WEBVTT 00:00:00.000 --> 00:00:01.976 align:middle line:90% [SQUEAKING] 00:00:01.976 --> 00:00:03.952 align:middle line:90% [RUSTLING] 00:00:03.952 --> 00:00:06.916 align:middle line:90% [CLICKING] 00:00:06.916 --> 00:00:18.707 align:middle line:90% 00:00:18.707 --> 00:00:19.290 align:middle line:90% BEN OLKEN: OK. 00:00:19.290 --> 00:00:20.810 align:middle line:90% So where are we? 00:00:20.810 --> 00:00:22.560 align:middle line:84% So we're just finishing up the last couple 00:00:22.560 --> 00:00:24.810 align:middle line:84% of things I wanted to talk about redistribution programs 00:00:24.810 --> 00:00:25.727 align:middle line:90% and transfer programs. 00:00:25.727 --> 00:00:28.890 align:middle line:84% And then I'm going to talk about tax, OK? 00:00:28.890 --> 00:00:29.550 align:middle line:90% All right. 00:00:29.550 --> 00:00:35.130 align:middle line:84% So did I talk about cash versus-- 00:00:35.130 --> 00:00:37.650 align:middle line:84% did I talk about the conceptual versus supply shocks, 00:00:37.650 --> 00:00:39.330 align:middle line:84% and man shocks, and all that stuff? 00:00:39.330 --> 00:00:39.900 align:middle line:90% OK, great. 00:00:39.900 --> 00:00:42.238 align:middle line:84% But I didn't show you the results, is that right? 00:00:42.238 --> 00:00:42.780 align:middle line:90% That's right? 00:00:42.780 --> 00:00:43.280 align:middle line:90% OK, right. 00:00:43.280 --> 00:00:45.810 align:middle line:90% So just to remind you, so yeah. 00:00:45.810 --> 00:00:48.540 align:middle line:84% I think this is what we went through last time, right? 00:00:48.540 --> 00:00:51.480 align:middle line:84% So the point about these cash versus in-kind programs 00:00:51.480 --> 00:00:57.210 align:middle line:84% is one thing you might think about is that a cash program is 00:00:57.210 --> 00:01:00.990 align:middle line:84% a demand, going to be a demand shock, 00:01:00.990 --> 00:01:05.850 align:middle line:84% whereas an in-kind program may also be a supply shock, right? 00:01:05.850 --> 00:01:12.180 align:middle line:90% So this is very basic, right? 00:01:12.180 --> 00:01:15.932 align:middle line:90% What happens if we do can-- 00:01:15.932 --> 00:01:17.640 align:middle line:84% you can actually do it in different ways. 00:01:17.640 --> 00:01:21.250 align:middle line:90% 00:01:21.250 --> 00:01:22.510 align:middle line:90% It's as simple as-- 00:01:22.510 --> 00:01:25.690 align:middle line:84% I should draw it, probably, right? 00:01:25.690 --> 00:01:30.018 align:middle line:84% So this is the supply of whatever the good is. 00:01:30.018 --> 00:01:31.310 align:middle line:90% This is the demand of the good. 00:01:31.310 --> 00:01:34.460 align:middle line:84% If I give you cash, I'm holding the supply of the good-- 00:01:34.460 --> 00:01:37.190 align:middle line:84% if I give you if a cash transfer, 00:01:37.190 --> 00:01:39.080 align:middle line:84% that's going to increase the demand. 00:01:39.080 --> 00:01:41.510 align:middle line:84% If I give an in-kind transfer, then 00:01:41.510 --> 00:01:45.020 align:middle line:84% the government may not only increase the demand, 00:01:45.020 --> 00:01:47.300 align:middle line:84% but also may change the total aggregate supply. 00:01:47.300 --> 00:01:51.390 align:middle line:84% And so there may be differences in terms of the price effects 00:01:51.390 --> 00:01:52.640 align:middle line:90% from these different programs. 00:01:52.640 --> 00:01:53.960 align:middle line:90% Yeah, Ahmed? 00:01:53.960 --> 00:01:57.860 align:middle line:84% AUDIENCE: The system [INAUDIBLE] transferred. 00:01:57.860 --> 00:02:00.072 align:middle line:84% Some say [INAUDIBLE] transfer is like rice. 00:02:00.072 --> 00:02:01.030 align:middle line:90% BEN OLKEN: For example. 00:02:01.030 --> 00:02:01.430 align:middle line:90% BEN OLKEN: For example. 00:02:01.430 --> 00:02:03.472 align:middle line:84% AUDIENCE: It has this little holes full of rice 00:02:03.472 --> 00:02:05.930 align:middle line:84% and it just [? send into ?] different provinces or regions. 00:02:05.930 --> 00:02:09.050 align:middle line:84% But it can also very well be that our government just buys 00:02:09.050 --> 00:02:11.240 align:middle line:90% from a market in that region. 00:02:11.240 --> 00:02:12.260 align:middle line:90% BEN OLKEN: Totally, yes. 00:02:12.260 --> 00:02:12.920 align:middle line:90% Exactly. 00:02:12.920 --> 00:02:14.840 align:middle line:90% And so you're totally right. 00:02:14.840 --> 00:02:15.920 align:middle line:90% If you think about-- 00:02:15.920 --> 00:02:17.270 align:middle line:90% it's actually not obvious. 00:02:17.270 --> 00:02:17.840 align:middle line:90% Where is it? 00:02:17.840 --> 00:02:22.970 align:middle line:84% If you think about housing, so this example works really well 00:02:22.970 --> 00:02:27.715 align:middle line:84% if you think about, say, public housing, right? 00:02:27.715 --> 00:02:29.090 align:middle line:84% In public housing, the government 00:02:29.090 --> 00:02:32.600 align:middle line:84% goes and builds houses as opposed to just providing 00:02:32.600 --> 00:02:33.170 align:middle line:90% vouchers. 00:02:33.170 --> 00:02:35.450 align:middle line:84% That's clearly a supply shock, right? 00:02:35.450 --> 00:02:37.240 align:middle line:84% I agree with you for something like rice. 00:02:37.240 --> 00:02:38.990 align:middle line:84% It's a little more complicated and depends 00:02:38.990 --> 00:02:40.945 align:middle line:84% on where is the government sourcing its rice. 00:02:40.945 --> 00:02:42.320 align:middle line:84% It's the cleanest example is they 00:02:42.320 --> 00:02:44.420 align:middle line:84% source it outside the country and bring it in. 00:02:44.420 --> 00:02:47.270 align:middle line:90% Obviously a supply shock, right? 00:02:47.270 --> 00:02:49.580 align:middle line:84% I think in some cases, if they are just 00:02:49.580 --> 00:02:52.670 align:middle line:84% going into the super local market 00:02:52.670 --> 00:02:54.732 align:middle line:84% and buying it right there, not a supply shock. 00:02:54.732 --> 00:02:56.690 align:middle line:84% And so actually, that's related to what they're 00:02:56.690 --> 00:02:57.773 align:middle line:90% going to do in this paper. 00:02:57.773 --> 00:02:59.930 align:middle line:90% So they're going to talk about-- 00:02:59.930 --> 00:03:03.320 align:middle line:84% they cut this, basically, by the remoteness of the regions. 00:03:03.320 --> 00:03:06.510 align:middle line:84% And the idea is that basically, it's related. 00:03:06.510 --> 00:03:09.770 align:middle line:84% So the basic idea is in really remote locations, what 00:03:09.770 --> 00:03:12.842 align:middle line:84% the government is doing is it probably is a net supply shock. 00:03:12.842 --> 00:03:14.300 align:middle line:84% It's sourcing the stuff externally, 00:03:14.300 --> 00:03:16.400 align:middle line:84% bringing it into the remote regions. 00:03:16.400 --> 00:03:20.870 align:middle line:84% In less remote regions, some combination of markets 00:03:20.870 --> 00:03:24.580 align:middle line:84% are more elastic because they are just thicker. 00:03:24.580 --> 00:03:25.580 align:middle line:90% They're thicker markets. 00:03:25.580 --> 00:03:28.520 align:middle line:84% So maybe the supply curve is more elastic, right? 00:03:28.520 --> 00:03:32.040 align:middle line:84% If a supply curve, obviously, is totally elastic, 00:03:32.040 --> 00:03:35.150 align:middle line:90% then this doesn't matter, right? 00:03:35.150 --> 00:03:38.250 align:middle line:84% So in those less remote regions, it could be that. 00:03:38.250 --> 00:03:39.938 align:middle line:84% Or it could be that you're buying 00:03:39.938 --> 00:03:41.480 align:middle line:84% in those regions and sourcing anyway. 00:03:41.480 --> 00:03:44.570 align:middle line:84% So for those two reasons, I think they focus 00:03:44.570 --> 00:03:46.770 align:middle line:90% on these more remote locations. 00:03:46.770 --> 00:03:48.920 align:middle line:90% That's a really good point. 00:03:48.920 --> 00:03:50.700 align:middle line:84% And just what's the point of this? 00:03:50.700 --> 00:03:54.890 align:middle line:84% The point is that if they look at the price of the goods, 00:03:54.890 --> 00:04:00.830 align:middle line:84% they find a substantial increase, a substantial change 00:04:00.830 --> 00:04:02.330 align:middle line:84% in the price, where the price is-- 00:04:02.330 --> 00:04:04.537 align:middle line:84% and by the way, this is relative to a control. 00:04:04.537 --> 00:04:05.870 align:middle line:90% Neither of these is significant. 00:04:05.870 --> 00:04:08.010 align:middle line:84% But the difference between them is significant. 00:04:08.010 --> 00:04:10.880 align:middle line:84% So they find that prices are lower in the in-kind areas 00:04:10.880 --> 00:04:15.740 align:middle line:84% relative to the cash areas in the remote locations. 00:04:15.740 --> 00:04:19.630 align:middle line:84% And then that can affect consumption 00:04:19.630 --> 00:04:22.130 align:middle line:84% and people's overall well-being from the program because you 00:04:22.130 --> 00:04:24.260 align:middle line:90% get less-- 00:04:24.260 --> 00:04:27.103 align:middle line:84% in real quantity terms, you get more 00:04:27.103 --> 00:04:29.270 align:middle line:84% in the in-kind areas because of these price effects. 00:04:29.270 --> 00:04:33.920 align:middle line:84% And that's basically what they argue. 00:04:33.920 --> 00:04:36.860 align:middle line:84% So the in-kind area is people's total consumption is actually 00:04:36.860 --> 00:04:40.770 align:middle line:90% higher for that reason. 00:04:40.770 --> 00:04:44.462 align:middle line:84% So let me actually not go through the details of this 00:04:44.462 --> 00:04:45.920 align:middle line:84% because it's a pretty simple point. 00:04:45.920 --> 00:04:46.640 align:middle line:84% But I just want to see if there are 00:04:46.640 --> 00:04:48.230 align:middle line:84% any just mention of those two things, which 00:04:48.230 --> 00:04:50.272 align:middle line:84% seem like they should be equivalent in principle, 00:04:50.272 --> 00:04:52.340 align:middle line:84% but actually have these different reasons 00:04:52.340 --> 00:04:53.670 align:middle line:90% for basic price theory reasons. 00:04:53.670 --> 00:04:54.920 align:middle line:90% Yeah, Tishara? 00:04:54.920 --> 00:04:57.635 align:middle line:84% AUDIENCE: How do we think about programs where government buys, 00:04:57.635 --> 00:05:00.410 align:middle line:84% say, rice at a high price from the producers 00:05:00.410 --> 00:05:03.162 align:middle line:84% but essentially gives it out for free [INAUDIBLE]?? 00:05:03.162 --> 00:05:04.370 align:middle line:90% BEN OLKEN: What, what's that? 00:05:04.370 --> 00:05:05.995 align:middle line:84% AUDIENCE: So government is buying, say, 00:05:05.995 --> 00:05:07.880 align:middle line:84% rice from the producers at a minimum support 00:05:07.880 --> 00:05:10.190 align:middle line:84% price, which could be higher than the market price, 00:05:10.190 --> 00:05:13.160 align:middle line:84% and then sells it, sends it to the consumers essentially 00:05:13.160 --> 00:05:15.310 align:middle line:84% for a very low price or for free. 00:05:15.310 --> 00:05:20.050 align:middle line:84% BEN OLKEN: Yeah, that's still going to have-- 00:05:20.050 --> 00:05:22.280 align:middle line:84% it's still going to look like this in the-- 00:05:22.280 --> 00:05:22.780 align:middle line:90% oh, sorry. 00:05:22.780 --> 00:05:23.080 align:middle line:90% No. 00:05:23.080 --> 00:05:24.190 align:middle line:84% That's a different program, sorry. 00:05:24.190 --> 00:05:25.960 align:middle line:84% So you're saying there's another program. 00:05:25.960 --> 00:05:28.970 align:middle line:90% 00:05:28.970 --> 00:05:31.970 align:middle line:84% I think you're saying-- so this would be the market level. 00:05:31.970 --> 00:05:34.850 align:middle line:84% And you're saying the government basically comes 00:05:34.850 --> 00:05:37.550 align:middle line:90% in and goes like this, right? 00:05:37.550 --> 00:05:40.050 align:middle line:84% I think you think of it like-- if you think of it like that, 00:05:40.050 --> 00:05:41.008 align:middle line:90% what's going to happen? 00:05:41.008 --> 00:05:44.400 align:middle line:84% So it's going to raise the price that producers are selling it 00:05:44.400 --> 00:05:44.900 align:middle line:90% at. 00:05:44.900 --> 00:05:46.910 align:middle line:84% It's going to lower the price people are consuming it at. 00:05:46.910 --> 00:05:48.410 align:middle line:84% And it's going to inefficiently increase 00:05:48.410 --> 00:05:49.970 align:middle line:84% the amount of the good, or it's going 00:05:49.970 --> 00:05:51.920 align:middle line:84% to increase the amount of the good consumed 00:05:51.920 --> 00:05:53.590 align:middle line:84% above and beyond the market level. 00:05:53.590 --> 00:05:55.340 align:middle line:84% And so the way you should think about this 00:05:55.340 --> 00:05:58.070 align:middle line:84% is like a redistribution program. 00:05:58.070 --> 00:06:00.890 align:middle line:84% First of all, we want to think about who 00:06:00.890 --> 00:06:03.170 align:middle line:84% the welfare of these people are, right? 00:06:03.170 --> 00:06:04.430 align:middle line:90% Are they rich or poor? 00:06:04.430 --> 00:06:07.250 align:middle line:84% Is that a good thing to be doing or not? 00:06:07.250 --> 00:06:10.137 align:middle line:84% Relative to who are these different people? 00:06:10.137 --> 00:06:11.720 align:middle line:84% Second thing is we should think about, 00:06:11.720 --> 00:06:13.820 align:middle line:84% do we actually want more consumption 00:06:13.820 --> 00:06:15.380 align:middle line:90% of this good or not, right? 00:06:15.380 --> 00:06:18.485 align:middle line:84% So for example, governments often-- 00:06:18.485 --> 00:06:20.840 align:middle line:90% 00:06:20.840 --> 00:06:22.340 align:middle line:84% and the third thing we want to think 00:06:22.340 --> 00:06:25.880 align:middle line:84% about is, who are the people who are induced to buy more of it 00:06:25.880 --> 00:06:26.870 align:middle line:90% and get the subsidy? 00:06:26.870 --> 00:06:30.030 align:middle line:84% Are they the people you want to be targeting? 00:06:30.030 --> 00:06:32.300 align:middle line:84% So if this is an inferior good, this actually 00:06:32.300 --> 00:06:33.368 align:middle line:90% may not look so bad. 00:06:33.368 --> 00:06:34.910 align:middle line:84% If this is a normal good, then you're 00:06:34.910 --> 00:06:37.682 align:middle line:84% actually giving a lot of the subsidy to richer people. 00:06:37.682 --> 00:06:39.140 align:middle line:84% And that might not be so efficient. 00:06:39.140 --> 00:06:40.765 align:middle line:84% And the other thing is, do you actually 00:06:40.765 --> 00:06:46.390 align:middle line:84% care that you're distorting the overall consumption 00:06:46.390 --> 00:06:47.390 align:middle line:90% of this particular good? 00:06:47.390 --> 00:06:48.682 align:middle line:90% This is a good think is being-- 00:06:48.682 --> 00:06:50.307 align:middle line:84% suppose you think this is a good that's 00:06:50.307 --> 00:06:52.007 align:middle line:84% being inefficiently underconsumed 00:06:52.007 --> 00:06:53.090 align:middle line:90% in the market equilibrium. 00:06:53.090 --> 00:06:55.710 align:middle line:84% You think maybe people are undernourished, for example, 00:06:55.710 --> 00:06:57.110 align:middle line:84% and you want them to consume more of this at the market 00:06:57.110 --> 00:06:58.580 align:middle line:90% price-- this looks great. 00:06:58.580 --> 00:07:00.988 align:middle line:84% If this is, I don't know, gasoline or something, 00:07:00.988 --> 00:07:02.780 align:middle line:84% where maybe you think that actually there's 00:07:02.780 --> 00:07:05.322 align:middle line:84% no particular reason we think we're under consuming gasoline, 00:07:05.322 --> 00:07:06.500 align:middle line:90% it's terrible. 00:07:06.500 --> 00:07:10.080 align:middle line:84% So I think that's the way I would think about it. 00:07:10.080 --> 00:07:12.260 align:middle line:84% But the key difference in that policy 00:07:12.260 --> 00:07:16.030 align:middle line:84% is it's changing the marginal price, right? 00:07:16.030 --> 00:07:17.020 align:middle line:90% And that's why it has-- 00:07:17.020 --> 00:07:20.960 align:middle line:84% whereas those things are not changing the marginal press. 00:07:20.960 --> 00:07:27.230 align:middle line:84% And the [INAUDIBLE] paper works through all this stuff with 00:07:27.230 --> 00:07:28.250 align:middle line:90% a poverty-- 00:07:28.250 --> 00:07:31.093 align:middle line:84% thinking about heterogeneity in the population, 00:07:31.093 --> 00:07:32.510 align:middle line:84% and poverty, and who do we think-- 00:07:32.510 --> 00:07:33.650 align:middle line:84% how do we think of all the different agents. 00:07:33.650 --> 00:07:36.607 align:middle line:84% And that's a very nice framework if you want to think about that 00:07:36.607 --> 00:07:38.940 align:middle line:84% and compare the inframarginal versus marginal transfers. 00:07:38.940 --> 00:07:41.460 align:middle line:90% 00:07:41.460 --> 00:07:44.270 align:middle line:90% Other questions? 00:07:44.270 --> 00:07:44.890 align:middle line:90% OK. 00:07:44.890 --> 00:07:48.855 align:middle line:84% Oh, so the other paper I wanted to mention-- 00:07:48.855 --> 00:07:50.980 align:middle line:84% and again, I'm just giving you just a brief flavor. 00:07:50.980 --> 00:07:53.397 align:middle line:84% There's an enormous literature, an expanding literature 00:07:53.397 --> 00:07:54.730 align:middle line:90% on the design of these programs. 00:07:54.730 --> 00:07:56.897 align:middle line:84% But I just wanted to mention a couple of things just 00:07:56.897 --> 00:07:58.605 align:middle line:84% to give you a sense of some of the things 00:07:58.605 --> 00:07:59.630 align:middle line:90% people are looking at. 00:07:59.630 --> 00:08:03.130 align:middle line:84% So how many of you have heard of conditional cash transfer 00:08:03.130 --> 00:08:04.770 align:middle line:90% programs? 00:08:04.770 --> 00:08:06.450 align:middle line:90% Maybe most of you, OK, right? 00:08:06.450 --> 00:08:09.480 align:middle line:84% So these are programs that are very common now 00:08:09.480 --> 00:08:11.447 align:middle line:90% in the developing world. 00:08:11.447 --> 00:08:13.530 align:middle line:84% They condition aid on fulfilling a set of criteria 00:08:13.530 --> 00:08:15.613 align:middle line:84% the government thinks are good, like for example, 00:08:15.613 --> 00:08:18.030 align:middle line:84% sending your kids to school or getting them health shocks, 00:08:18.030 --> 00:08:18.360 align:middle line:90% or whatever. 00:08:18.360 --> 00:08:20.235 align:middle line:84% You get the cash grants to do this stuff, OK? 00:08:20.235 --> 00:08:21.750 align:middle line:84% So the government is incentivizing 00:08:21.750 --> 00:08:23.460 align:middle line:90% these additional behaviors. 00:08:23.460 --> 00:08:26.160 align:middle line:84% And there were several randomized trials 00:08:26.160 --> 00:08:30.930 align:middle line:84% that showed that those did increase the desired behavior 00:08:30.930 --> 00:08:32.331 align:middle line:90% or the incentivized behavior. 00:08:32.331 --> 00:08:34.289 align:middle line:84% Not surprising because it's being incentivized. 00:08:34.289 --> 00:08:37.950 align:middle line:90% But it turns out to be there. 00:08:37.950 --> 00:08:40.682 align:middle line:84% And the justification for that is intergenerational, right? 00:08:40.682 --> 00:08:42.390 align:middle line:84% The idea is that many parents don't fully 00:08:42.390 --> 00:08:44.790 align:middle line:84% internalize the value of their kids' human capital 00:08:44.790 --> 00:08:45.330 align:middle line:90% acquisition. 00:08:45.330 --> 00:08:46.830 align:middle line:84% And governments can correct for this 00:08:46.830 --> 00:08:50.510 align:middle line:84% by incentivizing these behaviors. 00:08:50.510 --> 00:08:52.700 align:middle line:90% What this paper was-- 00:08:52.700 --> 00:08:55.760 align:middle line:84% first paper that I know of to, in a really rigorous way, 00:08:55.760 --> 00:09:00.710 align:middle line:84% compare what is the impact of the conditionality per se? 00:09:00.710 --> 00:09:05.420 align:middle line:84% So those programs before were bundled, 00:09:05.420 --> 00:09:08.307 align:middle line:84% where there's overall randomized evaluation 00:09:08.307 --> 00:09:10.640 align:middle line:84% of the conditional cash transfer developed from nothing, 00:09:10.640 --> 00:09:12.590 align:middle line:84% but didn't say, is this conditioning thing 00:09:12.590 --> 00:09:14.540 align:middle line:90% actually a good idea or not? 00:09:14.540 --> 00:09:17.000 align:middle line:90% And this paper is nice. 00:09:17.000 --> 00:09:20.180 align:middle line:84% It examines conditional and unconditional transfers 00:09:20.180 --> 00:09:22.483 align:middle line:90% in the same context. 00:09:22.483 --> 00:09:23.900 align:middle line:84% And so how do we think about that? 00:09:23.900 --> 00:09:27.150 align:middle line:84% Well, the first thing you would think about 00:09:27.150 --> 00:09:29.137 align:middle line:84% if you condition on something is, of course 00:09:29.137 --> 00:09:30.720 align:middle line:84% it's going to change the price, right? 