WEBVTT 00:00:00.000 --> 00:00:01.484 align:middle line:90% [SQUEAKING] 00:00:01.484 --> 00:00:02.960 align:middle line:90% [RUSTLING] 00:00:02.960 --> 00:00:04.928 align:middle line:90% [CLICKING] 00:00:04.928 --> 00:00:10.840 align:middle line:90% 00:00:10.840 --> 00:00:12.360 align:middle line:84% ROBERT M. TOWNSEND: So today, what I 00:00:12.360 --> 00:00:19.920 align:middle line:84% want to do is to tell you about this class and its contents. 00:00:19.920 --> 00:00:22.560 align:middle line:90% So it's 14.129. 00:00:22.560 --> 00:00:28.760 align:middle line:84% And the correct title is Blockchain and the Design 00:00:28.760 --> 00:00:30.800 align:middle line:90% of Financial Systems. 00:00:30.800 --> 00:00:35.160 align:middle line:84% It's incorrectly listed in the registrar's office 00:00:35.160 --> 00:00:38.540 align:middle line:84% as Advanced Contract Theory, which is true, 00:00:38.540 --> 00:00:43.960 align:middle line:84% but we're going to focus on this topic. 00:00:43.960 --> 00:00:45.960 align:middle line:84% And the way I'm going to organize today 00:00:45.960 --> 00:00:51.760 align:middle line:84% is to jump right in and tell you the contents of each 00:00:51.760 --> 00:00:54.840 align:middle line:90% of the lectures. 00:00:54.840 --> 00:00:58.360 align:middle line:84% It's a bit tricky in the sense that I 00:00:58.360 --> 00:01:01.750 align:middle line:84% don't want to actually give each one of those lectures. 00:01:01.750 --> 00:01:04.709 align:middle line:84% That's going to happen as the class evolves, 00:01:04.709 --> 00:01:08.850 align:middle line:84% but I want to give you a good sense of the topics we're 00:01:08.850 --> 00:01:11.770 align:middle line:84% going to be covering and a bit of the way 00:01:11.770 --> 00:01:15.570 align:middle line:84% we're going to be covering it, and in particular, featuring 00:01:15.570 --> 00:01:19.930 align:middle line:84% this blend of computer science and economics. 00:01:19.930 --> 00:01:25.410 align:middle line:84% And after I scroll through a summary of the lectures, 00:01:25.410 --> 00:01:28.410 align:middle line:84% then I will turn to the syllabus, 00:01:28.410 --> 00:01:32.690 align:middle line:84% tell you a review, again, more about the organization 00:01:32.690 --> 00:01:34.210 align:middle line:90% of the class. 00:01:34.210 --> 00:01:39.890 align:middle line:84% And I'll also mention a parallel class and its contents, which 00:01:39.890 --> 00:01:44.210 align:middle line:84% are relevant, but not a requirement. 00:01:44.210 --> 00:01:48.170 align:middle line:84% And that way, you'll see how everything fits together, 00:01:48.170 --> 00:01:50.130 align:middle line:90% ideally. 00:01:50.130 --> 00:01:58.780 align:middle line:84% So this is a synthesis of computer science and economics. 00:01:58.780 --> 00:02:03.460 align:middle line:84% The buzzwords, so to speak, of computer science 00:02:03.460 --> 00:02:06.780 align:middle line:84% will be cryptocurrency, blockchain, tokenization, 00:02:06.780 --> 00:02:12.100 align:middle line:84% platforms, and computational algorithms. 00:02:12.100 --> 00:02:16.060 align:middle line:84% And the buzzwords, so to speak, of economics 00:02:16.060 --> 00:02:19.260 align:middle line:84% are going to be contract theory, mechanism design, 00:02:19.260 --> 00:02:23.260 align:middle line:84% general equilibrium theory, and monetary theory. 00:02:23.260 --> 00:02:28.740 align:middle line:84% So we're going to try to get these items on the same page. 00:02:28.740 --> 00:02:34.900 align:middle line:84% We want to understand topically the assumptions and shortcomings 00:02:34.900 --> 00:02:40.300 align:middle line:84% of three objects-- distributed ledgers, smart contracts, 00:02:40.300 --> 00:02:41.940 align:middle line:90% and encryption-- 00:02:41.940 --> 00:02:45.020 align:middle line:84% and the potential impact that these technologies have 00:02:45.020 --> 00:02:50.420 align:middle line:84% on legacy systems and explore the possibilities 00:02:50.420 --> 00:02:56.740 align:middle line:84% for new financial design and the implications for regulation. 00:02:56.740 --> 00:02:59.440 align:middle line:84% So each lecture, as you shall see, 00:02:59.440 --> 00:03:01.980 align:middle line:84% or at least a cluster of lectures, 00:03:01.980 --> 00:03:07.140 align:middle line:84% features one or several of these computer science technologies 00:03:07.140 --> 00:03:11.060 align:middle line:84% and doing that in the context of historical or contemporary 00:03:11.060 --> 00:03:16.060 align:middle line:84% applications and in that context using 00:03:16.060 --> 00:03:18.940 align:middle line:90% the various economic tools. 00:03:18.940 --> 00:03:22.220 align:middle line:84% So examples that you'll see more momentarily-- 00:03:22.220 --> 00:03:24.460 align:middle line:84% community currencies, coordination 00:03:24.460 --> 00:03:29.340 align:middle line:84% and financial crashes, tokenized assets and trade fails, 00:03:29.340 --> 00:03:33.100 align:middle line:84% multilateral payment and trade credit set-offs, 00:03:33.100 --> 00:03:36.580 align:middle line:84% liquidity injections, and financial contagion, 00:03:36.580 --> 00:03:39.440 align:middle line:84% risk-sharing auctions, and online markets. 00:03:39.440 --> 00:03:42.300 align:middle line:90% 00:03:42.300 --> 00:03:47.060 align:middle line:84% So the goal here is to unpack these different computer science 00:03:47.060 --> 00:03:50.820 align:middle line:84% technologies and their assumptions, 00:03:50.820 --> 00:03:55.470 align:middle line:84% sometimes rearranging them rather than accepting 00:03:55.470 --> 00:04:00.230 align:middle line:90% conventional bundles and labels. 00:04:00.230 --> 00:04:04.110 align:middle line:84% There is a lot of difference of opinion 00:04:04.110 --> 00:04:09.870 align:middle line:84% about these technologies, a lot of hype and exaggeration 00:04:09.870 --> 00:04:14.990 align:middle line:84% on the pro side of what can be done with them, 00:04:14.990 --> 00:04:20.829 align:middle line:84% but also a lot of criticism on the negative side of Bitcoin 00:04:20.829 --> 00:04:23.470 align:middle line:90% and cryptocurrencies and so on. 00:04:23.470 --> 00:04:28.030 align:middle line:84% So we want to set this polarization aside and get 00:04:28.030 --> 00:04:31.070 align:middle line:90% in the middle and be objective. 00:04:31.070 --> 00:04:35.590 align:middle line:84% So for example, we can combine the authentication 00:04:35.590 --> 00:04:38.150 align:middle line:84% and commitment schemes of encryption 00:04:38.150 --> 00:04:42.250 align:middle line:84% with an automated execution engine on a distributed ledger. 00:04:42.250 --> 00:04:45.070 align:middle line:90% That's the standard package. 00:04:45.070 --> 00:04:47.830 align:middle line:84% But we can also do this with computer code 00:04:47.830 --> 00:04:54.480 align:middle line:84% without a consensus algorithm or even use trusted third parties 00:04:54.480 --> 00:04:56.800 align:middle line:90% and escrow accounts. 00:04:56.800 --> 00:05:02.880 align:middle line:84% So we're quite open to and should review alternative ways 00:05:02.880 --> 00:05:05.320 align:middle line:90% of implementation. 00:05:05.320 --> 00:05:08.080 align:middle line:84% Another way to put this-- different technologies 00:05:08.080 --> 00:05:12.940 align:middle line:84% allow different implementations of the central planner, 00:05:12.940 --> 00:05:14.820 align:middle line:90% which is econ jargon. 00:05:14.820 --> 00:05:16.600 align:middle line:90% You'll see more. 00:05:16.600 --> 00:05:19.280 align:middle line:84% That's a key construct of economics. 00:05:19.280 --> 00:05:23.360 align:middle line:84% We can gradually abstract away the centralized role 00:05:23.360 --> 00:05:27.520 align:middle line:84% of the central planner into concretely deployed mechanisms 00:05:27.520 --> 00:05:30.340 align:middle line:84% and into commercial applications. 00:05:30.340 --> 00:05:34.240 align:middle line:84% So we're shooting for this balanced perspective, 00:05:34.240 --> 00:05:39.440 align:middle line:84% avoiding the hype and ideological positions, 00:05:39.440 --> 00:05:43.320 align:middle line:84% focusing instead on ways to implement use cases, 00:05:43.320 --> 00:05:48.680 align:middle line:84% allowing policy designers to leverage 00:05:48.680 --> 00:05:53.850 align:middle line:84% these tools for the different problems that they're facing. 00:05:53.850 --> 00:05:56.270 align:middle line:90% OK, so lecture 1-- 00:05:56.270 --> 00:05:59.410 align:middle line:90% 2-- 1 is today-- 00:05:59.410 --> 00:06:03.730 align:middle line:90% is about the blockchain. 00:06:03.730 --> 00:06:06.570 align:middle line:84% On each of these slides, as I go through lectures 2, 00:06:06.570 --> 00:06:09.670 align:middle line:84% 3, et cetera, you'll see a longer title, 00:06:09.670 --> 00:06:13.210 align:middle line:84% which is more descriptive of the components. 00:06:13.210 --> 00:06:17.610 align:middle line:84% So this lecture 2 is about a unified view of distributed 00:06:17.610 --> 00:06:20.890 align:middle line:84% ledgers and financial accounts, thinking 00:06:20.890 --> 00:06:25.810 align:middle line:84% of them both as databases, distinct views of money, 00:06:25.810 --> 00:06:31.050 align:middle line:84% and also providing a larger community perspective. 00:06:31.050 --> 00:06:33.330 align:middle line:84% So blockchains and financial accounts 00:06:33.330 --> 00:06:38.690 align:middle line:84% are each a database of transactions. 00:06:38.690 --> 00:06:42.690 align:middle line:84% And so we're going to think about it that way. 00:06:42.690 --> 00:06:48.050 align:middle line:84% However, each is associated with a notion of money. 00:06:48.050 --> 00:06:51.450 align:middle line:84% As I've already anticipated, for the blockchain, 00:06:51.450 --> 00:06:55.810 align:middle line:84% we think of Bitcoin and cryptocurrency. 00:06:55.810 --> 00:07:01.330 align:middle line:84% And many people stop thinking about it after that. 00:07:01.330 --> 00:07:07.210 align:middle line:84% For financial accounts, money is a particular item on the balance 00:07:07.210 --> 00:07:09.730 align:middle line:90% sheet of the holder. 