WEBVTT 00:00:00.000 --> 00:00:00.500 align:middle line:90% 00:00:00.500 --> 00:00:01.922 align:middle line:90% [SQUEAKING] 00:00:01.922 --> 00:00:04.292 align:middle line:90% [RUSTLING] 00:00:04.292 --> 00:00:05.714 align:middle line:90% [CLICKING] 00:00:05.714 --> 00:00:09.545 align:middle line:90% 00:00:09.545 --> 00:00:10.920 align:middle line:84% JON GRUBER: So today, we're going 00:00:10.920 --> 00:00:12.587 align:middle line:84% to continue our discussion of chapter 8. 00:00:12.587 --> 00:00:15.280 align:middle line:84% Remember, last time, we went through the costs of we 00:00:15.280 --> 00:00:16.840 align:middle line:90% have a highway project. 00:00:16.840 --> 00:00:19.480 align:middle line:84% We talked about the costs in terms of labor, 00:00:19.480 --> 00:00:22.222 align:middle line:84% in terms of asphalt, and in terms of maintenance. 00:00:22.222 --> 00:00:23.680 align:middle line:84% We talked about the issues involved 00:00:23.680 --> 00:00:27.040 align:middle line:84% in measuring those costs and came up with the total cost. 00:00:27.040 --> 00:00:28.540 align:middle line:84% Then we turned to the benefits side. 00:00:28.540 --> 00:00:30.480 align:middle line:84% And there's two types of benefits. 00:00:30.480 --> 00:00:34.902 align:middle line:84% There's valuing time savings and valuing lives saved. 00:00:34.902 --> 00:00:36.360 align:middle line:84% For each of these benefits, there's 00:00:36.360 --> 00:00:39.720 align:middle line:84% fundamentally three methods that are 00:00:39.720 --> 00:00:43.560 align:middle line:84% used to value the benefits, three methods one can use. 00:00:43.560 --> 00:00:46.460 align:middle line:84% The first method is what we call the market value method. 00:00:46.460 --> 00:00:49.000 align:middle line:90% 00:00:49.000 --> 00:00:52.640 align:middle line:84% And that's basically using a market-revealed price 00:00:52.640 --> 00:00:55.100 align:middle line:84% to measure the value of something. 00:00:55.100 --> 00:00:58.100 align:middle line:84% So for time, the market-revealed price is wages. 00:00:58.100 --> 00:00:59.890 align:middle line:84% Remember, I talked about the fact 00:00:59.890 --> 00:01:04.519 align:middle line:84% that in a standard, no friction, standard economic model 00:01:04.519 --> 00:01:08.010 align:middle line:84% of the labor market, the wage is the appropriate measure 00:01:08.010 --> 00:01:10.050 align:middle line:90% of people's time. 00:01:10.050 --> 00:01:11.890 align:middle line:84% So in that world, we can use the wage 00:01:11.890 --> 00:01:14.137 align:middle line:84% to measure what time gained is worth. 00:01:14.137 --> 00:01:15.970 align:middle line:84% We then talked about why that may or may not 00:01:15.970 --> 00:01:17.330 align:middle line:90% be a good measure. 00:01:17.330 --> 00:01:19.337 align:middle line:90% So wage has pros and cons. 00:01:19.337 --> 00:01:21.170 align:middle line:84% Remember, it might be too big or too little, 00:01:21.170 --> 00:01:23.570 align:middle line:84% depending on various imperfections in the labor 00:01:23.570 --> 00:01:24.190 align:middle line:90% market. 00:01:24.190 --> 00:01:26.750 align:middle line:84% So in a perfect labor market with no imperfections, 00:01:26.750 --> 00:01:29.330 align:middle line:84% the wage would be the appropriate measure of time. 00:01:29.330 --> 00:01:33.530 align:middle line:84% In a labor market with imperfections, that may not be. 00:01:33.530 --> 00:01:36.170 align:middle line:84% The second approach we use is an approach 00:01:36.170 --> 00:01:40.450 align:middle line:90% we call contingent valuation. 00:01:40.450 --> 00:01:42.870 align:middle line:84% Now, this is-- once again, I've mentioned this before, 00:01:42.870 --> 00:01:44.890 align:middle line:84% but part of what you get from this course 00:01:44.890 --> 00:01:46.070 align:middle line:90% is a few trade secrets. 00:01:46.070 --> 00:01:48.308 align:middle line:84% And one of the trade secrets in economics 00:01:48.308 --> 00:01:50.850 align:middle line:84% is we make a lot of money by making up fancy terms for things 00:01:50.850 --> 00:01:53.110 align:middle line:90% which are really not very hard. 00:01:53.110 --> 00:01:56.450 align:middle line:84% Contingent valuation means asking people, 00:01:56.450 --> 00:01:59.410 align:middle line:84% asking people what stuff is worth. 00:01:59.410 --> 00:02:01.890 align:middle line:84% And that is basically saying, look, what's it worth to you 00:02:01.890 --> 00:02:04.250 align:middle line:84% to save an hour, and doing a survey 00:02:04.250 --> 00:02:06.930 align:middle line:84% and ask people what it's worth to save an hour. 00:02:06.930 --> 00:02:09.328 align:middle line:84% That seems like the standard place one would start. 00:02:09.328 --> 00:02:11.870 align:middle line:84% If I want to know an hour of time is worth, you'd ask people. 00:02:11.870 --> 00:02:14.250 align:middle line:84% The problem is people have no idea, 00:02:14.250 --> 00:02:16.350 align:middle line:84% that it's a very hard thing to know how to answer. 00:02:16.350 --> 00:02:18.515 align:middle line:90% What's an hour of my time worth? 00:02:18.515 --> 00:02:19.890 align:middle line:84% And it gets even more complicated 00:02:19.890 --> 00:02:21.810 align:middle line:84% when you ask them harder questions, 00:02:21.810 --> 00:02:26.410 align:middle line:84% like, how do we value various other potential benefits 00:02:26.410 --> 00:02:28.990 align:middle line:84% of regulation or damage for environmental considerations? 00:02:28.990 --> 00:02:30.448 align:middle line:84% There's a great article which talks 00:02:30.448 --> 00:02:32.742 align:middle line:84% about all the problems you have when you 00:02:32.742 --> 00:02:33.950 align:middle line:90% ask people their preferences. 00:02:33.950 --> 00:02:39.810 align:middle line:84% So for example, one problem is, whether a question 00:02:39.810 --> 00:02:42.030 align:middle line:84% is asked isolated or with other questions matter. 00:02:42.030 --> 00:02:44.090 align:middle line:84% So if you ask people how much they'd 00:02:44.090 --> 00:02:47.210 align:middle line:84% be willing to pay to improve the visibility of the Grand Canyon, 00:02:47.210 --> 00:02:50.770 align:middle line:84% if you ask them solely-- if you only ask that question 00:02:50.770 --> 00:02:53.090 align:middle line:84% and then ask them as part of a series of questions 00:02:53.090 --> 00:02:55.970 align:middle line:84% or randomly asked some people only the question-- other people 00:02:55.970 --> 00:02:56.880 align:middle line:90% ask five questions. 00:02:56.880 --> 00:02:58.100 align:middle line:90% That's the third. 00:02:58.100 --> 00:03:00.460 align:middle line:84% The second group values it one fifth as much 00:03:00.460 --> 00:03:03.303 align:middle line:84% as the first group, even though it's random. 00:03:03.303 --> 00:03:05.720 align:middle line:84% So when I'm asking the list, it makes the Grand Canyon one 00:03:05.720 --> 00:03:08.940 align:middle line:84% fifth as valuable, which is not plausible. 00:03:08.940 --> 00:03:10.760 align:middle line:84% There's also the order of issues matter. 00:03:10.760 --> 00:03:15.890 align:middle line:84% So if you ask people if they'd rather save whales or seals, 00:03:15.890 --> 00:03:19.620 align:middle line:84% if you ask about seals and then whales, 00:03:19.620 --> 00:03:23.940 align:middle line:84% you find that seals are worth about the same 00:03:23.940 --> 00:03:27.740 align:middle line:84% as whales, about 142 to save the seals and 195 to save 00:03:27.740 --> 00:03:28.240 align:middle line:90% the whales. 00:03:28.240 --> 00:03:30.900 align:middle line:84% So whales are worth about 25% more than seals. 00:03:30.900 --> 00:03:32.442 align:middle line:84% If you ask about the whales first, 00:03:32.442 --> 00:03:34.900 align:middle line:84% now seals are worth twice as much as-- now whales are worth 00:03:34.900 --> 00:03:36.020 align:middle line:90% twice as much as seals-- 00:03:36.020 --> 00:03:38.120 align:middle line:90% 172 to $85. 00:03:38.120 --> 00:03:39.580 align:middle line:90% So the order you ask matters. 00:03:39.580 --> 00:03:41.740 align:middle line:84% And then finally, there's something 00:03:41.740 --> 00:03:43.500 align:middle line:84% called the embedding effect, which 00:03:43.500 --> 00:03:45.708 align:middle line:84% is people have a hard time thinking about magnitudes. 00:03:45.708 --> 00:03:49.340 align:middle line:84% So if you ask people randomly, what is it worth to save 2,000 00:03:49.340 --> 00:03:53.040 align:middle line:84% birds, 20,000 birds, 200,000 birds, or 2 million birds, 00:03:53.040 --> 00:03:54.910 align:middle line:90% they give the same answer. 00:03:54.910 --> 00:03:57.493 align:middle line:84% It's because they don't know how to-- and it's not surprising. 00:03:57.493 --> 00:03:59.702 align:middle line:84% It's hard to think about what these things are worth. 00:03:59.702 --> 00:04:02.270 align:middle line:84% As a result, it's hard to use contingent valuation methods 00:04:02.270 --> 00:04:03.990 align:middle line:90% to value things. 00:04:03.990 --> 00:04:07.870 align:middle line:84% And that's why economists typically 00:04:07.870 --> 00:04:12.710 align:middle line:84% prefer a third approach, which we call the revealed preference 00:04:12.710 --> 00:04:20.372 align:middle line:84% approach, revealed preference, which is basically saying, 00:04:20.372 --> 00:04:22.330 align:middle line:84% people might not know how to answer a question, 00:04:22.330 --> 00:04:26.632 align:middle line:84% but their actions betray their true valuation. 00:04:26.632 --> 00:04:28.090 align:middle line:84% The fundamental rule of economics-- 00:04:28.090 --> 00:04:30.673 align:middle line:84% you want to know something about people, look at what they do, 00:04:30.673 --> 00:04:31.630 align:middle line:90% not what they say. 00:04:31.630 --> 00:04:33.630 align:middle line:84% And revealed preference methods are 00:04:33.630 --> 00:04:36.790 align:middle line:84% ways of using people's actions to value 00:04:36.790 --> 00:04:38.090 align:middle line:90% how much they value things. 00:04:38.090 --> 00:04:39.703 align:middle line:90% So for example, someone tell me. 00:04:39.703 --> 00:04:40.870 align:middle line:90% Raise your hand and tell me. 00:04:40.870 --> 00:04:43.412 align:middle line:84% If what you think about valuing how much an hour of your time 00:04:43.412 --> 00:04:46.590 align:middle line:84% is worth, how could I use a revealed behavior 00:04:46.590 --> 00:04:49.150 align:middle line:84% to tell me how much an hour of your time is worth? 00:04:49.150 --> 00:04:50.167 align:middle line:90% What could I do? 00:04:50.167 --> 00:04:52.250 align:middle line:84% I want to know what an hour of your time is worth. 00:04:52.250 --> 00:04:53.787 align:middle line:84% I see you do lots of things in life. 00:04:53.787 --> 00:04:54.620 align:middle line:90% I track you all day. 00:04:54.620 --> 00:04:56.160 align:middle line:90% I see everything you do. 00:04:56.160 --> 00:04:58.063 align:middle line:84% How could I use your decisions every day 00:04:58.063 --> 00:04:59.980 align:middle line:84% to tell me what an hour of your time is worth? 00:04:59.980 --> 00:05:04.200 align:middle line:90% 00:05:04.200 --> 00:05:05.185 align:middle line:90% Yeah? 00:05:05.185 --> 00:05:07.060 align:middle line:84% AUDIENCE: Without an additional introduction, 00:05:07.060 --> 00:05:09.880 align:middle line:84% you could see whether they use something like a grocery 00:05:09.880 --> 00:05:12.350 align:middle line:90% delivery app or a-- 00:05:12.350 --> 00:05:13.100 align:middle line:90% JON GRUBER: Great. 00:05:13.100 --> 00:05:14.140 align:middle line:90% That's a great idea. 00:05:14.140 --> 00:05:17.320 align:middle line:84% If you use a grocery delivery app, you pay $5 more, 00:05:17.320 --> 00:05:19.840 align:middle line:90% and you save X minutes. 00:05:19.840 --> 00:05:21.320 align:middle line:84% So you could look at, how much more 00:05:21.320 --> 00:05:23.360 align:middle line:84% are you willing to pay to save X minutes? 00:05:23.360 --> 00:05:26.000 align:middle line:84% You could look at two different ways people can drive home, 00:05:26.000 --> 00:05:28.332 align:middle line:84% one which might have a toll and one way which wouldn't 00:05:28.332 --> 00:05:29.540 align:middle line:90% have a toll but takes longer. 00:05:29.540 --> 00:05:32.840 align:middle line:84% You can actually set ways to avoid tolls. 00:05:32.840 --> 00:05:35.540 align:middle line:84% You could look at people who set the Waze in different ways-- 00:05:35.540 --> 00:05:38.160 align:middle line:84% Waze, different spellings-- and see how much 00:05:38.160 --> 00:05:40.480 align:middle line:90% they'll pay to avoid tolls. 00:05:40.480 --> 00:05:44.740 align:middle line:84% So you can look at people's decisions to try to reveal it. 00:05:44.740 --> 00:05:47.200 align:middle line:84% A classic way of doing this is looking 00:05:47.200 --> 00:05:49.800 align:middle line:84% at the most important single financial decision 00:05:49.800 --> 00:05:53.560 align:middle line:84% every American makes, which is to buy a house. 00:05:53.560 --> 00:05:56.400 align:middle line:84% Houses are worth different amounts 00:05:56.400 --> 00:05:57.860 align:middle line:90% depending on the commute. 00:05:57.860 --> 00:05:59.360 align:middle line:84% The value of the commute is embedded 00:05:59.360 --> 00:06:00.880 align:middle line:90% in the value of a house. 00:06:00.880 --> 00:06:05.565 align:middle line:84% The further you have to commute, the less a house is worth. 00:06:05.565 --> 00:06:07.440 align:middle line:84% Now, the problem, of course, is house value's 00:06:07.440 --> 00:06:10.560 align:middle line:90% determined by lots of things. 00:06:10.560 --> 00:06:13.200 align:middle line:84% So while it's true that in general, long commutes are 00:06:13.200 --> 00:06:15.240 align:middle line:84% associated with lower valued houses, 00:06:15.240 --> 00:06:17.540 align:middle line:84% it's also true that lots of other things matter. 00:06:17.540 --> 00:06:22.180 align:middle line:84% So for example, if you look at Everett, which is about-- 00:06:22.180 --> 00:06:24.260 align:middle line:84% which is about three miles from downtown Boston, 00:06:24.260 --> 00:06:28.920 align:middle line:84% most of the jobs in Boston, the average house price is $300,000. 00:06:28.920 --> 00:06:30.880 align:middle line:84% If you look at Lexington, where I live, 00:06:30.880 --> 00:06:32.820 align:middle line:84% which is three times as far away, 00:06:32.820 --> 00:06:34.632 align:middle line:84% the average house price is $1 million. 00:06:34.632 --> 00:06:36.840 align:middle line:84% So that would seem to be backwards, where three times 00:06:36.840 --> 00:06:38.460 align:middle line:84% farther away, our house prices are worth more. 00:06:38.460 --> 00:06:41.002 align:middle line:84% That's because we've got good schools and nice houses and all 00:06:41.002 --> 00:06:42.360 align:middle line:90% that stuff. 00:06:42.360 --> 00:06:44.480 align:middle line:90% So it's very hard. 00:06:44.480 --> 00:06:46.743 align:middle line:84% And likewise, all these decisions we've talked about, 00:06:46.743 --> 00:06:47.660 align:middle line:90% it's hard to use them. 00:06:47.660 --> 00:06:50.000 align:middle line:84% So the fact you use grocery delivery or not, maybe that's 00:06:50.000 --> 00:06:51.030 align:middle line:90% not about value of time. 00:06:51.030 --> 00:06:52.430 align:middle line:84% Maybe it's just you're socially awkward. 00:06:52.430 --> 00:06:53.888 align:middle line:84% You don't want to deal with people. 00:06:53.888 --> 00:06:56.730 align:middle line:84% Or the fact you drive one way or another, maybe it's just habit. 00:06:56.730 --> 00:06:59.690 align:middle line:84% So the problem with a revealed preference approach 00:06:59.690 --> 00:07:02.370 align:middle line:84% is that decisions are endogenous. 00:07:02.370 --> 00:07:05.395 align:middle line:84% If I want to try to value your time by comparing two things, 00:07:05.395 --> 00:07:06.770 align:middle line:84% I've got the problem we discussed 00:07:06.770 --> 00:07:09.650 align:middle line:84% in chapter 3 of the bias problem, 00:07:09.650 --> 00:07:12.290 align:middle line:84% that people who use Waze, who get 00:07:12.290 --> 00:07:15.250 align:middle line:84% groceries delivered and don't get delivered, aren't the same. 00:07:15.250 --> 00:07:16.630 align:middle line:90% So you can't just compare them. 00:07:16.630 --> 00:07:18.590 align:middle line:84% And people live in Lexington and Everett aren't just the same. 00:07:18.590 --> 00:07:20.210 align:middle line:90% You can't just compare them. 00:07:20.210 --> 00:07:22.690 align:middle line:84% So what we need is some kind of experimental or 00:07:22.690 --> 00:07:26.610 align:middle line:84% quasi-experimental method to look at how people value time. 00:07:26.610 --> 00:07:28.550 align:middle line:84% Probably my favorite example, in this case, 00:07:28.550 --> 00:07:32.872 align:middle line:84% for valuing time is this super cool paper that was done many 00:07:32.872 --> 00:07:35.330 align:middle line:84% years ago, actually in the mid '80s, but it's a really cool 00:07:35.330 --> 00:07:36.170 align:middle line:90% paper. 00:07:36.170 --> 00:07:39.990 align:middle line:84% What they said is, look, during the gas crisis of the 1970s-- 00:07:39.990 --> 00:07:42.890 align:middle line:84% many of you may know, in the 1970s, OPEC was formed, 00:07:42.890 --> 00:07:44.810 align:middle line:84% and gas prices went through the roof. 00:07:44.810 --> 00:07:47.690 align:middle line:84% As a result, gas prices got very high in the US. 00:07:47.690 --> 00:07:51.940 align:middle line:84% Politicians, responding to this, put a cap on gas prices 00:07:51.940 --> 00:07:54.020 align:middle line:84% and said that you could not-- gas stations 00:07:54.020 --> 00:07:55.980 align:middle line:84% could not charge beyond a certain amount. 00:07:55.980 --> 00:07:58.720 align:middle line:84% Well, we know from standard economics what that did. 00:07:58.720 --> 00:08:01.100 align:middle line:84% It created giant lines at gas stations 00:08:01.100 --> 00:08:03.378 align:middle line:84% because basically, gas was now artificially cheap 00:08:03.378 --> 00:08:04.920 align:middle line:84% because the government had capped it. 00:08:04.920 --> 00:08:07.860 align:middle line:84% So people waited in long lines to get gas. 00:08:07.860 --> 00:08:09.187 align:middle line:90% The problem with that is-- 00:08:09.187 --> 00:08:10.520 align:middle line:90% so the-- it was not the problem. 00:08:10.520 --> 00:08:12.353 align:middle line:84% The interesting fact that I didn't know when 00:08:12.353 --> 00:08:14.820 align:middle line:84% read this paper, it turned out those gas caps only applied 00:08:14.820 --> 00:08:16.580 align:middle line:90% to national chains. 00:08:16.580 --> 00:08:21.740 align:middle line:84% So your Mobils and Exxons had that cap, but Joe's Gas did not. 00:08:21.740 --> 00:08:24.220 align:middle line:84% So it turned out you had side by side, 00:08:24.220 --> 00:08:27.060 align:middle line:84% for random reasons, gas stations with cheap gas 00:08:27.060 --> 00:08:30.420 align:middle line:84% and long lines and expensive gas and short lines. 00:08:30.420 --> 00:08:32.840 align:middle line:84% And you could use that difference to value time, 00:08:32.840 --> 00:08:34.659 align:middle line:84% because essentially, you could choose. 00:08:34.659 --> 00:08:36.520 align:middle line:84% You could decide you want to go to Exxon, 00:08:36.520 --> 00:08:38.220 align:middle line:90% save money, and wait online. 00:08:38.220 --> 00:08:41.840 align:middle line:84% Or do I want to go to Joe's Gas, spend more, not wait online? 00:08:41.840 --> 00:08:43.460 align:middle line:90% And that's a true random choice. 00:08:43.460 --> 00:08:45.960 align:middle line:84% I mean, there might be some underlying preference for Exxon, 00:08:45.960 --> 00:08:47.980 align:middle line:84% but that's probably likely to be not a big deal. 00:08:47.980 --> 00:08:50.380 align:middle line:84% So that's kind of a cool quasi-experiment. 00:08:50.380 --> 00:08:51.820 align:middle line:90% So what did he found? 00:08:51.820 --> 00:08:55.180 align:middle line:84% He found that-- in today's dollars-- 00:08:55.180 --> 00:08:58.620 align:middle line:84% that you saved about $0.50 a gallon. 00:08:58.620 --> 00:09:05.220 align:middle line:84% Actually, precisely $0.54 a gallon savings-- 00:09:05.220 --> 00:09:10.380 align:middle line:84% $0.54 a gallon savings by going to Chevron, going to a chain. 00:09:10.380 --> 00:09:15.860 align:middle line:84% You waited, on average, 14.6 more minutes. 00:09:15.860 --> 00:09:21.300 align:middle line:84% So you saved 14.6 minutes to save $0.54 a gallon. 00:09:21.300 --> 00:09:30.700 align:middle line:84% And the typical purchase was 10.5 gallons. 00:09:30.700 --> 00:09:34.640 align:middle line:84% So what that meant was the typical person was waiting 14.6 00:09:34.640 --> 00:09:43.260 align:middle line:90% minutes to save $22.50. 00:09:43.260 --> 00:09:46.550 align:middle line:90% Was saving 14.6 minutes. 00:09:46.550 --> 00:09:47.310 align:middle line:90% Oh, I'm sorry. 00:09:47.310 --> 00:09:48.020 align:middle line:90% No, that's wrong. 00:09:48.020 --> 00:09:48.770 align:middle line:90% I don't mean that. 00:09:48.770 --> 00:09:54.510 align:middle line:84% What I mean is a typical person saved $27.50 per hour they 00:09:54.510 --> 00:09:55.130 align:middle line:90% waited. 