WEBVTT 00:00:00.000 --> 00:00:00.979 align:middle line:90% 00:00:00.979 --> 00:00:01.937 align:middle line:90% [SQUEAKING] 00:00:01.937 --> 00:00:04.332 align:middle line:90% [RUSTLING] 00:00:04.332 --> 00:00:07.206 align:middle line:90% [CLICKING] 00:00:07.206 --> 00:00:10.208 align:middle line:90% 00:00:10.208 --> 00:00:12.500 align:middle line:84% JON GRUBER: A big week out of the Economics Department. 00:00:12.500 --> 00:00:14.840 align:middle line:90% Big news is the Nobel Prize. 00:00:14.840 --> 00:00:16.880 align:middle line:84% Very timely for this class because just a couple 00:00:16.880 --> 00:00:19.080 align:middle line:84% of lectures ago, we talked about Daron Acemoglu. 00:00:19.080 --> 00:00:21.080 align:middle line:84% I called him out as the best political economist 00:00:21.080 --> 00:00:22.880 align:middle line:84% in the world, and the Nobel Prize 00:00:22.880 --> 00:00:26.860 align:middle line:84% validated that, along with his collaborators, one of whom, 00:00:26.860 --> 00:00:28.303 align:middle line:84% Simon Johnson at the Sloan School, 00:00:28.303 --> 00:00:30.220 align:middle line:84% is also a good friend and a co-author of mine. 00:00:30.220 --> 00:00:33.140 align:middle line:84% So it's really been a thrill to see them get this award. 00:00:33.140 --> 00:00:35.740 align:middle line:84% And let's remember why they got this award, 00:00:35.740 --> 00:00:37.000 align:middle line:90% because it's so timely. 00:00:37.000 --> 00:00:40.000 align:middle line:84% And they got this award because their work is timeless. 00:00:40.000 --> 00:00:42.232 align:middle line:84% But their work is particularly timely. 00:00:42.232 --> 00:00:43.940 align:middle line:84% Because you remember what I talked about? 00:00:43.940 --> 00:00:47.240 align:middle line:84% I talked about how, if you will, the line 00:00:47.240 --> 00:00:49.880 align:middle line:84% between government success and government failure 00:00:49.880 --> 00:00:51.800 align:middle line:90% can be rather thin. 00:00:51.800 --> 00:00:55.080 align:middle line:84% And that basically what their work was so important in showing 00:00:55.080 --> 00:00:58.680 align:middle line:84% out-- in showing was-- and I had the box in the book 00:00:58.680 --> 00:01:01.370 align:middle line:84% about their study of countries succeeded or failed, which 00:01:01.370 --> 00:01:04.090 align:middle line:84% is one of, probably, maybe the most famous economics 00:01:04.090 --> 00:01:06.570 align:middle line:84% article of the last 25 years and the main reason that they 00:01:06.570 --> 00:01:07.810 align:middle line:90% won the Nobel. 00:01:07.810 --> 00:01:10.450 align:middle line:84% It's that basically institutions matter, 00:01:10.450 --> 00:01:14.010 align:middle line:84% and societies that weaken their institutions, 00:01:14.010 --> 00:01:16.570 align:middle line:84% particularly the democratic institutions, 00:01:16.570 --> 00:01:21.090 align:middle line:84% can move from a society that succeeds to one that fails. 00:01:21.090 --> 00:01:24.530 align:middle line:84% And they talk about comparisons like North and South Korea. 00:01:24.530 --> 00:01:27.650 align:middle line:84% Or right across the border, US-Mexico. 00:01:27.650 --> 00:01:31.930 align:middle line:84% Or other examples of places that were very comparable 00:01:31.930 --> 00:01:35.650 align:middle line:84% until institutional differences took them off the rails 00:01:35.650 --> 00:01:38.410 align:middle line:84% and drove them in very different directions. 00:01:38.410 --> 00:01:40.810 align:middle line:84% Obviously, understated in that is 00:01:40.810 --> 00:01:45.970 align:middle line:84% where America is right now, and the importance of 00:01:45.970 --> 00:01:48.490 align:middle line:84% maintaining democratic institutions and the fragility 00:01:48.490 --> 00:01:51.490 align:middle line:84% of those Democratic institutions. 00:01:51.490 --> 00:01:55.970 align:middle line:84% And understanding that it's so critical for how 00:01:55.970 --> 00:01:57.330 align:middle line:90% we think about this election. 00:01:57.330 --> 00:01:58.788 align:middle line:84% And how we think about what happens 00:01:58.788 --> 00:02:01.257 align:middle line:84% after the election, the role that we play as a society, 00:02:01.257 --> 00:02:03.590 align:middle line:84% to make sure that it won't close our eyes to any erosion 00:02:03.590 --> 00:02:06.310 align:middle line:84% of the democratic institutions that could leave our kids 00:02:06.310 --> 00:02:08.830 align:middle line:84% and our grandkids with a failed state in the future. 00:02:08.830 --> 00:02:10.928 align:middle line:90% So it's a really timely award. 00:02:10.928 --> 00:02:11.970 align:middle line:90% It's an incredible award. 00:02:11.970 --> 00:02:12.887 align:middle line:90% It's a profound award. 00:02:12.887 --> 00:02:14.937 align:middle line:84% And I really just think it's wonderful for MIT 00:02:14.937 --> 00:02:16.770 align:middle line:84% and for the world that they gave this award. 00:02:16.770 --> 00:02:19.350 align:middle line:84% So I urge you to read up on their own 00:02:19.350 --> 00:02:21.270 align:middle line:84% on Simon and James and their work. 00:02:21.270 --> 00:02:22.358 align:middle line:90% Read Why Nations Fail. 00:02:22.358 --> 00:02:24.150 align:middle line:84% Be one of the more than million-plus people 00:02:24.150 --> 00:02:24.858 align:middle line:90% to buy that book. 00:02:24.858 --> 00:02:27.062 align:middle line:84% It's an incredible book, incredibly successful book. 00:02:27.062 --> 00:02:29.270 align:middle line:84% And spend some time really reflecting on those issues 00:02:29.270 --> 00:02:32.110 align:middle line:84% because they're important for everyone all around the world. 00:02:32.110 --> 00:02:33.270 align:middle line:90% All right? 00:02:33.270 --> 00:02:36.110 align:middle line:84% So with that in mind, let's get much more mundane. 00:02:36.110 --> 00:02:39.130 align:middle line:84% Let's talk about some specific social insurance programs. 00:02:39.130 --> 00:02:42.310 align:middle line:84% Once again, we're in the social insurance part of the course. 00:02:42.310 --> 00:02:44.150 align:middle line:84% We set the stage with chapter 12, 00:02:44.150 --> 00:02:46.773 align:middle line:84% setting up the fundamental trade off of social insurance. 00:02:46.773 --> 00:02:48.190 align:middle line:84% How much could-- subject smoothing 00:02:48.190 --> 00:02:51.870 align:middle line:84% does it do versus how much moral hazard does it cause? 00:02:51.870 --> 00:02:54.990 align:middle line:84% We talked about applying that in the context of Social Security. 00:02:54.990 --> 00:02:57.440 align:middle line:84% Now we're going to turn to three other of 00:02:57.440 --> 00:03:00.360 align:middle line:84% the major social insurance programs in our society. 00:03:00.360 --> 00:03:04.680 align:middle line:84% And these programs, one of them I'll talk about 00:03:04.680 --> 00:03:06.280 align:middle line:90% is unemployment insurance. 00:03:06.280 --> 00:03:08.593 align:middle line:84% That is something-- a perfect example of unemployment 00:03:08.593 --> 00:03:10.760 align:middle line:84% insurance matters was during COVID, when we ended up 00:03:10.760 --> 00:03:12.600 align:middle line:84% with 6 times as many unemployment insurance 00:03:12.600 --> 00:03:15.080 align:middle line:84% recipients as we ever had because so many people lost 00:03:15.080 --> 00:03:16.418 align:middle line:90% their jobs. 00:03:16.418 --> 00:03:18.460 align:middle line:84% We realized the importance of programs like this. 00:03:18.460 --> 00:03:20.925 align:middle line:84% And I'll come back to what happened during COVID. 00:03:20.925 --> 00:03:23.300 align:middle line:84% But I'm going to talk about each of these three programs. 00:03:23.300 --> 00:03:25.960 align:middle line:84% There's a lot of institutional detail to cover today. 00:03:25.960 --> 00:03:28.200 align:middle line:84% So once again, I'm going to try to go pretty-- 00:03:28.200 --> 00:03:29.632 align:middle line:90% keep it pretty shallow. 00:03:29.632 --> 00:03:31.840 align:middle line:84% Ask me questions about the features of these programs 00:03:31.840 --> 00:03:33.298 align:middle line:84% that you feel like you need to know 00:03:33.298 --> 00:03:35.080 align:middle line:90% more about to understand them. 00:03:35.080 --> 00:03:38.920 align:middle line:84% So let's start by talking about unemployment insurance, OK? 00:03:38.920 --> 00:03:40.800 align:middle line:84% Unemployment insurance is the insurance 00:03:40.800 --> 00:03:46.040 align:middle line:90% we provide against layoffs. 00:03:46.040 --> 00:03:47.740 align:middle line:90% So UI's for layoffs. 00:03:47.740 --> 00:03:48.960 align:middle line:90% What does that mean? 00:03:48.960 --> 00:03:51.480 align:middle line:84% It's not for people who quit their job. 00:03:51.480 --> 00:03:53.760 align:middle line:84% It's not for people who get fired for cause. 00:03:53.760 --> 00:03:55.738 align:middle line:84% It's people who lose their job because-- not 00:03:55.738 --> 00:03:57.780 align:middle line:84% because their employer said they are bad workers, 00:03:57.780 --> 00:04:00.660 align:middle line:84% because the employer cannot afford to employ them anymore. 00:04:00.660 --> 00:04:03.132 align:middle line:90% OK? 00:04:03.132 --> 00:04:07.220 align:middle line:84% So that is a very important feature of this. 00:04:07.220 --> 00:04:11.820 align:middle line:84% Moreover, it's a program that requires 00:04:11.820 --> 00:04:13.700 align:middle line:84% to receive the benefits that you're actively 00:04:13.700 --> 00:04:15.740 align:middle line:90% looking for new work. 00:04:15.740 --> 00:04:17.700 align:middle line:90% So you have to be laid off. 00:04:17.700 --> 00:04:20.140 align:middle line:90% You have to be searching. 00:04:20.140 --> 00:04:22.160 align:middle line:84% And you have to be searching for a new job. 00:04:22.160 --> 00:04:25.060 align:middle line:90% 00:04:25.060 --> 00:04:27.900 align:middle line:90% Searching for a new job, OK? 00:04:27.900 --> 00:04:31.460 align:middle line:84% So that's another big requirement. 00:04:31.460 --> 00:04:38.560 align:middle line:84% So basically, that's who the program is for. 00:04:38.560 --> 00:04:42.820 align:middle line:84% What it provides you is a weekly benefit. 00:04:42.820 --> 00:04:47.300 align:middle line:84% And that weekly benefit is a function of your earnings 00:04:47.300 --> 00:04:48.680 align:middle line:90% before you lost your job. 00:04:48.680 --> 00:04:54.870 align:middle line:84% So if you look at figure 14-1, this shows a typical 00:04:54.870 --> 00:04:56.130 align:middle line:90% unemployment benefit schedule. 00:04:56.130 --> 00:04:58.070 align:middle line:84% This is the schedule for Michigan, 00:04:58.070 --> 00:05:00.510 align:middle line:84% but it's a typical unemployment schedule, 00:05:00.510 --> 00:05:04.230 align:middle line:84% where basically what determines your benefit is 00:05:04.230 --> 00:05:05.210 align:middle line:90% your weekly wage. 00:05:05.210 --> 00:05:06.877 align:middle line:84% In the highest quarter of the past year, 00:05:06.877 --> 00:05:09.168 align:middle line:84% it's actually the highest quarter in the year that ends 00:05:09.168 --> 00:05:11.630 align:middle line:84% the quarter before this one, it's-- very technically. 00:05:11.630 --> 00:05:13.550 align:middle line:84% But basically, the point is you have 00:05:13.550 --> 00:05:15.910 align:middle line:84% to show some attachment to the labor force. 00:05:15.910 --> 00:05:18.670 align:middle line:84% So in if in your most active quarter 00:05:18.670 --> 00:05:24.070 align:middle line:84% last year you weren't earning at least $262, you don't get UI. 00:05:24.070 --> 00:05:26.750 align:middle line:84% If you weren't earning $262 a week in your most active 00:05:26.750 --> 00:05:28.230 align:middle line:84% quarter, then they say you weren't really attached 00:05:28.230 --> 00:05:29.022 align:middle line:90% to the labor force. 00:05:29.022 --> 00:05:30.470 align:middle line:90% We're not going to give you UI. 00:05:30.470 --> 00:05:33.010 align:middle line:84% Once you're earning that much, you get a minimum benefit, 00:05:33.010 --> 00:05:36.310 align:middle line:84% that minimum benefit of about $150-something a week. 00:05:36.310 --> 00:05:40.830 align:middle line:84% Your benefit then rises at about 50% replacement. 00:05:40.830 --> 00:05:42.330 align:middle line:84% So for every dollar or more of wage, 00:05:42.330 --> 00:05:45.510 align:middle line:84% you have 50% more benefit until you get a maximum benefit. 00:05:45.510 --> 00:05:48.110 align:middle line:84% Here, the maximum benefit in Michigan 00:05:48.110 --> 00:05:54.520 align:middle line:84% is $362, which you hit a little less than $700 in wages. 00:05:54.520 --> 00:05:59.600 align:middle line:84% Now, one interesting feature about UI is that it-- 00:05:59.600 --> 00:06:04.080 align:middle line:84% unlike Social Security, it is a state, not a federal program. 00:06:04.080 --> 00:06:05.920 align:middle line:84% So we have 50 versions of these benefit 00:06:05.920 --> 00:06:07.960 align:middle line:90% schedules across the states. 00:06:07.960 --> 00:06:11.920 align:middle line:84% That is very exciting for empirical work. 00:06:11.920 --> 00:06:14.120 align:middle line:84% Because remember chapter 3, what we want 00:06:14.120 --> 00:06:16.063 align:middle line:84% is these quasi-experimental changes. 00:06:16.063 --> 00:06:17.480 align:middle line:84% Well, when the program's national, 00:06:17.480 --> 00:06:20.000 align:middle line:84% there's not a lot quasi-experimental change 00:06:20.000 --> 00:06:22.680 align:middle line:84% that's not-- doesn't affect the whole country. 00:06:22.680 --> 00:06:25.240 align:middle line:84% Well, here you have states running experiments all the time 00:06:25.240 --> 00:06:26.000 align:middle line:90% for you. 00:06:26.000 --> 00:06:28.000 align:middle line:84% And so it's become an incredibly active area 00:06:28.000 --> 00:06:30.540 align:middle line:84% of empirical research is studying the effect of UI, 00:06:30.540 --> 00:06:32.840 align:middle line:84% because there's so much variation across states 00:06:32.840 --> 00:06:34.320 align:middle line:90% and over time. 00:06:34.320 --> 00:06:36.640 align:middle line:84% Overall, the average replacement rate for this program 00:06:36.640 --> 00:06:38.400 align:middle line:90% is about 44%. 00:06:38.400 --> 00:06:41.620 align:middle line:84% So similar to Social Security, replace about 44% of your wages. 00:06:41.620 --> 00:06:44.240 align:middle line:84% But obviously, it varies depending on how high 00:06:44.240 --> 00:06:45.680 align:middle line:90% your wages. 00:06:45.680 --> 00:06:47.820 align:middle line:90% So that's the benefit you get. 00:06:47.820 --> 00:06:51.700 align:middle line:90% 00:06:51.700 --> 00:06:57.740 align:middle line:84% You get that benefit if, for a certain amount of time-- 00:06:57.740 --> 00:07:02.120 align:middle line:84% typically in a standard economy where things are going well, 00:07:02.120 --> 00:07:03.980 align:middle line:90% it's 26 weeks. 00:07:03.980 --> 00:07:05.760 align:middle line:84% So as long as you've worked enough, 00:07:05.760 --> 00:07:08.340 align:middle line:84% as long as you've been working for basically a year, 00:07:08.340 --> 00:07:11.620 align:middle line:84% you can get 26 weeks of unemployment insurance benefits. 00:07:11.620 --> 00:07:14.375 align:middle line:84% However, that number can vary a lot. 00:07:14.375 --> 00:07:16.500 align:middle line:84% For example, if you've been working for less than a 00:07:16.500 --> 00:07:19.440 align:middle line:84% year, in some states, they'll let you have UI benefits, 00:07:19.440 --> 00:07:22.380 align:middle line:90% but for a few-- less time. 00:07:22.380 --> 00:07:25.780 align:middle line:84% States have what they call trigger rules, which 00:07:25.780 --> 00:07:28.460 align:middle line:84% are basically if the employer-- if the economic situation gets 00:07:28.460 --> 00:07:31.740 align:middle line:84% particularly bad in that state, they will extend benefits 00:07:31.740 --> 00:07:34.820 align:middle line:90% beyond 26 weeks. 00:07:34.820 --> 00:07:36.740 align:middle line:84% And then the federal government also 00:07:36.740 --> 00:07:40.580 align:middle line:84% issues emergency extensions when the whole national economy 00:07:40.580 --> 00:07:42.500 align:middle line:84% is in bad shape, like during COVID. 00:07:42.500 --> 00:07:43.680 align:middle line:90% During the Great Recession. 00:07:43.680 --> 00:07:47.420 align:middle line:84% UI benefits actually went up to 99 weeks at the peak. 00:07:47.420 --> 00:07:50.150 align:middle line:84% So they can be extended a lot when 00:07:50.150 --> 00:07:52.990 align:middle line:90% economic times are quite bad. 00:07:52.990 --> 00:07:54.870 align:middle line:84% So you get this replacement rate, which 00:07:54.870 --> 00:07:56.130 align:middle line:90% is a function of your wages. 00:07:56.130 --> 00:07:58.270 align:middle line:84% You get it for 26 weeks, but it varies according 00:07:58.270 --> 00:08:00.310 align:middle line:90% to economic conditions. 00:08:00.310 --> 00:08:03.110 align:middle line:90% OK? 00:08:03.110 --> 00:08:07.710 align:middle line:84% Importantly, you only get it if you claim it. 00:08:07.710 --> 00:08:12.510 align:middle line:84% So it turns out only about 3/4 of those who qualify for UI 00:08:12.510 --> 00:08:15.110 align:middle line:90% actually take it. 00:08:15.110 --> 00:08:16.547 align:middle line:90% Now, it's free money. 00:08:16.547 --> 00:08:18.130 align:middle line:84% I mean, if you're unemployed, why not? 00:08:18.130 --> 00:08:19.450 align:middle line:90% And that's a fascinating issue. 00:08:19.450 --> 00:08:21.190 align:middle line:84% We'll come back to over and over again 00:08:21.190 --> 00:08:23.710 align:middle line:84% when we talk about social insurance, which 00:08:23.710 --> 00:08:27.270 align:middle line:84% is what we call the take-up problem, which 00:08:27.270 --> 00:08:29.830 align:middle line:84% is that just because someone's entitled to something 00:08:29.830 --> 00:08:32.630 align:middle line:90% doesn't mean they take it. 00:08:32.630 --> 00:08:34.270 align:middle line:90% Well, why not? 00:08:34.270 --> 00:08:38.270 align:middle line:84% Well, there's basically two classes 00:08:38.270 --> 00:08:40.429 align:middle line:84% of reasons, three classes of reasons 00:08:40.429 --> 00:08:44.270 align:middle line:84% why people don't take benefits to which they're entitled. 00:08:44.270 --> 00:08:46.590 align:middle line:90% The first is just information. 00:08:46.590 --> 00:08:50.890 align:middle line:84% They just don't know that they get these benefits. 00:08:50.890 --> 00:08:53.810 align:middle line:90% The second is hassle costs. 00:08:53.810 --> 00:08:54.710 align:middle line:90% It's a pain. 00:08:54.710 --> 00:08:57.070 align:middle line:84% You got to maybe go wait in a line. 00:08:57.070 --> 00:08:59.130 align:middle line:84% You got to write a letter and, say, you 00:08:59.130 --> 00:09:01.010 align:middle line:90% got to produce documentation. 00:09:01.010 --> 00:09:03.290 align:middle line:90% It's a pain. 00:09:03.290 --> 00:09:06.650 align:middle line:84% And the third is stigma, which is maybe 00:09:06.650 --> 00:09:10.530 align:middle line:84% you feel bad taking a government handout. 00:09:10.530 --> 00:09:13.410 align:middle line:84% Some combination of these three reasons 00:09:13.410 --> 00:09:15.450 align:middle line:84% explains why take-up is less than full. 00:09:15.450 --> 00:09:17.270 align:middle line:84% We don't really know what they are. 00:09:17.270 --> 00:09:22.410 align:middle line:84% We do know people respond rationally to incentives. 00:09:22.410 --> 00:09:23.790 align:middle line:84% That is, the higher is a benefit, 00:09:23.790 --> 00:09:26.250 align:middle line:90% the higher is the take-up. 00:09:26.250 --> 00:09:28.513 align:middle line:84% But that doesn't really rule out any of these 00:09:28.513 --> 00:09:29.930 align:middle line:84% because maybe it's higher benefit. 00:09:29.930 --> 00:09:31.750 align:middle line:84% Sort of rules out the information angle. 00:09:31.750 --> 00:09:34.483 align:middle line:84% But both hassle and stigma, if it's a big benefit, 00:09:34.483 --> 00:09:36.650 align:middle line:84% you only put up with the hassle and a lot of stigma. 00:09:36.650 --> 00:09:40.010 align:middle line:84% So the fact that take-up rises with higher benefits 00:09:40.010 --> 00:09:42.530 align:middle line:84% kind of says information isn't the full story, 00:09:42.530 --> 00:09:45.490 align:middle line:84% but it can't really differentiate between these two. 00:09:45.490 --> 00:09:47.980 align:middle line:84% So if you look at all workers who 00:09:47.980 --> 00:09:51.300 align:middle line:84% separate from the jobs in America, only about a quarter 00:09:51.300 --> 00:09:52.540 align:middle line:90% get UI. 00:09:52.540 --> 00:09:54.020 align:middle line:84% We believe unemployment insurance is something everybody 00:09:54.020 --> 00:09:54.978 align:middle line:90% who's not working gets. 00:09:54.978 --> 00:09:55.740 align:middle line:90% It's not true. 00:09:55.740 --> 00:09:58.260 align:middle line:84% By the time you take people who were laid off, 00:09:58.260 --> 00:10:01.220 align:middle line:84% who are actively searching, and who decide to take it, 00:10:01.220 --> 00:10:05.340 align:middle line:84% only about 25% of those who are unemployed at a point in time 00:10:05.340 --> 00:10:08.140 align:middle line:90% are actually getting UI. 00:10:08.140 --> 00:10:08.940 align:middle line:90% OK? 00:10:08.940 --> 00:10:13.140 align:middle line:84% So basically, that's how the program works. 