00:09:30.720 --> 00:09:32.803 align:middle line:84% That's good if you want incentivize that behavior, 00:09:32.803 --> 00:09:33.690 align:middle line:90% right? 00:09:33.690 --> 00:09:38.340 align:middle line:84% But the other point, which I don't think people 00:09:38.340 --> 00:09:42.375 align:middle line:84% necessarily-- it's maybe obvious ex post, but wasn't, I think, 00:09:42.375 --> 00:09:44.250 align:middle line:84% obvious until people had written this paper-- 00:09:44.250 --> 00:09:48.420 align:middle line:84% is if there are people who are really far from the margin who 00:09:48.420 --> 00:09:50.910 align:middle line:84% are not going to take up the incentivized behavior, 00:09:50.910 --> 00:09:53.730 align:middle line:84% then you cut them off from the cash transfer, right? 00:09:53.730 --> 00:09:57.570 align:middle line:84% So a rigidly enforced conditional cash transfer 00:09:57.570 --> 00:10:00.160 align:middle line:84% is not going to be providing the transfer to people 00:10:00.160 --> 00:10:01.410 align:middle line:90% who can't meet the conditions. 00:10:01.410 --> 00:10:03.420 align:middle line:90% That's the whole point. 00:10:03.420 --> 00:10:05.920 align:middle line:84% And therefore, those households don't get the income effects 00:10:05.920 --> 00:10:06.490 align:middle line:90% from the transfer. 00:10:06.490 --> 00:10:07.600 align:middle line:90% And that may also be important. 00:10:07.600 --> 00:10:08.210 align:middle line:90% OK? 00:10:08.210 --> 00:10:09.520 align:middle line:90% So how do they show this? 00:10:09.520 --> 00:10:12.430 align:middle line:90% 00:10:12.430 --> 00:10:14.298 align:middle line:84% I should say, by the way, there's 00:10:14.298 --> 00:10:16.340 align:middle line:84% another paper, which I'm not going to talk about, 00:10:16.340 --> 00:10:18.160 align:middle line:84% which is about the labeling effect of the conditional cash 00:10:18.160 --> 00:10:20.032 align:middle line:84% transfer, which just says if you tell people 00:10:20.032 --> 00:10:22.240 align:middle line:84% the conditional cash transfer it's a conditional cash 00:10:22.240 --> 00:10:25.870 align:middle line:84% transfer, but don't actually enforce the conditions, 00:10:25.870 --> 00:10:29.230 align:middle line:84% people may think they're enforced and do them anyway. 00:10:29.230 --> 00:10:32.175 align:middle line:84% And that paper says a labeled cash transfer 00:10:32.175 --> 00:10:33.550 align:middle line:84% versus an actual conditional cash 00:10:33.550 --> 00:10:35.592 align:middle line:84% transfer-- you may get a lot of the benefits just 00:10:35.592 --> 00:10:38.800 align:middle line:84% from the labeling without this negative cost. 00:10:38.800 --> 00:10:40.333 align:middle line:90% But what do they do in any case? 00:10:40.333 --> 00:10:41.750 align:middle line:84% So what do they find in this case? 00:10:41.750 --> 00:10:42.667 align:middle line:90% So let's look at this. 00:10:42.667 --> 00:10:45.050 align:middle line:84% So what are some of the things I wanted to note? 00:10:45.050 --> 00:10:48.490 align:middle line:84% So the first is let's look at actual enroll-- 00:10:48.490 --> 00:10:51.700 align:middle line:84% so they condition on enrolling your kids at school. 00:10:51.700 --> 00:10:53.770 align:middle line:84% They have these self-reported enrollment. 00:10:53.770 --> 00:10:55.330 align:middle line:84% But then they have teacher-reported enrollment. 00:10:55.330 --> 00:10:57.038 align:middle line:84% And they have some arguments in the paper 00:10:57.038 --> 00:11:00.970 align:middle line:84% that the teacher-reported enrollment is the better one. 00:11:00.970 --> 00:11:03.168 align:middle line:84% So lo and behold, not super surprising, 00:11:03.168 --> 00:11:05.710 align:middle line:84% the conditional cash treatment, which is incentivizing people 00:11:05.710 --> 00:11:08.900 align:middle line:84% to go to school, results in more kids going to school. 00:11:08.900 --> 00:11:09.400 align:middle line:90% OK? 00:11:09.400 --> 00:11:11.380 align:middle line:90% So that's that. 00:11:11.380 --> 00:11:15.440 align:middle line:84% And you can see it for three years or whatever. 00:11:15.440 --> 00:11:19.480 align:middle line:84% And those things are statistically significantly 00:11:19.480 --> 00:11:23.570 align:middle line:84% different in most of the specifications. 00:11:23.570 --> 00:11:27.350 align:middle line:84% And they do better on their test scores, OK? 00:11:27.350 --> 00:11:28.100 align:middle line:90% Fine. 00:11:28.100 --> 00:11:29.810 align:middle line:84% So it seems like it works in terms-- 00:11:29.810 --> 00:11:33.450 align:middle line:84% the conditionality is important for the incentivized behaviors. 00:11:33.450 --> 00:11:39.420 align:middle line:84% On the other hand, they also then looked at early marriage, 00:11:39.420 --> 00:11:41.290 align:middle line:90% basically, as another outcome. 00:11:41.290 --> 00:11:44.490 align:middle line:84% So early marriage particularly for girls in the household. 00:11:44.490 --> 00:11:47.405 align:middle line:84% And early marriage is not incentivized outcome. 00:11:47.405 --> 00:11:48.780 align:middle line:84% But it's something that you might 00:11:48.780 --> 00:11:51.887 align:middle line:84% think that preventing early marriage 00:11:51.887 --> 00:11:53.970 align:middle line:84% might be something the government might care about 00:11:53.970 --> 00:11:56.760 align:middle line:84% in this context even though they're not incentivizing it. 00:11:56.760 --> 00:12:00.720 align:middle line:84% And what they find is that actually, the fraction who 00:12:00.720 --> 00:12:06.260 align:middle line:84% were ever married in this period is actually 00:12:06.260 --> 00:12:09.750 align:middle line:84% lower in the unconditional treatment 00:12:09.750 --> 00:12:11.190 align:middle line:84% than in the conditional treatment. 00:12:11.190 --> 00:12:14.030 align:middle line:84% So both of these treatments lower early marriage 00:12:14.030 --> 00:12:16.730 align:middle line:84% compared to the control, but it's larger 00:12:16.730 --> 00:12:18.630 align:middle line:90% in the unconditional treatment. 00:12:18.630 --> 00:12:21.550 align:middle line:84% And what they argue that's going on-- you see it here-- 00:12:21.550 --> 00:12:29.580 align:middle line:84% which is if you look among those who are not enrolled in school, 00:12:29.580 --> 00:12:34.162 align:middle line:84% so these are the people who would be not getting the-- 00:12:34.162 --> 00:12:35.370 align:middle line:90% are endogenous groups, right? 00:12:35.370 --> 00:12:37.740 align:middle line:84% So there's fewer people not enrolled 00:12:37.740 --> 00:12:40.830 align:middle line:84% in school in the conditional treatment 00:12:40.830 --> 00:12:43.230 align:middle line:84% than in the unconditional treatment, right? 00:12:43.230 --> 00:12:45.070 align:middle line:90% This is the share of the group. 00:12:45.070 --> 00:12:46.770 align:middle line:90% So that's not surprising. 00:12:46.770 --> 00:12:48.450 align:middle line:84% That's the condition that we just saw, 00:12:48.450 --> 00:12:52.410 align:middle line:84% that the conditional treatment reduces the number 00:12:52.410 --> 00:12:53.970 align:middle line:90% of people enrolled in school-- 00:12:53.970 --> 00:12:55.810 align:middle line:90% who are not enrolled, sorry. 00:12:55.810 --> 00:12:58.540 align:middle line:84% But there's a big effect of the program 00:12:58.540 --> 00:13:02.230 align:middle line:84% on early marriage for the not-enrolled group 00:13:02.230 --> 00:13:06.430 align:middle line:84% in the unconditional group, but not in the conditional group. 00:13:06.430 --> 00:13:07.300 align:middle line:90% And why is that? 00:13:07.300 --> 00:13:10.870 align:middle line:84% Because these people over here don't get the transfer. 00:13:10.870 --> 00:13:13.335 align:middle line:84% This is not-- this cross tab here is not 00:13:13.335 --> 00:13:15.460 align:middle line:84% super clean because this is an endogenous variable. 00:13:15.460 --> 00:13:17.043 align:middle line:84% There could be samples not shown here. 00:13:17.043 --> 00:13:19.390 align:middle line:84% But I think that this is a pretty natural explanation 00:13:19.390 --> 00:13:21.310 align:middle line:84% for-- but these effects are well-identified. 00:13:21.310 --> 00:13:22.977 align:middle line:84% And this is a pretty natural explanation 00:13:22.977 --> 00:13:24.910 align:middle line:84% for what's going on over here, right? 00:13:24.910 --> 00:13:26.778 align:middle line:84% So basically, what happens is there's-- 00:13:26.778 --> 00:13:29.320 align:middle line:84% I think the way to think about it is there's a bunch of kids, 00:13:29.320 --> 00:13:29.820 align:middle line:90% right? 00:13:29.820 --> 00:13:32.830 align:middle line:84% Some of them are on the margin for going to school or not. 00:13:32.830 --> 00:13:35.810 align:middle line:84% And they don't-- in the conditional treatment, 00:13:35.810 --> 00:13:37.060 align:middle line:90% some of them switch over here. 00:13:37.060 --> 00:13:38.180 align:middle line:90% They don't go to school-- 00:13:38.180 --> 00:13:39.880 align:middle line:84% I mean, they go to school, right? 00:13:39.880 --> 00:13:42.640 align:middle line:84% But there's another set of kids who are just not on the margin 00:13:42.640 --> 00:13:43.880 align:middle line:90% for going to school or not. 00:13:43.880 --> 00:13:46.210 align:middle line:84% And for those kids, the income effect 00:13:46.210 --> 00:13:47.380 align:middle line:90% could be really important. 00:13:47.380 --> 00:13:49.360 align:middle line:84% And you're taking away the income effect 00:13:49.360 --> 00:13:54.990 align:middle line:84% for the kids who are not enrolled 00:13:54.990 --> 00:13:56.550 align:middle line:90% in the conditional treatment. 00:13:56.550 --> 00:14:05.050 align:middle line:84% And so this basically analyzes this, 00:14:05.050 --> 00:14:10.627 align:middle line:84% I think, a little more, basically showing you 00:14:10.627 --> 00:14:11.710 align:middle line:90% the same thing it entails. 00:14:11.710 --> 00:14:12.340 align:middle line:90% Actually, maybe not. 00:14:12.340 --> 00:14:13.280 align:middle line:90% I don't need to go through that. 00:14:13.280 --> 00:14:14.980 align:middle line:84% But I think this is the key insight. 00:14:14.980 --> 00:14:16.820 align:middle line:90% Yeah, Ahmed. 00:14:16.820 --> 00:14:20.143 align:middle line:84% AUDIENCE: So what treatments are unconditional-- 00:14:20.143 --> 00:14:20.810 align:middle line:90% BEN OLKEN: What? 00:14:20.810 --> 00:14:24.230 align:middle line:84% AUDIENCE: When the treatment is unconditional, 00:14:24.230 --> 00:14:26.820 align:middle line:84% it affects a larger number of people. 00:14:26.820 --> 00:14:29.030 align:middle line:90% But then does this cost more? 00:14:29.030 --> 00:14:31.490 align:middle line:90% So are we comparing-- 00:14:31.490 --> 00:14:33.098 align:middle line:90% and is the treatment the same-- 00:14:33.098 --> 00:14:34.390 align:middle line:90% BEN OLKEN: It's more expensive. 00:14:34.390 --> 00:14:36.580 align:middle line:90% You're right. 00:14:36.580 --> 00:14:37.580 align:middle line:90% AUDIENCE: We are still-- 00:14:37.580 --> 00:14:38.840 align:middle line:90% BEN OLKEN: Because you're right. 00:14:38.840 --> 00:14:40.900 align:middle line:84% In the conditional treatment, you don't pay the-- 00:14:40.900 --> 00:14:43.730 align:middle line:84% well, actually, there's two different effects. 00:14:43.730 --> 00:14:46.490 align:middle line:84% Number one is, in practice, actually monitoring 00:14:46.490 --> 00:14:48.710 align:middle line:84% all the conditions is actually not free. 00:14:48.710 --> 00:14:50.900 align:middle line:84% So you do get a cost savings over here 00:14:50.900 --> 00:14:53.330 align:middle line:84% from not having to actually go and do all that monitoring 00:14:53.330 --> 00:14:53.880 align:middle line:90% or whatever. 00:14:53.880 --> 00:14:54.500 align:middle line:90% So that's one point. 00:14:54.500 --> 00:14:56.000 align:middle line:84% But on the other hand, you're right. 00:14:56.000 --> 00:14:59.300 align:middle line:84% You are saving-- quote "saving" unquote. 00:14:59.300 --> 00:15:02.923 align:middle line:84% You're not paying the cost of those transfers 00:15:02.923 --> 00:15:04.590 align:middle line:84% for the people that are getting cut off, 00:15:04.590 --> 00:15:07.340 align:middle line:84% which is 30% of the sample over here. 00:15:07.340 --> 00:15:08.210 align:middle line:90% So you're right. 00:15:08.210 --> 00:15:13.580 align:middle line:84% The transfers you're giving out are whatever. 00:15:13.580 --> 00:15:15.440 align:middle line:84% Here you're only giving out 70% of the cash 00:15:15.440 --> 00:15:17.030 align:middle line:84% compared what you're giving out here. 00:15:17.030 --> 00:15:17.960 align:middle line:90% So you're right. 00:15:17.960 --> 00:15:20.168 align:middle line:84% This program is probably a bit cheaper in that sense. 00:15:20.168 --> 00:15:22.008 align:middle line:90% 00:15:22.008 --> 00:15:24.550 align:middle line:84% I suspect these differences are sufficiently stark that maybe 00:15:24.550 --> 00:15:27.160 align:middle line:90% that's not going to do it. 00:15:27.160 --> 00:15:29.010 align:middle line:90% But it's a fair point. 00:15:29.010 --> 00:15:30.880 align:middle line:90% Other questions? 00:15:30.880 --> 00:15:31.380 align:middle line:90% OK. 00:15:31.380 --> 00:15:32.255 align:middle line:90% So I think that was-- 00:15:32.255 --> 00:15:35.845 align:middle line:90% 00:15:35.845 --> 00:15:38.680 align:middle line:84% I just want to mention that everyone always thinks 00:15:38.680 --> 00:15:40.510 align:middle line:90% that the conditions are good. 00:15:40.510 --> 00:15:43.780 align:middle line:84% But it's worth keeping those in mind as well. 00:15:43.780 --> 00:15:45.240 align:middle line:90% I'm not going to-- 00:15:45.240 --> 00:15:50.077 align:middle line:84% I wanted to just put that out there as a pointer-- 00:15:50.077 --> 00:15:51.660 align:middle line:84% those two papers, almost, as a pointer 00:15:51.660 --> 00:15:55.770 align:middle line:84% to a whole literature, which I don't have the bandwidth here 00:15:55.770 --> 00:15:58.120 align:middle line:90% to go through in detail. 00:15:58.120 --> 00:16:00.540 align:middle line:90% But understanding the design-- 00:16:00.540 --> 00:16:03.120 align:middle line:84% as countries are doing more of these transfer schemes 00:16:03.120 --> 00:16:05.520 align:middle line:84% and redistribution schemes, understanding the design 00:16:05.520 --> 00:16:07.650 align:middle line:84% questions involved in doing that properly 00:16:07.650 --> 00:16:10.450 align:middle line:84% is a really expanding area that people are thinking through. 00:16:10.450 --> 00:16:13.255 align:middle line:84% So for example, some recent work-- 00:16:13.255 --> 00:16:14.880 align:middle line:84% we have a paper on, how do you actually 00:16:14.880 --> 00:16:16.630 align:middle line:84% deliver the transfers effectively and does 00:16:16.630 --> 00:16:22.500 align:middle line:84% transparency that gives people information about what 00:16:22.500 --> 00:16:24.170 align:middle line:90% their rights are matter? 00:16:24.170 --> 00:16:25.920 align:middle line:84% One of our recent papers says yes on this. 00:16:25.920 --> 00:16:27.795 align:middle line:84% How about technology and reducing corruption? 00:16:27.795 --> 00:16:29.730 align:middle line:84% There's a lot of interest in what 00:16:29.730 --> 00:16:32.730 align:middle line:84% is the underlying technological infrastructure 00:16:32.730 --> 00:16:34.640 align:middle line:90% of these programs? 00:16:34.640 --> 00:16:37.410 align:middle line:84% smart cards, biometric authentication, 00:16:37.410 --> 00:16:40.230 align:middle line:84% digital payments-- all that stuff is entering. 00:16:40.230 --> 00:16:42.420 align:middle line:84% And understanding that has been a really active area 00:16:42.420 --> 00:16:45.780 align:middle line:84% of research in the last five years or so. 00:16:45.780 --> 00:16:48.540 align:middle line:84% And actually, even just changing the back-end system-- 00:16:48.540 --> 00:16:53.880 align:middle line:84% the nitty gritty of how the government reimburses itself 00:16:53.880 --> 00:16:56.320 align:middle line:84% for these programs can be really important. 00:16:56.320 --> 00:16:59.700 align:middle line:84% We have some papers on government 00:16:59.700 --> 00:17:01.313 align:middle line:84% versus private distribution of this. 00:17:01.313 --> 00:17:03.480 align:middle line:84% There's some work on how do you get people enrolled. 00:17:03.480 --> 00:17:05.310 align:middle line:84% There's questions about more stuff 00:17:05.310 --> 00:17:07.440 align:middle line:90% on the cash versus in-kind. 00:17:07.440 --> 00:17:09.030 align:middle line:84% Should you have transfers be lumpy? 00:17:09.030 --> 00:17:14.783 align:middle line:84% Should they be a series of big lump sums 00:17:14.783 --> 00:17:16.950 align:middle line:84% or a small series of payments, et cetera, et cetera. 00:17:16.950 --> 00:17:21.210 align:middle line:84% There's a number of questions that people are actually 00:17:21.210 --> 00:17:22.740 align:middle line:90% working, studying here. 00:17:22.740 --> 00:17:25.940 align:middle line:84% I'm actually in the middle of writing a lit review on this 00:17:25.940 --> 00:17:28.860 align:middle line:90% with several co-authors. 00:17:28.860 --> 00:17:31.193 align:middle line:84% But I want to mention that if you're interested in this, 00:17:31.193 --> 00:17:33.235 align:middle line:84% I think is a really active area, but there's also 00:17:33.235 --> 00:17:33.883 align:middle line:90% a lot going on. 00:17:33.883 --> 00:17:35.550 align:middle line:84% And so it's worth making sure you get up 00:17:35.550 --> 00:17:37.080 align:middle line:84% to speed if you're interested in these topics. 00:17:37.080 --> 00:17:39.090 align:middle line:84% But there's a lot of active research on this. 00:17:39.090 --> 00:17:41.265 align:middle line:84% OK, any last questions before I switch gears? 00:17:41.265 --> 00:17:44.670 align:middle line:90% 00:17:44.670 --> 00:17:46.800 align:middle line:90% And this is general-- 00:17:46.800 --> 00:17:47.340 align:middle line:90% oh, sorry. 00:17:47.340 --> 00:17:48.750 align:middle line:90% And one last thing. 00:17:48.750 --> 00:17:54.505 align:middle line:84% I think a lot of lower and middle income countries-- 00:17:54.505 --> 00:17:56.130 align:middle line:84% not the poorest countries in the world, 00:17:56.130 --> 00:17:58.980 align:middle line:84% but low middle income countries, are 00:17:58.980 --> 00:18:02.100 align:middle line:84% starting to think about social insurance programs now. 00:18:02.100 --> 00:18:04.140 align:middle line:84% How do we provide health insurance, disability 00:18:04.140 --> 00:18:07.950 align:middle line:84% insurance, weather insurance, crop insurance, 00:18:07.950 --> 00:18:11.770 align:middle line:90% those kinds of schemes. 00:18:11.770 --> 00:18:15.513 align:middle line:84% And I suspect-- and there's just beginning 00:18:15.513 --> 00:18:16.680 align:middle line:90% to be some research on that. 00:18:16.680 --> 00:18:19.820 align:middle line:84% And I suspect that's an important next frontier of how 00:18:19.820 --> 00:18:21.410 align:middle line:84% do we understand those challenges 00:18:21.410 --> 00:18:24.500 align:middle line:84% in these low information and low governance environments? 00:18:24.500 --> 00:18:25.310 align:middle line:90% I think that's-- 00:18:25.310 --> 00:18:28.190 align:middle line:84% I just wanted to mention that as an important next frontier 00:18:28.190 --> 00:18:30.510 align:middle line:90% of future research. 00:18:30.510 --> 00:18:31.010 align:middle line:90% OK. 00:18:31.010 --> 00:18:32.750 align:middle line:84% Any last questions before I switch gears? 00:18:32.750 --> 00:18:35.490 align:middle line:90% 00:18:35.490 --> 00:18:36.240 align:middle line:90% No? 00:18:36.240 --> 00:18:36.930 align:middle line:90% OK. 00:18:36.930 --> 00:18:37.475 align:middle line:90% Yeah? 00:18:37.475 --> 00:18:39.350 align:middle line:84% AUDIENCE: How should we compare this spending 00:18:39.350 --> 00:18:41.350 align:middle line:90% with infrastructure spending? 00:18:41.350 --> 00:18:41.880 align:middle line:84% BEN OLKEN: I'm sorry, [INAUDIBLE].. 00:18:41.880 --> 00:18:42.838 align:middle line:90% You've got to speak up. 00:18:42.838 --> 00:18:43.710 align:middle line:90% AUDIENCE: Sorry. 00:18:43.710 --> 00:18:45.720 align:middle line:84% How should we compare this spending with, say, 00:18:45.720 --> 00:18:48.262 align:middle line:90% infrastructure spending? 00:18:48.262 --> 00:18:48.970 align:middle line:90% Is that possible? 00:18:48.970 --> 00:18:50.440 align:middle line:84% Or do we know anything about that? 00:18:50.440 --> 00:18:59.188 align:middle line:90% 00:18:59.188 --> 00:19:00.675 align:middle line:90% AUDIENCE: [INAUDIBLE] 00:19:00.675 --> 00:19:01.300 align:middle line:90% AUDIENCE: Yeah. 00:19:01.300 --> 00:19:02.105 align:middle line:90% BEN OLKEN: What? 00:19:02.105 --> 00:19:02.680 align:middle line:90% Sorry? 