00:07:09.730 --> 00:07:14.810 align:middle line:84% For example, Fiat money or demand deposits 00:07:14.810 --> 00:07:16.810 align:middle line:90% in a commercial bank. 00:07:16.810 --> 00:07:21.730 align:middle line:84% So it's also a liability of an issuer, 00:07:21.730 --> 00:07:26.610 align:middle line:84% but the truth is that each of distributed ledgers 00:07:26.610 --> 00:07:30.330 align:middle line:84% and financial accounts are broader and do not 00:07:30.330 --> 00:07:36.190 align:middle line:84% require us to force either one into the construct of money. 00:07:36.190 --> 00:07:38.770 align:middle line:90% 00:07:38.770 --> 00:07:44.090 align:middle line:84% So these evocative pictures are more details 00:07:44.090 --> 00:07:50.060 align:middle line:84% coming next time on Thursday about financial accounts. 00:07:50.060 --> 00:07:53.780 align:middle line:84% And this is actually the blockchain picture of what's 00:07:53.780 --> 00:07:58.540 align:middle line:90% going on with the databases. 00:07:58.540 --> 00:08:03.660 align:middle line:84% So you basically have data that are hashed pairwise, 00:08:03.660 --> 00:08:06.800 align:middle line:84% hashed again, and hashed the third time. 00:08:06.800 --> 00:08:10.740 align:middle line:84% So this is basically an archive system 00:08:10.740 --> 00:08:16.740 align:middle line:84% that allows you to uncover the individual transactions 00:08:16.740 --> 00:08:20.740 align:middle line:90% on, say, Bitcoin. 00:08:20.740 --> 00:08:24.940 align:middle line:84% So block, chain, together-- that's 00:08:24.940 --> 00:08:30.280 align:middle line:84% where the terminology comes from as a database. 00:08:30.280 --> 00:08:30.780 align:middle line:90% OK? 00:08:30.780 --> 00:08:33.460 align:middle line:90% 00:08:33.460 --> 00:08:36.100 align:middle line:84% Both blockchains and financial accounts 00:08:36.100 --> 00:08:39.659 align:middle line:84% can incorporate multiple objects. 00:08:39.659 --> 00:08:44.380 align:middle line:84% And from a monetary perspective, that's fine. 00:08:44.380 --> 00:08:48.270 align:middle line:84% You have goods or assets or certificates 00:08:48.270 --> 00:08:52.510 align:middle line:84% that, in a monetary theory definition, 00:08:52.510 --> 00:08:55.630 align:middle line:90% appear frequently in exchange. 00:08:55.630 --> 00:08:58.450 align:middle line:84% So we have-- you can barely see it-- 00:08:58.450 --> 00:09:03.490 align:middle line:84% a matrix of transactions of grain, clothing, labor, money, 00:09:03.490 --> 00:09:06.310 align:middle line:90% IOUs, and so on. 00:09:06.310 --> 00:09:10.950 align:middle line:84% And we count up how often they are 00:09:10.950 --> 00:09:14.510 align:middle line:84% used as we are, say, looking at an individual household 00:09:14.510 --> 00:09:21.470 align:middle line:84% or aggregating up to a village or the US economy. 00:09:21.470 --> 00:09:24.030 align:middle line:84% So a money is something with a high velocity that 00:09:24.030 --> 00:09:27.510 align:middle line:84% appears frequently in exchange or the amount traded 00:09:27.510 --> 00:09:32.230 align:middle line:90% per unit outstanding is large. 00:09:32.230 --> 00:09:34.190 align:middle line:84% So it's fine to have multiple monies. 00:09:34.190 --> 00:09:36.790 align:middle line:84% In this picture that you can barely see, 00:09:36.790 --> 00:09:39.350 align:middle line:84% you would notice that grain is actually 00:09:39.350 --> 00:09:42.070 align:middle line:90% being used as well as money. 00:09:42.070 --> 00:09:46.160 align:middle line:84% Workers get paid in grain, and they use used the grain 00:09:46.160 --> 00:09:49.120 align:middle line:90% to purchase other items. 00:09:49.120 --> 00:09:52.800 align:middle line:84% And you have IOUs that are denominated in money and IOUs 00:09:52.800 --> 00:09:54.920 align:middle line:90% that are denominated in grain. 00:09:54.920 --> 00:09:56.860 align:middle line:90% And all of this is coexisting. 00:09:56.860 --> 00:10:01.320 align:middle line:84% And you have objects that rarely appear 00:10:01.320 --> 00:10:05.640 align:middle line:84% except in the bilateral exchange. 00:10:05.640 --> 00:10:09.600 align:middle line:84% The point here is that blockchains are broader 00:10:09.600 --> 00:10:14.740 align:middle line:84% and the financial accounts are broader. 00:10:14.740 --> 00:10:17.440 align:middle line:84% Blockchains in particular are a way 00:10:17.440 --> 00:10:22.400 align:middle line:90% to write and execute contracts. 00:10:22.400 --> 00:10:26.560 align:middle line:84% So you take the concept of a balance 00:10:26.560 --> 00:10:32.700 align:middle line:84% of an item and the transfer change in the balance, 00:10:32.700 --> 00:10:34.560 align:middle line:90% as in a transaction. 00:10:34.560 --> 00:10:40.280 align:middle line:84% That generalizes to the balance as the state and the transfer 00:10:40.280 --> 00:10:42.880 align:middle line:90% corresponds to state changes. 00:10:42.880 --> 00:10:45.500 align:middle line:84% So with that simple change in vocabulary, 00:10:45.500 --> 00:10:47.820 align:middle line:84% we're writing contracts on the blockchain. 00:10:47.820 --> 00:10:50.560 align:middle line:90% 00:10:50.560 --> 00:10:54.520 align:middle line:84% An interesting difference, tension really, 00:10:54.520 --> 00:10:58.560 align:middle line:84% appears with an individual versus community perspective. 00:10:58.560 --> 00:11:02.720 align:middle line:84% From the financial accounts perspective, 00:11:02.720 --> 00:11:06.560 align:middle line:84% where money is an asset, the tension 00:11:06.560 --> 00:11:14.500 align:middle line:84% is store the asset as a store of value on the one hand, 00:11:14.500 --> 00:11:16.920 align:middle line:90% or use it in transactions. 00:11:16.920 --> 00:11:21.640 align:middle line:84% And the more people are, shall we say, hoarding it, 00:11:21.640 --> 00:11:24.300 align:middle line:84% the less they're going to be using it for transactions. 00:11:24.300 --> 00:11:27.580 align:middle line:84% And we could have a liquidity shortage in the market, 00:11:27.580 --> 00:11:30.460 align:middle line:84% even though from each individual's point of view, 00:11:30.460 --> 00:11:34.240 align:middle line:90% they're being rational. 00:11:34.240 --> 00:11:38.080 align:middle line:84% Now, very much surprising from a financial accounting 00:11:38.080 --> 00:11:39.700 align:middle line:90% perspective-- 00:11:39.700 --> 00:11:45.370 align:middle line:84% if you go back to the invention of double entry bookkeeping 00:11:45.370 --> 00:11:51.490 align:middle line:84% in Italy with Pacioli, he embraced 00:11:51.490 --> 00:11:55.770 align:middle line:84% a community perspective, not individual accounting 00:11:55.770 --> 00:11:57.250 align:middle line:90% perspective. 00:11:57.250 --> 00:12:01.250 align:middle line:84% And for him, money-- yes, you could hold it, 00:12:01.250 --> 00:12:06.070 align:middle line:84% but it was really a liability, not an asset. 00:12:06.070 --> 00:12:10.770 align:middle line:84% It's a liability in the sense that there was an obligation 00:12:10.770 --> 00:12:16.030 align:middle line:84% to spend it to accomplish some community objective. 00:12:16.030 --> 00:12:19.730 align:middle line:90% 00:12:19.730 --> 00:12:21.970 align:middle line:90% I promised you balance. 00:12:21.970 --> 00:12:24.870 align:middle line:84% So nothing perfect about the blockchain. 00:12:24.870 --> 00:12:30.130 align:middle line:84% There's a theorem in computer science called the CAP theorem, 00:12:30.130 --> 00:12:34.450 align:middle line:84% which is an impossibility theorem that you cannot have all 00:12:34.450 --> 00:12:37.610 align:middle line:84% three of consistency, availability, 00:12:37.610 --> 00:12:41.140 align:middle line:84% and partition tolerance at the same time. 00:12:41.140 --> 00:12:43.240 align:middle line:90% And so you have to make choices. 00:12:43.240 --> 00:12:46.340 align:middle line:84% And that same trilemma can carry over 00:12:46.340 --> 00:12:50.060 align:middle line:84% and allow us to interpret village economies 00:12:50.060 --> 00:12:56.580 align:middle line:84% and contemporary US economies and so on in the sense of having 00:12:56.580 --> 00:12:57.920 align:middle line:90% to make choices. 00:12:57.920 --> 00:13:02.340 align:middle line:90% 00:13:02.340 --> 00:13:07.340 align:middle line:84% Well, again, just to repeat, I don't 00:13:07.340 --> 00:13:11.060 align:middle line:84% expect us all to completely understand these slides. 00:13:11.060 --> 00:13:15.140 align:middle line:84% I'm trying to give you an introduction to what we're 00:13:15.140 --> 00:13:19.940 align:middle line:84% going to be doing in more detail in this case on Thursday. 00:13:19.940 --> 00:13:26.420 align:middle line:84% So I'll resist the temptation to lecture in too much detail. 00:13:26.420 --> 00:13:33.300 align:middle line:84% So this part is having to do with the opportunities 00:13:33.300 --> 00:13:39.310 align:middle line:84% of the course, which is these so-called readings, which 00:13:39.310 --> 00:13:40.090 align:middle line:90% are featured. 00:13:40.090 --> 00:13:44.650 align:middle line:84% Here we have Agustin Carstens, "The Future Monetary System-- 00:13:44.650 --> 00:13:47.950 align:middle line:90% Vision to Reality" from the BIS. 00:13:47.950 --> 00:13:51.870 align:middle line:84% Relatively recently, this strange thing called 00:13:51.870 --> 00:13:57.510 align:middle line:84% "The Baby-Sitting Economy," which Paul Krugman reviewed, 00:13:57.510 --> 00:14:03.390 align:middle line:84% has to do with an object, which are babysitting coupons 00:14:03.390 --> 00:14:09.390 align:middle line:84% or certificates that were used in exchange and analogies 00:14:09.390 --> 00:14:12.550 align:middle line:90% to money and what can go wrong. 00:14:12.550 --> 00:14:18.550 align:middle line:84% Goldstein and co-authors explore this tension in real-time gross 00:14:18.550 --> 00:14:20.550 align:middle line:84% settlement between "Payments, Reserves, 00:14:20.550 --> 00:14:23.150 align:middle line:90% and Financial Fragility." 00:14:23.150 --> 00:14:30.390 align:middle line:84% And De Meijer is an article about Kenya, 00:14:30.390 --> 00:14:32.790 align:middle line:84% where cryptocurrency blockchains are 00:14:32.790 --> 00:14:39.920 align:middle line:84% being used to serve the purpose of targeting poor households 00:14:39.920 --> 00:14:44.040 align:middle line:90% in rural or urban communities. 00:14:44.040 --> 00:14:49.240 align:middle line:84% So in the lectures, I touch on these articles, 00:14:49.240 --> 00:14:53.080 align:middle line:84% but the main reason they're here is for your interest. 