00:10:13.140 --> 00:10:14.740 align:middle line:90% How is it paid for? 00:10:14.740 --> 00:10:17.580 align:middle line:84% It's paid for through a payroll tax. 00:10:17.580 --> 00:10:22.080 align:middle line:84% Your employer pays a payroll tax on your behalf. 00:10:22.080 --> 00:10:23.640 align:middle line:90% This payroll tax is small. 00:10:23.640 --> 00:10:26.620 align:middle line:84% It averages less than 1% of payroll. 00:10:26.620 --> 00:10:29.940 align:middle line:84% And it's partially experience-rated. 00:10:29.940 --> 00:10:35.687 align:middle line:90% Partially experience rated. 00:10:35.687 --> 00:10:36.520 align:middle line:90% What does that mean? 00:10:36.520 --> 00:10:38.103 align:middle line:84% Well, what does experience-rated mean? 00:10:38.103 --> 00:10:40.020 align:middle line:84% Experience-rated, we mentioned this before, 00:10:40.020 --> 00:10:43.340 align:middle line:84% is a system where you pay back what you collect. 00:10:43.340 --> 00:10:45.770 align:middle line:84% So in a perfectly experience-rated system, 00:10:45.770 --> 00:10:48.290 align:middle line:84% every dollar of benefits that your workers get you 00:10:48.290 --> 00:10:50.035 align:middle line:90% pay in higher taxes. 00:10:50.035 --> 00:10:51.410 align:middle line:84% That is not how the system works. 00:10:51.410 --> 00:10:53.530 align:middle line:84% It's a partially experience-rated system 00:10:53.530 --> 00:10:57.310 align:middle line:84% in that firms do not pay back exactly what their workers get. 00:10:57.310 --> 00:11:01.190 align:middle line:84% And we'll come back later to how that actually works in practice. 00:11:01.190 --> 00:11:02.510 align:middle line:90% That's UI. 00:11:02.510 --> 00:11:05.790 align:middle line:90% Questions about that, about UI? 00:11:05.790 --> 00:11:07.310 align:middle line:90% Yeah? 00:11:07.310 --> 00:11:07.910 align:middle line:90% Go ahead. 00:11:07.910 --> 00:11:10.270 align:middle line:84% AUDIENCE: Are there differences in how unemployment 00:11:10.270 --> 00:11:14.525 align:middle line:84% is defined based on whether the worker is on a W-2 or 1099? 00:11:14.525 --> 00:11:15.650 align:middle line:90% JON GRUBER: Great question. 00:11:15.650 --> 00:11:16.650 align:middle line:90% We'll come back to that. 00:11:16.650 --> 00:11:20.790 align:middle line:84% It really is designed for traditional workers on W-2. 00:11:20.790 --> 00:11:22.838 align:middle line:84% So as the economy shifts towards more gig work, 00:11:22.838 --> 00:11:25.130 align:middle line:84% it's presenting a challenge, which we saw during COVID. 00:11:25.130 --> 00:11:26.330 align:middle line:90% We'll come back to that. 00:11:26.330 --> 00:11:26.910 align:middle line:90% Yeah? 00:11:26.910 --> 00:11:29.330 align:middle line:84% AUDIENCE: So if you're a struggling with lots of layoffs, 00:11:29.330 --> 00:11:32.010 align:middle line:84% you'll get taxed more through these experiences? 00:11:32.010 --> 00:11:35.180 align:middle line:84% JON GRUBER: Yes, partially, up to a limit. 00:11:35.180 --> 00:11:36.930 align:middle line:84% If you have enough layoffs, at that point, 00:11:36.930 --> 00:11:38.772 align:middle line:90% they'll stop taxing you extra. 00:11:38.772 --> 00:11:40.230 align:middle line:84% So it's basically there's a ceiling 00:11:40.230 --> 00:11:41.650 align:middle line:90% on what your tax rate can be. 00:11:41.650 --> 00:11:44.250 align:middle line:90% 00:11:44.250 --> 00:11:46.910 align:middle line:84% And I want-- the normative tone in your question 00:11:46.910 --> 00:11:48.892 align:middle line:84% I want to come back to, which is, well, gee, 00:11:48.892 --> 00:11:51.350 align:middle line:84% it seems kind of mean to tax these firms when they're down. 00:11:51.350 --> 00:11:53.933 align:middle line:84% But let's come back to why you may or may not want to do that. 00:11:53.933 --> 00:11:55.370 align:middle line:90% We'll come back to that later. 00:11:55.370 --> 00:11:58.410 align:middle line:90% So that's UI. 00:11:58.410 --> 00:12:00.650 align:middle line:84% The second program we're going to talk about 00:12:00.650 --> 00:12:04.570 align:middle line:90% is Disability Insurance, or DI. 00:12:04.570 --> 00:12:10.690 align:middle line:84% DI is defined as a program for individuals 00:12:10.690 --> 00:12:16.890 align:middle line:84% who have become disabled and are unable to return to work. 00:12:16.890 --> 00:12:17.450 align:middle line:90% OK? 00:12:17.450 --> 00:12:20.050 align:middle line:84% Individuals who become disabled and are 00:12:20.050 --> 00:12:22.550 align:middle line:90% unable to return to work. 00:12:22.550 --> 00:12:27.130 align:middle line:84% So essentially, what happens if you become disabled? 00:12:27.130 --> 00:12:30.110 align:middle line:84% You have to first not work for five months. 00:12:30.110 --> 00:12:32.570 align:middle line:84% There's a five-month waiting period. 00:12:32.570 --> 00:12:34.270 align:middle line:84% Then, having not worked for five months, 00:12:34.270 --> 00:12:38.047 align:middle line:84% you can apply for disability insurance. 00:12:38.047 --> 00:12:40.130 align:middle line:84% There's an adjudication about whether you actually 00:12:40.130 --> 00:12:45.380 align:middle line:84% are disabled, and then you get a benefit. 00:12:45.380 --> 00:12:47.660 align:middle line:84% So let's talk about each of the pieces. 00:12:47.660 --> 00:12:50.460 align:middle line:84% First of all, what is the benefit? 00:12:50.460 --> 00:12:54.700 align:middle line:84% The benefit you get is you get your PIA, but at whatever age 00:12:54.700 --> 00:12:56.120 align:middle line:90% you're adjudicated disabled. 00:12:56.120 --> 00:12:58.460 align:middle line:84% So remember, your PIA, that's your full Social Security 00:12:58.460 --> 00:12:59.120 align:middle line:90% benefit. 00:12:59.120 --> 00:13:00.920 align:middle line:84% What you get if you're retired at 67, 00:13:00.920 --> 00:13:03.840 align:middle line:84% you get that at 30, or 35, or 40, whenever you're disabled. 00:13:03.840 --> 00:13:04.780 align:middle line:90% So it's pretty good. 00:13:04.780 --> 00:13:06.740 align:middle line:84% It's not great because your 35 highest 00:13:06.740 --> 00:13:10.020 align:middle line:90% years are not that high yet. 00:13:10.020 --> 00:13:11.760 align:middle line:84% But basically, if you tell someone at 50, 00:13:11.760 --> 00:13:14.427 align:middle line:84% they get to retire in their full Social Security benefit, that's 00:13:14.427 --> 00:13:15.820 align:middle line:90% not a bad deal for many people. 00:13:15.820 --> 00:13:16.400 align:middle line:90% OK? 00:13:16.400 --> 00:13:18.940 align:middle line:90% So benefits are fairly generous. 00:13:18.940 --> 00:13:19.600 align:middle line:90% OK? 00:13:19.600 --> 00:13:21.880 align:middle line:90% That's what you get. 00:13:21.880 --> 00:13:26.240 align:middle line:84% Moreover, once you are on DI after two years, 00:13:26.240 --> 00:13:27.240 align:middle line:90% you also get Medicare. 00:13:27.240 --> 00:13:29.500 align:middle line:84% You get universal health insurance coverage. 00:13:29.500 --> 00:13:33.100 align:middle line:84% So it's a benefit plus health insurance coverage. 00:13:33.100 --> 00:13:34.820 align:middle line:90% How is it decided? 00:13:34.820 --> 00:13:38.500 align:middle line:90% Well, people apply for DI. 00:13:38.500 --> 00:13:39.920 align:middle line:90% People apply for DI. 00:13:39.920 --> 00:13:44.470 align:middle line:84% And there's an adjudication done by a state board. 00:13:44.470 --> 00:13:47.190 align:middle line:84% And they basically decide, are you disabled? 00:13:47.190 --> 00:13:49.870 align:middle line:84% And the way the system works today is about 1/3 of people who 00:13:49.870 --> 00:13:52.910 align:middle line:84% apply are adjudicated as disabled. 00:13:52.910 --> 00:13:54.730 align:middle line:84% But if you're not attributed as disabled, 00:13:54.730 --> 00:13:59.070 align:middle line:84% you can appeal to what's called administrative court. 00:13:59.070 --> 00:14:01.190 align:middle line:84% And you could actually technically go all the way 00:14:01.190 --> 00:14:03.110 align:middle line:90% to the Supreme Court. 00:14:03.110 --> 00:14:05.670 align:middle line:84% And after all the appeals are taken into account, about half 00:14:05.670 --> 00:14:07.090 align:middle line:90% of the people apply get it. 00:14:07.090 --> 00:14:08.390 align:middle line:90% But a third get it right away. 00:14:08.390 --> 00:14:11.670 align:middle line:84% Of the other 2/3 that don't get it, a bunch get it on appeal. 00:14:11.670 --> 00:14:14.370 align:middle line:84% So then over time, roughly speaking, of those who apply, 00:14:14.370 --> 00:14:16.710 align:middle line:90% about half get it. 00:14:16.710 --> 00:14:21.110 align:middle line:84% Now, the trick is this adjudication is very difficult. 00:14:21.110 --> 00:14:22.510 align:middle line:84% It's very hard to tell someone is 00:14:22.510 --> 00:14:24.630 align:middle line:84% disabled because most disability are 00:14:24.630 --> 00:14:26.910 align:middle line:84% for things which are not like losing your arm 00:14:26.910 --> 00:14:28.630 align:middle line:90% or becoming paralyzed. 00:14:28.630 --> 00:14:30.950 align:middle line:84% Therefore, things like chronic back pain, 00:14:30.950 --> 00:14:33.110 align:middle line:84% or depression, or things like that. 00:14:33.110 --> 00:14:35.210 align:middle line:90% They're very hard to determine. 00:14:35.210 --> 00:14:38.670 align:middle line:84% Moreover, the way DI works is it technically says cannot go 00:14:38.670 --> 00:14:39.850 align:middle line:90% to work. 00:14:39.850 --> 00:14:43.090 align:middle line:84% You get DI if you cannot go back and do work substantially 00:14:43.090 --> 00:14:44.510 align:middle line:90% similar to what-- 00:14:44.510 --> 00:14:46.890 align:middle line:84% basically, you cannot do work you're qualified for. 00:14:46.890 --> 00:14:52.592 align:middle line:84% So if you're a high school dropout, lumberjack, and you-- 00:14:52.592 --> 00:14:54.050 align:middle line:84% your legs, and you become disabled, 00:14:54.050 --> 00:14:57.490 align:middle line:84% and you're in a wheelchair, clearly you're done. 00:14:57.490 --> 00:14:59.290 align:middle line:84% But if you're a software engineer, 00:14:59.290 --> 00:15:01.330 align:middle line:84% it's not entirely clear you are done 00:15:01.330 --> 00:15:03.210 align:middle line:90% because you're in a wheelchair. 00:15:03.210 --> 00:15:06.130 align:middle line:84% So basically, the adjudicator has this difficult decision 00:15:06.130 --> 00:15:08.650 align:middle line:84% deciding whether you're disabled and whether your disability 00:15:08.650 --> 00:15:11.490 align:middle line:84% means you can't do basically something that you're 00:15:11.490 --> 00:15:13.210 align:middle line:90% trained to do. 00:15:13.210 --> 00:15:16.050 align:middle line:84% And as a result of how difficult this is, 00:15:16.050 --> 00:15:17.970 align:middle line:84% they ran an experiment where they 00:15:17.970 --> 00:15:20.890 align:middle line:84% took a set of disability insurance adjudicators 00:15:20.890 --> 00:15:22.880 align:middle line:84% and gave them a set of files to evaluate. 00:15:22.880 --> 00:15:25.130 align:middle line:84% And a year later, they randomly gave them the same set 00:15:25.130 --> 00:15:26.550 align:middle line:90% of files to evaluate. 00:15:26.550 --> 00:15:29.608 align:middle line:84% And they changed their mind in almost a quarter of them. 00:15:29.608 --> 00:15:31.150 align:middle line:84% So it's very hard to figure this out. 00:15:31.150 --> 00:15:32.030 align:middle line:84% I mean, it's not that they're bad. 00:15:32.030 --> 00:15:32.655 align:middle line:90% It's their job. 00:15:32.655 --> 00:15:33.490 align:middle line:90% It's just hard. 00:15:33.490 --> 00:15:33.990 align:middle line:90% OK? 00:15:33.990 --> 00:15:36.850 align:middle line:84% So there's a lot of randomness in this process. 00:15:36.850 --> 00:15:40.420 align:middle line:84% Now, this program is a national program. 00:15:40.420 --> 00:15:43.198 align:middle line:84% In fact, it's part of the Social Security program. 00:15:43.198 --> 00:15:44.740 align:middle line:84% Social Security is technically called 00:15:44.740 --> 00:15:50.100 align:middle line:84% the OASDI, Old Age Security and Disability Insurance Program. 00:15:50.100 --> 00:15:53.500 align:middle line:84% So part of your payroll tax, your 6.2% payroll tax, 00:15:53.500 --> 00:15:54.820 align:middle line:90% is going to pay for DI. 00:15:54.820 --> 00:15:56.560 align:middle line:84% DI is actually a rather large share. 00:15:56.560 --> 00:16:00.020 align:middle line:84% DI is about 14% of the entire Social Security program. 00:16:00.020 --> 00:16:02.340 align:middle line:90% It's about $140 billion a year. 00:16:02.340 --> 00:16:04.460 align:middle line:90% So it's a big program. 00:16:04.460 --> 00:16:06.300 align:middle line:84% It's national, although the adjudication 00:16:06.300 --> 00:16:09.460 align:middle line:90% happens at the state level. 00:16:09.460 --> 00:16:12.980 align:middle line:84% So there's a little state variation, but mostly national. 00:16:12.980 --> 00:16:13.780 align:middle line:90% OK? 00:16:13.780 --> 00:16:17.260 align:middle line:90% Questions about that? 00:16:17.260 --> 00:16:19.940 align:middle line:84% Now, the third type of social insurance we'll cover 00:16:19.940 --> 00:16:22.840 align:middle line:84% is the nation's oldest social insurance program, 00:16:22.840 --> 00:16:25.180 align:middle line:90% workers' compensation. 00:16:25.180 --> 00:16:26.020 align:middle line:90% OK? 00:16:26.020 --> 00:16:29.980 align:middle line:84% Workers' compensation differs from DI and UI 00:16:29.980 --> 00:16:31.460 align:middle line:90% in an important way. 00:16:31.460 --> 00:16:36.160 align:middle line:84% This is DI and UI are government-provided insurance 00:16:36.160 --> 00:16:38.760 align:middle line:84% that you fund through a payroll tax. 00:16:38.760 --> 00:16:44.160 align:middle line:84% Workers' comp is mandated private insurance. 00:16:44.160 --> 00:16:46.040 align:middle line:84% This is insurance that covers you 00:16:46.040 --> 00:16:49.400 align:middle line:84% against getting injured on the job. 00:16:49.400 --> 00:16:52.120 align:middle line:84% So it's insurance to cover you getting injured on the job. 00:16:52.120 --> 00:16:56.880 align:middle line:84% The state specifies what you're going to get if you get injured. 00:16:56.880 --> 00:16:59.898 align:middle line:84% But it then says you have to go find insurance to cover that. 00:16:59.898 --> 00:17:01.440 align:middle line:84% It says, look, if you get the injury, 00:17:01.440 --> 00:17:02.732 align:middle line:90% you have to get paid this much. 00:17:02.732 --> 00:17:04.900 align:middle line:84% And you have to go find an insurer to buy that from. 00:17:04.900 --> 00:17:06.400 align:middle line:84% In some states, the state itself can 00:17:06.400 --> 00:17:08.260 align:middle line:84% be the insurer or one of the insurers, 00:17:08.260 --> 00:17:10.700 align:middle line:84% but generally people buy private insurance to cover this. 00:17:10.700 --> 00:17:14.280 align:middle line:84% So it is a mandated employer-provided insurance 00:17:14.280 --> 00:17:16.000 align:middle line:90% benefit. 00:17:16.000 --> 00:17:17.000 align:middle line:90% OK? 00:17:17.000 --> 00:17:19.140 align:middle line:84% We're going to show later in chapter 18, 00:17:19.140 --> 00:17:20.599 align:middle line:90% effectively it's the same thing. 00:17:20.599 --> 00:17:22.540 align:middle line:84% Employers pay money, workers get a benefit. 00:17:22.540 --> 00:17:25.160 align:middle line:84% It's just how you're labeling it. 00:17:25.160 --> 00:17:29.320 align:middle line:90% But it is different technically. 00:17:29.320 --> 00:17:32.720 align:middle line:84% And once again, it's partially experience rated. 00:17:32.720 --> 00:17:36.050 align:middle line:84% What you pay to your insurer depends partly 00:17:36.050 --> 00:17:38.368 align:middle line:84% on your own accident experience, but not fully. 00:17:38.368 --> 00:17:39.910 align:middle line:84% Essentially, the bigger your firm is, 00:17:39.910 --> 00:17:41.630 align:middle line:84% the more it's based on your own experience. 00:17:41.630 --> 00:17:43.047 align:middle line:84% The smaller your firm is, the less 00:17:43.047 --> 00:17:46.970 align:middle line:84% they base it on your experience, because accidents can be random. 00:17:46.970 --> 00:17:47.890 align:middle line:90% OK? 00:17:47.890 --> 00:17:51.170 align:middle line:84% Workers' comp has two components. 00:17:51.170 --> 00:17:53.690 align:middle line:84% There's a cash benefit, which you 00:17:53.690 --> 00:17:56.290 align:middle line:84% get if you're out of work, and coverage 00:17:56.290 --> 00:18:01.050 align:middle line:84% for your medical expenses arising from being hurt. 00:18:01.050 --> 00:18:04.810 align:middle line:84% So if you get hurt at work, it covers both your-- 00:18:04.810 --> 00:18:07.330 align:middle line:84% it both pays you a weekly stipend 00:18:07.330 --> 00:18:09.290 align:middle line:84% and covers your medical benefits. 00:18:09.290 --> 00:18:12.330 align:middle line:84% The weekly stipend varies across states, 00:18:12.330 --> 00:18:14.770 align:middle line:84% but is generally pretty generous. 00:18:14.770 --> 00:18:19.210 align:middle line:84% The typical workers' comp weekly stipend covers about 75% 00:18:19.210 --> 00:18:24.250 align:middle line:84% of lost wages, and it's not taxed. 00:18:24.250 --> 00:18:27.290 align:middle line:84% So when you count taxes, the after-tax replacement rate 00:18:27.290 --> 00:18:30.650 align:middle line:84% for workers' comp is on the order of 90%. 00:18:30.650 --> 00:18:33.140 align:middle line:84% So it's pretty generous replacement for the wages 00:18:33.140 --> 00:18:35.560 align:middle line:84% that you lose, but it varies across state. 00:18:35.560 --> 00:18:38.580 align:middle line:84% Indeed, my favorite table of any paper ever published is what I 00:18:38.580 --> 00:18:40.057 align:middle line:90% call the Gory Table, table 14-1. 00:18:40.057 --> 00:18:42.140 align:middle line:84% This is how much you get for losing different body 00:18:42.140 --> 00:18:44.140 align:middle line:90% parts in different states. 00:18:44.140 --> 00:18:46.502 align:middle line:84% So for example, if you're working in Indiana 00:18:46.502 --> 00:18:47.960 align:middle line:84% and you lose your arm and your leg, 00:18:47.960 --> 00:18:49.060 align:middle line:84% you might want to try to drag yourself 00:18:49.060 --> 00:18:50.260 align:middle line:84% across the border to Illinois, where 00:18:50.260 --> 00:18:51.680 align:middle line:90% you get three times as much. 00:18:51.680 --> 00:18:52.820 align:middle line:84% No, you can't, actually, because it's based 00:18:52.820 --> 00:18:53.960 align:middle line:90% on where your employer is. 00:18:53.960 --> 00:18:55.460 align:middle line:90% But it makes a good joke. 00:18:55.460 --> 00:18:58.123 align:middle line:84% Basically, there's a lot of variation. 00:18:58.123 --> 00:19:00.040 align:middle line:84% This table is kind of fun because you see, oh, 00:19:00.040 --> 00:19:03.400 align:middle line:84% is losing your foot really worth $90,000 or $150,000? 00:19:03.400 --> 00:19:06.100 align:middle line:84% But in some sense, it's more cute than useful. 00:19:06.100 --> 00:19:08.260 align:middle line:84% The vast majority of claims are actually 00:19:08.260 --> 00:19:10.020 align:middle line:84% what we call a temporary injury, which 00:19:10.020 --> 00:19:12.140 align:middle line:84% is the last column, which is what you 00:19:12.140 --> 00:19:14.300 align:middle line:90% get per week if you can't work. 00:19:14.300 --> 00:19:15.840 align:middle line:90% It's not some permanent damage. 00:19:15.840 --> 00:19:18.880 align:middle line:84% Mostly, workers' comp is not for permanent things. 00:19:18.880 --> 00:19:21.008 align:middle line:84% It's for temporary things that go wrong 00:19:21.008 --> 00:19:22.800 align:middle line:84% and you can't work for some period of time. 00:19:22.800 --> 00:19:25.540 align:middle line:84% But you even see that varies a lot over how much you get 00:19:25.540 --> 00:19:27.160 align:middle line:90% over 10 weeks-- varies a lot. 00:19:27.160 --> 00:19:29.700 align:middle line:84% Once you get to Illinois, you get 00:19:29.700 --> 00:19:31.220 align:middle line:84% twice as much as you get in Indiana 00:19:31.220 --> 00:19:33.680 align:middle line:90% if you're hurt for 10 weeks. 00:19:33.680 --> 00:19:34.300 align:middle line:90% OK? 00:19:34.300 --> 00:19:37.320 align:middle line:90% 00:19:37.320 --> 00:19:40.537 align:middle line:84% Workers' comp is essentially no-fault insurance. 00:19:40.537 --> 00:19:42.620 align:middle line:84% So you're buying insurance from an insurer saying, 00:19:42.620 --> 00:19:44.037 align:middle line:84% look, it doesn't matter if you got 00:19:44.037 --> 00:19:45.625 align:middle line:84% hurt because your employer tripped you 00:19:45.625 --> 00:19:46.500 align:middle line:90% as you're walking by. 00:19:46.500 --> 00:19:47.292 align:middle line:90% I guess not really. 00:19:47.292 --> 00:19:49.995 align:middle line:84% But if it was a shitty ladder and it collapses, 00:19:49.995 --> 00:19:52.120 align:middle line:84% that's the same as your employer has a great ladder 00:19:52.120 --> 00:19:53.243 align:middle line:90% and you fall off. 00:19:53.243 --> 00:19:54.660 align:middle line:84% Either way, you get workers' comp. 00:19:54.660 --> 00:19:55.740 align:middle line:90% It's no-fault insurance. 