00:19:02.680 --> 00:19:04.780 align:middle line:84% AUDIENCE: Maybe in terms of margin of value of-- 00:19:04.780 --> 00:19:06.610 align:middle line:84% BEN OLKEN: Well, that's exactly what I was going to say. 00:19:06.610 --> 00:19:07.270 align:middle line:90% Yes, exactly. 00:19:07.270 --> 00:19:08.682 align:middle line:90% I was trying to say, well, yes. 00:19:08.682 --> 00:19:10.640 align:middle line:84% You can compute the MBPF of all of these things 00:19:10.640 --> 00:19:11.980 align:middle line:90% and try to think them through. 00:19:11.980 --> 00:19:13.630 align:middle line:90% That's not a-- 00:19:13.630 --> 00:19:14.620 align:middle line:90% I think that the-- 00:19:14.620 --> 00:19:17.155 align:middle line:90% 00:19:17.155 --> 00:19:18.780 align:middle line:84% so that's what I was going to say, yes. 00:19:18.780 --> 00:19:20.570 align:middle line:84% You just figure out a welfare frame. 00:19:20.570 --> 00:19:22.290 align:middle line:90% So two points. 00:19:22.290 --> 00:19:25.830 align:middle line:84% The first is you need an augmented-- 00:19:25.830 --> 00:19:26.330 align:middle line:90% OK. 00:19:26.330 --> 00:19:28.655 align:middle line:84% So for those of you who have taken public finance, 00:19:28.655 --> 00:19:31.280 align:middle line:84% you've probably talked about the marginal value of public funds 00:19:31.280 --> 00:19:34.430 align:middle line:84% as a framework for just thinking about, what 00:19:34.430 --> 00:19:40.940 align:middle line:84% is the marginal benefit of dollar of government spending 00:19:40.940 --> 00:19:43.340 align:middle line:84% relative to the net cost to the government, where 00:19:43.340 --> 00:19:45.862 align:middle line:84% the net cost takes into account the effect of the spending 00:19:45.862 --> 00:19:47.070 align:middle line:90% on future government revenue? 00:19:47.070 --> 00:19:49.940 align:middle line:90% So that's the rough-- 00:19:49.940 --> 00:19:51.320 align:middle line:84% another way of saying it is, what 00:19:51.320 --> 00:19:55.640 align:middle line:84% is the recipient's willingness to pay for whatever you're 00:19:55.640 --> 00:19:59.060 align:middle line:84% providing relative to the net cost of the government, where 00:19:59.060 --> 00:20:00.830 align:middle line:84% the net cost takes into account effects 00:20:00.830 --> 00:20:01.770 align:middle line:90% on future government revenue? 00:20:01.770 --> 00:20:04.400 align:middle line:84% So some of this stuff actually looks free, has infinite MBPFs 00:20:04.400 --> 00:20:06.680 align:middle line:84% if basically the net cost of the government 00:20:06.680 --> 00:20:10.250 align:middle line:84% is zero because it increases people's future earnings so 00:20:10.250 --> 00:20:12.287 align:middle line:84% much they pay more taxes and so on and so forth. 00:20:12.287 --> 00:20:14.870 align:middle line:84% In principle, those approaches allow you to compare everything 00:20:14.870 --> 00:20:16.290 align:middle line:90% to everything else. 00:20:16.290 --> 00:20:18.450 align:middle line:84% The problem is you have to actually estimate-- 00:20:18.450 --> 00:20:19.590 align:middle line:90% you have to do two things. 00:20:19.590 --> 00:20:24.050 align:middle line:84% Number one, you have to estimate what are the gains, 00:20:24.050 --> 00:20:27.743 align:middle line:84% or what is the willingness to pay in some real sense? 00:20:27.743 --> 00:20:28.910 align:middle line:90% And what are the real costs? 00:20:28.910 --> 00:20:31.888 align:middle line:84% That's hard to do for some of these programs, 00:20:31.888 --> 00:20:33.930 align:middle line:84% particularly some of the infrastructure programs. 00:20:33.930 --> 00:20:37.900 align:middle line:84% The other point is you also need to augment that approach 00:20:37.900 --> 00:20:42.200 align:middle line:84% to make sure it takes into account redistribution effects. 00:20:42.200 --> 00:20:46.307 align:middle line:84% So the baseline MBPF considers a dollar 00:20:46.307 --> 00:20:47.390 align:middle line:90% is a dollar for everybody. 00:20:47.390 --> 00:20:48.460 align:middle line:84% But if programs are fundamentally 00:20:48.460 --> 00:20:50.668 align:middle line:84% about redistribution, you've got to take into account 00:20:50.668 --> 00:20:54.097 align:middle line:84% the fact that giving a dollar for a poor person 00:20:54.097 --> 00:20:56.430 align:middle line:84% is more valuable than a dollar for a rich person because 00:20:56.430 --> 00:20:57.280 align:middle line:90% of concave utility. 00:20:57.280 --> 00:21:00.050 align:middle line:84% And you've got to include that in the welfare calculation 00:21:00.050 --> 00:21:00.550 align:middle line:90% somehow. 00:21:00.550 --> 00:21:03.340 align:middle line:84% So the long and short of it is yes. 00:21:03.340 --> 00:21:04.060 align:middle line:90% You can do it. 00:21:04.060 --> 00:21:06.550 align:middle line:84% What you need to do is a careful cost benefit analysis 00:21:06.550 --> 00:21:09.310 align:middle line:84% or a careful market value of public funds. 00:21:09.310 --> 00:21:12.540 align:middle line:84% There's a bunch of inputs you need to do that. 00:21:12.540 --> 00:21:15.360 align:middle line:84% But you want to think of what is the net social return 00:21:15.360 --> 00:21:18.300 align:middle line:84% relative to the net social cost of the government taking-- 00:21:18.300 --> 00:21:21.105 align:middle line:84% weighted by some redistribution metric, roughly speaking. 00:21:21.105 --> 00:21:24.566 align:middle line:90% 00:21:24.566 --> 00:21:26.980 align:middle line:90% OK, other questions? 00:21:26.980 --> 00:21:31.210 align:middle line:84% And there's a [INAUDIBLE] and [INAUDIBLE] 00:21:31.210 --> 00:21:33.340 align:middle line:84% paper on some of the marginal value of public funds 00:21:33.340 --> 00:21:34.820 align:middle line:84% in the [? KJE ?] I think last year, 00:21:34.820 --> 00:21:37.030 align:middle line:84% which goes through some of this in a fair amount of detail 00:21:37.030 --> 00:21:38.110 align:middle line:90% of how do you think about that. 00:21:38.110 --> 00:21:40.660 align:middle line:84% I don't think it talks about the redistribution component a bit 00:21:40.660 --> 00:21:41.110 align:middle line:90% as much. 00:21:41.110 --> 00:21:42.985 align:middle line:84% But I think it's probably in there somewhere. 00:21:42.985 --> 00:21:46.448 align:middle line:84% And also, when our review paper is out, 00:21:46.448 --> 00:21:47.740 align:middle line:90% it'll all talk about that, too. 00:21:47.740 --> 00:21:50.993 align:middle line:90% 00:21:50.993 --> 00:21:51.660 align:middle line:90% Other questions? 00:21:51.660 --> 00:21:52.160 align:middle line:90% Yeah? 00:21:52.160 --> 00:21:54.460 align:middle line:84% AUDIENCE: What do MBPFs look like for these programs 00:21:54.460 --> 00:21:55.480 align:middle line:90% in developing countries? 00:21:55.480 --> 00:21:56.980 align:middle line:84% BEN OLKEN: I don't think it's been very well calculated 00:21:56.980 --> 00:21:58.022 align:middle line:90% for most of the programs. 00:21:58.022 --> 00:22:00.720 align:middle line:90% 00:22:00.720 --> 00:22:03.780 align:middle line:90% I think it's a great question. 00:22:03.780 --> 00:22:05.280 align:middle line:84% I don't think people have calculated 00:22:05.280 --> 00:22:07.180 align:middle line:90% for a lot of these programs. 00:22:07.180 --> 00:22:09.070 align:middle line:84% The other thing I'll also say is that the-- 00:22:09.070 --> 00:22:15.880 align:middle line:90% 00:22:15.880 --> 00:22:18.940 align:middle line:84% yeah, let me just leave it at that. 00:22:18.940 --> 00:22:23.170 align:middle line:84% But there's a lot of interest in some-- 00:22:23.170 --> 00:22:26.080 align:middle line:84% so there's starting to be some long-run evaluations of some 00:22:26.080 --> 00:22:29.230 align:middle line:84% of these programs, which are starting to say what happens 00:22:29.230 --> 00:22:31.778 align:middle line:84% to these programs to people 10, 20 years later to see, 00:22:31.778 --> 00:22:33.820 align:middle line:84% is it actually-- are they entering the workforce? 00:22:33.820 --> 00:22:36.670 align:middle line:84% Can we actually-- because one of the questions 00:22:36.670 --> 00:22:42.010 align:middle line:84% is, what is the long-run impact on people's ability 00:22:42.010 --> 00:22:44.260 align:middle line:84% to work, and ultimately pay taxes, and all that stuff? 00:22:44.260 --> 00:22:45.160 align:middle line:90% And are they in the tax net? 00:22:45.160 --> 00:22:47.338 align:middle line:84% I don't think that stuff has actually been fully 00:22:47.338 --> 00:22:48.380 align:middle line:90% thought through, frankly. 00:22:48.380 --> 00:22:50.880 align:middle line:84% And I think there's a lot of interesting issues in doing so. 00:22:50.880 --> 00:22:53.630 align:middle line:90% 00:22:53.630 --> 00:22:56.477 align:middle line:84% Often, for example, when I've seen some analyses like this, 00:22:56.477 --> 00:22:59.060 align:middle line:84% they'll say, well, we're going to look at some interim impact. 00:22:59.060 --> 00:23:00.477 align:middle line:84% They'll maybe do it in two stages. 00:23:00.477 --> 00:23:02.165 align:middle line:84% They'll say, well, there's an impact 00:23:02.165 --> 00:23:03.290 align:middle line:90% of this in going to school. 00:23:03.290 --> 00:23:05.390 align:middle line:84% And then we'll take the schooling regression 00:23:05.390 --> 00:23:06.480 align:middle line:90% and apply it to that. 00:23:06.480 --> 00:23:08.720 align:middle line:84% And we'll back something out of that. 00:23:08.720 --> 00:23:11.220 align:middle line:84% So that's some of the approaches people take. 00:23:11.220 --> 00:23:13.700 align:middle line:84% But I think that thinking about some of the stuff 00:23:13.700 --> 00:23:15.050 align:middle line:90% seriously would be interesting. 00:23:15.050 --> 00:23:16.800 align:middle line:84% And there'd be more work to be done there. 00:23:16.800 --> 00:23:19.030 align:middle line:90% 00:23:19.030 --> 00:23:21.580 align:middle line:90% Other questions? 00:23:21.580 --> 00:23:22.110 align:middle line:90% OK. 00:23:22.110 --> 00:23:24.010 align:middle line:84% So the flip side-- the next part of it is, well, 00:23:24.010 --> 00:23:25.885 align:middle line:84% how do we raise money to pay for all of this? 00:23:25.885 --> 00:23:28.560 align:middle line:84% And how do we think about the challenges of the public sector 00:23:28.560 --> 00:23:30.840 align:middle line:84% in a developing country context, OK? 00:23:30.840 --> 00:23:37.355 align:middle line:84% And there are, broadly speaking-- 00:23:37.355 --> 00:23:38.730 align:middle line:84% so what am I going to talk about? 00:23:38.730 --> 00:23:41.690 align:middle line:84% I'm going to talk about a bunch of things. 00:23:41.690 --> 00:23:44.260 align:middle line:84% First, I'm going to talk about what's 00:23:44.260 --> 00:23:46.990 align:middle line:84% different about the public, the tax 00:23:46.990 --> 00:23:49.207 align:middle line:84% side of things in a developing country context? 00:23:49.207 --> 00:23:51.290 align:middle line:84% And I'll talk about a couple of different theories 00:23:51.290 --> 00:23:52.510 align:middle line:90% for what's going on there. 00:23:52.510 --> 00:23:55.590 align:middle line:90% 00:23:55.590 --> 00:23:58.560 align:middle line:84% I'll then talk about, I think, two different ways 00:23:58.560 --> 00:24:01.050 align:middle line:84% of approaching the problem of tax enforcement. 00:24:01.050 --> 00:24:03.510 align:middle line:84% So one of them is related to the paper you guys read 00:24:03.510 --> 00:24:04.613 align:middle line:90% for today about the VAT. 00:24:04.613 --> 00:24:06.030 align:middle line:84% Essentially, you can think of that 00:24:06.030 --> 00:24:08.070 align:middle line:90% as through the tax structure. 00:24:08.070 --> 00:24:10.590 align:middle line:84% Are certain kinds of taxes or tax structures 00:24:10.590 --> 00:24:14.100 align:middle line:84% is going to work better in these low information environments? 00:24:14.100 --> 00:24:15.780 align:middle line:84% And the second is tax administration. 00:24:15.780 --> 00:24:16.950 align:middle line:84% Are there certain things we need to do about the way 00:24:16.950 --> 00:24:18.750 align:middle line:84% we collect taxes that are going to be 00:24:18.750 --> 00:24:22.020 align:middle line:84% different in these developing country contexts? 00:24:22.020 --> 00:24:23.550 align:middle line:84% And ultimately, does this matter? 00:24:23.550 --> 00:24:25.920 align:middle line:84% And then I also want to talk briefly, if I have time, 00:24:25.920 --> 00:24:28.770 align:middle line:84% about something called informal taxation, which 00:24:28.770 --> 00:24:33.000 align:middle line:84% is, I think, the analog to informal insurance on the tax 00:24:33.000 --> 00:24:33.990 align:middle line:90% side. 00:24:33.990 --> 00:24:36.730 align:middle line:90% So that's the outline. 00:24:36.730 --> 00:24:37.230 align:middle line:90% All right. 00:24:37.230 --> 00:24:39.750 align:middle line:84% So just stepping way back, if you take a public finance 00:24:39.750 --> 00:24:43.920 align:middle line:84% class, there's a vast literature on taxation issues, 00:24:43.920 --> 00:24:46.350 align:middle line:84% incidence, optimal income tax theory, capital taxation, 00:24:46.350 --> 00:24:48.100 align:middle line:84% consumption taxes, dynamic considerations, 00:24:48.100 --> 00:24:49.470 align:middle line:90% and so on and so forth. 00:24:49.470 --> 00:24:52.840 align:middle line:84% You take an entire semester on this stuff. 00:24:52.840 --> 00:24:54.960 align:middle line:84% I think by comparison, understanding 00:24:54.960 --> 00:24:58.530 align:middle line:84% how do we think about this stuff in a developing country context 00:24:58.530 --> 00:25:00.390 align:middle line:90% is we know relatively little. 00:25:00.390 --> 00:25:03.000 align:middle line:84% We a lot more than we did maybe even 10 years ago when 00:25:03.000 --> 00:25:05.320 align:middle line:84% I first started teaching this class, which is exciting. 00:25:05.320 --> 00:25:07.070 align:middle line:84% There's actually been a really active area 00:25:07.070 --> 00:25:09.630 align:middle line:84% of research, which is part of the reason I want to mention it 00:25:09.630 --> 00:25:11.880 align:middle line:84% for people who are interested in development economics 00:25:11.880 --> 00:25:12.428 align:middle line:90% questions. 00:25:12.428 --> 00:25:14.220 align:middle line:84% And the other thing is I think what we know 00:25:14.220 --> 00:25:18.000 align:middle line:84% suggests that there's something that is very much 00:25:18.000 --> 00:25:21.180 align:middle line:84% a development economics issue, thinking about something 00:25:21.180 --> 00:25:22.910 align:middle line:90% in the tax structure. 00:25:22.910 --> 00:25:23.910 align:middle line:90% And I think there's two. 00:25:23.910 --> 00:25:26.940 align:middle line:84% The first is-- and these are the same themes 00:25:26.940 --> 00:25:28.398 align:middle line:84% that I think came up in the stuff I 00:25:28.398 --> 00:25:29.565 align:middle line:90% talked about redistribution. 00:25:29.565 --> 00:25:31.500 align:middle line:84% So one issue is an information problem, right? 00:25:31.500 --> 00:25:33.375 align:middle line:84% In the same way we had information challenges 00:25:33.375 --> 00:25:35.490 align:middle line:84% of identifying who's poor, we also 00:25:35.490 --> 00:25:38.003 align:middle line:84% have information problems about figuring out how to tax 00:25:38.003 --> 00:25:39.420 align:middle line:84% and how much to tax people, right? 00:25:39.420 --> 00:25:42.880 align:middle line:84% So how do you levy an income tax on subsistence farmers, 00:25:42.880 --> 00:25:44.220 align:middle line:90% for example? 00:25:44.220 --> 00:25:48.520 align:middle line:90% Or even do you want to do that? 00:25:48.520 --> 00:25:52.500 align:middle line:84% How do you tax labor in a place where 00:25:52.500 --> 00:25:57.950 align:middle line:84% you have 50% of the economy is in the informal sector? 00:25:57.950 --> 00:26:02.330 align:middle line:84% The government just has much less information 00:26:02.330 --> 00:26:03.360 align:middle line:90% about a lot of people. 00:26:03.360 --> 00:26:05.755 align:middle line:84% And so how do you deal with that? 00:26:05.755 --> 00:26:08.130 align:middle line:84% And the second is, I think, on enforcement or governance, 00:26:08.130 --> 00:26:11.105 align:middle line:84% which is I think also related to the information problems, 00:26:11.105 --> 00:26:13.730 align:middle line:84% there's a lot of opportunity for corruption and other problems. 00:26:13.730 --> 00:26:17.180 align:middle line:84% And how do we think about the enforcement side? 00:26:17.180 --> 00:26:21.290 align:middle line:84% And I think that suggests that just the way we 00:26:21.290 --> 00:26:26.120 align:middle line:84% may want to think about taxation in a developing country context 00:26:26.120 --> 00:26:29.180 align:middle line:84% may not just be, we should think about 00:26:29.180 --> 00:26:32.480 align:middle line:90% what is the optimal Ramsey tax? 00:26:32.480 --> 00:26:34.130 align:middle line:90% And how do we set our taxes-- 00:26:34.130 --> 00:26:36.945 align:middle line:84% I think some of those issues are perhaps important. 00:26:36.945 --> 00:26:39.320 align:middle line:84% But some of these other issues may be really first order. 00:26:39.320 --> 00:26:43.020 align:middle line:84% And that's what I want to talk about today. 00:26:43.020 --> 00:26:44.440 align:middle line:90% OK. 00:26:44.440 --> 00:26:44.955 align:middle line:90% Yeah? 00:26:44.955 --> 00:26:46.330 align:middle line:84% AUDIENCE: I just have a question. 00:26:46.330 --> 00:26:48.520 align:middle line:84% How much of these issues are limited 00:26:48.520 --> 00:26:52.030 align:middle line:84% to the public sector and the [INAUDIBLE] most of them 00:26:52.030 --> 00:26:55.354 align:middle line:90% [INAUDIBLE] taxes? 00:26:55.354 --> 00:26:58.086 align:middle line:84% The same issues aren't true when these private companies 00:26:58.086 --> 00:27:02.450 align:middle line:84% [INAUDIBLE] figure out which customer is good, 00:27:02.450 --> 00:27:03.430 align:middle line:90% or something like that. 00:27:03.430 --> 00:27:03.920 align:middle line:90% Does that make sense? 00:27:03.920 --> 00:27:04.610 align:middle line:90% BEN OLKEN: Yeah. 00:27:04.610 --> 00:27:06.720 align:middle line:84% So I think it's a-- yeah, it's a good question. 00:27:06.720 --> 00:27:15.520 align:middle line:84% So let me hold that thought for a minute. 00:27:15.520 --> 00:27:18.610 align:middle line:84% I actually think that there are particular issues 00:27:18.610 --> 00:27:22.700 align:middle line:90% vis-a-vis tax. 00:27:22.700 --> 00:27:26.960 align:middle line:84% I think they are a bit different. 00:27:26.960 --> 00:27:28.850 align:middle line:84% So let me give you some examples. 00:27:28.850 --> 00:27:30.435 align:middle line:84% And then you can see what you think. 00:27:30.435 --> 00:27:32.060 align:middle line:84% AUDIENCE: OK, I just have [INAUDIBLE].. 00:27:32.060 --> 00:27:34.435 align:middle line:84% BEN OLKEN: So particular one of the big themes, actually, 00:27:34.435 --> 00:27:38.120 align:middle line:84% is about third party reporting, which is about formality. 00:27:38.120 --> 00:27:40.578 align:middle line:84% And that has a very specific tax structure. 00:27:40.578 --> 00:27:41.120 align:middle line:90% AUDIENCE: OK. 00:27:41.120 --> 00:27:42.110 align:middle line:84% BEN OLKEN: Is that what you're saying? 00:27:42.110 --> 00:27:42.735 align:middle line:90% AUDIENCE: Yeah. 00:27:42.735 --> 00:27:44.535 align:middle line:90% I guess what I'm [INAUDIBLE]. 00:27:44.535 --> 00:27:45.910 align:middle line:84% Are there any particular concerns 00:27:45.910 --> 00:27:50.122 align:middle line:84% about space or moving some things that [INAUDIBLE]?? 00:27:50.122 --> 00:27:51.830 align:middle line:84% BEN OLKEN: In general, it's the opposite. 00:27:51.830 --> 00:27:53.872 align:middle line:84% In general, governments have a lot better options 00:27:53.872 --> 00:27:55.250 align:middle line:90% than private firms, right? 00:27:55.250 --> 00:27:58.820 align:middle line:84% If you think about what can the IRS observe about you, 00:27:58.820 --> 00:28:01.400 align:middle line:84% they know a lot more than amazon.com. 00:28:01.400 --> 00:28:03.890 align:middle line:84% Maybe they don't, actually, in this day and age. 00:28:03.890 --> 00:28:06.350 align:middle line:84% But in principle, they can observe all kinds of things 00:28:06.350 --> 00:28:09.650 align:middle line:84% that the private sector can't because they 00:28:09.650 --> 00:28:11.660 align:middle line:84% can mandate that people report stuff to them. 00:28:11.660 --> 00:28:12.160 align:middle line:90% Right? 00:28:12.160 --> 00:28:15.080 align:middle line:84% So that mandatory reporting of information 00:28:15.080 --> 00:28:16.600 align:middle line:90% means that they can-- 00:28:16.600 --> 00:28:17.890 align:middle line:90% it's interesting. 00:28:17.890 --> 00:28:19.400 align:middle line:84% I just said, in the new tech world, 00:28:19.400 --> 00:28:21.650 align:middle line:84% they can figure out ways of backing a lot of the stuff 00:28:21.650 --> 00:28:23.240 align:middle line:84% out by what you click on or whatever. 