00:14:53.080 --> 00:14:56.720 align:middle line:84% These or other papers may be something 00:14:56.720 --> 00:14:59.120 align:middle line:90% you would like to explore. 00:14:59.120 --> 00:15:05.760 align:middle line:84% And there's, as I'll say later, no midterm, no exam. 00:15:05.760 --> 00:15:09.000 align:middle line:90% It's all about research. 00:15:09.000 --> 00:15:14.520 align:middle line:84% So you have the option to pursue what you're interested in most. 00:15:14.520 --> 00:15:19.240 align:middle line:90% And we provide some pathways. 00:15:19.240 --> 00:15:24.040 align:middle line:84% So at the end of each of these lecture two-page summaries, 00:15:24.040 --> 00:15:27.760 align:middle line:84% I'll just list some of the papers. 00:15:27.760 --> 00:15:32.520 align:middle line:84% OK, lecture 3 goes back to these tools-- in this case, 00:15:32.520 --> 00:15:34.360 align:middle line:90% distributed ledgers-- 00:15:34.360 --> 00:15:38.480 align:middle line:90% and what you can do with them. 00:15:38.480 --> 00:15:43.320 align:middle line:84% The longer title is Fragmented Markets, Policy Objectives, 00:15:43.320 --> 00:15:46.440 align:middle line:84% Regulatory Solution, and Distributed Ledgers 00:15:46.440 --> 00:15:49.720 align:middle line:90% as a Technology Solution. 00:15:49.720 --> 00:15:54.320 align:middle line:84% We're going to put some friction in the markets 00:15:54.320 --> 00:15:57.520 align:middle line:84% and then see what problem emerges 00:15:57.520 --> 00:16:00.800 align:middle line:90% and how to remedy the problem. 00:16:00.800 --> 00:16:06.200 align:middle line:84% Sometimes, there are regulatory solutions, 00:16:06.200 --> 00:16:10.200 align:middle line:84% as in this US national marketing system, the way 00:16:10.200 --> 00:16:12.920 align:middle line:90% the US does the regulation. 00:16:12.920 --> 00:16:19.200 align:middle line:84% And we will also feature the use of distributed ledgers 00:16:19.200 --> 00:16:23.840 align:middle line:84% as a solution to an information problem. 00:16:23.840 --> 00:16:27.380 align:middle line:84% Now, time and time again, when you're reading on the web, 00:16:27.380 --> 00:16:33.610 align:middle line:84% you'll come across a use case or an objective, 00:16:33.610 --> 00:16:36.930 align:middle line:84% something perhaps the private sector is doing 00:16:36.930 --> 00:16:39.210 align:middle line:90% or central banks are doing. 00:16:39.210 --> 00:16:43.010 align:middle line:84% And they just launch right into the application. 00:16:43.010 --> 00:16:44.130 align:middle line:90% Why? 00:16:44.130 --> 00:16:47.610 align:middle line:84% What are they trying to accomplish, exactly? 00:16:47.610 --> 00:16:51.330 align:middle line:84% Heaven forbid just use the new technologies 00:16:51.330 --> 00:16:56.970 align:middle line:84% for the sake of doing it without thinking about social welfare. 00:16:56.970 --> 00:17:00.710 align:middle line:90% So we will get into this. 00:17:00.710 --> 00:17:04.970 align:middle line:84% Economists have a criterion typically used, 00:17:04.970 --> 00:17:09.410 align:middle line:84% called Pareto optimality, which is a weak standard that 00:17:09.410 --> 00:17:13.210 align:middle line:84% says whatever is going on currently, 00:17:13.210 --> 00:17:16.530 align:middle line:84% you should not be able to make some participants better 00:17:16.530 --> 00:17:21.190 align:middle line:84% off without harming other participants 00:17:21.190 --> 00:17:24.750 align:middle line:84% because then there's a gain for everyone. 00:17:24.750 --> 00:17:29.210 align:middle line:84% So it could not possibly be efficient the way it is. 00:17:29.210 --> 00:17:34.640 align:middle line:84% So these slides have to do with prices should be equated, 00:17:34.640 --> 00:17:37.140 align:middle line:84% even if we have fragmented markets, 00:17:37.140 --> 00:17:39.260 align:middle line:90% which is what the US is doing. 00:17:39.260 --> 00:17:43.940 align:middle line:84% They look at the execution of stock market trades 00:17:43.940 --> 00:17:47.780 align:middle line:84% and have regulations that try to make 00:17:47.780 --> 00:17:53.020 align:middle line:84% sure people aren't being cheated and buying things more 00:17:53.020 --> 00:17:56.060 align:middle line:84% expensively than they would have if they had 00:17:56.060 --> 00:17:58.460 align:middle line:90% bought on a different exchange. 00:17:58.460 --> 00:18:03.260 align:middle line:84% The blockchain point of view is let's 00:18:03.260 --> 00:18:06.700 align:middle line:84% fragment the markets artificially, take 00:18:06.700 --> 00:18:13.420 align:middle line:84% that as a given, and then try to achieve 00:18:13.420 --> 00:18:15.540 align:middle line:84% what would have been a competitive outcome 00:18:15.540 --> 00:18:21.420 align:middle line:84% or an efficient outcome with agents constrained 00:18:21.420 --> 00:18:24.900 align:middle line:84% to be trading with each other pairwise in these fragmented 00:18:24.900 --> 00:18:28.260 align:middle line:90% markets. 00:18:28.260 --> 00:18:31.110 align:middle line:84% And what happens here is that even 00:18:31.110 --> 00:18:36.070 align:middle line:84% if you impose the correct prices, 00:18:36.070 --> 00:18:40.030 align:middle line:84% it is, in general, impossible for agents 00:18:40.030 --> 00:18:44.710 align:middle line:84% to solve the problem in a decentralized way, looking 00:18:44.710 --> 00:18:49.310 align:middle line:84% at their own histories or the histories of people 00:18:49.310 --> 00:18:50.450 align:middle line:90% that they've met. 00:18:50.450 --> 00:18:53.110 align:middle line:90% 00:18:53.110 --> 00:18:59.470 align:middle line:84% It's just like programming code and deciding what information 00:18:59.470 --> 00:19:01.870 align:middle line:84% the code is allowed to use or would 00:19:01.870 --> 00:19:07.470 align:middle line:84% be required to use in order to get to the objective. 00:19:07.470 --> 00:19:13.170 align:middle line:84% And bilateral credits and debits are, in general, not sufficient. 00:19:13.170 --> 00:19:16.190 align:middle line:84% You need to invoke this community 00:19:16.190 --> 00:19:21.350 align:middle line:84% perspective, like the common but distributed ledger. 00:19:21.350 --> 00:19:27.190 align:middle line:84% These pictures at the bottom are like a balance sheet 00:19:27.190 --> 00:19:32.750 align:middle line:84% at a point in time and the transactions 00:19:32.750 --> 00:19:35.590 align:middle line:90% that change the balance. 00:19:35.590 --> 00:19:41.910 align:middle line:84% And they evolve over time as an individual is trading. 00:19:41.910 --> 00:19:45.350 align:middle line:84% And having individuals see only their own accounts 00:19:45.350 --> 00:19:48.470 align:middle line:84% like this or even those of their trading partner 00:19:48.470 --> 00:19:52.230 align:middle line:84% will not allow you, in general, to get to the Walrasian outcome. 00:19:52.230 --> 00:19:54.030 align:middle line:84% Now, you may think there is a way 00:19:54.030 --> 00:20:01.110 align:middle line:84% to do better with money, just like a decentralized system 00:20:01.110 --> 00:20:05.350 align:middle line:84% where you carry around your own cash or e-credit 00:20:05.350 --> 00:20:10.470 align:middle line:84% and spend it as you wish, each person doing that. 00:20:10.470 --> 00:20:15.870 align:middle line:84% The problem is it takes an enormous amount of liquidity 00:20:15.870 --> 00:20:19.190 align:middle line:84% for agents to achieve the objective 00:20:19.190 --> 00:20:22.710 align:middle line:84% of the efficient Walrasian allocation. 00:20:22.710 --> 00:20:25.230 align:middle line:84% So the decentralized way in which 00:20:25.230 --> 00:20:30.080 align:middle line:84% we're used to thinking about money is actually inadequate. 00:20:30.080 --> 00:20:37.680 align:middle line:84% And in any event, in real-time gross settlement systems, 00:20:37.680 --> 00:20:40.640 align:middle line:84% there's a misnomer because if it really 00:20:40.640 --> 00:20:43.920 align:middle line:84% were instantaneous real-time, you'd 00:20:43.920 --> 00:20:47.080 align:middle line:84% have to have a lot of liquidity for banks 00:20:47.080 --> 00:20:53.160 align:middle line:84% to be able to honor presentation of commitments due. 00:20:53.160 --> 00:20:55.880 align:middle line:84% So almost simultaneously, with the coming 00:20:55.880 --> 00:20:57.720 align:middle line:84% of real-time gross settlement instead 00:20:57.720 --> 00:21:01.480 align:middle line:84% of end-of-day settlement, there came this quest 00:21:01.480 --> 00:21:06.320 align:middle line:84% to find liquidity-saving mechanisms. 00:21:06.320 --> 00:21:09.120 align:middle line:84% And Ostroy and Starr anticipates this. 00:21:09.120 --> 00:21:16.080 align:middle line:84% Another thing you could do is have a warehouse facility 00:21:16.080 --> 00:21:20.680 align:middle line:84% so nobody trades until they meet the big trader. 00:21:20.680 --> 00:21:22.980 align:middle line:90% And what goes wrong? 00:21:22.980 --> 00:21:24.600 align:middle line:90% Well, a couple of things. 00:21:24.600 --> 00:21:27.810 align:middle line:84% That trader would have to have enormous inventories 00:21:27.810 --> 00:21:30.530 align:middle line:84% of all the goods or assets in order 00:21:30.530 --> 00:21:35.770 align:middle line:84% to honor the requests for outflows. 00:21:35.770 --> 00:21:40.530 align:middle line:84% And likewise, having a very big trader like that, 00:21:40.530 --> 00:21:45.610 align:middle line:84% he or she will be tempted to exploit their market power, 00:21:45.610 --> 00:21:48.390 align:middle line:84% which I claim we also see in practice. 00:21:48.390 --> 00:21:52.570 align:middle line:84% And I'll show you more when we get to this lecture. 00:21:52.570 --> 00:21:55.150 align:middle line:84% And finally, you could think, well, why money? 00:21:55.150 --> 00:21:56.150 align:middle line:90% Why not credit? 00:21:56.150 --> 00:21:59.730 align:middle line:84% Let's just have deficit accounts, 00:21:59.730 --> 00:22:03.530 align:middle line:84% overdraft accounts, and let people spend out 00:22:03.530 --> 00:22:05.530 align:middle line:90% of those accounts. 00:22:05.530 --> 00:22:08.430 align:middle line:84% Well, that's a little bit like the end-of-day settlement. 