00:19:55.740 --> 00:19:56.983 align:middle line:90% It arose, actually, because-- 00:19:56.983 --> 00:19:58.400 align:middle line:84% it's actually kind of interesting. 00:19:58.400 --> 00:20:02.065 align:middle line:84% Before the program arose in the early 20th century, 00:20:02.065 --> 00:20:04.440 align:middle line:84% if you got hurt at work, the only way to compensation was 00:20:04.440 --> 00:20:06.480 align:middle line:90% to sue your employer. 00:20:06.480 --> 00:20:08.665 align:middle line:84% So it was an at-fault system based on a lawsuit. 00:20:08.665 --> 00:20:10.040 align:middle line:84% The problem was, I guess, lawyers 00:20:10.040 --> 00:20:11.980 align:middle line:84% hadn't figured out working on contingency yet. 00:20:11.980 --> 00:20:13.980 align:middle line:84% So if you're poor, you couldn't afford a lawyer. 00:20:13.980 --> 00:20:15.820 align:middle line:84% There wasn't Morgan and Morgan or whatever. 00:20:15.820 --> 00:20:16.320 align:middle line:90% OK? 00:20:16.320 --> 00:20:17.660 align:middle line:90% You couldn't afford a lawyer. 00:20:17.660 --> 00:20:21.185 align:middle line:84% So basically, poor people had no recourse was one problem. 00:20:21.185 --> 00:20:22.560 align:middle line:84% The other problem with the system 00:20:22.560 --> 00:20:26.400 align:middle line:84% is we have another term for lawyers in this class. 00:20:26.400 --> 00:20:27.860 align:middle line:90% They're called deadweight loss. 00:20:27.860 --> 00:20:28.595 align:middle line:90% [LAUGHING] 00:20:28.595 --> 00:20:30.220 align:middle line:84% Why are lawyers called deadweight loss? 00:20:30.220 --> 00:20:32.890 align:middle line:84% Well, all lawyers do is transfer money from party A to party B 00:20:32.890 --> 00:20:35.130 align:middle line:90% and take some out along the way. 00:20:35.130 --> 00:20:36.010 align:middle line:90% OK? 00:20:36.010 --> 00:20:38.570 align:middle line:84% So basically, the tort system has a deadweight loss 00:20:38.570 --> 00:20:39.498 align:middle line:90% associated with it. 00:20:39.498 --> 00:20:41.790 align:middle line:84% And the view was, rather than all this deadweight loss, 00:20:41.790 --> 00:20:44.210 align:middle line:84% why don't we just go out and give them a benefit? 00:20:44.210 --> 00:20:46.270 align:middle line:84% However, the benefit is no fault, 00:20:46.270 --> 00:20:49.410 align:middle line:84% which raises moral hazard concerns we'll come back to. 00:20:49.410 --> 00:20:53.210 align:middle line:84% OK, so that is how workers' comp works. 00:20:53.210 --> 00:20:53.810 align:middle line:90% Yeah? 00:20:53.810 --> 00:20:56.670 align:middle line:84% AUDIENCE: If you're hurt on the job, but you can still work, 00:20:56.670 --> 00:20:57.930 align:middle line:90% is that workers' comp-- 00:20:57.930 --> 00:20:59.570 align:middle line:84% JON GRUBER: If you're hurt on the job, 00:20:59.570 --> 00:21:03.210 align:middle line:84% basically the way it works is an adjudication, which 00:21:03.210 --> 00:21:07.190 align:middle line:84% is you go to a doctor and they say whether you can work or not. 00:21:07.190 --> 00:21:08.705 align:middle line:90% AUDIENCE: And if you can work? 00:21:08.705 --> 00:21:10.830 align:middle line:84% JON GRUBER: If the doctor says you're able to work, 00:21:10.830 --> 00:21:12.530 align:middle line:84% then you don't get workers' comp. 00:21:12.530 --> 00:21:14.030 align:middle line:84% AUDIENCE: And then my third question 00:21:14.030 --> 00:21:15.650 align:middle line:84% was, can you explain the workers' comp 00:21:15.650 --> 00:21:17.702 align:middle line:84% with the partially-experienced ratings again? 00:21:17.702 --> 00:21:18.410 align:middle line:90% JON GRUBER: Yeah. 00:21:18.410 --> 00:21:21.010 align:middle line:84% So what that means is full-experience rated would 00:21:21.010 --> 00:21:23.110 align:middle line:90% mean if Gruber Co. 00:21:23.110 --> 00:21:25.510 align:middle line:84% pays that $10,000 of benefits to my hurt workers, 00:21:25.510 --> 00:21:27.810 align:middle line:90% then I pay $10,000 back in tax. 00:21:27.810 --> 00:21:29.750 align:middle line:84% So basically it's just like they're just 00:21:29.750 --> 00:21:31.950 align:middle line:84% allowing me to smooth my consumption-- my firm 00:21:31.950 --> 00:21:33.510 align:middle line:90% spending over time. 00:21:33.510 --> 00:21:36.230 align:middle line:84% That for a large firm, that's basically what they do. 00:21:36.230 --> 00:21:38.430 align:middle line:84% But for a small firm, if Gruber Co. 00:21:38.430 --> 00:21:41.190 align:middle line:84% has two employees, it's pretty rare 00:21:41.190 --> 00:21:42.570 align:middle line:90% that one's going to get hurt. 00:21:42.570 --> 00:21:44.630 align:middle line:84% And that's a pretty big shock if one gets lost. 00:21:44.630 --> 00:21:46.630 align:middle line:84% So then they'll say, well, look, your tax rate's 00:21:46.630 --> 00:21:48.130 align:middle line:84% a function partly of your experience 00:21:48.130 --> 00:21:51.750 align:middle line:84% or partly also of just an average experience in the state. 00:21:51.750 --> 00:21:53.110 align:middle line:90% Yeah? 00:21:53.110 --> 00:21:56.010 align:middle line:84% AUDIENCE: Does this chart apply to things like fingers or toes? 00:21:56.010 --> 00:21:58.590 align:middle line:84% JON GRUBER: Yeah, they got everything. 00:21:58.590 --> 00:21:59.710 align:middle line:90% They have everything. 00:21:59.710 --> 00:22:03.150 align:middle line:84% Expands out toward every body part. 00:22:03.150 --> 00:22:04.750 align:middle line:90% All right. 00:22:04.750 --> 00:22:07.550 align:middle line:84% So other question about his work? 00:22:07.550 --> 00:22:09.770 align:middle line:84% So let's compare him with this overview. 00:22:09.770 --> 00:22:13.630 align:middle line:84% Let's look at table 14.2, which basically is a summary of how 00:22:13.630 --> 00:22:15.590 align:middle line:90% these programs compare. 00:22:15.590 --> 00:22:18.870 align:middle line:84% So essentially, what's the qualifying event? 00:22:18.870 --> 00:22:20.590 align:middle line:84% It's either unemployment, or disability, 00:22:20.590 --> 00:22:22.550 align:middle line:90% or on-the-job injury. 00:22:22.550 --> 00:22:24.310 align:middle line:90% How long does it last? 00:22:24.310 --> 00:22:28.280 align:middle line:84% 26 weeks for UI, although it can vary. 00:22:28.280 --> 00:22:30.183 align:middle line:90% Indefinitely for DI. 00:22:30.183 --> 00:22:32.100 align:middle line:84% Basically, once you're on DI, you never leave. 00:22:32.100 --> 00:22:33.225 align:middle line:90% Technically, you can leave. 00:22:33.225 --> 00:22:35.480 align:middle line:84% No one ever leaves once they're on DI. 00:22:35.480 --> 00:22:37.400 align:middle line:84% And then for workers' comp, it's indefinite 00:22:37.400 --> 00:22:41.200 align:middle line:84% until a doctor says you can go back to work. 00:22:41.200 --> 00:22:42.440 align:middle line:90% OK? 00:22:42.440 --> 00:22:46.760 align:middle line:84% How hard is it to verify whether you're actually qualified? 00:22:46.760 --> 00:22:50.160 align:middle line:84% Well, to verify whether you lost your job is pretty darn easy. 00:22:50.160 --> 00:22:53.840 align:middle line:84% To verify whether you're looking for work is really hard. 00:22:53.840 --> 00:22:57.960 align:middle line:84% To verify whether you're disabled is pretty hard. 00:22:57.960 --> 00:23:02.600 align:middle line:84% To verify if you're hurt at work and can't work is really hard. 00:23:02.600 --> 00:23:03.360 align:middle line:90% OK? 00:23:03.360 --> 00:23:05.008 align:middle line:84% Much harder because it's much harder 00:23:05.008 --> 00:23:07.300 align:middle line:84% to decide if you can go back to work today or tomorrow, 00:23:07.300 --> 00:23:09.440 align:middle line:84% versus if you could never go back to work. 00:23:09.440 --> 00:23:11.240 align:middle line:90% OK? 00:23:11.240 --> 00:23:12.060 align:middle line:90% What do you get? 00:23:12.060 --> 00:23:15.240 align:middle line:84% You get an average replacement of UI, about 45%. 00:23:15.240 --> 00:23:17.500 align:middle line:84% The average replacement for DI is about 60%. 00:23:17.500 --> 00:23:20.760 align:middle line:84% The average under workers' comp is about 90%. 00:23:20.760 --> 00:23:23.240 align:middle line:90% And does it vary across states? 00:23:23.240 --> 00:23:24.640 align:middle line:90% Worker's comp and UI does. 00:23:24.640 --> 00:23:26.570 align:middle line:90% DI really doesn't. 00:23:26.570 --> 00:23:29.090 align:middle line:84% So we're going to come back to this table 00:23:29.090 --> 00:23:31.210 align:middle line:84% later in the lecture and talk about what we've 00:23:31.210 --> 00:23:33.210 align:middle line:84% learned about these programs makes us think 00:23:33.210 --> 00:23:35.770 align:middle line:90% about the way they're set up. 00:23:35.770 --> 00:23:37.650 align:middle line:84% Now, one thing that's interesting 00:23:37.650 --> 00:23:41.370 align:middle line:84% is, how do you think about the duration of social insurance 00:23:41.370 --> 00:23:42.450 align:middle line:90% benefits? 00:23:42.450 --> 00:23:50.850 align:middle line:84% So if we go to figure 14-2, this figure shows-- 00:23:50.850 --> 00:23:52.270 align:middle line:84% this kind of cool, colored figure, 00:23:52.270 --> 00:23:55.530 align:middle line:84% shows the level and duration of unemployment insurance benefits 00:23:55.530 --> 00:23:56.585 align:middle line:90% around the world. 00:23:56.585 --> 00:23:58.710 align:middle line:84% So for instance, the blue line's the United States. 00:23:58.710 --> 00:24:01.570 align:middle line:84% You get about a 44% replacement rate and it ends. 00:24:01.570 --> 00:24:03.890 align:middle line:90% After six months, it goes to 0. 00:24:03.890 --> 00:24:05.970 align:middle line:84% If you compare that to Austria, you 00:24:05.970 --> 00:24:09.850 align:middle line:84% get about a 55% replacement rate that never runs out. 00:24:09.850 --> 00:24:12.570 align:middle line:84% As long as you're unemployed, you get the benefits. 00:24:12.570 --> 00:24:14.430 align:middle line:84% Belgium, it goes down, but stays high. 00:24:14.430 --> 00:24:18.010 align:middle line:84% Other states, it lasts longer, then goes to 0. 00:24:18.010 --> 00:24:19.970 align:middle line:90% Well, how do you think about-- 00:24:19.970 --> 00:24:24.210 align:middle line:84% how does our chapter 12 help us think about the optimal duration 00:24:24.210 --> 00:24:26.990 align:middle line:90% moderation of UI benefits? 00:24:26.990 --> 00:24:27.610 align:middle line:90% OK. 00:24:27.610 --> 00:24:28.750 align:middle line:90% Well, let's start. 00:24:28.750 --> 00:24:31.250 align:middle line:84% So let's think about the principle of chapter 12, 00:24:31.250 --> 00:24:34.030 align:middle line:84% about optimal social insurance determination, 00:24:34.030 --> 00:24:35.690 align:middle line:90% and what forces should drive. 00:24:35.690 --> 00:24:37.650 align:middle line:84% Unemployment insurance benefits to be longer 00:24:37.650 --> 00:24:41.403 align:middle line:84% and what forces should drive them to be shorter. 00:24:41.403 --> 00:24:43.070 align:middle line:84% Or in other words, under what conditions 00:24:43.070 --> 00:24:44.550 align:middle line:84% do you like a system like Austria's, where 00:24:44.550 --> 00:24:45.430 align:middle line:90% you get it forever? 00:24:45.430 --> 00:24:46.850 align:middle line:84% Under what conditions do you like the system of the US, 00:24:46.850 --> 00:24:49.017 align:middle line:84% where it just cuts off after a short period of time? 00:24:49.017 --> 00:24:52.310 align:middle line:90% 00:24:52.310 --> 00:24:55.190 align:middle line:84% What are the trade-offs in thinking about that? 00:24:55.190 --> 00:24:57.110 align:middle line:90% Yeah? 00:24:57.110 --> 00:24:58.580 align:middle line:90% AUDIENCE: [INAUDIBLE]. 00:24:58.580 --> 00:25:02.382 align:middle line:90% 00:25:02.382 --> 00:25:03.090 align:middle line:90% JON GRUBER: Yeah. 00:25:03.090 --> 00:25:05.070 align:middle line:84% So basically the cost, obviously, 00:25:05.070 --> 00:25:06.830 align:middle line:84% is the longer you let these benefits go, 00:25:06.830 --> 00:25:08.910 align:middle line:84% the less incentive there is to look for work. 00:25:08.910 --> 00:25:10.710 align:middle line:84% In Austria, if you don't love your job 00:25:10.710 --> 00:25:13.002 align:middle line:84% and you're sitting at home forever with 60% replacement 00:25:13.002 --> 00:25:14.750 align:middle line:84% rate, if they can't really verify how 00:25:14.750 --> 00:25:16.710 align:middle line:90% hard you're looking for work-- 00:25:16.710 --> 00:25:18.870 align:middle line:84% so there's a huge moral hazard cost. 00:25:18.870 --> 00:25:21.790 align:middle line:84% That's a clear disadvantage to making 00:25:21.790 --> 00:25:23.260 align:middle line:90% the benefits last a long time. 00:25:23.260 --> 00:25:26.520 align:middle line:84% I would say go US, we cut them off shorter. 00:25:26.520 --> 00:25:29.600 align:middle line:84% What is the counterargument for letting them last longer? 00:25:29.600 --> 00:25:30.100 align:middle line:90% Yeah, 00:25:30.100 --> 00:25:31.760 align:middle line:84% AUDIENCE: Maybe the job market sucks? 00:25:31.760 --> 00:25:32.840 align:middle line:84% JON GRUBER: Well, one argument is maybe 00:25:32.840 --> 00:25:34.007 align:middle line:90% the job market really sucks. 00:25:34.007 --> 00:25:36.423 align:middle line:84% So that would say the problem is you're kicking people off 00:25:36.423 --> 00:25:38.320 align:middle line:84% before they can plausibly find a job. 00:25:38.320 --> 00:25:41.520 align:middle line:90% What else? 00:25:41.520 --> 00:25:42.900 align:middle line:90% Going back to chapter 12. 00:25:42.900 --> 00:25:46.360 align:middle line:90% 00:25:46.360 --> 00:25:48.800 align:middle line:84% Remember, what drives consumption smoothing 00:25:48.800 --> 00:25:50.100 align:middle line:90% for social insurance programs? 00:25:50.100 --> 00:25:52.600 align:middle line:90% What drives that? 00:25:52.600 --> 00:25:56.580 align:middle line:84% What drives how much consumption smoothing you can do? 00:25:56.580 --> 00:25:57.080 align:middle line:90% Yeah? 00:25:57.080 --> 00:25:58.360 align:middle line:90% AUDIENCE: How much [INAUDIBLE]. 00:25:58.360 --> 00:26:00.180 align:middle line:84% JON GRUBER: How much self-insurance you replace. 00:26:00.180 --> 00:26:01.360 align:middle line:84% Let's think about how self-insurance 00:26:01.360 --> 00:26:02.318 align:middle line:90% works for unemployment. 00:26:02.318 --> 00:26:04.193 align:middle line:84% When you first lose your job, you're probably 00:26:04.193 --> 00:26:05.340 align:middle line:90% pretty well self-insured. 00:26:05.340 --> 00:26:06.300 align:middle line:90% You got some savings. 00:26:06.300 --> 00:26:06.960 align:middle line:90% You got a credit card. 00:26:06.960 --> 00:26:07.900 align:middle line:90% You can borrow up to the limit. 00:26:07.900 --> 00:26:09.817 align:middle line:84% Some friends will help you out, but the longer 00:26:09.817 --> 00:26:11.560 align:middle line:84% you're unemployed, the worse and worse 00:26:11.560 --> 00:26:14.080 align:middle line:84% your self-insurance is going to be. 00:26:14.080 --> 00:26:18.120 align:middle line:84% So in some sense, that UI does more consumption 00:26:18.120 --> 00:26:21.740 align:middle line:84% smoothing the longer you're out of work, which would say benefit 00:26:21.740 --> 00:26:25.020 align:middle line:84% should, if anything, rise over time. 00:26:25.020 --> 00:26:27.073 align:middle line:84% So the trade-off is on the one hand, 00:26:27.073 --> 00:26:29.240 align:middle line:84% the longer they're over time, the more moral hazard. 00:26:29.240 --> 00:26:30.657 align:middle line:84% On the other hand, the longer they 00:26:30.657 --> 00:26:33.420 align:middle line:84% last, the more consumption smoothing. 00:26:33.420 --> 00:26:36.500 align:middle line:84% And you have Stephen's point about the fact 00:26:36.500 --> 00:26:38.900 align:middle line:84% that if it's hard to find a new job, 00:26:38.900 --> 00:26:40.820 align:middle line:84% you might want them to be longer. 00:26:40.820 --> 00:26:42.283 align:middle line:90% OK? 00:26:42.283 --> 00:26:43.700 align:middle line:84% But then you have one last-- yeah? 00:26:43.700 --> 00:26:47.780 align:middle line:84% AUDIENCE: Or you could think about it as [INAUDIBLE]. 00:26:47.780 --> 00:26:49.740 align:middle line:84% Is it from the end of that [INAUDIBLE]? 00:26:49.740 --> 00:26:50.500 align:middle line:90% JON GRUBER: No. 00:26:50.500 --> 00:26:52.380 align:middle line:90% No, from the start. 00:26:52.380 --> 00:26:54.780 align:middle line:84% But then there's one last factor, 00:26:54.780 --> 00:26:59.085 align:middle line:84% which is that the fact-- that's another version of moral hazard, 00:26:59.085 --> 00:27:01.460 align:middle line:84% which is the longer you make it, the more people's skills 00:27:01.460 --> 00:27:03.940 align:middle line:84% deteriorate, and the harder it is to find a new job. 00:27:03.940 --> 00:27:05.080 align:middle line:84% So in other words, your moral hazard, 00:27:05.080 --> 00:27:06.280 align:middle line:84% not only keeping them from working. 00:27:06.280 --> 00:27:08.220 align:middle line:84% You're making it harder to find a new job. 00:27:08.220 --> 00:27:10.480 align:middle line:84% So the bottom line is it's a classic on the one hand, 00:27:10.480 --> 00:27:11.200 align:middle line:90% on the other hand thing. 00:27:11.200 --> 00:27:12.240 align:middle line:90% There's no right answer. 00:27:12.240 --> 00:27:14.720 align:middle line:84% But it's a great-- in one graph, in one example, 00:27:14.720 --> 00:27:17.740 align:middle line:84% it's a great way to understand the trade-offs of chapter 12. 00:27:17.740 --> 00:27:18.620 align:middle line:90% OK? 00:27:18.620 --> 00:27:21.510 align:middle line:90% Questions about that? 00:27:21.510 --> 00:27:26.232 align:middle line:90% Now, let's talk about benefits. 00:27:26.232 --> 00:27:27.190 align:middle line:90% So that's the duration. 00:27:27.190 --> 00:27:28.330 align:middle line:84% We're not talking about duration anymore. 00:27:28.330 --> 00:27:30.090 align:middle line:90% Let's talk about benefits level. 00:27:30.090 --> 00:27:31.670 align:middle line:84% Let's focus on the replacement rate 00:27:31.670 --> 00:27:34.830 align:middle line:84% as the key determinant of generosity of social insurance 00:27:34.830 --> 00:27:35.830 align:middle line:90% programs. 00:27:35.830 --> 00:27:39.070 align:middle line:84% And let's ask, how do we think about the optimal replacement 00:27:39.070 --> 00:27:39.710 align:middle line:90% rate? 00:27:39.710 --> 00:27:42.085 align:middle line:84% Well, the way we think about the optimal replacement rate 00:27:42.085 --> 00:27:44.590 align:middle line:84% is to trade off the consumption smoothing benefits 00:27:44.590 --> 00:27:47.430 align:middle line:90% against the moral hazard costs. 00:27:47.430 --> 00:27:49.630 align:middle line:84% And what's awesome about chapter 14 00:27:49.630 --> 00:27:52.490 align:middle line:84% is we have a large literature on both. 00:27:52.490 --> 00:27:55.030 align:middle line:84% So we can actually put them together. 00:27:55.030 --> 00:27:55.990 align:middle line:90% So let's start. 00:27:55.990 --> 00:27:57.990 align:middle line:84% There's a much smaller literature on consumption 00:27:57.990 --> 00:28:00.270 align:middle line:90% smoothing benefits. 00:28:00.270 --> 00:28:04.750 align:middle line:84% The way that literature works is essentially asking literally 00:28:04.750 --> 00:28:06.430 align:middle line:84% if benefits get more generous, are 00:28:06.430 --> 00:28:08.023 align:middle line:90% people's consumptions smoother? 00:28:08.023 --> 00:28:10.190 align:middle line:84% So I did the first article on this with unemployment 00:28:10.190 --> 00:28:13.530 align:middle line:84% insurance, where I asked if UI benefits are more generous, 00:28:13.530 --> 00:28:16.070 align:middle line:84% are people consumption more smooth when they lose their job? 00:28:16.070 --> 00:28:20.040 align:middle line:84% To the extent-- so think of a regression-- 00:28:20.040 --> 00:28:21.520 align:middle line:84% of the change in your consumption 00:28:21.520 --> 00:28:24.640 align:middle line:84% when you lose your job on how generous benefits are. 00:28:24.640 --> 00:28:25.820 align:middle line:90% Think of that regression. 00:28:25.820 --> 00:28:32.120 align:middle line:84% So delta C equals alpha plus beta times benefit. 00:28:32.120 --> 00:28:32.720 align:middle line:90% OK. 00:28:32.720 --> 00:28:35.720 align:middle line:84% If someone has 100% self-insurance, 00:28:35.720 --> 00:28:39.040 align:middle line:84% what should the coefficient beta be? 00:28:39.040 --> 00:28:40.540 align:middle line:84% Let's get some other folks involved. 00:28:40.540 --> 00:28:41.165 align:middle line:90% Come on, folks. 00:28:41.165 --> 00:28:42.720 align:middle line:90% Let's wake up. 00:28:42.720 --> 00:28:44.400 align:middle line:90% You got this. 00:28:44.400 --> 00:28:47.013 align:middle line:84% If someone has 100% self-insurance, 00:28:47.013 --> 00:28:48.680 align:middle line:84% what should the-- how should the benefit 00:28:48.680 --> 00:28:50.055 align:middle line:84% affect your change in consumption 00:28:50.055 --> 00:28:52.480 align:middle line:90% when you lose your job? 00:28:52.480 --> 00:28:53.265 align:middle line:90% Yeah? 00:28:53.265 --> 00:28:53.890 align:middle line:90% AUDIENCE: Zero. 00:28:53.890 --> 00:28:54.598 align:middle line:90% JON GRUBER: Zero. 00:28:54.598 --> 00:28:56.100 align:middle line:84% Because you're already self-insured. 