00:28:23.240 --> 00:28:25.860 align:middle line:84% But the government has a fundamental advantage, 00:28:25.860 --> 00:28:27.920 align:middle line:84% which is it can make you tell them stuff. 00:28:27.920 --> 00:28:30.830 align:middle line:84% And they can throw you in jail if you don't or if you 00:28:30.830 --> 00:28:31.640 align:middle line:90% lie about it. 00:28:31.640 --> 00:28:33.890 align:middle line:84% So that is a huge advantage the government 00:28:33.890 --> 00:28:37.730 align:middle line:84% has over the private sector in getting information. 00:28:37.730 --> 00:28:38.420 align:middle line:90% Right. 00:28:38.420 --> 00:28:40.760 align:middle line:90% OK. 00:28:40.760 --> 00:28:42.710 align:middle line:90% It can make you tell them stuff. 00:28:42.710 --> 00:28:45.260 align:middle line:84% It can make banks tell them stuff. 00:28:45.260 --> 00:28:48.350 align:middle line:84% Again, how that's actually a policy question 00:28:48.350 --> 00:28:50.510 align:middle line:84% and how much of that additional reporting we want-- 00:28:50.510 --> 00:28:53.570 align:middle line:84% and different countries make different choices 00:28:53.570 --> 00:28:57.380 align:middle line:84% about how much information they want the tax authority to have. 00:28:57.380 --> 00:28:59.750 align:middle line:84% And actually, if you've been reading the news in the US, 00:28:59.750 --> 00:29:02.840 align:middle line:84% for example, this question about should the IRS 00:29:02.840 --> 00:29:05.210 align:middle line:84% be able to access information about your bank balances 00:29:05.210 --> 00:29:06.440 align:middle line:84% and under what circumstances is actually 00:29:06.440 --> 00:29:08.030 align:middle line:84% an active question of policy debate, 00:29:08.030 --> 00:29:10.370 align:middle line:84% where people seem to actually disagree about this question, 00:29:10.370 --> 00:29:10.870 align:middle line:90% apparently. 00:29:10.870 --> 00:29:12.260 align:middle line:90% So OK. 00:29:12.260 --> 00:29:18.110 align:middle line:84% So anyway, so I think it is different. 00:29:18.110 --> 00:29:20.990 align:middle line:90% OK. 00:29:20.990 --> 00:29:22.720 align:middle line:90% So just a few overview facts. 00:29:22.720 --> 00:29:27.250 align:middle line:84% So this paper by Gordon and Li basically 00:29:27.250 --> 00:29:30.468 align:middle line:84% made the point that just overall-- the paper 00:29:30.468 --> 00:29:31.510 align:middle line:90% is a bit out of date now. 00:29:31.510 --> 00:29:34.873 align:middle line:84% But overall, the tax structure looks 00:29:34.873 --> 00:29:36.790 align:middle line:84% very different in developing countries than it 00:29:36.790 --> 00:29:37.730 align:middle line:90% looks in developed countries. 00:29:37.730 --> 00:29:39.190 align:middle line:84% So there's just a few facts here. 00:29:39.190 --> 00:29:40.773 align:middle line:84% Actually, let me pass out the handout. 00:29:40.773 --> 00:29:57.900 align:middle line:90% 00:29:57.900 --> 00:29:59.580 align:middle line:90% Oh, I want one. 00:29:59.580 --> 00:30:02.090 align:middle line:90% Thanks. 00:30:02.090 --> 00:30:02.870 align:middle line:90% OK. 00:30:02.870 --> 00:30:04.790 align:middle line:84% So this is too small to read on the screen. 00:30:04.790 --> 00:30:06.510 align:middle line:84% Maybe I can zoom in a little bit. 00:30:06.510 --> 00:30:10.500 align:middle line:90% 00:30:10.500 --> 00:30:13.400 align:middle line:90% OK. 00:30:13.400 --> 00:30:14.960 align:middle line:90% So what's the point? 00:30:14.960 --> 00:30:16.490 align:middle line:90% Several just facts. 00:30:16.490 --> 00:30:19.070 align:middle line:84% First is rich countries have a lot more tax 00:30:19.070 --> 00:30:21.370 align:middle line:84% revenue as a share of GDP than poor countries. 00:30:21.370 --> 00:30:21.870 align:middle line:90% OK? 00:30:21.870 --> 00:30:24.440 align:middle line:84% So in general, the size of the public sector 00:30:24.440 --> 00:30:26.540 align:middle line:90% looks really different. 00:30:26.540 --> 00:30:29.840 align:middle line:84% Number two, where they're getting their tax 00:30:29.840 --> 00:30:31.690 align:middle line:90% looks really different. 00:30:31.690 --> 00:30:35.520 align:middle line:84% So income taxes in general are a smaller share 00:30:35.520 --> 00:30:40.120 align:middle line:84% of revenue in rich countries than in poor countries. 00:30:40.120 --> 00:30:44.230 align:middle line:84% Corporate income tax is a much larger share 00:30:44.230 --> 00:30:46.505 align:middle line:84% of income taxes in poor countries 00:30:46.505 --> 00:30:47.630 align:middle line:90% relative to rich countries. 00:30:47.630 --> 00:30:50.715 align:middle line:84% So in the very poorest countries, half 00:30:50.715 --> 00:30:52.840 align:middle line:84% their income tax comes from corporate income taxes. 00:30:52.840 --> 00:30:55.060 align:middle line:84% In rich countries, it's a sixth or something. 00:30:55.060 --> 00:30:59.568 align:middle line:84% So rich countries are better at taxing individuals 00:30:59.568 --> 00:31:00.610 align:middle line:90% relative to corporations. 00:31:00.610 --> 00:31:03.280 align:middle line:90% Poor countries tax corporations. 00:31:03.280 --> 00:31:07.523 align:middle line:84% At least in this period, tariffs are a non-trivial share 00:31:07.523 --> 00:31:08.440 align:middle line:90% of government revenue. 00:31:08.440 --> 00:31:10.815 align:middle line:84% This probably was true in the historical US, for example, 00:31:10.815 --> 00:31:14.290 align:middle line:84% also, that tariffs used to be a much bigger deal than they are. 00:31:14.290 --> 00:31:16.240 align:middle line:90% So tariffs are non-trivial. 00:31:16.240 --> 00:31:18.730 align:middle line:84% They also talk about seigniorage as a function of revenue. 00:31:18.730 --> 00:31:22.180 align:middle line:84% Again, inflation has gone down up until the last six months. 00:31:22.180 --> 00:31:25.430 align:middle line:84% Inflation had gone down a lot, I think, around the world. 00:31:25.430 --> 00:31:27.310 align:middle line:84% So this may be a little bit out of date. 00:31:27.310 --> 00:31:29.290 align:middle line:84% But inflation was more of a big deal. 00:31:29.290 --> 00:31:32.110 align:middle line:84% And seigniorage income, which is the revenue the government gets 00:31:32.110 --> 00:31:34.623 align:middle line:84% for printing money, was a non-trivial share 00:31:34.623 --> 00:31:36.790 align:middle line:84% of the government budget in some of these countries. 00:31:36.790 --> 00:31:38.650 align:middle line:84% And the other fact is the estimates 00:31:38.650 --> 00:31:41.450 align:middle line:84% the informal sector as a share of GDP are much larger. 00:31:41.450 --> 00:31:44.440 align:middle line:84% So it looks like these countries are doing different stuff. 00:31:44.440 --> 00:31:47.020 align:middle line:90% And a question is why. 00:31:47.020 --> 00:31:53.510 align:middle line:84% And so here so there are two theories that can-- here 00:31:53.510 --> 00:31:55.280 align:middle line:84% are two theories that can explain why 00:31:55.280 --> 00:31:57.240 align:middle line:90% things look really different. 00:31:57.240 --> 00:32:00.050 align:middle line:84% So one is this theory by Gordon and Li. 00:32:00.050 --> 00:32:04.700 align:middle line:84% And the other is a theory by Kleven, Kreiner, and Saez. 00:32:04.700 --> 00:32:09.110 align:middle line:84% So this one is about The financial sector generating 00:32:09.110 --> 00:32:11.970 align:middle line:84% information, which I'll talk about in a second. 00:32:11.970 --> 00:32:14.540 align:middle line:84% And the other one is about third party reporting generating 00:32:14.540 --> 00:32:15.080 align:middle line:90% information. 00:32:15.080 --> 00:32:18.715 align:middle line:84% And underlying both of them is this question of the government 00:32:18.715 --> 00:32:20.090 align:middle line:84% needs the information to tax you. 00:32:20.090 --> 00:32:21.470 align:middle line:90% Otherwise, you evade. 00:32:21.470 --> 00:32:23.030 align:middle line:84% And where do you get the information? 00:32:23.030 --> 00:32:25.070 align:middle line:84% And why might that look different in these two 00:32:25.070 --> 00:32:26.280 align:middle line:90% different contexts? 00:32:26.280 --> 00:32:28.998 align:middle line:84% So I'm going to talk about this one very quickly. 00:32:28.998 --> 00:32:31.040 align:middle line:84% And this one I'll go through a super simple model 00:32:31.040 --> 00:32:31.790 align:middle line:90% to make the point. 00:32:31.790 --> 00:32:33.030 align:middle line:90% OK. 00:32:33.030 --> 00:32:35.783 align:middle line:84% So the Gordon and Li paper argues 00:32:35.783 --> 00:32:37.700 align:middle line:84% that the financial sector is really important. 00:32:37.700 --> 00:32:44.190 align:middle line:84% And the basic idea is that if you use the banks, 00:32:44.190 --> 00:32:46.400 align:middle line:84% the banks observe how much money is 00:32:46.400 --> 00:32:47.660 align:middle line:90% flowing through your accounts. 00:32:47.660 --> 00:32:50.675 align:middle line:84% Because that's what banks do, right? 00:32:50.675 --> 00:32:52.800 align:middle line:84% They have to keep track of how much money you have. 00:32:52.800 --> 00:32:56.250 align:middle line:84% So that creates a record of what you're doing. 00:32:56.250 --> 00:33:00.830 align:middle line:84% And the government can tap into that record 00:33:00.830 --> 00:33:04.270 align:middle line:84% and use that as information to tax you. 00:33:04.270 --> 00:33:05.220 align:middle line:90% OK? 00:33:05.220 --> 00:33:08.730 align:middle line:84% And they just argue, basically, that essentially, 00:33:08.730 --> 00:33:15.660 align:middle line:84% larger organizations need to use the financial sector. 00:33:15.660 --> 00:33:18.927 align:middle line:90% 00:33:18.927 --> 00:33:20.760 align:middle line:84% So sorry, you can do it in a couple of ways. 00:33:20.760 --> 00:33:22.760 align:middle line:84% Number one, there's more financial development 00:33:22.760 --> 00:33:24.350 align:middle line:84% for other reasons in rich countries, 00:33:24.350 --> 00:33:27.410 align:middle line:84% just like people are using banks for other reasons. 00:33:27.410 --> 00:33:31.220 align:middle line:84% And number two, the corporate sector really-- you 00:33:31.220 --> 00:33:34.340 align:middle line:84% can run a local pizza shop, maybe, in cash. 00:33:34.340 --> 00:33:37.040 align:middle line:84% But you can't run General Electric in cash 00:33:37.040 --> 00:33:38.960 align:middle line:84% because it's just too complicated. 00:33:38.960 --> 00:33:40.940 align:middle line:84% And you need to actually use banks. 00:33:40.940 --> 00:33:43.250 align:middle line:84% And there's another fact, which we'll 00:33:43.250 --> 00:33:45.890 align:middle line:90% talk about in 14742 that-- 00:33:45.890 --> 00:33:48.680 align:middle line:84% firms are just a lot bigger in developed countries 00:33:48.680 --> 00:33:51.410 align:middle line:90% for other sets of reasons. 00:33:51.410 --> 00:33:54.710 align:middle line:84% So little tiny mom and pop shops that maybe potentially 00:33:54.710 --> 00:33:57.860 align:middle line:84% could run in cash are a much larger share 00:33:57.860 --> 00:34:02.990 align:middle line:84% of what's happening in the corporate world in India 00:34:02.990 --> 00:34:06.960 align:middle line:84% than they are in the US, for example. 00:34:06.960 --> 00:34:11.100 align:middle line:84% And so that also gives you-- they 00:34:11.100 --> 00:34:15.179 align:middle line:84% argue that basic fact of the financial sector generating 00:34:15.179 --> 00:34:17.281 align:middle line:84% information generates a lot more of the pattern, 00:34:17.281 --> 00:34:18.239 align:middle line:90% some of these patterns. 00:34:18.239 --> 00:34:20.040 align:middle line:84% For example, they're going to say 00:34:20.040 --> 00:34:23.100 align:middle line:84% that's one of the reasons why developing countries are 00:34:23.100 --> 00:34:23.719 align:middle line:90% going to tax-- 00:34:23.719 --> 00:34:25.469 align:middle line:84% spend a lot of effort taxing corporations. 00:34:25.469 --> 00:34:27.386 align:middle line:84% Because at least they can see the corporations 00:34:27.386 --> 00:34:31.620 align:middle line:84% and see what they're doing if large corporations are 00:34:31.620 --> 00:34:34.620 align:middle line:84% inelastic in their use of the formal sector. 00:34:34.620 --> 00:34:36.179 align:middle line:84% And then they argue everything else 00:34:36.179 --> 00:34:40.360 align:middle line:84% has to do that, which is now-- so for example, 00:34:40.360 --> 00:34:42.750 align:middle line:84% we have this sector that's now being taxed a lot, which 00:34:42.750 --> 00:34:43.857 align:middle line:90% is the informal sector. 00:34:43.857 --> 00:34:45.690 align:middle line:84% So they argue maybe you have to have tariffs 00:34:45.690 --> 00:34:48.148 align:middle line:84% to protect them and make sure everyone's on a level playing 00:34:48.148 --> 00:34:48.969 align:middle line:90% field. 00:34:48.969 --> 00:34:51.420 align:middle line:84% And then you use inflation, they argue, maybe 00:34:51.420 --> 00:34:55.139 align:middle line:84% as an optional tax on the cash economy. 00:34:55.139 --> 00:34:58.920 align:middle line:84% So basically, if putting your money in the bank 00:34:58.920 --> 00:35:01.050 align:middle line:84% means it gets taxed, we also want 00:35:01.050 --> 00:35:03.490 align:middle line:84% to have an equivalent tax on your cash holdings 00:35:03.490 --> 00:35:05.430 align:middle line:84% since they do that through inflation. 00:35:05.430 --> 00:35:09.900 align:middle line:84% This maybe is a little too neat maybe. 00:35:09.900 --> 00:35:13.417 align:middle line:84% It's not totally obvious to me that in fact, actually, what's 00:35:13.417 --> 00:35:15.000 align:middle line:84% really happening in these countries is 00:35:15.000 --> 00:35:17.040 align:middle line:84% the central banker is saying actually, 00:35:17.040 --> 00:35:21.150 align:middle line:84% the optimal Taylor rule level of inflation in our economy 00:35:21.150 --> 00:35:23.790 align:middle line:84% is 10% because we want to correct 00:35:23.790 --> 00:35:27.990 align:middle line:84% this tax wedge between-- so this story is, I think, consistent. 00:35:27.990 --> 00:35:31.890 align:middle line:84% I'm not 100% sure it's how people are actually thinking. 00:35:31.890 --> 00:35:34.000 align:middle line:90% But that's the story. 00:35:34.000 --> 00:35:35.555 align:middle line:84% So their argument is it all flows 00:35:35.555 --> 00:35:36.680 align:middle line:90% from this financial sector. 00:35:36.680 --> 00:35:39.360 align:middle line:84% And everything else in the tax structure flows from that. 00:35:39.360 --> 00:35:40.720 align:middle line:84% I'm not going to go through that in a lot of detail. 00:35:40.720 --> 00:35:43.095 align:middle line:84% But I wanted to mention that as one argument. 00:35:43.095 --> 00:35:48.740 align:middle line:90% 00:35:48.740 --> 00:35:51.290 align:middle line:84% The second argument is this argument 00:35:51.290 --> 00:35:54.020 align:middle line:84% that Kleven, Kreiner and Saez have put forward. 00:35:54.020 --> 00:35:56.493 align:middle line:84% And this is related to a series of papers 00:35:56.493 --> 00:35:58.910 align:middle line:84% that they and others have done on this idea of third party 00:35:58.910 --> 00:35:59.630 align:middle line:90% reporting. 00:35:59.630 --> 00:36:01.550 align:middle line:84% And this is consistent with the VAT paper 00:36:01.550 --> 00:36:03.050 align:middle line:90% that you guys read for today. 00:36:03.050 --> 00:36:07.460 align:middle line:84% So the basic idea of a lot of taxation 00:36:07.460 --> 00:36:10.250 align:middle line:84% is how does the government know stuff? 00:36:10.250 --> 00:36:12.860 align:middle line:84% Well, it gets information from two different parties 00:36:12.860 --> 00:36:13.460 align:middle line:90% and compares. 00:36:13.460 --> 00:36:15.980 align:middle line:84% And if you get different answers, 00:36:15.980 --> 00:36:17.420 align:middle line:84% there's something funky going on. 00:36:17.420 --> 00:36:22.460 align:middle line:84% And so what's called third party reporting 00:36:22.460 --> 00:36:25.442 align:middle line:84% of the same information-- so the government is the first party. 00:36:25.442 --> 00:36:26.150 align:middle line:90% I'm second party. 00:36:26.150 --> 00:36:27.275 align:middle line:90% I report to the government. 00:36:27.275 --> 00:36:29.390 align:middle line:84% Having someone else report that same information 00:36:29.390 --> 00:36:31.990 align:middle line:90% makes it harder for me to lie. 00:36:31.990 --> 00:36:32.490 align:middle line:90% OK? 00:36:32.490 --> 00:36:37.500 align:middle line:84% So here's a simple model that can explain 00:36:37.500 --> 00:36:40.860 align:middle line:84% why is this relevant and why is this potentially a development 00:36:40.860 --> 00:36:42.430 align:middle line:90% phenomenon? 00:36:42.430 --> 00:36:46.350 align:middle line:84% So most taxes, they argue, are collected through third party 00:36:46.350 --> 00:36:47.100 align:middle line:90% firms-- 00:36:47.100 --> 00:36:50.310 align:middle line:84% third parties, such as employers, which double report. 00:36:50.310 --> 00:36:53.490 align:middle line:84% So for example, I work for MIT, right? 00:36:53.490 --> 00:36:56.220 align:middle line:84% And every year at the end of the year, 00:36:56.220 --> 00:36:59.160 align:middle line:84% I tell the IRS how much I think I earned. 00:36:59.160 --> 00:37:02.880 align:middle line:84% And MIT tells the IRS how much they paid me. 00:37:02.880 --> 00:37:04.440 align:middle line:84% And the IRS can just check if I'm 00:37:04.440 --> 00:37:06.240 align:middle line:84% reporting something different than what MIT reported. 00:37:06.240 --> 00:37:07.050 align:middle line:90% That's a red flag. 00:37:07.050 --> 00:37:07.800 align:middle line:90% And I get audited. 00:37:07.800 --> 00:37:10.403 align:middle line:84% Or actually, they send me a correction letter. 00:37:10.403 --> 00:37:12.570 align:middle line:84% They would send me a correction letter saying you've 00:37:12.570 --> 00:37:13.612 align:middle line:90% underreported your taxes. 00:37:13.612 --> 00:37:15.860 align:middle line:90% Please pay, OK? 00:37:15.860 --> 00:37:19.430 align:middle line:84% And there's some work even in modern economies. 00:37:19.430 --> 00:37:22.255 align:middle line:84% There's these same sets of co-authors, 00:37:22.255 --> 00:37:24.630 align:middle line:84% or some combination of them have a nice paper in Denmark, 00:37:24.630 --> 00:37:29.090 align:middle line:84% for example, showing that underreporting of income 00:37:29.090 --> 00:37:31.880 align:middle line:84% is very, very low when it's third party reported, 00:37:31.880 --> 00:37:34.100 align:middle line:84% and very high, even in Denmark, when 00:37:34.100 --> 00:37:36.300 align:middle line:90% it's not third party reported. 00:37:36.300 --> 00:37:37.150 align:middle line:90% OK? 00:37:37.150 --> 00:37:40.740 align:middle line:84% And so I think that's probably true in lots of places, right? 00:37:40.740 --> 00:37:45.645 align:middle line:90% So this may be for-- yeah. 00:37:45.645 --> 00:37:48.160 align:middle line:90% 00:37:48.160 --> 00:37:53.350 align:middle line:84% And the idea of this paper is to argue that this third party 00:37:53.350 --> 00:37:54.580 align:middle line:90% reporting-- 00:37:54.580 --> 00:37:58.960 align:middle line:84% it's easier to make this thing work in really large firms. 00:37:58.960 --> 00:38:01.660 align:middle line:84% And then they're going to rely on the fact I just told you, 00:38:01.660 --> 00:38:04.840 align:middle line:84% which is that firms are much larger in the developed world 00:38:04.840 --> 00:38:07.100 align:middle line:90% than the developing world. 00:38:07.100 --> 00:38:08.740 align:middle line:90% So what's the argument? 00:38:08.740 --> 00:38:12.765 align:middle line:84% So imagine there's a firm with n employees. 00:38:12.765 --> 00:38:16.830 align:middle line:84% It generates some surplus, big W, and some wages, w1, wN. 00:38:16.830 --> 00:38:22.810 align:middle line:84% And the total wage bill is just the sum of the wages. 00:38:22.810 --> 00:38:27.010 align:middle line:84% Third party reporting is that the firm and the employees 00:38:27.010 --> 00:38:29.800 align:middle line:90% report w bar to the government. 00:38:29.800 --> 00:38:31.765 align:middle line:84% And the government taxes you based on w bar. 00:38:31.765 --> 00:38:34.430 align:middle line:90% 00:38:34.430 --> 00:38:38.950 align:middle line:90% And imagine if w is not-- 00:38:38.950 --> 00:38:41.300 align:middle line:84% the firm has to keep track of what it's doing. 00:38:41.300 --> 00:38:43.600 align:middle line:90% So it has a set of books, right? 00:38:43.600 --> 00:38:47.560 align:middle line:84% And people inside the firm know what w is. 00:38:47.560 --> 00:38:48.730 align:middle line:90% And they know what w bar is. 00:38:48.730 --> 00:38:52.900 align:middle line:84% And if those are different, people understand that, right? 