WEBVTT 00:00:00.000 --> 00:00:02.259 We need to talk about alchemy. Alchemy, okay. 00:00:02.399 --> 00:00:05.240 You mean like the old medieval pseudoscience? 00:00:05.320 --> 00:00:07.559 Exactly. People in robes, you know, dusty labs 00:00:07.559 --> 00:00:09.880 trying to turn lead into gold. Right. But I'm 00:00:09.880 --> 00:00:12.000 not talking about that kind. Not the potions 00:00:12.000 --> 00:00:15.539 and wizards kind. I'm talking about modern alchemy. 00:00:15.619 --> 00:00:17.940 Okay. The kind that happens in a glass office 00:00:17.940 --> 00:00:21.059 in Manhattan, 50 stories up. The kind where you 00:00:21.059 --> 00:00:24.280 take a pile of absolute garbage, I mean, loans 00:00:24.280 --> 00:00:26.399 that are never, ever going to be paid back, and 00:00:26.399 --> 00:00:28.820 you wave a mathematical wand over them, and suddenly... 00:00:28.820 --> 00:00:31.140 They're gold. They're gold. They're AAA rated. 00:00:31.239 --> 00:00:34.020 They are, on paper, the safest investment in 00:00:34.020 --> 00:00:36.759 the entire world. And then, inevitably, the spell 00:00:36.759 --> 00:00:41.159 breaks. And the world blows up. Welcome to the 00:00:41.159 --> 00:00:44.520 Deep Dive. Today, we are tackling the villain 00:00:44.520 --> 00:00:47.939 of the 2008 story, the smoking gun. The three 00:00:47.939 --> 00:00:49.840 letters that just keep coming up in every single 00:00:49.840 --> 00:00:52.020 documentary, every book, every lawsuit. Every 00:00:52.020 --> 00:00:54.179 angry conversation about the financial crisis. 00:00:54.280 --> 00:00:57.340 The CDO. The collateralized debt obligation. 00:00:57.799 --> 00:01:00.280 Warren Buffett didn't mince words about these 00:01:00.280 --> 00:01:03.020 things. He famously called derivatives like these 00:01:03.020 --> 00:01:05.680 financial weapons of mass destruction. He did. 00:01:05.840 --> 00:01:07.939 And it's a phrase that gets quoted all the time. 00:01:08.269 --> 00:01:10.170 And usually when people use phrases like that, 00:01:10.209 --> 00:01:11.810 they're being dramatic, right, for headlines. 00:01:12.170 --> 00:01:14.790 But in this case, looking at the source material 00:01:14.790 --> 00:01:17.069 we have, the Financial Crisis Inquiry Report, 00:01:17.250 --> 00:01:20.230 the really deep investigative work from journalists 00:01:20.230 --> 00:01:23.989 like Michael Lewis, Bethany McLean, I mean, he 00:01:23.989 --> 00:01:27.109 was barely scratching the surface. It is the 00:01:27.109 --> 00:01:30.290 defining acronym of a decade. But what's so fascinating, 00:01:30.489 --> 00:01:32.950 I think, is that the reputation, that villain 00:01:32.950 --> 00:01:36.450 status. it often obscures what the thing actually 00:01:36.450 --> 00:01:39.189 is. Totally. It's so easy to just call it a weapon. 00:01:39.329 --> 00:01:41.870 It is so much harder to understand the mechanics 00:01:41.870 --> 00:01:44.430 of why it was built in the first place. And that 00:01:44.430 --> 00:01:46.329 is our mission today. We are going to move past 00:01:46.329 --> 00:01:48.760 the buzzwords. We're going to ignore the stigma 00:01:48.760 --> 00:01:50.459 for a minute and just look at the mechanics because 00:01:50.459 --> 00:01:52.019 and this is where it gets really interesting. 00:01:52.219 --> 00:01:54.980 This wasn't designed to destroy the economy. 