00:28:56.100 --> 00:28:57.500 align:middle line:84% Your consumption is already going to be smooth. 00:28:57.500 --> 00:28:58.360 align:middle line:90% It shouldn't matter. 00:28:58.360 --> 00:29:01.120 align:middle line:84% If likewise you have zero self-insurance, 00:29:01.120 --> 00:29:02.823 align:middle line:90% that coefficient should be 1. 00:29:02.823 --> 00:29:05.240 align:middle line:84% So it says that the regression from that-- the coefficient 00:29:05.240 --> 00:29:07.310 align:middle line:84% from that regression tells you both how much 00:29:07.310 --> 00:29:09.060 align:middle line:84% consumption smoothing the program is doing 00:29:09.060 --> 00:29:10.760 align:middle line:90% and how much crowded it's doing. 00:29:10.760 --> 00:29:12.160 align:middle line:84% And the answer through regression 00:29:12.160 --> 00:29:19.540 align:middle line:84% is beta is around 0.3, that basically people are pretty well 00:29:19.540 --> 00:29:22.940 align:middle line:84% self-insured, not fully self-insured. 00:29:22.940 --> 00:29:26.380 align:middle line:84% Beta is not 0-- but reasonably well self-insured. 00:29:26.380 --> 00:29:27.580 align:middle line:90% And actually, I've then-- 00:29:27.580 --> 00:29:29.380 align:middle line:84% I and others have then gone to a series of articles 00:29:29.380 --> 00:29:31.280 align:middle line:84% showing whether self-insurance comes from. 00:29:31.280 --> 00:29:35.380 align:middle line:84% It comes from people save more when UI is less generous. 00:29:35.380 --> 00:29:39.180 align:middle line:84% And their spouses go to work more when UI is less generous. 00:29:39.180 --> 00:29:43.060 align:middle line:84% And as a result, there's a decent amount of self-insurance. 00:29:43.060 --> 00:29:44.060 align:middle line:90% OK? 00:29:44.060 --> 00:29:48.340 align:middle line:84% So there's some consumption smoothing, not a ton. 00:29:48.340 --> 00:29:57.700 align:middle line:84% Now, this sort of analysis has been done for other programs, 00:29:57.700 --> 00:29:59.780 align:middle line:90% as well. 00:29:59.780 --> 00:30:01.400 align:middle line:84% If you look at disability insurance, 00:30:01.400 --> 00:30:03.820 align:middle line:84% there's a really neat study from the Netherlands 00:30:03.820 --> 00:30:05.820 align:middle line:90% by my colleague David Autor. 00:30:05.820 --> 00:30:10.420 align:middle line:84% And what David Autor finds is that basically-- 00:30:10.420 --> 00:30:12.840 align:middle line:84% that basically beta is much bigger. 00:30:12.840 --> 00:30:19.430 align:middle line:84% So this is-- so beta UI is smaller than beta DI. 00:30:19.430 --> 00:30:22.110 align:middle line:84% The disability insurance does a lot more consumption 00:30:22.110 --> 00:30:23.770 align:middle line:84% smoothing and a lot less crowd-out, 00:30:23.770 --> 00:30:25.290 align:middle line:90% which is not surprising. 00:30:25.290 --> 00:30:26.790 align:middle line:90% Why? 00:30:26.790 --> 00:30:28.362 align:middle line:84% Why is it not surprising that there 00:30:28.362 --> 00:30:30.070 align:middle line:84% should be more consumption smoothing done 00:30:30.070 --> 00:30:33.630 align:middle line:90% by DI than there is by UI? 00:30:33.630 --> 00:30:34.670 align:middle line:90% Yeah? 00:30:34.670 --> 00:30:36.750 align:middle line:84% AUDIENCE: DI is a longer period of time. 00:30:36.750 --> 00:30:37.833 align:middle line:90% JON GRUBER: Longer period? 00:30:37.833 --> 00:30:38.750 align:middle line:90% It's a more serious-- 00:30:38.750 --> 00:30:40.070 align:middle line:84% you can't self-insure against losing 00:30:40.070 --> 00:30:41.330 align:middle line:90% your job the rest of your life. 00:30:41.330 --> 00:30:42.690 align:middle line:84% It's pretty hard to self-insure against. 00:30:42.690 --> 00:30:44.990 align:middle line:84% You can self-insure on being against being out of work 00:30:44.990 --> 00:30:49.210 align:middle line:84% for 10 weeks, so it's not surprising. 00:30:49.210 --> 00:30:54.390 align:middle line:84% Now, the one we don't know is beta workers' comp. 00:30:54.390 --> 00:30:57.390 align:middle line:84% This has never really been estimated. 00:30:57.390 --> 00:30:58.670 align:middle line:90% OK? 00:30:58.670 --> 00:31:04.070 align:middle line:84% My guess is it would kind be in the middle here in the sense 00:31:04.070 --> 00:31:08.590 align:middle line:84% that on the one hand, getting injured 00:31:08.590 --> 00:31:14.130 align:middle line:84% is probably less anticipatable than losing your job. 00:31:14.130 --> 00:31:18.690 align:middle line:84% So that says people are probably less well self-insured. 00:31:18.690 --> 00:31:21.575 align:middle line:84% On the other hand, it's not as permanent as being disabled. 00:31:21.575 --> 00:31:23.450 align:middle line:84% So people probably have at least some savings 00:31:23.450 --> 00:31:25.410 align:middle line:84% they could use to help themselves out. 00:31:25.410 --> 00:31:27.270 align:middle line:84% So we don't know where beta WC is, 00:31:27.270 --> 00:31:29.530 align:middle line:84% but it's probably in the middle there. 00:31:29.530 --> 00:31:30.210 align:middle line:90% OK. 00:31:30.210 --> 00:31:33.210 align:middle line:84% This is a way to put numbers on the intuition 00:31:33.210 --> 00:31:36.050 align:middle line:84% from last time, which is this fundamental trade off 00:31:36.050 --> 00:31:39.770 align:middle line:84% between self-insurance and consumption smoothing. 00:31:39.770 --> 00:31:40.770 align:middle line:90% OK. 00:31:40.770 --> 00:31:41.950 align:middle line:90% Questions about that? 00:31:41.950 --> 00:31:44.690 align:middle line:90% 00:31:44.690 --> 00:31:48.370 align:middle line:84% Now there's a relatively small literature on that. 00:31:48.370 --> 00:31:50.490 align:middle line:84% Since economists are grumpy, there's 00:31:50.490 --> 00:31:52.810 align:middle line:84% a much larger literature on moral hazard. 00:31:52.810 --> 00:31:54.430 align:middle line:90% That's more fun to talk about. 00:31:54.430 --> 00:31:55.970 align:middle line:84% It's not that fun to talk about whether smoothing consump-- 00:31:55.970 --> 00:31:57.170 align:middle line:84% It's really fun to talk about people 00:31:57.170 --> 00:31:59.130 align:middle line:84% perniciously not going to work because they're 00:31:59.130 --> 00:32:00.530 align:middle line:90% getting benefits. 00:32:00.530 --> 00:32:04.570 align:middle line:84% So there's a massive literature on the moral hazard 00:32:04.570 --> 00:32:08.668 align:middle line:84% effect of workers' comp, of UI, massive literature 00:32:08.668 --> 00:32:10.710 align:middle line:84% on the moral hazard effect of all these programs. 00:32:10.710 --> 00:32:12.150 align:middle line:84% So I want to go through them one by one. 00:32:12.150 --> 00:32:13.760 align:middle line:84% The consensus in the literature is pretty small. 00:32:13.760 --> 00:32:15.180 align:middle line:90% That's all I have to say. 00:32:15.180 --> 00:32:17.200 align:middle line:84% The moral hazard literature is very large, 00:32:17.200 --> 00:32:19.860 align:middle line:84% so I want to go through each program one by one. 00:32:19.860 --> 00:32:24.540 align:middle line:84% And the moral hazard literature is large 00:32:24.540 --> 00:32:26.580 align:middle line:84% because, for example, for UI, you 00:32:26.580 --> 00:32:30.180 align:middle line:84% have all these wonderful state quasi-experiments to study. 00:32:30.180 --> 00:32:34.660 align:middle line:84% So a lot of studies of the form figure 14-3. 00:32:34.660 --> 00:32:37.580 align:middle line:90% 00:32:37.580 --> 00:32:40.780 align:middle line:84% Figure 14-3 is an example of how you set up a quasi-experimental 00:32:40.780 --> 00:32:42.360 align:middle line:90% study of unemployment insurance. 00:32:42.360 --> 00:32:43.820 align:middle line:84% Let's describe what we have here. 00:32:43.820 --> 00:32:47.380 align:middle line:84% We have a state that used to have the blue replacement 00:32:47.380 --> 00:32:52.260 align:middle line:84% schedule and then changes it, raises its benefits, 00:32:52.260 --> 00:32:55.340 align:middle line:84% and adds the reds, and replace the upper segment with the red. 00:32:55.340 --> 00:32:57.700 align:middle line:84% So their new budget, their new state benefit schedule 00:32:57.700 --> 00:32:59.320 align:middle line:90% is the same up to $700. 00:32:59.320 --> 00:33:01.660 align:middle line:90% But then it's higher after $700. 00:33:01.660 --> 00:33:03.540 align:middle line:84% So they move from all-blue schedule 00:33:03.540 --> 00:33:05.120 align:middle line:90% to the blue plus red schedule. 00:33:05.120 --> 00:33:07.900 align:middle line:90% 00:33:07.900 --> 00:33:10.020 align:middle line:84% The way quasi-experiment analysis works 00:33:10.020 --> 00:33:13.150 align:middle line:90% of that is, they say, OK, look. 00:33:13.150 --> 00:33:17.750 align:middle line:84% You've got two groups, those who earn below 700. 00:33:17.750 --> 00:33:20.310 align:middle line:90% Nothing changed for them. 00:33:20.310 --> 00:33:24.790 align:middle line:84% Those who are in above 800, something changed for them. 00:33:24.790 --> 00:33:28.150 align:middle line:84% So let's look and ask, over time, 00:33:28.150 --> 00:33:30.790 align:middle line:84% do you see a different behavior in those 00:33:30.790 --> 00:33:34.150 align:middle line:84% who earn above $800 versus those who earn below $700? 00:33:34.150 --> 00:33:35.770 align:middle line:90% The difference in difference. 00:33:35.770 --> 00:33:37.167 align:middle line:90% We discussed this. 00:33:37.167 --> 00:33:38.250 align:middle line:90% We discussed in chapter 3. 00:33:38.250 --> 00:33:40.950 align:middle line:84% Go and remind yourself, and look at the appendix to chapter 14 00:33:40.950 --> 00:33:42.750 align:middle line:84% to remind yourself how this all works. 00:33:42.750 --> 00:33:45.350 align:middle line:84% But essentially, we do a difference in difference. 00:33:45.350 --> 00:33:47.150 align:middle line:84% The difference in difference here 00:33:47.150 --> 00:33:53.390 align:middle line:84% is group-- we basically look at group H 00:33:53.390 --> 00:33:59.230 align:middle line:84% after minus H before is one difference minus L 00:33:59.230 --> 00:34:01.510 align:middle line:90% after minus L before. 00:34:01.510 --> 00:34:03.630 align:middle line:84% And let's talk what's going on here. 00:34:03.630 --> 00:34:04.608 align:middle line:90% You could just do this. 00:34:04.608 --> 00:34:06.650 align:middle line:84% You could say, well, look, the benefit increased. 00:34:06.650 --> 00:34:08.909 align:middle line:84% Let's look at what happened to group H. 00:34:08.909 --> 00:34:11.250 align:middle line:84% Did their unemployment durations go up? 00:34:11.250 --> 00:34:13.449 align:middle line:90% What would be wrong with that? 00:34:13.449 --> 00:34:13.969 align:middle line:90% Yeah? 00:34:13.969 --> 00:34:16.548 align:middle line:90% AUDIENCE: [INAUDIBLE]. 00:34:16.548 --> 00:34:17.590 align:middle line:90% JON GRUBER: Well, no, no. 00:34:17.590 --> 00:34:20.310 align:middle line:84% Because you're doing it based on their-- 00:34:20.310 --> 00:34:22.889 align:middle line:84% AUDIENCE: Well, you do not have their control group. 00:34:22.889 --> 00:34:25.170 align:middle line:84% JON GRUBER: But why do you need a control group? 00:34:25.170 --> 00:34:27.170 align:middle line:84% AUDIENCE: See if there were other effects there. 00:34:27.170 --> 00:34:28.650 align:middle line:84% JON GRUBER: What other effects are you particularly worried 00:34:28.650 --> 00:34:29.150 align:middle line:90% about? 00:34:29.150 --> 00:34:30.530 align:middle line:90% AUDIENCE: I don't know. 00:34:30.530 --> 00:34:31.570 align:middle line:84% JON GRUBER: Maybe they raise the benefits 00:34:31.570 --> 00:34:32.870 align:middle line:90% because it was a bad time. 00:34:32.870 --> 00:34:35.159 align:middle line:84% Maybe states rate benefits when there's a recession. 00:34:35.159 --> 00:34:37.409 align:middle line:84% So of course you're going to find higher benefits lead 00:34:37.409 --> 00:34:38.650 align:middle line:84% to longer-- people being unemployed more, 00:34:38.650 --> 00:34:40.929 align:middle line:84% because they raised benefits during a recession. 00:34:40.929 --> 00:34:45.130 align:middle line:84% So the idea is that this-- think of HA minus HB 00:34:45.130 --> 00:34:50.409 align:middle line:84% as being the effect of UI, the UI moral hazard, 00:34:50.409 --> 00:34:53.670 align:middle line:84% plus the effect of economic conditions. 00:34:53.670 --> 00:34:56.290 align:middle line:90% 00:34:56.290 --> 00:34:59.590 align:middle line:84% But that's where this group comes in, which, say, well, 00:34:59.590 --> 00:35:02.310 align:middle line:84% the economic conditions also effect anyone below $700. 00:35:02.310 --> 00:35:05.330 align:middle line:84% So here you have the effect of just econ conditions. 00:35:05.330 --> 00:35:08.780 align:middle line:84% So by taking the difference in difference, 00:35:08.780 --> 00:35:14.580 align:middle line:84% you end up with just the UI moral hazard effect. 00:35:14.580 --> 00:35:16.680 align:middle line:84% And that's the idea of this approach. 00:35:16.680 --> 00:35:19.300 align:middle line:90% 00:35:19.300 --> 00:35:19.900 align:middle line:90% Yeah? 00:35:19.900 --> 00:35:21.020 align:middle line:90% Question? 00:35:21.020 --> 00:35:22.500 align:middle line:90% You look puzzled. 00:35:22.500 --> 00:35:25.100 align:middle line:84% AUDIENCE: Yeah, for the DID approach, 00:35:25.100 --> 00:35:27.007 align:middle line:84% I guess one of the main threats to validity 00:35:27.007 --> 00:35:29.340 align:middle line:84% would be the comparison between the control [INAUDIBLE]. 00:35:29.340 --> 00:35:31.040 align:middle line:84% JON GRUBER: So let's put this in English. 00:35:31.040 --> 00:35:33.165 align:middle line:84% You clearly know how to think about this framework. 00:35:33.165 --> 00:35:34.700 align:middle line:90% Let's put this in English. 00:35:34.700 --> 00:35:37.660 align:middle line:84% What's wrong with-- how would you criticize 00:35:37.660 --> 00:35:38.920 align:middle line:90% what I just laid out here? 00:35:38.920 --> 00:35:41.110 align:middle line:84% I criticized just using this comparison. 00:35:41.110 --> 00:35:42.860 align:middle line:84% How would you criticize what I-- go ahead. 00:35:42.860 --> 00:35:43.760 align:middle line:84% See if you can put it in English. 00:35:43.760 --> 00:35:45.500 align:middle line:84% How would you criticize what I just laid out here? 00:35:45.500 --> 00:35:48.460 align:middle line:84% AUDIENCE: If you have a control [INAUDIBLE] not the same people. 00:35:48.460 --> 00:35:50.127 align:middle line:84% JON GRUBER: They're not the same people. 00:35:50.127 --> 00:35:53.188 align:middle line:84% But that's OK as long as they're different for random reasons. 00:35:53.188 --> 00:35:55.480 align:middle line:84% But why might they not be different for random reasons? 00:35:55.480 --> 00:35:57.940 align:middle line:84% AUDIENCE: Well, because these are two different states. 00:35:57.940 --> 00:35:59.420 align:middle line:90% JON GRUBER: No, it's one state. 00:35:59.420 --> 00:36:00.460 align:middle line:90% One state. 00:36:00.460 --> 00:36:01.080 align:middle line:90% Yeah? 00:36:01.080 --> 00:36:02.420 align:middle line:84% AUDIENCE: Because the income is different. 00:36:02.420 --> 00:36:03.200 align:middle line:84% JON GRUBER: Their income is different. 00:36:03.200 --> 00:36:04.242 align:middle line:90% And why does that matter? 00:36:04.242 --> 00:36:06.080 align:middle line:84% They're different before income beforehand. 00:36:06.080 --> 00:36:08.240 align:middle line:84% AUDIENCE: It reflects demographic differences 00:36:08.240 --> 00:36:08.940 align:middle line:90% between the two. 00:36:08.940 --> 00:36:09.857 align:middle line:90% JON GRUBER: That's OK. 00:36:09.857 --> 00:36:10.460 align:middle line:90% That's OK. 00:36:10.460 --> 00:36:11.780 align:middle line:84% If these groups are different, that's OK. 00:36:11.780 --> 00:36:12.420 align:middle line:90% You know why? 00:36:12.420 --> 00:36:14.003 align:middle line:84% Because I'm comparing before to after. 00:36:14.003 --> 00:36:16.087 align:middle line:84% So it doesn't matter these guys are in these guys. 00:36:16.087 --> 00:36:18.720 align:middle line:84% So I'm looking at what happens over time within the group. 00:36:18.720 --> 00:36:21.120 align:middle line:84% So any fixed difference between groups 00:36:21.120 --> 00:36:23.640 align:middle line:90% is picked up by this, by this. 00:36:23.640 --> 00:36:24.200 align:middle line:90% OK? 00:36:24.200 --> 00:36:26.680 align:middle line:84% The fact that these guys are in these guys doesn't matter. 00:36:26.680 --> 00:36:28.597 align:middle line:84% Because I'm looking at what happened over time 00:36:28.597 --> 00:36:30.000 align:middle line:90% within each group. 00:36:30.000 --> 00:36:30.940 align:middle line:90% Yeah? 00:36:30.940 --> 00:36:33.827 align:middle line:90% AUDIENCE: [INAUDIBLE]. 00:36:33.827 --> 00:36:34.660 align:middle line:90% JON GRUBER: Exactly. 00:36:34.660 --> 00:36:35.702 align:middle line:90% The groups are different. 00:36:35.702 --> 00:36:36.620 align:middle line:90% That's not a problem. 00:36:36.620 --> 00:36:40.360 align:middle line:84% If the effect of the recession is different, that is a problem. 00:36:40.360 --> 00:36:42.720 align:middle line:84% What if recessions affect low-income workers differently 00:36:42.720 --> 00:36:44.360 align:middle line:90% than high-income workers? 00:36:44.360 --> 00:36:45.040 align:middle line:90% OK? 00:36:45.040 --> 00:36:47.385 align:middle line:84% So basically I should have broken this down in piece. 00:36:47.385 --> 00:36:48.760 align:middle line:84% Let's say your first estimate was 00:36:48.760 --> 00:36:51.480 align:middle line:90% just I'm going to look at HB. 00:36:51.480 --> 00:36:54.200 align:middle line:84% Well, that's bad because the recession could have happened. 00:36:54.200 --> 00:36:57.640 align:middle line:84% Then you say, well, look at HA minus HB. 00:36:57.640 --> 00:36:59.597 align:middle line:90% Well, that's bad because-- 00:36:59.597 --> 00:37:00.680 align:middle line:90% I'm sorry, you look at HB. 00:37:00.680 --> 00:37:02.120 align:middle line:84% You say, well, that's bad because people just 00:37:02.120 --> 00:37:03.060 align:middle line:90% are different. 00:37:03.060 --> 00:37:06.710 align:middle line:84% HA minus B gets you this, then you difference out from this. 00:37:06.710 --> 00:37:08.130 align:middle line:84% This, by differencing, you get rid 00:37:08.130 --> 00:37:10.890 align:middle line:84% of any distinguishing differences within the group. 00:37:10.890 --> 00:37:12.370 align:middle line:84% But we don't get rid of is the fact 00:37:12.370 --> 00:37:15.690 align:middle line:84% that economic conditions-- what if this is econ H and this 00:37:15.690 --> 00:37:17.570 align:middle line:90% is econ L? 00:37:17.570 --> 00:37:20.090 align:middle line:84% What if the recession affects them differently? 00:37:20.090 --> 00:37:21.350 align:middle line:90% Then you're not getting UIMH. 00:37:21.350 --> 00:37:26.770 align:middle line:84% You're getting UIMH plus another term, which is econ H minus econ 00:37:26.770 --> 00:37:30.530 align:middle line:90% L. What do we call that term? 00:37:30.530 --> 00:37:32.290 align:middle line:90% Bias. 00:37:32.290 --> 00:37:37.070 align:middle line:84% So the bias only goes away if econ H equals econ L. Well, 00:37:37.070 --> 00:37:40.403 align:middle line:90% how would you figure that out? 00:37:40.403 --> 00:37:42.070 align:middle line:84% What would you do to solve that problem? 00:37:42.070 --> 00:37:43.690 align:middle line:90% Are you just stuck? 00:37:43.690 --> 00:37:45.250 align:middle line:90% No, you're not stuck. 00:37:45.250 --> 00:37:47.010 align:middle line:90% What do you do? 00:37:47.010 --> 00:37:49.490 align:middle line:84% Someone actually already gave the answer. 00:37:49.490 --> 00:37:50.210 align:middle line:90% Yeah? 00:37:50.210 --> 00:37:51.610 align:middle line:84% AUDIENCE: You have to make sure that they're randomly 00:37:51.610 --> 00:37:52.610 align:middle line:90% distributed. 00:37:52.610 --> 00:37:54.330 align:middle line:84% JON GRUBER: Well, no, we know they're not randomly distributed 00:37:54.330 --> 00:37:55.450 align:middle line:90% because they start with-- 00:37:55.450 --> 00:37:57.130 align:middle line:84% I mean, people are bottom below $700 00:37:57.130 --> 00:38:00.850 align:middle line:84% or above $800 are not randomly distributed. 00:38:00.850 --> 00:38:02.590 align:middle line:84% What has to be randomly distributed? 00:38:02.590 --> 00:38:05.420 align:middle line:90% What has to be random? 00:38:05.420 --> 00:38:08.100 align:middle line:84% Or what-- how would you-- actually, 00:38:08.100 --> 00:38:10.820 align:middle line:84% let's just take a step back and ask-- 00:38:10.820 --> 00:38:12.580 align:middle line:84% think about what I'm assuming here. 00:38:12.580 --> 00:38:15.420 align:middle line:84% I'm assuming that the effect of economic conditions on these two 00:38:15.420 --> 00:38:18.220 align:middle line:84% groups in this state are the same. 00:38:18.220 --> 00:38:19.440 align:middle line:90% How might I test that? 00:38:19.440 --> 00:38:22.020 align:middle line:90% 00:38:22.020 --> 00:38:23.020 align:middle line:90% Yeah? 00:38:23.020 --> 00:38:24.540 align:middle line:84% AUDIENCE: If you look historically. 00:38:24.540 --> 00:38:26.123 align:middle line:84% JON GRUBER: I could look historically? 