00:38:52.900 --> 00:38:56.650 align:middle line:84% So in particular, if w is less than w bar, 00:38:56.650 --> 00:39:00.280 align:middle line:84% any employee can go to the IRS and report 00:39:00.280 --> 00:39:02.368 align:middle line:90% that you're cheating. 00:39:02.368 --> 00:39:03.260 align:middle line:90% OK? 00:39:03.260 --> 00:39:05.150 align:middle line:84% And if someone announces you, the government 00:39:05.150 --> 00:39:08.980 align:middle line:84% applies a fine theta for all the evaded taxes plus 00:39:08.980 --> 00:39:09.980 align:middle line:90% collects the back taxes. 00:39:09.980 --> 00:39:12.313 align:middle line:84% So basically, if you get denounced, you pay 1 plus theta 00:39:12.313 --> 00:39:16.480 align:middle line:84% times the amount times w minus w bar, OK? 00:39:16.480 --> 00:39:21.940 align:middle line:84% So if everybody can agree not to denounce you, 00:39:21.940 --> 00:39:25.008 align:middle line:84% then we all just report w bar equals 0 to the government. 00:39:25.008 --> 00:39:26.050 align:middle line:90% And there's some surplus. 00:39:26.050 --> 00:39:28.210 align:middle line:90% And we split the surplus, OK? 00:39:28.210 --> 00:39:30.930 align:middle line:84% The problem is that a single employee 00:39:30.930 --> 00:39:33.450 align:middle line:90% can denounce the cheating, OK? 00:39:33.450 --> 00:39:37.830 align:middle line:84% And this can happen for two reasons. 00:39:37.830 --> 00:39:40.875 align:middle line:84% There can be random shocks, which 00:39:40.875 --> 00:39:42.750 align:middle line:84% is if someone makes a mistake or someone gets 00:39:42.750 --> 00:39:45.150 align:middle line:90% annoyed or aggravated. 00:39:45.150 --> 00:39:48.900 align:middle line:84% Or you can actually have rational whistleblowing, OK? 00:39:48.900 --> 00:39:51.000 align:middle line:84% And both of these are going to make 00:39:51.000 --> 00:39:58.890 align:middle line:84% it harder to have misreporting in larger firms. 00:39:58.890 --> 00:40:00.390 align:middle line:90% So why is that? 00:40:00.390 --> 00:40:02.820 align:middle line:84% So here's the random shock version. 00:40:02.820 --> 00:40:05.400 align:middle line:90% It's very simple. 00:40:05.400 --> 00:40:09.360 align:middle line:84% So if w is not equal to w bar, then 00:40:09.360 --> 00:40:10.860 align:middle line:84% imagine that each employee denounces 00:40:10.860 --> 00:40:12.180 align:middle line:90% with probability epsilon. 00:40:12.180 --> 00:40:12.960 align:middle line:90% What is epsilon? 00:40:12.960 --> 00:40:14.550 align:middle line:90% It's I get really mad at you. 00:40:14.550 --> 00:40:16.140 align:middle line:84% And I go to the IRS and turn you in. 00:40:16.140 --> 00:40:18.150 align:middle line:84% And actually, if you've talked to the IRS, 00:40:18.150 --> 00:40:19.650 align:middle line:84% how do they learn stuff, it's really 00:40:19.650 --> 00:40:21.570 align:middle line:90% a lot of cases like that. 00:40:21.570 --> 00:40:24.330 align:middle line:84% Some employer has some disagreement with the employee. 00:40:24.330 --> 00:40:25.860 align:middle line:84% The employee gets really aggravated 00:40:25.860 --> 00:40:27.600 align:middle line:84% and is like, well, I'm going to show you, 00:40:27.600 --> 00:40:32.980 align:middle line:84% and marches into the IRS office and denounces the firm, OK? 00:40:32.980 --> 00:40:36.660 align:middle line:84% So I think this idea that people get upset with some probability 00:40:36.660 --> 00:40:39.570 align:middle line:84% and go to denounce the firm is actually 00:40:39.570 --> 00:40:42.487 align:middle line:84% a pretty reasonable summary of what happens. 00:40:42.487 --> 00:40:44.820 align:middle line:84% So actually, in fact, we had a student a couple of years 00:40:44.820 --> 00:40:48.870 align:middle line:84% ago who was looking at not tax, but but a related thing 00:40:48.870 --> 00:40:51.270 align:middle line:84% of Medicare and Medicaid fraud in the US 00:40:51.270 --> 00:40:52.830 align:middle line:84% and talked to a bunch of prosecutors. 00:40:52.830 --> 00:40:53.520 align:middle line:90% And they were like, yes. 00:40:53.520 --> 00:40:55.110 align:middle line:84% This is exactly how we learn about stuff. 00:40:55.110 --> 00:40:56.970 align:middle line:84% Some disgruntled employee walks in and is like, 00:40:56.970 --> 00:40:58.500 align:middle line:84% I'm going to tell you what's actually going on. 00:40:58.500 --> 00:40:59.980 align:middle line:90% And they're like, please do. 00:40:59.980 --> 00:41:01.400 align:middle line:90% And OK. 00:41:01.400 --> 00:41:03.240 align:middle line:90% So OK. 00:41:03.240 --> 00:41:09.285 align:middle line:90% So what's going to happen? 00:41:09.285 --> 00:41:11.160 align:middle line:84% Well, this is almost obvious when you say it. 00:41:11.160 --> 00:41:12.420 align:middle line:84% What's the probability the firm's 00:41:12.420 --> 00:41:14.790 align:middle line:84% going to get away with it is 1 minus epsilon to the nth 00:41:14.790 --> 00:41:18.270 align:middle line:84% power, where n is the number of employees you have, right? 00:41:18.270 --> 00:41:21.990 align:middle line:84% So everything basically follows from that, right? 00:41:21.990 --> 00:41:26.790 align:middle line:84% So what is the probability that you get away with it, 00:41:26.790 --> 00:41:30.400 align:middle line:84% essentially, is going to be declining with that. 00:41:30.400 --> 00:41:33.690 align:middle line:84% So at some point, it's going to basically 00:41:33.690 --> 00:41:36.218 align:middle line:84% be not worth it for you to do that. 00:41:36.218 --> 00:41:38.010 align:middle line:84% And in particular, they give this condition 00:41:38.010 --> 00:41:40.470 align:middle line:84% that they're going to evade if 1 minus epsilon to the nth 00:41:40.470 --> 00:41:43.618 align:middle line:84% is greater than theta over 1 plus theta. 00:41:43.618 --> 00:41:44.910 align:middle line:90% You can work this out yourself. 00:41:44.910 --> 00:41:46.500 align:middle line:90% I don't think it's crucial. 00:41:46.500 --> 00:41:48.510 align:middle line:90% This is the key point, right? 00:41:48.510 --> 00:41:50.370 align:middle line:84% The more employees you have, the more likely 00:41:50.370 --> 00:41:51.990 align:middle line:84% that someone's going to get aggravated and denounce you 00:41:51.990 --> 00:41:52.350 align:middle line:90% to the government. 00:41:52.350 --> 00:41:54.060 align:middle line:84% And therefore, really small firms 00:41:54.060 --> 00:41:56.122 align:middle line:84% can probably sustain this with high probability. 00:41:56.122 --> 00:41:57.330 align:middle line:90% And therefore, it's worth it. 00:41:57.330 --> 00:41:59.170 align:middle line:84% For really large firms, the probably you get 00:41:59.170 --> 00:42:00.360 align:middle line:90% denounced is really high. 00:42:00.360 --> 00:42:03.720 align:middle line:84% And because the penalty is applied to all-- 00:42:03.720 --> 00:42:05.280 align:middle line:84% and the key thing driving this model 00:42:05.280 --> 00:42:08.460 align:middle line:84% is it only takes one person to denounce. 00:42:08.460 --> 00:42:10.293 align:middle line:84% But once I denounce, I catch everybody. 00:42:10.293 --> 00:42:11.460 align:middle line:90% And you pay the big penalty. 00:42:11.460 --> 00:42:19.710 align:middle line:84% So the penalty per person is constant. 00:42:19.710 --> 00:42:22.910 align:middle line:84% But the probability is increasing with them. 00:42:22.910 --> 00:42:24.467 align:middle line:90% OK, so that's that model. 00:42:24.467 --> 00:42:26.550 align:middle line:84% The other model-- oh, I didn't actually put it up. 00:42:26.550 --> 00:42:28.260 align:middle line:84% The other model is actually, I think, even 00:42:28.260 --> 00:42:29.760 align:middle line:84% a little more interesting, actually, 00:42:29.760 --> 00:42:31.470 align:middle line:84% which is this whistleblower idea. 00:42:31.470 --> 00:42:33.900 align:middle line:90% And the whistleblower idea is-- 00:42:33.900 --> 00:42:36.510 align:middle line:84% and this actually-- the US actually does exactly-- 00:42:36.510 --> 00:42:37.530 align:middle line:84% it's actually one of the few countries 00:42:37.530 --> 00:42:38.790 align:middle line:90% I know that does exactly this. 00:42:38.790 --> 00:42:44.700 align:middle line:84% So the US's whistleblower policy is if you walk into the IRS 00:42:44.700 --> 00:42:49.840 align:middle line:84% and you say MIT has been cheating on its taxes 00:42:49.840 --> 00:42:56.080 align:middle line:84% and here's how, they pay you, the whistleblower, between 15% 00:42:56.080 --> 00:42:58.840 align:middle line:90% to 30% of what they recover. 00:42:58.840 --> 00:43:00.330 align:middle line:90% OK? 00:43:00.330 --> 00:43:02.640 align:middle line:90% So why is that a great policy? 00:43:02.640 --> 00:43:07.050 align:middle line:84% Because now as the firm gets large, 00:43:07.050 --> 00:43:10.770 align:middle line:84% any individual employee can walk into the IRS 00:43:10.770 --> 00:43:12.810 align:middle line:90% and take the whole 30%, right? 00:43:12.810 --> 00:43:16.050 align:middle line:84% So if for a firm of size 2, for example, 00:43:16.050 --> 00:43:17.610 align:middle line:90% this is not worth it, right? 00:43:17.610 --> 00:43:20.310 align:middle line:84% Because you're only getting, say, 30%. 00:43:20.310 --> 00:43:22.270 align:middle line:84% But you were getting half the surplus. 00:43:22.270 --> 00:43:23.890 align:middle line:84% So 30% is less than half the surplus. 00:43:23.890 --> 00:43:25.170 align:middle line:90% So you're worse off. 00:43:25.170 --> 00:43:29.320 align:middle line:84% For a firm of size 10,000, this is a really awesome deal, 00:43:29.320 --> 00:43:29.820 align:middle line:90% right? 00:43:29.820 --> 00:43:31.278 align:middle line:84% And therefore, equilibrium people-- 00:43:31.278 --> 00:43:33.612 align:middle line:84% they don't actually have to pay very much in equilibrium 00:43:33.612 --> 00:43:34.950 align:middle line:90% because nobody cheats, right? 00:43:34.950 --> 00:43:39.120 align:middle line:84% So in equilibrium, this is free, but prevents cheating 00:43:39.120 --> 00:43:41.670 align:middle line:90% in large firms. 00:43:41.670 --> 00:43:46.940 align:middle line:84% So I think this idea of these whistleblower rewards, 00:43:46.940 --> 00:43:50.030 align:middle line:84% actually-- this one, I think, is more the developing country 00:43:50.030 --> 00:43:50.563 align:middle line:90% version. 00:43:50.563 --> 00:43:52.730 align:middle line:84% Because I haven't seen a lot of developing countries 00:43:52.730 --> 00:43:53.897 align:middle line:90% do this whistleblower thing. 00:43:53.897 --> 00:44:02.930 align:middle line:84% But the same argument, I think, also could be there. 00:44:02.930 --> 00:44:05.590 align:middle line:84% But in any case, both of these have the principle 00:44:05.590 --> 00:44:08.620 align:middle line:84% that as the economy grows, the optimal firm 00:44:08.620 --> 00:44:10.130 align:middle line:90% size is going to increase. 00:44:10.130 --> 00:44:11.050 align:middle line:84% And therefore, you're going to end up 00:44:11.050 --> 00:44:12.217 align:middle line:90% in these really large firms. 00:44:12.217 --> 00:44:14.710 align:middle line:84% And in really large firms, it's basically, 00:44:14.710 --> 00:44:18.070 align:middle line:84% it's just too risky to cheat because someone will probably 00:44:18.070 --> 00:44:19.270 align:middle line:90% denounce you, OK? 00:44:19.270 --> 00:44:21.380 align:middle line:84% And I think there's probably something to that. 00:44:21.380 --> 00:44:22.240 align:middle line:90% Yeah. 00:44:22.240 --> 00:44:23.990 align:middle line:84% AUDIENCE: It's actually the case that it's 00:44:23.990 --> 00:44:28.570 align:middle line:84% a firm [INAUDIBLE] individuals who actually have information 00:44:28.570 --> 00:44:33.130 align:middle line:84% about the taxes or a tax rep within the firm also 00:44:33.130 --> 00:44:34.887 align:middle line:84% [INAUDIBLE] proportionately to the firm? 00:44:34.887 --> 00:44:36.470 align:middle line:84% BEN OLKEN: Yeah, it's a good question. 00:44:36.470 --> 00:44:37.360 align:middle line:90% So you're right. 00:44:37.360 --> 00:44:39.340 align:middle line:84% So in this model, everybody can denounce. 00:44:39.340 --> 00:44:40.818 align:middle line:84% You might think as the firm grows, 00:44:40.818 --> 00:44:42.610 align:middle line:84% we have a specialized accounting department 00:44:42.610 --> 00:44:44.200 align:middle line:84% who becomes more and more specialized. 00:44:44.200 --> 00:44:45.340 align:middle line:84% And maybe it's only three accountants 00:44:45.340 --> 00:44:46.610 align:middle line:84% actually have all the information. 00:44:46.610 --> 00:44:48.340 align:middle line:84% The problem is every individual knows they were 00:44:48.340 --> 00:44:49.465 align:middle line:90% cheating on your own taxes. 00:44:49.465 --> 00:44:52.055 align:middle line:90% 00:44:52.055 --> 00:44:53.430 align:middle line:84% AUDIENCE: Every individual knows? 00:44:53.430 --> 00:44:54.097 align:middle line:90% BEN OLKEN: Yeah. 00:44:54.097 --> 00:44:58.780 align:middle line:84% Because even in the model with an accounting department, 00:44:58.780 --> 00:45:03.330 align:middle line:84% you know what you actually got paid. 00:45:03.330 --> 00:45:05.610 align:middle line:84% And you know what the firm told you 00:45:05.610 --> 00:45:07.770 align:middle line:90% to write down on your taxes. 00:45:07.770 --> 00:45:10.580 align:middle line:84% So everybody knows that the firm is cheating. 00:45:10.580 --> 00:45:13.155 align:middle line:84% And that could be enough to start the process. 00:45:13.155 --> 00:45:15.030 align:middle line:84% You could at least walk into the IRS and say, 00:45:15.030 --> 00:45:17.438 align:middle line:84% I don't know how prevalent this is, but here's what 00:45:17.438 --> 00:45:18.230 align:middle line:90% they told me to do. 00:45:18.230 --> 00:45:20.060 align:middle line:84% And here's what they told my buddy to do. 00:45:20.060 --> 00:45:21.350 align:middle line:90% And maybe that's enough. 00:45:21.350 --> 00:45:23.775 align:middle line:84% So you're right that that creation 00:45:23.775 --> 00:45:25.400 align:middle line:84% of specialized knowledge may undo this. 00:45:25.400 --> 00:45:27.410 align:middle line:84% But I think the key point is everybody 00:45:27.410 --> 00:45:30.380 align:middle line:84% has to be complicit in the cheating. 00:45:30.380 --> 00:45:34.060 align:middle line:90% 00:45:34.060 --> 00:45:36.620 align:middle line:90% OK. 00:45:36.620 --> 00:45:37.850 align:middle line:90% So this paper by Jensen-- 00:45:37.850 --> 00:45:41.990 align:middle line:84% I think this is actually now forthcoming in the AER-- 00:45:41.990 --> 00:45:46.040 align:middle line:84% takes seriously this idea that basically, individual income 00:45:46.040 --> 00:45:48.518 align:middle line:84% taxation only works in formal employment relationships. 00:45:48.518 --> 00:45:50.060 align:middle line:84% So another way of saying it is that-- 00:45:50.060 --> 00:45:54.236 align:middle line:90% 00:45:54.236 --> 00:45:55.450 align:middle line:90% so what's the idea? 00:45:55.450 --> 00:45:56.830 align:middle line:90% He says, look. 00:45:56.830 --> 00:45:59.080 align:middle line:84% If I'm an employee in a formal firm, 00:45:59.080 --> 00:46:01.330 align:middle line:84% then the firm is double reporting, right? 00:46:01.330 --> 00:46:03.940 align:middle line:84% And maybe the government can try to collect it. 00:46:03.940 --> 00:46:07.000 align:middle line:84% But if I'm in an informal firm, or if I'm 00:46:07.000 --> 00:46:08.680 align:middle line:84% self-employed or whatever, forget it. 00:46:08.680 --> 00:46:11.170 align:middle line:90% The government has no hope, OK? 00:46:11.170 --> 00:46:14.200 align:middle line:84% And Jensen asks, well, what are the implications 00:46:14.200 --> 00:46:18.490 align:middle line:84% of this for the tax structure more broadly? 00:46:18.490 --> 00:46:21.700 align:middle line:84% And in particular, he basically postulates 00:46:21.700 --> 00:46:27.040 align:middle line:84% that countries realize that when the employment share-- 00:46:27.040 --> 00:46:30.668 align:middle line:84% so what are the implications for how we design the tax code, OK? 00:46:30.668 --> 00:46:32.710 align:middle line:84% And he argues that basically, when the employment 00:46:32.710 --> 00:46:35.050 align:middle line:84% share in a given income decile is too low, 00:46:35.050 --> 00:46:36.580 align:middle line:84% governments basically just give up 00:46:36.580 --> 00:46:38.455 align:middle line:84% on taxing the people altogether and just make 00:46:38.455 --> 00:46:40.005 align:middle line:90% them exempt from income tax. 00:46:40.005 --> 00:46:42.220 align:middle line:90% They don't even try, OK? 00:46:42.220 --> 00:46:47.170 align:middle line:84% And then he basically argues that as countries develop, 00:46:47.170 --> 00:46:50.380 align:middle line:84% as countries become larger, the share of people who are working 00:46:50.380 --> 00:46:51.910 align:middle line:84% as employees in those larger firms-- 00:46:51.910 --> 00:46:57.910 align:middle line:84% that increases and moves down the income percentiles. 00:46:57.910 --> 00:47:00.340 align:middle line:84% And therefore, countries are going to change their tax 00:47:00.340 --> 00:47:02.650 align:middle line:90% structure endogenously. 00:47:02.650 --> 00:47:03.150 align:middle line:90% OK? 00:47:03.150 --> 00:47:05.400 align:middle line:84% So I'm going to focus just on some cross-country facts 00:47:05.400 --> 00:47:08.050 align:middle line:90% for now, which I think are cool. 00:47:08.050 --> 00:47:08.550 align:middle line:90% Sorry. 00:47:08.550 --> 00:47:10.590 align:middle line:84% These are first the US over time. 00:47:10.590 --> 00:47:12.100 align:middle line:90% So what is this? 00:47:12.100 --> 00:47:18.335 align:middle line:84% So this is the employee share, OK? 00:47:18.335 --> 00:47:23.050 align:middle line:90% 00:47:23.050 --> 00:47:26.920 align:middle line:84% The employee share by income decile, OK? 00:47:26.920 --> 00:47:29.500 align:middle line:84% So this says what fraction of people 00:47:29.500 --> 00:47:33.730 align:middle line:84% are employees in part of a firm or self-employed? 00:47:33.730 --> 00:47:36.490 align:middle line:84% The informal sector, we're not really going to focus on here. 00:47:36.490 --> 00:47:37.420 align:middle line:90% So what do you see? 00:47:37.420 --> 00:47:43.540 align:middle line:84% So in the 1870s, it looks like basically, most 00:47:43.540 --> 00:47:46.510 align:middle line:84% of the people in the United States in most income deciles-- 00:47:46.510 --> 00:47:49.810 align:middle line:84% it was only the employee share only 00:47:49.810 --> 00:47:53.340 align:middle line:84% happens at the very top of the income distribution. 00:47:53.340 --> 00:47:54.870 align:middle line:84% And we have no income tax in 1870. 00:47:54.870 --> 00:48:00.350 align:middle line:84% So the US income tax was introduced in 1918, 1919. 00:48:00.350 --> 00:48:03.310 align:middle line:84% I don't remember exactly-- approximately in 1918 00:48:03.310 --> 00:48:05.480 align:middle line:90% plus or minus a couple of years. 00:48:05.480 --> 00:48:06.150 align:middle line:90% 1912? 00:48:06.150 --> 00:48:07.720 align:middle line:90% Does anyone remember? 00:48:07.720 --> 00:48:09.040 align:middle line:90% It's in the 1910s. 00:48:09.040 --> 00:48:10.960 align:middle line:84% There's a Constitutional Amendment to it. 00:48:10.960 --> 00:48:12.800 align:middle line:90% Anyway, approximately then. 00:48:12.800 --> 00:48:15.770 align:middle line:90% Certainly not here and yes here. 00:48:15.770 --> 00:48:23.790 align:middle line:84% And so this bar is what is where is the exemption of the income 00:48:23.790 --> 00:48:24.713 align:middle line:90% tax? 00:48:24.713 --> 00:48:26.880 align:middle line:84% So almost all countries say people below some income 00:48:26.880 --> 00:48:28.530 align:middle line:90% don't pay income tax. 00:48:28.530 --> 00:48:31.380 align:middle line:84% And he says, well, where is that threshold in the income 00:48:31.380 --> 00:48:32.960 align:middle line:90% distribution? 00:48:32.960 --> 00:48:35.350 align:middle line:84% And you can see that basically what's 00:48:35.350 --> 00:48:38.350 align:middle line:84% happening in the US over time is two things. 00:48:38.350 --> 00:48:41.640 align:middle line:84% As the US grows, the employee share is really changing. 00:48:41.640 --> 00:48:42.140 align:middle line:90% OK? 00:48:42.140 --> 00:48:44.350 align:middle line:84% So it's always the case that richer people are 00:48:44.350 --> 00:48:45.580 align:middle line:90% more likely to be employees. 00:48:45.580 --> 00:48:48.280 align:middle line:84% But where that thing is happening 00:48:48.280 --> 00:48:51.700 align:middle line:90% is moving down the distribution. 00:48:51.700 --> 00:48:55.420 align:middle line:84% And as that's happening, the income tax exemption threshold 00:48:55.420 --> 00:48:57.240 align:middle line:90% also is moving to the left. 00:48:57.240 --> 00:48:57.740 align:middle line:90% OK? 00:48:57.740 --> 00:48:59.240 align:middle line:90% And that's the US over time. 00:48:59.240 --> 00:49:07.360 align:middle line:84% So the next point is that if you look across countries 00:49:07.360 --> 00:49:09.230 align:middle line:84% you see very similar patterns, OK? 00:49:09.230 --> 00:49:13.360 align:middle line:84% So India looks a lot more like the historical US, right? 