00:09:55.130 --> 00:09:57.650 align:middle line:84% If you do that math, per hour someone waited, 00:09:57.650 --> 00:10:00.410 align:middle line:90% they saved $27.50. 00:10:00.410 --> 00:10:02.390 align:middle line:90% So that's the value of time. 00:10:02.390 --> 00:10:03.310 align:middle line:90% We can value time. 00:10:03.310 --> 00:10:06.310 align:middle line:90% People randomly had to choose. 00:10:06.310 --> 00:10:10.190 align:middle line:84% And their preference reveals that they would pay up to $27.50 00:10:10.190 --> 00:10:11.502 align:middle line:90% to save an hour. 00:10:11.502 --> 00:10:13.710 align:middle line:84% But beyond that, they'd rather-- they wouldn't rather 00:10:13.710 --> 00:10:15.110 align:middle line:90% save the hour. 00:10:15.110 --> 00:10:18.590 align:middle line:84% What's very cool about that is in today's dollars, that's 00:10:18.590 --> 00:10:21.997 align:middle line:84% almost exactly the average wage in America, 00:10:21.997 --> 00:10:23.830 align:middle line:84% that actually the estimate you get from that 00:10:23.830 --> 00:10:26.497 align:middle line:84% is very similar to the estimate you get from the first approach. 00:10:26.497 --> 00:10:27.910 align:middle line:90% That's pretty cool. 00:10:27.910 --> 00:10:29.410 align:middle line:84% These approaches are very different. 00:10:29.410 --> 00:10:30.250 align:middle line:90% And these two align. 00:10:30.250 --> 00:10:31.670 align:middle line:90% That's very good news. 00:10:31.670 --> 00:10:33.628 align:middle line:84% So it seems like we have a pretty good estimate 00:10:33.628 --> 00:10:34.670 align:middle line:90% of the value of time. 00:10:34.670 --> 00:10:38.450 align:middle line:84% The average wage in the US was about $23. 00:10:38.450 --> 00:10:40.510 align:middle line:84% So it was a little bit higher but close enough 00:10:40.510 --> 00:10:41.830 align:middle line:90% for government work. 00:10:41.830 --> 00:10:44.982 align:middle line:84% So basically, it looks like you got a pretty similar estimate 00:10:44.982 --> 00:10:45.940 align:middle line:90% of what time is worth-- 00:10:45.940 --> 00:10:48.600 align:middle line:90% 23, $27 an hour. 00:10:48.600 --> 00:10:53.280 align:middle line:84% So that is an example of how we use these methods. 00:10:53.280 --> 00:10:55.820 align:middle line:84% Now let's come to the more interesting case, 00:10:55.820 --> 00:10:58.680 align:middle line:90% which is valuing lives. 00:10:58.680 --> 00:11:01.000 align:middle line:84% Let's ask the first question, which is-- 00:11:01.000 --> 00:11:05.720 align:middle line:84% God, that sounds awfully morally difficult and ugly to value 00:11:05.720 --> 00:11:06.220 align:middle line:90% lives. 00:11:06.220 --> 00:11:08.052 align:middle line:84% I mean, if you go to your cocktail party, 00:11:08.052 --> 00:11:10.260 align:middle line:84% and they're like, oh my god, we got to save this kid. 00:11:10.260 --> 00:11:10.820 align:middle line:90% He's down the well. 00:11:10.820 --> 00:11:12.200 align:middle line:84% And you're like, well, it depends how much 00:11:12.200 --> 00:11:13.540 align:middle line:90% it's going to cost to save him. 00:11:13.540 --> 00:11:15.720 align:middle line:84% Your friends look at you like you're an asshole. 00:11:15.720 --> 00:11:17.580 align:middle line:90% And you're just an economist. 00:11:17.580 --> 00:11:19.600 align:middle line:84% And that's why they call us the dismal science, 00:11:19.600 --> 00:11:21.142 align:middle line:84% because we make statements like that. 00:11:21.142 --> 00:11:24.240 align:middle line:84% But in fact, society does not leave us a choice. 00:11:24.240 --> 00:11:26.312 align:middle line:84% We cannot choose whether or not to value lives. 00:11:26.312 --> 00:11:28.020 align:middle line:84% We have to choose whether to value lives, 00:11:28.020 --> 00:11:31.600 align:middle line:84% because all the time, policymakers face decisions that 00:11:31.600 --> 00:11:33.840 align:middle line:90% turn on the value of a life. 00:11:33.840 --> 00:11:39.662 align:middle line:84% So some fun examples in the book-- so in 1993, GM-- 00:11:39.662 --> 00:11:41.120 align:middle line:84% swear to God, it's a true example-- 00:11:41.120 --> 00:11:46.450 align:middle line:84% GM made a new truck that was exploding. 00:11:46.450 --> 00:11:48.250 align:middle line:84% It turned out they'd put the gas tank 00:11:48.250 --> 00:11:49.550 align:middle line:90% on the outside of the truck. 00:11:49.550 --> 00:11:50.690 align:middle line:90% I'm not kidding you. 00:11:50.690 --> 00:11:52.370 align:middle line:84% Literally, they built the gas tank on the outside. 00:11:52.370 --> 00:11:54.828 align:middle line:84% It turns out when the gas tank is on the outside the truck, 00:11:54.828 --> 00:11:56.930 align:middle line:90% it's more likely to explode. 00:11:56.930 --> 00:11:59.950 align:middle line:84% GM probably should have known that. 00:11:59.950 --> 00:12:01.190 align:middle line:90% And so there was a recall. 00:12:01.190 --> 00:12:01.950 align:middle line:90% There's recall. 00:12:01.950 --> 00:12:04.117 align:middle line:84% They said, look-- so the government said, GM, you've 00:12:04.117 --> 00:12:07.090 align:middle line:84% got to recall all these trucks, getting them off the roads. 00:12:07.090 --> 00:12:12.950 align:middle line:84% So the government estimated that the recall would cost $1 billion 00:12:12.950 --> 00:12:15.190 align:middle line:84% and would save at most 32 more lives. 00:12:15.190 --> 00:12:16.530 align:middle line:90% 32 lives. 00:12:16.530 --> 00:12:19.800 align:middle line:84% So that's a value of life of $1 billion divided by 32, 00:12:19.800 --> 00:12:21.570 align:middle line:90% or $31 million. 00:12:21.570 --> 00:12:23.970 align:middle line:84% So GM could save those lives, $31 million. 00:12:23.970 --> 00:12:26.770 align:middle line:84% GM went back and said, wait a second, 00:12:26.770 --> 00:12:28.330 align:middle line:90% that's hugely inefficient. 00:12:28.330 --> 00:12:34.770 align:middle line:84% What if instead we said we are going to buy 200,000 child 00:12:34.770 --> 00:12:38.570 align:middle line:84% safety seats and provide all this money to educate people 00:12:38.570 --> 00:12:40.770 align:middle line:84% about drunk driving and wearing seat belts? 00:12:40.770 --> 00:12:45.850 align:middle line:84% And they estimated that that would save at least 50 lives 00:12:45.850 --> 00:12:50.250 align:middle line:84% at a cost of at least-- save at least 50 lives 00:12:50.250 --> 00:12:55.450 align:middle line:84% at a cost of $50 million, or only $1 million per life saved. 00:12:55.450 --> 00:12:58.050 align:middle line:84% So GM said, look, you can make us do this recall, 00:12:58.050 --> 00:13:00.350 align:middle line:84% but that's going to be $31 million per life saved. 00:13:00.350 --> 00:13:04.690 align:middle line:84% Or we can save more lives for way less money. 00:13:04.690 --> 00:13:07.170 align:middle line:90% And as you can imagine, their-- 00:13:07.170 --> 00:13:09.390 align:middle line:84% GM even saying that caused a massive uproar. 00:13:09.390 --> 00:13:12.530 align:middle line:84% And they had to recall the cars, but they were right. 00:13:12.530 --> 00:13:14.910 align:middle line:90% That's a rational calculation. 00:13:14.910 --> 00:13:16.330 align:middle line:90% Let's look at another example. 00:13:16.330 --> 00:13:19.090 align:middle line:84% In 1999, there was a commuter train crash 00:13:19.090 --> 00:13:22.270 align:middle line:84% at London's Paddington Station, killing 31 people 00:13:22.270 --> 00:13:24.750 align:middle line:84% because basically, the brakes failed on the train. 00:13:24.750 --> 00:13:26.810 align:middle line:84% Basically, the safety system failed. 00:13:26.810 --> 00:13:29.250 align:middle line:84% So basically, London wanted to introduce a new safety 00:13:29.250 --> 00:13:30.630 align:middle line:90% system for their trains. 00:13:30.630 --> 00:13:32.050 align:middle line:90% And they had a choice. 00:13:32.050 --> 00:13:35.090 align:middle line:84% They could put in a system called the Train Protection 00:13:35.090 --> 00:13:37.370 align:middle line:84% Warning System, which would stop any train traveling 00:13:37.370 --> 00:13:43.060 align:middle line:84% under 75 miles an hour, for $700 million. 00:13:43.060 --> 00:13:45.360 align:middle line:84% For $700 million, they could put in this one system. 00:13:45.360 --> 00:13:49.580 align:middle line:84% Or for $3 billion, 3 to $9 billion, 00:13:49.580 --> 00:13:52.780 align:middle line:84% they could stop trains traveling at any speed. 00:13:52.780 --> 00:13:55.740 align:middle line:84% They could have a better system for 3 to $9 billion 00:13:55.740 --> 00:14:01.300 align:middle line:84% that would save one to three more lives per year. 00:14:01.300 --> 00:14:05.260 align:middle line:84% So if we think this system was going to cost-- 00:14:05.260 --> 00:14:08.540 align:middle line:84% last 30 to 50 years, that would mean it 00:14:08.540 --> 00:14:12.620 align:middle line:84% would-- the cost would be between 20 and $300 million 00:14:12.620 --> 00:14:14.060 align:middle line:90% per life saved. 00:14:14.060 --> 00:14:17.120 align:middle line:84% So by going from one system to the better system, 00:14:17.120 --> 00:14:20.380 align:middle line:84% we could save as many as three lives a year at a cost of 20 00:14:20.380 --> 00:14:22.620 align:middle line:90% to $300 million per life saved. 00:14:22.620 --> 00:14:24.620 align:middle line:90% Worth it? 00:14:24.620 --> 00:14:26.320 align:middle line:90% That's pretty expensive. 00:14:26.320 --> 00:14:27.493 align:middle line:90% They decide not to do it. 00:14:27.493 --> 00:14:28.160 align:middle line:90% Were they wrong? 00:14:28.160 --> 00:14:28.620 align:middle line:90% Were they right? 00:14:28.620 --> 00:14:29.720 align:middle line:84% That's what we need to think about. 00:14:29.720 --> 00:14:32.220 align:middle line:84% But the point is, they couldn't avoid that decision. 00:14:32.220 --> 00:14:34.520 align:middle line:84% And let's talk about the most pertinent example to us, 00:14:34.520 --> 00:14:36.100 align:middle line:90% which is COVID. 00:14:36.100 --> 00:14:38.180 align:middle line:84% This decision was pertinent during COVID. 00:14:38.180 --> 00:14:42.830 align:middle line:84% Now, first of all, if-- sometimes it's a no-brainer. 00:14:42.830 --> 00:14:47.910 align:middle line:84% So for example, aggressive testing for COVID 00:14:47.910 --> 00:14:51.730 align:middle line:84% basically cost a tiny amount of money per life saved. 00:14:51.730 --> 00:14:54.930 align:middle line:84% Or vaccines cost a tiny amount of money per life saved. 00:14:54.930 --> 00:14:56.670 align:middle line:90% Those are no-brainers. 00:14:56.670 --> 00:14:58.750 align:middle line:84% But what about all the shutdowns? 00:14:58.750 --> 00:15:00.070 align:middle line:90% That's a harder case. 00:15:00.070 --> 00:15:03.350 align:middle line:84% People have actually valued that and tried to value 00:15:03.350 --> 00:15:06.630 align:middle line:84% how much we lost in economic activity versus how many lives 00:15:06.630 --> 00:15:08.630 align:middle line:90% we saved by the shutdowns. 00:15:08.630 --> 00:15:14.630 align:middle line:84% And they estimate that basically, the shutdowns 00:15:14.630 --> 00:15:19.990 align:middle line:84% cost the US about $7.2 trillion in foregone economic activity 00:15:19.990 --> 00:15:24.990 align:middle line:84% and saved 1.24 million lives, falsely precise but roughly 00:15:24.990 --> 00:15:29.950 align:middle line:84% speaking, for a cost of about $6 million per life saved. 00:15:29.950 --> 00:15:31.090 align:middle line:90% That's a big number. 00:15:31.090 --> 00:15:31.870 align:middle line:90% How big a number? 00:15:31.870 --> 00:15:34.230 align:middle line:84% Should we pay $6 million to save a life? 00:15:34.230 --> 00:15:36.015 align:middle line:90% So you cannot avoid this. 00:15:36.015 --> 00:15:37.640 align:middle line:84% And certainly in your lifetimes, you've 00:15:37.640 --> 00:15:40.280 align:middle line:84% seen it in bright lights in COVID, which 00:15:40.280 --> 00:15:41.760 align:middle line:90% is, how do we think about-- 00:15:41.760 --> 00:15:44.720 align:middle line:84% let's take $6 million, if that was worth it or not. 00:15:44.720 --> 00:15:47.160 align:middle line:84% So once again, we can go to the same three methods 00:15:47.160 --> 00:15:49.720 align:middle line:84% to try to decide if $6 million a life worth it. 00:15:49.720 --> 00:15:51.600 align:middle line:90% The first is market value. 00:15:51.600 --> 00:15:53.480 align:middle line:84% What's the market value of a life? 00:15:53.480 --> 00:15:56.240 align:middle line:90% Well, once again, we have wages. 00:15:56.240 --> 00:15:59.320 align:middle line:84% You could say basically, the value of your time is wages. 00:15:59.320 --> 00:16:02.440 align:middle line:84% People work on average about 20% of their life. 00:16:02.440 --> 00:16:07.280 align:middle line:84% So we can take the average wage, take a 40-hour week-- 00:16:07.280 --> 00:16:10.280 align:middle line:84% or take the average wage applied to every hour you're alive. 00:16:10.280 --> 00:16:14.040 align:middle line:84% And what you get from that is a value of life about $4 and 1/2 00:16:14.040 --> 00:16:16.820 align:middle line:90% million. 00:16:16.820 --> 00:16:19.600 align:middle line:90% So that's below the $6 million. 00:16:19.600 --> 00:16:22.460 align:middle line:84% Now, clearly, that has some flaws. 00:16:22.460 --> 00:16:25.180 align:middle line:90% What might be wrong with that? 00:16:25.180 --> 00:16:25.680 align:middle line:90% Yeah? 00:16:25.680 --> 00:16:30.320 align:middle line:84% AUDIENCE: The wage disparities among industries, positions. 00:16:30.320 --> 00:16:33.460 align:middle line:84% JON GRUBER: Well, that-- this is a very important question. 00:16:33.460 --> 00:16:34.700 align:middle line:90% This is why we're unpopular. 00:16:34.700 --> 00:16:35.620 align:middle line:90% That is not a problem. 00:16:35.620 --> 00:16:38.660 align:middle line:84% Why is that not a problem, that there's wage disparities? 00:16:38.660 --> 00:16:40.700 align:middle line:84% Why is that not necessarily a problem? 00:16:40.700 --> 00:16:44.225 align:middle line:84% When would it not be a problem that there's wage disparities? 00:16:44.225 --> 00:16:45.100 align:middle line:90% Let's make it simple. 00:16:45.100 --> 00:16:54.940 align:middle line:84% Imagine workers in-- garbage men make less than finance bros. 00:16:54.940 --> 00:16:56.680 align:middle line:84% Why is that not necessarily a problem? 00:16:56.680 --> 00:17:00.177 align:middle line:84% Garbage men lives may be worth less than finance bro lives. 00:17:00.177 --> 00:17:02.760 align:middle line:84% AUDIENCE: That's not a problem if the deaths you're preventing 00:17:02.760 --> 00:17:04.119 align:middle line:90% are randomly distributed. 00:17:04.119 --> 00:17:06.119 align:middle line:90% JON GRUBER: No. 00:17:06.119 --> 00:17:08.700 align:middle line:84% AUDIENCE: It's not a problem if it's a competitive market. 00:17:08.700 --> 00:17:10.825 align:middle line:84% JON GRUBER: If the marginal product of finance bros 00:17:10.825 --> 00:17:13.103 align:middle line:84% is higher, then they're worth more. 00:17:13.103 --> 00:17:15.520 align:middle line:84% That's the kind of shit we got to deal with as economists. 00:17:15.520 --> 00:17:21.579 align:middle line:84% So basically, essentially, the point is, in our economic model, 00:17:21.579 --> 00:17:24.079 align:middle line:84% the value of life is a function of many things, one of which 00:17:24.079 --> 00:17:26.680 align:middle line:84% is the marginal product of what you produce in society. 00:17:26.680 --> 00:17:28.865 align:middle line:84% Now, you go further and argue, as is probably right, 00:17:28.865 --> 00:17:31.240 align:middle line:84% that what finance bros makes does not necessarily reflect 00:17:31.240 --> 00:17:34.160 align:middle line:90% their underlying social value. 00:17:34.160 --> 00:17:38.670 align:middle line:84% No offense to all you future finance bros and sisters. 00:17:38.670 --> 00:17:42.990 align:middle line:84% And then maybe we start to have an issue with wages. 00:17:42.990 --> 00:17:43.990 align:middle line:90% But let's be very clear. 00:17:43.990 --> 00:17:45.450 align:middle line:84% Just like Exxon, we have to start with what 00:17:45.450 --> 00:17:47.030 align:middle line:90% the basic economic model says. 00:17:47.030 --> 00:17:50.492 align:middle line:84% The basic economic model says that essentially, 00:17:50.492 --> 00:17:52.450 align:middle line:84% in a perfect labor market, your wage represents 00:17:52.450 --> 00:17:54.492 align:middle line:84% your marginal product, and the value of your life 00:17:54.492 --> 00:17:57.850 align:middle line:84% is your marginal product applied to all your hours. 00:17:57.850 --> 00:18:00.310 align:middle line:84% So one problem with that, of course, 00:18:00.310 --> 00:18:03.850 align:middle line:84% is we think wages are a bad indicator of marginal product. 00:18:03.850 --> 00:18:05.590 align:middle line:84% This isn't a problem of valuing time, 00:18:05.590 --> 00:18:07.332 align:middle line:84% but it is a problem of valuing life. 00:18:07.332 --> 00:18:09.290 align:middle line:84% We have all the other problems we talked about, 00:18:09.290 --> 00:18:11.330 align:middle line:84% a reason why people can't set the value of wage 00:18:11.330 --> 00:18:12.732 align:middle line:90% equal to leisure. 00:18:12.732 --> 00:18:14.690 align:middle line:84% We also have the problem that your wage may not 00:18:14.690 --> 00:18:17.610 align:middle line:84% reflect how much people like you, 00:18:17.610 --> 00:18:20.450 align:middle line:84% once again coming back to the finance bros. 00:18:20.450 --> 00:18:25.090 align:middle line:84% So basically, different people will be differentially sad 00:18:25.090 --> 00:18:26.170 align:middle line:90% if you die. 00:18:26.170 --> 00:18:28.670 align:middle line:84% And that should presumably reflect the value of your life 00:18:28.670 --> 00:18:28.850 align:middle line:90% too. 00:18:28.850 --> 00:18:29.290 align:middle line:90% Yeah. 00:18:29.290 --> 00:18:29.915 align:middle line:90% AUDIENCE: Yeah. 00:18:29.915 --> 00:18:31.850 align:middle line:90% I got a corollary to that. 00:18:31.850 --> 00:18:34.940 align:middle line:84% A person's value is not the same as the value 00:18:34.940 --> 00:18:36.497 align:middle line:90% that they are able to produce. 00:18:36.497 --> 00:18:37.580 align:middle line:90% JON GRUBER: Yeah, exactly. 00:18:37.580 --> 00:18:41.100 align:middle line:84% So market value is not great for lives. 00:18:41.100 --> 00:18:44.403 align:middle line:84% That's not really going to fly, although in legal circumstances, 00:18:44.403 --> 00:18:45.320 align:middle line:90% typically what's used. 00:18:45.320 --> 00:18:47.680 align:middle line:84% If you look at typically-- if you look at damages case, 00:18:47.680 --> 00:18:49.120 align:middle line:90% people are killed by accident. 00:18:49.120 --> 00:18:52.540 align:middle line:84% They'll often use forgone wages as a measure. 00:18:52.540 --> 00:18:54.480 align:middle line:84% Second would be contingent valuation. 00:18:54.480 --> 00:18:55.037 align:middle line:90% Ask people. 00:18:55.037 --> 00:18:57.120 align:middle line:84% Now, you can't ask people what their life's worth. 00:18:57.120 --> 00:18:57.760 align:middle line:90% That's dumb. 00:18:57.760 --> 00:19:01.140 align:middle line:84% But you can ask people questions like, how much would you 00:19:01.140 --> 00:19:03.880 align:middle line:84% pay to reduce the risk of your life? 00:19:03.880 --> 00:19:05.500 align:middle line:84% So for example, you could say, look, 00:19:05.500 --> 00:19:08.850 align:middle line:84% you're about to take a flight in an airline. 00:19:08.850 --> 00:19:10.600 align:middle line:84% You're about to take a flight from A to B. 00:19:10.600 --> 00:19:11.480 align:middle line:90% You have two choices. 00:19:11.480 --> 00:19:14.637 align:middle line:84% One is an airline where it crashes once every 50,000 times, 00:19:14.637 --> 00:19:17.220 align:middle line:84% and one's an airline where it crashes every once every 500,000 00:19:17.220 --> 00:19:17.840 align:middle line:90% times. 00:19:17.840 --> 00:19:19.923 align:middle line:84% How much more would you pay for the second airline 00:19:19.923 --> 00:19:21.227 align:middle line:90% than the first? 00:19:21.227 --> 00:19:23.060 align:middle line:84% And then you could say, well, that gives you 00:19:23.060 --> 00:19:26.660 align:middle line:84% an incremental value of a life, because basically, you're 00:19:26.660 --> 00:19:28.700 align:middle line:90% getting a probability of dying. 00:19:28.700 --> 00:19:31.100 align:middle line:84% They'll tell you-- they say if it's $100 more, well, 00:19:31.100 --> 00:19:33.780 align:middle line:90% then that gives you $100 is-- 00:19:33.780 --> 00:19:35.540 align:middle line:84% you divide that by the incremental change 00:19:35.540 --> 00:19:36.820 align:middle line:90% in the value of dying. 00:19:36.820 --> 00:19:39.580 align:middle line:90% That gives you a value of life. 00:19:39.580 --> 00:19:42.400 align:middle line:84% The problem with that is when you use that method, 00:19:42.400 --> 00:19:46.100 align:middle line:84% you get the tight estimate of 1 million to $32 million 00:19:46.100 --> 00:19:47.162 align:middle line:90% per life saved. 