00:38:26.123 --> 00:38:28.340 align:middle line:84% I could ask the last time there's a recession, 00:38:28.340 --> 00:38:30.380 align:middle line:90% was econ H and econ L the same? 00:38:30.380 --> 00:38:33.420 align:middle line:84% So I could literally try to control for this term. 00:38:33.420 --> 00:38:35.800 align:middle line:84% I could literally put this in the regression. 00:38:35.800 --> 00:38:36.800 align:middle line:90% That's one way to do it. 00:38:36.800 --> 00:38:39.525 align:middle line:90% What else could I do? 00:38:39.525 --> 00:38:41.400 align:middle line:84% Once again, somebody already gave the answer. 00:38:41.400 --> 00:38:43.150 align:middle line:84% I, a minute ago, actually gave the answer. 00:38:43.150 --> 00:38:44.960 align:middle line:90% I snuck it in. 00:38:44.960 --> 00:38:45.460 align:middle line:90% Yeah? 00:38:45.460 --> 00:38:47.335 align:middle line:84% AUDIENCE: Think of that as a different state. 00:38:47.335 --> 00:38:48.780 align:middle line:90% JON GRUBER: A different state. 00:38:48.780 --> 00:38:52.300 align:middle line:84% Look at the neighboring state, where any recession probably 00:38:52.300 --> 00:38:55.420 align:middle line:90% hits them about the same. 00:38:55.420 --> 00:38:58.180 align:middle line:84% But so therefore, this term should be the same 00:38:58.180 --> 00:38:59.900 align:middle line:90% in the neighboring state. 00:38:59.900 --> 00:39:02.612 align:middle line:84% So if you can test your steer assumption 00:39:02.612 --> 00:39:04.820 align:middle line:84% by doing the same regression for a neighboring state, 00:39:04.820 --> 00:39:06.480 align:middle line:90% you should get 0. 00:39:06.480 --> 00:39:08.940 align:middle line:84% If you get 0, then your assumption is correct. 00:39:08.940 --> 00:39:11.098 align:middle line:84% If you don't, then it's problematic. 00:39:11.098 --> 00:39:13.140 align:middle line:84% Typically what we'll do in that situation is say, 00:39:13.140 --> 00:39:16.180 align:middle line:84% well, let's difference out what happens to the other state. 00:39:16.180 --> 00:39:17.930 align:middle line:84% That's called the difference in difference 00:39:17.930 --> 00:39:18.980 align:middle line:90% in difference estimator. 00:39:18.980 --> 00:39:21.480 align:middle line:84% Now, that may not be good enough because the other state may 00:39:21.480 --> 00:39:22.772 align:middle line:90% be different for other reasons. 00:39:22.772 --> 00:39:24.438 align:middle line:84% So then there's other things you can do. 00:39:24.438 --> 00:39:26.240 align:middle line:84% And this has given birth to an entire range 00:39:26.240 --> 00:39:28.420 align:middle line:84% of specification check technologies 00:39:28.420 --> 00:39:31.800 align:middle line:84% we do in economics, which we don't have time to get into. 00:39:31.800 --> 00:39:34.360 align:middle line:84% But this is the basic idea, is the basic idea 00:39:34.360 --> 00:39:36.120 align:middle line:84% of quasi-experimental economics, it's 00:39:36.120 --> 00:39:39.480 align:middle line:84% a game of whack-a-mole on identification, 00:39:39.480 --> 00:39:41.672 align:middle line:84% essentially saying, here's my strategy. 00:39:41.672 --> 00:39:43.880 align:middle line:84% Now I've got to whack down all the potential problems 00:39:43.880 --> 00:39:45.200 align:middle line:90% with that strategy. 00:39:45.200 --> 00:39:48.040 align:middle line:84% With a randomized controlled trial, you don't. 00:39:48.040 --> 00:39:49.020 align:middle line:90% There's no problem. 00:39:49.020 --> 00:39:50.600 align:middle line:90% You've randomized by definition. 00:39:50.600 --> 00:39:54.200 align:middle line:84% That's why randomized controlled trials are very much in favor. 00:39:54.200 --> 00:39:56.720 align:middle line:84% Because the problem with whack-a-mole is it's 00:39:56.720 --> 00:39:58.120 align:middle line:90% really hard to prove. 00:39:58.120 --> 00:40:00.620 align:middle line:84% All you can do is disprove it a million ways. 00:40:00.620 --> 00:40:02.870 align:middle line:84% And our poor students back there, Ahmet and Valerie, 00:40:02.870 --> 00:40:05.850 align:middle line:84% are writing papers where they're going to have to run 8,000 00:40:05.850 --> 00:40:07.902 align:middle line:84% of these graphs and tests and show that-- 00:40:07.902 --> 00:40:10.110 align:middle line:84% because basically, what happens in economic seminars? 00:40:10.110 --> 00:40:12.635 align:middle line:84% You go there, and it's a bunch of people finding moles. 00:40:12.635 --> 00:40:13.510 align:middle line:90% What about that mole? 00:40:13.510 --> 00:40:14.070 align:middle line:90% What about that mole? 00:40:14.070 --> 00:40:16.153 align:middle line:84% You got to whack it, whack it, whack it, whack it. 00:40:16.153 --> 00:40:17.130 align:middle line:90% OK? 00:40:17.130 --> 00:40:18.920 align:middle line:84% Ultimately, the believable the paper 00:40:18.920 --> 00:40:21.170 align:middle line:84% is how believable you could be at whacking those moles 00:40:21.170 --> 00:40:23.490 align:middle line:84% and how much people really believe you've gotten them. 00:40:23.490 --> 00:40:26.090 align:middle line:84% But that is-- just to be clear, while this course really 00:40:26.090 --> 00:40:29.450 align:middle line:84% advocates quasi-experimental methods as a solution 00:40:29.450 --> 00:40:32.330 align:middle line:84% to understand a lot of empirical problems, it's not perfect. 00:40:32.330 --> 00:40:32.830 align:middle line:90% OK? 00:40:32.830 --> 00:40:34.872 align:middle line:84% It's only perfect is to the extent you can really 00:40:34.872 --> 00:40:36.930 align:middle line:84% feel confident you've whacked all the moles. 00:40:36.930 --> 00:40:37.650 align:middle line:90% OK? 00:40:37.650 --> 00:40:40.370 align:middle line:90% Questions about that? 00:40:40.370 --> 00:40:41.250 align:middle line:90% All right. 00:40:41.250 --> 00:40:44.090 align:middle line:90% Now, so what do we find? 00:40:44.090 --> 00:40:47.810 align:middle line:84% When we do that for UI, we find a reasonably large moral hazard 00:40:47.810 --> 00:40:48.310 align:middle line:90% effect. 00:40:48.310 --> 00:40:50.130 align:middle line:90% The consensus estimate. 00:40:50.130 --> 00:40:53.730 align:middle line:84% It's actually been revised in a recent paper. 00:40:53.730 --> 00:40:56.067 align:middle line:84% The consensus-- actually, it's an interesting story, 00:40:56.067 --> 00:40:58.150 align:middle line:84% a little bit of a sad story about it for academia. 00:40:58.150 --> 00:41:01.180 align:middle line:84% The consensus estimate was that the elasticity 00:41:01.180 --> 00:41:04.540 align:middle line:84% of how long people stayed unemployed 00:41:04.540 --> 00:41:08.380 align:middle line:84% with respect to the benefit was about 0.8, 00:41:08.380 --> 00:41:11.980 align:middle line:84% that for every 10% benefits were higher people stayed unemployed 00:41:11.980 --> 00:41:15.700 align:middle line:84% about 8% longer, which is a pretty big effect. 00:41:15.700 --> 00:41:18.180 align:middle line:84% There's been a recent article which suggests maybe 00:41:18.180 --> 00:41:21.100 align:middle line:84% too big an effect, because it's noted something 00:41:21.100 --> 00:41:24.540 align:middle line:84% that's a frequent test for what we call publication bias. 00:41:24.540 --> 00:41:28.140 align:middle line:84% There's a bias that's, quite frankly, the fault of publishing 00:41:28.140 --> 00:41:31.560 align:middle line:84% companies, which is that if a result is significant, 00:41:31.560 --> 00:41:34.380 align:middle line:84% it is more likely to get published than if it's not. 00:41:34.380 --> 00:41:36.780 align:middle line:84% And the way you test for publication bias is you 00:41:36.780 --> 00:41:39.820 align:middle line:84% look at the distribution of significant statistics. 00:41:39.820 --> 00:41:42.460 align:middle line:84% There's a massive spike at exactly the cutoff 00:41:42.460 --> 00:41:43.418 align:middle line:90% where it's significant. 00:41:43.418 --> 00:41:45.502 align:middle line:84% That suggests maybe people waited till that point, 00:41:45.502 --> 00:41:47.020 align:middle line:84% and then they published the article. 00:41:47.020 --> 00:41:49.478 align:middle line:84% There's a recent article which shows that's huge in this UI 00:41:49.478 --> 00:41:52.500 align:middle line:84% literature and suggests now that the right estimate is around 00:41:52.500 --> 00:41:56.032 align:middle line:84% 0.5, that there's been a lot of public-- that it's still-- 00:41:56.032 --> 00:41:58.240 align:middle line:84% the moral hazard, is still real, but maybe not as bad 00:41:58.240 --> 00:41:59.800 align:middle line:90% as we thought. 00:41:59.800 --> 00:42:00.642 align:middle line:90% It's a side note. 00:42:00.642 --> 00:42:02.100 align:middle line:84% Doesn't really matter for you guys. 00:42:02.100 --> 00:42:03.517 align:middle line:84% It's just an interesting side note 00:42:03.517 --> 00:42:05.685 align:middle line:84% about the hazards of published research. 00:42:05.685 --> 00:42:07.560 align:middle line:84% The bottom line is moral hazard is real here. 00:42:07.560 --> 00:42:11.000 align:middle line:90% 0.5 to 0.8, that's a big range. 00:42:11.000 --> 00:42:15.480 align:middle line:84% But that raises an interesting question. 00:42:15.480 --> 00:42:18.920 align:middle line:84% If higher unemployment insurance benefits 00:42:18.920 --> 00:42:21.360 align:middle line:84% cause you to stay out of work longer, 00:42:21.360 --> 00:42:25.680 align:middle line:84% either 50% longer or 80% longer, whatever it, is that bad? 00:42:25.680 --> 00:42:29.360 align:middle line:84% In fact, why are you not sure that's bad? 00:42:29.360 --> 00:42:32.720 align:middle line:84% Let's say it is true that higher unemployment insurance 00:42:32.720 --> 00:42:39.360 align:middle line:84% benefits cause you to stay out of work longer causally. 00:42:39.360 --> 00:42:41.360 align:middle line:84% What determines if that's actually 00:42:41.360 --> 00:42:42.940 align:middle line:90% efficient or inefficient? 00:42:42.940 --> 00:42:45.800 align:middle line:90% 00:42:45.800 --> 00:42:46.640 align:middle line:90% Yeah? 00:42:46.640 --> 00:42:49.400 align:middle line:84% AUDIENCE: Well, wouldn't that suggest that people value 00:42:49.400 --> 00:42:52.600 align:middle line:84% leisure more than they are allowed to? 00:42:52.600 --> 00:42:54.420 align:middle line:84% JON GRUBER: Well, let's go to allowed to. 00:42:54.420 --> 00:43:00.250 align:middle line:90% 00:43:00.250 --> 00:43:02.150 align:middle line:84% Certainly people are valuing leisure more. 00:43:02.150 --> 00:43:05.930 align:middle line:84% Certainly they stay at a lower value of leisure. 00:43:05.930 --> 00:43:08.030 align:middle line:84% To push your argument, the way-- by allowed to, 00:43:08.030 --> 00:43:09.822 align:middle line:84% you have this view that people can't really 00:43:09.822 --> 00:43:11.970 align:middle line:84% take advantage of their preferences for leisure. 00:43:11.970 --> 00:43:14.178 align:middle line:84% What would cause you not to be able to take advantage 00:43:14.178 --> 00:43:16.370 align:middle line:90% of a preference for leisure? 00:43:16.370 --> 00:43:18.690 align:middle line:84% The fact that you don't have money in the bank. 00:43:18.690 --> 00:43:20.810 align:middle line:84% If everyone could borrow-- let's say 00:43:20.810 --> 00:43:22.890 align:middle line:84% everyone could freely borrow, and let's say 00:43:22.890 --> 00:43:25.090 align:middle line:90% they truly needed a week off. 00:43:25.090 --> 00:43:27.863 align:middle line:84% They just take the week off and borrow to pay-- to borrow 00:43:27.863 --> 00:43:30.030 align:middle line:84% and pay it back by working hard later in the future. 00:43:30.030 --> 00:43:31.490 align:middle line:90% You can't do that. 00:43:31.490 --> 00:43:35.130 align:middle line:84% And in particular, the concern is what if you need more time 00:43:35.130 --> 00:43:37.410 align:middle line:90% to find a good job? 00:43:37.410 --> 00:43:42.650 align:middle line:84% What if, in fact, the problem is that UI is too short? 00:43:42.650 --> 00:43:45.010 align:middle line:84% And with a short UI, people are rushed 00:43:45.010 --> 00:43:47.210 align:middle line:84% into jobs that are not the best fit for them. 00:43:47.210 --> 00:43:50.530 align:middle line:84% We don't want the brain surgeon working at McDonald's, OK? 00:43:50.530 --> 00:43:52.530 align:middle line:84% But it's going to take the brain surgeon a while 00:43:52.530 --> 00:43:53.690 align:middle line:90% to find a good job. 00:43:53.690 --> 00:43:58.150 align:middle line:84% And the brain surgeon can't borrow to pay their mortgage. 00:43:58.150 --> 00:43:59.930 align:middle line:84% They're going to rush into a bad job. 00:43:59.930 --> 00:44:02.190 align:middle line:84% It's hard to look for a job when you already have one. 00:44:02.190 --> 00:44:03.430 align:middle line:84% Once you have the brain-- this is 00:44:03.430 --> 00:44:05.490 align:middle line:84% a very important point and subtle point in heart. 00:44:05.490 --> 00:44:07.390 align:middle line:84% So let's stay in it for a minute. 00:44:07.390 --> 00:44:11.093 align:middle line:84% Once again, we've got the brain surgeon loses the job. 00:44:11.093 --> 00:44:12.510 align:middle line:84% We're going to ask, are we worried 00:44:12.510 --> 00:44:14.510 align:middle line:84% the brain surgeon is going to rush back and take 00:44:14.510 --> 00:44:15.550 align:middle line:90% an inappropriate job? 00:44:15.550 --> 00:44:18.470 align:middle line:84% The answer is no, as long as credit markets 00:44:18.470 --> 00:44:20.190 align:middle line:90% work appropriately. 00:44:20.190 --> 00:44:21.582 align:middle line:84% If credit markets work perfectly, 00:44:21.582 --> 00:44:23.290 align:middle line:84% the brain surgeon would not have to rush. 00:44:23.290 --> 00:44:24.370 align:middle line:84% Because the brain surgeon can say, look, 00:44:24.370 --> 00:44:25.790 align:middle line:90% I'm going to be rich again. 00:44:25.790 --> 00:44:27.890 align:middle line:84% I'll just borrow against my future wealth, 00:44:27.890 --> 00:44:30.270 align:middle line:84% and I'll take as long as it needs to find a good job. 00:44:30.270 --> 00:44:33.110 align:middle line:84% The problem is, banks won't make that loan. 00:44:33.110 --> 00:44:36.950 align:middle line:84% And as a result, they face what we call a liquidity constraint. 00:44:36.950 --> 00:44:39.990 align:middle line:84% They cannot get the money they need to find the best job 00:44:39.990 --> 00:44:42.310 align:middle line:90% for them, maybe. 00:44:42.310 --> 00:44:43.862 align:middle line:90% So they need more time. 00:44:43.862 --> 00:44:46.070 align:middle line:84% So the fact that unemployment insurance is giving you 00:44:46.070 --> 00:44:48.950 align:middle line:90% more time might be good. 00:44:48.950 --> 00:44:51.130 align:middle line:84% It might be helping people match the jobs they want. 00:44:51.130 --> 00:44:52.030 align:middle line:90% Yeah? 00:44:52.030 --> 00:44:55.240 align:middle line:84% AUDIENCE: It isn't a certainty that at approach you 00:44:55.240 --> 00:44:58.280 align:middle line:84% will find a certain job you're looking for. 00:44:58.280 --> 00:45:00.120 align:middle line:84% It's the credit market constrains you 00:45:00.120 --> 00:45:01.337 align:middle line:90% with the obstacles. 00:45:01.337 --> 00:45:02.920 align:middle line:84% JON GRUBER: Credit market [INAUDIBLE]. 00:45:02.920 --> 00:45:05.640 align:middle line:84% But if it's not a certainty, the credit market 00:45:05.640 --> 00:45:08.600 align:middle line:84% still functions to charge a higher interest rate. 00:45:08.600 --> 00:45:09.862 align:middle line:90% AUDIENCE: And then I guess-- 00:45:09.862 --> 00:45:11.320 align:middle line:84% JON GRUBER: But the point is that-- 00:45:11.320 --> 00:45:13.640 align:middle line:84% AUDIENCE: [INAUDIBLE] it doesn't work out well for the brain 00:45:13.640 --> 00:45:14.640 align:middle line:90% surgeon who ends up-- 00:45:14.640 --> 00:45:16.520 align:middle line:84% JON GRUBER: No, the brain surgeon 00:45:16.520 --> 00:45:18.520 align:middle line:84% who ends up not getting a job will just default, 00:45:18.520 --> 00:45:20.280 align:middle line:90% and the bank will lead it. 00:45:20.280 --> 00:45:22.960 align:middle line:84% But that's why they'll have to charge a higher interest rate. 00:45:22.960 --> 00:45:25.720 align:middle line:90% So now we have this conundrum. 00:45:25.720 --> 00:45:28.078 align:middle line:84% Well, shit, we know more unemployment insurance benefits 00:45:28.078 --> 00:45:29.620 align:middle line:84% cause guys to stay unemployed longer. 00:45:29.620 --> 00:45:32.480 align:middle line:84% But how do we tell is that helping 00:45:32.480 --> 00:45:35.720 align:middle line:84% them-- is that good in the sense it's helping them find better 00:45:35.720 --> 00:45:39.820 align:middle line:84% jobs or bad in the sense just subsidize them sit on the couch? 00:45:39.820 --> 00:45:42.820 align:middle line:84% Remember, we talked last time about the distortion. 00:45:42.820 --> 00:45:44.360 align:middle line:84% Sitting on the couch is inefficient. 00:45:44.360 --> 00:45:45.860 align:middle line:84% Because if it's sitting on the couch 00:45:45.860 --> 00:45:47.180 align:middle line:84% you wouldn't have done otherwise, 00:45:47.180 --> 00:45:48.440 align:middle line:84% but you're doing because of these benefits, that's 00:45:48.440 --> 00:45:49.480 align:middle line:90% inefficient. 00:45:49.480 --> 00:45:52.210 align:middle line:84% So how do we determine whether UI benefits 00:45:52.210 --> 00:45:54.650 align:middle line:84% are causing you to get a better job or sit on the couch? 00:45:54.650 --> 00:45:57.650 align:middle line:90% What would you do? 00:45:57.650 --> 00:45:59.090 align:middle line:90% How could you test that? 00:45:59.090 --> 00:46:00.270 align:middle line:90% How would you test that? 00:46:00.270 --> 00:46:03.350 align:middle line:90% 00:46:03.350 --> 00:46:03.850 align:middle line:90% Yeah? 00:46:03.850 --> 00:46:05.090 align:middle line:90% Go ahead. 00:46:05.090 --> 00:46:09.330 align:middle line:84% AUDIENCE: Could you graph the change in productivity or income 00:46:09.330 --> 00:46:11.590 align:middle line:90% from the old job to the new? 00:46:11.590 --> 00:46:12.730 align:middle line:90% JON GRUBER: Exactly. 00:46:12.730 --> 00:46:13.810 align:middle line:90% You don't have to-- 00:46:13.810 --> 00:46:16.230 align:middle line:84% once again, you're answering in more level detail we need. 00:46:16.230 --> 00:46:20.710 align:middle line:84% But the bottom line is if it's helping you find a better job, 00:46:20.710 --> 00:46:23.210 align:middle line:84% we should see you end up with a better wage. 00:46:23.210 --> 00:46:26.650 align:middle line:84% So the test here, instead of a delta C regression 00:46:26.650 --> 00:46:28.650 align:middle line:84% like I was running-- where's my regression? 00:46:28.650 --> 00:46:30.525 align:middle line:84% I don't know where that was-- yeah, up there. 00:46:30.525 --> 00:46:35.290 align:middle line:84% Instead of regressing delta C, do delta wage. 00:46:35.290 --> 00:46:39.230 align:middle line:84% What happens to your wage from your old job to your new job 00:46:39.230 --> 00:46:41.690 align:middle line:90% as UI gets more generous? 00:46:41.690 --> 00:46:44.650 align:middle line:84% If more generous UI is improving your search, 00:46:44.650 --> 00:46:47.770 align:middle line:84% you should see a positive coefficient there. 00:46:47.770 --> 00:46:50.237 align:middle line:84% If more generous UI is not improving your search, 00:46:50.237 --> 00:46:51.070 align:middle line:90% you should see zero. 00:46:51.070 --> 00:46:55.230 align:middle line:90% And the answer is you see zero. 00:46:55.230 --> 00:46:58.467 align:middle line:84% The bottom line is guys are sitting around, OK? 00:46:58.467 --> 00:47:00.550 align:middle line:84% That basically there's no evidence that's actually 00:47:00.550 --> 00:47:02.130 align:middle line:90% helping them find better jobs. 00:47:02.130 --> 00:47:03.950 align:middle line:90% It's just subsidizing leisure. 00:47:03.950 --> 00:47:05.382 align:middle line:90% Yeah? 00:47:05.382 --> 00:47:08.043 align:middle line:90% AUDIENCE: [INAUDIBLE]. 00:47:08.043 --> 00:47:09.210 align:middle line:90% JON GRUBER: Gina, I'm sorry. 00:47:09.210 --> 00:47:10.398 align:middle line:90% You got-- louder and slower. 00:47:10.398 --> 00:47:11.190 align:middle line:90% AUDIENCE: Oh, yeah. 00:47:11.190 --> 00:47:13.830 align:middle line:84% What happened with the type of work 00:47:13.830 --> 00:47:18.150 align:middle line:84% that you do that is not [INAUDIBLE] related [INAUDIBLE] 00:47:18.150 --> 00:47:19.310 align:middle line:90% childbirth? 00:47:19.310 --> 00:47:21.270 align:middle line:84% JON GRUBER: Great, great question. 00:47:21.270 --> 00:47:22.572 align:middle line:90% That's a great question. 00:47:22.572 --> 00:47:24.030 align:middle line:84% We'll come back later to childcare. 00:47:24.030 --> 00:47:26.155 align:middle line:84% If you think there's a market failure in the people 00:47:26.155 --> 00:47:28.970 align:middle line:84% under-providing child care, that could be an argument for saying, 00:47:28.970 --> 00:47:30.628 align:middle line:84% well, gee, maybe the fact, their wages 00:47:30.628 --> 00:47:32.170 align:middle line:84% don't go up with providing childcare, 00:47:32.170 --> 00:47:33.410 align:middle line:90% maybe we're mismeasuring it. 00:47:33.410 --> 00:47:34.618 align:middle line:90% You really want total margin. 