00:01:55.120 --> 00:01:57.519 No, not at all. It was designed to manage risk. 00:01:57.560 --> 00:02:00.540 To manage risk. Precisely. It was a genuine attempt 00:02:00.540 --> 00:02:05.599 to use math and financial structure to make investing 00:02:05.599 --> 00:02:08.580 safer, more efficient. And the tragedy or maybe 00:02:08.580 --> 00:02:11.159 the irony is how a tool designed to disperse 00:02:11.159 --> 00:02:14.080 risk ended up concentrating it. Concentrating 00:02:14.080 --> 00:02:16.729 it to the point of. Well, systemic failure. Global 00:02:16.729 --> 00:02:19.009 systemic failure. It really is a story about 00:02:19.009 --> 00:02:21.030 alchemy. We have these documents and they all 00:02:21.030 --> 00:02:24.250 paint this picture of bankers, rating agencies, 00:02:24.409 --> 00:02:27.750 investors, all sort of holding hands and agreeing 00:02:27.750 --> 00:02:29.990 to believe in a fantasy because the money was 00:02:29.990 --> 00:02:31.909 just too good. It's a story about incentives. 00:02:32.129 --> 00:02:35.120 It's a story about a blind faith in models. And 00:02:35.120 --> 00:02:37.180 maybe most importantly, it's about what happens 00:02:37.180 --> 00:02:39.360 when you completely separate the person lending 00:02:39.360 --> 00:02:41.680 the money from the person who's actually taking 00:02:41.680 --> 00:02:43.740 the risk. OK, so let's start at the very bottom. 00:02:43.780 --> 00:02:45.919 Before we get into the crash and the drama, we 00:02:45.919 --> 00:02:47.919 have to define the beast. If you're listening 00:02:47.919 --> 00:02:50.000 and you've heard CDO a thousand times but never 00:02:50.000 --> 00:02:52.780 really looked under the hood, what in the most 00:02:52.780 --> 00:02:56.039 basic sense is a collateralized debt obligation. 00:02:56.159 --> 00:03:00.099 OK. At its heart, a CDO is a type of what's called 00:03:00.099 --> 00:03:02.860 a structured asset -backed security. Now, that 00:03:02.860 --> 00:03:04.819 sounds technical, but just think of it as a promise. 00:03:05.000 --> 00:03:08.539 A promise. It is a promise to pay investors back, 00:03:08.719 --> 00:03:12.080 but in a very specific sequence. And that payment 00:03:12.080 --> 00:03:14.599 is based on the cash flow that's collected from 00:03:14.599 --> 00:03:16.840 a big pool of assets. Okay, pool of assets. That's 00:03:16.840 --> 00:03:18.960 the collateral part of the name, right? That's 00:03:18.960 --> 00:03:22.639 the C. Correct. A CDO is essentially a box. You 00:03:22.639 --> 00:03:24.789 take a whole bunch of debt. And originally, this 00:03:24.789 --> 00:03:27.430 was things like corporate debt, bank loans, stuff 00:03:27.430 --> 00:03:30.750 relatively. But as we'll see, it eventually became 00:03:30.750 --> 00:03:33.889 completely dominated by mortgage backed securities 00:03:33.889 --> 00:03:37.169 or MBS. You put all these loans into the box. 00:03:37.270 --> 00:03:40.590 Those loans pay interest every month. That interest 00:03:40.590 --> 00:03:43.569 is the cash flow. So if I'm an investor, I give 00:03:43.569 --> 00:03:45.770 you money. You put it in this box with a bunch 00:03:45.770 --> 00:03:47.370 of other people's money. You go out and buy a 00:03:47.370 --> 00:03:49.370 thousand mortgages. And when those homeowners 00:03:49.370 --> 00:03:53.449 pay their bills. That is the basic idea, yes. 00:03:54.189 --> 00:03:56.610 But there's a crucial twist. If it were just 00:03:56.610 --> 00:03:59.250 a pool where everyone shared the risk equally, 00:03:59.449 --> 00:04:01.449 it would just be a mutual fund, basically. Right. 00:04:01.550 --> 00:04:04.349 A CDO is different. It's sliced into sections. 00:04:04.409 --> 00:04:06.550 And these sections are called tranches. Tranches. 00:04:06.550 --> 00:04:08.270 That's the French word for slice, right? Like 00:04:08.270 --> 00:04:10.849 a slice of cake. It is. And to understand why 00:04:10.849 --> 00:04:13.729 these exist, we have to use the analogy that 00:04:13.729 --> 00:04:15.849 comes up in almost every single explanation because 00:04:15.849 --> 00:04:18.649 it is just so accurate. We have to talk about 00:04:18.649 --> 00:04:22.699 the waterfall. Ah, yes. The water flowing into 00:04:22.699 --> 00:04:25.519 cups analogy. I want to slow down here because 00:04:25.519 --> 00:04:27.879 this is really the mechanism. This is the piece 00:04:27.879 --> 00:04:30.500 of financial engineering that fooled everyone. 