00:49:13.360 --> 00:49:16.300 align:middle line:84% Only employee share is really high. 00:49:16.300 --> 00:49:18.530 align:middle line:90% Very high income tax threshold. 00:49:18.530 --> 00:49:20.800 align:middle line:84% Indonesia-- a bit richer in his data. 00:49:20.800 --> 00:49:22.900 align:middle line:84% It's a bit to the left of the threshold. 00:49:22.900 --> 00:49:25.690 align:middle line:84% And again, this pattern is similar. 00:49:25.690 --> 00:49:28.318 align:middle line:84% If I were to show you Mexico, it's what you expect. 00:49:28.318 --> 00:49:29.860 align:middle line:84% It's going to look like a little more 00:49:29.860 --> 00:49:32.380 align:middle line:84% to the left on both the black line and those crosses 00:49:32.380 --> 00:49:34.290 align:middle line:90% and so on and so forth. 00:49:34.290 --> 00:49:39.330 align:middle line:84% And then you can plot all these different countries. 00:49:39.330 --> 00:49:42.330 align:middle line:84% And you can see a couple of things. 00:49:42.330 --> 00:49:48.530 align:middle line:84% The first is what is the overall employee share? 00:49:48.530 --> 00:49:51.600 align:middle line:84% What fraction of people are employees versus self-employed? 00:49:51.600 --> 00:49:53.480 align:middle line:84% You can see this general pattern we 00:49:53.480 --> 00:49:55.370 align:middle line:84% were seeing that this is moving to the left 00:49:55.370 --> 00:49:57.890 align:middle line:84% and getting higher as we become richer-- 00:49:57.890 --> 00:50:00.560 align:middle line:84% that seems to be true looking at a cross-section of countries 00:50:00.560 --> 00:50:01.980 align:middle line:90% around the world, OK? 00:50:01.980 --> 00:50:03.920 align:middle line:84% So richer countries have high employee shares. 00:50:03.920 --> 00:50:09.220 align:middle line:84% Poorer countries have much lower employee shares, OK? 00:50:09.220 --> 00:50:13.840 align:middle line:84% You also see that the personal income tax 00:50:13.840 --> 00:50:16.540 align:middle line:84% base is what share of the population 00:50:16.540 --> 00:50:22.330 align:middle line:84% is above that black line required 00:50:22.330 --> 00:50:24.340 align:middle line:90% to pay personal income taxes? 00:50:24.340 --> 00:50:27.170 align:middle line:84% That's also declining with income. 00:50:27.170 --> 00:50:31.510 align:middle line:84% So richer countries are trying to tax 00:50:31.510 --> 00:50:33.160 align:middle line:90% further down the distribution. 00:50:33.160 --> 00:50:36.550 align:middle line:84% And this says what fraction-- this is the most amazing graph 00:50:36.550 --> 00:50:37.600 align:middle line:90% of this paper. 00:50:37.600 --> 00:50:42.790 align:middle line:84% This is what fraction of the total employees 00:50:42.790 --> 00:50:45.760 align:middle line:84% are in the personal income tax base? 00:50:45.760 --> 00:50:47.860 align:middle line:84% So it's the ratio of those two things, right? 00:50:47.860 --> 00:50:50.740 align:middle line:84% What fraction of the people who are employees 00:50:50.740 --> 00:50:55.840 align:middle line:84% are above the black line is almost constant as a fraction 00:50:55.840 --> 00:50:57.564 align:middle line:90% of development. 00:50:57.564 --> 00:50:58.510 align:middle line:90% OK? 00:50:58.510 --> 00:51:01.540 align:middle line:84% So this is basically arguing that as countries-- what's 00:51:01.540 --> 00:51:04.900 align:middle line:84% happening is countries are realizing that they can only 00:51:04.900 --> 00:51:06.460 align:middle line:90% tax people who are employees. 00:51:06.460 --> 00:51:09.250 align:middle line:84% And they wait until-- and they're setting-- 00:51:09.250 --> 00:51:11.170 align:middle line:84% they're designing their tax schedule 00:51:11.170 --> 00:51:16.240 align:middle line:84% to take into account the fact that basically, it's 00:51:16.240 --> 00:51:18.310 align:middle line:84% only worth it to go after the employees. 00:51:18.310 --> 00:51:20.610 align:middle line:84% And they're doing that almost a constant-- not exactly 00:51:20.610 --> 00:51:21.110 align:middle line:90% constant. 00:51:21.110 --> 00:51:23.193 align:middle line:84% It's still the case that it's a little bit higher. 00:51:23.193 --> 00:51:27.407 align:middle line:84% But I think this is remarkably flat by comparison. 00:51:27.407 --> 00:51:28.990 align:middle line:84% AUDIENCE: I think that [INAUDIBLE]---- 00:51:28.990 --> 00:51:29.657 align:middle line:90% BEN OLKEN: What? 00:51:29.657 --> 00:51:31.300 align:middle line:90% AUDIENCE: I have a question. 00:51:31.300 --> 00:51:34.297 align:middle line:84% What happens with informal employees? 00:51:34.297 --> 00:51:36.005 align:middle line:84% BEN OLKEN: This is only formal employees. 00:51:36.005 --> 00:51:37.228 align:middle line:90% AUDIENCE: OK, [INAUDIBLE]. 00:51:37.228 --> 00:51:39.020 align:middle line:84% BEN OLKEN: Should only be formal employees. 00:51:39.020 --> 00:51:40.490 align:middle line:84% I believe it's only formal employees. 00:51:40.490 --> 00:51:42.698 align:middle line:84% I'm not 100% sure that's how they calculate the data. 00:51:42.698 --> 00:51:44.490 align:middle line:84% But it should only be formal employees. 00:51:44.490 --> 00:51:47.788 align:middle line:84% The model is only about formal employees. 00:51:47.788 --> 00:51:49.830 align:middle line:84% I don't remember the details of how he constructs 00:51:49.830 --> 00:51:50.776 align:middle line:84% this from all different data sets 00:51:50.776 --> 00:51:52.984 align:middle line:84% to know if it's formal or informal employee examples, 00:51:52.984 --> 00:51:55.630 align:middle line:84% but it certainly should be based on the model. 00:51:55.630 --> 00:51:58.040 align:middle line:90% Yeah? 00:51:58.040 --> 00:52:04.310 align:middle line:84% AUDIENCE: What if informality is [INAUDIBLE] imagine 00:52:04.310 --> 00:52:06.554 align:middle line:84% maybe everybody shows the same [INAUDIBLE] in the US, 00:52:06.554 --> 00:52:11.380 align:middle line:84% but [INAUDIBLE] there are some countries that 00:52:11.380 --> 00:52:14.700 align:middle line:84% hire the same employee and just choose to have them be formally 00:52:14.700 --> 00:52:17.075 align:middle line:84% on the books [? and social security ?] 00:52:17.075 --> 00:52:21.320 align:middle line:84% or just have them pay the cash [INAUDIBLE].. 00:52:21.320 --> 00:52:24.583 align:middle line:84% So it's [INAUDIBLE] salary [INAUDIBLE].. 00:52:24.583 --> 00:52:25.250 align:middle line:90% BEN OLKEN: Yeah. 00:52:25.250 --> 00:52:28.207 align:middle line:84% So let me apologize that I don't remember 00:52:28.207 --> 00:52:30.290 align:middle line:84% exactly how he's constructing this data, to answer 00:52:30.290 --> 00:52:30.915 align:middle line:90% Sal's question. 00:52:30.915 --> 00:52:33.080 align:middle line:84% So I don't remember that question well enough, 00:52:33.080 --> 00:52:34.380 align:middle line:90% and I should. 00:52:34.380 --> 00:52:37.490 align:middle line:84% But I think the point is this is an equilibrium relationship. 00:52:37.490 --> 00:52:39.840 align:middle line:84% I think that's the way I would think about this graph. 00:52:39.840 --> 00:52:42.915 align:middle line:84% I guess the broader point I'm trying to-- so I think this is 00:52:42.915 --> 00:52:44.825 align:middle line:90% a, I think, very-- 00:52:44.825 --> 00:52:46.950 align:middle line:84% and there's another part of this looking at the US. 00:52:46.950 --> 00:52:51.600 align:middle line:84% But I think to me, what I think is really pretty remarkable 00:52:51.600 --> 00:52:54.540 align:middle line:84% about this is showing that at least it looks like-- 00:52:54.540 --> 00:52:57.540 align:middle line:84% and the data is really very consistent with countries 00:52:57.540 --> 00:53:02.010 align:middle line:84% choosing their threshold of who we're even trying 00:53:02.010 --> 00:53:03.933 align:middle line:84% to tax, taking into account what's going on 00:53:03.933 --> 00:53:05.100 align:middle line:90% in the employment structure. 00:53:05.100 --> 00:53:07.110 align:middle line:84% I agree with you that there's a whole range 00:53:07.110 --> 00:53:10.182 align:middle line:84% of additional questions to think about-- formal versus informal, 00:53:10.182 --> 00:53:11.140 align:middle line:90% and so on and so forth. 00:53:11.140 --> 00:53:12.540 align:middle line:84% So I don't think this is by any means the last word on this. 00:53:12.540 --> 00:53:14.580 align:middle line:84% But I just think the point I want to illustrate 00:53:14.580 --> 00:53:20.790 align:middle line:84% is that the idea is that thinking 00:53:20.790 --> 00:53:24.500 align:middle line:84% about the previous model, the data is 00:53:24.500 --> 00:53:27.000 align:middle line:84% consistent with the idea that countries realize you can only 00:53:27.000 --> 00:53:30.720 align:middle line:84% tax people who are in these employment 00:53:30.720 --> 00:53:33.480 align:middle line:84% relationships in equilibrium and are adjusting their tax 00:53:33.480 --> 00:53:34.410 align:middle line:90% code implicitly. 00:53:34.410 --> 00:53:36.960 align:middle line:84% And therefore, as countries develop, 00:53:36.960 --> 00:53:38.820 align:middle line:84% one thing that naturally happens is 00:53:38.820 --> 00:53:42.160 align:middle line:84% because they work in larger companies 00:53:42.160 --> 00:53:44.910 align:middle line:84% and that happens as part of the development process, that 00:53:44.910 --> 00:53:46.618 align:middle line:84% almost always has implications for what's 00:53:46.618 --> 00:53:49.360 align:middle line:90% going on on the tax side. 00:53:49.360 --> 00:53:51.650 align:middle line:90% Questions? 00:53:51.650 --> 00:53:52.150 align:middle line:90% OK. 00:53:52.150 --> 00:53:55.120 align:middle line:90% 00:53:55.120 --> 00:54:01.920 align:middle line:84% I think one question that this may raise 00:54:01.920 --> 00:54:06.810 align:middle line:84% is can you improve on the ability of firms 00:54:06.810 --> 00:54:08.650 align:middle line:90% to report on their workers? 00:54:08.650 --> 00:54:12.060 align:middle line:84% So can we-- if this whole thing was 00:54:12.060 --> 00:54:14.430 align:middle line:84% taken as exogenous epsilon thing, 00:54:14.430 --> 00:54:16.230 align:middle line:90% but can we change this, right? 00:54:16.230 --> 00:54:21.690 align:middle line:84% Can firms change epsilons, or the probability 00:54:21.690 --> 00:54:22.620 align:middle line:90% of getting a report? 00:54:22.620 --> 00:54:24.120 align:middle line:84% Can governments change epsilon, the probability 00:54:24.120 --> 00:54:25.290 align:middle line:90% of getting a report? 00:54:25.290 --> 00:54:27.310 align:middle line:84% And I just want to mention one paper here, 00:54:27.310 --> 00:54:29.023 align:middle line:84% which I think is a really nice paper that 00:54:29.023 --> 00:54:29.940 align:middle line:90% tries to look at this. 00:54:29.940 --> 00:54:33.245 align:middle line:84% I think this paper doesn't-- it has some data limitations. 00:54:33.245 --> 00:54:35.620 align:middle line:84% So I don't think totally is the last word on the subject. 00:54:35.620 --> 00:54:37.412 align:middle line:84% So I just want to mention this as, I think, 00:54:37.412 --> 00:54:40.695 align:middle line:84% an exciting, interesting question to think about. 00:54:40.695 --> 00:54:41.820 align:middle line:90% So what does this paper do? 00:54:41.820 --> 00:54:43.230 align:middle line:90% So this paper says, look. 00:54:43.230 --> 00:54:46.110 align:middle line:84% Under Mexico's old pension system, 00:54:46.110 --> 00:54:47.820 align:middle line:84% workers' pensions were essentially 00:54:47.820 --> 00:54:50.355 align:middle line:84% a flat function of how much your wages were reported. 00:54:50.355 --> 00:54:53.796 align:middle line:90% 00:54:53.796 --> 00:54:56.678 align:middle line:84% Sorry, they were flat function-- they were not a function of how 00:54:56.678 --> 00:54:57.720 align:middle line:90% much your wages reported. 00:54:57.720 --> 00:54:59.520 align:middle line:84% They were just a function of years of work. 00:54:59.520 --> 00:55:02.010 align:middle line:84% So you didn't care what the government 00:55:02.010 --> 00:55:04.590 align:middle line:90% was reporting on your behalf. 00:55:04.590 --> 00:55:07.942 align:middle line:84% In the new system, you basically, essentially 00:55:07.942 --> 00:55:09.900 align:middle line:84% get some fraction put into your pension account 00:55:09.900 --> 00:55:11.692 align:middle line:84% as a fraction of how much the government is 00:55:11.692 --> 00:55:13.003 align:middle line:90% reporting of your income. 00:55:13.003 --> 00:55:13.690 align:middle line:90% OK? 00:55:13.690 --> 00:55:16.107 align:middle line:84% So now all of a sudden, you go from a world where you have 00:55:16.107 --> 00:55:19.110 align:middle line:84% an incentive to care about what the government is telling about 00:55:19.110 --> 00:55:22.380 align:middle line:84% your income from one where you don't. 00:55:22.380 --> 00:55:24.690 align:middle line:84% And they try to establish a different diff 00:55:24.690 --> 00:55:26.190 align:middle line:84% by saying after the reform, maybe 00:55:26.190 --> 00:55:28.080 align:middle line:84% young workers have a greater incentive 00:55:28.080 --> 00:55:29.747 align:middle line:84% to make sure firms report wages honestly 00:55:29.747 --> 00:55:30.960 align:middle line:90% than before the reform. 00:55:30.960 --> 00:55:35.010 align:middle line:84% And they find better match in wages before and after. 00:55:35.010 --> 00:55:36.180 align:middle line:90% The paper, as I said-- 00:55:36.180 --> 00:55:37.783 align:middle line:90% there's empirical challenges. 00:55:37.783 --> 00:55:39.700 align:middle line:84% They don't quite-- so it's a little bit tough. 00:55:39.700 --> 00:55:41.010 align:middle line:90% But I think this question of-- 00:55:41.010 --> 00:55:43.135 align:middle line:84% I just wanted to mention this as one example of how 00:55:43.135 --> 00:55:46.873 align:middle line:90% do you think about improving-- 00:55:46.873 --> 00:55:48.540 align:middle line:84% how do you think about changing epsilon? 00:55:48.540 --> 00:55:50.665 align:middle line:84% Actually, this was something that I actually myself 00:55:50.665 --> 00:55:51.210 align:middle line:90% tried to do. 00:55:51.210 --> 00:55:54.870 align:middle line:84% I tried to develop an RCT on this topic in Indonesia. 00:55:54.870 --> 00:55:56.490 align:middle line:90% It didn't end up working out. 00:55:56.490 --> 00:55:59.280 align:middle line:84% But we were thinking of well, maybe the government 00:55:59.280 --> 00:56:02.340 align:middle line:84% could just start sending people text messages and say hey, 00:56:02.340 --> 00:56:04.890 align:middle line:84% here's what the government-- here's 00:56:04.890 --> 00:56:07.450 align:middle line:84% what was reported for your wages this month 00:56:07.450 --> 00:56:13.482 align:middle line:84% FYI as a way of increasing that epsilon. 00:56:13.482 --> 00:56:14.940 align:middle line:84% But I think more generally, I think 00:56:14.940 --> 00:56:16.320 align:middle line:90% this is an interesting question. 00:56:16.320 --> 00:56:17.430 align:middle line:90% The epsilon isn't fixed. 00:56:17.430 --> 00:56:19.180 align:middle line:84% And maybe you can do something about that. 00:56:19.180 --> 00:56:21.021 align:middle line:84% So I just wanted to put that up there. 00:56:21.021 --> 00:56:23.270 align:middle line:90% OK, questions? 00:56:23.270 --> 00:56:23.770 align:middle line:90% OK. 00:56:23.770 --> 00:56:25.395 align:middle line:84% So now I wanted to talk about the paper 00:56:25.395 --> 00:56:28.170 align:middle line:84% that you guys read, actually, sorry, for last time, which 00:56:28.170 --> 00:56:29.310 align:middle line:90% is about value added tax. 00:56:29.310 --> 00:56:31.740 align:middle line:84% And the reason I want to mention this is I 00:56:31.740 --> 00:56:33.240 align:middle line:84% feel like, as I mentioned, there are 00:56:33.240 --> 00:56:36.630 align:middle line:84% two different approaches to what we can do about this problem. 00:56:36.630 --> 00:56:38.610 align:middle line:84% Or sorry, the first set of papers was just 00:56:38.610 --> 00:56:40.440 align:middle line:84% like, why does this look different 00:56:40.440 --> 00:56:41.920 align:middle line:90% in the developing world? 00:56:41.920 --> 00:56:44.477 align:middle line:84% The second set of papers are like, what could we possibly 00:56:44.477 --> 00:56:45.060 align:middle line:90% do about this? 00:56:45.060 --> 00:56:47.740 align:middle line:84% Number one is on the tax design side. 00:56:47.740 --> 00:56:50.460 align:middle line:84% And the other is on the tax enforcement side. 00:56:50.460 --> 00:56:53.750 align:middle line:90% 00:56:53.750 --> 00:57:00.072 align:middle line:84% So this paper on the VAT is an example of the tax structure 00:57:00.072 --> 00:57:01.530 align:middle line:84% itself, designing the tax structure 00:57:01.530 --> 00:57:05.010 align:middle line:84% itself to make sure that it's third party reported, OK? 00:57:05.010 --> 00:57:09.300 align:middle line:84% So I think for all the non-Americans in the room 00:57:09.300 --> 00:57:12.083 align:middle line:84% know what a value-added tax is, or knew prior to reading this 00:57:12.083 --> 00:57:13.500 align:middle line:84% because every country in the world 00:57:13.500 --> 00:57:18.510 align:middle line:84% basically has one except for the United States, approximately. 00:57:18.510 --> 00:57:22.770 align:middle line:84% So basically, the idea is that firms are taxed on output 00:57:22.770 --> 00:57:26.730 align:middle line:84% but receive a tax credit for what they already 00:57:26.730 --> 00:57:28.510 align:middle line:90% pay in their inputs. 00:57:28.510 --> 00:57:29.310 align:middle line:90% OK? 00:57:29.310 --> 00:57:32.700 align:middle line:84% And the other thing you have to note is that imports are taxed 00:57:32.700 --> 00:57:34.770 align:middle line:90% and exports are not taxed. 00:57:34.770 --> 00:57:37.710 align:middle line:84% So at the very end of this at the VAT chain, 00:57:37.710 --> 00:57:38.760 align:middle line:90% you have to rebate. 00:57:38.760 --> 00:57:40.320 align:middle line:84% And the reason is you want everything 00:57:40.320 --> 00:57:42.940 align:middle line:84% to be on an equal playing field domestically. 00:57:42.940 --> 00:57:45.180 align:middle line:84% So if you were taxing all domestic production 00:57:45.180 --> 00:57:47.575 align:middle line:84% at, say, 10%, we want to tax imports at 10% 00:57:47.575 --> 00:57:49.825 align:middle line:84% so the imports aren't cheaper than the domestic stuff, 00:57:49.825 --> 00:57:51.180 align:middle line:90% so they're equally taxed. 00:57:51.180 --> 00:57:54.360 align:middle line:84% And the way that the world has worked this out 00:57:54.360 --> 00:57:56.760 align:middle line:84% is that when things come into your border, 00:57:56.760 --> 00:57:58.500 align:middle line:84% you tax them at your local VAT rate. 00:57:58.500 --> 00:58:00.720 align:middle line:84% So you have a uniform tariff at your VAT rate. 00:58:00.720 --> 00:58:03.840 align:middle line:84% And when you export things, the exporting firm 00:58:03.840 --> 00:58:09.250 align:middle line:84% gets a rebate equal to the sum of the VAT on the input chain. 00:58:09.250 --> 00:58:13.950 align:middle line:84% So essentially, everything is taxed at its own domestic VAT 00:58:13.950 --> 00:58:15.990 align:middle line:84% rate regardless of where it is produced. 00:58:15.990 --> 00:58:17.850 align:middle line:90% That's the principle. 00:58:17.850 --> 00:58:21.990 align:middle line:84% I will note that in a lot of developing countries that I've 00:58:21.990 --> 00:58:24.465 align:middle line:90% been engaged with, this step-- 00:58:24.465 --> 00:58:27.140 align:middle line:90% 00:58:27.140 --> 00:58:32.600 align:middle line:90% it ends up being challenging. 00:58:32.600 --> 00:58:35.650 align:middle line:84% So governments love taking money in. 00:58:35.650 --> 00:58:39.380 align:middle line:84% They do not like sending money out as a general rule. 00:58:39.380 --> 00:58:41.905 align:middle line:84% And so they put a lot-- and also, you 00:58:41.905 --> 00:58:43.430 align:middle line:84% can be worried about a lot of fraud. 00:58:43.430 --> 00:58:48.730 align:middle line:84% So they put a lot of checks and cross checks 00:58:48.730 --> 00:58:51.670 align:middle line:84% on those VAT rebates at the export stage. 00:58:51.670 --> 00:58:56.065 align:middle line:84% And that can be an important impediment to trade, actually. 00:58:56.065 --> 00:58:57.940 align:middle line:84% And so actually, thinking about that question 00:58:57.940 --> 00:59:02.140 align:middle line:84% of how do you make that VAT exporting step work properly 00:59:02.140 --> 00:59:09.230 align:middle line:84% is at least, in some cases that I've encountered, a challenge. 