00:47:34.618 --> 00:47:36.160 align:middle line:84% This shouldn't be-- delta wage should 00:47:36.160 --> 00:47:37.330 align:middle line:90% be delta marginal product. 00:47:37.330 --> 00:47:39.538 align:middle line:84% Marginal product includes all the things you produce, 00:47:39.538 --> 00:47:40.670 align:middle line:90% including childcare. 00:47:40.670 --> 00:47:42.210 align:middle line:90% That's technically right. 00:47:42.210 --> 00:47:44.950 align:middle line:84% So it's not a full-- it's not a perfect test. 00:47:44.950 --> 00:47:45.630 align:middle line:90% OK. 00:47:45.630 --> 00:47:48.670 align:middle line:84% But clearly we think there's pretty significant moral hazard 00:47:48.670 --> 00:47:50.880 align:middle line:90% going on with UI. 00:47:50.880 --> 00:47:51.800 align:middle line:90% OK? 00:47:51.800 --> 00:47:54.560 align:middle line:90% So that's the evidence for UI 00:47:54.560 --> 00:47:56.760 align:middle line:90% Now we come to DI. 00:47:56.760 --> 00:47:58.160 align:middle line:90% Here's the problem with DI. 00:47:58.160 --> 00:47:59.660 align:middle line:84% You can do the same kind of study, 00:47:59.660 --> 00:48:03.640 align:middle line:90% except it doesn't vary by state. 00:48:03.640 --> 00:48:04.860 align:middle line:90% So what are you going to do? 00:48:04.860 --> 00:48:09.080 align:middle line:84% Well, in that case, you turn to a different empirical strategy, 00:48:09.080 --> 00:48:14.240 align:middle line:84% which is shown in figure 14-5, a regression discontinuity 00:48:14.240 --> 00:48:16.200 align:middle line:90% strategy. 00:48:16.200 --> 00:48:20.640 align:middle line:90% This is from a study in-- 00:48:20.640 --> 00:48:29.500 align:middle line:84% I believe it was in the Netherlands again. 00:48:29.500 --> 00:48:31.360 align:middle line:84% What the Netherlands-- the Netherlands 00:48:31.360 --> 00:48:35.000 align:middle line:84% has a program where the benefits discretely, at age 45, 00:48:35.000 --> 00:48:37.400 align:middle line:90% get more generous. 00:48:37.400 --> 00:48:40.060 align:middle line:84% Exactly at age 45, benefits get more generous. 00:48:40.060 --> 00:48:43.920 align:middle line:84% What the graph on the top shows-- 00:48:43.920 --> 00:48:48.380 align:middle line:84% the graph on the top figure 14.5 shows 00:48:48.380 --> 00:48:51.500 align:middle line:84% how claiming rates for disability insurance rates 00:48:51.500 --> 00:48:52.180 align:middle line:90% with age. 00:48:52.180 --> 00:48:53.080 align:middle line:90% Now, they go up. 00:48:53.080 --> 00:48:54.720 align:middle line:84% The older you are, the more likely you are to be disabled. 00:48:54.720 --> 00:48:56.020 align:middle line:90% That's not a surprise. 00:48:56.020 --> 00:48:59.140 align:middle line:84% But there's a huge jump exactly at 45, 00:48:59.140 --> 00:49:02.460 align:middle line:84% and there's no reason for this discrete jump at 45 00:49:02.460 --> 00:49:04.673 align:middle line:90% on the benefits change. 00:49:04.673 --> 00:49:06.340 align:middle line:84% So this isn't a difference in difference 00:49:06.340 --> 00:49:07.298 align:middle line:90% in a traditional sense. 00:49:07.298 --> 00:49:10.420 align:middle line:84% This is literally-- it's like time series, in a way. 00:49:10.420 --> 00:49:15.540 align:middle line:84% It's just using a very sharp event and saying, 00:49:15.540 --> 00:49:18.420 align:middle line:84% this sharp event is exactly consistent 00:49:18.420 --> 00:49:21.600 align:middle line:84% with there being an incentive effect of this. 00:49:21.600 --> 00:49:23.980 align:middle line:84% And then the bottom graph shows what 00:49:23.980 --> 00:49:26.260 align:middle line:84% happens to-- so let me explain how you 00:49:26.260 --> 00:49:27.820 align:middle line:90% interpret a graph like this. 00:49:27.820 --> 00:49:30.980 align:middle line:90% The x-axis is age. 00:49:30.980 --> 00:49:33.803 align:middle line:84% The y-axis is how much income people get from DI. 00:49:33.803 --> 00:49:35.220 align:middle line:84% Think of it as how often you claim 00:49:35.220 --> 00:49:37.860 align:middle line:90% DI, the odds of claiming DI. 00:49:37.860 --> 00:49:43.220 align:middle line:84% Each dot is-- I think it's a month of birth, the average. 00:49:43.220 --> 00:49:44.940 align:middle line:84% So each dot is the average amount 00:49:44.940 --> 00:49:49.270 align:middle line:84% of income for die for that month of birth in the Netherlands. 00:49:49.270 --> 00:49:50.650 align:middle line:90% So the dots are general-- 00:49:50.650 --> 00:49:53.470 align:middle line:84% And then the lines that are in red 00:49:53.470 --> 00:49:55.390 align:middle line:84% are the estimated regression coefficients 00:49:55.390 --> 00:49:58.270 align:middle line:84% on either side of the vertical line. 00:49:58.270 --> 00:50:00.350 align:middle line:84% See, there's generally a positive slope 00:50:00.350 --> 00:50:03.302 align:middle line:84% before and after the vertical line. 00:50:03.302 --> 00:50:05.510 align:middle line:84% That's just saying people are more likely to claim DI 00:50:05.510 --> 00:50:06.490 align:middle line:90% as they get older. 00:50:06.490 --> 00:50:12.110 align:middle line:84% But it's the jump in that slope at 45 that is really striking. 00:50:12.110 --> 00:50:14.870 align:middle line:84% And if you look at the bottom, you do the same graph 00:50:14.870 --> 00:50:17.630 align:middle line:90% but with whether people work. 00:50:17.630 --> 00:50:19.830 align:middle line:84% And you see there's a corresponding jump 00:50:19.830 --> 00:50:24.670 align:middle line:90% down at age 45. 00:50:24.670 --> 00:50:27.870 align:middle line:84% So that is a regression discontinuity approach 00:50:27.870 --> 00:50:30.435 align:middle line:84% to showing that disability insurance has 00:50:30.435 --> 00:50:32.310 align:middle line:84% an effect on people's labor supply decisions. 00:50:32.310 --> 00:50:33.970 align:middle line:90% That is, this is profound. 00:50:33.970 --> 00:50:39.950 align:middle line:84% This is saying people who could actually work are claiming DI. 00:50:39.950 --> 00:50:42.750 align:middle line:84% If DI was just people who truly had career-ending injuries 00:50:42.750 --> 00:50:46.080 align:middle line:84% and could never work again, this line should be flat. 00:50:46.080 --> 00:50:49.400 align:middle line:84% There shouldn't be-- the benefit level shouldn't matter. 00:50:49.400 --> 00:50:51.340 align:middle line:84% If you truly couldn't work again, 00:50:51.340 --> 00:50:53.757 align:middle line:84% then you should just claim it when you can't work, and not 00:50:53.757 --> 00:50:54.800 align:middle line:90% claim it when you can. 00:50:54.800 --> 00:50:58.200 align:middle line:84% Clearly this is not saying that, clearly, some people claiming DI 00:50:58.200 --> 00:51:00.620 align:middle line:84% could work and are claiming DI instead. 00:51:00.620 --> 00:51:03.120 align:middle line:84% And that's why you get this relationship. 00:51:03.120 --> 00:51:05.460 align:middle line:84% There's a lot of fun studies in DI like this. 00:51:05.460 --> 00:51:08.100 align:middle line:84% I've got a study in Canada where there was a difference 00:51:08.100 --> 00:51:09.558 align:middle line:84% in different study where, actually, 00:51:09.558 --> 00:51:12.440 align:middle line:84% one province of Canada changed their benefit and others didn't. 00:51:12.440 --> 00:51:13.940 align:middle line:90% There's a really cool study. 00:51:13.940 --> 00:51:16.080 align:middle line:84% Another empirical approach that's 00:51:16.080 --> 00:51:19.720 align:middle line:84% really cool in catching on now is 00:51:19.720 --> 00:51:23.680 align:middle line:84% to use the fact that while DI is a national program, 00:51:23.680 --> 00:51:26.640 align:middle line:84% the people who adjudicate whether you're disabled 00:51:26.640 --> 00:51:29.840 align:middle line:84% are humans, and those humans have tendencies. 00:51:29.840 --> 00:51:32.280 align:middle line:84% And it turns out you get randomly assigned 00:51:32.280 --> 00:51:34.760 align:middle line:90% to who adjudicate your case. 00:51:34.760 --> 00:51:38.460 align:middle line:84% So let's say that Ahmet applies for disability. 00:51:38.460 --> 00:51:41.580 align:middle line:84% He gets randomly assigned to me, and I'm a tough-ass judge. 00:51:41.580 --> 00:51:43.500 align:middle line:90% I never approve anything. 00:51:43.500 --> 00:51:46.300 align:middle line:84% And Valerie applies for DI, and she gets assigned 00:51:46.300 --> 00:51:47.760 align:middle line:90% to Paul, who's a loose judge. 00:51:47.760 --> 00:51:50.020 align:middle line:90% He approves everything. 00:51:50.020 --> 00:51:52.640 align:middle line:84% Then we actually have-- since that assignment is random, 00:51:52.640 --> 00:51:54.180 align:middle line:90% we have an experiment. 00:51:54.180 --> 00:51:56.608 align:middle line:84% What happens to people like Ahmet, 00:51:56.608 --> 00:51:58.900 align:middle line:84% who are unlikely to get DI, versus people like Valerie, 00:51:58.900 --> 00:52:00.478 align:middle line:90% who are likely to get it? 00:52:00.478 --> 00:52:02.020 align:middle line:84% And what you find is a similar thing, 00:52:02.020 --> 00:52:04.140 align:middle line:84% that people are unlikely to get DI end up 00:52:04.140 --> 00:52:07.343 align:middle line:84% working a lot more than people who are likely to get it. 00:52:07.343 --> 00:52:09.260 align:middle line:84% So all these approaches give a similar answer. 00:52:09.260 --> 00:52:11.100 align:middle line:84% So the standard difference in difference, 00:52:11.100 --> 00:52:13.860 align:middle line:84% the regression discontinuity, the random assignment 00:52:13.860 --> 00:52:15.460 align:middle line:90% across adjudicators. 00:52:15.460 --> 00:52:17.820 align:middle line:84% And they all say the elasticity-- 00:52:17.820 --> 00:52:19.220 align:middle line:90% so this is the elasticity of UI. 00:52:19.220 --> 00:52:23.900 align:middle line:84% The elasticity of labor supply with respect to DI benefits 00:52:23.900 --> 00:52:26.060 align:middle line:90% is around 0.3. 00:52:26.060 --> 00:52:29.100 align:middle line:84% So smaller, considerably smaller than what you 00:52:29.100 --> 00:52:31.020 align:middle line:90% get for unemployment insurance. 00:52:31.020 --> 00:52:36.460 align:middle line:84% Then it's true, higher benefits cause more use of disability. 00:52:36.460 --> 00:52:39.180 align:middle line:84% But the effect is not as large, which is once again not 00:52:39.180 --> 00:52:41.590 align:middle line:90% surprising. 00:52:41.590 --> 00:52:42.330 align:middle line:90% It's easy. 00:52:42.330 --> 00:52:45.870 align:middle line:84% Remember, moral hazard is going to rise in-- 00:52:45.870 --> 00:52:49.790 align:middle line:84% moral hazard is going to rise in difficulty of observability 00:52:49.790 --> 00:52:51.150 align:middle line:90% or verifiability. 00:52:51.150 --> 00:52:53.130 align:middle line:84% The harder it is to verify something, 00:52:53.130 --> 00:52:55.310 align:middle line:84% the more moral hazard there's going to be. 00:52:55.310 --> 00:52:58.285 align:middle line:84% Well, it's hard to verify if you're looking for work. 00:52:58.285 --> 00:53:00.410 align:middle line:84% It's pretty easy to verify-- it's easier to verify, 00:53:00.410 --> 00:53:03.510 align:middle line:84% but not perfect, if you're disabled. 00:53:03.510 --> 00:53:06.170 align:middle line:90% OK, that's disability insurance. 00:53:06.170 --> 00:53:07.910 align:middle line:84% Finally, we come to workers' comp. 00:53:07.910 --> 00:53:10.610 align:middle line:84% Workers' comp, once again, it's a state program. 00:53:10.610 --> 00:53:12.090 align:middle line:90% So we have the same of approach. 00:53:12.090 --> 00:53:15.430 align:middle line:84% Indeed, we can even do a difference in difference 00:53:15.430 --> 00:53:16.890 align:middle line:90% in difference within a state. 00:53:16.890 --> 00:53:20.470 align:middle line:84% Figure 14-7 is from a famous study by Alan Krueger where 00:53:20.470 --> 00:53:23.330 align:middle line:84% the State of Minnesota changed its schedule, 00:53:23.330 --> 00:53:25.950 align:middle line:90% but in multiple places. 00:53:25.950 --> 00:53:27.870 align:middle line:84% So you could actually compare, say, group E 00:53:27.870 --> 00:53:31.910 align:middle line:84% to group D and group C to group B to group A. 00:53:31.910 --> 00:53:34.430 align:middle line:84% And then since they're very arbitrarily close to each other, 00:53:34.430 --> 00:53:36.710 align:middle line:84% you can actually learn less about recessionary effects 00:53:36.710 --> 00:53:39.270 align:middle line:84% because you have a bunch of comparisons. 00:53:39.270 --> 00:53:41.850 align:middle line:84% And the bottom line is he found-- so 00:53:41.850 --> 00:53:44.650 align:middle line:84% for example, he-- what the numbers below 00:53:44.650 --> 00:53:47.230 align:middle line:84% are the elasticities you get at each segment. 00:53:47.230 --> 00:53:50.010 align:middle line:84% Or it's, I'm sorry, the delta duration in each segment. 00:53:50.010 --> 00:53:52.450 align:middle line:84% So in other words, duration went up a lot for segments 00:53:52.450 --> 00:53:56.010 align:middle line:84% E, C, and A, and barely up for segments B and D, 00:53:56.010 --> 00:53:59.170 align:middle line:84% which is consistent with this being a causal effect 00:53:59.170 --> 00:54:02.250 align:middle line:90% of workers' comp. 00:54:02.250 --> 00:54:04.670 align:middle line:84% Here's what's striking about the workers' comp literature. 00:54:04.670 --> 00:54:06.530 align:middle line:90% You get a huge elasticity. 00:54:06.530 --> 00:54:11.930 align:middle line:84% Krueger's paper implies that elasticity of 1.8. 00:54:11.930 --> 00:54:15.270 align:middle line:84% For every 10% benefits, workers' comp benefits go up. 00:54:15.270 --> 00:54:17.930 align:middle line:90% You get 18% longer durations. 00:54:17.930 --> 00:54:20.070 align:middle line:84% The literature says it's found a smaller number. 00:54:20.070 --> 00:54:23.250 align:middle line:84% But I would say the consensus-- probably the consensus-- 00:54:23.250 --> 00:54:28.690 align:middle line:84% workers' comp elasticity is around 1, higher than UI, 00:54:28.690 --> 00:54:32.410 align:middle line:84% that basically people are very elastic in their decisions 00:54:32.410 --> 00:54:36.690 align:middle line:84% on how long they stay out of work with respect 00:54:36.690 --> 00:54:40.322 align:middle line:84% to the generosity of workers' comp benefits. 00:54:40.322 --> 00:54:42.780 align:middle line:84% Now, workers' comp, there's all sorts of other fun studies, 00:54:42.780 --> 00:54:44.220 align:middle line:90% as well. 00:54:44.220 --> 00:54:46.180 align:middle line:84% In workers' comp, there's something 00:54:46.180 --> 00:54:49.420 align:middle line:90% called the Monday Effect. 00:54:49.420 --> 00:54:51.380 align:middle line:84% Let me explain the Monday Effect. 00:54:51.380 --> 00:54:53.500 align:middle line:84% Workers' comp, roughly speaking, 90% 00:54:53.500 --> 00:54:55.700 align:middle line:84% of workers' comp for two kinds of injuries-- 00:54:55.700 --> 00:55:01.430 align:middle line:84% lacerations, you get cut, or sprains and-- 00:55:01.430 --> 00:55:03.120 align:middle line:90% or sprains like you get hurt. 00:55:03.120 --> 00:55:04.980 align:middle line:84% You hurt your back, you hurt your shoulder, 00:55:04.980 --> 00:55:06.000 align:middle line:90% or something like that. 00:55:06.000 --> 00:55:08.140 align:middle line:84% Roughly speaking, that's the two types 00:55:08.140 --> 00:55:10.940 align:middle line:90% of injuries for workers' comp. 00:55:10.940 --> 00:55:18.860 align:middle line:84% Which type of injury do you think is most likely on Mondays 00:55:18.860 --> 00:55:21.380 align:middle line:84% relative to the other days of the week? 00:55:21.380 --> 00:55:24.740 align:middle line:84% What happens to the ratio of sprains and strains 00:55:24.740 --> 00:55:30.050 align:middle line:84% to lacerations on Monday versus the rest of the week? 00:55:30.050 --> 00:55:31.220 align:middle line:90% AUDIENCE: [INAUDIBLE]. 00:55:31.220 --> 00:55:32.340 align:middle line:90% JON GRUBER: Why? 00:55:32.340 --> 00:55:35.460 align:middle line:84% AUDIENCE: Because it's much less easy to verify. 00:55:35.460 --> 00:55:37.790 align:middle line:84% JON GRUBER: Well, it's easy to verify what? 00:55:37.790 --> 00:55:40.170 align:middle line:84% AUDIENCE: Whether you're going to actually be injured. 00:55:40.170 --> 00:55:42.910 align:middle line:84% JON GRUBER: Well, not that you're injured. 00:55:42.910 --> 00:55:45.448 align:middle line:90% In particular, what? 00:55:45.448 --> 00:55:46.990 align:middle line:84% AUDIENCE: If you got injured at work. 00:55:46.990 --> 00:55:48.350 align:middle line:90% JON GRUBER: At work. 00:55:48.350 --> 00:55:52.010 align:middle line:84% Because as long as you can drag yourself into work on Monday, 00:55:52.010 --> 00:55:53.750 align:middle line:90% you get hurt at softball. 00:55:53.750 --> 00:55:56.918 align:middle line:84% If you sprain your ACL at softball, 00:55:56.918 --> 00:55:59.210 align:middle line:84% you can call in and say, I sprained my ACL at softball. 00:55:59.210 --> 00:56:00.690 align:middle line:84% They'll say, great, we'll see when you heal up. 00:56:00.690 --> 00:56:02.310 align:middle line:84% Or you can drag yourself to work on Monday and trip, 00:56:02.310 --> 00:56:04.227 align:middle line:84% and say, oh my God, I sprained my ACL at work. 00:56:04.227 --> 00:56:06.990 align:middle line:84% They say, OK, here's $1,000 a week. 00:56:06.990 --> 00:56:07.810 align:middle line:90% OK? 00:56:07.810 --> 00:56:09.690 align:middle line:84% If you're lacerated, that's hard to do, 00:56:09.690 --> 00:56:14.490 align:middle line:84% unless you want to go in and reopen it or something at work. 00:56:14.490 --> 00:56:17.590 align:middle line:84% And once again, you see that Monday effect in great detail, 00:56:17.590 --> 00:56:20.110 align:middle line:84% that basically the ratio of sprains and strains 00:56:20.110 --> 00:56:21.790 align:middle line:84% to laceration is much higher on Monday. 00:56:21.790 --> 00:56:24.950 align:middle line:84% Sort of a clever test for moral hazard. 00:56:24.950 --> 00:56:26.730 align:middle line:84% The bottom line is moral hazard is real, 00:56:26.730 --> 00:56:31.070 align:middle line:84% and it looks like pretty large for workers' compensation. 00:56:31.070 --> 00:56:33.150 align:middle line:90% OK? 00:56:33.150 --> 00:56:34.330 align:middle line:90% Questions about that? 00:56:34.330 --> 00:56:37.170 align:middle line:90% 00:56:37.170 --> 00:56:39.610 align:middle line:90% Let's go back now. 00:56:39.610 --> 00:56:40.675 align:middle line:90% Yeah? 00:56:40.675 --> 00:56:42.050 align:middle line:84% AUDIENCE: So we've like, I guess, 00:56:42.050 --> 00:56:44.290 align:middle line:84% demonstrated like we're-- excuse me-- 00:56:44.290 --> 00:56:47.050 align:middle line:84% moral hazard is bad [INAUDIBLE] bad in the sense that 00:56:47.050 --> 00:56:47.910 align:middle line:90% it's inefficient. 00:56:47.910 --> 00:56:50.490 align:middle line:84% But in terms of if a government has 00:56:50.490 --> 00:56:54.070 align:middle line:84% a policy, such as workers' comp or disability insurance, 00:56:54.070 --> 00:56:58.350 align:middle line:84% and there is workers' comp and people choose to exploit that, 00:56:58.350 --> 00:57:01.547 align:middle line:84% is that necessarily bad for the government? 00:57:01.547 --> 00:57:02.630 align:middle line:90% JON GRUBER: Yes, it's bad. 00:57:02.630 --> 00:57:04.630 align:middle line:84% Well, it's bad from the government's perspective 00:57:04.630 --> 00:57:06.850 align:middle line:84% because the government represents society. 00:57:06.850 --> 00:57:08.750 align:middle line:84% And if you take it when you're not qualified, 00:57:08.750 --> 00:57:09.507 align:middle line:90% there's two costs. 00:57:09.507 --> 00:57:11.590 align:middle line:84% One is we have more tax revenue we got to collect. 00:57:11.590 --> 00:57:14.650 align:middle line:84% And two is you're not producing goods that make society richer. 00:57:14.650 --> 00:57:16.450 align:middle line:90% So the pie is smaller. 00:57:16.450 --> 00:57:17.880 align:middle line:90% Yeah? 00:57:17.880 --> 00:57:21.050 align:middle line:84% AUDIENCE: [INAUDIBLE] harder to verify [INAUDIBLE] 00:57:21.050 --> 00:57:24.212 align:middle line:84% affects more the moral hazard of the adverse selection? 00:57:24.212 --> 00:57:26.670 align:middle line:84% JON GRUBER: We're not talking about adverse selection here. 00:57:26.670 --> 00:57:29.650 align:middle line:84% There's no choice of insurance here. 00:57:29.650 --> 00:57:31.170 align:middle line:90% Yeah? 00:57:31.170 --> 00:57:33.570 align:middle line:84% AUDIENCE: Has it been looked into with workers' comp 00:57:33.570 --> 00:57:36.900 align:middle line:84% whether increasing benefits lets people 00:57:36.900 --> 00:57:38.760 align:middle line:90% stay longer and fully recover? 00:57:38.760 --> 00:57:40.020 align:middle line:90% JON GRUBER: Great question. 00:57:40.020 --> 00:57:42.273 align:middle line:84% Great question, which is-- this is 00:57:42.273 --> 00:57:45.200 align:middle line:84% sort of like what I talked about with unemployment insurance. 00:57:45.200 --> 00:57:46.700 align:middle line:84% I said, longer duration unemployment 00:57:46.700 --> 00:57:48.560 align:middle line:84% might be good because it helps you find a better job. 00:57:48.560 --> 00:57:50.310 align:middle line:84% Longer durations out of work might be good 00:57:50.310 --> 00:57:51.780 align:middle line:90% because you come back healthier. 