00:44:08.130 align:middle line:84% And that's going to lead there to be-- 00:44:08.130 --> 00:44:10.470 align:middle line:84% it looks like they're not borrowing as much k 00:44:10.470 --> 00:44:12.900 align:middle line:84% as they might have wanted to if these information 00:44:12.900 --> 00:44:15.000 align:middle line:90% asymmetries hadn't been there. 00:44:15.000 --> 00:44:17.810 align:middle line:90% 00:44:17.810 --> 00:44:19.180 align:middle line:90% So that's model one. 00:44:19.180 --> 00:44:22.010 align:middle line:90% That's the moral hazard model. 00:44:22.010 --> 00:44:25.412 align:middle line:84% Here's another model, which is the adverse selection model. 00:44:25.412 --> 00:44:26.620 align:middle line:90% This is an alternative story. 00:44:26.620 --> 00:44:28.953 align:middle line:84% And going back to what I was talking about with Tishana, 00:44:28.953 --> 00:44:32.260 align:middle line:84% this is kind of like screening heterogeneity among projects. 00:44:32.260 --> 00:44:33.940 align:middle line:84% So now, in this model, we're going 00:44:33.940 --> 00:44:36.880 align:middle line:84% to suppose that p is a fixed characteristic 00:44:36.880 --> 00:44:38.150 align:middle line:90% of an individual. 00:44:38.150 --> 00:44:41.310 align:middle line:84% Some people have high p, some people have low p. 00:44:41.310 --> 00:44:43.230 align:middle line:90% OK. 00:44:43.230 --> 00:44:45.990 align:middle line:84% So in this model, we say, look, suppose the bank offers 00:44:45.990 --> 00:44:48.350 align:middle line:90% an interest rate, r. 00:44:48.350 --> 00:44:50.750 align:middle line:90% Who's going to take that loan? 00:44:50.750 --> 00:44:53.410 align:middle line:90% Who wants that loan? 00:44:53.410 --> 00:44:58.940 align:middle line:84% Well, I'm going to take that loan if, again, 00:44:58.940 --> 00:45:00.320 align:middle line:90% same little liability. 00:45:00.320 --> 00:45:03.820 align:middle line:84% The project fails, probability 1 minus p, I pay nothing. 00:45:03.820 --> 00:45:07.630 align:middle line:84% Probability, p, I get f of k r of p. 00:45:07.630 --> 00:45:08.770 align:middle line:90% That's my return. 00:45:08.770 --> 00:45:11.710 align:middle line:84% And I have to pay this gross interest rate, r, on the amount 00:45:11.710 --> 00:45:13.015 align:middle line:90% I borrow, which is k minus w. 00:45:13.015 --> 00:45:19.220 align:middle line:90% 00:45:19.220 --> 00:45:22.750 align:middle line:84% So p is going to drop out, but if r is greater than-- 00:45:22.750 --> 00:45:25.000 align:middle line:84% so big R, the return, is greater than r, 00:45:25.000 --> 00:45:29.950 align:middle line:84% the amount I have to pay back, r times the amount I borrow, 00:45:29.950 --> 00:45:34.830 align:middle line:84% divide it by f of k, then I'm going to want that. 00:45:34.830 --> 00:45:36.480 align:middle line:90% OK. 00:45:36.480 --> 00:45:41.020 align:middle line:84% So if you recall from the beginning, 00:45:41.020 --> 00:45:52.410 align:middle line:84% we had that r prime was less than zero, 00:45:52.410 --> 00:45:55.148 align:middle line:90% that we had the lower-- 00:45:55.148 --> 00:45:57.690 align:middle line:84% the riskier projects were the ones that were paying the most. 00:45:57.690 --> 00:46:00.655 align:middle line:84% The higher return projects were paying less. 00:46:00.655 --> 00:46:01.155 align:middle line:90% OK. 00:46:01.155 --> 00:46:04.210 align:middle line:90% 00:46:04.210 --> 00:46:06.010 align:middle line:84% What that means is, there's going 00:46:06.010 --> 00:46:11.990 align:middle line:84% to be a cutoff value where people 00:46:11.990 --> 00:46:14.700 align:middle line:84% with p less than some cutoff value 00:46:14.700 --> 00:46:16.700 align:middle line:84% are going to be the ones who are going to pick-- 00:46:16.700 --> 00:46:18.440 align:middle line:90% going to take up the loan. 00:46:18.440 --> 00:46:21.290 align:middle line:90% 00:46:21.290 --> 00:46:25.850 align:middle line:84% r of p less than some cutoff is the same as saying 00:46:25.850 --> 00:46:30.650 align:middle line:84% that p is going to be less than some cutoff, 00:46:30.650 --> 00:46:32.670 align:middle line:84% so the super risky projects up to some cutoff 00:46:32.670 --> 00:46:34.545 align:middle line:84% are the ones that are going to take the loan. 00:46:34.545 --> 00:46:36.860 align:middle line:90% 00:46:36.860 --> 00:46:38.350 align:middle line:90% OK. 00:46:38.350 --> 00:46:43.840 align:middle line:84% So we can define p tilde of r as exactly that cutoff, 00:46:43.840 --> 00:46:47.500 align:middle line:84% so we can just define p tilde, such that r of p tilde 00:46:47.500 --> 00:46:49.210 align:middle line:90% is exactly equal to this. 00:46:49.210 --> 00:46:51.700 align:middle line:84% And everybody with a p less than p tilde will borrow. 00:46:51.700 --> 00:46:54.830 align:middle line:90% 00:46:54.830 --> 00:46:55.330 align:middle line:90% OK. 00:46:55.330 --> 00:46:58.880 align:middle line:90% 00:46:58.880 --> 00:47:00.450 align:middle line:90% And so what happens? 00:47:00.450 --> 00:47:03.490 align:middle line:90% Well, as I r-- 00:47:03.490 --> 00:47:06.880 align:middle line:84% so I increase the interest rate-- 00:47:06.880 --> 00:47:10.810 align:middle line:84% the p tilde is going to shift to the left. 00:47:10.810 --> 00:47:12.515 align:middle line:84% Because if I increase the interest rate, 00:47:12.515 --> 00:47:13.765 align:middle line:90% the return's got to be higher. 00:47:13.765 --> 00:47:16.240 align:middle line:84% Higher return projects were the ones with the lower p. 00:47:16.240 --> 00:47:18.190 align:middle line:84% So again, higher interest rates are 00:47:18.190 --> 00:47:20.500 align:middle line:84% going to shift that threshold to the left, 00:47:20.500 --> 00:47:22.860 align:middle line:84% and we're going get riskier projects. 00:47:22.860 --> 00:47:26.180 align:middle line:84% So we're also going to get that same correlation that we had 00:47:26.180 --> 00:47:28.070 align:middle line:84% from the moral hazard model, which 00:47:28.070 --> 00:47:30.590 align:middle line:84% is a correlation between the interest rate 00:47:30.590 --> 00:47:32.420 align:middle line:84% and the riskiness of the projects, 00:47:32.420 --> 00:47:35.895 align:middle line:84% although it's happening through this different channel. 00:47:35.895 --> 00:47:37.770 align:middle line:84% One was this ex-ante, who wants the projects. 00:47:37.770 --> 00:47:48.120 align:middle line:84% The other is the ex-post, who is the interest rates. 00:47:48.120 --> 00:47:50.490 align:middle line:84% And intuitively what's going on in this model, what's 00:47:50.490 --> 00:47:55.590 align:middle line:84% going on here is like, look, imagine the return is the-- 00:47:55.590 --> 00:47:58.740 align:middle line:84% in the good state, I have to pay back a lot. 00:47:58.740 --> 00:48:00.000 align:middle line:90% Who's going to want that? 00:48:00.000 --> 00:48:03.330 align:middle line:84% Only people for whom the good state is really, really good, 00:48:03.330 --> 00:48:05.925 align:middle line:84% and those are the people who were risky. 00:48:05.925 --> 00:48:14.090 align:middle line:90% 00:48:14.090 --> 00:48:17.290 align:middle line:84% So again, we have this negative correlation here. 00:48:17.290 --> 00:48:20.550 align:middle line:84% Now, what's the market equilibrium here? 00:48:20.550 --> 00:48:24.260 align:middle line:84% Well, again, we have that rho, that r times the amount you 00:48:24.260 --> 00:48:26.600 align:middle line:84% get-- so here, we have a slightly different condition. 00:48:26.600 --> 00:48:29.570 align:middle line:90% What is the p? 00:48:29.570 --> 00:48:31.190 align:middle line:90% For a given r, who borrows? 