00:04:30.680 --> 00:04:33.379 It absolutely is. So imagine the cash coming 00:04:33.379 --> 00:04:35.879 from all those mortgages, all those monthly payments 00:04:35.879 --> 00:04:39.759 is water. It flows out of the pool of loans and 00:04:39.759 --> 00:04:42.920 it hits a stack of cups. But the cups aren't 00:04:42.920 --> 00:04:44.920 side by side where they'd all fill up at the 00:04:44.920 --> 00:04:47.220 same time. They're stacked vertically. One on 00:04:47.220 --> 00:04:49.120 top of the other. Exactly. So you've got a top 00:04:49.120 --> 00:04:51.500 cup, a middle cup, and a bottom cup. And gravity 00:04:51.500 --> 00:04:53.920 does all the work. The water flows from the pipe 00:04:53.920 --> 00:04:55.800 and it goes into the top cup first. Always the 00:04:55.800 --> 00:04:58.699 top cup. Always. This represents the senior tranches. 00:04:58.779 --> 00:05:01.040 These are the investors who are first in line 00:05:01.040 --> 00:05:03.459 to get paid. Because they get the water first, 00:05:03.540 --> 00:05:06.160 they are the safest. Even if the flow of water 00:05:06.160 --> 00:05:08.540 slows down a little bit, say a few homeowners 00:05:08.540 --> 00:05:10.980 stop paying their mortgages, the top cup will 00:05:10.980 --> 00:05:14.120 probably still get full. So cup A, the top cup, 00:05:14.240 --> 00:05:17.810 is safe. It's boring. This is for, I don't know, 00:05:17.870 --> 00:05:21.209 grandma's pension fund. Precisely. Because it's 00:05:21.209 --> 00:05:23.990 so safe, it pays the lowest interest rate. But, 00:05:23.990 --> 00:05:27.069 and this is key, it gets that coveted AAA rating 00:05:27.069 --> 00:05:29.889 from the credit agencies. It is the last to lose. 00:05:30.230 --> 00:05:33.339 Okay, so my boring cup is full of water. What 00:05:33.339 --> 00:05:36.040 happens next? Well, once that top cup is completely 00:05:36.040 --> 00:05:39.019 full, the water spills over the rim and starts 00:05:39.019 --> 00:05:41.220 to flow into the next cup down. This would be 00:05:41.220 --> 00:05:43.360 the mezzanine tranches. Right. The middle ground. 00:05:43.620 --> 00:05:46.360 These are rated anywhere from AA all the way 00:05:46.360 --> 00:05:49.620 down to BB. Now, these investors only get paid 00:05:49.620 --> 00:05:52.300 if the senior guys are fully satisfied. So they 00:05:52.300 --> 00:05:54.360 take on more risk. Because if the water flow 00:05:54.360 --> 00:05:56.500 slows down a lot, it might run out before it 00:05:56.500 --> 00:05:58.480 ever reaches their cup. Exactly. It might just 00:05:58.480 --> 00:06:00.759 trickle in or not get there at all. So because 00:06:00.759 --> 00:06:02.600 they're taking more risk, they demand a higher 00:06:02.600 --> 00:06:04.939 yield. They want more profit for that uncertainty. 00:06:05.060 --> 00:06:07.639 Makes sense. And then at the very, very bottom 00:06:07.639 --> 00:06:09.980 of the stack. You have the equity tranche. The 00:06:09.980 --> 00:06:12.439 toxic waste. Or the gold mine, I guess, depending 00:06:12.439 --> 00:06:14.740 on your perspective. Both are accurate descriptions. 