00:22:08.430 --> 00:22:12.010 align:middle line:84% And the problem is people show up having bought more 00:22:12.010 --> 00:22:14.770 align:middle line:90% in value than they sold. 00:22:14.770 --> 00:22:18.650 align:middle line:84% And they can't honor the commitment 00:22:18.650 --> 00:22:21.530 align:middle line:84% to balance their account at the end. 00:22:21.530 --> 00:22:25.420 align:middle line:84% So that's like a default, and we see that a lot. 00:22:25.420 --> 00:22:33.060 align:middle line:84% So this theory article is amazingly 00:22:33.060 --> 00:22:35.500 align:middle line:84% capturing a lot of the current institutions we 00:22:35.500 --> 00:22:37.300 align:middle line:90% see, but remember. 00:22:37.300 --> 00:22:39.620 align:middle line:84% We're talking about a blockchain-- 00:22:39.620 --> 00:22:44.020 align:middle line:84% distributed ledgers as a way to solve those problems. 00:22:44.020 --> 00:22:49.820 align:middle line:84% OK, so again, a hint of things to explore. 00:22:49.820 --> 00:22:55.120 align:middle line:84% Chester Spatt has written, with his co-author, 00:22:55.120 --> 00:22:57.320 align:middle line:84% this "Regulating Market Microstructure." 00:22:57.320 --> 00:23:01.460 align:middle line:84% You can read all about what goes on in the US in terms 00:23:01.460 --> 00:23:05.780 align:middle line:84% of regulation and other related articles, 00:23:05.780 --> 00:23:08.220 align:middle line:84% or this liquidity savings mechanisms 00:23:08.220 --> 00:23:10.100 align:middle line:84% that I've already talked about, as described 00:23:10.100 --> 00:23:12.140 align:middle line:90% in Martin and McAndrews. 00:23:12.140 --> 00:23:20.980 align:middle line:84% This Alcazar has to do with a very concentrated provision 00:23:20.980 --> 00:23:24.860 align:middle line:90% of core banking services-- 00:23:24.860 --> 00:23:27.260 align:middle line:84% in other words, imperfect competition. 00:23:27.260 --> 00:23:33.620 align:middle line:84% And the "Tri-Party" paper is about these enormous potential 00:23:33.620 --> 00:23:37.180 align:middle line:84% liabilities that the Federal Reserve took on in the operation 00:23:37.180 --> 00:23:42.500 align:middle line:84% of the repo markets, often having within middle-of-day 00:23:42.500 --> 00:23:47.380 align:middle line:84% liabilities, larger than the entire US money stock. 00:23:47.380 --> 00:23:50.720 align:middle line:84% Of course, they figured it out eventually and tried to fix it. 00:23:50.720 --> 00:23:56.060 align:middle line:84% So again, as your interests emerge-- 00:23:56.060 --> 00:23:58.300 align:middle line:84% and you can browse and then decide no, I don't 00:23:58.300 --> 00:24:00.540 align:middle line:90% want to go down that path. 00:24:00.540 --> 00:24:03.220 align:middle line:90% That's fine. 00:24:03.220 --> 00:24:04.720 align:middle line:90% OK. 00:24:04.720 --> 00:24:07.880 align:middle line:84% So featured distributed ledgers-- 00:24:07.880 --> 00:24:10.080 align:middle line:84% now, I want to feature smart contracts, 00:24:10.080 --> 00:24:14.900 align:middle line:84% and also how you can use them, in this case, 00:24:14.900 --> 00:24:18.380 align:middle line:84% as a solution to a coordination problem. 00:24:18.380 --> 00:24:26.670 align:middle line:84% So here we're going to have the commodity space being locations 00:24:26.670 --> 00:24:30.810 align:middle line:84% and dates and randomly realized dates of the world, 00:24:30.810 --> 00:24:33.310 align:middle line:84% but we're going to stick with this Pareto criterion 00:24:33.310 --> 00:24:35.350 align:middle line:90% for efficiency. 00:24:35.350 --> 00:24:40.430 align:middle line:84% And with fragmented markets, we will 00:24:40.430 --> 00:24:45.750 align:middle line:84% try to implement the solution with privately issued securities 00:24:45.750 --> 00:24:47.230 align:middle line:90% or monies. 00:24:47.230 --> 00:24:53.830 align:middle line:84% They will be high velocity and circulate in exchange, 00:24:53.830 --> 00:24:58.750 align:middle line:84% but it's going to lead to a coordination problem 00:24:58.750 --> 00:25:02.190 align:middle line:90% and to crises. 00:25:02.190 --> 00:25:05.910 align:middle line:84% And the solution will come with this multi-agent smart contract. 00:25:05.910 --> 00:25:10.390 align:middle line:84% So this picture is meant to evoke that agents 00:25:10.390 --> 00:25:11.610 align:middle line:90% are getting partitioned. 00:25:11.610 --> 00:25:13.810 align:middle line:84% They're either in one location or another. 00:25:13.810 --> 00:25:15.390 align:middle line:90% We settle for two. 00:25:15.390 --> 00:25:18.690 align:middle line:84% They meet pairwise like that in trade, 00:25:18.690 --> 00:25:21.480 align:middle line:84% but then they get buffered around and sorted 00:25:21.480 --> 00:25:24.720 align:middle line:84% with different agents in yet other locations. 00:25:24.720 --> 00:25:28.700 align:middle line:84% And they can issue these securities that circulate. 00:25:28.700 --> 00:25:32.000 align:middle line:84% One can issue a promise to pay at date 4, 00:25:32.000 --> 00:25:36.680 align:middle line:84% issue it to 2, 2 carries it to 4, 4 trades to 3, 00:25:36.680 --> 00:25:40.040 align:middle line:84% and 3 presents it for redemption. 00:25:40.040 --> 00:25:43.000 align:middle line:84% So this private security, this IOU, 00:25:43.000 --> 00:25:48.800 align:middle line:84% will circulate in this model with high velocity. 00:25:48.800 --> 00:25:49.780 align:middle line:90% That's fine. 00:25:49.780 --> 00:25:52.080 align:middle line:90% It's a privately issued money. 00:25:52.080 --> 00:25:56.680 align:middle line:84% The problem is there are other privately issued monies issued 00:25:56.680 --> 00:25:59.320 align:middle line:90% in the other location. 00:25:59.320 --> 00:26:07.720 align:middle line:84% And if you don't have this coordination problem solved, 00:26:07.720 --> 00:26:12.440 align:middle line:84% they may incorrectly anticipate what's going on in other 00:26:12.440 --> 00:26:16.300 align:middle line:84% locations where assets are being issued, 00:26:16.300 --> 00:26:19.570 align:middle line:90% but they cannot see that. 00:26:19.570 --> 00:26:24.650 align:middle line:84% And if they guess incorrectly, you'll get a crash. 00:26:24.650 --> 00:26:28.650 align:middle line:84% Now, we see this in theory very clearly because we've 00:26:28.650 --> 00:26:29.910 align:middle line:90% simulated the model. 00:26:29.910 --> 00:26:32.810 align:middle line:84% And I'll show you that in this lecture 4. 00:26:32.810 --> 00:26:35.730 align:middle line:90% 00:26:35.730 --> 00:26:38.650 align:middle line:90% We've also seen it historically. 00:26:38.650 --> 00:26:48.170 align:middle line:84% So Walter Bagehot was describing the money markets in London, 00:26:48.170 --> 00:26:50.210 align:middle line:84% where these bills of exchange were 00:26:50.210 --> 00:26:55.970 align:middle line:90% circulating with high velocity. 00:26:55.970 --> 00:26:59.170 align:middle line:84% But then there are these periodic crashes. 00:26:59.170 --> 00:27:04.570 align:middle line:84% And I was over the break in London 00:27:04.570 --> 00:27:08.250 align:middle line:84% at the central bank, went looking. 00:27:08.250 --> 00:27:09.090 align:middle line:90% It occurred to me-- 00:27:09.090 --> 00:27:11.450 align:middle line:84% I was too early for my appointment-- 00:27:11.450 --> 00:27:12.990 align:middle line:90% to look for Lombard Street. 00:27:12.990 --> 00:27:16.780 align:middle line:90% Lombard Street is still there. 00:27:16.780 --> 00:27:18.580 align:middle line:84% And that's where the first clearinghouse 00:27:18.580 --> 00:27:21.060 align:middle line:90% was ever established. 00:27:21.060 --> 00:27:27.380 align:middle line:84% So it was real, and it was the center of the world's money 00:27:27.380 --> 00:27:28.740 align:middle line:90% markets. 00:27:28.740 --> 00:27:35.460 align:middle line:84% And what was going on then several hundred years ago 00:27:35.460 --> 00:27:36.540 align:middle line:90% is still-- 00:27:36.540 --> 00:27:40.980 align:middle line:84% we still face those problems today. 00:27:40.980 --> 00:27:42.400 align:middle line:90% There are related things. 00:27:42.400 --> 00:27:45.060 align:middle line:84% There's a concern in low and middle income countries 00:27:45.060 --> 00:27:49.840 align:middle line:84% with this digitization that is emerging 00:27:49.840 --> 00:27:53.220 align:middle line:84% and the fact that it's not coordinated. 00:27:53.220 --> 00:27:55.820 align:middle line:84% So parties will be having to basically 00:27:55.820 --> 00:27:59.060 align:middle line:84% cover balances without having the assets, 00:27:59.060 --> 00:28:01.840 align:middle line:90% like a big liability. 00:28:01.840 --> 00:28:04.300 align:middle line:84% That's a concern in some of these countries. 00:28:04.300 --> 00:28:11.080 align:middle line:84% And likewise, even with Decentralized Finance, or DeFi, 00:28:11.080 --> 00:28:16.700 align:middle line:84% which are these markets for cryptocurrencies, 00:28:16.700 --> 00:28:25.580 align:middle line:84% there is a concern that we don't have enough liquidity 00:28:25.580 --> 00:28:28.260 align:middle line:84% or that there is a liquidity problem. 00:28:28.260 --> 00:28:30.580 align:middle line:84% So the solution to these problems 00:28:30.580 --> 00:28:34.180 align:middle line:84% turns out to be available now with the distributed 00:28:34.180 --> 00:28:36.680 align:middle line:90% ledger and smart contracts. 00:28:36.680 --> 00:28:39.340 align:middle line:90% 00:28:39.340 --> 00:28:47.460 align:middle line:84% So these articles are using the Pareto criterion 00:28:47.460 --> 00:28:51.660 align:middle line:84% to explore whether or not you need to intervene. 00:28:51.660 --> 00:28:56.300 align:middle line:84% In this case in the US, the paper by Gorton 00:28:56.300 --> 00:29:02.020 align:middle line:84% is about those bills of exchange in England and how-- 00:29:02.020 --> 00:29:06.180 align:middle line:84% he calls it private money production without banks. 00:29:06.180 --> 00:29:09.180 align:middle line:84% Bagehot, I've just shown you on the screen. 00:29:09.180 --> 00:29:13.630 align:middle line:84% Sargent and Wallace is the paper about real bills 00:29:13.630 --> 00:29:16.230 align:middle line:84% versus the quantity theory, having 00:29:16.230 --> 00:29:20.430 align:middle line:90% to do with causes of inflation. 