00:48:31.190 --> 00:48:35.690 align:middle line:84% Well, it's everybody with interest rate with p 00:48:35.690 --> 00:48:38.030 align:middle line:90% less than that cutoff value. 00:48:38.030 --> 00:48:40.570 align:middle line:84% So what's the average probability of success? 00:48:40.570 --> 00:48:50.569 align:middle line:90% 00:48:50.569 --> 00:48:56.570 align:middle line:84% It's just the expectation-- this is e meaning expectation, not 00:48:56.570 --> 00:48:58.295 align:middle line:90% e meaning the expected return. 00:48:58.295 --> 00:49:02.045 align:middle line:90% 00:49:02.045 --> 00:49:03.170 align:middle line:90% So sorry for the imitation. 00:49:03.170 --> 00:49:06.450 align:middle line:84% It's the expectation of p conditional on p less than p 00:49:06.450 --> 00:49:06.950 align:middle line:90% tilde. 00:49:06.950 --> 00:49:11.740 align:middle line:90% 00:49:11.740 --> 00:49:14.560 align:middle line:84% We know that as p tilde goes down, 00:49:14.560 --> 00:49:16.150 align:middle line:84% the expectation of p tilde goes down. 00:49:16.150 --> 00:49:18.272 align:middle line:84% But beyond that, it's the shape of that thing 00:49:18.272 --> 00:49:20.230 align:middle line:84% depends on just what the distribution of p's is 00:49:20.230 --> 00:49:22.910 align:middle line:90% in the population. 00:49:22.910 --> 00:49:23.810 align:middle line:90% OK. 00:49:23.810 --> 00:49:27.830 align:middle line:84% That also is going to mean that r, the interest rate charge, 00:49:27.830 --> 00:49:30.080 align:middle line:84% is going to be equal to rho divided by the expectation 00:49:30.080 --> 00:49:30.290 align:middle line:90% thing. 00:49:30.290 --> 00:49:31.707 align:middle line:84% If this thing is less than 1, that 00:49:31.707 --> 00:49:34.190 align:middle line:84% means that the rs that going to be charged to people 00:49:34.190 --> 00:49:36.140 align:middle line:90% are going to be higher than rho. 00:49:36.140 --> 00:49:38.000 align:middle line:84% So we're also going to have higher interest 00:49:38.000 --> 00:49:39.470 align:middle line:90% rates than equilibrium. 00:49:39.470 --> 00:49:44.090 align:middle line:84% It's in the first best or whatever, then that rho thing. 00:49:44.090 --> 00:49:46.710 align:middle line:84% But the reasoning is a bit different. 00:49:46.710 --> 00:49:48.433 align:middle line:84% It's all about this ex-ante screening. 00:49:48.433 --> 00:49:49.850 align:middle line:84% Going back to your point, Tishana, 00:49:49.850 --> 00:49:50.630 align:middle line:84% this is where the banks are going 00:49:50.630 --> 00:49:52.310 align:middle line:84% to want to invest in doing everything 00:49:52.310 --> 00:49:55.400 align:middle line:84% that we can say to screen people, kind of ex-ante 00:49:55.400 --> 00:49:57.000 align:middle line:84% and get rid of the people who are 00:49:57.000 --> 00:49:58.250 align:middle line:90% going to be kind of lower rho. 00:49:58.250 --> 00:49:59.625 align:middle line:84% Because for a given interest rate 00:49:59.625 --> 00:50:02.300 align:middle line:84% that the bank's charging, conditional on charging 00:50:02.300 --> 00:50:05.900 align:middle line:84% r, if it can weed out some people who are the worst, 00:50:05.900 --> 00:50:06.932 align:middle line:90% it's going to do better. 00:50:06.932 --> 00:50:08.390 align:middle line:84% That's not necessarily equilibrium, 00:50:08.390 --> 00:50:10.880 align:middle line:84% but that's going to be the bank's incentive function. 00:50:10.880 --> 00:50:12.830 align:middle line:90% So I announce r. 00:50:12.830 --> 00:50:15.590 align:middle line:84% I know that everyone with p less and p tilde would like it. 00:50:15.590 --> 00:50:17.290 align:middle line:90% How can I make even more money? 00:50:17.290 --> 00:50:17.840 align:middle line:84% I'm going to be able to break even, 00:50:17.840 --> 00:50:19.298 align:middle line:84% but I can make even more money if I 00:50:19.298 --> 00:50:21.515 align:middle line:84% can find the really low people and knock them out. 00:50:21.515 --> 00:50:23.390 align:middle line:84% I want to get people who are just as close as 00:50:23.390 --> 00:50:24.680 align:middle line:90% possible to p tilde. 00:50:24.680 --> 00:50:26.330 align:middle line:84% And you can think, a lot of what banks 00:50:26.330 --> 00:50:27.950 align:middle line:84% are doing in that screening process 00:50:27.950 --> 00:50:29.780 align:middle line:84% is trying to figure out who are the good borrowers, who 00:50:29.780 --> 00:50:32.120 align:middle line:84% are the bad borrowers, can I remove the really bad ones. 00:50:32.120 --> 00:50:35.410 align:middle line:90% 00:50:35.410 --> 00:50:37.900 align:middle line:84% So again, we can draw that same kind of equilibrium. 00:50:37.900 --> 00:50:41.860 align:middle line:84% Here, this thing is the same market clearing condition, 00:50:41.860 --> 00:50:44.020 align:middle line:90% but now, we have a different-- 00:50:44.020 --> 00:50:45.520 align:middle line:84% instead of a moral hazard condition, 00:50:45.520 --> 00:50:48.500 align:middle line:84% we have an adverse selection condition, 00:50:48.500 --> 00:50:52.120 align:middle line:84% which says that here is the expectation of p, 00:50:52.120 --> 00:50:54.160 align:middle line:84% given that p is less than p tilde, where 00:50:54.160 --> 00:50:56.500 align:middle line:90% p tilde is a function of r. 00:50:56.500 --> 00:51:02.230 align:middle line:84% And so this curve here doesn't come from project choices. 00:51:02.230 --> 00:51:05.080 align:middle line:84% This is just a question of, what is the distribution of p's 00:51:05.080 --> 00:51:06.350 align:middle line:90% in the population. 00:51:06.350 --> 00:51:11.950 align:middle line:84% This is just like a property of the p 00:51:11.950 --> 00:51:14.430 align:middle line:90% function of the population. 00:51:14.430 --> 00:51:19.340 align:middle line:84% What is the distribution of p's And again, you 00:51:19.340 --> 00:51:21.413 align:middle line:84% can draw that in funky different ways. 00:51:21.413 --> 00:51:22.330 align:middle line:90% It can look like this. 00:51:22.330 --> 00:51:23.038 align:middle line:90% It can intersect. 00:51:23.038 --> 00:51:27.200 align:middle line:84% It can go on with equilibria and so on and so forth. 00:51:27.200 --> 00:51:31.540 align:middle line:84% It's got to be downward sloping, but beyond that, unclear. 00:51:31.540 --> 00:51:33.600 align:middle line:90% OK. 00:51:33.600 --> 00:51:37.060 align:middle line:90% Clear so far? 00:51:37.060 --> 00:51:38.810 align:middle line:90% OK. 00:51:38.810 --> 00:51:42.840 align:middle line:84% So both of those models, the moral hazard 00:51:42.840 --> 00:51:46.680 align:middle line:84% model and the adverse selection model, 00:51:46.680 --> 00:51:52.720 align:middle line:84% are going to feature high interest rates, or at least 00:51:52.720 --> 00:51:54.470 align:middle line:84% both of those are reasons why the interest 00:51:54.470 --> 00:51:56.770 align:middle line:84% rate will be higher than the market clearance. 00:51:56.770 --> 00:52:00.350 align:middle line:84% This is why the market for, say, microfinance, the interest rate 00:52:00.350 --> 00:52:04.010 align:middle line:90% would be higher than, say, the-- 00:52:04.010 --> 00:52:06.770 align:middle line:84% Imagine you have a simple model where microfinance institutions 00:52:06.770 --> 00:52:09.290 align:middle line:84% are raising capital on Wall Street 00:52:09.290 --> 00:52:11.972 align:middle line:90% and lending it to small lenders. 