00:19:47.162 --> 00:19:48.120 align:middle line:90% People just don't know. 00:19:48.120 --> 00:19:50.500 align:middle line:90% I mean, how can you know. 00:19:50.500 --> 00:19:52.020 align:middle line:84% And that's why economists once again 00:19:52.020 --> 00:19:55.300 align:middle line:84% turn to our favorite method, which is a revealed preference. 00:19:55.300 --> 00:20:00.580 align:middle line:84% How do you get people to reveal their preference for their life? 00:20:00.580 --> 00:20:02.900 align:middle line:90% How do you do that? 00:20:02.900 --> 00:20:03.540 align:middle line:90% Yeah. 00:20:03.540 --> 00:20:06.040 align:middle line:84% AUDIENCE: Very similar to the second but slightly different, 00:20:06.040 --> 00:20:07.180 align:middle line:90% we can say, if you-- 00:20:07.180 --> 00:20:10.780 align:middle line:84% I don't know-- consume normal foods versus organic foods 00:20:10.780 --> 00:20:12.427 align:middle line:90% and there's this probability. 00:20:12.427 --> 00:20:13.260 align:middle line:90% JON GRUBER: Exactly. 00:20:13.260 --> 00:20:18.060 align:middle line:84% You look at choices that are made that reveal 00:20:18.060 --> 00:20:19.838 align:middle line:90% that differ in their risk. 00:20:19.838 --> 00:20:22.380 align:middle line:84% And in particular, one of the most significant choices people 00:20:22.380 --> 00:20:26.660 align:middle line:90% make is working conditions. 00:20:26.660 --> 00:20:31.230 align:middle line:84% So if you are someone who is an unskilled worker 50 years ago 00:20:31.230 --> 00:20:33.670 align:middle line:84% in West Virginia, you could become 00:20:33.670 --> 00:20:36.750 align:middle line:84% an administrative assistant, or you could become a coal worker. 00:20:36.750 --> 00:20:41.430 align:middle line:84% Those had very different risk profiles. 00:20:41.430 --> 00:20:44.030 align:middle line:84% And you could ask-- but they had basically similar skill 00:20:44.030 --> 00:20:46.295 align:middle line:90% background requirements. 00:20:46.295 --> 00:20:48.170 align:middle line:84% You work at a grocery store, be a coal miner. 00:20:48.170 --> 00:20:49.170 align:middle line:90% Let's make that better. 00:20:49.170 --> 00:20:51.850 align:middle line:84% Similar requirements, but very, very different risk profiles. 00:20:51.850 --> 00:20:53.790 align:middle line:84% And you can look at, how much higher 00:20:53.790 --> 00:20:56.550 align:middle line:84% are the wages in jobs that are riskier? 00:20:56.550 --> 00:21:00.390 align:middle line:84% So if a job that has a 1% chance of killing you-- 00:21:00.390 --> 00:21:05.670 align:middle line:84% a 1% higher chance of killing you every year pays $1,000 per 00:21:05.670 --> 00:21:10.070 align:middle line:84% year more, then that says your life is worth a thousand over 00:21:10.070 --> 00:21:13.310 align:middle line:90% 0.01, or $1 million. 00:21:13.310 --> 00:21:15.590 align:middle line:84% So you basically look at how much more 00:21:15.590 --> 00:21:18.750 align:middle line:84% you have to pay to get people to take jobs that risk their-- 00:21:18.750 --> 00:21:21.750 align:middle line:84% that change their life risk by a certain amount. 00:21:21.750 --> 00:21:24.470 align:middle line:84% And there's a huge literature doing this. 00:21:24.470 --> 00:21:28.230 align:middle line:84% This is the economists' go-to method for valuing life. 00:21:28.230 --> 00:21:30.520 align:middle line:84% And the best estimates from this literature 00:21:30.520 --> 00:21:38.058 align:middle line:84% are that a life is worth about $10.5 million, 00:21:38.058 --> 00:21:40.600 align:middle line:84% which is a big number, bigger than the COVID number we talked 00:21:40.600 --> 00:21:41.380 align:middle line:90% about before. 00:21:41.380 --> 00:21:45.660 align:middle line:90% 00:21:45.660 --> 00:21:49.040 align:middle line:84% However, while I agree with most economists, this best method 00:21:49.040 --> 00:21:51.640 align:middle line:84% to value life, and it's the number I use in my work, 00:21:51.640 --> 00:21:55.020 align:middle line:84% in references, there are a number of problems with this. 00:21:55.020 --> 00:21:57.473 align:middle line:90% 00:21:57.473 --> 00:21:59.140 align:middle line:84% What are some of the problems with this? 00:21:59.140 --> 00:22:01.598 align:middle line:84% What are some of the problems of using some real preference 00:22:01.598 --> 00:22:02.190 align:middle line:90% method? 00:22:02.190 --> 00:22:03.200 align:middle line:90% Yeah. 00:22:03.200 --> 00:22:05.540 align:middle line:84% AUDIENCE: Determining the risk of someone's occupation. 00:22:05.540 --> 00:22:06.248 align:middle line:90% JON GRUBER: Yeah. 00:22:06.248 --> 00:22:08.920 align:middle line:84% Well, not just determining it but the person knowing it. 00:22:08.920 --> 00:22:10.960 align:middle line:84% I mean, I can measure it statistically, 00:22:10.960 --> 00:22:12.643 align:middle line:90% but the person has to know it. 00:22:12.643 --> 00:22:14.060 align:middle line:84% So there's an information problem. 00:22:14.060 --> 00:22:14.400 align:middle line:90% Yeah. 00:22:14.400 --> 00:22:16.240 align:middle line:84% AUDIENCE: People don't necessarily have-- 00:22:16.240 --> 00:22:19.240 align:middle line:84% like coal miners don't have free choice in-- 00:22:19.240 --> 00:22:24.120 align:middle line:84% JON GRUBER: They don't say a free choice in what they do. 00:22:24.120 --> 00:22:24.740 align:middle line:90% Great point. 00:22:24.740 --> 00:22:26.360 align:middle line:90% Other points? 00:22:26.360 --> 00:22:30.130 align:middle line:84% Other reasons why this might be problematic? 00:22:30.130 --> 00:22:31.250 align:middle line:90% Yeah. 00:22:31.250 --> 00:22:33.170 align:middle line:84% AUDIENCE: The difference between coal mining 00:22:33.170 --> 00:22:37.330 align:middle line:84% versus working in a grocery store might be significant. 00:22:37.330 --> 00:22:38.570 align:middle line:90% JON GRUBER: Great. 00:22:38.570 --> 00:22:40.752 align:middle line:84% The only difference is not just that it kills you. 00:22:40.752 --> 00:22:41.710 align:middle line:90% There's lots of things. 00:22:41.710 --> 00:22:43.170 align:middle line:90% The working conditions suck. 00:22:43.170 --> 00:22:44.530 align:middle line:90% You might get injured. 00:22:44.530 --> 00:22:47.627 align:middle line:84% This compensating differential, which by the way, 00:22:47.627 --> 00:22:50.210 align:middle line:84% this is a concept-- the concept of compensating differentials, 00:22:50.210 --> 00:22:53.170 align:middle line:84% I should have mentioned, goes back to Adam Smith. 00:22:53.170 --> 00:22:55.530 align:middle line:84% He actually talked about this in 1776. 00:22:55.530 --> 00:22:58.410 align:middle line:84% It's The Wealth of Nations, the first important economics text. 00:22:58.410 --> 00:23:00.607 align:middle line:90% He talks about this idea. 00:23:00.607 --> 00:23:02.690 align:middle line:84% But the problem is that compensating differentials 00:23:02.690 --> 00:23:04.065 align:middle line:84% with two jobs is the differential 00:23:04.065 --> 00:23:07.103 align:middle line:84% for all the amenities between the two jobs, not just death. 00:23:07.103 --> 00:23:08.770 align:middle line:84% And it's hard to find two jobs identical 00:23:08.770 --> 00:23:10.273 align:middle line:90% except for the risk of death. 00:23:10.273 --> 00:23:12.690 align:middle line:84% Typically, the risk of death comes with other health risks 00:23:12.690 --> 00:23:13.850 align:middle line:90% as well. 00:23:13.850 --> 00:23:15.590 align:middle line:84% The other problem is, in some sense, 00:23:15.590 --> 00:23:18.250 align:middle line:84% you're overstating the value of life 00:23:18.250 --> 00:23:21.210 align:middle line:84% if you use this differential, because you're actually 00:23:21.210 --> 00:23:25.830 align:middle line:84% valuing the value of everything that's crappy about this job. 00:23:25.830 --> 00:23:26.670 align:middle line:90% Other problems? 00:23:26.670 --> 00:23:29.990 align:middle line:90% 00:23:29.990 --> 00:23:30.490 align:middle line:90% Yeah. 00:23:30.490 --> 00:23:32.890 align:middle line:84% AUDIENCE: Even within jobs, risks for individuals 00:23:32.890 --> 00:23:34.810 align:middle line:84% may differ, and individuals may know that. 00:23:34.810 --> 00:23:36.628 align:middle line:84% JON GRUBER: So this is very interesting. 00:23:36.628 --> 00:23:38.170 align:middle line:84% This is a very important point I want 00:23:38.170 --> 00:23:40.530 align:middle line:84% to come to that's come home very clearly to me 00:23:40.530 --> 00:23:46.750 align:middle line:84% with my daughter's work, which is when you use this method, 00:23:46.750 --> 00:23:48.770 align:middle line:84% you only get-- and this comes to the point 00:23:48.770 --> 00:23:51.330 align:middle line:84% also about West Virginia workers having choice. 00:23:51.330 --> 00:23:53.650 align:middle line:84% When you use this method, you only 00:23:53.650 --> 00:23:56.290 align:middle line:84% get the value for the workers who 00:23:56.290 --> 00:24:00.050 align:middle line:84% are on the margin of deciding between the two jobs. 00:24:00.050 --> 00:24:03.050 align:middle line:84% So I don't get your guys' value of life 00:24:03.050 --> 00:24:04.930 align:middle line:84% by comparing coal miners and grocery 00:24:04.930 --> 00:24:05.990 align:middle line:90% workers in West Virginia. 00:24:05.990 --> 00:24:07.690 align:middle line:90% You guys are nothing like that. 00:24:07.690 --> 00:24:10.690 align:middle line:84% So if there's a difference in value of life, 00:24:10.690 --> 00:24:13.130 align:middle line:84% that give a very different answer. 00:24:13.130 --> 00:24:14.970 align:middle line:90% My daughter's an EMT. 00:24:14.970 --> 00:24:20.270 align:middle line:84% It's an incredibly hard, stressful, and skillful job. 00:24:20.270 --> 00:24:23.570 align:middle line:84% She had to study six months, pass an exam, very hard job. 00:24:23.570 --> 00:24:27.000 align:middle line:84% She took a pay cut, moving from Starbucks to EMT. 00:24:27.000 --> 00:24:27.500 align:middle line:90% Why? 00:24:27.500 --> 00:24:30.260 align:middle line:90% 00:24:30.260 --> 00:24:30.800 align:middle line:90% Why? 00:24:30.800 --> 00:24:31.300 align:middle line:90% Yeah. 00:24:31.300 --> 00:24:33.500 align:middle line:84% AUDIENCE: She was [INAUDIBLE] job. 00:24:33.500 --> 00:24:37.260 align:middle line:84% JON GRUBER: Yeah, she's an adrenaline junkie. 00:24:37.260 --> 00:24:38.360 align:middle line:90% She wants that job. 00:24:38.360 --> 00:24:39.680 align:middle line:90% She'd do it for almost free. 00:24:39.680 --> 00:24:40.542 align:middle line:90% She's in school now. 00:24:40.542 --> 00:24:42.000 align:middle line:84% I told her, you don't have to work. 00:24:42.000 --> 00:24:43.250 align:middle line:90% I'll pay your living expenses. 00:24:43.250 --> 00:24:44.420 align:middle line:90% She said no, I have to work. 00:24:44.420 --> 00:24:47.420 align:middle line:84% So basically, the point is if there's 00:24:47.420 --> 00:24:50.340 align:middle line:90% differential tastes for risk-- 00:24:50.340 --> 00:24:52.720 align:middle line:84% so the simple example-- imagine a simple example. 00:24:52.720 --> 00:24:54.095 align:middle line:84% Imagine the workers are deciding, 00:24:54.095 --> 00:24:58.060 align:middle line:84% and there's 900 coal miners jobs, 900 coal mining 00:24:58.060 --> 00:24:58.960 align:middle line:90% jobs to be filled. 00:24:58.960 --> 00:25:00.820 align:middle line:90% And there's 2,000 people. 00:25:00.820 --> 00:25:02.700 align:middle line:84% And 1,000 of them are risk lovers, 00:25:02.700 --> 00:25:05.100 align:middle line:90% and 1,000 are risk avoiders. 00:25:05.100 --> 00:25:07.580 align:middle line:84% Who's going to take the coal miner job? 00:25:07.580 --> 00:25:09.137 align:middle line:90% The 900 risk lovers. 00:25:09.137 --> 00:25:11.220 align:middle line:84% So they're not going to get paid a whole lot more, 00:25:11.220 --> 00:25:13.140 align:middle line:84% because they're already risk lovers. 00:25:13.140 --> 00:25:14.580 align:middle line:84% So you're getting an underestimate 00:25:14.580 --> 00:25:17.060 align:middle line:84% of the value of life if you look at what actually takes. 00:25:17.060 --> 00:25:19.285 align:middle line:84% If you took my daughter's job, you'd say, gee, 00:25:19.285 --> 00:25:21.160 align:middle line:84% you don't have to-- people don't really care. 00:25:21.160 --> 00:25:23.618 align:middle line:84% They're happy to put in putting stressful, life-threatening 00:25:23.618 --> 00:25:24.340 align:middle line:90% situations. 00:25:24.340 --> 00:25:25.090 align:middle line:90% Well, they're not. 00:25:25.090 --> 00:25:27.190 align:middle line:84% It's just people that take those jobs are. 00:25:27.190 --> 00:25:31.030 align:middle line:84% It's not an appropriate estimate for the average person. 00:25:31.030 --> 00:25:33.590 align:middle line:84% And that's an important bias here as well, 00:25:33.590 --> 00:25:36.410 align:middle line:84% that you're not getting the overall estimate. 00:25:36.410 --> 00:25:38.770 align:middle line:84% This is fundamentally at the econometric level 00:25:38.770 --> 00:25:41.090 align:middle line:84% what Josh Angrist, my colleague here at MIT, 00:25:41.090 --> 00:25:45.550 align:middle line:84% won the Nobel Prize for, was pointing out, basically, 00:25:45.550 --> 00:25:48.230 align:middle line:84% that whenever we do economic estimates, 00:25:48.230 --> 00:25:50.190 align:middle line:84% we get the estimate not necessarily 00:25:50.190 --> 00:25:53.510 align:middle line:84% for the average person but for the marginal person whose 00:25:53.510 --> 00:25:54.385 align:middle line:90% behavior has changed. 00:25:54.385 --> 00:25:55.760 align:middle line:84% If you look back at chapter 3, we 00:25:55.760 --> 00:25:57.230 align:middle line:84% don't talk about this in chapter 3. 00:25:57.230 --> 00:25:59.270 align:middle line:84% Look at chapter 3, all the examples. 00:25:59.270 --> 00:26:02.150 align:middle line:84% We're looking at deviations, like a welfare benefit changes 00:26:02.150 --> 00:26:03.392 align:middle line:90% by $100. 00:26:03.392 --> 00:26:05.850 align:middle line:84% We're looking at people who get-- whose behavior's changed. 00:26:05.850 --> 00:26:07.510 align:middle line:90% That's a small subset-- 00:26:07.510 --> 00:26:08.598 align:middle line:90% slice of the population. 00:26:08.598 --> 00:26:10.890 align:middle line:84% If they are not representative of the whole population, 00:26:10.890 --> 00:26:12.682 align:middle line:84% we're not getting the right average answer. 00:26:12.682 --> 00:26:15.190 align:middle line:84% We're just getting the right marginal answer. 00:26:15.190 --> 00:26:18.370 align:middle line:84% So this is an example of that here, which is-- and it's 00:26:18.370 --> 00:26:20.430 align:middle line:84% a particularly worrisome example because there's 00:26:20.430 --> 00:26:23.050 align:middle line:84% a set of people who may love risk or be less risk averse, 00:26:23.050 --> 00:26:26.950 align:middle line:84% and they're the ones who you're getting an estimate for. 00:26:26.950 --> 00:26:28.430 align:middle line:90% Questions about that? 00:26:28.430 --> 00:26:29.030 align:middle line:90% Yeah. 00:26:29.030 --> 00:26:30.990 align:middle line:84% AUDIENCE: So based on this argument, 00:26:30.990 --> 00:26:34.270 align:middle line:84% would you say that 10.5 million is an overestimate? 00:26:34.270 --> 00:26:35.200 align:middle line:90% JON GRUBER: Well, no. 00:26:35.200 --> 00:26:36.950 align:middle line:84% So for that argument, I'd say 10.5 million 00:26:36.950 --> 00:26:38.010 align:middle line:90% is an underestimate. 00:26:38.010 --> 00:26:40.385 align:middle line:84% Other arguments might say it's an overestimate because it 00:26:40.385 --> 00:26:42.510 align:middle line:84% accounts for injury and all the stuff that's 00:26:42.510 --> 00:26:43.950 align:middle line:90% crappy about the jobs. 00:26:43.950 --> 00:26:46.030 align:middle line:84% So that's why I end up using it, because I don't 00:26:46.030 --> 00:26:47.510 align:middle line:90% know which way the bias goes. 00:26:47.510 --> 00:26:48.682 align:middle line:90% There's also the fact. 00:26:48.682 --> 00:26:49.890 align:middle line:90% There's the information fact. 00:26:49.890 --> 00:26:50.790 align:middle line:84% There's another fact, which is we talked 00:26:50.790 --> 00:26:52.230 align:middle line:84% about behavioral economics and people are not 00:26:52.230 --> 00:26:53.890 align:middle line:84% good at optimizing over these things. 00:26:53.890 --> 00:26:55.550 align:middle line:84% So those people may not be making this decision 00:26:55.550 --> 00:26:57.370 align:middle line:84% appropriately, even if they have perfect information. 00:26:57.370 --> 00:26:59.245 align:middle line:84% They may not make the decision appropriately. 00:26:59.245 --> 00:27:01.210 align:middle line:84% So there's a lot of biases to this. 00:27:01.210 --> 00:27:03.808 align:middle line:84% I think there's less bias to this than these other methods. 00:27:03.808 --> 00:27:05.350 align:middle line:84% And I think there's no-- there's been 00:27:05.350 --> 00:27:07.142 align:middle line:84% no evidence of the bias systematically goes 00:27:07.142 --> 00:27:08.870 align:middle line:90% one way or another. 00:27:08.870 --> 00:27:11.010 align:middle line:84% So that's why I tend to use that estimate. 00:27:11.010 --> 00:27:14.710 align:middle line:90% 00:27:14.710 --> 00:27:20.902 align:middle line:84% Another fun thing about value of life 00:27:20.902 --> 00:27:23.360 align:middle line:84% is, should the value of life-- how should the value of life 00:27:23.360 --> 00:27:24.300 align:middle line:90% vary with age? 00:27:24.300 --> 00:27:26.920 align:middle line:90% 00:27:26.920 --> 00:27:28.317 align:middle line:90% So you're a policymaker. 00:27:28.317 --> 00:27:28.900 align:middle line:90% You're a king. 00:27:28.900 --> 00:27:30.620 align:middle line:84% You don't worry about political correctness. 00:27:30.620 --> 00:27:32.495 align:middle line:84% You get to decide how much to spend on saving 00:27:32.495 --> 00:27:33.760 align:middle line:90% people of different ages. 00:27:33.760 --> 00:27:35.480 align:middle line:84% How would you change the value of-- how 00:27:35.480 --> 00:27:37.772 align:middle line:84% do you think the value of life-- what should be the age 00:27:37.772 --> 00:27:40.360 align:middle line:90% profile of the value of life? 00:27:40.360 --> 00:27:42.300 align:middle line:90% Yeah. 00:27:42.300 --> 00:27:46.420 align:middle line:90% AUDIENCE: [INAUDIBLE] should be. 00:27:46.420 --> 00:27:47.420 align:middle line:90% JON GRUBER: Should what? 00:27:47.420 --> 00:27:49.492 align:middle line:84% AUDIENCE: It should peak around 20, 30. 00:27:49.492 --> 00:27:50.200 align:middle line:90% JON GRUBER: Yeah. 00:27:50.200 --> 00:27:53.240 align:middle line:84% So basically, it should be low when 00:27:53.240 --> 00:27:55.560 align:middle line:84% you're a baby, peak around-- not actually 20 or 30, 00:27:55.560 --> 00:27:56.940 align:middle line:84% because your marginal product is always going. 00:27:56.940 --> 00:27:58.480 align:middle line:84% It should really peak-- because remember, 00:27:58.480 --> 00:27:59.772 align:middle line:90% it's not just marginal product. 00:27:59.772 --> 00:28:01.500 align:middle line:84% It's also attachments you have to people. 00:28:01.500 --> 00:28:03.375 align:middle line:84% It's all the things that are value having you 00:28:03.375 --> 00:28:04.780 align:middle line:90% alive in society. 00:28:04.780 --> 00:28:06.280 align:middle line:84% Those probably peak-- I don't know-- 00:28:06.280 --> 00:28:09.160 align:middle line:84% 50, 60, something like that, when 00:28:09.160 --> 00:28:12.520 align:middle line:84% you're at the peak of your earnings power 00:28:12.520 --> 00:28:16.340 align:middle line:84% and when you maximize your attachments to the world. 00:28:16.340 --> 00:28:20.642 align:middle line:84% And then it should fall again, because you're not earning. 00:28:20.642 --> 00:28:21.350 align:middle line:90% People are dying. 00:28:21.350 --> 00:28:24.250 align:middle line:84% You're not as attached to them, whatever. 00:28:24.250 --> 00:28:27.570 align:middle line:84% If you actually do estimates by age of the value of a life, 00:28:27.570 --> 00:28:31.330 align:middle line:84% it actually looks like this and then like that. 00:28:31.330 --> 00:28:31.997 align:middle line:90% It doesn't drop. 00:28:31.997 --> 00:28:33.663 align:middle line:84% That is, people who are 90 think they're 00:28:33.663 --> 00:28:35.450 align:middle line:84% worth just as much as people who are 50, 00:28:35.450 --> 00:28:36.930 align:middle line:84% even though our models would say they don't, which 00:28:36.930 --> 00:28:39.410 align:middle line:84% is once again says that's a problem with the market value 00:28:39.410 --> 00:28:41.170 align:middle line:90% approach. 00:28:41.170 --> 00:28:43.710 align:middle line:84% So that's another-- so this is a fascinating literature. 00:28:43.710 --> 00:28:45.937 align:middle line:84% Here's another really cool-- yeah. 00:28:45.937 --> 00:28:48.270 align:middle line:84% AUDIENCE: That line comes from self-assessments or other 00:28:48.270 --> 00:28:48.490 align:middle line:90% people? 