00:06:15.220 --> 00:06:18.279 It's sometimes called the residual or the first 00:06:18.279 --> 00:06:21.639 loss piece. This is the bottom cup. It only gets 00:06:21.639 --> 00:06:23.920 water if everyone above it is completely full. 00:06:24.060 --> 00:06:27.779 So if the water flow slows down, even a tiny 00:06:27.779 --> 00:06:30.720 bit, if just a few percent of homeowners default, 00:06:31.040 --> 00:06:34.779 this cup stays bone dry. You're wiped out. These 00:06:34.779 --> 00:06:37.000 investors get wiped out first. It's designed 00:06:37.000 --> 00:06:40.160 that way. But there's always a but. But if the 00:06:40.160 --> 00:06:42.620 water is flowing at full force, if everyone pays 00:06:42.620 --> 00:06:45.019 their mortgage on time, this bottom cup gets 00:06:45.019 --> 00:06:47.019 everything that's left over. And the upside can 00:06:47.019 --> 00:06:49.560 be uncapped. You could make 20, 30 percent returns 00:06:49.560 --> 00:06:51.699 in a world where safe government bonds are paying 00:06:51.699 --> 00:06:54.259 5 percent. So theoretically, you've created a 00:06:54.259 --> 00:06:56.860 product for everyone. You have this super safe 00:06:56.860 --> 00:06:59.579 senior slice. for the pension funds who are legally 00:06:59.579 --> 00:07:01.980 required to hold safe assets. You've got the 00:07:01.980 --> 00:07:03.959 middle slices for people with a bit more appetite 00:07:03.959 --> 00:07:06.240 for risk. And then you have the equity slice 00:07:06.240 --> 00:07:08.120 for the hedge funds who really want to gamble 00:07:08.120 --> 00:07:11.000 for those massive returns. That is the theoretical 00:07:11.000 --> 00:07:14.319 brilliance of the CDO. You take one big pool 00:07:14.319 --> 00:07:16.699 of loans, which might be a mix of good and bad, 00:07:16.800 --> 00:07:19.500 and by prioritizing the payments through this 00:07:19.500 --> 00:07:22.660 waterfall structure, you create a safe asset 00:07:22.660 --> 00:07:25.720 out of it. You're not eliminating risk. No, you're 00:07:25.720 --> 00:07:27.959 not eliminating it. You are slicing it up and 00:07:27.959 --> 00:07:30.579 selling it to the people who theoretically want 00:07:30.579 --> 00:07:33.040 to hold that specific amount of risk. It sounds 00:07:33.040 --> 00:07:35.040 so logical when you say it like that. It sounds 00:07:35.040 --> 00:07:36.860 like it creates efficiency in the market. It's 00:07:36.860 --> 00:07:40.100 supposed to. But obviously got complicated. Let's 00:07:40.100 --> 00:07:41.839 talk about the cast of characters involved here, 00:07:41.879 --> 00:07:43.540 because it's not just a bank and an investor. 00:07:44.060 --> 00:07:46.939 It's a whole ecosystem of people. And they're 00:07:46.939 --> 00:07:49.660 all taking fees at every single step of the way. 00:07:49.779 --> 00:07:52.660 It is a highly specialized assembly line. And 00:07:52.660 --> 00:07:54.420 you have to understand all the players to see 00:07:54.420 --> 00:07:56.699 where things went wrong. First, you have the 00:07:56.699 --> 00:07:59.160 issuer. This is usually what's called a special 00:07:59.160 --> 00:08:02.399 purpose entity or an SPE. A separate legal robot. 00:08:02.660 --> 00:08:05.019 Essentially, yes. The bank doesn't issue the 00:08:05.019 --> 00:08:07.620 CDO directly from its own balance sheet. They 00:08:07.620 --> 00:08:10.079 create a shell company, usually incorporated 00:08:10.079 --> 00:08:13.199 in the Cayman Islands. I always see the Cayman 00:08:13.199 --> 00:08:15.220 Islands mentioned in these stories. Is that just 00:08:15.220 --> 00:08:17.449 because of... bankers like the beach? It would 00:08:17.449 --> 00:08:21.430 be a nice perk, but no. It's purely for tax neutrality 00:08:21.430 --> 00:08:23.850 and what's called bankruptcy remoteness. Okay, 00:08:23.910 --> 00:08:26.860 break that down. Tax neutrality. By setting up 00:08:26.860 --> 00:08:30.579 in the Caymans, the CDO avoids paying U .S. federal 00:08:30.579 --> 00:08:33.679 income tax on its global income. It's a pass 00:08:33.679 --> 00:08:35.740 -through vehicle. It makes it cleaner for global 00:08:35.740 --> 00:08:38.500 investors. And bankruptcy remoteness. That means 00:08:38.500 --> 00:08:41.799 if the parent banks say Merrill Lynch goes bankrupt, 00:08:42.220 --> 00:08:46.059 its creditors can't come after the assets inside 00:08:46.059 --> 00:08:49.580 the CDO. The CDO is legally separate. It's remote. 