00:59:09.230 --> 00:59:09.730 align:middle line:90% OK. 00:59:09.730 --> 00:59:11.443 align:middle line:84% And so what why do a value added tax? 00:59:11.443 --> 00:59:13.860 align:middle line:84% If everyone participates, it's equivalent to a consumption 00:59:13.860 --> 00:59:14.387 align:middle line:90% tax, right? 00:59:14.387 --> 00:59:15.220 align:middle line:90% So this is the idea. 00:59:15.220 --> 00:59:19.125 align:middle line:84% So the idea here of our VAT chain-- 00:59:19.125 --> 00:59:27.150 align:middle line:90% 00:59:27.150 --> 00:59:32.590 align:middle line:84% so I can't remember the way Dina draws it in the paper. 00:59:32.590 --> 00:59:35.125 align:middle line:90% Which way did she draw it? 00:59:35.125 --> 00:59:36.940 align:middle line:84% OK, she goes to the supplier on top, right? 00:59:36.940 --> 00:59:37.940 align:middle line:90% So we have the supplier. 00:59:37.940 --> 00:59:40.360 align:middle line:90% 00:59:40.360 --> 00:59:43.870 align:middle line:90% The chain goes down, right? 00:59:43.870 --> 00:59:50.870 align:middle line:90% Middle and retail, OK? 00:59:50.870 --> 00:59:58.040 align:middle line:84% So we have-- so this is goods are moving down the chart, 00:59:58.040 --> 00:59:58.580 align:middle line:90% right? 00:59:58.580 --> 01:00:00.740 align:middle line:84% So the idea is that this person-- 01:00:00.740 --> 01:00:05.780 align:middle line:84% this is the raw supplier pays VAT on this piece here. 01:00:05.780 --> 01:00:08.635 align:middle line:84% This person pays VAT on this minus this. 01:00:08.635 --> 01:00:10.010 align:middle line:84% So they pay VAT on this and this. 01:00:10.010 --> 01:00:11.480 align:middle line:90% And the retail pays the bottom. 01:00:11.480 --> 01:00:14.870 align:middle line:84% And this is equivalent to just charging a sales 01:00:14.870 --> 01:00:17.710 align:middle line:90% tax at the very bottom, right? 01:00:17.710 --> 01:00:19.900 align:middle line:84% If it's uniform, basically, you're 01:00:19.900 --> 01:00:21.968 align:middle line:84% paying a tax on the whole production chain. 01:00:21.968 --> 01:00:24.010 align:middle line:84% It's just that here it's collected in these three 01:00:24.010 --> 01:00:27.950 align:middle line:84% different pieces whereas here it's collected all at the end. 01:00:27.950 --> 01:00:31.467 align:middle line:84% So in the US, for example, you pay a sales tax 01:00:31.467 --> 01:00:33.550 align:middle line:84% when you buy things in Massachusetts, for example, 01:00:33.550 --> 01:00:34.092 align:middle line:90% of whatever-- 01:00:34.092 --> 01:00:35.440 align:middle line:90% 6 and 1/4%. 01:00:35.440 --> 01:00:36.760 align:middle line:90% They're taxing it down here. 01:00:36.760 --> 01:00:38.950 align:middle line:84% In a VAT, they're trying to do the same thing-- tax 01:00:38.950 --> 01:00:40.908 align:middle line:84% consumption-- but they're taxing it all the way 01:00:40.908 --> 01:00:42.020 align:middle line:90% through the chain. 01:00:42.020 --> 01:00:50.120 align:middle line:84% And the idea of this is that basically, you 01:00:50.120 --> 01:00:53.870 align:middle line:84% get better reporting because in a sales tax, 01:00:53.870 --> 01:00:59.180 align:middle line:84% the retailer has all kinds of incentives not to underreport 01:00:59.180 --> 01:01:00.308 align:middle line:90% the total sale price. 01:01:00.308 --> 01:01:01.850 align:middle line:84% And the government loses those taxes. 01:01:01.850 --> 01:01:03.420 align:middle line:84% That's still true here, by the way. 01:01:03.420 --> 01:01:05.220 align:middle line:84% There's nothing going on over here. 01:01:05.220 --> 01:01:07.040 align:middle line:84% But in all the previous steps, you 01:01:07.040 --> 01:01:09.360 align:middle line:84% have incentives that go in opposite directions. 01:01:09.360 --> 01:01:12.710 align:middle line:84% So this person would like to underreport how much they 01:01:12.710 --> 01:01:18.420 align:middle line:90% are selling their good for. 01:01:18.420 --> 01:01:20.970 align:middle line:84% But this person over here would like 01:01:20.970 --> 01:01:22.678 align:middle line:84% to make that number as big as possible 01:01:22.678 --> 01:01:24.720 align:middle line:84% because they get a credit for that amount, right? 01:01:24.720 --> 01:01:28.140 align:middle line:84% And so the whole idea from governments 01:01:28.140 --> 01:01:30.090 align:middle line:84% is we get accurate reporting of information 01:01:30.090 --> 01:01:32.940 align:middle line:84% if we have information reported by both parties. 01:01:32.940 --> 01:01:35.790 align:middle line:84% The whole idea of the VAT is that for everything 01:01:35.790 --> 01:01:37.470 align:middle line:84% except for the very last step, we 01:01:37.470 --> 01:01:40.350 align:middle line:84% can generate double reporting of information. 01:01:40.350 --> 01:01:44.220 align:middle line:84% And the government can actually use that to figure things out. 01:01:44.220 --> 01:01:45.390 align:middle line:90% OK. 01:01:45.390 --> 01:01:46.530 align:middle line:90% So that's the basic idea. 01:01:46.530 --> 01:01:49.240 align:middle line:84% And this paper is about testing whether that idea actually 01:01:49.240 --> 01:01:49.740 align:middle line:90% works. 01:01:49.740 --> 01:01:53.760 align:middle line:90% And there are these VAT chains. 01:01:53.760 --> 01:01:55.330 align:middle line:84% I think in your comments, I just want 01:01:55.330 --> 01:01:56.580 align:middle line:90% to mention a couple of things. 01:01:56.580 --> 01:02:02.642 align:middle line:84% You guys mentioned that one thing that actually 01:02:02.642 --> 01:02:05.100 align:middle line:84% has to happen, which you'll see as we talk with this paper, 01:02:05.100 --> 01:02:11.760 align:middle line:84% is doing this third party reporting is actually tricky. 01:02:11.760 --> 01:02:17.810 align:middle line:84% So it requires the government collecting tax ID numbers 01:02:17.810 --> 01:02:20.120 align:middle line:84% for everyone to match every transection, or at least 01:02:20.120 --> 01:02:23.070 align:middle line:84% every firm at the firm-to-firm level. 01:02:23.070 --> 01:02:27.127 align:middle line:84% It turns out that actually is harder than you may think. 01:02:27.127 --> 01:02:29.210 align:middle line:84% And at least in some contexts that I've worked at, 01:02:29.210 --> 01:02:30.860 align:middle line:90% they collect all that data. 01:02:30.860 --> 01:02:35.210 align:middle line:84% But actually, they try to actually analyze it and figure 01:02:35.210 --> 01:02:36.710 align:middle line:84% out if people are actually cheating. 01:02:36.710 --> 01:02:39.590 align:middle line:84% It turns out to be a very labor-intensive 01:02:39.590 --> 01:02:41.960 align:middle line:84% and surprisingly hard to automate 01:02:41.960 --> 01:02:45.650 align:middle line:84% process because the world is messy, right? 01:02:45.650 --> 01:02:48.920 align:middle line:84% It's a lot simpler if I have one employer. 01:02:48.920 --> 01:02:54.260 align:middle line:84% And the IRS can pretty easily check my income tax. 01:02:54.260 --> 01:03:00.050 align:middle line:84% For a complicated world where a firm has got thousands 01:03:00.050 --> 01:03:05.090 align:middle line:84% of clients and hundreds of suppliers, doing all this stuff 01:03:05.090 --> 01:03:08.240 align:middle line:90% can be surprisingly tricky. 01:03:08.240 --> 01:03:12.460 align:middle line:90% So someone mentioned automating. 01:03:12.460 --> 01:03:16.265 align:middle line:84% What is the impact of digital payments or automating stuff? 01:03:16.265 --> 01:03:17.890 align:middle line:84% I think actually understanding all that 01:03:17.890 --> 01:03:22.570 align:middle line:84% and how that implies what governments can tax as firms 01:03:22.570 --> 01:03:24.160 align:middle line:90% are developing and becoming-- 01:03:24.160 --> 01:03:26.368 align:middle line:84% countries are developing and becoming more electronic 01:03:26.368 --> 01:03:29.065 align:middle line:84% or whatever is an interesting area to think about. 01:03:29.065 --> 01:03:31.157 align:middle line:90% 01:03:31.157 --> 01:03:31.990 align:middle line:90% I had another thing. 01:03:31.990 --> 01:03:32.990 align:middle line:90% What was I going to say? 01:03:32.990 --> 01:03:40.020 align:middle line:90% 01:03:40.020 --> 01:03:41.190 align:middle line:90% Anyway, OK. 01:03:41.190 --> 01:03:41.880 align:middle line:90% All right. 01:03:41.880 --> 01:03:45.837 align:middle line:84% So what she talks about in the paper 01:03:45.837 --> 01:03:48.420 align:middle line:84% is that there's apparently two different kinds of tax evasion. 01:03:48.420 --> 01:03:50.850 align:middle line:84% She calls them collusive evasion and unilateral evasion. 01:03:50.850 --> 01:03:57.070 align:middle line:84% So collusive evasion is we omit an entire transection 01:03:57.070 --> 01:04:01.150 align:middle line:84% completely, whereas unilateral invasion is we just 01:04:01.150 --> 01:04:02.920 align:middle line:90% disagree on these amounts. 01:04:02.920 --> 01:04:04.180 align:middle line:90% OK? 01:04:04.180 --> 01:04:06.040 align:middle line:84% And she also is going to try to understand 01:04:06.040 --> 01:04:11.797 align:middle line:90% what the impacts of this are. 01:04:11.797 --> 01:04:13.380 align:middle line:84% I think after I went through all this. 01:04:13.380 --> 01:04:16.470 align:middle line:84% So she argues that basically, what happens 01:04:16.470 --> 01:04:21.850 align:middle line:90% if I shock this firm over here? 01:04:21.850 --> 01:04:27.240 align:middle line:84% Well, if I was just-- and I make sure this firm becomes honest-- 01:04:27.240 --> 01:04:33.570 align:middle line:84% well, in collusive evasion, all of a sudden, 01:04:33.570 --> 01:04:36.640 align:middle line:84% that whole transaction appears on the books. 01:04:36.640 --> 01:04:37.510 align:middle line:90% Yeah, so OK. 01:04:37.510 --> 01:04:40.060 align:middle line:84% So what happens, basically-- inclusive evasion-- 01:04:40.060 --> 01:04:49.090 align:middle line:84% when I shock this firm, that new thing goes up. 01:04:49.090 --> 01:04:52.330 align:middle line:84% And this firm can actually start reporting it as an input 01:04:52.330 --> 01:04:55.120 align:middle line:90% and maybe reduce their VAT. 01:04:55.120 --> 01:04:59.740 align:middle line:84% Over here, what happens is they all 01:04:59.740 --> 01:05:01.900 align:middle line:90% have to move towards agreement. 01:05:01.900 --> 01:05:05.363 align:middle line:84% And so those could be two different effects. 01:05:05.363 --> 01:05:07.030 align:middle line:84% What I want to talk about a little more, 01:05:07.030 --> 01:05:08.770 align:middle line:84% though, is what she actually does 01:05:08.770 --> 01:05:10.260 align:middle line:90% in terms of the experiments. 01:05:10.260 --> 01:05:11.635 align:middle line:84% So she does two different things. 01:05:11.635 --> 01:05:14.210 align:middle line:90% 01:05:14.210 --> 01:05:24.700 align:middle line:84% The first thing she does is she does a spillover experiment, 01:05:24.700 --> 01:05:27.390 align:middle line:84% which is to say she tells these firms that they're going 01:05:27.390 --> 01:05:33.600 align:middle line:90% to be suspected of tax evasion. 01:05:33.600 --> 01:05:38.490 align:middle line:84% They're told they're going to be audited. 01:05:38.490 --> 01:05:40.170 align:middle line:90% They then go and audit them. 01:05:40.170 --> 01:05:43.680 align:middle line:84% And they see what happened, what actually happened. 01:05:43.680 --> 01:05:45.960 align:middle line:84% And the thing I want to note here 01:05:45.960 --> 01:05:49.680 align:middle line:84% is this audit is the thing required to actually 01:05:49.680 --> 01:05:50.970 align:middle line:90% go and match everything up. 01:05:50.970 --> 01:05:53.310 align:middle line:84% So she only knows what happens to the upstream 01:05:53.310 --> 01:05:55.560 align:middle line:84% and the downstream people in the audit group 01:05:55.560 --> 01:05:58.800 align:middle line:84% because doing that whole matching actually is manual 01:05:58.800 --> 01:06:03.240 align:middle line:84% and requires the audit to actually take place, OK? 01:06:03.240 --> 01:06:06.930 align:middle line:84% In the deterrence experiment, she basically 01:06:06.930 --> 01:06:09.070 align:middle line:84% sends some random fraction of people 01:06:09.070 --> 01:06:11.070 align:middle line:84% in different parts of the chain a letter saying, 01:06:11.070 --> 01:06:14.430 align:middle line:84% you're likely to be audited, and looks at their responsiveness. 01:06:14.430 --> 01:06:15.270 align:middle line:90% OK? 01:06:15.270 --> 01:06:18.840 align:middle line:84% So these experiments are doing two different things. 01:06:18.840 --> 01:06:21.390 align:middle line:84% And so this one, I think, is pretty clear, right? 01:06:21.390 --> 01:06:23.610 align:middle line:84% This is basically testing for the VAT chain. 01:06:23.610 --> 01:06:25.440 align:middle line:84% This one, I think, has the cleanest test. 01:06:25.440 --> 01:06:29.160 align:middle line:84% This one basically says if this VAT chain thing is actually 01:06:29.160 --> 01:06:33.120 align:middle line:84% happening and I tell you all of a sudden you're 01:06:33.120 --> 01:06:35.190 align:middle line:84% going to be audited, the question 01:06:35.190 --> 01:06:36.810 align:middle line:84% is not just what your response is. 01:06:36.810 --> 01:06:42.810 align:middle line:84% But does that feed through to your upstream and downstream 01:06:42.810 --> 01:06:46.270 align:middle line:90% people through this VAT chain? 01:06:46.270 --> 01:06:51.310 align:middle line:84% The second piece, the deterrence piece, is a little less clear, 01:06:51.310 --> 01:06:52.030 align:middle line:90% I think. 01:06:52.030 --> 01:06:54.690 align:middle line:84% In the deterrence piece, she basically says, look. 01:06:54.690 --> 01:07:00.030 align:middle line:84% In the VAT world in general, where should evasion 01:07:00.030 --> 01:07:01.980 align:middle line:90% be highest, right? 01:07:01.980 --> 01:07:03.990 align:middle line:84% We think evasion should be highest down here 01:07:03.990 --> 01:07:06.180 align:middle line:90% at the retail level. 01:07:06.180 --> 01:07:09.990 align:middle line:84% So if you believe that what's happening in the deterrence 01:07:09.990 --> 01:07:13.530 align:middle line:84% experiment is that it moves you from whatever you were doing 01:07:13.530 --> 01:07:17.950 align:middle line:84% now to no cheating, then you expect a larger responsiveness 01:07:17.950 --> 01:07:18.450 align:middle line:90% down here. 01:07:18.450 --> 01:07:20.400 align:middle line:84% Because there was more cheating to begin with. 01:07:20.400 --> 01:07:23.670 align:middle line:84% But I think, as one of you pointed out in your comments, 01:07:23.670 --> 01:07:25.380 align:middle line:84% this one is a little more complicated 01:07:25.380 --> 01:07:33.730 align:middle line:84% because you could also have the view of maybe the auditor can't 01:07:33.730 --> 01:07:36.460 align:middle line:84% figure it out down here because there's way of figuring out 01:07:36.460 --> 01:07:38.060 align:middle line:90% what the truth is over here. 01:07:38.060 --> 01:07:40.750 align:middle line:84% So I think this one, the deterrence one 01:07:40.750 --> 01:07:43.240 align:middle line:84% is only interpretable if you think of the deterrence 01:07:43.240 --> 01:07:48.010 align:middle line:84% as moving from whatever you were doing before to the truth. 01:07:48.010 --> 01:07:50.020 align:middle line:84% Then I think her interpretation of using 01:07:50.020 --> 01:07:55.330 align:middle line:84% that to measure differential amounts of cheating is right. 01:07:55.330 --> 01:07:58.547 align:middle line:84% If you say, well, it's not clear what 01:07:58.547 --> 01:08:00.130 align:middle line:84% we're learning from the audit unless-- 01:08:00.130 --> 01:08:05.480 align:middle line:90% 01:08:05.480 --> 01:08:09.350 align:middle line:84% because that is related to where you are in the VAT chain, 01:08:09.350 --> 01:08:10.640 align:middle line:90% it's a little bit less clear. 01:08:10.640 --> 01:08:14.662 align:middle line:90% 01:08:14.662 --> 01:08:16.620 align:middle line:84% She's trying to infer something about the level 01:08:16.620 --> 01:08:18.960 align:middle line:84% from-- when she measures the derivative, 01:08:18.960 --> 01:08:23.490 align:middle line:84% she measures what is D taxes, D audit for the three 01:08:23.490 --> 01:08:24.450 align:middle line:90% different places. 01:08:24.450 --> 01:08:26.580 align:middle line:84% She's trying to infer something about the level 01:08:26.580 --> 01:08:29.040 align:middle line:84% of preperiod evasion from that derivative. 01:08:29.040 --> 01:08:31.770 align:middle line:84% And I guess the point I'm making is that only is valid 01:08:31.770 --> 01:08:35.160 align:middle line:84% if you basically assume that the shock is getting you 01:08:35.160 --> 01:08:36.960 align:middle line:84% to the same point in all those places. 01:08:36.960 --> 01:08:39.240 align:middle line:84% And I think that's not totally obvious. 01:08:39.240 --> 01:08:41.189 align:middle line:84% So that's why I like the spillover one better. 01:08:41.189 --> 01:08:42.803 align:middle line:84% I think the spillover one's cleaner. 01:08:42.803 --> 01:08:44.470 align:middle line:84% Questions about the experimental design? 01:08:44.470 --> 01:08:47.632 align:middle line:90% 01:08:47.632 --> 01:08:49.080 align:middle line:90% OK. 01:08:49.080 --> 01:08:51.680 align:middle line:90% So what does she find? 01:08:51.680 --> 01:08:53.310 align:middle line:90% Oops. 01:08:53.310 --> 01:08:57.330 align:middle line:84% She finds-- this is the spillover stuff. 01:08:57.330 --> 01:09:07.399 align:middle line:84% So this says do you pay more VAT if you're 01:09:07.399 --> 01:09:08.569 align:middle line:90% going to be audited, right? 01:09:08.569 --> 01:09:09.740 align:middle line:90% Then yes, you do. 01:09:09.740 --> 01:09:11.450 align:middle line:90% OK. 01:09:11.450 --> 01:09:13.680 align:middle line:84% That just says that's the own effect. 01:09:13.680 --> 01:09:18.870 align:middle line:84% And the question is are there upstream or downstream effects? 01:09:18.870 --> 01:09:22.100 align:middle line:84% And the answer is there are pretty strong upstream effects, 01:09:22.100 --> 01:09:22.600 align:middle line:90% OK? 01:09:22.600 --> 01:09:26.760 align:middle line:84% So what this basically says is I audit these people over here, 01:09:26.760 --> 01:09:27.630 align:middle line:90% right? 01:09:27.630 --> 01:09:31.439 align:middle line:84% And these people also pay more taxes. 01:09:31.439 --> 01:09:32.310 align:middle line:90% OK? 01:09:32.310 --> 01:09:33.550 align:middle line:90% And that's it. 01:09:33.550 --> 01:09:37.490 align:middle line:84% And I think that's the key point. 01:09:37.490 --> 01:09:39.660 align:middle line:84% Why-- the retail people not pay more taxes? 01:09:39.660 --> 01:09:42.630 align:middle line:84% Well, because they have a lot more flexibility 01:09:42.630 --> 01:09:46.410 align:middle line:90% at the bottom end, right? 01:09:46.410 --> 01:09:47.880 align:middle line:84% So people who are downstream, they 01:09:47.880 --> 01:09:51.302 align:middle line:84% can adjust in the three chain version. 01:09:51.302 --> 01:09:53.010 align:middle line:84% In the real world, it's more complicated. 01:09:53.010 --> 01:09:55.438 align:middle line:84% But certainly, the more downstream you are-- 01:09:55.438 --> 01:09:57.480 align:middle line:84% the retail people are the ones that have the most 01:09:57.480 --> 01:10:00.900 align:middle line:90% ability to evade things. 01:10:00.900 --> 01:10:05.090 align:middle line:84% So that's the key point over here 01:10:05.090 --> 01:10:07.580 align:middle line:84% is that it seems to spill over upstream. 01:10:07.580 --> 01:10:10.280 align:middle line:84% And I think that's the best evidence that we have 01:10:10.280 --> 01:10:14.330 align:middle line:84% that actually, these chains seem to be mattering. 01:10:14.330 --> 01:10:17.150 align:middle line:84% In the deterrence paper-- this is too small, sorry. 01:10:17.150 --> 01:10:19.675 align:middle line:84% So let me zoom in on it since you don't have the handout. 01:10:19.675 --> 01:10:26.670 align:middle line:90% 01:10:26.670 --> 01:10:30.312 align:middle line:84% The key point here is just a heterogeneous treatment effect. 01:10:30.312 --> 01:10:31.770 align:middle line:84% So where does the deterrence effect 01:10:31.770 --> 01:10:32.882 align:middle line:90% have the largest effects? 01:10:32.882 --> 01:10:34.590 align:middle line:84% It has the largest effects for the people 01:10:34.590 --> 01:10:36.660 align:middle line:90% who are more downstream. 01:10:36.660 --> 01:10:39.060 align:middle line:84% So that basically is arguing that there was more 01:10:39.060 --> 01:10:40.830 align:middle line:84% cheating happening-- under the assumption 01:10:40.830 --> 01:10:43.080 align:middle line:84% that you can learn about the level from the derivative 01:10:43.080 --> 01:10:46.200 align:middle line:84% that I talked about before, she's arguing there's more 01:10:46.200 --> 01:10:47.280 align:middle line:90% cheating down there. 01:10:47.280 --> 01:10:50.360 align:middle line:90% 01:10:50.360 --> 01:10:54.560 align:middle line:84% OK, that's what I wanted to say about this paper. 