00:28:48.490 --> 00:28:49.890 align:middle line:84% JON GRUBER: This is from compensating differential 00:28:49.890 --> 00:28:51.970 align:middle line:84% studies, because when people are older now, 00:28:51.970 --> 00:28:52.990 align:middle line:84% you might say they're not working. 00:28:52.990 --> 00:28:54.610 align:middle line:84% We have also consumption-based studies of the kind 00:28:54.610 --> 00:28:56.318 align:middle line:84% that you suggested, like how much more we 00:28:56.318 --> 00:28:59.610 align:middle line:90% pay for an airbag in a car. 00:28:59.610 --> 00:29:02.290 align:middle line:84% AUDIENCE: Would we consider it like your potential 00:29:02.290 --> 00:29:03.630 align:middle line:90% to create value? 00:29:03.630 --> 00:29:06.110 align:middle line:84% So as a child, you have your whole life ahead of you. 00:29:06.110 --> 00:29:07.930 align:middle line:84% JON GRUBER: Well, that would be-- 00:29:07.930 --> 00:29:08.990 align:middle line:90% that's a great point. 00:29:08.990 --> 00:29:14.190 align:middle line:84% Many people would say, well, it should be potential. 00:29:14.190 --> 00:29:16.995 align:middle line:84% This is a replacement value model. 00:29:16.995 --> 00:29:18.620 align:middle line:84% I mean, this says, the first day you're 00:29:18.620 --> 00:29:21.100 align:middle line:84% born, you're worth the least, because you're not 00:29:21.100 --> 00:29:23.700 align:middle line:84% earning anything, and you're replaceable nine days-- 00:29:23.700 --> 00:29:24.200 align:middle line:90% month. 00:29:24.200 --> 00:29:25.360 align:middle line:90% So that's awful. 00:29:25.360 --> 00:29:27.920 align:middle line:84% As a parent, I can't imagine anything worse than saying that. 00:29:27.920 --> 00:29:30.573 align:middle line:84% But that is what the economic model would be saying. 00:29:30.573 --> 00:29:32.740 align:middle line:84% But if you're saying potential, but the point is you 00:29:32.740 --> 00:29:34.040 align:middle line:90% could replace that potential. 00:29:34.040 --> 00:29:36.340 align:middle line:84% So that's kind of the idea of why it's low. 00:29:36.340 --> 00:29:39.220 align:middle line:84% Once again, why we're not popular at parties. 00:29:39.220 --> 00:29:44.540 align:middle line:84% So this is actually a really cool study. 00:29:44.540 --> 00:29:46.940 align:middle line:84% In February 2003, the state of Oregon 00:29:46.940 --> 00:29:50.580 align:middle line:84% had massive budgetary problems and laid off a bunch of police. 00:29:50.580 --> 00:29:52.940 align:middle line:84% So they couldn't do traffic stops anymore. 00:29:52.940 --> 00:29:56.700 align:middle line:84% And people drove faster and died more on the road. 00:29:56.700 --> 00:30:00.460 align:middle line:84% So this study looked at how many more deaths were caused 00:30:00.460 --> 00:30:03.380 align:middle line:84% by driving faster versus how much the state saved 00:30:03.380 --> 00:30:05.460 align:middle line:90% by laying off the police. 00:30:05.460 --> 00:30:10.060 align:middle line:84% What they found was there would have been 2,167 fewer 00:30:10.060 --> 00:30:15.180 align:middle line:84% fatalities, and the state saved $670 million, 00:30:15.180 --> 00:30:18.800 align:middle line:84% which implies a value of life of $309,000. 00:30:18.800 --> 00:30:22.200 align:middle line:84% That is if life is worth more than $309,000, 00:30:22.200 --> 00:30:24.380 align:middle line:84% they should not have laid off the troopers. 00:30:24.380 --> 00:30:27.100 align:middle line:84% Life is definitely worth more than $309,000. 00:30:27.100 --> 00:30:29.500 align:middle line:84% So by that metric, it was a huge mistake 00:30:29.500 --> 00:30:32.260 align:middle line:84% to lay off these troopers, which once again speaks to the fact 00:30:32.260 --> 00:30:35.638 align:middle line:84% that society is not always pulling the same direction. 00:30:35.638 --> 00:30:36.680 align:middle line:90% They had a budget crisis. 00:30:36.680 --> 00:30:37.680 align:middle line:90% They had to do something. 00:30:37.680 --> 00:30:39.800 align:middle line:84% There's no social planner saying, oh, don't do it. 00:30:39.800 --> 00:30:40.600 align:middle line:90% Life's worth more than that. 00:30:40.600 --> 00:30:41.620 align:middle line:90% I'll give you money. 00:30:41.620 --> 00:30:43.420 align:middle line:84% So that's an example of the problems 00:30:43.420 --> 00:30:47.980 align:middle line:84% that we face in society, making these hard decisions. 00:30:47.980 --> 00:30:50.020 align:middle line:90% Questions about that? 00:30:50.020 --> 00:30:51.900 align:middle line:84% What I really like about this lecture 00:30:51.900 --> 00:30:53.878 align:middle line:84% is cost-benefit analysis seems like it'd 00:30:53.878 --> 00:30:55.420 align:middle line:84% be almost the most-- almost certainly 00:30:55.420 --> 00:30:57.040 align:middle line:84% the most boring thing in the world, 00:30:57.040 --> 00:30:59.180 align:middle line:84% but it turns out to be really interesting and fun 00:30:59.180 --> 00:31:01.680 align:middle line:84% because there's all these really interesting moral questions 00:31:01.680 --> 00:31:05.020 align:middle line:90% and ethical issues that come up. 00:31:05.020 --> 00:31:05.980 align:middle line:90% Yeah? 00:31:05.980 --> 00:31:10.180 align:middle line:84% AUDIENCE: So my question now is, given all these problems 00:31:10.180 --> 00:31:13.420 align:middle line:84% we run into, societally, are we accepting 00:31:13.420 --> 00:31:16.710 align:middle line:84% that there is one cost of life for every person 00:31:16.710 --> 00:31:17.793 align:middle line:90% or that they differ? 00:31:17.793 --> 00:31:20.210 align:middle line:84% JON GRUBER: Well, not just one cost life for every person. 00:31:20.210 --> 00:31:21.990 align:middle line:90% There's a great-- there's age. 00:31:21.990 --> 00:31:23.170 align:middle line:90% There's income. 00:31:23.170 --> 00:31:23.980 align:middle line:90% There's countries. 00:31:23.980 --> 00:31:26.230 align:middle line:84% How do they value in life in Bangladesh versus valuing 00:31:26.230 --> 00:31:27.510 align:middle line:90% a life in the US? 00:31:27.510 --> 00:31:29.230 align:middle line:84% There's all these difficult issues. 00:31:29.230 --> 00:31:33.270 align:middle line:84% And basically, we don't have very clear answers. 00:31:33.270 --> 00:31:36.830 align:middle line:84% So basically, what we do is we have to pick a number. 00:31:36.830 --> 00:31:39.790 align:middle line:90% The government picks a number. 00:31:39.790 --> 00:31:44.070 align:middle line:84% But that number even varies across government agencies. 00:31:44.070 --> 00:31:46.690 align:middle line:84% So it's really, really a challenging issue. 00:31:46.690 --> 00:31:48.730 align:middle line:84% But once again, because it's challenging, 00:31:48.730 --> 00:31:49.910 align:middle line:84% we can't throw up our hands and say, we're 00:31:49.910 --> 00:31:50.730 align:middle line:90% not going to deal with it. 00:31:50.730 --> 00:31:51.730 align:middle line:90% We have to deal with it. 00:31:51.730 --> 00:31:53.110 align:middle line:90% It's not a choice. 00:31:53.110 --> 00:31:55.212 align:middle line:90% The COVID example shows that. 00:31:55.212 --> 00:31:57.670 align:middle line:84% So we just have to keep working on trying to find it and do 00:31:57.670 --> 00:31:58.752 align:middle line:90% better and better job. 00:31:58.752 --> 00:32:00.210 align:middle line:84% Some things will never be resolved. 00:32:00.210 --> 00:32:03.150 align:middle line:84% We'll never resolve how to value a life in Bangladesh 00:32:03.150 --> 00:32:05.110 align:middle line:84% versus the US, because fundamentally, economics 00:32:05.110 --> 00:32:07.110 align:middle line:84% would say the life in Bangladesh is worth less. 00:32:07.110 --> 00:32:09.750 align:middle line:84% Fundamentally, morality would say it's not. 00:32:09.750 --> 00:32:12.600 align:middle line:90% And that's an unresolvable gap. 00:32:12.600 --> 00:32:15.580 align:middle line:84% That we're never going to resolve. 00:32:15.580 --> 00:32:17.168 align:middle line:90% Yeah, go ahead. 00:32:17.168 --> 00:32:18.960 align:middle line:84% AUDIENCE: Is there a difference between how 00:32:18.960 --> 00:32:21.960 align:middle line:84% someone values their own life and people value that person's 00:32:21.960 --> 00:32:22.460 align:middle line:90% life? 00:32:22.460 --> 00:32:23.100 align:middle line:90% JON GRUBER: Great question. 00:32:23.100 --> 00:32:25.220 align:middle line:84% That's another problem with this measure, 00:32:25.220 --> 00:32:26.933 align:middle line:84% which is it's what I think about my life. 00:32:26.933 --> 00:32:28.100 align:middle line:90% I may think nobody likes me. 00:32:28.100 --> 00:32:31.140 align:middle line:84% And a bunch of people might like me. 00:32:31.140 --> 00:32:33.480 align:middle line:84% Or I think people like me, and they don't. 00:32:33.480 --> 00:32:35.580 align:middle line:84% So that's another problem with this measure. 00:32:35.580 --> 00:32:36.980 align:middle line:90% But this is really interesting. 00:32:36.980 --> 00:32:38.680 align:middle line:84% So this at this point, you can be pretty frustrated, 00:32:38.680 --> 00:32:39.660 align:middle line:84% like, well, I don't know what to do. 00:32:39.660 --> 00:32:40.952 align:middle line:90% I don't know what's life worth. 00:32:40.952 --> 00:32:45.560 align:middle line:84% But we can realize there's an important thing we can obviously 00:32:45.560 --> 00:32:50.040 align:middle line:84% do, which is the fact is the government every day makes 00:32:50.040 --> 00:32:52.480 align:middle line:90% decisions that save lives. 00:32:52.480 --> 00:32:54.400 align:middle line:84% And we can ask, are they making them 00:32:54.400 --> 00:32:56.320 align:middle line:90% in the right priority order? 00:32:56.320 --> 00:33:02.200 align:middle line:84% So if you look at table 8-3, this looks at the social cost 00:33:02.200 --> 00:33:05.640 align:middle line:84% of a variety of regulations the government puts in place and how 00:33:05.640 --> 00:33:07.820 align:middle line:84% many lives and what it cost per life saved. 00:33:07.820 --> 00:33:12.770 align:middle line:84% So for example, enforcing that lighters are childproof 00:33:12.770 --> 00:33:15.470 align:middle line:84% does not cost the lighter producers much money 00:33:15.470 --> 00:33:17.290 align:middle line:90% but saves a ton of lives. 00:33:17.290 --> 00:33:21.210 align:middle line:84% So you say, basically, it costs $140,000 per life saved 00:33:21.210 --> 00:33:23.130 align:middle line:90% by childproofing lighters. 00:33:23.130 --> 00:33:26.690 align:middle line:84% On the other hand, regulating cattle 00:33:26.690 --> 00:33:29.210 align:middle line:84% feed so we don't get mad cow disease has 00:33:29.210 --> 00:33:33.270 align:middle line:84% a tiny chance of affecting our health but costs a huge amount. 00:33:33.270 --> 00:33:35.843 align:middle line:84% So that's $250 million per life saved. 00:33:35.843 --> 00:33:37.510 align:middle line:84% So what you can do is you can say, look, 00:33:37.510 --> 00:33:39.190 align:middle line:84% even if I don't know what the number is, 00:33:39.190 --> 00:33:40.850 align:middle line:84% I know we should put more money towards things 00:33:40.850 --> 00:33:43.330 align:middle line:84% on the bottom of this list and less money towards things-- 00:33:43.330 --> 00:33:44.810 align:middle line:84% more effort towards things at the bottom of this list 00:33:44.810 --> 00:33:47.530 align:middle line:84% and less effort towards things at the top of the list. 00:33:47.530 --> 00:33:50.490 align:middle line:84% So at least it can help you with relative answers 00:33:50.490 --> 00:33:53.970 align:middle line:84% or what we call cost effectiveness analysis. 00:33:53.970 --> 00:33:57.610 align:middle line:84% Cost effectiveness analysis is not necessarily, 00:33:57.610 --> 00:33:59.730 align:middle line:90% what's the value? 00:33:59.730 --> 00:34:02.030 align:middle line:84% But what should we do relatively? 00:34:02.030 --> 00:34:04.490 align:middle line:84% What action should we take relative to others? 00:34:04.490 --> 00:34:08.449 align:middle line:84% So at least putting valuation on lives 00:34:08.449 --> 00:34:11.650 align:middle line:84% can at least help us rank different interventions, 00:34:11.650 --> 00:34:14.690 align:middle line:84% even if the absolute level is really hard to pin down. 00:34:14.690 --> 00:34:16.530 align:middle line:90% So there is value in that. 00:34:16.530 --> 00:34:19.270 align:middle line:84% So to the extent we're a little depressed about the fact 00:34:19.270 --> 00:34:20.889 align:middle line:84% it's hard to know how to value this, at least there 00:34:20.889 --> 00:34:22.972 align:middle line:84% is a useful piece, which by doing these exercises, 00:34:22.972 --> 00:34:26.250 align:middle line:84% we can at least rank things and say, hey, among decisions, 00:34:26.250 --> 00:34:28.210 align:middle line:84% this would be the most efficient decision. 00:34:28.210 --> 00:34:29.210 align:middle line:90% Yeah. 00:34:29.210 --> 00:34:32.010 align:middle line:84% AUDIENCE: It probably varies from between each of these. 00:34:32.010 --> 00:34:34.010 align:middle line:90% How linear are these numbers? 00:34:34.010 --> 00:34:36.332 align:middle line:84% If you were to pour in a certain amount of money-- 00:34:36.332 --> 00:34:37.790 align:middle line:84% JON GRUBER: Yeah, no, that's great. 00:34:37.790 --> 00:34:40.830 align:middle line:84% It may be like we can't do more to childproof lighters. 00:34:40.830 --> 00:34:42.170 align:middle line:90% They're childproofed. 00:34:42.170 --> 00:34:43.730 align:middle line:90% I agree. 00:34:43.730 --> 00:34:46.230 align:middle line:84% But think of this more along the decisions we'll talk about, 00:34:46.230 --> 00:34:48.969 align:middle line:84% like, for example, expanding health insurance programs. 00:34:48.969 --> 00:34:51.330 align:middle line:84% You could expand health insurance programs, 00:34:51.330 --> 00:34:57.865 align:middle line:84% or you could subsidize better highways. 00:34:57.865 --> 00:34:58.990 align:middle line:90% They save different things. 00:34:58.990 --> 00:35:01.770 align:middle line:84% You could use the valuation to try to get at that. 00:35:01.770 --> 00:35:03.650 align:middle line:90% Yeah. 00:35:03.650 --> 00:35:08.680 align:middle line:84% AUDIENCE: Does this include marginal benefits of scaling, 00:35:08.680 --> 00:35:11.180 align:middle line:84% because once you have every car with a seat belt, presumably 00:35:11.180 --> 00:35:11.820 align:middle line:90% producing-- 00:35:11.820 --> 00:35:13.653 align:middle line:84% JON GRUBER: That's a similar question, which 00:35:13.653 --> 00:35:16.020 align:middle line:84% is, it's not linear, and there could be cliffs at which 00:35:16.020 --> 00:35:18.460 align:middle line:84% beyond it, which it doesn't help. 00:35:18.460 --> 00:35:19.060 align:middle line:90% Yeah. 00:35:19.060 --> 00:35:20.720 align:middle line:84% AUDIENCE: How much does altruism matter in this? 00:35:20.720 --> 00:35:22.012 align:middle line:90% Like, how do we [? define if ?] 00:35:22.012 --> 00:35:23.700 align:middle line:90% I'm a vegan or vegetarian? 00:35:23.700 --> 00:35:26.658 align:middle line:84% I don't necessarily-- I mean, I wouldn't care for myself [? 00:35:26.658 --> 00:35:27.700 align:middle line:90% or anything like that. ?] 00:35:27.700 --> 00:35:29.658 align:middle line:84% JON GRUBER: No, I mean-- well, actually, that's 00:35:29.658 --> 00:35:32.420 align:middle line:84% a great question, which is this whole discussion is taking 00:35:32.420 --> 00:35:35.680 align:middle line:84% this theoretical perspective of the social planner, 00:35:35.680 --> 00:35:37.740 align:middle line:84% that I'm one benevolent person trying 00:35:37.740 --> 00:35:40.380 align:middle line:90% to do what's best for society. 00:35:40.380 --> 00:35:42.000 align:middle line:84% As we'll discuss more next chapter, 00:35:42.000 --> 00:35:43.680 align:middle line:84% we'll talk about political economics. 00:35:43.680 --> 00:35:46.420 align:middle line:84% Right now, people don't seem to feel that way about our society. 00:35:46.420 --> 00:35:48.862 align:middle line:84% People aren't benevolent towards their fellow citizens, 00:35:48.862 --> 00:35:50.320 align:middle line:84% so that's another factor to put in. 00:35:50.320 --> 00:35:52.840 align:middle line:84% But I think that's the last issue I want to worry about. 00:35:52.840 --> 00:35:54.673 align:middle line:84% Let's assume this benevolent social planner. 00:35:54.673 --> 00:35:57.240 align:middle line:90% It makes life a lot easier. 00:35:57.240 --> 00:35:59.140 align:middle line:90% Other questions, thoughts? 00:35:59.140 --> 00:36:05.820 align:middle line:84% OK, so let me talk about one more issue, which 00:36:05.820 --> 00:36:08.190 align:middle line:84% is, what about the fact that a lot of these benefits 00:36:08.190 --> 00:36:10.353 align:middle line:90% happen in the future? 00:36:10.353 --> 00:36:12.770 align:middle line:84% Now, we talked last time, when costs happen in the future, 00:36:12.770 --> 00:36:13.290 align:middle line:90% we discount it. 00:36:13.290 --> 00:36:14.873 align:middle line:84% We put it in present discounted value. 00:36:14.873 --> 00:36:16.950 align:middle line:84% And you can do a similar thing with benefits. 00:36:16.950 --> 00:36:20.750 align:middle line:84% But it's a little, once again, ethically difficult, especially 00:36:20.750 --> 00:36:22.790 align:middle line:90% something like global warming. 00:36:22.790 --> 00:36:27.030 align:middle line:84% Even with a very low discount rate, lives in 100 years 00:36:27.030 --> 00:36:30.130 align:middle line:84% aren't worth a whole lot relative to lives today. 00:36:30.130 --> 00:36:32.130 align:middle line:84% And lives of 500 years are worth almost nothing, 00:36:32.130 --> 00:36:34.190 align:middle line:90% by the time you discount them. 00:36:34.190 --> 00:36:36.070 align:middle line:84% Do we really want to feel that way? 00:36:36.070 --> 00:36:38.990 align:middle line:84% So imagine global warming worked as no one died for 200 years 00:36:38.990 --> 00:36:41.790 align:middle line:84% and then everyone died, or not everyone died, but then 00:36:41.790 --> 00:36:43.630 align:middle line:90% a huge number of people died. 00:36:43.630 --> 00:36:47.670 align:middle line:84% Well, if you said, gee, I could save 1/10 of those people today. 00:36:47.670 --> 00:36:50.870 align:middle line:84% I might rather do that, because by the time I discount, 00:36:50.870 --> 00:36:53.150 align:middle line:84% those people in 200 years aren't worth that much. 00:36:53.150 --> 00:36:55.190 align:middle line:84% Well, that sounds kind of weird, doesn't it? 00:36:55.190 --> 00:36:58.150 align:middle line:84% People aren't worth something in the future? 00:36:58.150 --> 00:37:00.470 align:middle line:84% So that's another tricky ethical issue, which is, 00:37:00.470 --> 00:37:02.073 align:middle line:90% how do you discount the future? 00:37:02.073 --> 00:37:04.490 align:middle line:84% And once again, the standard model has got a clear answer. 00:37:04.490 --> 00:37:06.300 align:middle line:84% You should discount it because we'd 00:37:06.300 --> 00:37:08.550 align:middle line:84% save that money today of all that money in the future. 00:37:08.550 --> 00:37:12.110 align:middle line:84% But money does no good when the Earth is on fire. 00:37:12.110 --> 00:37:14.610 align:middle line:84% Our standard model is assuming it's a substitution of money, 00:37:14.610 --> 00:37:16.890 align:middle line:84% that money can solve the problem, but money can't. 00:37:16.890 --> 00:37:18.910 align:middle line:84% 200 years from now, it's too late. 00:37:18.910 --> 00:37:20.813 align:middle line:90% Money can't solve that problem. 00:37:20.813 --> 00:37:22.230 align:middle line:84% So for global warming, many people 00:37:22.230 --> 00:37:24.617 align:middle line:84% say we should use a lower or no discount rate, 00:37:24.617 --> 00:37:26.950 align:middle line:84% because you can't easily trade off the money between now 00:37:26.950 --> 00:37:27.770 align:middle line:90% and the future. 00:37:27.770 --> 00:37:29.770 align:middle line:84% So for example, think about a different example. 00:37:29.770 --> 00:37:34.670 align:middle line:84% Imagine investing in food today versus investing 00:37:34.670 --> 00:37:36.970 align:middle line:84% in food in the future or something like that. 00:37:36.970 --> 00:37:38.750 align:middle line:84% Then you would want to use discount rate. 00:37:38.750 --> 00:37:41.310 align:middle line:84% You want to say, well, if we invest in food today-- 00:37:41.310 --> 00:37:43.107 align:middle line:84% if we invest in food in the future, 00:37:43.107 --> 00:37:44.690 align:middle line:84% that's money we could put aside today. 00:37:44.690 --> 00:37:45.690 align:middle line:84% We'll have more money in the future. 00:37:45.690 --> 00:37:47.190 align:middle line:90% We want to account for that. 00:37:47.190 --> 00:37:49.507 align:middle line:84% But if it's stopping global warming in 500 years, 00:37:49.507 --> 00:37:51.590 align:middle line:84% you can't do that, because the money in the future 00:37:51.590 --> 00:37:53.270 align:middle line:84% is no good if we're all underwater. 00:37:53.270 --> 00:37:55.308 align:middle line:84% So that's another difficult-- once again, 00:37:55.308 --> 00:37:56.850 align:middle line:84% not to make this even more difficult. 