00:08:49.820 --> 00:08:52.519 It protects the CDO investors from the bank's 00:08:52.519 --> 00:08:54.600 problems. And it protects the bank's balance 00:08:54.600 --> 00:08:57.419 sheet from the CDO. That's the idea. It holds 00:08:57.419 --> 00:08:59.519 the assets so the parent bank doesn't have to. 00:08:59.679 --> 00:09:02.059 OK, so we hear the Cayman robot who's actually 00:09:02.059 --> 00:09:04.220 running the show. Who's picking the loans? That 00:09:04.220 --> 00:09:06.460 would be the asset manager. This is the person 00:09:06.460 --> 00:09:09.100 or the firm who is responsible for selecting 00:09:09.100 --> 00:09:11.320 the loans or the bonds that go into the box. 00:09:11.440 --> 00:09:13.679 They're the stock pickers, essentially. They're 00:09:13.679 --> 00:09:16.379 the bond pickers. Yeah. Their job is to manage 00:09:16.379 --> 00:09:18.559 a portfolio. And for that, they earn fees. They 00:09:18.559 --> 00:09:22.059 get senior fees, subordinated fees. They're supposed 00:09:22.059 --> 00:09:24.120 to be the experts, the ones picking the winners. 00:09:24.480 --> 00:09:27.340 Supposed to be. But as we'll see later, they're 00:09:27.340 --> 00:09:29.360 often just picking whatever the bank told them 00:09:29.360 --> 00:09:31.879 to pick to get a deal done. Or whatever was available, 00:09:32.120 --> 00:09:35.120 which became a huge problem. Next, you have the 00:09:35.120 --> 00:09:37.440 underwriter. The big investment bank. Usually, 00:09:37.440 --> 00:09:41.649 yes. A Merrill Lynch, a Citigroup. a bear stearns 00:09:41.649 --> 00:09:44.590 they are the architects they structure the deal 00:09:44.590 --> 00:09:47.129 they decide how big the slices are what the interest 00:09:47.129 --> 00:09:49.710 rates will be and most importantly they sell 00:09:49.710 --> 00:09:52.509 those slices to investors they are the salesmen 00:09:52.509 --> 00:09:54.450 and we can't forget the gatekeepers the ones 00:09:54.450 --> 00:09:58.340 who made the alchemy seem Ah, yes. The rating 00:09:58.340 --> 00:10:02.340 agencies, Moody's, S &P, Fitch, they are absolutely 00:10:02.340 --> 00:10:04.940 critical to this entire process. Why so critical? 00:10:05.080 --> 00:10:06.759 Because they look at the tranches and they assign 00:10:06.759 --> 00:10:10.080 a grade. AAA for the top cup, AA for the next, 00:10:10.200 --> 00:10:12.100 all the way down to junk for the very bottom. 00:10:12.320 --> 00:10:14.840 Without their stamp of approval, especially that 00:10:14.840 --> 00:10:17.320 AAA stamp, the senior tranches could not be sold. 00:10:17.720 --> 00:10:20.100 Because many of the biggest investors, like pension 00:10:20.100 --> 00:10:23.500 funds and insurance companies, have rules. They 00:10:23.500 --> 00:10:25.580 are only allowed to buy assets that are rated 00:10:25.580 --> 00:10:28.700 AAA. So without that rating, the biggest pool 00:10:28.700 --> 00:10:31.240 of money in the world is off limits. So the rating 00:10:31.240 --> 00:10:33.519 agencies held the keys to the kingdom. They did. 00:10:33.759 --> 00:10:37.210 And finally, you have the trustee. The trustee 00:10:37.210 --> 00:10:39.830 sounds important. They're basically the accountant, 00:10:40.029 --> 00:10:43.070 a bank usually. They hold the legal title to 00:10:43.070 --> 00:10:44.970 the assets in the pool, and they are the ones 00:10:44.970 --> 00:10:47.309 who run the waterfall models every month to figure 00:10:47.309 --> 00:10:49.730 out who gets paid what and in what order. So 00:10:49.730 --> 00:10:52.090 that's the anatomy. It's a machine. A machine 00:10:52.090 --> 00:10:54.429 designed to process debt and spit out securities. 