00:52:11.972 --> 00:52:13.430 align:middle line:84% This already tells you there should 00:52:13.430 --> 00:52:16.010 align:middle line:84% be reasons why the interest rate is paid by small borrowers, 00:52:16.010 --> 00:52:18.030 align:middle line:84% or it tells you why the interest rate paid by the borrower 00:52:18.030 --> 00:52:19.572 align:middle line:84% should be a lot higher than the level 00:52:19.572 --> 00:52:21.900 align:middle line:84% that the bank is charging on interest rates. 00:52:21.900 --> 00:52:24.550 align:middle line:84% But it comes from the default channel. 00:52:24.550 --> 00:52:25.080 align:middle line:90% All right. 00:52:25.080 --> 00:52:30.360 align:middle line:90% 00:52:30.360 --> 00:52:33.242 align:middle line:84% So those models predict high interest rates, 00:52:33.242 --> 00:52:34.950 align:middle line:84% but the reason for the high interest rate 00:52:34.950 --> 00:52:36.367 align:middle line:84% is to cover the expected defaults. 00:52:36.367 --> 00:52:39.540 align:middle line:90% 00:52:39.540 --> 00:52:43.680 align:middle line:84% So one unfortunate fact for this model 00:52:43.680 --> 00:52:46.800 align:middle line:84% is that, at least in microfinance, 00:52:46.800 --> 00:52:50.730 align:middle line:84% at least in some cases, interest rates are high. 00:52:50.730 --> 00:52:54.540 align:middle line:84% But often repayment rates are actually high too. 00:52:54.540 --> 00:52:55.790 align:middle line:90% OK. 00:52:55.790 --> 00:52:59.740 align:middle line:84% So then that doesn't work in this model 00:52:59.740 --> 00:53:01.780 align:middle line:84% because the only reason in both those models 00:53:01.780 --> 00:53:04.580 align:middle line:84% that the rates were high was to cover the defaults. 00:53:04.580 --> 00:53:09.610 align:middle line:84% So to the extent this is also a force, 00:53:09.610 --> 00:53:15.100 align:middle line:84% if there are reasons for that wedge beyond the default rates, 00:53:15.100 --> 00:53:17.050 align:middle line:84% we'd add something else to the model. 00:53:17.050 --> 00:53:18.710 align:middle line:84% One simple way to do it, by the way, 00:53:18.710 --> 00:53:21.727 align:middle line:84% is if you thought that the lenders were not competitive, 00:53:21.727 --> 00:53:23.185 align:middle line:84% then they're just charging markups. 00:53:23.185 --> 00:53:25.448 align:middle line:90% 00:53:25.448 --> 00:53:27.490 align:middle line:84% It's not in the slide, but if the lenders are not 00:53:27.490 --> 00:53:29.260 align:middle line:84% competitive, then, of course, they're 00:53:29.260 --> 00:53:34.550 align:middle line:84% going to jack up the interest rate and make more profits. 00:53:34.550 --> 00:53:35.050 align:middle line:90% So fine. 00:53:35.050 --> 00:53:37.425 align:middle line:90% That could be part of it. 00:53:37.425 --> 00:53:38.800 align:middle line:84% And by the way, there are reasons 00:53:38.800 --> 00:53:40.508 align:middle line:84% to think that there may be noncompetition 00:53:40.508 --> 00:53:42.670 align:middle line:90% in lending markets. 00:53:42.670 --> 00:53:44.087 align:middle line:84% In particular, this whole question 00:53:44.087 --> 00:53:45.503 align:middle line:84% that we were talking about getting 00:53:45.503 --> 00:53:47.380 align:middle line:84% to know people's p's and monitoring them, 00:53:47.380 --> 00:53:51.250 align:middle line:84% there's an important rule for information. 00:53:51.250 --> 00:53:54.880 align:middle line:84% So in particular, imagine that I'm Tishana's lender, 00:53:54.880 --> 00:53:57.460 align:middle line:84% and she and I have been working together for years. 00:53:57.460 --> 00:54:02.560 align:middle line:84% I may have information about her p from my years of interaction 00:54:02.560 --> 00:54:05.020 align:middle line:84% with her that, if Rabajit chooses another lender, 00:54:05.020 --> 00:54:06.370 align:middle line:90% Rabajit doesn't have. 00:54:06.370 --> 00:54:09.652 align:middle line:84% So that may mean I have some advantage of lending to Tishana 00:54:09.652 --> 00:54:11.860 align:middle line:84% over Rabajit, and I'd be able to use that and extract 00:54:11.860 --> 00:54:13.970 align:middle line:90% some rents from that. 00:54:13.970 --> 00:54:17.500 align:middle line:84% So in particular, there may be good reasons 00:54:17.500 --> 00:54:19.600 align:middle line:84% to think that the model of perfect competition 00:54:19.600 --> 00:54:21.340 align:middle line:84% is not necessarily the only model. 00:54:21.340 --> 00:54:25.200 align:middle line:84% It's not necessarily a perfect model for this. 00:54:25.200 --> 00:54:26.430 align:middle line:90% That's one. 00:54:26.430 --> 00:54:29.230 align:middle line:84% But here's another model that also could be relevant, 00:54:29.230 --> 00:54:30.920 align:middle line:90% which is monitoring costs. 00:54:30.920 --> 00:54:33.710 align:middle line:84% And in particular, this monitoring cost idea 00:54:33.710 --> 00:54:35.903 align:middle line:90% captures the-- 00:54:35.903 --> 00:54:39.988 align:middle line:90% 00:54:39.988 --> 00:54:41.780 align:middle line:84% this actually is not going to quite capture 00:54:41.780 --> 00:54:44.090 align:middle line:84% it the way it's going to be written down here, 00:54:44.090 --> 00:54:45.650 align:middle line:84% but it captures the idea that banks 00:54:45.650 --> 00:54:47.960 align:middle line:84% are spending a lot of money doing stuff 00:54:47.960 --> 00:54:49.935 align:middle line:90% to make sure people repay. 00:54:49.935 --> 00:54:50.435 align:middle line:90% OK. 00:54:50.435 --> 00:54:54.720 align:middle line:90% 00:54:54.720 --> 00:54:57.740 align:middle line:84% So one version of monitoring is, we have to spend a lot of money 00:54:57.740 --> 00:55:00.180 align:middle line:84% upfront figuring out who the good p's and who are the bad 00:55:00.180 --> 00:55:00.680 align:middle line:90% p's. 00:55:00.680 --> 00:55:02.300 align:middle line:90% That's one version of it. 00:55:02.300 --> 00:55:05.863 align:middle line:84% A second version is the moral hazard version, I think, 00:55:05.863 --> 00:55:08.280 align:middle line:84% that we were talking about earlier, which is to say, look, 00:55:08.280 --> 00:55:09.290 align:middle line:84% I'm going to meet with you every single month 00:55:09.290 --> 00:55:11.832 align:middle line:84% and make sure you're doing what you said you were going to do 00:55:11.832 --> 00:55:17.540 align:middle line:84% and not deviating and doing a worse project. 00:55:17.540 --> 00:55:24.290 align:middle line:84% But a third one is, maybe I can actually just spend some effort 00:55:24.290 --> 00:55:26.820 align:middle line:90% to get you to repay. 00:55:26.820 --> 00:55:31.100 align:middle line:84% So that's what this model is about, 00:55:31.100 --> 00:55:35.300 align:middle line:84% so let's think about monitoring here. 00:55:35.300 --> 00:55:37.377 align:middle line:84% So let's abstract from the project 00:55:37.377 --> 00:55:38.960 align:middle line:84% choices for the moment, just to make-- 00:55:38.960 --> 00:55:40.910 align:middle line:90% and the project's failing. 00:55:40.910 --> 00:55:44.390 align:middle line:84% But instead, we're going to think about repayment. 00:55:44.390 --> 00:55:46.610 align:middle line:84% So as before in the previous model, 00:55:46.610 --> 00:55:50.680 align:middle line:84% suppose that if I invest k dollars, I get 00:55:50.680 --> 00:55:55.480 align:middle line:84% a gross return, f of k, the gross interest rate, r. 00:55:55.480 --> 00:55:58.270 align:middle line:84% And so if I have wealth, w, and I want to invest k, 00:55:58.270 --> 00:55:59.717 align:middle line:90% I need to borrow k minus w. 00:55:59.717 --> 00:56:02.050 align:middle line:84% And the amount I'm supposed to pay is k minus w times r. 00:56:02.050 --> 00:56:04.530 align:middle line:90% That's the same as before. 