00:37:56.850 --> 00:37:59.690 align:middle line:84% It's another tricky issue that we have there. 00:37:59.690 --> 00:38:01.350 align:middle line:90% Let's put it all together. 00:38:01.350 --> 00:38:06.280 align:middle line:84% We put it all together in table 8-4. 00:38:06.280 --> 00:38:08.720 align:middle line:84% We already talked about the cost of this project 00:38:08.720 --> 00:38:12.920 align:middle line:84% being $253 million in present value. 00:38:12.920 --> 00:38:15.500 align:middle line:84% We're going to then have time saved and lives saved, 00:38:15.500 --> 00:38:16.620 align:middle line:84% but those are going to go off into the future. 00:38:16.620 --> 00:38:17.953 align:middle line:90% So we're going to discount them. 00:38:17.953 --> 00:38:19.820 align:middle line:84% Let's use the save 7% discount rate. 00:38:19.820 --> 00:38:22.520 align:middle line:84% Let's ignore the global warming issue I just talked about. 00:38:22.520 --> 00:38:24.340 align:middle line:84% And you end up with benefits that are much, 00:38:24.340 --> 00:38:26.160 align:middle line:90% much larger than costs. 00:38:26.160 --> 00:38:28.760 align:middle line:84% So that would say you should do this project, 00:38:28.760 --> 00:38:31.080 align:middle line:84% that the value of time saved and lives saved 00:38:31.080 --> 00:38:34.440 align:middle line:84% is massive relative to the costs. 00:38:34.440 --> 00:38:38.920 align:middle line:84% It turns out that lives are worth so much 00:38:38.920 --> 00:38:43.023 align:middle line:84% that when you're saving lives, most projects want to get done. 00:38:43.023 --> 00:38:45.440 align:middle line:84% So it turns out that once a project involves saving lives, 00:38:45.440 --> 00:38:47.315 align:middle line:84% it's going to be a pretty inefficient project 00:38:47.315 --> 00:38:51.600 align:middle line:84% not to want to do it, because the number is so massive. 00:38:51.600 --> 00:38:54.200 align:middle line:90% $10.5 million a life is a lot. 00:38:54.200 --> 00:38:57.920 align:middle line:90% So that's where we end up. 00:38:57.920 --> 00:39:00.240 align:middle line:90% Questions about that? 00:39:00.240 --> 00:39:03.050 align:middle line:84% Let me talk about a couple more things before we stop. 00:39:03.050 --> 00:39:06.890 align:middle line:84% The first is how to-- if you ever find yourself 00:39:06.890 --> 00:39:10.810 align:middle line:84% in this situation, how to avoid the mistakes noneconomists make. 00:39:10.810 --> 00:39:12.530 align:middle line:84% And a common mistake noneconomists 00:39:12.530 --> 00:39:14.403 align:middle line:84% make is double-counting benefits. 00:39:14.403 --> 00:39:16.070 align:middle line:84% So for example, if you do this analysis, 00:39:16.070 --> 00:39:17.630 align:middle line:84% some would say, well, that's good, 00:39:17.630 --> 00:39:20.090 align:middle line:84% but you've understated the benefits because you've also 00:39:20.090 --> 00:39:21.970 align:middle line:90% created all these jobs. 00:39:21.970 --> 00:39:22.670 align:middle line:90% What about that? 00:39:22.670 --> 00:39:24.410 align:middle line:84% What are the value of creating the jobs? 00:39:24.410 --> 00:39:26.970 align:middle line:90% What would your answer be? 00:39:26.970 --> 00:39:29.410 align:middle line:84% Why would you explain that's wrong? 00:39:29.410 --> 00:39:30.330 align:middle line:90% Yeah. 00:39:30.330 --> 00:39:33.330 align:middle line:84% AUDIENCE: We already accounted for the opportunity cost. 00:39:33.330 --> 00:39:35.430 align:middle line:84% JON GRUBER: OK, that's kind of a nerdy answer. 00:39:35.430 --> 00:39:39.090 align:middle line:84% Like, they'll be like, OK, I don't understand you. 00:39:39.090 --> 00:39:42.070 align:middle line:84% Well, how would you answer them in more intuitive terms? 00:39:42.070 --> 00:39:42.570 align:middle line:90% Yeah. 00:39:42.570 --> 00:39:44.530 align:middle line:84% AUDIENCE: They would have found a job elsewhere. 00:39:44.530 --> 00:39:46.863 align:middle line:84% JON GRUBER: Yeah, they would have found a job elsewhere, 00:39:46.863 --> 00:39:50.290 align:middle line:84% not creating a job or taking a job from somewhere else. 00:39:50.290 --> 00:39:52.090 align:middle line:84% Or, for example, if someone said, 00:39:52.090 --> 00:39:54.850 align:middle line:84% yeah, well, not only did you save time. 00:39:54.850 --> 00:39:57.443 align:middle line:84% You raised the value of houses, because now it's 00:39:57.443 --> 00:39:59.110 align:middle line:84% quicker to get back and forth from work. 00:39:59.110 --> 00:40:02.260 align:middle line:90% Why shouldn't that be counted? 00:40:02.260 --> 00:40:04.820 align:middle line:84% Why shouldn't you count raising the value of time? 00:40:04.820 --> 00:40:05.320 align:middle line:90% I'm sorry. 00:40:05.320 --> 00:40:06.340 align:middle line:84% Why shouldn't you count raising the value 00:40:06.340 --> 00:40:07.520 align:middle line:90% of houses along the highway? 00:40:07.520 --> 00:40:10.520 align:middle line:90% 00:40:10.520 --> 00:40:11.880 align:middle line:90% And the answer is you shouldn't. 00:40:11.880 --> 00:40:14.060 align:middle line:90% Why not? 00:40:14.060 --> 00:40:16.240 align:middle line:84% Think about why those houses are worth more. 00:40:16.240 --> 00:40:20.500 align:middle line:90% 00:40:20.500 --> 00:40:23.660 align:middle line:90% Why are those houses worth more? 00:40:23.660 --> 00:40:24.713 align:middle line:90% Yeah, Alec. 00:40:24.713 --> 00:40:26.380 align:middle line:84% AUDIENCE: Because of driving time saved, 00:40:26.380 --> 00:40:26.980 align:middle line:90% which is already included. 00:40:26.980 --> 00:40:28.080 align:middle line:84% JON GRUBER: It would be double counting. 00:40:28.080 --> 00:40:29.247 align:middle line:90% You've already counted that. 00:40:29.247 --> 00:40:30.997 align:middle line:84% They're worth more because you saved time. 00:40:30.997 --> 00:40:32.380 align:middle line:90% You already counted saving time. 00:40:32.380 --> 00:40:35.740 align:middle line:84% These are the kinds of mistakes that people commonly 00:40:35.740 --> 00:40:39.580 align:middle line:84% make in this that we need to use our economic intuition to avoid. 00:40:39.580 --> 00:40:43.940 align:middle line:84% Now, the last point I want to come to-- 00:40:43.940 --> 00:40:46.100 align:middle line:84% I'll discuss this early in my office hours-- 00:40:46.100 --> 00:40:48.961 align:middle line:90% is distributional concerns. 00:40:48.961 --> 00:40:50.586 align:middle line:84% This course so far has been efficiency, 00:40:50.586 --> 00:40:51.480 align:middle line:90% efficiency, efficiency. 00:40:51.480 --> 00:40:52.500 align:middle line:84% And we don't care rich versus poor. 00:40:52.500 --> 00:40:54.660 align:middle line:84% We discussed a little bit here, but largely, we 00:40:54.660 --> 00:40:56.040 align:middle line:90% haven't talked about that. 00:40:56.040 --> 00:40:58.260 align:middle line:84% But distributional concerns can be really critical 00:40:58.260 --> 00:40:59.720 align:middle line:84% when doing cost-benefit analysis, 00:40:59.720 --> 00:41:01.137 align:middle line:84% because the costs and benefits can 00:41:01.137 --> 00:41:03.240 align:middle line:84% accrue to very different types of people. 00:41:03.240 --> 00:41:05.740 align:middle line:84% For example, a lot of the most serious environmental damage 00:41:05.740 --> 00:41:07.740 align:middle line:84% in our country has been done to some of the most 00:41:07.740 --> 00:41:10.180 align:middle line:90% disadvantaged communities. 00:41:10.180 --> 00:41:12.680 align:middle line:84% So if we said, Should we clean up this environmental damage? 00:41:12.680 --> 00:41:15.300 align:middle line:84% we might say, well, overall, it's not worth it. 00:41:15.300 --> 00:41:16.495 align:middle line:90% Or we might even say more. 00:41:16.495 --> 00:41:19.120 align:middle line:84% Well, it's certainly not worth it, because they're poor people, 00:41:19.120 --> 00:41:21.300 align:middle line:90% so their lives are worth less. 00:41:21.300 --> 00:41:23.900 align:middle line:90% But that's worrisome. 00:41:23.900 --> 00:41:26.340 align:middle line:84% How do we deal with that in a rigorous way? 00:41:26.340 --> 00:41:29.500 align:middle line:84% What economists do is they say we have distributional weights 00:41:29.500 --> 00:41:30.960 align:middle line:90% or social welfare weights. 00:41:30.960 --> 00:41:33.940 align:middle line:84% Remember, let's go back to our standard utilitarian social 00:41:33.940 --> 00:41:35.100 align:middle line:90% welfare model. 00:41:35.100 --> 00:41:37.220 align:middle line:84% In our standard utilitarian social welfare model, 00:41:37.220 --> 00:41:39.660 align:middle line:84% we want to redistribute income until everyone's 00:41:39.660 --> 00:41:43.020 align:middle line:84% marginal utility of income is the same. 00:41:43.020 --> 00:41:45.020 align:middle line:84% Well, poor people have a higher marginal utility 00:41:45.020 --> 00:41:46.580 align:middle line:90% of income than rich people. 00:41:46.580 --> 00:41:49.660 align:middle line:84% So that would say we want to redistribute from rich to poor. 00:41:49.660 --> 00:41:52.660 align:middle line:84% That would say that a project which taxes the rich 00:41:52.660 --> 00:41:56.700 align:middle line:84% and distributes the poor should get a special plus up 00:41:56.700 --> 00:41:58.710 align:middle line:84% relative to a project which is just tax the rich 00:41:58.710 --> 00:42:00.793 align:middle line:84% and benefits the rich, or even worse, tax the poor 00:42:00.793 --> 00:42:02.590 align:middle line:90% and benefits the rich. 00:42:02.590 --> 00:42:05.430 align:middle line:84% That's what we call a distributional weight. 00:42:05.430 --> 00:42:08.270 align:middle line:84% Distributional weight, W-E-I-G-H-T, 00:42:08.270 --> 00:42:12.310 align:middle line:84% would be that basically, you're going to put extra value based 00:42:12.310 --> 00:42:15.670 align:middle line:84% on who's helped and who pays for this project. 00:42:15.670 --> 00:42:18.510 align:middle line:84% And that value would be weights which are higher on the poorer 00:42:18.510 --> 00:42:21.365 align:middle line:84% people because in a utilitarian social welfare function, 00:42:21.365 --> 00:42:23.490 align:middle line:84% they have a higher marginal utility of consumption, 00:42:23.490 --> 00:42:25.990 align:middle line:84% so valuing resources then more highly. 00:42:25.990 --> 00:42:26.770 align:middle line:90% Yeah, Steven. 00:42:26.770 --> 00:42:32.252 align:middle line:84% AUDIENCE: [INAUDIBLE] highways [INAUDIBLE]. 00:42:32.252 --> 00:42:33.710 align:middle line:84% I remember a city where the highway 00:42:33.710 --> 00:42:37.130 align:middle line:84% was built over a historically marginalized community. 00:42:37.130 --> 00:42:41.470 align:middle line:84% And crime there is an issue there that-- 00:42:41.470 --> 00:42:43.350 align:middle line:84% there's crime in that area of the city that 00:42:43.350 --> 00:42:46.970 align:middle line:84% may have been exacerbated by having that project over them, 00:42:46.970 --> 00:42:47.497 align:middle line:90% for example. 00:42:47.497 --> 00:42:49.330 align:middle line:84% When you have lives saved numbers like this, 00:42:49.330 --> 00:42:51.288 align:middle line:84% I imagine that there's not-- it's hard for them 00:42:51.288 --> 00:42:52.650 align:middle line:90% to capture that information. 00:42:52.650 --> 00:42:54.230 align:middle line:90% [INAUDIBLE] 00:42:54.230 --> 00:42:58.578 align:middle line:84% JON GRUBER: Yeah, no, I mean, this is once again-- once again, 00:42:58.578 --> 00:43:00.120 align:middle line:84% you can look at any problem like this 00:43:00.120 --> 00:43:01.940 align:middle line:84% and view it as an opportunity or challenge. 00:43:01.940 --> 00:43:03.410 align:middle line:84% You can say, God, this is really fucking hard 00:43:03.410 --> 00:43:04.993 align:middle line:84% and throw your hands up and walk away. 00:43:04.993 --> 00:43:07.040 align:middle line:84% Or you could say, that's really cool. 00:43:07.040 --> 00:43:09.540 align:middle line:84% Let me try to add that into my analysis and figure that out. 00:43:09.540 --> 00:43:11.393 align:middle line:84% That's not been done well, but it should be. 00:43:11.393 --> 00:43:13.560 align:middle line:84% We should look at all the benefits and all the costs 00:43:13.560 --> 00:43:14.948 align:middle line:90% and to whom they're attributed. 00:43:14.948 --> 00:43:17.240 align:middle line:84% And that would be exactly the kind of interesting thing 00:43:17.240 --> 00:43:19.400 align:middle line:84% to build in these kind of cost-benefit analyses. 00:43:19.400 --> 00:43:19.900 align:middle line:90% Yeah. 00:43:19.900 --> 00:43:22.200 align:middle line:84% AUDIENCE: [INAUDIBLE] if you assume 00:43:22.200 --> 00:43:24.960 align:middle line:84% that there is some marginal damage to society 00:43:24.960 --> 00:43:26.770 align:middle line:90% by having income inequality? 00:43:26.770 --> 00:43:28.520 align:middle line:84% JON GRUBER: We're going to talk in chapter 00:43:28.520 --> 00:43:30.760 align:middle line:84% 17 more about theories of income inequality. 00:43:30.760 --> 00:43:33.500 align:middle line:84% But essentially, there are many theories of income inequality 00:43:33.500 --> 00:43:35.520 align:middle line:84% which lead you to the same spot, which 00:43:35.520 --> 00:43:38.008 align:middle line:84% is from today's perspective, we'd 00:43:38.008 --> 00:43:39.800 align:middle line:84% be better off taking a dollar from rich guy 00:43:39.800 --> 00:43:42.503 align:middle line:84% and giving it to a poor guy, and there's many ways to get there. 00:43:42.503 --> 00:43:43.920 align:middle line:84% But at the end of the day, you get 00:43:43.920 --> 00:43:47.120 align:middle line:84% to that same spot, which fundamentally most Americans-- 00:43:47.120 --> 00:43:49.640 align:middle line:84% not all, far from all-- but most Americans 00:43:49.640 --> 00:43:53.360 align:middle line:84% would agree that all else equal, taking a dollar from a rich guy, 00:43:53.360 --> 00:43:57.240 align:middle line:84% giving it to a poor guy would make society better off. 00:43:57.240 --> 00:44:01.800 align:middle line:84% And we'll talk at length about that in chapter 17. 00:44:01.800 --> 00:44:05.280 align:middle line:84% Now, speaking of difficulty of getting Americans 00:44:05.280 --> 00:44:08.640 align:middle line:84% to agree on things, let's now turn to chapter 9 00:44:08.640 --> 00:44:11.080 align:middle line:90% and political economy. 00:44:11.080 --> 00:44:13.720 align:middle line:84% As I said, the fourth fundamental question-- 00:44:13.720 --> 00:44:16.040 align:middle line:84% in the first lecture, the fourth fundamental question 00:44:16.040 --> 00:44:19.160 align:middle line:84% about economics is, why do governments do what they do? 00:44:19.160 --> 00:44:21.880 align:middle line:84% At one level, this is the ultimate overarching question 00:44:21.880 --> 00:44:24.520 align:middle line:84% of public finance, which is I spend 00:44:24.520 --> 00:44:28.080 align:middle line:84% this whole lecture assuming some benevolent social planner. 00:44:28.080 --> 00:44:31.200 align:middle line:84% And it's hard to teach this chapter without revealing 00:44:31.200 --> 00:44:33.120 align:middle line:90% one's preferences. 00:44:33.120 --> 00:44:35.440 align:middle line:84% But I certainly think it's hard to imagine Donald Trump 00:44:35.440 --> 00:44:37.160 align:middle line:90% as a benevolent social planner. 00:44:37.160 --> 00:44:40.040 align:middle line:84% You may like or dislike him, but certainly, that would not 00:44:40.040 --> 00:44:41.800 align:middle line:90% be your description of him. 00:44:41.800 --> 00:44:47.320 align:middle line:84% So basically, the question is, how do we think about a world 00:44:47.320 --> 00:44:50.340 align:middle line:84% where decisions are not made by some theoretical social planner 00:44:50.340 --> 00:44:51.598 align:middle line:90% but by real people? 00:44:51.598 --> 00:44:53.890 align:middle line:84% And that's the field of political economy, which builds 00:44:53.890 --> 00:44:55.588 align:middle line:90% in government decision-making. 00:44:55.588 --> 00:44:57.130 align:middle line:84% Now I'll only do a lecture and a half 00:44:57.130 --> 00:44:59.188 align:middle line:84% on this, which is very unfortunate. 00:44:59.188 --> 00:45:01.730 align:middle line:84% Fortunately, our department has the best political economists 00:45:01.730 --> 00:45:03.670 align:middle line:90% in the world teaching here. 00:45:03.670 --> 00:45:07.142 align:middle line:84% Daron Acemoglu, who wrote the book Why Nations Fail, probably 00:45:07.142 --> 00:45:08.850 align:middle line:84% the most important political economy book 00:45:08.850 --> 00:45:11.490 align:middle line:84% ever written, or at least in the last century. 00:45:11.490 --> 00:45:15.250 align:middle line:84% Ben Olken, who does amazing work on how we think about governance 00:45:15.250 --> 00:45:16.850 align:middle line:90% in developing countries. 00:45:16.850 --> 00:45:18.490 align:middle line:84% Abhijit Banerjee, Nobel Prize winner 00:45:18.490 --> 00:45:21.970 align:middle line:84% who's done incredible work about the theory of political economy. 00:45:21.970 --> 00:45:24.570 align:middle line:84% We have incredible resource in our department, 00:45:24.570 --> 00:45:26.602 align:middle line:84% and there's multiple courses you can take. 00:45:26.602 --> 00:45:28.310 align:middle line:84% So I urge you, if you want to learn more, 00:45:28.310 --> 00:45:29.227 align:middle line:90% to take those classes. 00:45:29.227 --> 00:45:32.090 align:middle line:84% Unfortunately, here, I only have about a lecture and a half. 00:45:32.090 --> 00:45:35.590 align:middle line:84% So here's how we're going to think about political economy. 00:45:35.590 --> 00:45:38.330 align:middle line:84% Next time, we'll do the fun gossipy stuff. 00:45:38.330 --> 00:45:40.950 align:middle line:84% Now we're going to do the more rigorous economic stuff. 00:45:40.950 --> 00:45:42.870 align:middle line:84% Here's how we're going to think about political economy. 00:45:42.870 --> 00:45:44.620 align:middle line:84% How we're going to think about it is we're 00:45:44.620 --> 00:45:46.690 align:middle line:84% going to proceed in several steps. 00:45:46.690 --> 00:45:50.770 align:middle line:84% The first step is going to be, imagine a theoretical world 00:45:50.770 --> 00:45:53.500 align:middle line:84% where we could implement the perfect form 00:45:53.500 --> 00:45:55.740 align:middle line:90% of political economy. 00:45:55.740 --> 00:45:59.500 align:middle line:84% Then we'll say, OK, that doesn't exist. 00:45:59.500 --> 00:46:02.900 align:middle line:84% A second alternative would be, imagine a world where 00:46:02.900 --> 00:46:05.040 align:middle line:84% literally everybody, every American, 00:46:05.040 --> 00:46:08.980 align:middle line:84% voted to choose every item, no politicians, 00:46:08.980 --> 00:46:11.540 align:middle line:84% or what we call direct democracy, 00:46:11.540 --> 00:46:14.978 align:middle line:84% which occurs in some cases but not in most. 00:46:14.978 --> 00:46:16.520 align:middle line:84% And then third, and most interesting, 00:46:16.520 --> 00:46:19.540 align:middle line:84% we'll talk next time about representative democracy, which 00:46:19.540 --> 00:46:21.540 align:middle line:84% is what we have and most countries have, 00:46:21.540 --> 00:46:23.415 align:middle line:84% where we don't literally make every decision. 00:46:23.415 --> 00:46:25.060 align:middle line:84% We elect people who make the decisions. 00:46:25.060 --> 00:46:27.102 align:middle line:84% So that's the three steps we're going to proceed. 00:46:27.102 --> 00:46:30.100 align:middle line:84% And we're going to start with an example of what we call 00:46:30.100 --> 00:46:33.220 align:middle line:90% a model of unanimous consent. 00:46:33.220 --> 00:46:37.020 align:middle line:84% So this first model is a model of unanimous consent, which 00:46:37.020 --> 00:46:40.060 align:middle line:84% is, imagine a world where we can only 00:46:40.060 --> 00:46:43.940 align:middle line:90% do things if everyone agrees. 00:46:43.940 --> 00:46:47.980 align:middle line:84% And imagine a world where we could actually make that happen. 00:46:47.980 --> 00:46:49.940 align:middle line:90% We get everyone to agree. 00:46:49.940 --> 00:46:53.950 align:middle line:84% That is a world of what we use-- what we call Lindahl pricing, 00:46:53.950 --> 00:46:57.390 align:middle line:84% named after a Swedish economist who developed this model 00:46:57.390 --> 00:47:02.550 align:middle line:84% in the early 20th century, where he basically said, look, 00:47:02.550 --> 00:47:07.190 align:middle line:84% as long as we charge every individual their marginal 00:47:07.190 --> 00:47:09.670 align:middle line:84% willingness to pay for a public good, 00:47:09.670 --> 00:47:12.950 align:middle line:90% we can get unanimous consent. 00:47:12.950 --> 00:47:17.590 align:middle line:84% To see that, let's go to figure 9-1, 00:47:17.590 --> 00:47:23.270 align:middle line:84% which is a replication back from an earlier-- 00:47:23.270 --> 00:47:26.570 align:middle line:84% this replication example from an earlier chapter, chapter 7, 00:47:26.570 --> 00:47:28.713 align:middle line:90% which was demand for fireworks. 00:47:28.713 --> 00:47:31.130 align:middle line:84% And we talked about marginal willingness to pay fireworks. 