00:10:54.889 --> 00:10:57.509 But before it became this weapon of mass destruction, 00:10:57.830 --> 00:10:59.710 it was actually seen as a good thing, right? 00:11:00.120 --> 00:11:02.080 Let's get into the origin story. Yeah, it wasn't 00:11:02.080 --> 00:11:04.779 always evil. In fact, the precursors to the CDO 00:11:04.779 --> 00:11:06.679 were actually government -sponsored programs. 00:11:06.820 --> 00:11:09.419 We have to go back to the 1970s. Okay. You have 00:11:09.419 --> 00:11:13.279 Ginnie Mae in 1970 and Freddie Mac in 1971. They 00:11:13.279 --> 00:11:16.000 created the very first mortgage -backed securities. 00:11:16.600 --> 00:11:18.919 And the goal there was noble, right? It was to 00:11:18.919 --> 00:11:21.820 help. liquidity in the housing market. Exactly. 00:11:22.399 --> 00:11:25.559 The idea was if a local bank in Ohio can sell 00:11:25.559 --> 00:11:27.779 its mortgages to Ginnie Mae, it gets cash back 00:11:27.779 --> 00:11:30.159 immediately. And with that cash, it can then 00:11:30.159 --> 00:11:32.799 lend to more people to buy homes. It was all 00:11:32.799 --> 00:11:35.820 about expanding the American dream of homeownership. 00:11:35.840 --> 00:11:37.379 It greased the wheels of the housing market. 00:11:37.539 --> 00:11:40.019 It did. But of course, the private market wanted 00:11:40.019 --> 00:11:43.700 in on the action. In 1977, Salomon Brothers created 00:11:43.700 --> 00:11:47.809 the first private label, MBS. It actually failed 00:11:47.809 --> 00:11:49.690 at first. It was a bit ahead of its time. And 00:11:49.690 --> 00:11:51.230 this is when those famous names get involved. 00:11:51.350 --> 00:11:54.409 Yes. You have legends like Louis Ranieri at Salomon 00:11:54.409 --> 00:11:58.070 Brothers and Larry Fink, who is now, of course, 00:11:58.149 --> 00:12:00.009 the head of BlackRock, the biggest asset manager 00:12:00.009 --> 00:12:03.190 in the world. They are credited with really inventing 00:12:03.190 --> 00:12:05.409 the business of securitization as we know it, 00:12:05.450 --> 00:12:07.549 pooling mortgages and slicing them up. So when 00:12:07.549 --> 00:12:11.029 did the first official CDO with that name appear? 00:12:11.529 --> 00:12:14.350 That showed up in 1987. It was issued by Drexel 00:12:14.350 --> 00:12:16.970 Burnham Lambert. Oh, Michael Milken's firm, the 00:12:16.970 --> 00:12:19.629 Junk Bond Kings. It's the very same. They issued 00:12:19.629 --> 00:12:22.070 it for a bank called Imperial Savings Association. 00:12:22.710 --> 00:12:27.289 1987. So this technology, this basic idea, was 00:12:27.289 --> 00:12:29.830 around for a full 20 years before the crisis. 00:12:30.429 --> 00:12:32.990 It wasn't some new invention in 2005. Not at 00:12:32.990 --> 00:12:36.070 all. And through the 1990s and the early 2000s, 00:12:36.070 --> 00:12:38.649 we had what you might call the good CDO era. 00:12:39.250 --> 00:12:42.470 And the key word during this period was diversification. 00:12:42.570 --> 00:12:45.120 Right. Because early on, they weren't just stuffing 00:12:45.120 --> 00:12:47.200 these things with mortgages from one place. No, 00:12:47.240 --> 00:12:49.559 absolutely not. They were multi -sector CDOs. 00:12:49.639 --> 00:12:52.139 The collateral pools were incredibly diverse. 