01:10:54.560 --> 01:10:59.350 align:middle line:84% But do you have any comments or questions from having read it? 01:10:59.350 --> 01:11:03.180 align:middle line:84% [? Kadesh, ?] you look like you have formulated a question. 01:11:03.180 --> 01:11:06.740 align:middle line:84% AUDIENCE: I just wanted to [INAUDIBLE] that the deterrence 01:11:06.740 --> 01:11:09.986 align:middle line:84% experiment-- do you think that was designed 01:11:09.986 --> 01:11:11.745 align:middle line:90% with this question in mind? 01:11:11.745 --> 01:11:13.870 align:middle line:84% Or do you think there was-- they learned something. 01:11:13.870 --> 01:11:19.990 align:middle line:84% And then was like, oh, maybe we should test the [INAUDIBLE]?? 01:11:19.990 --> 01:11:21.700 align:middle line:90% BEN OLKEN: Oh, did they go in-- 01:11:21.700 --> 01:11:23.620 align:middle line:90% did they go into this-- 01:11:23.620 --> 01:11:28.290 align:middle line:84% AUDIENCE: With that test, with this theory in mind. 01:11:28.290 --> 01:11:31.383 align:middle line:84% BEN OLKEN: I can't speak to what she was thinking when 01:11:31.383 --> 01:11:32.550 align:middle line:90% she designed the experiment. 01:11:32.550 --> 01:11:33.530 align:middle line:90% But I suspect they did. 01:11:33.530 --> 01:11:36.135 align:middle line:84% I think this is the key question in VAT 01:11:36.135 --> 01:11:37.760 align:middle line:84% those people were trying to understand, 01:11:37.760 --> 01:11:39.052 align:middle line:90% is understanding the spillover. 01:11:39.052 --> 01:11:43.160 align:middle line:84% So I think they were trying to go into it with the two things. 01:11:43.160 --> 01:11:45.260 align:middle line:84% I think the reason for the two experiments 01:11:45.260 --> 01:11:47.270 align:middle line:90% is the deterrence experiment-- 01:11:47.270 --> 01:11:49.220 align:middle line:90% she can measure the outcomes. 01:11:49.220 --> 01:11:54.050 align:middle line:84% She doesn't need to know who the suppliers and the clients are. 01:11:54.050 --> 01:11:55.050 align:middle line:90% You see what I'm saying? 01:11:55.050 --> 01:11:58.620 align:middle line:84% So the advantage of the deterrence experiment 01:11:58.620 --> 01:12:00.360 align:middle line:90% is you get this enormous sample. 01:12:00.360 --> 01:12:03.358 align:middle line:84% And you don't need to know the whole VAT chain. 01:12:03.358 --> 01:12:05.400 align:middle line:84% So I think there, she can try to say, well, look. 01:12:05.400 --> 01:12:06.817 align:middle line:84% I'm trying to understand something 01:12:06.817 --> 01:12:07.860 align:middle line:90% in a really big sample. 01:12:07.860 --> 01:12:09.300 align:middle line:90% What can I possibly learn? 01:12:09.300 --> 01:12:13.440 align:middle line:84% And she doesn't know who the suppliers and the clients are. 01:12:13.440 --> 01:12:15.150 align:middle line:84% The reason for the audit experiment 01:12:15.150 --> 01:12:16.983 align:middle line:84% and why she had to do that on a small sample 01:12:16.983 --> 01:12:22.360 align:middle line:84% is she only can learn who the suppliers and the clients 01:12:22.360 --> 01:12:24.700 align:middle line:84% are once they do that manual audit process. 01:12:24.700 --> 01:12:27.117 align:middle line:84% At least in this context, it's not sufficiently automated. 01:12:27.117 --> 01:12:28.900 align:middle line:84% So if you had a fully automated system, 01:12:28.900 --> 01:12:31.420 align:middle line:84% you could run the spillovers in this experiment, too. 01:12:31.420 --> 01:12:34.870 align:middle line:84% But she can't do that because they don't have that data. 01:12:34.870 --> 01:12:36.550 align:middle line:84% AUDIENCE: So the deterrence thing 01:12:36.550 --> 01:12:40.760 align:middle line:84% is just adding this kind of [INAUDIBLE] validity? 01:12:40.760 --> 01:12:41.990 align:middle line:90% BEN OLKEN: Yeah. 01:12:41.990 --> 01:12:44.520 align:middle line:84% I think it's adding a different piece. 01:12:44.520 --> 01:12:46.770 align:middle line:84% Under the assumption that I was saying before that you 01:12:46.770 --> 01:12:49.103 align:middle line:84% can learn about the preperiod level from the derivative, 01:12:49.103 --> 01:12:51.480 align:middle line:84% it is testing a different aspect of this question also, 01:12:51.480 --> 01:12:52.450 align:middle line:90% which is to say, look. 01:12:52.450 --> 01:12:54.117 align:middle line:84% There's two questions you might believe. 01:12:54.117 --> 01:12:56.105 align:middle line:90% One is you'd say-- 01:12:56.105 --> 01:12:58.230 align:middle line:84% two questions about the VAT from the simple theory. 01:12:58.230 --> 01:12:59.100 align:middle line:90% One says, look. 01:12:59.100 --> 01:13:01.630 align:middle line:90% This thing spills over. 01:13:01.630 --> 01:13:03.630 align:middle line:84% The other says there should be way more cheating 01:13:03.630 --> 01:13:06.090 align:middle line:84% at the very bottom, at the very downstream. 01:13:06.090 --> 01:13:11.618 align:middle line:84% And so she's arguing that by learning about this derivative, 01:13:11.618 --> 01:13:14.160 align:middle line:84% you're learning that maybe there is much more cheating there. 01:13:14.160 --> 01:13:15.868 align:middle line:84% It's certainly consistent with that idea. 01:13:15.868 --> 01:13:19.920 align:middle line:90% 01:13:19.920 --> 01:13:20.790 align:middle line:90% Other questions? 01:13:20.790 --> 01:13:21.345 align:middle line:90% Yeah, Ahmed. 01:13:21.345 --> 01:13:23.220 align:middle line:84% AUDIENCE: Signing returns [? to the letter ?] 01:13:23.220 --> 01:13:25.710 align:middle line:90% is quite cheap, right? 01:13:25.710 --> 01:13:29.705 align:middle line:84% It does feel like there's a simple way of increasing 01:13:29.705 --> 01:13:32.870 align:middle line:90% her complexity on this-- 01:13:32.870 --> 01:13:34.620 align:middle line:84% BEN OLKEN: Well, the question is what is-- 01:13:34.620 --> 01:13:38.316 align:middle line:90% 01:13:38.316 --> 01:13:40.130 align:middle line:90% you can't deter people-- 01:13:40.130 --> 01:13:41.880 align:middle line:84% you have to have rational-- you ultimately 01:13:41.880 --> 01:13:43.770 align:middle line:84% have to deal with rational expectations. 01:13:43.770 --> 01:13:45.670 align:middle line:84% So you can't tell everyone they're 01:13:45.670 --> 01:13:47.420 align:middle line:84% going to be audited with 100% probability, 01:13:47.420 --> 01:13:49.130 align:middle line:84% and not audited with 100% probability. 01:13:49.130 --> 01:13:50.510 align:middle line:90% Eventually, they'll it out. 01:13:50.510 --> 01:13:53.760 align:middle line:84% AUDIENCE: But for the short term, it's free money. 01:13:53.760 --> 01:13:54.260 align:middle line:90% [INAUDIBLE] 01:13:54.260 --> 01:13:56.660 align:middle line:84% BEN OLKEN: But in the long run, it has a lot of-- 01:13:56.660 --> 01:14:01.100 align:middle line:84% I think this was worded in a way to be honest and truthful. 01:14:01.100 --> 01:14:03.260 align:middle line:90% But I think that you have to-- 01:14:03.260 --> 01:14:05.390 align:middle line:90% and that suggests that's true. 01:14:05.390 --> 01:14:07.807 align:middle line:84% But it's not a free money machine. 01:14:07.807 --> 01:14:10.140 align:middle line:84% I think you probably have to actually increase the audit 01:14:10.140 --> 01:14:11.940 align:middle line:90% rate, too, ultimately. 01:14:11.940 --> 01:14:16.140 align:middle line:84% But there is this basic fact that many governments are not 01:14:16.140 --> 01:14:21.190 align:middle line:84% spending enough resources on enforcement, which 01:14:21.190 --> 01:14:24.002 align:middle line:84% I will argue in the next paper, which I'll probably teach you 01:14:24.002 --> 01:14:25.960 align:middle line:84% at beginning next class, at least in Indonesia, 01:14:25.960 --> 01:14:29.650 align:middle line:84% and is actually a very active debate in the US right now. 01:14:29.650 --> 01:14:32.260 align:middle line:84% Should we spend more money on tax enforcement? 01:14:32.260 --> 01:14:36.190 align:middle line:84% And many estimates conclude that governments are not 01:14:36.190 --> 01:14:38.710 align:middle line:84% spending nearly enough money on tax enforcement. 01:14:38.710 --> 01:14:43.500 align:middle line:84% It's a very, very high return activity. 01:14:43.500 --> 01:14:44.850 align:middle line:90% Yeah? 01:14:44.850 --> 01:14:47.932 align:middle line:84% AUDIENCE: In a case where a supplier and a buyer 01:14:47.932 --> 01:14:51.380 align:middle line:84% are bargaining over the tax, why do we expect it-- 01:14:51.380 --> 01:14:53.760 align:middle line:90% why do you expect them to-- 01:14:53.760 --> 01:14:56.014 align:middle line:84% for a solution to marketing game to be 01:14:56.014 --> 01:14:58.460 align:middle line:84% the true amount, the correct amount 01:14:58.460 --> 01:15:01.870 align:middle line:84% versus just some solution to some marketing-- 01:15:01.870 --> 01:15:04.236 align:middle line:84% BEN OLKEN: Well, the tax code doesn't care. 01:15:04.236 --> 01:15:05.490 align:middle line:90% AUDIENCE: Oh. 01:15:05.490 --> 01:15:06.420 align:middle line:90% Why is that? 01:15:06.420 --> 01:15:12.025 align:middle line:84% BEN OLKEN: Well, imagine I move this line over here. 01:15:12.025 --> 01:15:13.650 align:middle line:84% As long as they agree, the government's 01:15:13.650 --> 01:15:16.578 align:middle line:84% still getting taxed on this whole amount. 01:15:16.578 --> 01:15:19.367 align:middle line:84% AUDIENCE: So they're probably going to say [INAUDIBLE].. 01:15:19.367 --> 01:15:21.200 align:middle line:84% BEN OLKEN: For as long as they agree on what 01:15:21.200 --> 01:15:23.900 align:middle line:90% they're going to report-- 01:15:23.900 --> 01:15:26.330 align:middle line:90% that's the beauty of the VAT. 01:15:26.330 --> 01:15:28.850 align:middle line:90% If we're poor-- if we say-- 01:15:28.850 --> 01:15:36.445 align:middle line:84% for everyone except the bottom layer, my sales to you 01:15:36.445 --> 01:15:37.820 align:middle line:84% are how much-- what I sell to you 01:15:37.820 --> 01:15:39.360 align:middle line:90% is how much you buy from me. 01:15:39.360 --> 01:15:42.800 align:middle line:84% So if I say I'm selling to you at a lower price, 01:15:42.800 --> 01:15:44.060 align:middle line:90% I pay less tax. 01:15:44.060 --> 01:15:47.330 align:middle line:84% But you pay more tax because your inputs go down. 01:15:47.330 --> 01:15:50.520 align:middle line:84% So at some level, the government doesn't care. 01:15:50.520 --> 01:15:51.520 align:middle line:90% You see what I'm saying? 01:15:51.520 --> 01:15:53.970 align:middle line:84% AUDIENCE: It was meant for firm size stuff, right? 01:15:53.970 --> 01:15:56.428 align:middle line:84% BEN OLKEN: Yes, it might have implications for the economy. 01:15:56.428 --> 01:15:59.363 align:middle line:84% I think also, I think maybe for the deterrence reasons, 01:15:59.363 --> 01:16:00.780 align:middle line:84% maybe for the denouncement reasons 01:16:00.780 --> 01:16:11.623 align:middle line:84% we talked about before, there's a focal point on the truth. 01:16:11.623 --> 01:16:13.290 align:middle line:84% It is an interesting question, actually. 01:16:13.290 --> 01:16:15.210 align:middle line:90% There may be some-- 01:16:15.210 --> 01:16:17.340 align:middle line:84% I suspect there's some good microtheory bargaining 01:16:17.340 --> 01:16:19.375 align:middle line:84% answer that I don't know the answer to 01:16:19.375 --> 01:16:20.500 align:middle line:90% for why that might be true. 01:16:20.500 --> 01:16:23.190 align:middle line:84% But I don't know it off the top of my head. 01:16:23.190 --> 01:16:23.725 align:middle line:90% Yeah. 01:16:23.725 --> 01:16:25.350 align:middle line:84% AUDIENCE: Is there any real literature, 01:16:25.350 --> 01:16:28.757 align:middle line:84% if I'm really into this kind of systems, to organized crime-- 01:16:28.757 --> 01:16:29.840 align:middle line:90% BEN OLKEN: So what, sorry? 01:16:29.840 --> 01:16:33.710 align:middle line:84% AUDIENCE: To organized crime and money laundering? 01:16:33.710 --> 01:16:35.030 align:middle line:90% BEN OLKEN: Tell me more. 01:16:35.030 --> 01:16:37.430 align:middle line:84% AUDIENCE: So the general idea is that maybe you're 01:16:37.430 --> 01:16:38.210 align:middle line:90% organized crime. 01:16:38.210 --> 01:16:44.427 align:middle line:84% You use small retail stores to just laundering money 01:16:44.427 --> 01:16:45.010 align:middle line:90% through there. 01:16:45.010 --> 01:16:48.430 align:middle line:84% And then if you set up a system where 01:16:48.430 --> 01:16:51.470 align:middle line:84% you have double reporting like this one-- 01:16:51.470 --> 01:16:52.720 align:middle line:90% BEN OLKEN: You make it harder. 01:16:52.720 --> 01:16:54.165 align:middle line:84% AUDIENCE: It would make it harder to money-- 01:16:54.165 --> 01:16:54.510 align:middle line:90% uh, yeah. 01:16:54.510 --> 01:16:54.820 align:middle line:90% BEN OLKEN: Yeah. 01:16:54.820 --> 01:16:55.320 align:middle line:90% So sorry. 01:16:55.320 --> 01:16:58.370 align:middle line:84% So why is money laundering relevant here? 01:16:58.370 --> 01:17:01.893 align:middle line:84% So money laundering is almost the inverse problem, right? 01:17:01.893 --> 01:17:04.060 align:middle line:84% The money laundering problem is the inverse problem. 01:17:04.060 --> 01:17:05.518 align:middle line:84% I have a bunch of cash it's illegal 01:17:05.518 --> 01:17:08.950 align:middle line:90% that I want to make look legal. 01:17:08.950 --> 01:17:11.620 align:middle line:84% And so the way that they do that is you basically, 01:17:11.620 --> 01:17:15.040 align:middle line:84% you have a business that has a lot of where the actual income 01:17:15.040 --> 01:17:17.080 align:middle line:90% level is not reportable. 01:17:17.080 --> 01:17:19.300 align:middle line:84% And you overstate your income, essentially. 01:17:19.300 --> 01:17:23.260 align:middle line:90% So you buy a cash business. 01:17:23.260 --> 01:17:25.180 align:middle line:84% The money launderer runs a cash business 01:17:25.180 --> 01:17:27.520 align:middle line:84% and overstates the amount of actual business going 01:17:27.520 --> 01:17:31.900 align:middle line:84% through as a way of mixing the cash they want to make legal 01:17:31.900 --> 01:17:33.460 align:middle line:90% with the legal cash. 01:17:33.460 --> 01:17:36.250 align:middle line:84% And because the government can't observe the retail side, 01:17:36.250 --> 01:17:37.730 align:middle line:90% you can overdo it. 01:17:37.730 --> 01:17:39.230 align:middle line:84% That's the idea of money laundering. 01:17:39.230 --> 01:17:40.660 align:middle line:84% It's the inverse problem of this. 01:17:40.660 --> 01:17:42.967 align:middle line:84% People want to pay more taxes, right? 01:17:42.967 --> 01:17:44.800 align:middle line:84% So yes, there could be implications to that. 01:17:44.800 --> 01:17:47.110 align:middle line:90% I haven't thought them through. 01:17:47.110 --> 01:17:49.390 align:middle line:84% I think as long as-- that's why the money laundering 01:17:49.390 --> 01:17:52.412 align:middle line:90% guys work at the retail level. 01:17:52.412 --> 01:17:54.370 align:middle line:84% It's that same point that the government really 01:17:54.370 --> 01:17:56.740 align:middle line:84% doesn't know what's happening at that very last piece. 01:17:56.740 --> 01:17:59.080 align:middle line:84% And that actually leads me to the very last thing 01:17:59.080 --> 01:18:01.028 align:middle line:90% I want to say on this topic. 01:18:01.028 --> 01:18:01.570 align:middle line:90% Actually, no. 01:18:01.570 --> 01:18:03.778 align:middle line:84% Sorry there's two things I want to say on this topic. 01:18:03.778 --> 01:18:06.820 align:middle line:84% Number one is one cool thing about the VAT 01:18:06.820 --> 01:18:08.890 align:middle line:84% is that in principle, governments 01:18:08.890 --> 01:18:11.350 align:middle line:84% are collecting huge amounts of data 01:18:11.350 --> 01:18:14.830 align:middle line:84% on who's trading with who, at least in the formal network. 01:18:14.830 --> 01:18:17.560 align:middle line:84% And I think a bunch of researchers have realized this 01:18:17.560 --> 01:18:20.980 align:middle line:84% and that this is a cool data set, very hard 01:18:20.980 --> 01:18:23.415 align:middle line:84% to obtain because it's tax data-- but a cool data set 01:18:23.415 --> 01:18:24.790 align:middle line:84% in principle on the whole network 01:18:24.790 --> 01:18:27.530 align:middle line:84% of supplier-supplier supply chain relationships. 01:18:27.530 --> 01:18:31.122 align:middle line:84% And I think Dave Donaldson and Dina Pomeranz 01:18:31.122 --> 01:18:33.580 align:middle line:84% and several co-authors actually have paper using this data, 01:18:33.580 --> 01:18:35.038 align:middle line:84% as do, I think, a few other people. 01:18:35.038 --> 01:18:37.880 align:middle line:84% So it's an exciting area people are starting to think through. 01:18:37.880 --> 01:18:40.190 align:middle line:84% And the second thing I wanted to mention is-- 01:18:40.190 --> 01:18:41.690 align:middle line:84% the last thing I would say is, well, 01:18:41.690 --> 01:18:44.630 align:middle line:84% what can you do at the very end of the chain? 01:18:44.630 --> 01:18:48.250 align:middle line:84% And so Joanna Naritomi has a paper on this, which basically 01:18:48.250 --> 01:18:51.845 align:middle line:84% says can we find a way to incentivize people 01:18:51.845 --> 01:18:53.720 align:middle line:84% at the very bottom of the chain to be honest? 01:18:53.720 --> 01:18:57.100 align:middle line:84% And I'll just note, if you've ever been to a store 01:18:57.100 --> 01:19:00.430 align:middle line:84% where it says if we don't give you a receipt, 01:19:00.430 --> 01:19:03.570 align:middle line:84% your purchase is free, what are they doing there? 01:19:03.570 --> 01:19:04.570 align:middle line:90% Why are they doing that? 01:19:04.570 --> 01:19:05.680 align:middle line:90% It's the same idea. 01:19:05.680 --> 01:19:07.720 align:middle line:84% They're trying to incentivize the consumer 01:19:07.720 --> 01:19:10.630 align:middle line:84% to make sure that the transection is reported 01:19:10.630 --> 01:19:12.700 align:middle line:90% through the cash machine, right? 01:19:12.700 --> 01:19:15.610 align:middle line:84% In that case, usually, it's about the employer worrying 01:19:15.610 --> 01:19:19.600 align:middle line:84% about the clerk stealing the resources, which is why 01:19:19.600 --> 01:19:20.930 align:middle line:90% they want it to be reported. 01:19:20.930 --> 01:19:22.840 align:middle line:84% But the same idea is, how can we incentivize 01:19:22.840 --> 01:19:27.430 align:middle line:84% the very retail consumer to make sure things are properly 01:19:27.430 --> 01:19:28.430 align:middle line:90% reported? 01:19:28.430 --> 01:19:31.150 align:middle line:84% And so she studies a policy in Brazil, 01:19:31.150 --> 01:19:35.800 align:middle line:84% where the VAT receipt issued at the final thing 01:19:35.800 --> 01:19:37.190 align:middle line:90% was also a lottery ticket. 01:19:37.190 --> 01:19:37.690 align:middle line:90% OK? 01:19:37.690 --> 01:19:43.210 align:middle line:84% So for every dollar of actual legally reported VAT, 01:19:43.210 --> 01:19:44.600 align:middle line:90% you get to enter some lottery. 01:19:44.600 --> 01:19:46.420 align:middle line:84% And the government gives you prizes, OK? 01:19:46.420 --> 01:19:48.160 align:middle line:90% And it's very small. 01:19:48.160 --> 01:19:53.410 align:middle line:84% So the value is 2% of the total purchase price. 01:19:53.410 --> 01:19:55.712 align:middle line:84% So I don't know what the VAT rate in Brazil 01:19:55.712 --> 01:19:56.920 align:middle line:90% is, but let's say it was 10%. 01:19:56.920 --> 01:20:00.000 align:middle line:90% 01:20:00.000 --> 01:20:02.230 align:middle line:84% And even if the retail sector is, I don't know, 30%, 01:20:02.230 --> 01:20:04.990 align:middle line:84% it's still a small share of the total VAT 01:20:04.990 --> 01:20:07.330 align:middle line:84% the government is collecting even in this last piece. 01:20:07.330 --> 01:20:09.322 align:middle line:84% But it's at least some reason that if you 01:20:09.322 --> 01:20:10.780 align:middle line:84% were at least totally indifferent-- 01:20:10.780 --> 01:20:13.240 align:middle line:84% you shouldn't be totally indifferent. 01:20:13.240 --> 01:20:15.190 align:middle line:84% And she finds that actually, this 01:20:15.190 --> 01:20:20.620 align:middle line:84% leads to a huge impact, a 24% impact in reported revenues 01:20:20.620 --> 01:20:21.815 align:middle line:90% from the very bottom firms. 01:20:21.815 --> 01:20:23.440 align:middle line:84% And so that suggests that maybe there's 01:20:23.440 --> 01:20:26.230 align:middle line:84% something you can do to incentivize people down here as 01:20:26.230 --> 01:20:27.280 align:middle line:90% well. 01:20:27.280 --> 01:20:28.280 align:middle line:90% OK, I'm out of time. 01:20:28.280 --> 01:20:29.510 align:middle line:90% I'll stop here. 01:20:29.510 --> 01:20:35.350 align:middle line:84% I'm going to finish up by talking about tax 01:20:35.350 --> 01:20:38.490 align:middle line:90% administration next time. 01:20:38.490 --> 01:21:02.000 align:middle line:90%