00:57:51.780 --> 00:57:54.140 align:middle line:84% And there's not great evidence on that. 00:57:54.140 --> 00:57:57.640 align:middle line:84% I'm suspecting we will find that even if there's effect, 00:57:57.640 --> 00:57:59.640 align:middle line:84% it's small relative to the amount of extra time, 00:57:59.640 --> 00:58:00.773 align:middle line:90% people stay out of work. 00:58:00.773 --> 00:58:01.440 align:middle line:90% But it could be. 00:58:01.440 --> 00:58:02.190 align:middle line:90% It could be there. 00:58:02.190 --> 00:58:04.290 align:middle line:84% It's a great point and a very parallel point. 00:58:04.290 --> 00:58:06.540 align:middle line:84% It shows a great understanding of this parallel point, 00:58:06.540 --> 00:58:08.160 align:middle line:84% that just because you find an effect, 00:58:08.160 --> 00:58:09.827 align:middle line:84% there could be multiple interpretations. 00:58:09.827 --> 00:58:13.560 align:middle line:84% You got to take the extra step to do that interpretation. 00:58:13.560 --> 00:58:14.060 align:middle line:90% OK? 00:58:14.060 --> 00:58:17.780 align:middle line:90% So let's go back. 00:58:17.780 --> 00:58:21.240 align:middle line:84% Now here's what's cool is let's go back to table 14-2 and say, 00:58:21.240 --> 00:58:23.220 align:middle line:90% how did the government do? 00:58:23.220 --> 00:58:28.780 align:middle line:84% Well let's compare the next-to-last row of table 14-2, 00:58:28.780 --> 00:58:32.740 align:middle line:84% which shows the replacement rates for these three programs. 00:58:32.740 --> 00:58:33.760 align:middle line:90% What do we know? 00:58:33.760 --> 00:58:38.320 align:middle line:84% We know unemployment insurance has a modest consumption 00:58:38.320 --> 00:58:44.120 align:middle line:84% smoothing effect and a reasonably large moral hazard 00:58:44.120 --> 00:58:47.930 align:middle line:84% effect, which suggests benefits shouldn't be too generous. 00:58:47.930 --> 00:58:49.180 align:middle line:90% And they're not that generous. 00:58:49.180 --> 00:58:51.520 align:middle line:90% 44% replacement rate. 00:58:51.520 --> 00:58:56.960 align:middle line:84% Disability insurance has a smaller moral hazard effect 00:58:56.960 --> 00:58:58.680 align:middle line:84% and a larger consumption smoothing 00:58:58.680 --> 00:59:01.160 align:middle line:84% effect, which says benefits should be more generous. 00:59:01.160 --> 00:59:02.160 align:middle line:90% And they are. 00:59:02.160 --> 00:59:04.040 align:middle line:90% They're 60%. 00:59:04.040 --> 00:59:07.560 align:middle line:84% Workers' comp has an unknown, but probably mixed 00:59:07.560 --> 00:59:10.940 align:middle line:84% consumption smoothing benefit and a large moral hazard effect, 00:59:10.940 --> 00:59:13.160 align:middle line:84% which has benefits-- should be, if anything, similar 00:59:13.160 --> 00:59:14.560 align:middle line:90% or lower than UI. 00:59:14.560 --> 00:59:17.520 align:middle line:84% They're actually bigger than the other two programs. 00:59:17.520 --> 00:59:22.672 align:middle line:84% So this tells us workers' comp is too generous relative-- now, 00:59:22.672 --> 00:59:24.880 align:middle line:84% maybe too generous relative to some ultimate-- if God 00:59:24.880 --> 00:59:27.400 align:middle line:84% came down and spoke to us, but certainly too generous 00:59:27.400 --> 00:59:29.720 align:middle line:84% relative to the other two programs. 00:59:29.720 --> 00:59:32.770 align:middle line:84% It's hard to justify a world where there's 00:59:32.770 --> 00:59:37.130 align:middle line:84% a bigger moral hazard and likely not that much bigger consumption 00:59:37.130 --> 00:59:40.850 align:middle line:84% smoothing, and the benefit's so much larger. 00:59:40.850 --> 00:59:41.610 align:middle line:90% OK? 00:59:41.610 --> 00:59:43.650 align:middle line:84% That is cool, because we've actually 00:59:43.650 --> 00:59:46.330 align:middle line:84% used theory and empirical evidence 00:59:46.330 --> 00:59:48.090 align:middle line:90% to draw a policy conclusion. 00:59:48.090 --> 00:59:50.490 align:middle line:84% Now the next, of course, we get policymakers to listen. 00:59:50.490 --> 00:59:51.950 align:middle line:90% That's a harder thing. 00:59:51.950 --> 00:59:52.450 align:middle line:90% OK? 00:59:52.450 --> 00:59:54.010 align:middle line:90% That's chapter 9 stuff. 00:59:54.010 --> 00:59:54.850 align:middle line:90% But this is just a-- 00:59:54.850 --> 00:59:56.890 align:middle line:84% I love this point because this is really 00:59:56.890 --> 00:59:58.570 align:middle line:90% what makes economics fun. 00:59:58.570 --> 01:00:00.370 align:middle line:90% It's like, OK, we got a theory. 01:00:00.370 --> 01:00:01.970 align:middle line:90% We developed the theory. 01:00:01.970 --> 01:00:05.690 align:middle line:84% Now that theory suggests some key parameters we must know. 01:00:05.690 --> 01:00:07.710 align:middle line:84% We've estimated those parameters. 01:00:07.710 --> 01:00:10.050 align:middle line:84% And that gives us lessons for policy. 01:00:10.050 --> 01:00:12.170 align:middle line:84% And that's why I find this chapter very exciting. 01:00:12.170 --> 01:00:12.670 align:middle line:90% OK. 01:00:12.670 --> 01:00:13.390 align:middle line:90% Steven? 01:00:13.390 --> 01:00:15.290 align:middle line:84% AUDIENCE: [INAUDIBLE] that workers' comp 01:00:15.290 --> 01:00:18.268 align:middle line:84% is more generous than the others with the inclination 01:00:18.268 --> 01:00:20.810 align:middle line:84% that it should be more in line with the others, but not clear 01:00:20.810 --> 01:00:22.710 align:middle line:84% whether the others should become more generous? 01:00:22.710 --> 01:00:23.543 align:middle line:90% JON GRUBER: Exactly. 01:00:23.543 --> 01:00:24.630 align:middle line:90% No, I mean, I-- yeah. 01:00:24.630 --> 01:00:27.770 align:middle line:84% I haven't told you whether 44% is the right number. 01:00:27.770 --> 01:00:29.070 align:middle line:90% So really, it's a relative-- 01:00:29.070 --> 01:00:31.320 align:middle line:84% I still haven't told you exactly what the numbers are. 01:00:31.320 --> 01:00:33.700 align:middle line:84% But it's sort of a relative conclusion. 01:00:33.700 --> 01:00:35.260 align:middle line:90% All right? 01:00:35.260 --> 01:00:38.740 align:middle line:90% Now, two other things to cover. 01:00:38.740 --> 01:00:40.580 align:middle line:90% And I'm going to have to-- 01:00:40.580 --> 01:00:42.940 align:middle line:84% we are having a little celebration for our Nobel Prize 01:00:42.940 --> 01:00:43.863 align:middle line:90% winner at 4:00. 01:00:43.863 --> 01:00:45.780 align:middle line:84% So I'm going to have to whip out of here right 01:00:45.780 --> 01:00:46.763 align:middle line:90% when class is over. 01:00:46.763 --> 01:00:48.180 align:middle line:84% So let me cover a couple of things 01:00:48.180 --> 01:00:50.660 align:middle line:90% pretty quickly before I go. 01:00:50.660 --> 01:00:53.700 align:middle line:84% The first thing is, what about the role of firms? 01:00:53.700 --> 01:00:56.480 align:middle line:84% And what about the role of imperfect experience rating? 01:00:56.480 --> 01:00:59.820 align:middle line:84% So to understand what I mean by that, look at figure 14-8. 01:00:59.820 --> 01:01:07.060 align:middle line:84% Figure 14-8 shows the tax rate that firms face in Vermont. 01:01:07.060 --> 01:01:08.700 align:middle line:84% Let me explain what this graph shows. 01:01:08.700 --> 01:01:14.960 align:middle line:84% On the x-axis is basically a number 01:01:14.960 --> 01:01:20.300 align:middle line:84% which represents the extent to which a firm has taken 01:01:20.300 --> 01:01:22.030 align:middle line:90% advantage of the UI system. 01:01:22.030 --> 01:01:23.280 align:middle line:90% It's called the benefit ratio. 01:01:23.280 --> 01:01:25.740 align:middle line:84% It's the ratio of the benefits that's been paid out 01:01:25.740 --> 01:01:26.958 align:middle line:90% to the payroll of the firm. 01:01:26.958 --> 01:01:28.500 align:middle line:84% You don't need to understand it other 01:01:28.500 --> 01:01:30.260 align:middle line:84% than know the more you move to the right, 01:01:30.260 --> 01:01:34.080 align:middle line:84% the more the firm uses the UI system, the more layoffs it has. 01:01:34.080 --> 01:01:37.600 align:middle line:84% On the y-axis is the tax rate the firm pays. 01:01:37.600 --> 01:01:41.920 align:middle line:84% The black line would be perfect experience rating. 01:01:41.920 --> 01:01:45.320 align:middle line:84% That would be a system after some minimum to make the program 01:01:45.320 --> 01:01:46.760 align:middle line:90% balance. 01:01:46.760 --> 01:01:50.920 align:middle line:84% The slope of that would be such that every dollar of firm 01:01:50.920 --> 01:01:54.000 align:middle line:84% paid out in benefits, they'd pay back in taxes. 01:01:54.000 --> 01:01:56.060 align:middle line:84% Relative to that line, you notice two things. 01:01:56.060 --> 01:02:02.160 align:middle line:84% First of all, the red line is always below it. 01:02:02.160 --> 01:02:04.160 align:middle line:84% So it's imperfect experience rating that way. 01:02:04.160 --> 01:02:05.780 align:middle line:84% Second of all, there's flat parts. 01:02:05.780 --> 01:02:10.200 align:middle line:84% There's these steps as opposed to being smooth. 01:02:10.200 --> 01:02:12.980 align:middle line:84% And third and most important, there's a maximum rate. 01:02:12.980 --> 01:02:15.480 align:middle line:84% Someone asked, do you always pay ever and ever higher taxes? 01:02:15.480 --> 01:02:16.380 align:middle line:90% No, you don't. 01:02:16.380 --> 01:02:17.820 align:middle line:84% Once you hit that maximum rate, doesn't matter. 01:02:17.820 --> 01:02:19.320 align:middle line:84% You can lay off anybody you want. 01:02:19.320 --> 01:02:21.020 align:middle line:84% You're only paying the maximum rate. 01:02:21.020 --> 01:02:24.260 align:middle line:84% So you're never going to pay more than 7.7% of payroll, 01:02:24.260 --> 01:02:26.800 align:middle line:84% no matter how many people you lay off. 01:02:26.800 --> 01:02:29.450 align:middle line:84% So those are the three senses in which this is an imperfect 01:02:29.450 --> 01:02:32.210 align:middle line:90% experience rating. 01:02:32.210 --> 01:02:35.530 align:middle line:84% So the question is, how do we feel about that? 01:02:35.530 --> 01:02:37.730 align:middle line:84% Well, let's come back to the normative implication 01:02:37.730 --> 01:02:39.737 align:middle line:90% that was raised before. 01:02:39.737 --> 01:02:42.070 align:middle line:84% If you think about what unemployment insurance tax does, 01:02:42.070 --> 01:02:44.610 align:middle line:84% it says, hey, what experience rating does 01:02:44.610 --> 01:02:46.790 align:middle line:84% says, hey, you, firm, just laid off a bunch of guys. 01:02:46.790 --> 01:02:48.490 align:middle line:90% I'm going to raise your taxes. 01:02:48.490 --> 01:02:51.430 align:middle line:90% That sounds kind of harsh. 01:02:51.430 --> 01:02:52.930 align:middle line:84% So in that sense, you can understand 01:02:52.930 --> 01:02:55.690 align:middle line:84% the argument for imperfect experience rating. 01:02:55.690 --> 01:02:58.970 align:middle line:84% But what market failure does that argument rely on? 01:02:58.970 --> 01:03:02.170 align:middle line:84% What do you need for that argument to hold? 01:03:02.170 --> 01:03:06.432 align:middle line:84% Or how would you argue against that argument? 01:03:06.432 --> 01:03:07.890 align:middle line:84% Why would you say, well, firms that 01:03:07.890 --> 01:03:10.717 align:middle line:84% have laid off a lot of people shouldn't be subsidized? 01:03:10.717 --> 01:03:12.550 align:middle line:84% Because they have another thing they can do. 01:03:12.550 --> 01:03:15.250 align:middle line:90% 01:03:15.250 --> 01:03:16.448 align:middle line:90% What else could they do? 01:03:16.448 --> 01:03:18.490 align:middle line:84% How could they avail themselves of self-insurance 01:03:18.490 --> 01:03:21.030 align:middle line:84% in this situation, just like people can? 01:03:21.030 --> 01:03:24.570 align:middle line:90% 01:03:24.570 --> 01:03:25.830 align:middle line:90% In a perfect world. 01:03:25.830 --> 01:03:26.657 align:middle line:90% Yeah, Steven? 01:03:26.657 --> 01:03:27.990 align:middle line:90% AUDIENCE: Allocate more savings. 01:03:27.990 --> 01:03:29.698 align:middle line:84% JON GRUBER: Well, they can't really save, 01:03:29.698 --> 01:03:30.887 align:middle line:90% but they could borrow. 01:03:30.887 --> 01:03:32.470 align:middle line:84% They go to the bank and say, look, I'm 01:03:32.470 --> 01:03:34.678 align:middle line:84% laying off a bunch of guys now, but I'm fundamentally 01:03:34.678 --> 01:03:35.470 align:middle line:90% a sound firm. 01:03:35.470 --> 01:03:37.070 align:middle line:90% Give me money. 01:03:37.070 --> 01:03:40.190 align:middle line:84% And probably the bank may be more 01:03:40.190 --> 01:03:42.350 align:middle line:84% likely to do when the person goes 01:03:42.350 --> 01:03:45.950 align:middle line:84% because they have physical assets the bank can claim. 01:03:45.950 --> 01:03:48.630 align:middle line:84% So in a perfect credit market, this argument 01:03:48.630 --> 01:03:50.362 align:middle line:90% actually doesn't hold. 01:03:50.362 --> 01:03:52.070 align:middle line:84% In a perfect credit market, you shouldn't 01:03:52.070 --> 01:03:53.590 align:middle line:84% be subsidizing firms to lay people off 01:03:53.590 --> 01:03:55.510 align:middle line:84% because if they're a good risk, they can just 01:03:55.510 --> 01:03:57.350 align:middle line:90% borrow the money they need. 01:03:57.350 --> 01:03:59.650 align:middle line:84% If they're a bad risk, what are you doing? 01:03:59.650 --> 01:04:02.830 align:middle line:84% You're systematically subsidizing bad firms. 01:04:02.830 --> 01:04:05.630 align:middle line:84% So if you think about what imperfect experience rating is 01:04:05.630 --> 01:04:06.870 align:middle line:90% doing, it's two things. 01:04:06.870 --> 01:04:09.310 align:middle line:84% It's providing insurance to firms 01:04:09.310 --> 01:04:12.390 align:middle line:84% that are good bets that could come back, 01:04:12.390 --> 01:04:16.070 align:middle line:84% but providing subsidies to firms that are bad bets and just 01:04:16.070 --> 01:04:18.430 align:middle line:84% shouldn't exist in the first place. 01:04:18.430 --> 01:04:19.630 align:middle line:90% Alex, question? 01:04:19.630 --> 01:04:20.310 align:middle line:90% Yeah. 01:04:20.310 --> 01:04:22.650 align:middle line:84% So let me give you the perfect example. 01:04:22.650 --> 01:04:24.160 align:middle line:90% Let's talk about Canada. 01:04:24.160 --> 01:04:24.840 align:middle line:90% OK. 01:04:24.840 --> 01:04:27.600 align:middle line:84% Let's talk about five minute about the famous Canada story. 01:04:27.600 --> 01:04:29.640 align:middle line:84% In the Atlantic provinces of Canada, 01:04:29.640 --> 01:04:31.140 align:middle line:84% the economy basically has tanked. 01:04:31.140 --> 01:04:32.920 align:middle line:84% It was all natural resource economy. 01:04:32.920 --> 01:04:34.508 align:middle line:90% It's basically tanked. 01:04:34.508 --> 01:04:36.300 align:middle line:84% They have an unemployment insurance system. 01:04:36.300 --> 01:04:37.675 align:middle line:84% They rely on fishing, and fishing 01:04:37.675 --> 01:04:39.200 align:middle line:90% can't make any money anymore. 01:04:39.200 --> 01:04:41.158 align:middle line:84% They had an unemployment insurance system which 01:04:41.158 --> 01:04:44.200 align:middle line:90% had the following features. 01:04:44.200 --> 01:04:46.502 align:middle line:84% You got a replacement rate of 60%. 01:04:46.502 --> 01:04:48.460 align:middle line:84% So not outrageous, a little higher than the US. 01:04:48.460 --> 01:04:50.520 align:middle line:84% Not outrageous-- 60% replacement rate. 01:04:50.520 --> 01:04:55.280 align:middle line:84% But to get that, you had to work 10 weeks a year, 01:04:55.280 --> 01:04:58.520 align:middle line:84% and then you got the replacement for the other 42 weeks. 01:04:58.520 --> 01:05:00.640 align:middle line:84% So as long as you worked 10 weeks a year, 01:05:00.640 --> 01:05:04.360 align:middle line:90% you got UI for 42 weeks at 60% 01:05:04.360 --> 01:05:06.960 align:middle line:84% Now imagine you and four friends just want 01:05:06.960 --> 01:05:08.960 align:middle line:84% to think about setting up a fishing business. 01:05:08.960 --> 01:05:09.460 align:middle line:90% OK? 01:05:09.460 --> 01:05:10.418 align:middle line:90% This is a hard example. 01:05:10.418 --> 01:05:11.060 align:middle line:90% So just listen. 01:05:11.060 --> 01:05:12.460 align:middle line:84% Don't take notes, just listen to this. 01:05:12.460 --> 01:05:15.000 align:middle line:84% You and four friends that want to set up a fishing business, 01:05:15.000 --> 01:05:15.800 align:middle line:90% OK? 01:05:15.800 --> 01:05:18.708 align:middle line:84% The fishing boat can work 50 weeks a year. 01:05:18.708 --> 01:05:20.500 align:middle line:84% Let's leave the other two out for vacation. 01:05:20.500 --> 01:05:22.535 align:middle line:84% So you can only take it out 50 weeks a year. 01:05:22.535 --> 01:05:24.410 align:middle line:84% So each of you and your friends would take it 01:05:24.410 --> 01:05:27.170 align:middle line:90% for 10 of those weeks, OK? 01:05:27.170 --> 01:05:31.410 align:middle line:84% Let's say that with the fishing boat, 01:05:31.410 --> 01:05:37.650 align:middle line:84% you catch $8,000 worth of fish for every 10 weeks, 01:05:37.650 --> 01:05:42.250 align:middle line:84% or a total of $40,000 for the year. 01:05:42.250 --> 01:05:43.990 align:middle line:84% That is the amount you can produce. 01:05:43.990 --> 01:05:46.370 align:middle line:84% So each of you have worked 10 weeks a year, 01:05:46.370 --> 01:05:50.170 align:middle line:84% and you'll produce $40,000 of income. 01:05:50.170 --> 01:05:55.210 align:middle line:84% Now, what if you said to the authorities that after 10 weeks 01:05:55.210 --> 01:05:58.090 align:middle line:90% each of you got laid off? 01:05:58.090 --> 01:05:59.250 align:middle line:90% How would that work? 01:05:59.250 --> 01:06:03.690 align:middle line:84% Well, the first person, they would go out, fish. 01:06:03.690 --> 01:06:05.490 align:middle line:90% OK? 01:06:05.490 --> 01:06:09.130 align:middle line:84% Oh, and I also say that your wages are $8,000 for every 10 01:06:09.130 --> 01:06:11.570 align:middle line:90% weeks or $800 a week. 01:06:11.570 --> 01:06:17.090 align:middle line:84% I say that you earn $800 a week and you're 01:06:17.090 --> 01:06:19.330 align:middle line:90% laid off after 10 weeks. 01:06:19.330 --> 01:06:20.970 align:middle line:90% What does person A get? 01:06:20.970 --> 01:06:24.630 align:middle line:84% They get the $8,000 during the 10 weeks they work, 01:06:24.630 --> 01:06:27.150 align:middle line:90% plus they get 50-- 01:06:27.150 --> 01:06:35.150 align:middle line:90% 40-- 42 weeks of 60% of $800. 01:06:35.150 --> 01:06:38.310 align:middle line:84% So they get-- that's what they get for the rest of the year. 01:06:38.310 --> 01:06:40.190 align:middle line:84% Then person B works their 10 weeks, 01:06:40.190 --> 01:06:42.270 align:middle line:90% and they say they were laid off. 01:06:42.270 --> 01:06:44.490 align:middle line:84% That person C works their 10 weeks. 01:06:44.490 --> 01:06:46.150 align:middle line:84% Then they say they were laid off. 01:06:46.150 --> 01:06:52.110 align:middle line:84% When you put this together, you have created a firm that is 01:06:52.110 --> 01:07:00.830 align:middle line:84% earning $141,000 on a actual product of $40,000. 01:07:00.830 --> 01:07:06.430 align:middle line:84% And there is no experience rating in Canada, none. 01:07:06.430 --> 01:07:10.565 align:middle line:84% So you've just magically ripped off the government for $100,000. 01:07:10.565 --> 01:07:12.690 align:middle line:84% That's the problem with imperfect experience rating 01:07:12.690 --> 01:07:15.370 align:middle line:84% or, in this case, no experience rating is it subsidizes firms. 01:07:15.370 --> 01:07:17.570 align:middle line:90% This firm should not exist. 01:07:17.570 --> 01:07:18.070 align:middle line:90% OK? 01:07:18.070 --> 01:07:20.760 align:middle line:84% This firm-- as long as those workers having something they 01:07:20.760 --> 01:07:23.320 align:middle line:84% can do elsewhere that earns more than $8,000 a year, 01:07:23.320 --> 01:07:24.658 align:middle line:90% this firm should not exist. 01:07:24.658 --> 01:07:26.700 align:middle line:84% But it does because the government subsidizes it. 01:07:26.700 --> 01:07:29.400 align:middle line:84% That is the cost of imperfect experience rating. 01:07:29.400 --> 01:07:33.280 align:middle line:84% The benefit is it offsets credit market imperfections. 01:07:33.280 --> 01:07:37.640 align:middle line:84% The cost is that it subsidizes inefficient firms. 01:07:37.640 --> 01:07:38.680 align:middle line:90% Yeah? 01:07:38.680 --> 01:07:41.440 align:middle line:84% AUDIENCE: [INAUDIBLE] a couple of times about [INAUDIBLE] 01:07:41.440 --> 01:07:44.080 align:middle line:84% are the biggest obstacles between-- or what 01:07:44.080 --> 01:07:46.360 align:middle line:84% are the biggest separators between the current credit 01:07:46.360 --> 01:07:47.320 align:middle line:90% market [INAUDIBLE]? 01:07:47.320 --> 01:07:48.820 align:middle line:84% JON GRUBER: And so the biggest thing 01:07:48.820 --> 01:07:51.600 align:middle line:84% is, basically, it's essentially-- 01:07:51.600 --> 01:07:53.760 align:middle line:90% it's asymmetric information. 