00:29:20.430 --> 00:29:22.430 align:middle line:84% The quantity theory is that inflation 00:29:22.430 --> 00:29:25.110 align:middle line:84% is determined by the quantity of money, 00:29:25.110 --> 00:29:28.470 align:middle line:84% and we should be careful that all monies be limited 00:29:28.470 --> 00:29:30.470 align:middle line:90% in supply for that reason. 00:29:30.470 --> 00:29:34.390 align:middle line:84% The real bills doctrine says no, that's wrong. 00:29:34.390 --> 00:29:37.830 align:middle line:84% Financial intermediation is fine as long 00:29:37.830 --> 00:29:42.110 align:middle line:84% as there is real backing for the liabilities, 00:29:42.110 --> 00:29:44.910 align:middle line:84% the so-called real bills doctrine. 00:29:44.910 --> 00:29:48.350 align:middle line:84% And this is the basis of a debate that is still 00:29:48.350 --> 00:29:54.430 align:middle line:84% going on today that is relevant for central banks like the US 00:29:54.430 --> 00:29:57.110 align:middle line:90% Federal Reserve. 00:29:57.110 --> 00:29:59.790 align:middle line:84% And I've mentioned the liquidity fragmentation. 00:29:59.790 --> 00:30:00.870 align:middle line:90% OK. 00:30:00.870 --> 00:30:05.470 align:middle line:84% Lecture 5 goes further in exploring 00:30:05.470 --> 00:30:10.070 align:middle line:84% tokenized and programmable assets. 00:30:10.070 --> 00:30:13.760 align:middle line:84% So now, the ledger has become dynamic. 00:30:13.760 --> 00:30:18.440 align:middle line:84% We can talk about atomic trade and settlement 00:30:18.440 --> 00:30:22.280 align:middle line:84% and do that on platforms and compare 00:30:22.280 --> 00:30:24.680 align:middle line:84% these tokenized immediate settlement 00:30:24.680 --> 00:30:29.800 align:middle line:84% assets on the blockchain with the current legacy systems 00:30:29.800 --> 00:30:32.720 align:middle line:90% where there are trade fails. 00:30:32.720 --> 00:30:37.400 align:middle line:84% So the buzzwords are tokenization 00:30:37.400 --> 00:30:41.480 align:middle line:84% and programmable assets that are made possible 00:30:41.480 --> 00:30:46.840 align:middle line:90% by multilateral smart contracts. 00:30:46.840 --> 00:30:50.520 align:middle line:84% Now, when you think about the accounting point of view 00:30:50.520 --> 00:30:55.000 align:middle line:84% of legacy systems, one thing that 00:30:55.000 --> 00:31:00.600 align:middle line:84% hits you is the balance sheets, assets, and liabilities 00:31:00.600 --> 00:31:07.160 align:middle line:84% should not be just simple static representations of a trader's 00:31:07.160 --> 00:31:09.160 align:middle line:90% assets and liabilities. 00:31:09.160 --> 00:31:15.170 align:middle line:84% It should be a dynamic, explicit representation of future assets 00:31:15.170 --> 00:31:17.250 align:middle line:84% and liabilities that they contract 00:31:17.250 --> 00:31:23.290 align:middle line:84% for under these multilateral smart contracts. 00:31:23.290 --> 00:31:30.690 align:middle line:84% And likewise, you can agree for trades into the future 00:31:30.690 --> 00:31:34.330 align:middle line:90% and commit to them now. 00:31:34.330 --> 00:31:36.610 align:middle line:84% So you have this odd juxtaposition 00:31:36.610 --> 00:31:41.990 align:middle line:84% of immediate settlement in the sense of not only trading, 00:31:41.990 --> 00:31:44.190 align:middle line:84% but committing to the outcome of the trade, 00:31:44.190 --> 00:31:51.410 align:middle line:84% whereas in legacy systems, it's end-of-day or t plus 3. 00:31:51.410 --> 00:31:59.210 align:middle line:84% And you may or may not get the asset you thought you purchased. 00:31:59.210 --> 00:32:05.970 align:middle line:84% And the guy with the money may not show up or vice versa. 00:32:05.970 --> 00:32:09.430 align:middle line:84% The asset is not there when you need it. 00:32:09.430 --> 00:32:12.090 align:middle line:84% So this little example here at the bottom 00:32:12.090 --> 00:32:18.610 align:middle line:84% is about three agents, simple prototype, 00:32:18.610 --> 00:32:21.610 align:middle line:90% where it's asset lending. 00:32:21.610 --> 00:32:26.130 align:middle line:84% And they want to hold different levels of the asset 00:32:26.130 --> 00:32:32.330 align:middle line:84% at different points in time as an example of multilateral asset 00:32:32.330 --> 00:32:33.410 align:middle line:90% trades. 00:32:33.410 --> 00:32:37.330 align:middle line:84% I mentioned the legacy system and trade fails. 00:32:37.330 --> 00:32:43.170 align:middle line:84% And it's shocking at first, when you first read about it, 00:32:43.170 --> 00:32:45.930 align:middle line:90% the magnitude of the fails. 00:32:45.930 --> 00:32:55.530 align:middle line:84% So for example, here are treasuries in it 00:32:55.530 --> 00:32:58.410 align:middle line:90% says, billions of dollars. 00:32:58.410 --> 00:33:02.550 align:middle line:84% But 10,000 billion-- we're into the trillions. 00:33:02.550 --> 00:33:06.050 align:middle line:84% So we're measuring here the value 00:33:06.050 --> 00:33:11.820 align:middle line:84% in dollars of trades that were agreed to, but not executed. 00:33:11.820 --> 00:33:18.340 align:middle line:84% So it's, quote, "arguably a big problem." 00:33:18.340 --> 00:33:22.740 align:middle line:84% Now, the blockchain isn't a cure-all, either, 00:33:22.740 --> 00:33:26.300 align:middle line:84% because various problems would emerge, as will 00:33:26.300 --> 00:33:29.660 align:middle line:90% be clear from that example. 00:33:29.660 --> 00:33:31.780 align:middle line:84% If it were the case the asset can only 00:33:31.780 --> 00:33:35.980 align:middle line:84% be programmed by someone who currently holds it, 00:33:35.980 --> 00:33:40.740 align:middle line:84% then entering into a trade could reveal information 00:33:40.740 --> 00:33:44.100 align:middle line:84% that they would rather keep private, 00:33:44.100 --> 00:33:47.140 align:middle line:84% which, in turn, may mean that they're not 00:33:47.140 --> 00:33:49.820 align:middle line:90% going to get a good price. 00:33:49.820 --> 00:33:51.700 align:middle line:84% And there are potential solutions 00:33:51.700 --> 00:33:55.540 align:middle line:84% having to do with centralizing these markets 00:33:55.540 --> 00:34:04.260 align:middle line:84% or being able to use the conditional release 00:34:04.260 --> 00:34:10.070 align:middle line:84% function on the tokenized asset, which we can talk about more 00:34:10.070 --> 00:34:14.230 align:middle line:90% later, or with encryption. 00:34:14.230 --> 00:34:17.550 align:middle line:84% But in general, these systems that emerge 00:34:17.550 --> 00:34:21.190 align:middle line:84% may be only partially interoperable. 00:34:21.190 --> 00:34:25.790 align:middle line:84% And even the tokenization raises issues of trust 00:34:25.790 --> 00:34:29.150 align:middle line:84% if, in fact, it's a pre-existing asset that 00:34:29.150 --> 00:34:32.710 align:middle line:84% has to be put in escrow in order to allow us to tokenize 00:34:32.710 --> 00:34:34.429 align:middle line:90% and write code on it. 00:34:34.429 --> 00:34:40.310 align:middle line:84% You're relying on the issuer to hold it to the escrow account 00:34:40.310 --> 00:34:44.730 align:middle line:84% to be valid and for them not to cheat and use the asset. 00:34:44.730 --> 00:34:46.630 align:middle line:84% It's got to be there when you need it. 00:34:46.630 --> 00:34:51.230 align:middle line:84% So there are various explorations going on 00:34:51.230 --> 00:34:55.110 align:middle line:84% among domestic central banks and international agencies 00:34:55.110 --> 00:34:59.110 align:middle line:84% for a cross-border exchange problem. 00:34:59.110 --> 00:35:03.680 align:middle line:84% Such a blockchain has been proposed for foreign exchange 00:35:03.680 --> 00:35:04.620 align:middle line:90% transaction. 00:35:04.620 --> 00:35:08.400 align:middle line:84% And the BIS has now switched gears a bit 00:35:08.400 --> 00:35:15.640 align:middle line:84% and is proposing unified ledgers even for domestic systems. 00:35:15.640 --> 00:35:19.320 align:middle line:84% But we have to be careful that, in fact, we 00:35:19.320 --> 00:35:22.600 align:middle line:84% don't end up violating the coherence guarantee, which 00:35:22.600 --> 00:35:26.240 align:middle line:84% is a big feature of what you can do 00:35:26.240 --> 00:35:28.540 align:middle line:84% with these multilateral smart contracts. 00:35:28.540 --> 00:35:31.840 align:middle line:90% So again, these readings-- 00:35:31.840 --> 00:35:37.520 align:middle line:84% read all about settlement fails, understand better 00:35:37.520 --> 00:35:42.480 align:middle line:84% this risk of tokenization, be mindful of what 00:35:42.480 --> 00:35:46.440 align:middle line:84% it means to have programmable money with this coherence 00:35:46.440 --> 00:35:47.640 align:middle line:90% guarantee. 00:35:47.640 --> 00:35:50.880 align:middle line:84% And these two lines at the end have 00:35:50.880 --> 00:35:54.160 align:middle line:84% to do with what we're seeing out there 00:35:54.160 --> 00:35:56.040 align:middle line:84% currently from the Swiss National 00:35:56.040 --> 00:36:01.440 align:middle line:84% Bank and other consortia of large banks. 00:36:01.440 --> 00:36:06.090 align:middle line:84% So again, each of these, the last in particular, 00:36:06.090 --> 00:36:13.850 align:middle line:84% are options for you to explore in your readings. 00:36:13.850 --> 00:36:17.070 align:middle line:84% How does it work, exactly, critique it, 00:36:17.070 --> 00:36:20.850 align:middle line:90% understand it better-- 00:36:20.850 --> 00:36:26.090 align:middle line:90% would all be potential projects. 00:36:26.090 --> 00:36:31.330 align:middle line:84% OK, so moving on, we come to algorithmic flows 00:36:31.330 --> 00:36:34.570 align:middle line:84% on networks as solution to multilateral settlement 00:36:34.570 --> 00:36:35.410 align:middle line:90% problems. 00:36:35.410 --> 00:36:38.690 align:middle line:84% Or the longer title is How to Improve 00:36:38.690 --> 00:36:40.810 align:middle line:90% Financial Infrastructure-- 00:36:40.810 --> 00:36:44.130 align:middle line:84% Taking Advantage of What We Know about Network Cycles 00:36:44.130 --> 00:36:49.210 align:middle line:84% and Chains Using Algorithms to Maximize Timely 00:36:49.210 --> 00:36:52.710 align:middle line:84% Payment of Multilateral Trade Credit Offsets, 00:36:52.710 --> 00:36:56.570 align:middle line:84% or alternatively, back to the US repo market-- 00:36:56.570 --> 00:37:02.850 align:middle line:84% what to do about credit guarantees, time permitting. 