00:47:31.130 --> 00:47:32.870 align:middle line:90% Back then, it was Ben and Jerry. 00:47:32.870 --> 00:47:34.410 align:middle line:90% Now it's Ava and Jack. 00:47:34.410 --> 00:47:36.070 align:middle line:84% So the funny thing is when I wrote 00:47:36.070 --> 00:47:40.805 align:middle line:84% this textbook 20 years ago, 22 years ago, 00:47:40.805 --> 00:47:42.930 align:middle line:84% every example involved the names of one of my kids. 00:47:42.930 --> 00:47:45.553 align:middle line:84% I've now had to change it because that's not really fair. 00:47:45.553 --> 00:47:47.470 align:middle line:84% And I was very uptight about my kids being mad 00:47:47.470 --> 00:47:48.950 align:middle line:90% at me, Ava, Jack, and Sam. 00:47:48.950 --> 00:47:50.090 align:middle line:90% And Sam is now Rachel. 00:47:50.090 --> 00:47:53.190 align:middle line:84% But Tom was Sam, and I was very upset about my kids 00:47:53.190 --> 00:47:55.430 align:middle line:84% being mad at me, so I carefully went through and made 00:47:55.430 --> 00:47:57.110 align:middle line:84% sure I used every kid's names exactly 00:47:57.110 --> 00:47:58.165 align:middle line:90% the same number of times. 00:47:58.165 --> 00:47:59.790 align:middle line:84% And the ironic part is, none of my kids 00:47:59.790 --> 00:48:01.288 align:middle line:90% have ever opened the book. 00:48:01.288 --> 00:48:02.830 align:middle line:84% So I spent all that time for nothing. 00:48:02.830 --> 00:48:04.530 align:middle line:84% Anyway, Ava and Jack or two of my kids. 00:48:04.530 --> 00:48:06.910 align:middle line:84% And let's imagine that they have a willingness 00:48:06.910 --> 00:48:08.167 align:middle line:90% to pay for fireworks. 00:48:08.167 --> 00:48:09.750 align:middle line:84% And let's say Ava's willingness to pay 00:48:09.750 --> 00:48:12.070 align:middle line:84% is depicted in the first-- in panel a. 00:48:12.070 --> 00:48:13.390 align:middle line:90% Jack's in panel b. 00:48:13.390 --> 00:48:16.550 align:middle line:84% And the total social willingness to pay, in panel C. 00:48:16.550 --> 00:48:18.470 align:middle line:84% Just like we do with Ben and Jerry, 00:48:18.470 --> 00:48:20.430 align:middle line:84% it's their willingness to pay for fireworks. 00:48:20.430 --> 00:48:24.733 align:middle line:84% OK, so Ava's willing to pay $1 for the first firework. 00:48:24.733 --> 00:48:26.650 align:middle line:84% And by the time you get to the 100th firework, 00:48:26.650 --> 00:48:28.510 align:middle line:90% it's worth nothing to her. 00:48:28.510 --> 00:48:31.310 align:middle line:84% Jack's willing to pay $3 for the first firework, 00:48:31.310 --> 00:48:34.270 align:middle line:84% but also the time you get to 100 fireworks, worth nothing to him. 00:48:34.270 --> 00:48:36.870 align:middle line:84% And so society as a whole is willing to pay 00:48:36.870 --> 00:48:41.710 align:middle line:84% $4 for the first firework, with its diminishing demand. 00:48:41.710 --> 00:48:44.823 align:middle line:84% And if the marginal cost of fireworks is $1, 00:48:44.823 --> 00:48:47.240 align:middle line:84% then the optimal number of fireworks to provide is provide 00:48:47.240 --> 00:48:47.980 align:middle line:90% is 75. 00:48:47.980 --> 00:48:49.438 align:middle line:84% That's the graphical implementation 00:48:49.438 --> 00:48:52.120 align:middle line:84% of the Samuelson rule, the sum of the marginal willingness 00:48:52.120 --> 00:48:54.280 align:middle line:90% to pay. 00:48:54.280 --> 00:48:55.995 align:middle line:90% So here's what Lindahl said. 00:48:55.995 --> 00:48:57.620 align:middle line:84% What if we set up the following system? 00:48:57.620 --> 00:48:58.540 align:middle line:90% It's in your textbook. 00:48:58.540 --> 00:48:59.740 align:middle line:90% So just talk it through. 00:48:59.740 --> 00:49:01.120 align:middle line:90% It's hard to write it all down. 00:49:01.120 --> 00:49:05.240 align:middle line:84% Imagine the government says, we're 00:49:05.240 --> 00:49:08.800 align:middle line:84% going to set a set of tax prices for fireworks. 00:49:08.800 --> 00:49:10.880 align:middle line:84% For every firework we choose, we're 00:49:10.880 --> 00:49:14.240 align:middle line:84% going to charge each person a certain amount of money. 00:49:14.240 --> 00:49:15.460 align:middle line:90% Fireworks cost $1 each. 00:49:15.460 --> 00:49:18.385 align:middle line:84% It could be $0.50 each, 90/10, whatever it is. 00:49:18.385 --> 00:49:19.760 align:middle line:84% We're going to charge each person 00:49:19.760 --> 00:49:22.060 align:middle line:90% a tax price for each firework. 00:49:22.060 --> 00:49:23.260 align:middle line:90% It's going to add up to $1. 00:49:23.260 --> 00:49:24.180 align:middle line:90% That's the only constraint. 00:49:24.180 --> 00:49:24.920 align:middle line:90% Fireworks are $1. 00:49:24.920 --> 00:49:26.380 align:middle line:90% We're not going to make money. 00:49:26.380 --> 00:49:27.380 align:middle line:90% It's going to add to $1. 00:49:27.380 --> 00:49:29.793 align:middle line:90% But we set these tax prices. 00:49:29.793 --> 00:49:31.960 align:middle line:84% Each individual would then say, well, at that price, 00:49:31.960 --> 00:49:34.548 align:middle line:84% this is how many fireworks I want. 00:49:34.548 --> 00:49:36.840 align:middle line:84% Well, the government, by doing that over and over again 00:49:36.840 --> 00:49:39.543 align:middle line:84% can essentially trace out the demand curve. 00:49:39.543 --> 00:49:41.460 align:middle line:84% So I go to people and say over and over again, 00:49:41.460 --> 00:49:44.240 align:middle line:84% how many fireworks do you want at $0.10, at $0.20, at $0.30? 00:49:44.240 --> 00:49:46.230 align:middle line:90% I trace out the demand curve. 00:49:46.230 --> 00:49:48.728 align:middle line:84% Once I have the domain curve and know marginal cost, 00:49:48.728 --> 00:49:50.770 align:middle line:84% I can figure out the optimal number of fireworks. 00:49:50.770 --> 00:49:52.170 align:middle line:84% And then I can charge people what 00:49:52.170 --> 00:49:53.770 align:middle line:90% they said it was worth to them. 00:49:53.770 --> 00:49:56.590 align:middle line:84% So in this case, if the optimal number of fireworks is 75, 00:49:56.590 --> 00:49:58.890 align:middle line:90% I say, look, it's 75 fireworks. 00:49:58.890 --> 00:50:02.130 align:middle line:90% Fireworks cost $1. 00:50:02.130 --> 00:50:03.270 align:middle line:90% Fireworks cost $1. 00:50:03.270 --> 00:50:04.938 align:middle line:90% So that's $75. 00:50:04.938 --> 00:50:06.230 align:middle line:90% How am I going to pay for that? 00:50:06.230 --> 00:50:08.130 align:middle line:84% Well, Eva said, for 75 fireworks, 00:50:08.130 --> 00:50:11.650 align:middle line:90% she's willing to pay $0.25 each. 00:50:11.650 --> 00:50:15.210 align:middle line:84% Jack said, for 75 fireworks, willing to pay $0.75 each. 00:50:15.210 --> 00:50:20.010 align:middle line:84% So I charge Jack $0.75 on 75 fireworks, so like $49. 00:50:20.010 --> 00:50:23.790 align:middle line:84% I charge Eva $0.25 for 75 fireworks. 00:50:23.790 --> 00:50:25.050 align:middle line:90% It's like $21. 00:50:25.050 --> 00:50:27.110 align:middle line:90% And it adds up-- or $26. 00:50:27.110 --> 00:50:30.550 align:middle line:84% It adds up to the $75, whatever the numbers are. 00:50:30.550 --> 00:50:31.970 align:middle line:90% I forget what I have here. 00:50:31.970 --> 00:50:37.330 align:middle line:84% I charged Jack-- so I charge-- whatever, I just 00:50:37.330 --> 00:50:38.350 align:middle line:90% charge the amount. 00:50:38.350 --> 00:50:44.620 align:middle line:84% So basically, this delivers the Samuelson equilibrium. 00:50:44.620 --> 00:50:49.380 align:middle line:84% Each person you are adding up to the proper number of fireworks. 00:50:49.380 --> 00:50:52.700 align:middle line:84% You're covering your costs, and everybody is paying 00:50:52.700 --> 00:50:55.620 align:middle line:90% their true marginal value. 00:50:55.620 --> 00:50:56.980 align:middle line:90% You're done. 00:50:56.980 --> 00:50:59.860 align:middle line:84% And there's no reason to deviate. 00:50:59.860 --> 00:51:02.040 align:middle line:84% Ava and Jack are both perfectly happy. 00:51:02.040 --> 00:51:05.200 align:middle line:84% They'd be happier paying less, but they're on the domain curve, 00:51:05.200 --> 00:51:07.020 align:middle line:90% so they're happy. 00:51:07.020 --> 00:51:09.500 align:middle line:84% So you've achieved the first best. 00:51:09.500 --> 00:51:13.460 align:middle line:84% So that would be the ideal political economy system, 00:51:13.460 --> 00:51:17.380 align:middle line:84% would be we could get the first best Samuelson equilibrium. 00:51:17.380 --> 00:51:19.940 align:middle line:90% Now, in fact, we can't do that. 00:51:19.940 --> 00:51:24.165 align:middle line:84% And we can't do that for the three reasons I laid out. 00:51:24.165 --> 00:51:26.040 align:middle line:84% Was it last lecture or a couple lectures ago? 00:51:26.040 --> 00:51:27.780 align:middle line:90% I think it was last lecture. 00:51:27.780 --> 00:51:33.620 align:middle line:84% The first reason is the preference revelation problem, 00:51:33.620 --> 00:51:35.700 align:middle line:84% which is if I set up this system, 00:51:35.700 --> 00:51:37.980 align:middle line:90% people have an incentive to lie. 00:51:37.980 --> 00:51:39.980 align:middle line:90% Let's work through the math. 00:51:39.980 --> 00:51:44.140 align:middle line:84% Imagine that Jack lied and said his preferences were the same 00:51:44.140 --> 00:51:46.420 align:middle line:90% as Ava's. 00:51:46.420 --> 00:51:48.680 align:middle line:84% In that case, we would say, well, 00:51:48.680 --> 00:51:51.860 align:middle line:84% if Jack and Ava were the same, we would end up producing 50 00:51:51.860 --> 00:51:53.860 align:middle line:90% fireworks at $0.50 each. 00:51:53.860 --> 00:51:56.620 align:middle line:84% If there were two Avas, the optimal number of fireworks is 00:51:56.620 --> 00:51:59.060 align:middle line:90% 50 fireworks, $0.50 each. 00:51:59.060 --> 00:52:00.320 align:middle line:90% What would that mean? 00:52:00.320 --> 00:52:05.140 align:middle line:84% That would mean that what Ava pays-- 00:52:05.140 --> 00:52:11.860 align:middle line:84% Ava used to pay $0.25 for 75 fireworks, or $18.75. 00:52:11.860 --> 00:52:16.900 align:middle line:84% She used to pay $18.75 and get the same 75 fireworks Jack got. 00:52:16.900 --> 00:52:18.080 align:middle line:90% Now she pays more. 00:52:18.080 --> 00:52:20.000 align:middle line:84% She pays $25 and gets fewer fireworks. 00:52:20.000 --> 00:52:21.540 align:middle line:90% She gets 50. 00:52:21.540 --> 00:52:25.800 align:middle line:84% Now, Jack gets fewer too, but he doesn't pay that much less. 00:52:25.800 --> 00:52:27.680 align:middle line:84% Jack gets fewer too, but he pays a ton less. 00:52:27.680 --> 00:52:28.460 align:middle line:90% I'm sorry. 00:52:28.460 --> 00:52:34.940 align:middle line:84% Jack used to pay $56.25 for 75 fireworks. 00:52:34.940 --> 00:52:38.100 align:middle line:84% Now he pays $25 for 50 fireworks, 00:52:38.100 --> 00:52:41.310 align:middle line:84% so he gets fewer fireworks, but he pays a ton less. 00:52:41.310 --> 00:52:43.270 align:middle line:90% What's happened here? 00:52:43.270 --> 00:52:46.390 align:middle line:84% Jack has become a free rider on Ava. 00:52:46.390 --> 00:52:49.110 align:middle line:84% By lying about his preferences, he's made-- 00:52:49.110 --> 00:52:51.430 align:middle line:90% forced Ava to contribute more. 00:52:51.430 --> 00:52:52.370 align:middle line:90% He's hurt himself. 00:52:52.370 --> 00:52:54.670 align:middle line:84% He gets fewer fireworks, but he's paying less than half 00:52:54.670 --> 00:52:57.750 align:middle line:84% as much to get 3/4-- to get 2/3 of many fireworks. 00:52:57.750 --> 00:53:00.430 align:middle line:84% Ava is paying more to get fewer fireworks. 00:53:00.430 --> 00:53:02.830 align:middle line:84% So by lying, he's managed to make himself better off 00:53:02.830 --> 00:53:04.990 align:middle line:90% at Ava's expense. 00:53:04.990 --> 00:53:07.668 align:middle line:84% So there's a huge incentive to lie. 00:53:07.668 --> 00:53:08.710 align:middle line:90% That's the first problem. 00:53:08.710 --> 00:53:11.230 align:middle line:90% People get that? 00:53:11.230 --> 00:53:12.630 align:middle line:90% Huge incentive to lie. 00:53:12.630 --> 00:53:16.390 align:middle line:84% Second problem is people may not know. 00:53:16.390 --> 00:53:21.190 align:middle line:84% We don't know what things are worth. 00:53:21.190 --> 00:53:25.710 align:middle line:84% People may not know what fireworks are worth. 00:53:25.710 --> 00:53:30.950 align:middle line:84% And the third problem is aggregation, 00:53:30.950 --> 00:53:34.448 align:middle line:84% which is this is all well and good in a two-person society. 00:53:34.448 --> 00:53:36.990 align:middle line:84% But if we had to collect every person in America's preference 00:53:36.990 --> 00:53:41.017 align:middle line:84% for every single public good we provide at every possible price, 00:53:41.017 --> 00:53:43.600 align:middle line:84% the entire population would work for the government collecting 00:53:43.600 --> 00:53:45.580 align:middle line:90% that data from each other. 00:53:45.580 --> 00:53:47.480 align:middle line:90% It wouldn't work. 00:53:47.480 --> 00:53:50.720 align:middle line:84% So this system is a great theoretical benchmark 00:53:50.720 --> 00:53:53.840 align:middle line:84% of how you achieve the unanimous first best, 00:53:53.840 --> 00:53:55.460 align:middle line:90% but it's irrelevant in practice. 00:53:55.460 --> 00:53:58.978 align:middle line:90% 00:53:58.978 --> 00:54:01.520 align:middle line:84% What it does highlight-- and we'll come back to this in later 00:54:01.520 --> 00:54:02.360 align:middle line:90% lectures-- 00:54:02.360 --> 00:54:04.920 align:middle line:84% is that decisions will be more popular, 00:54:04.920 --> 00:54:07.920 align:middle line:84% the extent to which they can link the taxes people pay 00:54:07.920 --> 00:54:09.880 align:middle line:90% and the benefits they get. 00:54:09.880 --> 00:54:11.760 align:middle line:84% One way to make a decision popular-- and this 00:54:11.760 --> 00:54:14.177 align:middle line:84% will come to explain a lot of decisions governments make-- 00:54:14.177 --> 00:54:17.262 align:middle line:84% is try to link to get closer to Lindahl equilibrium 00:54:17.262 --> 00:54:18.720 align:middle line:84% by linking what people actually pay 00:54:18.720 --> 00:54:20.180 align:middle line:84% and the benefits they actually get. 00:54:20.180 --> 00:54:23.440 align:middle line:90% 00:54:23.440 --> 00:54:24.825 align:middle line:90% Yeah. 00:54:24.825 --> 00:54:26.700 align:middle line:84% AUDIENCE: I think that makes sense in theory. 00:54:26.700 --> 00:54:28.242 align:middle line:84% But in practice, wouldn't most people 00:54:28.242 --> 00:54:30.280 align:middle line:84% say that they would want lower taxes if that 00:54:30.280 --> 00:54:32.420 align:middle line:84% meant not spending more money on roads, 00:54:32.420 --> 00:54:34.240 align:middle line:90% even though we all benefit. 00:54:34.240 --> 00:54:36.860 align:middle line:84% But people behaviorally are going to say, 00:54:36.860 --> 00:54:37.950 align:middle line:90% I want lower taxes. 00:54:37.950 --> 00:54:38.700 align:middle line:90% JON GRUBER: Right. 00:54:38.700 --> 00:54:40.820 align:middle line:84% So basically, first of all, it's a couple of things. 00:54:40.820 --> 00:54:42.237 align:middle line:84% Maybe they don't realize what it's 00:54:42.237 --> 00:54:44.160 align:middle line:84% going to be like to have shitty roads. 00:54:44.160 --> 00:54:45.975 align:middle line:84% Second of all, they may be lying. 00:54:45.975 --> 00:54:48.100 align:middle line:84% And third of all, how do you put that all together? 00:54:48.100 --> 00:54:50.440 align:middle line:84% Once again, the tax benefit linkage would be, 00:54:50.440 --> 00:54:53.480 align:middle line:84% how do you raise money in a way that 00:54:53.480 --> 00:54:56.960 align:middle line:84% more directly links the benefits of better roads 00:54:56.960 --> 00:54:59.120 align:middle line:90% to the taxes people pay? 00:54:59.120 --> 00:55:03.200 align:middle line:90% Well, let me ask you a question? 00:55:03.200 --> 00:55:06.005 align:middle line:84% At what level of government do we fund road repairs? 00:55:06.005 --> 00:55:07.880 align:middle line:84% The federal government, the state government, 00:55:07.880 --> 00:55:10.497 align:middle line:90% or the city government? 00:55:10.497 --> 00:55:12.080 align:middle line:84% AUDIENCE: Here, it's going to be city. 00:55:12.080 --> 00:55:13.520 align:middle line:90% JON GRUBER: City. 00:55:13.520 --> 00:55:17.630 align:middle line:84% Why does that help answer your next question? 00:55:17.630 --> 00:55:18.292 align:middle line:90% Alec, yeah. 00:55:18.292 --> 00:55:19.750 align:middle line:84% AUDIENCE: Their taxpayers are going 00:55:19.750 --> 00:55:21.160 align:middle line:84% to be the ones directly benefiting from the roads. 00:55:21.160 --> 00:55:23.827 align:middle line:84% JON GRUBER: Yeah, if I'm taxing you to pay for roads in Alabama, 00:55:23.827 --> 00:55:25.120 align:middle line:90% you're like, screw that. 00:55:25.120 --> 00:55:27.160 align:middle line:84% So basically, that's exactly-- that's why 00:55:27.160 --> 00:55:29.160 align:middle line:84% this model can actually help explain 00:55:29.160 --> 00:55:30.420 align:middle line:90% decisions governments make. 00:55:30.420 --> 00:55:32.960 align:middle line:84% And we'll come back to that in chapter 10. 00:55:32.960 --> 00:55:33.800 align:middle line:90% All right. 00:55:33.800 --> 00:55:35.380 align:middle line:90% So that's the first method. 00:55:35.380 --> 00:55:37.630 align:middle line:84% Well, it has some interesting directional predictions, 00:55:37.630 --> 00:55:38.910 align:middle line:90% obviously relevant in reality. 00:55:38.910 --> 00:55:41.330 align:middle line:84% Second method says, OK, let's move 00:55:41.330 --> 00:55:44.050 align:middle line:84% from unanimous decision-making, which 00:55:44.050 --> 00:55:46.690 align:middle line:84% never works, to the common decision-making mode used 00:55:46.690 --> 00:55:50.010 align:middle line:84% in the world, which is majority decision-making. 00:55:50.010 --> 00:55:52.530 align:middle line:84% And the best way to tell if a majority of people 00:55:52.530 --> 00:55:57.210 align:middle line:84% want something is with direct democracy. 00:55:57.210 --> 00:56:00.810 align:middle line:84% Let's just have everybody vote over everything. 00:56:00.810 --> 00:56:05.610 align:middle line:84% If you vote, you solve the aggregation problem 00:56:05.610 --> 00:56:08.730 align:middle line:90% because you're voting. 00:56:08.730 --> 00:56:13.038 align:middle line:84% And you potentially-- and we'll come back to this-- 00:56:13.038 --> 00:56:15.330 align:middle line:84% solve the revelation problem, because there's no reason 00:56:15.330 --> 00:56:16.413 align:middle line:90% to lie when you're voting. 00:56:16.413 --> 00:56:19.210 align:middle line:84% But we'll come back to why that may not be true. 00:56:19.210 --> 00:56:22.142 align:middle line:84% So direct democracy-- and direct democracy 00:56:22.142 --> 00:56:23.350 align:middle line:90% has a long history in the US. 00:56:23.350 --> 00:56:25.410 align:middle line:84% The first town meeting in the US that's recorded 00:56:25.410 --> 00:56:30.450 align:middle line:84% was held in 1657, where the-- in 1657, the residents 00:56:30.450 --> 00:56:33.130 align:middle line:84% of Huntington, New York, held a meeting 00:56:33.130 --> 00:56:36.500 align:middle line:84% to vote to hire the first schoolmaster of their town. 00:56:36.500 --> 00:56:38.760 align:middle line:84% There's still town meetings held all over New England. 00:56:38.760 --> 00:56:41.343 align:middle line:84% Many decisions in New England are still made by town meetings. 00:56:41.343 --> 00:56:42.247 align:middle line:90% People get together. 00:56:42.247 --> 00:56:44.080 align:middle line:84% And they vote on what they're going to have. 00:56:44.080 --> 00:56:46.540 align:middle line:84% Literally, there's no politician making the decision. 00:56:46.540 --> 00:56:48.620 align:middle line:84% This actually led to an interesting case 00:56:48.620 --> 00:56:50.065 align:middle line:90% in New Hampshire. 00:56:50.065 --> 00:56:51.440 align:middle line:84% The problem with that, of course, 00:56:51.440 --> 00:56:53.247 align:middle line:90% is many people don't show up. 00:56:53.247 --> 00:56:55.080 align:middle line:84% And what happened in a town in New Hampshire 00:56:55.080 --> 00:56:58.420 align:middle line:84% is there was this guy who was a very extreme antigovernment 00:56:58.420 --> 00:56:59.540 align:middle line:90% person. 00:56:59.540 --> 00:57:02.020 align:middle line:84% And he realized in this town that town meetings 00:57:02.020 --> 00:57:03.340 align:middle line:90% were poorly attended. 00:57:03.340 --> 00:57:04.880 align:middle line:84% So he gathered a few of his friends, 00:57:04.880 --> 00:57:06.255 align:middle line:84% and they went to the town meeting 00:57:06.255 --> 00:57:09.580 align:middle line:84% and voted to basically end public financing of all things 00:57:09.580 --> 00:57:10.500 align:middle line:90% in the town. 00:57:10.500 --> 00:57:12.625 align:middle line:84% It basically turned into a purely capitalistic town 00:57:12.625 --> 00:57:17.020 align:middle line:84% with no government, no schools, no fire department, nothing. 