00:56:04.530 --> 00:56:06.350 align:middle line:90% OK. 00:56:06.350 --> 00:56:11.030 align:middle line:84% But now, suppose that it's not the good state or the state 00:56:11.030 --> 00:56:11.660 align:middle line:90% anymore. 00:56:11.660 --> 00:56:13.670 align:middle line:84% Suppose that basically I can just 00:56:13.670 --> 00:56:23.610 align:middle line:84% choose to stiff you and not repay by paying a cost eta. 00:56:23.610 --> 00:56:24.900 align:middle line:90% Right, eta? 00:56:24.900 --> 00:56:26.730 align:middle line:90% Yes. 00:56:26.730 --> 00:56:30.151 align:middle line:84% We don't have an eta variant yet, do we? 00:56:30.151 --> 00:56:31.026 align:middle line:90% STUDENT: [INAUDIBLE]. 00:56:31.026 --> 00:56:31.362 align:middle line:90% PROFESSOR: Huh? 00:56:31.362 --> 00:56:33.190 align:middle line:84% STUDENT: It's already [INAUDIBLE] omicron, 00:56:33.190 --> 00:56:34.928 align:middle line:90% so presumably it just wasn't-- 00:56:34.928 --> 00:56:36.220 align:middle line:90% PROFESSOR: It wasn't a big one. 00:56:36.220 --> 00:56:37.137 align:middle line:90% STUDENT: --a big deal. 00:56:37.137 --> 00:56:39.285 align:middle line:84% PROFESSOR: Yeah, they skipped a couple recently. 00:56:39.285 --> 00:56:41.410 align:middle line:84% They skipped nu apparently because everyone thought 00:56:41.410 --> 00:56:42.910 align:middle line:84% it was going to be called the "new variant," 00:56:42.910 --> 00:56:44.618 align:middle line:84% and they thought that would be confusing. 00:56:44.618 --> 00:56:48.010 align:middle line:90% So anyway, eta, OK. 00:56:48.010 --> 00:56:50.320 align:middle line:84% By cost, eta, that's proportional to the amount 00:56:50.320 --> 00:56:51.730 align:middle line:90% invested. 00:56:51.730 --> 00:56:53.420 align:middle line:84% So that's what's going to happen? 00:56:53.420 --> 00:56:57.760 align:middle line:84% So lenders are going to provide financing up to the point 00:56:57.760 --> 00:56:59.747 align:middle line:84% where the borrower is going to repay. 00:56:59.747 --> 00:57:01.330 align:middle line:84% So this is a slightly different model, 00:57:01.330 --> 00:57:03.910 align:middle line:84% but in this case in particular, they'd like to get paid back. 00:57:03.910 --> 00:57:10.300 align:middle line:90% 00:57:10.300 --> 00:57:11.950 align:middle line:90% What's the borrower's choice? 00:57:11.950 --> 00:57:15.543 align:middle line:84% Well, the borrower could pay back and get f of k. 00:57:15.543 --> 00:57:17.460 align:middle line:84% And the borrower's net return if they pay back 00:57:17.460 --> 00:57:20.150 align:middle line:84% is f of k minus the amount they have to pay. 00:57:20.150 --> 00:57:25.820 align:middle line:84% Or the borrower could pay this default cost and get f of k 00:57:25.820 --> 00:57:29.420 align:middle line:90% minus nu of k-- eta of k. 00:57:29.420 --> 00:57:31.370 align:middle line:84% And you can think of this as like, 00:57:31.370 --> 00:57:36.050 align:middle line:84% they come and seize my factory and take some share of it. 00:57:36.050 --> 00:57:41.160 align:middle line:90% 00:57:41.160 --> 00:57:44.200 align:middle line:84% So it's a slightly different model. 00:57:44.200 --> 00:57:54.270 align:middle line:84% So if we rearrange terms here, the kind 00:57:54.270 --> 00:58:01.710 align:middle line:84% of key quality condition is that r k minus w has to equal eta k, 00:58:01.710 --> 00:58:05.700 align:middle line:84% or that k over w is equal to r over r minus eta. 00:58:05.700 --> 00:58:06.200 align:middle line:90% OK. 00:58:06.200 --> 00:58:11.520 align:middle line:90% 00:58:11.520 --> 00:58:14.310 align:middle line:84% So in this model, this is actually 00:58:14.310 --> 00:58:16.702 align:middle line:84% going to give a maximum amount the bank 00:58:16.702 --> 00:58:17.910 align:middle line:90% is going to want to lend you. 00:58:17.910 --> 00:58:20.880 align:middle line:90% 00:58:20.880 --> 00:58:24.650 align:middle line:84% With amount of borrowing, it's going to be increasing in w. 00:58:24.650 --> 00:58:33.270 align:middle line:84% The more w you have, the more you can borrow, 00:58:33.270 --> 00:58:34.770 align:middle line:90% and decreasing in r. 00:58:34.770 --> 00:58:37.190 align:middle line:90% 00:58:37.190 --> 00:58:37.690 align:middle line:90% OK. 00:58:37.690 --> 00:58:40.780 align:middle line:90% 00:58:40.780 --> 00:58:45.530 align:middle line:84% So now, let's suppose there's a monitoring technology. 00:58:45.530 --> 00:58:50.210 align:middle line:84% So the lender needs to spend some resources in order 00:58:50.210 --> 00:58:51.710 align:middle line:84% to make the borrower want to repay. 00:58:51.710 --> 00:58:53.540 align:middle line:84% Basically, in other words, eta is 00:58:53.540 --> 00:58:55.790 align:middle line:84% equal to zero if the borrower doesn't spend any money. 00:58:55.790 --> 00:58:57.410 align:middle line:84% But if the borrower does some stuff, 00:58:57.410 --> 00:59:00.345 align:middle line:90% they can get you to repay. 00:59:00.345 --> 00:59:02.220 align:middle line:84% Maybe that's like finding out where you live. 00:59:02.220 --> 00:59:05.115 align:middle line:90% 00:59:05.115 --> 00:59:07.740 align:middle line:84% If you think about a lot of the microfinance institutions we'll 00:59:07.740 --> 00:59:09.990 align:middle line:84% talk about next time, they might be clustered together 00:59:09.990 --> 00:59:10.890 align:middle line:90% in groups of people. 00:59:10.890 --> 00:59:12.998 align:middle line:84% And the basic idea is, if you defaulted, 00:59:12.998 --> 00:59:15.540 align:middle line:84% other people in your group can impose social sanctions on you 00:59:15.540 --> 00:59:16.290 align:middle line:90% if you to default. 00:59:16.290 --> 00:59:18.990 align:middle line:84% There's some cost we have to create to get you to repay. 00:59:18.990 --> 00:59:20.657 align:middle line:84% And in this model, they're going to say, 00:59:20.657 --> 00:59:22.620 align:middle line:90% suppose those are a fixed cost. 00:59:22.620 --> 00:59:24.795 align:middle line:84% It's basically a cost per borrower. 00:59:24.795 --> 00:59:28.190 align:middle line:90% 00:59:28.190 --> 00:59:30.890 align:middle line:84% And there are some estimates from this old paper by Aleem, 00:59:30.890 --> 00:59:35.630 align:middle line:84% which says that maybe they're pretty high, particularly 00:59:35.630 --> 00:59:36.590 align:middle line:90% for very small loans. 00:59:36.590 --> 00:59:39.320 align:middle line:84% And this cost would be a reason why basically microfinance has 00:59:39.320 --> 00:59:42.050 align:middle line:84% really high interest rates, because if I 00:59:42.050 --> 00:59:44.600 align:middle line:90% have to find out, yes-- 00:59:44.600 --> 00:59:46.280 align:middle line:84% Fixed cost may be a little bit much, 00:59:46.280 --> 00:59:48.530 align:middle line:84% like it's probably cheaper to find out 00:59:48.530 --> 00:59:56.330 align:middle line:84% if Joe's dry cleaner is a better investment. 00:59:56.330 --> 00:59:58.520 align:middle line:84% Or Joe's tailor is probably easier 00:59:58.520 --> 01:00:01.848 align:middle line:84% to figure out if they're worth lending to then the Gap 01:00:01.848 --> 01:00:02.390 align:middle line:90% or something. 01:00:02.390 --> 01:00:04.760 align:middle line:90% It's probably not totally fixed. 01:00:04.760 --> 01:00:09.290 align:middle line:84% But the idea that it has this large fixed component, 01:00:09.290 --> 01:00:12.170 align:middle line:84% or it's definitely not increasing very much, 01:00:12.170 --> 01:00:13.520 align:middle line:90% seems kind of reasonable. 01:00:13.520 --> 01:00:15.440 align:middle line:84% And if that's true, that's going to give us 01:00:15.440 --> 01:00:17.452 align:middle line:84% a reasoning why microfinance is making 01:00:17.452 --> 01:00:18.660 align:middle line:90% lots and lots of small loans. 