00:37:02.850 --> 00:37:07.390 align:middle line:84% So we put the obligations of, say, 00:37:07.390 --> 00:37:14.370 align:middle line:84% trader I to pay trader J on a network diagram. 00:37:14.370 --> 00:37:17.730 align:middle line:84% So for example, A owes 2 units to B, 00:37:17.730 --> 00:37:21.450 align:middle line:90% B owes 2 units to C, et cetera. 00:37:21.450 --> 00:37:27.570 align:middle line:84% C owes 1 unit to A. So those edges, 00:37:27.570 --> 00:37:29.890 align:middle line:84% when they connect the two nodes, represent 00:37:29.890 --> 00:37:34.130 align:middle line:84% a liability for one party and an asset for the other party 00:37:34.130 --> 00:37:36.810 align:middle line:90% if it's paid. 00:37:36.810 --> 00:37:40.290 align:middle line:84% But agents can be indirectly connected 00:37:40.290 --> 00:37:42.550 align:middle line:90% in various complicated ways. 00:37:42.550 --> 00:37:45.250 align:middle line:90% 00:37:45.250 --> 00:37:54.610 align:middle line:84% In fact, when you look at an actual picture of networks-- 00:37:54.610 --> 00:37:56.770 align:middle line:90% this is Italian data-- 00:37:56.770 --> 00:38:02.380 align:middle line:84% you can actually see the individual nodes and the edges 00:38:02.380 --> 00:38:09.300 align:middle line:84% because they're just tens of thousands of transactions. 00:38:09.300 --> 00:38:13.300 align:middle line:84% So you can begin to conjure up this interconnectedness, which 00:38:13.300 --> 00:38:15.480 align:middle line:90% is very complicated. 00:38:15.480 --> 00:38:18.580 align:middle line:84% In fact, individual agents are very 00:38:18.580 --> 00:38:22.780 align:middle line:84% unlikely to know much at all about the overall structure 00:38:22.780 --> 00:38:24.620 align:middle line:90% of the network. 00:38:24.620 --> 00:38:27.500 align:middle line:84% They'll know their own assets and liabilities, 00:38:27.500 --> 00:38:30.500 align:middle line:84% maybe a bit about their partners, 00:38:30.500 --> 00:38:33.940 align:middle line:84% but not much more than that, typically. 00:38:33.940 --> 00:38:37.540 align:middle line:84% The second problem is even if we had 00:38:37.540 --> 00:38:40.980 align:middle line:84% a multilateral smart contract node that 00:38:40.980 --> 00:38:45.340 align:middle line:84% did have the complete knowledge of the network, 00:38:45.340 --> 00:38:48.500 align:middle line:84% like a community ledger, it would still 00:38:48.500 --> 00:38:52.900 align:middle line:84% have to determine the ordering of the payments. 00:38:52.900 --> 00:38:58.230 align:middle line:84% Does A pay B, B pay C, then C pays A? 00:38:58.230 --> 00:39:04.030 align:middle line:84% But B has to be paid first in order to pay C, et cetera. 00:39:04.030 --> 00:39:08.070 align:middle line:84% It becomes a very large combinatorial problem 00:39:08.070 --> 00:39:12.190 align:middle line:84% to figure out the correct judicious ordering 00:39:12.190 --> 00:39:14.690 align:middle line:90% of the payments. 00:39:14.690 --> 00:39:18.630 align:middle line:84% And finally, even if there were a liquidity pool or some kind 00:39:18.630 --> 00:39:23.310 align:middle line:84% of overdraft facility that, in principle, could maximize 00:39:23.310 --> 00:39:26.030 align:middle line:84% the amount cleared, you'd still have 00:39:26.030 --> 00:39:32.750 align:middle line:84% to decide where to inject the liquidity, how much to inject, 00:39:32.750 --> 00:39:35.790 align:middle line:84% and how to recover that liquidity at the collection 00:39:35.790 --> 00:39:40.830 align:middle line:84% point in such a way that the system still remains balanced. 00:39:40.830 --> 00:39:43.490 align:middle line:84% So these problems are very typical. 00:39:43.490 --> 00:39:45.150 align:middle line:84% Here we're, in this lecture, going 00:39:45.150 --> 00:39:49.110 align:middle line:84% to feature the problem of trade credit overdues 00:39:49.110 --> 00:39:54.790 align:middle line:84% and how to have this multilateral set-off, 00:39:54.790 --> 00:40:00.520 align:middle line:84% either clearing enclosed cycles or injecting liquidity 00:40:00.520 --> 00:40:03.260 align:middle line:90% through chains. 00:40:03.260 --> 00:40:10.680 align:middle line:84% You can better see that here, where this, what's left of it, 00:40:10.680 --> 00:40:16.540 align:middle line:84% is going to turn out to be a closed cycle, where one is paid, 00:40:16.540 --> 00:40:19.560 align:middle line:84% paid, paid, and it clears, but the red 00:40:19.560 --> 00:40:22.600 align:middle line:84% is the liquidity injection from the source 00:40:22.600 --> 00:40:30.760 align:middle line:84% to the target, which also clears a lot more of the trade. 00:40:30.760 --> 00:40:32.800 align:middle line:90% Oh. 00:40:32.800 --> 00:40:40.260 align:middle line:84% So for this lecture, I have invited Tomaž Fleischman, 00:40:40.260 --> 00:40:45.240 align:middle line:84% who is a collaborator in this work, who 00:40:45.240 --> 00:40:51.080 align:middle line:84% is a computer scientist and also helps run 00:40:51.080 --> 00:40:53.120 align:middle line:90% a company with a blockchain. 00:40:53.120 --> 00:41:02.240 align:middle line:84% So he will tell you much more in that lecture, lecture 6. 00:41:02.240 --> 00:41:06.160 align:middle line:84% And there's, indeed, Fleischman with his co-authors, 00:41:06.160 --> 00:41:08.700 align:middle line:90% "Liquidity-Saving Obligation." 00:41:08.700 --> 00:41:09.660 align:middle line:90% There's a lot to read. 00:41:09.660 --> 00:41:14.040 align:middle line:90% I just picked one of his papers. 00:41:14.040 --> 00:41:19.400 align:middle line:84% You may, depending on how far you are in your studies, 00:41:19.400 --> 00:41:23.680 align:middle line:84% recognize this Cormen, et al. book on Introduction 00:41:23.680 --> 00:41:24.700 align:middle line:90% to Algorithms. 00:41:24.700 --> 00:41:31.640 align:middle line:84% It's co-authored with MIT professors. 00:41:31.640 --> 00:41:39.080 align:middle line:84% And again, time permitting, we could get into the repo market. 00:41:39.080 --> 00:41:43.100 align:middle line:84% I put this here because if we don't do it in class, 00:41:43.100 --> 00:41:46.040 align:middle line:84% then that's an option for one of you in terms 00:41:46.040 --> 00:41:48.420 align:middle line:90% of your own research project. 00:41:48.420 --> 00:41:51.600 align:middle line:90% 00:41:51.600 --> 00:41:55.970 align:middle line:84% So the previous slide, lecture 6, was something static, 00:41:55.970 --> 00:42:02.890 align:middle line:84% but you can make these networks dynamic and stochastic. 00:42:02.890 --> 00:42:08.650 align:middle line:84% So this is a picture of the federal funds market. 00:42:08.650 --> 00:42:11.810 align:middle line:84% And you can see, depending on the time of day, 00:42:11.810 --> 00:42:15.970 align:middle line:84% the evolution of the network where these traders are, again, 00:42:15.970 --> 00:42:18.890 align:middle line:90% are connected with edges. 00:42:18.890 --> 00:42:23.410 align:middle line:84% So in this case, we go back to the basic general equilibrium 00:42:23.410 --> 00:42:28.730 align:middle line:84% idea of keeping track of dates and states of the world, 00:42:28.730 --> 00:42:33.010 align:middle line:84% but in this case, not just for incomes, but also 00:42:33.010 --> 00:42:36.290 align:middle line:90% for who's trading with whom. 00:42:36.290 --> 00:42:39.930 align:middle line:90% So that's a larger state. 00:42:39.930 --> 00:42:43.450 align:middle line:84% And it becomes a risk-sharing problem 00:42:43.450 --> 00:42:47.490 align:middle line:84% to try to stabilize consumption out of income. 00:42:47.490 --> 00:42:51.660 align:middle line:84% And then think about injecting liquidity in this context. 00:42:51.660 --> 00:42:55.380 align:middle line:84% So if extra liquidity were available, say, 00:42:55.380 --> 00:43:00.020 align:middle line:84% from some new monetary policy-- because that is not the standard 00:43:00.020 --> 00:43:03.640 align:middle line:84% way that monetary policy is implemented, 00:43:03.640 --> 00:43:08.700 align:middle line:84% but it could be implemented on these blockchains-- 00:43:08.700 --> 00:43:10.122 align:middle line:90% who should get the liquidity? 00:43:10.122 --> 00:43:11.580 align:middle line:84% And the answer is going to turn out 00:43:11.580 --> 00:43:17.760 align:middle line:84% to be the most valued node or recipient of the liquidity, 00:43:17.760 --> 00:43:21.020 align:middle line:84% assuming that they have to get the money in advance, 00:43:21.020 --> 00:43:25.000 align:middle line:84% would be a trader that participates in market clusters, 00:43:25.000 --> 00:43:30.140 align:middle line:84% even when the number of total participants is low, 00:43:30.140 --> 00:43:34.700 align:middle line:84% when there's aggregate correlated shocks to those 00:43:34.700 --> 00:43:38.820 align:middle line:84% in the cluster where average incomes and yields are 00:43:38.820 --> 00:43:40.780 align:middle line:84% low for those in the cluster, and where 00:43:40.780 --> 00:43:42.400 align:middle line:90% agents are very risk-averse. 00:43:42.400 --> 00:43:46.820 align:middle line:84% So we have a criterion to determine 00:43:46.820 --> 00:43:48.870 align:middle line:90% the most valued players. 00:43:48.870 --> 00:43:52.950 align:middle line:84% You would see this in large financial markets 00:43:52.950 --> 00:43:57.950 align:middle line:84% if we thought of a bond which pays off 00:43:57.950 --> 00:44:00.110 align:middle line:90% when someone is in the market. 00:44:00.110 --> 00:44:03.910 align:middle line:84% An individual I-specific bond-- the value of that bond 00:44:03.910 --> 00:44:06.990 align:middle line:84% would be the value of this key player. 00:44:06.990 --> 00:44:10.190 align:middle line:84% Or in village settings-- and we've done this-- 00:44:10.190 --> 00:44:12.230 align:middle line:84% key players turn out to be people 00:44:12.230 --> 00:44:17.130 align:middle line:84% who are, as the theory says, participating with other agents, 00:44:17.130 --> 00:44:20.350 align:middle line:84% even when there aren't very many transactions. 