00:57:17.020 --> 00:57:17.923 align:middle line:90% People got upset. 00:57:17.923 --> 00:57:20.340 align:middle line:84% They had another town meeting where people actually showed 00:57:20.340 --> 00:57:22.140 align:middle line:90% up and voted him down. 00:57:22.140 --> 00:57:24.320 align:middle line:90% But this is the challenge. 00:57:24.320 --> 00:57:27.000 align:middle line:84% Aggregation only works if people show up. 00:57:27.000 --> 00:57:28.780 align:middle line:84% So that's an example of direct democracy. 00:57:28.780 --> 00:57:31.780 align:middle line:84% More generally, many of you may be from states 00:57:31.780 --> 00:57:34.895 align:middle line:84% where there are ballot referenda or ballot initiatives, which 00:57:34.895 --> 00:57:36.770 align:middle line:84% are when you go to vote for your politicians, 00:57:36.770 --> 00:57:39.390 align:middle line:84% you also vote on a series of questions. 00:57:39.390 --> 00:57:43.710 align:middle line:84% A referendum means that-- is a case where there's 00:57:43.710 --> 00:57:45.790 align:middle line:84% been a law that's already passed, 00:57:45.790 --> 00:57:50.630 align:middle line:84% and citizens can basically ratify or not ratify it. 00:57:50.630 --> 00:57:56.790 align:middle line:84% An initiative is when voters can actually decide an issue. 00:57:56.790 --> 00:58:00.210 align:middle line:84% Now, initiatives are-- the popularity 00:58:00.210 --> 00:58:02.210 align:middle line:84% of these methods have waxed and waned over time. 00:58:02.210 --> 00:58:04.650 align:middle line:90% 24 states allow initiatives. 00:58:04.650 --> 00:58:07.950 align:middle line:90% It started with Oregon in 1904. 00:58:07.950 --> 00:58:11.830 align:middle line:84% More than 2,600 initiatives have been in state ballots. 00:58:11.830 --> 00:58:15.790 align:middle line:84% And about 41% have passed, although 60% of all initiative 00:58:15.790 --> 00:58:17.390 align:middle line:84% activities occurs in only six states-- 00:58:17.390 --> 00:58:20.170 align:middle line:84% Arizona, California, Colorado, North Dakota, 00:58:20.170 --> 00:58:22.315 align:middle line:90% Oregon, and Washington. 00:58:22.315 --> 00:58:24.190 align:middle line:84% The popularity of initiatives waxes and wanes 00:58:24.190 --> 00:58:26.730 align:middle line:90% over time, as do the topics. 00:58:26.730 --> 00:58:29.310 align:middle line:84% So early in their lives, initiatives 00:58:29.310 --> 00:58:31.450 align:middle line:90% were used for things like taxes. 00:58:31.450 --> 00:58:34.250 align:middle line:84% So the most famous, of which we'll discuss in chapter 10, 00:58:34.250 --> 00:58:36.710 align:middle line:84% was the passage in California in 1978 00:58:36.710 --> 00:58:39.990 align:middle line:84% of proposition 13, which limit how much revenue towns could 00:58:39.990 --> 00:58:42.670 align:middle line:90% collect in property taxes. 00:58:42.670 --> 00:58:45.630 align:middle line:90% That was an early initiative. 00:58:45.630 --> 00:58:49.630 align:middle line:84% Then, basically, the earliest-- there 00:58:49.630 --> 00:58:52.830 align:middle line:84% were things like election rules, alcohol regulation, labor laws, 00:58:52.830 --> 00:58:54.930 align:middle line:84% and government, were first things on. 00:58:54.930 --> 00:58:56.190 align:middle line:90% Then there was taxes. 00:58:56.190 --> 00:58:58.230 align:middle line:84% Now, the most popular initiative, of course, 00:58:58.230 --> 00:58:59.950 align:middle line:90% is marijuana legalization. 00:58:59.950 --> 00:59:03.230 align:middle line:84% But there's lots of other things that are on ballots as well. 00:59:03.230 --> 00:59:05.190 align:middle line:84% And for Massachusetts, we've had initiatives 00:59:05.190 --> 00:59:06.630 align:middle line:84% about charter schools, initiatives 00:59:06.630 --> 00:59:08.150 align:middle line:90% about the minimum wage. 00:59:08.150 --> 00:59:09.990 align:middle line:84% Lots of things are decided by initiative. 00:59:09.990 --> 00:59:10.490 align:middle line:90% Yeah. 00:59:10.490 --> 00:59:12.510 align:middle line:84% AUDIENCE: How up to date is marijuana legalization 00:59:12.510 --> 00:59:14.950 align:middle line:84% [INAUDIBLE] most popular, because I feel like abortion may 00:59:14.950 --> 00:59:15.533 align:middle line:90% have certain-- 00:59:15.533 --> 00:59:16.830 align:middle line:90% JON GRUBER: Abortion may have. 00:59:16.830 --> 00:59:19.390 align:middle line:84% Now, by this round, abortion's on a lot of ballots. 00:59:19.390 --> 00:59:21.450 align:middle line:90% That's a common example. 00:59:21.450 --> 00:59:21.970 align:middle line:90% Good point. 00:59:21.970 --> 00:59:23.190 align:middle line:90% Yeah. 00:59:23.190 --> 00:59:26.330 align:middle line:84% AUDIENCE: Are there ever federal referendums or initiatives? 00:59:26.330 --> 00:59:28.310 align:middle line:90% JON GRUBER: No, never. 00:59:28.310 --> 00:59:30.110 align:middle line:90% AUDIENCE: Is there a reason why? 00:59:30.110 --> 00:59:31.900 align:middle line:84% JON GRUBER: It's just not the way the US government is done. 00:59:31.900 --> 00:59:32.900 align:middle line:84% Now, there are in other countries. 00:59:32.900 --> 00:59:34.920 align:middle line:84% Switzerland makes almost all its decisions 00:59:34.920 --> 00:59:37.660 align:middle line:84% by full majority voting on initiatives, 00:59:37.660 --> 00:59:41.360 align:middle line:84% things like what the tax rate should be. 00:59:41.360 --> 00:59:44.560 align:middle line:90% That is not done in the US. 00:59:44.560 --> 00:59:46.040 align:middle line:84% There are state-level initiatives 00:59:46.040 --> 00:59:48.360 align:middle line:84% on things like tax levels but never 00:59:48.360 --> 00:59:49.860 align:middle line:90% done federally that I know of. 00:59:49.860 --> 00:59:53.100 align:middle line:84% I could be misspeaking for the record, but that I know of, 00:59:53.100 --> 00:59:54.280 align:middle line:90% there's none. 00:59:54.280 --> 00:59:59.600 align:middle line:84% OK, now let's-- what I want to do in the remaining time is 00:59:59.600 --> 01:00:02.740 align:middle line:84% write down the economic model of majority voting. 01:00:02.740 --> 01:00:04.240 align:middle line:84% And in particular, I'm going to talk 01:00:04.240 --> 01:00:07.280 align:middle line:84% about when majority voting actually works 01:00:07.280 --> 01:00:08.500 align:middle line:90% and when it doesn't work. 01:00:08.500 --> 01:00:10.800 align:middle line:84% I'll define what working and doesn't working means. 01:00:10.800 --> 01:00:15.000 align:middle line:84% So let's talk about, what principles would we 01:00:15.000 --> 01:00:16.840 align:middle line:90% like in a voting system? 01:00:16.840 --> 01:00:19.480 align:middle line:84% The fundamental principle we'd like in a voting system 01:00:19.480 --> 01:00:23.280 align:middle line:84% is that it consistently aggregates preferences. 01:00:23.280 --> 01:00:25.880 align:middle line:84% The fundamental goal of a voting system 01:00:25.880 --> 01:00:27.940 align:middle line:84% is that it consistently aggregates preferences. 01:00:27.940 --> 01:00:32.530 align:middle line:84% What do I mean by consistently aggregating preferences? 01:00:32.530 --> 01:00:34.530 align:middle line:90% I mean three things. 01:00:34.530 --> 01:00:39.890 align:middle line:84% The first is it must achieve dominance, 01:00:39.890 --> 01:00:45.130 align:middle line:84% which means that if one choice is preferred by all the voters, 01:00:45.130 --> 01:00:50.890 align:middle line:84% by a majority of voters, that choice should be chosen. 01:00:50.890 --> 01:00:51.590 align:middle line:90% No, all voters. 01:00:51.590 --> 01:00:52.090 align:middle line:90% I'm sorry. 01:00:52.090 --> 01:00:54.410 align:middle line:84% If there's one choice preferred by all voters, 01:00:54.410 --> 01:00:55.750 align:middle line:90% that should be chosen. 01:00:55.750 --> 01:00:58.510 align:middle line:84% If that's not true, the system is not consistently 01:00:58.510 --> 01:00:59.510 align:middle line:90% aggregating preferences. 01:00:59.510 --> 01:01:01.050 align:middle line:84% If everyone wants something and the voting system 01:01:01.050 --> 01:01:02.850 align:middle line:84% doesn't deliver it, then you're clearly not 01:01:02.850 --> 01:01:05.250 align:middle line:84% aggregating preferences appropriately. 01:01:05.250 --> 01:01:08.830 align:middle line:84% The second is our old friend from kindergarten-- 01:01:08.830 --> 01:01:10.450 align:middle line:90% transitivity. 01:01:10.450 --> 01:01:12.530 align:middle line:84% And I say kindergarten because you've probably-- 01:01:12.530 --> 01:01:14.410 align:middle line:84% now in all the math classes you guys have been in, you're sick 01:01:14.410 --> 01:01:15.535 align:middle line:90% of the term "transitivity." 01:01:15.535 --> 01:01:17.347 align:middle line:90% You all know what it means. 01:01:17.347 --> 01:01:18.430 align:middle line:90% So it's just transitivity. 01:01:18.430 --> 01:01:20.055 align:middle line:84% We want if a choice A is preferred to B 01:01:20.055 --> 01:01:23.510 align:middle line:84% and B is preferred to C, we need A to be preferred to C. 01:01:23.510 --> 01:01:25.890 align:middle line:84% That's the second condition we'd like. 01:01:25.890 --> 01:01:29.130 align:middle line:84% So those are two I would say noncontroversial decisions. 01:01:29.130 --> 01:01:32.050 align:middle line:84% The third decision is a little more controversial 01:01:32.050 --> 01:01:34.210 align:middle line:84% but in political economy generally assumed 01:01:34.210 --> 01:01:38.170 align:middle line:84% to be important, which is what we call independence 01:01:38.170 --> 01:01:40.570 align:middle line:90% of irrelevant alternatives. 01:01:40.570 --> 01:01:43.570 align:middle line:84% Independence of irrelevant alternatives. 01:01:43.570 --> 01:01:48.690 align:middle line:84% And this system says that the choice between A and B 01:01:48.690 --> 01:01:53.730 align:middle line:84% should not be affected by the random introduction of C. 01:01:53.730 --> 01:01:58.310 align:middle line:84% So let's say, ignoring the financing-- ignore the money. 01:01:58.310 --> 01:02:01.730 align:middle line:84% Let's say your town's choosing between erecting a statue 01:02:01.730 --> 01:02:02.610 align:middle line:90% and building-- 01:02:02.610 --> 01:02:06.290 align:middle line:90% 01:02:06.290 --> 01:02:08.670 align:middle line:90% basically a statue and a park. 01:02:08.670 --> 01:02:11.330 align:middle line:84% You're going to build a park or a big monument. 01:02:11.330 --> 01:02:15.530 align:middle line:84% And your town votes and prefers the statue to the park. 01:02:15.530 --> 01:02:17.970 align:middle line:84% Now, let's say a new option is introduced of a new police 01:02:17.970 --> 01:02:20.090 align:middle line:90% station. 01:02:20.090 --> 01:02:22.770 align:middle line:84% Now, that could win, but it should not 01:02:22.770 --> 01:02:27.820 align:middle line:84% affect people's preferences on a statue versus a park. 01:02:27.820 --> 01:02:29.420 align:middle line:84% The fact that this third alternative-- 01:02:29.420 --> 01:02:30.800 align:middle line:84% I'm not saying the alternative couldn't win. 01:02:30.800 --> 01:02:32.592 align:middle line:84% I'm just saying relative ranking should not 01:02:32.592 --> 01:02:37.220 align:middle line:84% change when a random independent alternative gets introduced. 01:02:37.220 --> 01:02:39.120 align:middle line:84% That is our third condition we impose. 01:02:39.120 --> 01:02:41.180 align:middle line:84% And that's what political economists impose, 01:02:41.180 --> 01:02:42.900 align:middle line:84% is the conditions of a consistent voting 01:02:42.900 --> 01:02:46.100 align:middle line:90% system, consistent aggregation. 01:02:46.100 --> 01:02:49.340 align:middle line:84% Now, what we're going to show you 01:02:49.340 --> 01:02:52.700 align:middle line:84% is when majority voting systems do and do not 01:02:52.700 --> 01:02:54.740 align:middle line:90% meet this condition. 01:02:54.740 --> 01:02:56.780 align:middle line:84% So let's start with an example of a town that's 01:02:56.780 --> 01:03:00.980 align:middle line:84% deciding on levels of funding for public education. 01:03:00.980 --> 01:03:02.940 align:middle line:90% Let's go to table 9-1. 01:03:02.940 --> 01:03:07.030 align:middle line:90% 01:03:07.030 --> 01:03:17.020 align:middle line:84% In table 9-1, there are-- we are voting over whether to have 01:03:17.020 --> 01:03:19.540 align:middle line:84% how-- whether we should vote-- how we should fund the level 01:03:19.540 --> 01:03:20.320 align:middle line:90% of school. 01:03:20.320 --> 01:03:22.580 align:middle line:84% And there's three choices-- high, medium, and low. 01:03:22.580 --> 01:03:26.150 align:middle line:84% High means great schools, high taxes. 01:03:26.150 --> 01:03:28.530 align:middle line:84% Medium means medium schools, medium taxes. 01:03:28.530 --> 01:03:31.032 align:middle line:84% Low means crappy schools, low taxes. 01:03:31.032 --> 01:03:32.490 align:middle line:84% And there's three groups of voters. 01:03:32.490 --> 01:03:34.550 align:middle line:90% There's parents. 01:03:34.550 --> 01:03:35.442 align:middle line:90% There's elders. 01:03:35.442 --> 01:03:37.150 align:middle line:84% And there's young couples who aren't sure 01:03:37.150 --> 01:03:40.555 align:middle line:84% if they're going to have kids before they move. 01:03:40.555 --> 01:03:42.430 align:middle line:84% And then we have, for each group, the ranking 01:03:42.430 --> 01:03:43.950 align:middle line:90% of their choices. 01:03:43.950 --> 01:03:47.132 align:middle line:84% Parents choose-- their first choice is high funding. 01:03:47.132 --> 01:03:48.090 align:middle line:90% They want good schools. 01:03:48.090 --> 01:03:49.670 align:middle line:90% They have to pay high taxes. 01:03:49.670 --> 01:03:50.530 align:middle line:90% Second, medium. 01:03:50.530 --> 01:03:52.390 align:middle line:90% Third, low. 01:03:52.390 --> 01:03:53.490 align:middle line:90% Elders are the reverse. 01:03:53.490 --> 01:03:56.070 align:middle line:84% These are selfish elders, not altruistic elders. 01:03:56.070 --> 01:03:57.843 align:middle line:84% Elders are like, I want low taxes. 01:03:57.843 --> 01:03:59.510 align:middle line:84% I don't care how crappy the schools are. 01:03:59.510 --> 01:04:00.570 align:middle line:90% The worst outcome for me. 01:04:00.570 --> 01:04:04.870 align:middle line:84% So I want low taxes, then medium, then high. 01:04:04.870 --> 01:04:09.870 align:middle line:84% The young couple, they're not really sure. 01:04:09.870 --> 01:04:12.765 align:middle line:84% So they're like, well, I'm not sure I'm going to be here. 01:04:12.765 --> 01:04:13.890 align:middle line:90% So I want to hedge my bets. 01:04:13.890 --> 01:04:15.110 align:middle line:90% My best is medium. 01:04:15.110 --> 01:04:17.710 align:middle line:84% My worst is low, because I want to be stuck 01:04:17.710 --> 01:04:19.870 align:middle line:90% in a town with crappy schools. 01:04:19.870 --> 01:04:22.750 align:middle line:84% I'm sorry my middle is low, because if the schools are 01:04:22.750 --> 01:04:24.060 align:middle line:90% crappy, I'll just move. 01:04:24.060 --> 01:04:25.360 align:middle line:90% The worst for me is high. 01:04:25.360 --> 01:04:28.290 align:middle line:84% I pay high taxes in a town that may not even stay in. 01:04:28.290 --> 01:04:30.540 align:middle line:84% So I mean, I'm not saying these are right preferences. 01:04:30.540 --> 01:04:32.240 align:middle line:84% But I'm saying these are not crazy preferences for people 01:04:32.240 --> 01:04:33.360 align:middle line:90% to have. 01:04:33.360 --> 01:04:36.240 align:middle line:84% So imagine those are people's preferences. 01:04:36.240 --> 01:04:40.040 align:middle line:84% If you had those preferences use majority voting, 01:04:40.040 --> 01:04:42.832 align:middle line:84% you will deliver a consistent outcome. 01:04:42.832 --> 01:04:45.040 align:middle line:84% Now, majority voting-- let's think of majority voting 01:04:45.040 --> 01:04:46.880 align:middle line:90% as happening pairwise. 01:04:46.880 --> 01:04:48.640 align:middle line:90% Think of pairwise voting. 01:04:48.640 --> 01:04:52.228 align:middle line:84% So for example, no matter how you have pairwise voting, 01:04:52.228 --> 01:04:54.020 align:middle line:84% you always end up with consistent outcomes. 01:04:54.020 --> 01:04:57.920 align:middle line:84% So for example, imagine we voted on H versus L. Look 01:04:57.920 --> 01:04:58.860 align:middle line:90% at this table. 01:04:58.860 --> 01:05:02.300 align:middle line:84% If we had H versus L, what would win? 01:05:02.300 --> 01:05:02.800 align:middle line:90% Yeah. 01:05:02.800 --> 01:05:03.400 align:middle line:90% AUDIENCE: L. 01:05:03.400 --> 01:05:04.400 align:middle line:90% JON GRUBER: L would win. 01:05:04.400 --> 01:05:06.980 align:middle line:84% L is preferred to H by two out of three voters. 01:05:06.980 --> 01:05:08.560 align:middle line:90% So L would win. 01:05:08.560 --> 01:05:12.280 align:middle line:90% So now what about H versus M? 01:05:12.280 --> 01:05:12.800 align:middle line:90% Same person. 01:05:12.800 --> 01:05:13.780 align:middle line:90% H versus M? 01:05:13.780 --> 01:05:14.520 align:middle line:90% AUDIENCE: M. 01:05:14.520 --> 01:05:15.640 align:middle line:90% JON GRUBER: M wins. 01:05:15.640 --> 01:05:17.620 align:middle line:84% So we got our two semifinal winners. 01:05:17.620 --> 01:05:20.540 align:middle line:84% So let's go to the finals, M versus L. What wins? 01:05:20.540 --> 01:05:21.520 align:middle line:90% AUDIENCE: M wins. 01:05:21.520 --> 01:05:22.480 align:middle line:90% JON GRUBER: M wins. 01:05:22.480 --> 01:05:25.100 align:middle line:84% And it turns out, no matter what order I did those in, 01:05:25.100 --> 01:05:26.340 align:middle line:90% you'd get the same answer. 01:05:26.340 --> 01:05:28.040 align:middle line:90% M would win. 01:05:28.040 --> 01:05:29.920 align:middle line:84% No matter what order I did those in, 01:05:29.920 --> 01:05:31.680 align:middle line:90% you would get the same answer. 01:05:31.680 --> 01:05:34.180 align:middle line:90% M will always win in this case. 01:05:34.180 --> 01:05:37.762 align:middle line:84% Majority voting works, as opposed to-- you might say, 01:05:37.762 --> 01:05:39.220 align:middle line:84% that's-- what do you mean by works? 01:05:39.220 --> 01:05:41.440 align:middle line:84% Let's talk about an example where it doesn't work. 01:05:41.440 --> 01:05:45.080 align:middle line:84% Imagine that the elderly person moves out, 01:05:45.080 --> 01:05:47.835 align:middle line:84% and they are replaced by someone who either 01:05:47.835 --> 01:05:50.460 align:middle line:84% is going to send their kids to the public schools if their good 01:05:50.460 --> 01:05:52.135 align:middle line:84% or the private schools if they're not. 01:05:52.135 --> 01:05:54.260 align:middle line:84% So we're going to turn, flip over to the next page. 01:05:54.260 --> 01:05:55.860 align:middle line:90% Go to table 9-2. 01:05:55.860 --> 01:05:57.753 align:middle line:90% Oh, you lost table 9-2. 01:05:57.753 --> 01:05:58.920 align:middle line:90% OK, we don't have table 9-2. 01:05:58.920 --> 01:06:03.560 align:middle line:90% 01:06:03.560 --> 01:06:05.220 align:middle line:90% Table 9-2 looks like this. 01:06:05.220 --> 01:06:06.940 align:middle line:84% And this when I copy from the book 01:06:06.940 --> 01:06:09.560 align:middle line:84% so I don't screw this one up, although I will say you guys 01:06:09.560 --> 01:06:12.120 align:middle line:84% were most active when I screwed that graph up. 01:06:12.120 --> 01:06:14.100 align:middle line:84% So I'm not going to screw stuff up on purpose, 01:06:14.100 --> 01:06:16.160 align:middle line:84% but I want to keep that level of activity. 01:06:16.160 --> 01:06:20.650 align:middle line:84% So now let's imagine you've got your public school parents. 01:06:20.650 --> 01:06:25.010 align:middle line:84% You've got your private school parents. 01:06:25.010 --> 01:06:27.070 align:middle line:84% And you've got your young couples. 01:06:27.070 --> 01:06:30.090 align:middle line:90% 01:06:30.090 --> 01:06:32.858 align:middle line:84% First choice, second choice, third choice. 01:06:32.858 --> 01:06:34.150 align:middle line:90% So we know the first and third. 01:06:34.150 --> 01:06:40.050 align:middle line:84% This is H, M, L. This is M, L, H. 01:06:40.050 --> 01:06:43.410 align:middle line:84% For the private school parents, they have an interesting choice. 01:06:43.410 --> 01:06:46.070 align:middle line:84% Their first choice is crappy schools. 01:06:46.070 --> 01:06:48.250 align:middle line:84% They send the kid to private school. 01:06:48.250 --> 01:06:51.770 align:middle line:84% But their second choice is H. And their worst choice 01:06:51.770 --> 01:06:52.570 align:middle line:90% is M. Why? 01:06:52.570 --> 01:06:54.230 align:middle line:90% Because M is the worst. 01:06:54.230 --> 01:06:57.150 align:middle line:84% They pay medium taxes, and they can't send their kids to school. 01:06:57.150 --> 01:07:00.170 align:middle line:84% At least with H, they'll then send their kids to local school. 01:07:00.170 --> 01:07:02.130 align:middle line:84% But the worst is a town with pretty high taxes 01:07:02.130 --> 01:07:03.917 align:middle line:90% and mediocre schools. 