01:00:18.660 --> 01:00:19.790 align:middle line:84% It's going to have to have really high interest 01:00:19.790 --> 01:00:23.100 align:middle line:84% rates because they have to cover this fixed monitoring cost. 01:00:23.100 --> 01:00:25.168 align:middle line:84% So basically, if you're making a small loan, 01:00:25.168 --> 01:00:26.210 align:middle line:90% you have this fixed cost. 01:00:26.210 --> 01:00:28.310 align:middle line:90% You have to recoup it. 01:00:28.310 --> 01:00:30.350 align:middle line:84% In terms of interest rate, it's going 01:00:30.350 --> 01:00:33.170 align:middle line:84% to be a large percent of a small base, 01:00:33.170 --> 01:00:34.610 align:middle line:84% whereas if it's a really big loan, 01:00:34.610 --> 01:00:36.680 align:middle line:84% that fixed cost is going to be much smaller. 01:00:36.680 --> 01:00:39.960 align:middle line:84% That's basically what's going on in this model. 01:00:39.960 --> 01:00:40.830 align:middle line:90% OK. 01:00:40.830 --> 01:00:42.220 align:middle line:90% So how is that going to work? 01:00:42.220 --> 01:00:45.930 align:middle line:84% So if it's a fixed cost, then the lender's 01:00:45.930 --> 01:00:49.890 align:middle line:84% going to just break even, so their return 01:00:49.890 --> 01:00:53.050 align:middle line:84% has to equal the market return plus the fixed cost. 01:00:53.050 --> 01:00:56.370 align:middle line:84% And so that's kind of it basically. 01:00:56.370 --> 01:00:58.465 align:middle line:84% If the amount they're borrowing is really small, 01:00:58.465 --> 01:01:00.840 align:middle line:84% then we only have to raise the interest rate a little bit 01:01:00.840 --> 01:01:01.882 align:middle line:90% to cover this fixed cost. 01:01:01.882 --> 01:01:03.725 align:middle line:84% If that borrowing is really big-- sorry, 01:01:03.725 --> 01:01:05.100 align:middle line:84% if the borrowing is really small, 01:01:05.100 --> 01:01:07.183 align:middle line:84% if you're raising interest rate a lot to cover it, 01:01:07.183 --> 01:01:09.730 align:middle line:84% if the fixed cost of the amount is really small, we don't. 01:01:09.730 --> 01:01:10.230 align:middle line:90% OK. 01:01:10.230 --> 01:01:13.180 align:middle line:90% 01:01:13.180 --> 01:01:14.890 align:middle line:84% So if we had this condition from before 01:01:14.890 --> 01:01:18.850 align:middle line:84% of what the optimal level of the credit constraint is, 01:01:18.850 --> 01:01:24.010 align:middle line:84% we can just substitute that in and get the return. 01:01:24.010 --> 01:01:27.160 align:middle line:84% It's going to be equal to the market return plus this markup 01:01:27.160 --> 01:01:29.920 align:middle line:84% term, which is going to be a function of their wealth. 01:01:29.920 --> 01:01:32.860 align:middle line:84% That's just combining these two points, 01:01:32.860 --> 01:01:37.540 align:middle line:84% and they're basically for really-- 01:01:37.540 --> 01:01:40.870 align:middle line:84% and so what's the point-- for really rich borrowers. 01:01:40.870 --> 01:01:45.070 align:middle line:90% 01:01:45.070 --> 01:01:50.010 align:middle line:84% For really rich borrowers, they can borrow a lot, 01:01:50.010 --> 01:01:53.490 align:middle line:84% and therefore, we can spread that fixed 01:01:53.490 --> 01:01:54.720 align:middle line:90% cost over a small amount. 01:01:54.720 --> 01:01:57.850 align:middle line:84% For really poor borrowers, that credit isn't really binding. 01:01:57.850 --> 01:01:59.340 align:middle line:84% We're not going to be willing to lend them very much, 01:01:59.340 --> 01:02:01.500 align:middle line:84% and therefore, that interest rate we're going to charge them 01:02:01.500 --> 01:02:02.660 align:middle line:90% is going to be a lot. 01:02:02.660 --> 01:02:03.660 align:middle line:90% And then you just solve. 01:02:03.660 --> 01:02:05.650 align:middle line:84% And then, of course, that has this additional thing, 01:02:05.650 --> 01:02:07.930 align:middle line:84% which is the interest is in an equilibrium condition, 01:02:07.930 --> 01:02:10.138 align:middle line:84% so you have to do a little algebra to solve that out. 01:02:10.138 --> 01:02:13.990 align:middle line:84% But that's the basic logic of what's going on. 01:02:13.990 --> 01:02:19.050 align:middle line:84% So in this model, some people will never be able to borrow. 01:02:19.050 --> 01:02:24.270 align:middle line:84% And this condition over here, if fee is greater than eta w, 01:02:24.270 --> 01:02:26.760 align:middle line:90% then there's no equilibrium. 01:02:26.760 --> 01:02:31.782 align:middle line:84% Either there's no solution where you can cover the monitoring 01:02:31.782 --> 01:02:33.990 align:middle line:84% cost with the interest rate because the industry gets 01:02:33.990 --> 01:02:36.448 align:middle line:84% too high, and then they won't repay and so on and so forth. 01:02:36.448 --> 01:02:38.068 align:middle line:90% There's no solution to that. 01:02:38.068 --> 01:02:40.110 align:middle line:84% If it's less than that, then there is a solution, 01:02:40.110 --> 01:02:42.693 align:middle line:84% and then you have this condition I said before that r is going 01:02:42.693 --> 01:02:44.560 align:middle line:90% down with wealth going up. 01:02:44.560 --> 01:02:46.920 align:middle line:84% So this is a model where we have people 01:02:46.920 --> 01:02:49.080 align:middle line:84% who are going to be actually credit constrained. 01:02:49.080 --> 01:02:51.548 align:middle line:84% Some people just cannot borrow at all. 01:02:51.548 --> 01:02:53.340 align:middle line:84% it's not this, you can borrow a little bit. 01:02:53.340 --> 01:02:56.110 align:middle line:84% Here, some people are just totally cut out of the market. 01:02:56.110 --> 01:02:58.140 align:middle line:84% And we also have this thing now where 01:02:58.140 --> 01:03:01.977 align:middle line:84% the people who are borrowing a little 01:03:01.977 --> 01:03:04.560 align:middle line:84% are the ones who are paying the high interest rates, which is, 01:03:04.560 --> 01:03:07.170 align:middle line:84% if you think back to one of the motivating factors I kind 01:03:07.170 --> 01:03:08.878 align:middle line:84% of gave you at the beginning of the talk, 01:03:08.878 --> 01:03:10.960 align:middle line:84% this is now something that explains that. 01:03:10.960 --> 01:03:12.670 align:middle line:84% And that also is consistent why we 01:03:12.670 --> 01:03:14.670 align:middle line:84% often-- consistent with finding the fact that we 01:03:14.670 --> 01:03:19.180 align:middle line:84% have all of these very small loans having very high rates. 01:03:19.180 --> 01:03:21.697 align:middle line:84% We also get-- maybe I won't make so much of this-- 01:03:21.697 --> 01:03:24.030 align:middle line:84% but we have this multiplier property, which is basically 01:03:24.030 --> 01:03:28.200 align:middle line:84% that dr, d rho, how the rate charged depends 01:03:28.200 --> 01:03:30.750 align:middle line:90% on underlying market rate. 01:03:30.750 --> 01:03:33.832 align:middle line:84% This thing is going to be less than-- 01:03:33.832 --> 01:03:36.030 align:middle line:84% this thing is going to be a multiplier that's 01:03:36.030 --> 01:03:39.150 align:middle line:84% going to make the rate be very sensitive to the underlying 01:03:39.150 --> 01:03:41.168 align:middle line:84% rate, so it also kind of moves around a lot. 01:03:41.168 --> 01:03:42.960 align:middle line:84% But I'm not going to focus on that as much. 01:03:42.960 --> 01:03:45.630 align:middle line:90% 01:03:45.630 --> 01:03:48.328 align:middle line:84% So what did this model add to these other things? 