00:44:20.350 --> 00:44:25.550 align:middle line:84% And those agents turn out to get higher consumption on average, 00:44:25.550 --> 00:44:30.670 align:middle line:84% as if they were receiving a premium, all of which 00:44:30.670 --> 00:44:32.070 align:middle line:90% is informal. 00:44:32.070 --> 00:44:35.910 align:middle line:84% And they don't use that language. 00:44:35.910 --> 00:44:39.950 align:middle line:90% And then a juxtaposition. 00:44:39.950 --> 00:44:42.990 align:middle line:84% There is another literature that seems 00:44:42.990 --> 00:44:45.350 align:middle line:90% to start out the same way-- 00:44:45.350 --> 00:44:49.310 align:middle line:84% financial markets with nodes and interconnectedness-- 00:44:49.310 --> 00:44:54.210 align:middle line:84% but it's a literature about contagion. 00:44:54.210 --> 00:44:59.250 align:middle line:84% There the idea is that traders get hit with an adverse shock, 00:44:59.250 --> 00:45:02.910 align:middle line:84% like a disease, and it's contagious, 00:45:02.910 --> 00:45:05.590 align:middle line:90% and it spreads to other traders. 00:45:05.590 --> 00:45:10.750 align:middle line:84% And that argues for limiting markets. 00:45:10.750 --> 00:45:13.850 align:middle line:84% It's exactly the opposite of the risk-sharing point of view, 00:45:13.850 --> 00:45:18.070 align:middle line:84% where you want to increase the number of traders 00:45:18.070 --> 00:45:22.230 align:middle line:84% in order to allow risk mitigation. 00:45:22.230 --> 00:45:24.470 align:middle line:84% And this picture is just meant to tell you 00:45:24.470 --> 00:45:28.590 align:middle line:84% that these two notions of financially central players 00:45:28.590 --> 00:45:31.630 align:middle line:90% are different from one another. 00:45:31.630 --> 00:45:37.550 align:middle line:84% And I will go into more detail on that later. 00:45:37.550 --> 00:45:41.870 align:middle line:84% Here's two papers, this Contagion! book-- 00:45:41.870 --> 00:45:47.000 align:middle line:84% exclamation mark-- Systemic Risk in-- that's the framework 00:45:47.000 --> 00:45:50.240 align:middle line:90% policymakers are using today. 00:45:50.240 --> 00:45:53.920 align:middle line:84% This is the way they think about financial markets. 00:45:53.920 --> 00:45:57.440 align:middle line:84% And so they're trying to limit the damage that 00:45:57.440 --> 00:46:00.920 align:middle line:84% would happen if something went wrong 00:46:00.920 --> 00:46:04.000 align:middle line:84% rather than the other perspective, which is 00:46:04.000 --> 00:46:07.320 align:middle line:90% to try to make markets stick. 00:46:07.320 --> 00:46:12.500 align:middle line:84% So you can explore the book with its many chapters. 00:46:12.500 --> 00:46:16.640 align:middle line:84% And Martin Sumner has a very nice review piece 00:46:16.640 --> 00:46:18.860 align:middle line:90% about financial contagion. 00:46:18.860 --> 00:46:21.720 align:middle line:90% 00:46:21.720 --> 00:46:23.220 align:middle line:90% OK. 00:46:23.220 --> 00:46:31.120 align:middle line:84% So then we get into mechanism design, and in particular, 00:46:31.120 --> 00:46:36.700 align:middle line:84% incentives to follow protocols and notions of trust. 00:46:36.700 --> 00:46:39.920 align:middle line:90% 00:46:39.920 --> 00:46:44.290 align:middle line:84% So the longer title is Private Information, 00:46:44.290 --> 00:46:48.530 align:middle line:84% Incentives to Report and Take Actions, the Byzantine Generals 00:46:48.530 --> 00:46:51.370 align:middle line:84% Problem, Differences between Economics 00:46:51.370 --> 00:46:55.970 align:middle line:84% and Computer Science Approaches to Trust. 00:46:55.970 --> 00:47:00.210 align:middle line:84% So when we do mechanism design, we're 00:47:00.210 --> 00:47:02.890 align:middle line:84% clear about the incentives to report the truth 00:47:02.890 --> 00:47:11.690 align:middle line:84% and take appropriate actions, as with an insurance 00:47:11.690 --> 00:47:14.770 align:middle line:84% company prepared to cover the loss of a client, 00:47:14.770 --> 00:47:17.210 align:middle line:84% but the loss is private to the agent, who 00:47:17.210 --> 00:47:23.530 align:middle line:84% will be inclined to lie about it in order to get the indemnity, 00:47:23.530 --> 00:47:27.130 align:middle line:84% although there are ways to mitigate that problem. 00:47:27.130 --> 00:47:31.290 align:middle line:84% Implementation looks like we might need that central planner 00:47:31.290 --> 00:47:37.450 align:middle line:84% that I mentioned earlier, implementing the smart contract 00:47:37.450 --> 00:47:43.460 align:middle line:84% code, but we really don't need the central planner at all. 00:47:43.460 --> 00:47:46.380 align:middle line:90% We just need the code. 00:47:46.380 --> 00:47:50.100 align:middle line:84% The code can be validated by agents in advance 00:47:50.100 --> 00:47:53.300 align:middle line:84% with a public commitment to carry out the plan. 00:47:53.300 --> 00:47:57.300 align:middle line:84% The payouts that might be necessary can be put in escrow 00:47:57.300 --> 00:48:01.020 align:middle line:84% and programmed, as with the dynamic assets, 00:48:01.020 --> 00:48:03.460 align:middle line:90% to be available when needed. 00:48:03.460 --> 00:48:07.580 align:middle line:84% Messages about claims can be recorded and put 00:48:07.580 --> 00:48:12.380 align:middle line:84% into this database of the contract. 00:48:12.380 --> 00:48:16.860 align:middle line:84% And implementation does not require implementation 00:48:16.860 --> 00:48:18.340 align:middle line:90% on the blockchain. 00:48:18.340 --> 00:48:21.860 align:middle line:84% This layer 1, layer 2 terminology 00:48:21.860 --> 00:48:27.640 align:middle line:84% has to do with having the code with the smart contract 00:48:27.640 --> 00:48:32.420 align:middle line:84% taking messages from the agents about the state of their ledgers 00:48:32.420 --> 00:48:36.820 align:middle line:84% and assigning new balance sheet positions based 00:48:36.820 --> 00:48:38.400 align:middle line:90% on what they claim. 00:48:38.400 --> 00:48:41.180 align:middle line:90% 00:48:41.180 --> 00:48:44.650 align:middle line:84% And I mentioned at the beginning this pragmatic point of view, 00:48:44.650 --> 00:48:47.110 align:middle line:84% that you don't have to have every piece of the blockchain 00:48:47.110 --> 00:48:50.210 align:middle line:84% in order to make good use of the technology. 00:48:50.210 --> 00:48:53.670 align:middle line:90% This is an example of that. 00:48:53.670 --> 00:48:58.870 align:middle line:84% In contrast, the computer scientists 00:48:58.870 --> 00:49:03.590 align:middle line:84% tend to rely on parallel computing notions of failure. 00:49:03.590 --> 00:49:09.190 align:middle line:84% So the idea is a given computer can fail or give 00:49:09.190 --> 00:49:11.750 align:middle line:84% an incorrect answer, but if you're 00:49:11.750 --> 00:49:15.210 align:middle line:84% able to put a bound on the probability, 00:49:15.210 --> 00:49:19.870 align:middle line:84% then if you have a sufficient number of computers replicating 00:49:19.870 --> 00:49:23.030 align:middle line:84% what should be the same outcome, you can determine 00:49:23.030 --> 00:49:24.790 align:middle line:90% the truth of the matter. 00:49:24.790 --> 00:49:31.030 align:middle line:84% And they use the same framework to think about rogue traders. 00:49:31.030 --> 00:49:38.030 align:middle line:84% That is to say a certain number of evil people 00:49:38.030 --> 00:49:41.590 align:middle line:84% could not follow a protocol, but you 00:49:41.590 --> 00:49:45.930 align:middle line:84% design the protocol that's tolerant to that, 00:49:45.930 --> 00:49:50.990 align:middle line:84% as in Byzantine fault-tolerant validation algorithms. 00:49:50.990 --> 00:49:56.910 align:middle line:84% The Bitcoin with its proof of work is, again, a way to-- 00:49:56.910 --> 00:50:03.510 align:middle line:84% by effectively randomly choosing the proposed truth 00:50:03.510 --> 00:50:09.550 align:middle line:84% of the validation, the odds that you would choose a rogue trader 00:50:09.550 --> 00:50:13.830 align:middle line:84% to do the value can be small if there's 00:50:13.830 --> 00:50:18.390 align:middle line:84% a large number of people trying to do 00:50:18.390 --> 00:50:20.630 align:middle line:90% the proof of work algorithm. 00:50:20.630 --> 00:50:24.550 align:middle line:84% And this bottom part here has to do with the Byzantine generals 00:50:24.550 --> 00:50:30.110 align:middle line:84% problem, having to do with two generals who have to communicate 00:50:30.110 --> 00:50:32.630 align:middle line:90% with one another. 00:50:32.630 --> 00:50:33.730 align:middle line:90% One gets the signal. 00:50:33.730 --> 00:50:35.710 align:middle line:90% The enemy is prepared. 00:50:35.710 --> 00:50:39.640 align:middle line:84% You don't want to attack if the enemy is it's prepared. 00:50:39.640 --> 00:50:42.820 align:middle line:84% But even if the enemy is unprepared, 00:50:42.820 --> 00:50:45.020 align:middle line:84% you have to communicate to the other general. 00:50:45.020 --> 00:50:47.880 align:middle line:84% And these messages may fail to arrive 00:50:47.880 --> 00:50:49.760 align:middle line:90% with a certain probability. 00:50:49.760 --> 00:50:54.880 align:middle line:84% There is a protocol that works quite well if agents follow it, 00:50:54.880 --> 00:51:02.000 align:middle line:84% which is for the second general to attack 00:51:02.000 --> 00:51:03.960 align:middle line:90% under certain conditions. 00:51:03.960 --> 00:51:08.280 align:middle line:84% I won't go into the details, but that protocol 00:51:08.280 --> 00:51:14.320 align:middle line:84% is vulnerable to the economic concept of incentives 00:51:14.320 --> 00:51:18.600 align:middle line:90% and Bayesian Nash equilibrium. 00:51:18.600 --> 00:51:22.260 align:middle line:84% Traders can convince themselves to never attack, actually, 00:51:22.260 --> 00:51:26.840 align:middle line:84% even when the enemy is unprepared 00:51:26.840 --> 00:51:29.440 align:middle line:84% because they're second-guessing and third-guessing 00:51:29.440 --> 00:51:31.720 align:middle line:90% what the other guy is doing. 00:51:31.720 --> 00:51:34.800 align:middle line:84% So we'll talk about that more in lecture 8. 00:51:34.800 --> 00:51:38.330 align:middle line:84% Here are two articles on computing and mechanism 00:51:38.330 --> 00:51:42.450 align:middle line:84% design and distributed ledgers and the governance of money 00:51:42.450 --> 00:51:46.330 align:middle line:84% that both have to do with computing and validation 00:51:46.330 --> 00:51:47.110 align:middle line:90% algorithms. 