01:07:03.917 --> 01:07:06.250 align:middle line:84% They'd either rather have a town with low tax and crappy 01:07:06.250 --> 01:07:07.930 align:middle line:84% schools-- they'll go to private school-- or a town 01:07:07.930 --> 01:07:09.070 align:middle line:84% with high taxes and good schools. 01:07:09.070 --> 01:07:10.445 align:middle line:84% Then they'll go to public school. 01:07:10.445 --> 01:07:13.730 align:middle line:84% Once again, not crazy preference for someone to have. 01:07:13.730 --> 01:07:17.970 align:middle line:84% But with these preferences, we get a very different outcome. 01:07:17.970 --> 01:07:20.060 align:middle line:84% Let's do the same voting as before. 01:07:20.060 --> 01:07:22.900 align:middle line:90% H versus L, what wins here? 01:07:22.900 --> 01:07:25.756 align:middle line:90% H versus L? 01:07:25.756 --> 01:07:26.628 align:middle line:90% AUDIENCE: L. 01:07:26.628 --> 01:07:27.420 align:middle line:90% JON GRUBER: L wins. 01:07:27.420 --> 01:07:28.820 align:middle line:90% L gets two votes. 01:07:28.820 --> 01:07:29.900 align:middle line:90% H gets one. 01:07:29.900 --> 01:07:30.980 align:middle line:90% OK, L wins. 01:07:30.980 --> 01:07:31.900 align:middle line:90% H versus M? 01:07:31.900 --> 01:07:35.124 align:middle line:90% 01:07:35.124 --> 01:07:35.986 align:middle line:90% AUDIENCE: H. 01:07:35.986 --> 01:07:37.820 align:middle line:90% JON GRUBER: H wins. 01:07:37.820 --> 01:07:42.460 align:middle line:84% So H wins over M. And H wins versus L. 01:07:42.460 --> 01:07:45.480 align:middle line:84% So you might think we're done, but what about L versus M? 01:07:45.480 --> 01:07:50.220 align:middle line:90% 01:07:50.220 --> 01:07:51.220 align:middle line:90% So I'm sorry. 01:07:51.220 --> 01:07:59.580 align:middle line:84% L wins over H. H wins over M. So we've got L versus-- 01:07:59.580 --> 01:08:00.900 align:middle line:90% but imagine we did-- 01:08:00.900 --> 01:08:06.300 align:middle line:84% OK, so now let's do L versus H. Who wins that? 01:08:06.300 --> 01:08:07.300 align:middle line:90% No, we already did that. 01:08:07.300 --> 01:08:08.180 align:middle line:90% I'm sorry. 01:08:08.180 --> 01:08:16.930 align:middle line:84% L beats H. H beats M. So L versus M, what do we get? 01:08:16.930 --> 01:08:17.430 align:middle line:90% AUDIENCE: M. 01:08:17.430 --> 01:08:20.430 align:middle line:84% JON GRUBER: M, which is inconsistent. 01:08:20.430 --> 01:08:23.069 align:middle line:90% It violates transitivity. 01:08:23.069 --> 01:08:26.029 align:middle line:84% And in fact, no matter how you do this vote, 01:08:26.029 --> 01:08:28.390 align:middle line:84% you're going to violate transitivity. 01:08:28.390 --> 01:08:31.229 align:middle line:84% You're never going to get the condition of aggregation. 01:08:31.229 --> 01:08:33.470 align:middle line:84% No matter how you do this vote, the order 01:08:33.470 --> 01:08:36.069 align:middle line:84% is going to violate transitivity. 01:08:36.069 --> 01:08:41.149 align:middle line:84% And indeed, no matter how you set up the voting mechanism, 01:08:41.149 --> 01:08:43.170 align:middle line:84% you can never guarantee a consistent outcome. 01:08:43.170 --> 01:08:45.090 align:middle line:84% I say, well, this is one voting system, Jon. 01:08:45.090 --> 01:08:46.210 align:middle line:90% What about other voting systems? 01:08:46.210 --> 01:08:47.649 align:middle line:84% Let's consider some other systems. 01:08:47.649 --> 01:08:51.510 align:middle line:84% Imagine, let's let everyone vote in their first choice. 01:08:51.510 --> 01:08:54.950 align:middle line:84% Well, in the first case, if we let everyone vote 01:08:54.950 --> 01:08:57.115 align:middle line:84% on their first choice, we'd get a tie. 01:08:57.115 --> 01:08:58.490 align:middle line:84% In the second case, we get a tie. 01:08:58.490 --> 01:08:59.450 align:middle line:90% It wouldn't help us. 01:08:59.450 --> 01:09:01.050 align:middle line:84% So even first choice doesn't help us. 01:09:01.050 --> 01:09:03.008 align:middle line:84% Let's do another thing like my dorm used to do. 01:09:03.008 --> 01:09:04.109 align:middle line:90% We'll do weighted voting. 01:09:04.109 --> 01:09:05.370 align:middle line:90% 3 points for first choice. 01:09:05.370 --> 01:09:06.450 align:middle line:90% 2 points, second choice. 01:09:06.450 --> 01:09:08.430 align:middle line:90% 1 point for third choice. 01:09:08.430 --> 01:09:11.319 align:middle line:84% In the first example, M would win with 7 points. 01:09:11.319 --> 01:09:12.736 align:middle line:84% In the second example, once again, 01:09:12.736 --> 01:09:15.327 align:middle line:84% it's going to be a three-way tie with 6 points each. 01:09:15.327 --> 01:09:17.410 align:middle line:84% So no matter how do you set up this voting system, 01:09:17.410 --> 01:09:20.310 align:middle line:84% you're not going to get a consistent outcome. 01:09:20.310 --> 01:09:23.649 align:middle line:84% Indeed-- let me just come to this exciting punchline. 01:09:23.649 --> 01:09:27.050 align:middle line:84% Indeed, this is the proof, a graphical proof, 01:09:27.050 --> 01:09:29.430 align:middle line:84% of a very complicated set of math that I cannot do-- 01:09:29.430 --> 01:09:30.810 align:middle line:84% because I'm not going to win the Nobel Prize, 01:09:30.810 --> 01:09:33.510 align:middle line:84% but someone who won the Nobel Prize did do it, Kenneth Arrow-- 01:09:33.510 --> 01:09:38.590 align:middle line:84% that proves mathematically that you cannot ever have a system 01:09:38.590 --> 01:09:41.950 align:middle line:84% which ever and always satisfies these three conditions, 01:09:41.950 --> 01:09:44.430 align:middle line:90% except a dictator. 01:09:44.430 --> 01:09:46.510 align:middle line:84% Absent a dictator, you can never have 01:09:46.510 --> 01:09:48.670 align:middle line:84% a system which satisfies all three 01:09:48.670 --> 01:09:50.870 align:middle line:84% of those properties in generality. 01:09:50.870 --> 01:09:52.687 align:middle line:90% Alec, question. 01:09:52.687 --> 01:09:54.770 align:middle line:84% AUDIENCE: So how often is this actually a problem? 01:09:54.770 --> 01:09:57.050 align:middle line:84% Because to create this, you created a very artificial-- 01:09:57.050 --> 01:09:57.930 align:middle line:84% JON GRUBER: Let's talk about that. 01:09:57.930 --> 01:09:58.570 align:middle line:90% Great point. 01:09:58.570 --> 01:10:01.590 align:middle line:90% Let's go to the figure. 01:10:01.590 --> 01:10:02.810 align:middle line:90% Let's go to the figure. 01:10:02.810 --> 01:10:06.400 align:middle line:90% The figure shows figure a. 01:10:06.400 --> 01:10:08.150 align:middle line:84% So what we're doing here is we're mapping. 01:10:08.150 --> 01:10:11.110 align:middle line:84% On the x-axis is the levels of school spending. 01:10:11.110 --> 01:10:16.040 align:middle line:84% On the y-axis is preferences-- first, second, third choice. 01:10:16.040 --> 01:10:19.280 align:middle line:84% So look at, for example, the green line. 01:10:19.280 --> 01:10:21.720 align:middle line:90% The green line is parents. 01:10:21.720 --> 01:10:24.020 align:middle line:84% High spending is the highest utility. 01:10:24.020 --> 01:10:26.780 align:middle line:84% That's at point A. Lowest spending is the lowest utility. 01:10:26.780 --> 01:10:29.080 align:middle line:84% That's at point B. People see that? 01:10:29.080 --> 01:10:30.060 align:middle line:90% Now look at the elders. 01:10:30.060 --> 01:10:31.840 align:middle line:90% It's the opposite line. 01:10:31.840 --> 01:10:34.000 align:middle line:90% Now look at the young couples. 01:10:34.000 --> 01:10:37.480 align:middle line:90% It peaks in the middle. 01:10:37.480 --> 01:10:43.160 align:middle line:84% Now go to figure B. The difference of private school 01:10:43.160 --> 01:10:48.600 align:middle line:84% parents is it goes down then back up. 01:10:48.600 --> 01:10:52.560 align:middle line:84% The red line is different than any of the lines 01:10:52.560 --> 01:10:56.840 align:middle line:84% in panel A, which all have what we call a single peak. 01:10:56.840 --> 01:11:00.200 align:middle line:84% So as long as preferences are single peaked, 01:11:00.200 --> 01:11:01.740 align:middle line:90% majority voting works. 01:11:01.740 --> 01:11:03.160 align:middle line:84% So the restriction you can impose 01:11:03.160 --> 01:11:04.160 align:middle line:84% to get around the impossibility-- 01:11:04.160 --> 01:11:05.493 align:middle line:90% the impossibility theorem holds. 01:11:05.493 --> 01:11:06.868 align:middle line:84% You have to impose a restriction. 01:11:06.868 --> 01:11:09.320 align:middle line:84% The restriction is what we call single-peaked preferences, 01:11:09.320 --> 01:11:12.720 align:middle line:84% preferences where the local maximum is the global maximum. 01:11:12.720 --> 01:11:15.290 align:middle line:90% 01:11:15.290 --> 01:11:18.130 align:middle line:84% That's not true in the second panel. 01:11:18.130 --> 01:11:20.590 align:middle line:84% When is that likely or not likely to be true? 01:11:20.590 --> 01:11:23.450 align:middle line:84% And The example, Alec, was not that weird. 01:11:23.450 --> 01:11:26.250 align:middle line:84% In fact, when-- the situation was likely to be violated 01:11:26.250 --> 01:11:27.730 align:middle line:84% are exactly situations when there's 01:11:27.730 --> 01:11:31.050 align:middle line:84% a good private alternative to the public good. 01:11:31.050 --> 01:11:32.730 align:middle line:84% That's exactly the conditions in which 01:11:32.730 --> 01:11:34.780 align:middle line:84% it is likely to get violated, because those 01:11:34.780 --> 01:11:36.530 align:middle line:84% are the conditions which people might want 01:11:36.530 --> 01:11:38.810 align:middle line:90% one extreme but not the middle. 01:11:38.810 --> 01:11:42.450 align:middle line:84% Generally, people prefer the middle. 01:11:42.450 --> 01:11:45.070 align:middle line:84% Or they have linear or they have monotonic rankings. 01:11:45.070 --> 01:11:47.010 align:middle line:84% The fact that this is nonmonotonicity, where 01:11:47.010 --> 01:11:48.430 align:middle line:90% they like the middle worst. 01:11:48.430 --> 01:11:50.570 align:middle line:84% That's because there's this private option. 01:11:50.570 --> 01:11:53.030 align:middle line:90% So Alec is generally right. 01:11:53.030 --> 01:11:55.090 align:middle line:84% Generally, it's a pretty safe assumption 01:11:55.090 --> 01:11:57.770 align:middle line:84% that preference is single peaked but not always. 01:11:57.770 --> 01:11:59.790 align:middle line:84% And the theoretical proposition is important, 01:11:59.790 --> 01:12:04.810 align:middle line:84% which is not always means that it will not always work. 01:12:04.810 --> 01:12:09.710 align:middle line:84% So basically, we restrict preferences. 01:12:09.710 --> 01:12:12.050 align:middle line:84% If we restrict preferences to be single peaked, 01:12:12.050 --> 01:12:15.030 align:middle line:84% we get that majority voting works, 01:12:15.030 --> 01:12:16.890 align:middle line:90% but we get more than that. 01:12:16.890 --> 01:12:20.810 align:middle line:84% We actually get an even better, more powerful outcome, which 01:12:20.810 --> 01:12:25.050 align:middle line:84% is that basically, if preferences are single peaked, 01:12:25.050 --> 01:12:27.890 align:middle line:84% you can prove mathematically what we call the median voter 01:12:27.890 --> 01:12:28.590 align:middle line:90% theorem. 01:12:28.590 --> 01:12:31.610 align:middle line:90% 01:12:31.610 --> 01:12:35.130 align:middle line:84% The median voter theorem states that if preferences 01:12:35.130 --> 01:12:39.170 align:middle line:84% are single peaked, the outcome of majority voting 01:12:39.170 --> 01:12:43.090 align:middle line:84% were the outcome preferred by the median voter. 01:12:43.090 --> 01:12:46.610 align:middle line:84% So you don't even need to know anybody else's preferences 01:12:46.610 --> 01:12:50.507 align:middle line:84% as long as I count people up-- if there's 1,001 people, 01:12:50.507 --> 01:12:52.590 align:middle line:84% I don't need to know the 1,000 people preferences. 01:12:52.590 --> 01:12:54.610 align:middle line:84% I just need to know the one guy who's the median voter, 01:12:54.610 --> 01:12:55.950 align:middle line:90% because they're the swing vote. 01:12:55.950 --> 01:12:57.293 align:middle line:90% Their preferences determine. 01:12:57.293 --> 01:12:59.710 align:middle line:84% So all you need to know is the median voter's preferences. 01:12:59.710 --> 01:13:01.230 align:middle line:90% We call them swing voters now. 01:13:01.230 --> 01:13:03.642 align:middle line:84% Back in the old day, we called them median voters. 01:13:03.642 --> 01:13:05.350 align:middle line:84% I don't care if there's a million people, 01:13:05.350 --> 01:13:06.058 align:middle line:90% 2 million people. 01:13:06.058 --> 01:13:08.310 align:middle line:84% As long as I can decide who the median voter is, 01:13:08.310 --> 01:13:10.210 align:middle line:84% I just need to know what they want, 01:13:10.210 --> 01:13:12.060 align:middle line:84% which is incredibly powerful in terms 01:13:12.060 --> 01:13:14.340 align:middle line:84% of reducing our information that we require 01:13:14.340 --> 01:13:15.800 align:middle line:90% to understand decisions. 01:13:15.800 --> 01:13:18.458 align:middle line:84% That's an unbelievably powerful theorem. 01:13:18.458 --> 01:13:20.000 align:middle line:84% It doesn't matter how big society is. 01:13:20.000 --> 01:13:23.260 align:middle line:84% I just need to know the median voter's preferences. 01:13:23.260 --> 01:13:25.860 align:middle line:90% That's the good news. 01:13:25.860 --> 01:13:29.640 align:middle line:84% The good news is that in a model with single-peak preferences 01:13:29.640 --> 01:13:32.580 align:middle line:84% with majority voting, the median voter will always 01:13:32.580 --> 01:13:33.560 align:middle line:90% determine the outcome. 01:13:33.560 --> 01:13:35.180 align:middle line:90% Yeah. 01:13:35.180 --> 01:13:36.982 align:middle line:84% AUDIENCE: These conditions-- is the thing 01:13:36.982 --> 01:13:38.940 align:middle line:84% that these conditions are necessary for to have 01:13:38.940 --> 01:13:42.140 align:middle line:84% a consistent top of the ranking or consistent in aggregate. 01:13:42.140 --> 01:13:43.940 align:middle line:84% JON GRUBER: Consistent voting mechanism, 01:13:43.940 --> 01:13:46.140 align:middle line:84% that voting delivers the outcome that 01:13:46.140 --> 01:13:48.580 align:middle line:90% meets these three conditions. 01:13:48.580 --> 01:13:52.460 align:middle line:84% So the good news is we can meet these three 01:13:52.460 --> 01:13:55.020 align:middle line:84% conditions with single-peak preference with majority voting. 01:13:55.020 --> 01:13:58.020 align:middle line:84% And even better news is always knows who the median voter is 01:13:58.020 --> 01:14:00.020 align:middle line:90% and what their preferences are. 01:14:00.020 --> 01:14:02.460 align:middle line:90% But here's the bad news. 01:14:02.460 --> 01:14:05.580 align:middle line:90% Is that outcome efficient? 01:14:05.580 --> 01:14:07.392 align:middle line:84% Is the median voter outcome efficient? 01:14:07.392 --> 01:14:08.100 align:middle line:90% When would it be? 01:14:08.100 --> 01:14:10.230 align:middle line:90% And when would it not be? 01:14:10.230 --> 01:14:13.590 align:middle line:84% And here, I'll accept a graphical intuition, not 01:14:13.590 --> 01:14:16.190 align:middle line:84% that kind of graph but thinking about statistics. 01:14:16.190 --> 01:14:18.690 align:middle line:84% When would the median voter-- remember, what's efficiency? 01:14:18.690 --> 01:14:21.550 align:middle line:84% Remember the Samuelson efficiency condition. 01:14:21.550 --> 01:14:23.790 align:middle line:84% When will the median voter outcome 01:14:23.790 --> 01:14:25.630 align:middle line:84% achieve the Samuelson efficiency condition? 01:14:25.630 --> 01:14:26.612 align:middle line:90% Or when will it not? 01:14:26.612 --> 01:14:28.570 align:middle line:84% There's actually a mathematical answer to this. 01:14:28.570 --> 01:14:29.185 align:middle line:90% Yeah. 01:14:29.185 --> 01:14:31.310 align:middle line:84% AUDIENCE: [INAUDIBLE] preference is the same thing. 01:14:31.310 --> 01:14:32.030 align:middle line:90% JON GRUBER: Yeah. 01:14:32.030 --> 01:14:34.470 align:middle line:84% And what would that mean in terms of a distribution? 01:14:34.470 --> 01:14:37.390 align:middle line:84% It would mean the median equals the mean, 01:14:37.390 --> 01:14:42.510 align:middle line:84% because the Samuelson condition is that the sum of values 01:14:42.510 --> 01:14:44.950 align:middle line:90% gets set equal to the cost. 01:14:44.950 --> 01:14:48.190 align:middle line:84% That means on average, you want the average person's value 01:14:48.190 --> 01:14:51.872 align:middle line:90% to be the average cost. 01:14:51.872 --> 01:14:53.830 align:middle line:84% In a median voter system that will only be true 01:14:53.830 --> 01:14:56.022 align:middle line:90% if the median equals the mean. 01:14:56.022 --> 01:14:57.730 align:middle line:84% But if the median doesn't equal the mean, 01:14:57.730 --> 01:14:58.530 align:middle line:90% it will no longer be true. 01:14:58.530 --> 01:15:00.780 align:middle line:84% In particular, if there's more intensity of preference 01:15:00.780 --> 01:15:02.590 align:middle line:84% on one side than another, you will not 01:15:02.590 --> 01:15:06.150 align:middle line:84% get the efficient outcome from the median voter theorem. 01:15:06.150 --> 01:15:09.640 align:middle line:84% So for example, imagine your town is considering 01:15:09.640 --> 01:15:11.180 align:middle line:90% building a statue to you. 01:15:11.180 --> 01:15:12.240 align:middle line:90% You got in MIT. 01:15:12.240 --> 01:15:12.875 align:middle line:90% Life's made. 01:15:12.875 --> 01:15:13.500 align:middle line:90% You're all set. 01:15:13.500 --> 01:15:15.167 align:middle line:84% They don't care what you're going to do. 01:15:15.167 --> 01:15:16.400 align:middle line:90% You already deserve a statue. 01:15:16.400 --> 01:15:21.000 align:middle line:84% And let's imagine there's 1,001 voters in your town. 01:15:21.000 --> 01:15:22.520 align:middle line:84% And the statue will, conveniently 01:15:22.520 --> 01:15:28.280 align:middle line:84% for our mathematical purposes, cost $40,040. 01:15:28.280 --> 01:15:31.640 align:middle line:84% And the town is going to finance it by an equal tax on everyone, 01:15:31.640 --> 01:15:33.920 align:middle line:90% of $40, conveniently enough. 01:15:33.920 --> 01:15:38.480 align:middle line:84% So a tax of $40 will finance building the statue to you. 01:15:38.480 --> 01:15:41.160 align:middle line:84% Now, let's say everyone in this town knows you. 01:15:41.160 --> 01:15:45.040 align:middle line:84% And let's say that 500 of the people in this town 01:15:45.040 --> 01:15:47.600 align:middle line:84% appropriately recognize you are awesome. 01:15:47.600 --> 01:15:54.120 align:middle line:84% And each of those 500 people would happily pay up to $100 01:15:54.120 --> 01:15:57.800 align:middle line:84% to have the statue, for a value of $50,000. 01:15:57.800 --> 01:16:01.800 align:middle line:84% So the Samuelson rule would say the statue should get built. 01:16:01.800 --> 01:16:05.080 align:middle line:84% If I aggregate-- assuming the other people's value 01:16:05.080 --> 01:16:09.680 align:middle line:84% isn't negative, the statue should get built. 01:16:09.680 --> 01:16:11.800 align:middle line:84% Assuming even if everyone else is zero, 01:16:11.800 --> 01:16:15.280 align:middle line:84% the statue should get built, because basically, 01:16:15.280 --> 01:16:20.200 align:middle line:84% $50,000 social value exceeds the $40,000 in costs. 01:16:20.200 --> 01:16:22.160 align:middle line:90% But let's say you have a vote. 01:16:22.160 --> 01:16:25.720 align:middle line:84% And the other 501 don't know you, don't like you, whatever. 01:16:25.720 --> 01:16:26.902 align:middle line:90% And they vote no. 01:16:26.902 --> 01:16:27.860 align:middle line:90% What's going to happen? 01:16:27.860 --> 01:16:29.880 align:middle line:84% Will the statue get built or not? 01:16:29.880 --> 01:16:32.920 align:middle line:84% No, because the median voter doesn't like you. 01:16:32.920 --> 01:16:35.920 align:middle line:84% The median voter values you at less than $40. 01:16:35.920 --> 01:16:38.200 align:middle line:84% So the statue won't get built. So we do not 01:16:38.200 --> 01:16:40.520 align:middle line:84% meet the Samuelson condition through the median voter 01:16:40.520 --> 01:16:42.200 align:middle line:90% theorem. 01:16:42.200 --> 01:16:44.383 align:middle line:84% So the median voter theorem will only 01:16:44.383 --> 01:16:45.800 align:middle line:84% meet the Samuelson condition if we 01:16:45.800 --> 01:16:47.300 align:middle line:84% have equal intensity of preferences, 01:16:47.300 --> 01:16:50.560 align:middle line:84% or if the median distribution equals the mean. 01:16:50.560 --> 01:16:54.000 align:middle line:84% Only in that case will the convenience of the median voter 01:16:54.000 --> 01:16:57.000 align:middle line:84% theorem meet the efficiency of the Samuelson condition, which 01:16:57.000 --> 01:16:58.760 align:middle line:90% is going to be pretty rare. 01:16:58.760 --> 01:17:01.870 align:middle line:90% So let me stop there. 01:17:01.870 --> 01:17:06.000 align:middle line:90%