01:03:48.328 --> 01:03:49.120 align:middle line:90% Well, a few things. 01:03:49.120 --> 01:03:51.390 align:middle line:84% So it added a wedge between the cost of capital 01:03:51.390 --> 01:03:53.310 align:middle line:90% in the interest rate. 01:03:53.310 --> 01:03:55.155 align:middle line:84% That was given by high monitoring cost. 01:03:55.155 --> 01:03:57.992 align:middle line:90% 01:03:57.992 --> 01:03:59.950 align:middle line:84% It gave you a sensitivity, which I talked about 01:03:59.950 --> 01:04:02.320 align:middle line:90% in the last slide. 01:04:02.320 --> 01:04:05.020 align:middle line:84% It also implies that subsidizing the cost of capital 01:04:05.020 --> 01:04:07.690 align:middle line:84% can lead to welfare gains because r will go down, 01:04:07.690 --> 01:04:11.450 align:middle line:84% and this kind of allows for us to borrow more. 01:04:11.450 --> 01:04:13.300 align:middle line:84% It also suggests that these monitoring costs 01:04:13.300 --> 01:04:15.175 align:middle line:84% are really important because these monitoring 01:04:15.175 --> 01:04:18.640 align:middle line:84% costs are what's excluding people from credit. 01:04:18.640 --> 01:04:21.220 align:middle line:84% And this question of, how do we think 01:04:21.220 --> 01:04:23.362 align:middle line:84% about what kinds of monitoring costs 01:04:23.362 --> 01:04:24.820 align:middle line:84% are going to be really effective is 01:04:24.820 --> 01:04:25.990 align:middle line:84% going to be something we're going to talk about a lot 01:04:25.990 --> 01:04:27.790 align:middle line:84% more next time in the microcredit lectures. 01:04:27.790 --> 01:04:30.760 align:middle line:90% 01:04:30.760 --> 01:04:31.420 align:middle line:90% OK. 01:04:31.420 --> 01:04:32.110 align:middle line:90% Yes? 01:04:32.110 --> 01:04:34.930 align:middle line:90% STUDENT: Two slides ago-- 01:04:34.930 --> 01:04:36.543 align:middle line:90% PROFESSOR: This slide? 01:04:36.543 --> 01:04:37.210 align:middle line:90% STUDENT: Before. 01:04:37.210 --> 01:04:41.110 align:middle line:90% 01:04:41.110 --> 01:04:41.820 align:middle line:90% What is cost for? 01:04:41.820 --> 01:04:45.780 align:middle line:90% 01:04:45.780 --> 01:04:49.330 align:middle line:84% PROFESSOR: Oh, these lenders are not wildly profitable. 01:04:49.330 --> 01:04:50.660 align:middle line:90% That's all it means. 01:04:50.660 --> 01:04:52.803 align:middle line:84% So we're seeing these big spreads, 01:04:52.803 --> 01:04:54.970 align:middle line:84% but it's not like the microfinance organizations are 01:04:54.970 --> 01:04:57.340 align:middle line:90% making tons and tons of money. 01:04:57.340 --> 01:04:58.450 align:middle line:90% So that was rejected. 01:04:58.450 --> 01:04:59.380 align:middle line:84% Like I said, the other thing is, maybe 01:04:59.380 --> 01:05:00.580 align:middle line:84% they're not competitive in certain events. 01:05:00.580 --> 01:05:01.660 align:middle line:84% This bullet point says, well, maybe it's 01:05:01.660 --> 01:05:03.470 align:middle line:84% not that they're totally making lots of rounds. 01:05:03.470 --> 01:05:04.720 align:middle line:84% I think it could be part of it, but I 01:05:04.720 --> 01:05:06.280 align:middle line:84% think in general these organizations are not 01:05:06.280 --> 01:05:07.000 align:middle line:90% wildly profitable. 01:05:07.000 --> 01:05:08.800 align:middle line:84% A lot of them actually are losing money, or certainly 01:05:08.800 --> 01:05:11.370 align:middle line:84% kind of breaking even, which suggests that it's not-- 01:05:11.370 --> 01:05:14.460 align:middle line:84% although I do think the rent thing could be part of it, 01:05:14.460 --> 01:05:16.260 align:middle line:84% it's not mostly about rents, and that's 01:05:16.260 --> 01:05:17.802 align:middle line:84% why we kind of need these real costs. 01:05:17.802 --> 01:05:21.250 align:middle line:90% 01:05:21.250 --> 01:05:22.330 align:middle line:90% OK. 01:05:22.330 --> 01:05:27.610 align:middle line:84% So the final or penultimate thing 01:05:27.610 --> 01:05:30.360 align:middle line:84% I want to talk about today is this paper by-- 01:05:30.360 --> 01:05:31.875 align:middle line:90% I want to talk about the theory. 01:05:31.875 --> 01:05:33.250 align:middle line:84% So basically, we had three models 01:05:33.250 --> 01:05:34.420 align:middle line:90% that you should keep in mind-- 01:05:34.420 --> 01:05:36.253 align:middle line:84% moral hazard model, adverse selection model, 01:05:36.253 --> 01:05:37.850 align:middle line:90% and monitoring cost model. 01:05:37.850 --> 01:05:43.390 align:middle line:84% So those are three, I would say, frameworks or tools 01:05:43.390 --> 01:05:46.002 align:middle line:84% that you should keep in mind as you think through some 01:05:46.002 --> 01:05:46.960 align:middle line:90% of these credit issues. 01:05:46.960 --> 01:05:49.570 align:middle line:90% 01:05:49.570 --> 01:05:53.400 align:middle line:84% So this Karlan and Zinman paper was basically 01:05:53.400 --> 01:05:57.330 align:middle line:84% trying to understand, how do we think about, can we 01:05:57.330 --> 01:05:59.310 align:middle line:84% differentiate between the first two models, 01:05:59.310 --> 01:06:02.510 align:middle line:84% adverse selection models and moral hazard models. 01:06:02.510 --> 01:06:07.930 align:middle line:84% And prior to this paper, there was a bunch of previous papers 01:06:07.930 --> 01:06:13.270 align:middle line:84% who said, look, can we test for asymmetric information 01:06:13.270 --> 01:06:17.920 align:middle line:84% by looking for, say, a positive correlation 01:06:17.920 --> 01:06:23.417 align:middle line:84% between the interest rate and the default rate risk. 01:06:23.417 --> 01:06:25.250 align:middle line:84% That was kind of that was coming out before, 01:06:25.250 --> 01:06:26.170 align:middle line:84% but they didn't really say, was it 01:06:26.170 --> 01:06:28.587 align:middle line:84% adverse selection or hazard because both those models kind 01:06:28.587 --> 01:06:29.940 align:middle line:90% of produced the same thing. 01:06:29.940 --> 01:06:31.565 align:middle line:84% So they will say, well, can we actually 01:06:31.565 --> 01:06:35.180 align:middle line:84% do something to differentiate between those two. 01:06:35.180 --> 01:06:38.680 align:middle line:90% 01:06:38.680 --> 01:06:40.312 align:middle line:84% So again, adverse selection is the idea 01:06:40.312 --> 01:06:42.020 align:middle line:84% that some people are riskier than others, 01:06:42.020 --> 01:06:43.520 align:middle line:84% and it's a characteristic about you. 01:06:43.520 --> 01:06:45.280 align:middle line:84% And moral hazard is, people can choose 01:06:45.280 --> 01:06:50.810 align:middle line:84% how much risk they want to take, and that's an ex-post choice. 01:06:50.810 --> 01:06:55.120 align:middle line:84% And so these are going to look similar in the cross-section. 01:06:55.120 --> 01:06:57.230 align:middle line:90% We basically talked about that. 01:06:57.230 --> 01:06:59.620 align:middle line:84% But the idea of their paper is to say, well, 01:06:59.620 --> 01:07:04.060 align:middle line:84% can experimentally we do something to tease this apart. 01:07:04.060 --> 01:07:08.800 align:middle line:84% And this paper pioneered this experimental design, 01:07:08.800 --> 01:07:11.350 align:middle line:84% but it's now been used for many, many, many different kinds 01:07:11.350 --> 01:07:12.230 align:middle line:90% of things. 