01:07:53.760 --> 01:07:55.580 align:middle line:84% So in a world of perfect information, 01:07:55.580 --> 01:07:58.220 align:middle line:84% I could go to you, Steven Bank, and I could say, 01:07:58.220 --> 01:08:00.880 align:middle line:90% hey, look, here's my firm. 01:08:00.880 --> 01:08:04.720 align:middle line:84% We basically can show you that we'll be viable again 01:08:04.720 --> 01:08:06.360 align:middle line:90% in two years, loan me money. 01:08:06.360 --> 01:08:07.800 align:middle line:90% And the bank would. 01:08:07.800 --> 01:08:09.820 align:middle line:84% First of all, that information is cooked. 01:08:09.820 --> 01:08:11.988 align:middle line:84% They can't trust-- you can't trust them. 01:08:11.988 --> 01:08:14.280 align:middle line:84% Second of all, you may disagree on your interpretations 01:08:14.280 --> 01:08:15.045 align:middle line:90% of that data. 01:08:15.045 --> 01:08:16.420 align:middle line:84% So even with perfect information, 01:08:16.420 --> 01:08:17.878 align:middle line:84% there could be bargaining problems. 01:08:17.878 --> 01:08:20.170 align:middle line:84% So for those reasons, the credit market is not perfect, 01:08:20.170 --> 01:08:22.870 align:middle line:84% both information imperfections and bargaining problems. 01:08:22.870 --> 01:08:24.578 align:middle line:84% The reason Coase doesn't work, these kind 01:08:24.578 --> 01:08:27.490 align:middle line:84% of bargaining failures could apply here, as well. 01:08:27.490 --> 01:08:30.210 align:middle line:84% But it's very important to come back to lecture 1. 01:08:30.210 --> 01:08:32.090 align:middle line:84% With perfect credit markets, there 01:08:32.090 --> 01:08:34.512 align:middle line:84% is no argument for imperfect experience rating. 01:08:34.512 --> 01:08:36.470 align:middle line:84% Remember, the argument for imperfect-- in fact, 01:08:36.470 --> 01:08:37.569 align:middle line:90% there's not really-- 01:08:37.569 --> 01:08:38.850 align:middle line:84% basically, firms should just-- there's 01:08:38.850 --> 01:08:40.970 align:middle line:84% not really that much of an argument for UI necessarily. 01:08:40.970 --> 01:08:41.890 align:middle line:84% I mean, there's still argument for UI 01:08:41.890 --> 01:08:43.490 align:middle line:90% because workers switch firms. 01:08:43.490 --> 01:08:45.948 align:middle line:84% But there's not an argument for imperfect experience rating 01:08:45.948 --> 01:08:47.250 align:middle line:90% with perfect credit markets. 01:08:47.250 --> 01:08:48.719 align:middle line:90% OK? 01:08:48.719 --> 01:08:52.490 align:middle line:84% Now, the last thing I want to cover in the last 10 minutes 01:08:52.490 --> 01:08:57.649 align:middle line:84% is what does this all mean for how we set up these programs? 01:08:57.649 --> 01:09:01.330 align:middle line:84% We talked about what it means for the benefit level. 01:09:01.330 --> 01:09:05.330 align:middle line:84% But these programs are more than their benefit levels. 01:09:05.330 --> 01:09:05.890 align:middle line:90% OK? 01:09:05.890 --> 01:09:07.950 align:middle line:84% These programs have lots of aspects. 01:09:07.950 --> 01:09:13.810 align:middle line:84% So for example, one thing that these programs differ in-- 01:09:13.810 --> 01:09:16.590 align:middle line:84% could differ in is how well they're targeted. 01:09:16.590 --> 01:09:21.010 align:middle line:84% You could move from one uniform program to one that's different. 01:09:21.010 --> 01:09:22.550 align:middle line:84% So for example, in Europe the way 01:09:22.550 --> 01:09:26.670 align:middle line:84% disability insurance works is there's two levels of benefits. 01:09:26.670 --> 01:09:30.130 align:middle line:84% Essentially, they adjudicate how disabled, you are. 01:09:30.130 --> 01:09:32.630 align:middle line:84% And if you're like a no-brainer, you get a higher benefit 01:09:32.630 --> 01:09:34.510 align:middle line:90% than if they're not sure. 01:09:34.510 --> 01:09:36.950 align:middle line:84% They're trying to target it more. 01:09:36.950 --> 01:09:39.847 align:middle line:84% You can imagine a similar thing in unemployment insurance. 01:09:39.847 --> 01:09:41.430 align:middle line:84% Turns out there's really fundamentally 01:09:41.430 --> 01:09:44.950 align:middle line:84% two types of unemployed people, those that are temporarily 01:09:44.950 --> 01:09:48.090 align:middle line:84% laid off and those that are truly permanently laid off. 01:09:48.090 --> 01:09:50.390 align:middle line:84% A temporary layoff is like my aunt in Minnesota, 01:09:50.390 --> 01:09:52.202 align:middle line:84% who every summer her firm said, we're 01:09:52.202 --> 01:09:53.910 align:middle line:84% going to shut the plant for three months, 01:09:53.910 --> 01:09:56.510 align:middle line:84% and you're coming back in the fall. 01:09:56.510 --> 01:09:59.430 align:middle line:84% For her, there was relatively little consumption smoothing 01:09:59.430 --> 01:10:00.697 align:middle line:90% from UI. 01:10:00.697 --> 01:10:02.030 align:middle line:90% Indeed, I showed in my article-- 01:10:02.030 --> 01:10:04.370 align:middle line:84% I ran those regressions-- that for temporary layoffs, 01:10:04.370 --> 01:10:07.070 align:middle line:90% that coefficient was like 0. 01:10:07.070 --> 01:10:09.270 align:middle line:84% For permanent layoffs, where you're not sure 01:10:09.270 --> 01:10:11.030 align:middle line:84% when you're coming back, self-insurance 01:10:11.030 --> 01:10:12.170 align:middle line:90% doesn't work as well. 01:10:12.170 --> 01:10:14.870 align:middle line:84% And for permanent layoffs, that coefficient was bigger. 01:10:14.870 --> 01:10:17.920 align:middle line:84% So you could target the generosity of UI benefits 01:10:17.920 --> 01:10:19.960 align:middle line:90% to the type of layoff. 01:10:19.960 --> 01:10:23.200 align:middle line:84% The point is can exploit our fundamental lesson from chapter 01:10:23.200 --> 01:10:27.280 align:middle line:84% 12 to try to find places where self-insurance is poor 01:10:27.280 --> 01:10:29.760 align:middle line:84% and moral hazard is low to really target 01:10:29.760 --> 01:10:31.320 align:middle line:84% the most generous benefits, giving 01:10:31.320 --> 01:10:33.240 align:middle line:90% the same thing to everyone. 01:10:33.240 --> 01:10:36.200 align:middle line:84% So that's one kind of thing besides worrying about benefit 01:10:36.200 --> 01:10:39.120 align:middle line:90% generosity we can do. 01:10:39.120 --> 01:10:42.600 align:middle line:84% Another thing we could do is we could say, wait a second. 01:10:42.600 --> 01:10:46.360 align:middle line:90% Could workers insure themselves? 01:10:46.360 --> 01:10:49.200 align:middle line:90% Imagine the following UI system. 01:10:49.200 --> 01:10:54.520 align:middle line:84% Basically, you were required, instead of paying UI tax, 01:10:54.520 --> 01:10:56.200 align:middle line:90% to have a fund. 01:10:56.200 --> 01:10:57.860 align:middle line:84% And you put money into that fund. 01:10:57.860 --> 01:11:00.770 align:middle line:84% And when you lose your job, you draw down from that fund. 01:11:00.770 --> 01:11:02.020 align:middle line:90% Now, the government helps out. 01:11:02.020 --> 01:11:04.420 align:middle line:84% If you lose your job before you've been working too long, 01:11:04.420 --> 01:11:05.003 align:middle line:90% they can help. 01:11:05.003 --> 01:11:06.180 align:middle line:90% They can top off your fund. 01:11:06.180 --> 01:11:08.140 align:middle line:84% If you stay unemployed too long, they can top off your fund, 01:11:08.140 --> 01:11:08.720 align:middle line:90% whatever. 01:11:08.720 --> 01:11:11.740 align:middle line:84% But the point is, why not make you absorb the risk? 01:11:11.740 --> 01:11:13.640 align:middle line:90% Why do we want to do that? 01:11:13.640 --> 01:11:16.187 align:middle line:84% Well, essentially, it's like a Samaritan's Dilemma problem. 01:11:16.187 --> 01:11:18.520 align:middle line:84% If you lose your job, we're going to be helping you out. 01:11:18.520 --> 01:11:21.540 align:middle line:84% So we're going to force you to self-insure by saving. 01:11:21.540 --> 01:11:23.260 align:middle line:84% Rather than taking some payroll tax 01:11:23.260 --> 01:11:24.540 align:middle line:84% and throwing to the government, we're 01:11:24.540 --> 01:11:25.873 align:middle line:90% going to force you to self-fund. 01:11:25.873 --> 01:11:27.980 align:middle line:84% It's like privatized Social Security. 01:11:27.980 --> 01:11:29.480 align:middle line:84% Whereas privatized Social Security 01:11:29.480 --> 01:11:30.520 align:middle line:84% had a bunch of problems, like the fact, 01:11:30.520 --> 01:11:32.220 align:middle line:84% we have this legacy debt, those problems 01:11:32.220 --> 01:11:34.500 align:middle line:90% don't exist for privatized UI. 01:11:34.500 --> 01:11:36.820 align:middle line:84% There's no reason we couldn't do that, we couldn't have 01:11:36.820 --> 01:11:38.460 align:middle line:84% a system where basically people save for themselves 01:11:38.460 --> 01:11:39.860 align:middle line:90% and their own accounts. 01:11:39.860 --> 01:11:41.480 align:middle line:90% And why would we want that? 01:11:41.480 --> 01:11:43.380 align:middle line:90% Because moral hazard goes away. 01:11:43.380 --> 01:11:47.140 align:middle line:84% If it's my own money, I don't suffer moral hazard. 01:11:47.140 --> 01:11:50.040 align:middle line:84% One way to see that is a really cool experiment 01:11:50.040 --> 01:11:53.260 align:middle line:84% they ran where they ran an experiment where 01:11:53.260 --> 01:11:56.420 align:middle line:84% instead of saying to people, you lost your job. 01:11:56.420 --> 01:12:00.220 align:middle line:84% You're eligible for up to 20 weeks of $500 a week. 01:12:00.220 --> 01:12:01.640 align:middle line:90% We said, you lost your job. 01:12:01.640 --> 01:12:04.340 align:middle line:84% Here's a check, a lump sum check that 01:12:04.340 --> 01:12:07.420 align:middle line:84% approximates the typical amount of money you get from UI. 01:12:07.420 --> 01:12:10.140 align:middle line:84% People found new jobs right away. 01:12:10.140 --> 01:12:15.070 align:middle line:84% So basically we could think about moving towards actually 01:12:15.070 --> 01:12:17.390 align:middle line:84% fighting moral hazard, but still providing insurance 01:12:17.390 --> 01:12:20.230 align:middle line:84% by basically forcing people to self-insure. 01:12:20.230 --> 01:12:21.733 align:middle line:84% That's one direction we could go. 01:12:21.733 --> 01:12:23.650 align:middle line:84% So that's an interesting thing to think about. 01:12:23.650 --> 01:12:25.733 align:middle line:84% The reason to think about-- the reason [INAUDIBLE] 01:12:25.733 --> 01:12:28.530 align:middle line:84% to think about is that-- a question was raised earlier. 01:12:28.530 --> 01:12:31.070 align:middle line:84% We're moving into a world where traditional employment 01:12:31.070 --> 01:12:32.550 align:middle line:90% relationships are eroding. 01:12:32.550 --> 01:12:35.350 align:middle line:84% How do you think about unemployment for an Uber driver? 01:12:35.350 --> 01:12:36.770 align:middle line:84% When is an Uber driver unemployed? 01:12:36.770 --> 01:12:39.770 align:middle line:84% Is it when people don't want to go from place A to place B? 01:12:39.770 --> 01:12:41.860 align:middle line:90% Then they're unemployed, right? 01:12:41.860 --> 01:12:46.070 align:middle line:84% So it's hard to define unemployment for many gig worker 01:12:46.070 --> 01:12:47.550 align:middle line:90% specialties. 01:12:47.550 --> 01:12:50.550 align:middle line:90% So UI doesn't really work. 01:12:50.550 --> 01:12:54.150 align:middle line:84% And this was a fundamental problem we faced during COVID. 01:12:54.150 --> 01:12:55.582 align:middle line:90% COVID hits. 01:12:55.582 --> 01:12:56.290 align:middle line:90% What do you have? 01:12:56.290 --> 01:12:59.950 align:middle line:84% You have a system which has two flaws. 01:12:59.950 --> 01:13:03.022 align:middle line:84% First of all, it's not generous enough. 01:13:03.022 --> 01:13:04.730 align:middle line:84% And what I mean by that is in two senses. 01:13:04.730 --> 01:13:06.970 align:middle line:84% First of all, for some low-income workers, 01:13:06.970 --> 01:13:09.190 align:middle line:84% UI wasn't enough to pay their bills 01:13:09.190 --> 01:13:12.200 align:middle line:84% if they were going to be unemployed for a long time. 01:13:12.200 --> 01:13:16.160 align:middle line:84% Second of all, the second of which wasn't generous enough 01:13:16.160 --> 01:13:18.410 align:middle line:90% is the sense in which-- 01:13:18.410 --> 01:13:19.660 align:middle line:90% so first of all, let me think. 01:13:19.660 --> 01:13:20.160 align:middle line:90% There's two. 01:13:20.160 --> 01:13:22.633 align:middle line:84% So one is it wasn't big enough to pay the bills. 01:13:22.633 --> 01:13:24.300 align:middle line:84% What was the other thing I was thinking? 01:13:24.300 --> 01:13:25.008 align:middle line:90% I don't remember. 01:13:25.008 --> 01:13:27.020 align:middle line:84% Anyway, so it's viewed as not generous enough, 01:13:27.020 --> 01:13:29.480 align:middle line:84% not big enough to pay people's bills. 01:13:29.480 --> 01:13:32.120 align:middle line:84% The second problem is that it's not 01:13:32.120 --> 01:13:35.400 align:middle line:84% designed for all these people who lost their gig jobs. 01:13:35.400 --> 01:13:38.840 align:middle line:84% So we set up two temporary programs during COVID 01:13:38.840 --> 01:13:40.920 align:middle line:84% to try to address these problems. 01:13:40.920 --> 01:13:44.660 align:middle line:84% The first was we gave everyone in America, 01:13:44.660 --> 01:13:47.240 align:middle line:84% in addition to their usual UI benefits-- oh, I 01:13:47.240 --> 01:13:48.620 align:middle line:84% know what the other reason we wanted generous was. 01:13:48.620 --> 01:13:49.120 align:middle line:90% I remember. 01:13:49.120 --> 01:13:50.813 align:middle line:84% We didn't want people to go to work. 01:13:50.813 --> 01:13:51.980 align:middle line:90% We wanted them to stay home. 01:13:51.980 --> 01:13:54.480 align:middle line:90% It wasn't moral hazard anymore. 01:13:54.480 --> 01:13:58.063 align:middle line:84% Indeed, this is a key insight I'm diverting here 01:13:58.063 --> 01:13:58.980 align:middle line:90% to go back to nothing. 01:13:58.980 --> 01:14:01.440 align:middle line:90% We think about durations. 01:14:01.440 --> 01:14:05.400 align:middle line:84% UI should be more generous when the economy is bad. 01:14:05.400 --> 01:14:06.240 align:middle line:90% Why? 01:14:06.240 --> 01:14:08.880 align:middle line:84% Because moral hazard is smaller and consumption smoothing 01:14:08.880 --> 01:14:09.940 align:middle line:90% is bigger. 01:14:09.940 --> 01:14:12.040 align:middle line:84% Think about a good economy versus a bad economy. 01:14:12.040 --> 01:14:15.620 align:middle line:84% In a good economy, you can always find a new job. 01:14:15.620 --> 01:14:17.980 align:middle line:84% So moral hazard could be bad because you're turning down 01:14:17.980 --> 01:14:18.940 align:middle line:90% those opportunities. 01:14:18.940 --> 01:14:21.007 align:middle line:84% And you have much better access to self-insurance 01:14:21.007 --> 01:14:22.840 align:middle line:84% because your friends have money to loan you. 01:14:22.840 --> 01:14:24.220 align:middle line:90% Your spouse can work. 01:14:24.220 --> 01:14:26.413 align:middle line:84% In a bad economy, you can't find the jobs. 01:14:26.413 --> 01:14:28.580 align:middle line:84% So there's no moral hazard if there's no job to get, 01:14:28.580 --> 01:14:31.260 align:middle line:84% and your spouse can't self-insure you. 01:14:31.260 --> 01:14:32.780 align:middle line:84% So what should happen is UI should 01:14:32.780 --> 01:14:38.780 align:middle line:84% get more generous in bad times, less generous than good times. 01:14:38.780 --> 01:14:40.997 align:middle line:84% With COVID and extra piece, well, it was a bad time. 01:14:40.997 --> 01:14:42.580 align:middle line:84% We didn't actually want people at work 01:14:42.580 --> 01:14:44.700 align:middle line:90% because it was contagious. 01:14:44.700 --> 01:14:48.093 align:middle line:84% For that reason, we wanted a much more generous UI system. 01:14:48.093 --> 01:14:50.260 align:middle line:84% Now, in a perfect world, you'd say, well, let's just 01:14:50.260 --> 01:14:52.420 align:middle line:90% raise the replacement rate. 01:14:52.420 --> 01:14:54.620 align:middle line:84% Let's say, well, instead of we think 01:14:54.620 --> 01:14:59.240 align:middle line:84% 44% is not enough in bad times, let's raise it to 80%, 90%, 01:14:59.240 --> 01:15:01.020 align:middle line:90% whatever. 01:15:01.020 --> 01:15:02.860 align:middle line:84% The problem-- what we did instead 01:15:02.860 --> 01:15:07.620 align:middle line:84% was say everyone who gets UI gets a flat $600 as well. 01:15:07.620 --> 01:15:08.690 align:middle line:90% Why is that different? 01:15:08.690 --> 01:15:10.910 align:middle line:84% Well, that's different because $600 means very different things 01:15:10.910 --> 01:15:11.870 align:middle line:90% to different people. 01:15:11.870 --> 01:15:14.950 align:middle line:84% For someone very low-income, many people in America 01:15:14.950 --> 01:15:17.510 align:middle line:90% replace rates well above 100%. 01:15:17.510 --> 01:15:19.830 align:middle line:84% Because $600 plus UI benefit is well above what 01:15:19.830 --> 01:15:21.110 align:middle line:90% they're earning beforehand. 01:15:21.110 --> 01:15:23.210 align:middle line:90% That's never good. 01:15:23.210 --> 01:15:24.690 align:middle line:84% It's never a good idea to provide-- 01:15:24.690 --> 01:15:27.710 align:middle line:84% I mean, it's basically you don't-- you're already smoothing 01:15:27.710 --> 01:15:31.595 align:middle line:84% the consumption, and you're just adding extra moral hazard. 01:15:31.595 --> 01:15:32.970 align:middle line:84% Whereas some people who are rich, 01:15:32.970 --> 01:15:35.750 align:middle line:84% it wasn't enough to actually let them still pay their mortgage. 01:15:35.750 --> 01:15:38.750 align:middle line:84% So why didn't we just raise replacement rates? 01:15:38.750 --> 01:15:41.118 align:middle line:84% Well, this comes back to chapter 9, 01:15:41.118 --> 01:15:42.910 align:middle line:84% because the administrative state in America 01:15:42.910 --> 01:15:44.230 align:middle line:90% is starved for resources. 01:15:44.230 --> 01:15:47.910 align:middle line:84% Because basically of opposition to government spending, 01:15:47.910 --> 01:15:51.950 align:middle line:84% many government agencies are not up to doing what they should do. 01:15:51.950 --> 01:15:54.350 align:middle line:84% The UI computer systems-- have any of you guys 01:15:54.350 --> 01:15:58.110 align:middle line:84% ever heard of a programming language called COBOL 01:15:58.110 --> 01:16:00.870 align:middle line:90% that existed when I was a kid? 01:16:00.870 --> 01:16:01.430 align:middle line:90% OK. 01:16:01.430 --> 01:16:04.750 align:middle line:84% UI system these days are still written in COBOL. 01:16:04.750 --> 01:16:08.025 align:middle line:84% Basically, these systems have not been kept up. 01:16:08.025 --> 01:16:09.650 align:middle line:84% And the state administrator said, look, 01:16:09.650 --> 01:16:11.070 align:middle line:84% you want us to get money people right away. 01:16:11.070 --> 01:16:12.210 align:middle line:84% It's going to take us like three years 01:16:12.210 --> 01:16:14.470 align:middle line:84% to fix our systems so we can change replacement rates. 01:16:14.470 --> 01:16:16.650 align:middle line:84% Let's just send out a flat check. 01:16:16.650 --> 01:16:19.690 align:middle line:84% So this is a cost to society of not investing 01:16:19.690 --> 01:16:22.570 align:middle line:90% government resources OK? 01:16:22.570 --> 01:16:25.780 align:middle line:84% The other thing they did was they said, look, 01:16:25.780 --> 01:16:28.030 align:middle line:84% we will not require you to have had a traditional job. 01:16:28.030 --> 01:16:31.650 align:middle line:84% If you say you had a job and you lost it, we'll send you a check. 01:16:31.650 --> 01:16:33.390 align:middle line:84% My daughter is a creative writer. 01:16:33.390 --> 01:16:34.470 align:middle line:90% She writes online. 01:16:34.470 --> 01:16:36.190 align:middle line:84% She doesn't have-- she's a gig worker. 01:16:36.190 --> 01:16:38.190 align:middle line:84% She just said, look, this is what I was making-- 01:16:38.190 --> 01:16:40.690 align:middle line:84% she self-asserted what she was making beforehand. 01:16:40.690 --> 01:16:43.138 align:middle line:84% She didn't lie, I'm proud to say-- self-asserted 01:16:43.138 --> 01:16:44.430 align:middle line:90% what she was making beforehand. 01:16:44.430 --> 01:16:46.513 align:middle line:84% They said, great, we'll send you x percent of that 01:16:46.513 --> 01:16:47.490 align:middle line:90% plus $600 a week. 01:16:47.490 --> 01:16:49.210 align:middle line:90% Other people did lie. 01:16:49.210 --> 01:16:52.670 align:middle line:84% And this led to massive fraud in the UI system. 01:16:52.670 --> 01:16:54.395 align:middle line:84% Once again, we need a better system. 01:16:54.395 --> 01:16:56.770 align:middle line:84% We need to advance our systems to deal with this changing 01:16:56.770 --> 01:16:58.690 align:middle line:90% nature of work in the economy. 01:16:58.690 --> 01:16:59.530 align:middle line:90% All right. 01:16:59.530 --> 01:17:02.320 align:middle line:90% So that is a lot to absorb. 01:17:02.320 --> 01:17:06.000 align:middle line:90%