00:51:47.110 --> 00:51:51.770 align:middle line:84% So again, these are options for you 00:51:51.770 --> 00:51:56.410 align:middle line:84% to consider when you're thinking about reporting in or doing 00:51:56.410 --> 00:51:58.690 align:middle line:90% your research proposal. 00:51:58.690 --> 00:52:04.430 align:middle line:84% So we finally get, in lecture 9, to the third leg, so to speak, 00:52:04.430 --> 00:52:05.870 align:middle line:90% of these new technologies. 00:52:05.870 --> 00:52:10.530 align:middle line:84% We covered distributed ledgers and smart contracts. 00:52:10.530 --> 00:52:14.170 align:middle line:90% And now, we get to encryption. 00:52:14.170 --> 00:52:17.210 align:middle line:84% This little picture over there is 00:52:17.210 --> 00:52:21.290 align:middle line:84% meant to remind you that keeping secrets has been 00:52:21.290 --> 00:52:23.650 align:middle line:90% with us for a very long time. 00:52:23.650 --> 00:52:30.010 align:middle line:84% In Mesopotamia, they used to create tokens 00:52:30.010 --> 00:52:34.130 align:middle line:84% as an invoice of what's in the shipment. 00:52:34.130 --> 00:52:37.780 align:middle line:84% And they put those tokens in a clay envelope 00:52:37.780 --> 00:52:42.540 align:middle line:84% and baked it hard and then wrote little pictures 00:52:42.540 --> 00:52:47.883 align:middle line:84% of the internal tokens that accompanied the shipment. 00:52:47.883 --> 00:52:49.300 align:middle line:84% When you think about it, you begin 00:52:49.300 --> 00:52:54.020 align:middle line:84% to realize that it would be very, very difficult for someone 00:52:54.020 --> 00:53:00.500 align:middle line:84% to steal the shipment without the knowledge of the recipient 00:53:00.500 --> 00:53:03.020 align:middle line:90% when it finally gets there. 00:53:03.020 --> 00:53:06.740 align:middle line:84% So it's a bit like private and public keys, 00:53:06.740 --> 00:53:11.160 align:middle line:84% which are the centerpiece now of modern encryption. 00:53:11.160 --> 00:53:16.780 align:middle line:84% It allows people to send a secret message. 00:53:16.780 --> 00:53:19.940 align:middle line:84% And the recipient can decipher it. 00:53:19.940 --> 00:53:25.740 align:middle line:84% For outgoing messages, it allows a receiver to authenticate 00:53:25.740 --> 00:53:29.820 align:middle line:84% the origin of the message, know who sent it so the sender cannot 00:53:29.820 --> 00:53:35.900 align:middle line:84% repudiate it, and verify the message has not been modified. 00:53:35.900 --> 00:53:42.380 align:middle line:84% So these are very powerful tools related 00:53:42.380 --> 00:53:47.380 align:middle line:84% to hashes and cryptographic puzzles, 00:53:47.380 --> 00:53:53.540 align:middle line:84% also to fully homomorphic encryption, which, 00:53:53.540 --> 00:53:57.020 align:middle line:84% with the math of it, is like setting up parallel spaces that 00:53:57.020 --> 00:54:08.740 align:middle line:84% are one-to-one, the actual text versus the encoded system. 00:54:08.740 --> 00:54:13.180 align:middle line:84% So you can do addition and multiplication 00:54:13.180 --> 00:54:17.200 align:middle line:84% on these encrypted systems and end up with the correct answer, 00:54:17.200 --> 00:54:24.020 align:middle line:84% even without knowing the actual inputs. 00:54:24.020 --> 00:54:27.500 align:middle line:84% And multiparty computation is another leg 00:54:27.500 --> 00:54:35.210 align:middle line:84% of this including zero-knowledge proofs, which we will go into. 00:54:35.210 --> 00:54:38.470 align:middle line:90% 00:54:38.470 --> 00:54:42.110 align:middle line:84% And these are papers, one by this Harvard professor 00:54:42.110 --> 00:54:45.310 align:middle line:84% on zero-knowledge mechanisms, which 00:54:45.310 --> 00:54:49.390 align:middle line:90% is a mechanism design problem. 00:54:49.390 --> 00:54:52.350 align:middle line:90% And again, back to-- 00:54:52.350 --> 00:54:57.950 align:middle line:84% see, in practice, chain link, layer zero, and this new Dfinity 00:54:57.950 --> 00:55:03.030 align:middle line:84% internet computer in your readings to see. 00:55:03.030 --> 00:55:06.470 align:middle line:84% So these are actual implementations 00:55:06.470 --> 00:55:11.070 align:middle line:84% that industry and central banks are adopting. 00:55:11.070 --> 00:55:16.830 align:middle line:84% Then in lecture 10, we will go through designs using encryption 00:55:16.830 --> 00:55:19.850 align:middle line:90% for auctions, for example. 00:55:19.850 --> 00:55:22.270 align:middle line:84% We can encrypt the bids and nevertheless determine 00:55:22.270 --> 00:55:25.110 align:middle line:90% the winner. 00:55:25.110 --> 00:55:27.870 align:middle line:84% We don't need some third-party auctioneer 00:55:27.870 --> 00:55:32.560 align:middle line:90% who is in a position to cheat. 00:55:32.560 --> 00:55:37.600 align:middle line:84% Back to that insurance example with agents experiencing shocks 00:55:37.600 --> 00:55:41.040 align:middle line:84% to their balance sheets, which they would like to have insured 00:55:41.040 --> 00:55:44.640 align:middle line:84% but don't want to reveal to third parties 00:55:44.640 --> 00:55:51.200 align:middle line:84% the state of their balances, we will show how to handle that. 00:55:51.200 --> 00:55:54.200 align:middle line:84% And the information here gets scrambled up 00:55:54.200 --> 00:55:56.320 align:middle line:90% so no one has the whole picture. 00:55:56.320 --> 00:56:00.540 align:middle line:84% And yet somehow, amazingly, the code can do what you want, 00:56:00.540 --> 00:56:04.360 align:middle line:84% including randomize at the crucial key moments, 00:56:04.360 --> 00:56:07.040 align:middle line:84% even though the code, quote unquote, "is not a person," 00:56:07.040 --> 00:56:09.960 align:middle line:84% and does not know the underlying true state. 00:56:09.960 --> 00:56:13.680 align:middle line:84% So you get these miracles happening 00:56:13.680 --> 00:56:17.200 align:middle line:84% using fully homomorphic encryption 00:56:17.200 --> 00:56:20.160 align:middle line:90% and multiparty computation. 00:56:20.160 --> 00:56:27.200 align:middle line:84% And you can use this also for centralized matching of agents 00:56:27.200 --> 00:56:31.030 align:middle line:84% who submit supply and demand schedules, 00:56:31.030 --> 00:56:31.950 align:middle line:90% but they're encrypted. 00:56:31.950 --> 00:56:35.210 align:middle line:84% And nevertheless, you can basically 00:56:35.210 --> 00:56:38.810 align:middle line:84% intersect supply and demand schedules in the encrypted space 00:56:38.810 --> 00:56:47.170 align:middle line:84% and figure out the equilibrium outcome. 00:56:47.170 --> 00:56:51.150 align:middle line:84% And you can put encryption, as in these papers, 00:56:51.150 --> 00:56:56.690 align:middle line:84% including this work that MIT is doing with Visa on encryption 00:56:56.690 --> 00:56:59.970 align:middle line:90% in the blockchain. 00:56:59.970 --> 00:57:05.050 align:middle line:84% And finally, we get lecture 11, which has 00:57:05.050 --> 00:57:08.090 align:middle line:90% to do with this juxtaposition-- 00:57:08.090 --> 00:57:12.290 align:middle line:84% on the one hand, Dubey, "Price-Quantity Strategic Market 00:57:12.290 --> 00:57:15.570 align:middle line:84% Games," which is a classic economics paper 00:57:15.570 --> 00:57:20.530 align:middle line:84% on an implementation that gets you to the Walrasian outcome. 00:57:20.530 --> 00:57:23.610 align:middle line:84% And on the other hand, we have SPEEDEX 00:57:23.610 --> 00:57:29.170 align:middle line:84% as a decentralized exchange that bears not a small amount 00:57:29.170 --> 00:57:31.410 align:middle line:90% of similarity to Dubey. 00:57:31.410 --> 00:57:35.530 align:middle line:84% And we need to put those things together somehow and deal 00:57:35.530 --> 00:57:39.130 align:middle line:84% with the fact that we have many, many agents, each 00:57:39.130 --> 00:57:41.170 align:middle line:90% with their strategies. 00:57:41.170 --> 00:57:45.370 align:middle line:84% And it is typically hard to find the Nash equilibrium 00:57:45.370 --> 00:57:47.410 align:middle line:90% that we're looking for. 00:57:47.410 --> 00:57:50.850 align:middle line:84% It's computationally complex, certainly 00:57:50.850 --> 00:57:53.970 align:middle line:84% at the level of individual agents. 00:57:53.970 --> 00:57:56.010 align:middle line:84% So this ongoing work-- the idea is 00:57:56.010 --> 00:58:02.690 align:middle line:84% to allow the algorithm to compute the [? ideal ?] 00:58:02.690 --> 00:58:04.810 align:middle line:84% candidate for the equilibrium and then 00:58:04.810 --> 00:58:08.870 align:middle line:84% signal to the agents what their strategy ought to be. 00:58:08.870 --> 00:58:12.690 align:middle line:84% And they know that the mechanism operates that way, 00:58:12.690 --> 00:58:15.850 align:middle line:84% so they don't have to have endless conjectures 00:58:15.850 --> 00:58:21.230 align:middle line:84% and what-if statements about what the other agents are doing. 00:58:21.230 --> 00:58:26.580 align:middle line:84% So that work is based in part on Daskalakis. 00:58:26.580 --> 00:58:30.900 align:middle line:84% And Costas is, as you guys know, a computer science professor 00:58:30.900 --> 00:58:32.420 align:middle line:90% here. 00:58:32.420 --> 00:58:36.300 align:middle line:90% So those are the 11 lectures. 00:58:36.300 --> 00:58:40.020 align:middle line:84% I hope you did not find it too bewildering. 00:58:40.020 --> 00:58:43.860 align:middle line:84% It is hard to go through it without getting 00:58:43.860 --> 00:58:45.120 align:middle line:90% into the details. 00:58:45.120 --> 00:58:51.500 align:middle line:84% And yet we always move on to yet another lecture. 00:58:51.500 --> 00:58:57.722 align:middle line:84% I was taught somewhere that if you're teaching well, 00:58:57.722 --> 00:58:59.180 align:middle line:84% you tell students what you're going 00:58:59.180 --> 00:59:03.940 align:middle line:84% to do and then do it, and then tell them what you did. 00:59:03.940 --> 00:59:08.890 align:middle line:84% So this was in the spirit of here's what we're going to do. 00:59:08.890 --> 00:59:25.000 align:middle line:90%