01:07:12.230 --> 01:07:13.730 align:middle line:84% And I think we may have already seen it already, 01:07:13.730 --> 01:07:15.147 align:middle line:84% but this is the original paper, so 01:07:15.147 --> 01:07:18.360 align:middle line:84% I wanted to go through it a little bit. 01:07:18.360 --> 01:07:22.340 align:middle line:84% So their idea was, adverse selection is driven 01:07:22.340 --> 01:07:25.310 align:middle line:90% by the ex-ante interest rate. 01:07:25.310 --> 01:07:28.180 align:middle line:84% Moral hazard is driven by the ex-post interest rate. 01:07:28.180 --> 01:07:32.420 align:middle line:84% And in the world, these are the same. 01:07:32.420 --> 01:07:34.640 align:middle line:84% The interest rate that you get when you-- 01:07:34.640 --> 01:07:36.110 align:middle line:84% like, I advertise and say, hey, I'm 01:07:36.110 --> 01:07:40.880 align:middle line:84% offering a loan at whatever, 10%. 01:07:40.880 --> 01:07:43.680 align:middle line:84% If you sign up for that loan, the rate is 10%. 01:07:43.680 --> 01:07:47.760 align:middle line:84% So those things in general, in the world, are the same. 01:07:47.760 --> 01:07:53.020 align:middle line:84% But maybe in an experiment, we can differentiate between them, 01:07:53.020 --> 01:07:56.800 align:middle line:84% and we can do that by surprising people with an interest rate, 01:07:56.800 --> 01:07:58.800 align:middle line:84% with a change in the interest rate after they've 01:07:58.800 --> 01:08:00.550 align:middle line:90% accepted the loan. 01:08:00.550 --> 01:08:03.540 align:middle line:84% So you advertise one rate, you get take up on one thing. 01:08:03.540 --> 01:08:06.930 align:middle line:84% You change it ex-post, and you get something else. 01:08:06.930 --> 01:08:17.229 align:middle line:84% And in particular, you can't, I think, make people worse off. 01:08:17.229 --> 01:08:19.151 align:middle line:84% I can't advertise 10% and surprise you to 20% 01:08:19.151 --> 01:08:21.609 align:middle line:84% because, number one, people would be really annoyed at you, 01:08:21.609 --> 01:08:24.279 align:middle line:84% and number two, you still get a selection effect because people 01:08:24.279 --> 01:08:26.020 align:middle line:90% would drop out again. 01:08:26.020 --> 01:08:27.910 align:middle line:84% But on the other hand, almost everyone 01:08:27.910 --> 01:08:30.620 align:middle line:84% is going to be happy with an improvement in their terms. 01:08:30.620 --> 01:08:34.710 align:middle line:84% So if I advertise a rate of 20%, and then surprise some people 01:08:34.710 --> 01:08:36.762 align:middle line:84% with a rate of 10%, there's going 01:08:36.762 --> 01:08:38.470 align:middle line:84% to be essentially no selection difference 01:08:38.470 --> 01:08:40.303 align:middle line:84% because everyone who wanted the thing at 20% 01:08:40.303 --> 01:08:41.689 align:middle line:90% will be happy to take it at 10%. 01:08:41.689 --> 01:08:43.314 align:middle line:84% So you can hold your selection constant 01:08:43.314 --> 01:08:46.660 align:middle line:84% but get a different kind of interest rate ex-post. 01:08:46.660 --> 01:08:50.380 align:middle line:90% So that's basically their idea. 01:08:50.380 --> 01:08:54.490 align:middle line:84% They have this other dynamic incentive thing, 01:08:54.490 --> 01:08:57.490 align:middle line:84% but that's, I think, neither here nor there. 01:08:57.490 --> 01:08:59.500 align:middle line:90% The basic idea is this. 01:08:59.500 --> 01:09:05.500 align:middle line:84% We just say, look, we offer people either a high rate 01:09:05.500 --> 01:09:08.319 align:middle line:90% or a low rate ex-ante. 01:09:08.319 --> 01:09:12.130 align:middle line:84% If you sign up for the loan with the low rate, 01:09:12.130 --> 01:09:14.035 align:middle line:90% you get the low rate. 01:09:14.035 --> 01:09:15.910 align:middle line:84% If you sign up for the loan with a high rate, 01:09:15.910 --> 01:09:17.560 align:middle line:84% with some probability, we surprise you 01:09:17.560 --> 01:09:19.220 align:middle line:90% and give you the low rates. 01:09:19.220 --> 01:09:24.069 align:middle line:84% So now, we have three boxes, and if we compare the high offers 01:09:24.069 --> 01:09:27.399 align:middle line:90% to the high offer-- 01:09:27.399 --> 01:09:30.700 align:middle line:84% and this middle box is the key box. 01:09:30.700 --> 01:09:33.460 align:middle line:84% Because if we compare the high offer, if we hold fixed 01:09:33.460 --> 01:09:36.220 align:middle line:84% the high offer rate and look at the actual contract rate, 01:09:36.220 --> 01:09:38.620 align:middle line:84% this box to this box tells us the moral hazard 01:09:38.620 --> 01:09:40.930 align:middle line:84% effect, because you both selected in at, say, 01:09:40.930 --> 01:09:43.090 align:middle line:84% 20%, but only some of you experienced 20%, 01:09:43.090 --> 01:09:44.500 align:middle line:90% and some of you experience 10%. 01:09:44.500 --> 01:09:48.850 align:middle line:84% And if we compare this box to this box, 01:09:48.850 --> 01:09:51.430 align:middle line:84% we get the pure selection effect because both of you guys 01:09:51.430 --> 01:09:53.649 align:middle line:84% experienced 10% ex-post, but here, 01:09:53.649 --> 01:09:55.900 align:middle line:84% you were selected in as having a 20% rate, and here, 01:09:55.900 --> 01:09:58.440 align:middle line:84% you're selected as having a 10% rate. 01:09:58.440 --> 01:10:02.887 align:middle line:84% And this design is useful for distinguishing 01:10:02.887 --> 01:10:04.470 align:middle line:84% between selection and moral hazard not 01:10:04.470 --> 01:10:07.480 align:middle line:84% just in credit, but kind of everywhere basically. 01:10:07.480 --> 01:10:09.870 align:middle line:84% And so people have done this thing in all other settings, 01:10:09.870 --> 01:10:13.230 align:middle line:84% trying to use this surprise improvement to differentiate 01:10:13.230 --> 01:10:14.690 align:middle line:84% between selection and moral hazard. 01:10:14.690 --> 01:10:16.470 align:middle line:84% It's a very general purpose experimental design. 01:10:16.470 --> 01:10:17.910 align:middle line:84% It's also, I think, a nice example 01:10:17.910 --> 01:10:20.820 align:middle line:84% of how an experiment can let you do something that you could not 01:10:20.820 --> 01:10:22.572 align:middle line:90% do in the real world. 01:10:22.572 --> 01:10:24.780 align:middle line:84% It's only because you have this ability to manipulate 01:10:24.780 --> 01:10:26.473 align:middle line:90% and ex-post change people's-- 01:10:26.473 --> 01:10:28.890 align:middle line:84% I mean, you could find a natural experiment to mimic this, 01:10:28.890 --> 01:10:31.320 align:middle line:84% but in general, it'll be hard to do. 01:10:31.320 --> 01:10:39.450 align:middle line:90% 01:10:39.450 --> 01:10:42.410 align:middle line:90% OK, is this clear? 01:10:42.410 --> 01:10:42.910 align:middle line:90% Yeah? 01:10:42.910 --> 01:10:44.170 align:middle line:90% STUDENT: One quick question. 01:10:44.170 --> 01:10:47.680 align:middle line:84% In the design that they're doing, 01:10:47.680 --> 01:10:51.730 align:middle line:84% the information of the adverse selection, 01:10:51.730 --> 01:10:57.730 align:middle line:84% is there some bundle of the selection type, 01:10:57.730 --> 01:11:03.640 align:middle line:84% but also potentially selection on moral hazard basically? 01:11:03.640 --> 01:11:05.770 align:middle line:84% Like, you wouldn't be able to distinguish that just 01:11:05.770 --> 01:11:08.168 align:middle line:90% from this, but you can-- 01:11:08.168 --> 01:11:10.210 align:middle line:84% PROFESSOR: Sorry, say again what you mean, Aaron. 01:11:10.210 --> 01:11:12.190 align:middle line:84% STUDENT: So with adverse selection, 01:11:12.190 --> 01:11:14.710 align:middle line:84% there's both I'm selecting on my p, 01:11:14.710 --> 01:11:16.555 align:middle line:84% but also I know how I would respond. 01:11:16.555 --> 01:11:20.950 align:middle line:90% 01:11:20.950 --> 01:11:23.590 align:middle line:84% There's also, I guess, this notion 01:11:23.590 --> 01:11:27.460 align:middle line:84% that you can select on how you would respond 01:11:27.460 --> 01:11:31.040 align:middle line:90% to different incentives. 01:11:31.040 --> 01:11:32.580 align:middle line:84% PROFESSOR: Oh, I think I understand. 01:11:32.580 --> 01:11:35.900 align:middle line:84% So you're saying, in a more general model, 01:11:35.900 --> 01:11:38.360 align:middle line:84% in addition to selecting on just having 01:11:38.360 --> 01:11:40.350 align:middle line:84% fixed p's or totally endogenous p's, 01:11:40.350 --> 01:11:42.350 align:middle line:84% I have another thing, which is, some people have 01:11:42.350 --> 01:11:43.370 align:middle line:90% different p of rs. 01:11:43.370 --> 01:11:44.150 align:middle line:90% Is that what you're saying? 01:11:44.150 --> 01:11:44.520 align:middle line:90% STUDENT: Right. 01:11:44.520 --> 01:11:45.020 align:middle line:90% Exactly. 01:11:45.020 --> 01:11:47.570 align:middle line:84% Yeah, so it's not anything we've discussed previously, 01:11:47.570 --> 01:11:49.278 align:middle line:84% but there's this other dimension that you 01:11:49.278 --> 01:11:51.470 align:middle line:84% could model that isn't captured by the comparisons 01:11:51.470 --> 01:11:52.603 align:middle line:90% that we're doing here. 01:11:52.603 --> 01:11:54.770 align:middle line:84% PROFESSOR: So OK, let's think that through together. 01:11:54.770 --> 01:11:56.878 align:middle line:84% So what's going to be picked up here? 01:11:56.878 --> 01:11:58.670 align:middle line:84% So you're saying, we have three dimensions. 01:11:58.670 --> 01:12:01.553 align:middle line:84% We have heterogeneity in p; ex-ante, your choice of p; 01:12:01.553 --> 01:12:03.970 align:middle line:84% and then you have this other thing where people may differ 01:12:03.970 --> 01:12:05.175 align:middle line:90% in their p of r function. 01:12:05.175 --> 01:12:05.800 align:middle line:90% STUDENT: Right. 01:12:05.800 --> 01:12:10.600 align:middle line:84% PROFESSOR: So how does that affect your interpretation 01:12:10.600 --> 01:12:11.870 align:middle line:90% here? 01:12:11.870 --> 01:12:14.240 align:middle line:84% STUDENT: I guess my interpretation 01:12:14.240 --> 01:12:17.510 align:middle line:84% is that the hidden information effect that they're picking up 01:12:17.510 --> 01:12:22.637 align:middle line:84% is some composite of the two different types of selection, 01:12:22.637 --> 01:12:24.595 align:middle line:84% that it's both selection on types and selection 01:12:24.595 --> 01:12:26.158 align:middle line:90% on slips basically. 01:12:26.158 --> 01:12:26.950 align:middle line:90% PROFESSOR: Hold on. 01:12:26.950 --> 01:12:28.100 align:middle line:90% Let me think for a minute. 01:12:28.100 --> 01:12:32.437 align:middle line:90% So it's clearly the case-- 01:12:32.437 --> 01:12:33.520 align:middle line:90% yes, I think that's right. 01:12:33.520 --> 01:12:35.145 align:middle line:84% So it's clearly the case, if we compare 01:12:35.145 --> 01:12:37.540 align:middle line:84% these people who had the same selection, some of them 01:12:37.540 --> 01:12:39.832 align:middle line:84% got the lower interest rate, that's a pure moral hazard 01:12:39.832 --> 01:12:41.500 align:middle line:90% effect. 01:12:41.500 --> 01:12:44.060 align:middle line:84% And if they compare this versus this, then-- 01:12:44.060 --> 01:12:47.888 align:middle line:90% 01:12:47.888 --> 01:12:48.680 align:middle line:90% yeah, you're right. 01:12:48.680 --> 01:12:51.440 align:middle line:84% I mean, yes, we don't know what-- 01:12:51.440 --> 01:12:54.020 align:middle line:84% they're experiencing the same p ex-post also. 01:12:54.020 --> 01:12:57.050 align:middle line:84% So what we're getting is like, in my example, 01:12:57.050 --> 01:12:58.700 align:middle line:90% we're both lowered to 10%. 01:12:58.700 --> 01:13:01.343 align:middle line:84% We're getting group number one experience 01:13:01.343 --> 01:13:03.260 align:middle line:84% at 10% and group number two experience at 10%. 01:13:03.260 --> 01:13:05.390 align:middle line:84% So you're right, we don't know what 01:13:05.390 --> 01:13:08.640 align:middle line:84% was the slope of their selection thing. 01:13:08.640 --> 01:13:11.330 align:middle line:84% But we are at least evaluating it at the same point, 01:13:11.330 --> 01:13:13.860 align:middle line:84% so it's still clearly picking up, 01:13:13.860 --> 01:13:15.470 align:middle line:84% I think, an adverse selection point. 01:13:15.470 --> 01:13:17.428 align:middle line:84% But yes, exactly, whether the adverse selection 01:13:17.428 --> 01:13:19.470 align:middle line:84% is a level or a slope, we are at least 01:13:19.470 --> 01:13:20.678 align:middle line:90% evaluating at the same point. 01:13:20.678 --> 01:13:24.164 align:middle line:90% 01:13:24.164 --> 01:13:25.870 align:middle line:84% One thing I'll say is that actually 01:13:25.870 --> 01:13:27.460 align:middle line:84% if you read the details of this paper, 01:13:27.460 --> 01:13:31.630 align:middle line:84% it's not the perfect setting to map into the model. 01:13:31.630 --> 01:13:34.450 align:middle line:84% It's consumer credit, and so it's a little harder 01:13:34.450 --> 01:13:35.250 align:middle line:90% to think about. 01:13:35.250 --> 01:13:37.458 align:middle line:84% I mean, of course, we can think about the same issues 01:13:37.458 --> 01:13:39.110 align:middle line:84% in consumer credit as in firm credit, 01:13:39.110 --> 01:13:41.568 align:middle line:84% but I don't think it necessarily maps to that model that we 01:13:41.568 --> 01:13:44.350 align:middle line:84% had in mind as nicely as maybe it could. 01:13:44.350 --> 01:13:49.090 align:middle line:84% But I think this paper was really innovative for creating 01:13:49.090 --> 01:13:50.740 align:middle line:90% this kind of general purpose. 01:13:50.740 --> 01:13:53.090 align:middle line:84% I mean, there may be another paper like it. 01:13:53.090 --> 01:13:55.340 align:middle line:84% This is the one I certainly know of that did it first. 01:13:55.340 --> 01:13:57.710 align:middle line:84% There may be others beforehand, but it certainly-- 01:13:57.710 --> 01:14:01.850 align:middle line:90% 01:14:01.850 --> 01:14:04.850 align:middle line:84% this design that they use here, I think, became very popular, 01:14:04.850 --> 01:14:09.320 align:middle line:84% and I think people have used it in a wide variety of settings. 01:14:09.320 --> 01:14:14.780 align:middle line:84% So in the end, the results are not super duper strong. 01:14:14.780 --> 01:14:16.220 align:middle line:84% They find a little bit of evidence 01:14:16.220 --> 01:14:21.860 align:middle line:84% of the contract rate, not a huge amount on the offer rate. 01:14:21.860 --> 01:14:24.260 align:middle line:84% But I want to focus on this paper 01:14:24.260 --> 01:14:26.982 align:middle line:84% less for the particular results they have, 01:14:26.982 --> 01:14:28.940 align:middle line:84% but more for that design of how you might think 01:14:28.940 --> 01:14:30.857 align:middle line:84% about distinguishing between adverse selection 01:14:30.857 --> 01:14:32.500 align:middle line:90% and moral hazard. 01:14:32.500 --> 01:14:33.940 align:middle line:90% OK, so I'll stop here. 01:14:33.940 --> 01:14:36.160 align:middle line:84% And then I will talk about microfinance, 01:14:36.160 --> 01:14:39.610 align:middle line:84% including the reading you guys did on Wednesday. 01:14:39.610 --> 01:14:41.460 align:middle line:90% OK, thanks. 01:14:41.460 --> 01:14:53.000 align:middle line:90%