00:12:52.220 --> 00:12:54.440 They'd have aircraft lease debt in there. So 00:12:54.440 --> 00:12:57.000 loans to airlines to buy planes. Exactly. They 00:12:57.000 --> 00:12:59.100 had loans for manufactured housing. They had 00:12:59.100 --> 00:13:02.639 student loans, credit card debt, corporate bonds 00:13:02.639 --> 00:13:05.039 from all different industries. And the theory, 00:13:05.139 --> 00:13:07.419 the sales pitch, was that... All these things 00:13:07.419 --> 00:13:09.779 were uncorrelated. Precisely. The safety theory 00:13:09.779 --> 00:13:12.980 was if the airline industry crashes because of, 00:13:12.980 --> 00:13:15.460 say, a pandemic, people are probably still paying 00:13:15.460 --> 00:13:17.700 their student loans. If the manufacturing sector 00:13:17.700 --> 00:13:20.399 in the Midwest slows down, people in California 00:13:20.399 --> 00:13:22.460 are still using their credit cards. One part 00:13:22.460 --> 00:13:24.580 of the economy might be weak, but the others 00:13:24.580 --> 00:13:26.879 will be strong. Right. If one sector takes a 00:13:26.879 --> 00:13:29.409 hit, the others buoy the ship. Diversification 00:13:29.409 --> 00:13:31.950 was the primary selling point. And for investors, 00:13:32.149 --> 00:13:34.909 this was a fantastic deal. Why? Because these 00:13:34.909 --> 00:13:37.750 CDOs offered returns that were two, sometimes 00:13:37.750 --> 00:13:40.289 three, percentage points higher than regular 00:13:40.289 --> 00:13:42.610 corporate bonds, even though they had the exact 00:13:42.610 --> 00:13:46.009 same credit rating. So I could buy a AAA -rated 00:13:46.009 --> 00:13:50.980 CDO tranche and get, say... 7 % interest when 00:13:50.980 --> 00:13:53.580 a AAA -rated corporate bond from General Electric 00:13:53.580 --> 00:13:56.360 was only paying 5%. Exactly that. It felt like 00:13:56.360 --> 00:13:58.500 a free lunch. The magic word on Wall Street, 00:13:58.679 --> 00:14:02.480 arbitrage. A free lunch. But then something shifted. 00:14:03.129 --> 00:14:05.450 The machine went into overdrive. We enter the 00:14:05.450 --> 00:14:07.889 engine of growth. Why did these things explode 00:14:07.889 --> 00:14:10.190 in popularity? How did we go from this niche, 00:14:10.330 --> 00:14:13.070 diversified product to a global obsession that 00:14:13.070 --> 00:14:15.149 brought the system to its knees? It comes down 00:14:15.149 --> 00:14:18.169 to two massive, powerful forces colliding at 00:14:18.169 --> 00:14:20.809 the same time. Bank incentives and investor demand. 00:14:21.110 --> 00:14:22.909 Okay, let's start with the banks. Why were they 00:14:22.909 --> 00:14:24.809 so utterly obsessed with making more and more 00:14:24.809 --> 00:14:27.629 CDOs? It's all about capital requirements. Banks 00:14:27.629 --> 00:14:29.710 are regulated. They have to keep a certain amount 00:14:29.710 --> 00:14:31.590 of their own money capital in reserve against 00:14:31.590 --> 00:14:34.019 the loans they hold. on their books. As a cushion 00:14:34.019 --> 00:14:36.039 in case things go bad. As a cushion, exactly. 00:14:36.299 --> 00:14:39.679 So if a bank holds a billion dollars in mortgages 00:14:39.679 --> 00:14:42.679 on its balance sheet, that ties up a lot of their 00:14:42.679 --> 00:14:44.700 capital. That's capital they can't use for anything 00:14:44.700 --> 00:14:48.779 else. But if they can securitize those loans, 00:14:48.899 --> 00:14:51.360 turn them into a CDO and sell them off, those 00:14:51.360 --> 00:14:54.139 loans are gone. They are off the books. Capital 00:14:54.139 --> 00:14:56.519 is freed up. It's totally freed up. They can 00:14:56.519 --> 00:14:59.039 immediately turn around and lend that same money 00:14:59.039 --> 00:15:02.960 out again to a new borrower. And every time they 00:15:02.960 --> 00:15:05.340 originate a loan, they get a fee. So the whole 00:15:05.340 --> 00:15:07.519 model shifted. Completely. It shifted from what 00:15:07.519 --> 00:15:10.759 was called originate to hold, where the bank 00:15:10.759 --> 00:15:13.559 makes a loan and holds it for 30 years. So they 00:15:13.559 --> 00:15:16.159 really care if you pay it back. Right. To originate 00:15:16.159 --> 00:15:19.659 to distribute. Volume over quality. They didn't 00:15:19.659 --> 00:15:21.360 care if the borrower could pay the loan back 00:15:21.360 --> 00:15:23.220 in 30 years. They only care if they could sell 00:15:23.220 --> 00:15:25.580 that loan to a CDO manager next week. It turned 00:15:25.580 --> 00:15:28.179 banks from vaults into pipelines. Just keep the 00:15:28.179 --> 00:15:30.419 product moving. Okay, so that's the supply side. 00:15:30.679 --> 00:15:33.379 The banks are incentivized to create as much 00:15:33.379 --> 00:15:35.840 of this stuff as possible. What about the demand 00:15:35.840 --> 00:15:38.519 side? Who was buying all of this? Well, you had 00:15:38.519 --> 00:15:41.240 this massive, massive wall of money that was 00:15:41.240 --> 00:15:43.159 sloshing around the globe looking for a home. 00:15:43.759 --> 00:15:47.059 Economists call it the global savings glut. A 00:15:47.059 --> 00:15:50.259 glut of savings. Between the year 2000 and 2007, 00:15:50.659 --> 00:15:53.779 global fixed income investments, that's the bond 00:15:53.779 --> 00:15:58.139 market, doubled. It went to $70 trillion. $70 00:15:58.139 --> 00:16:01.240 trillion. That is a staggering amount of money. 00:16:01.740 --> 00:16:03.500 Where was it all coming from? It was coming from 00:16:03.500 --> 00:16:05.759 everywhere. You had emerging market economies, 00:16:06.000 --> 00:16:09.120 especially China, running huge trade surpluses. 00:16:09.299 --> 00:16:11.360 They were selling us goods and getting paid in 00:16:11.360 --> 00:16:13.000 dollars and they had to invest those dollars 00:16:13.000 --> 00:16:15.700 somewhere. You had oil producing nations in the 00:16:15.700 --> 00:16:17.960 Middle East that were just flush with cash as 00:16:17.960 --> 00:16:20.960 oil prices skyrocketed. All of this money needed 00:16:20.960 --> 00:16:23.700 a safe place to go to earn interest. But U .S. 00:16:23.720 --> 00:16:25.759 Treasury bonds, the safest investment in the 00:16:25.759 --> 00:16:28.179 world, were paying peanuts at the time. Exactly. 00:16:28.200 --> 00:16:30.019 You have to remember the context. This is the 00:16:30.019 --> 00:16:32.769 period after the... bubble burst. It's after 00:16:32.769 --> 00:16:35.690 9 -11. The Federal Reserve, under Alan Greenspan, 00:16:35.889 --> 00:16:38.870 had slashed interest rates to historic lows to 00:16:38.870 --> 00:16:41.129 stimulate the economy. So investors were desperate. 00:16:41.309 --> 00:16:43.129 They were desperate for yield. They needed profit. 00:16:43.490 --> 00:16:45.190 At the same time, they were scared. They'd just 00:16:45.190 --> 00:16:47.230 been burned by the stock market crash. So they 00:16:47.230 --> 00:16:49.509 also wanted safety. They wanted that AAA rating. 00:16:49.750 --> 00:16:52.409 So you have this paradox. I want high returns, 00:16:52.549 --> 00:16:55.490 but I also want zero risk. Which is impossible, 00:16:55.789 --> 00:16:59.230 usually. But the CDO seemed to be the perfect 00:16:59.230 --> 00:17:02.340 solution. It offered the yield of a riskier asset. 00:17:02.440 --> 00:17:04.420 Remember, it pays better than regular bonds, 00:17:04.599 --> 00:17:07.359 but it had that triple A stamp of safety from 00:17:07.359 --> 00:17:10.559 Moody's or S &P. It looked like a safe Treasury 00:17:10.559 --> 00:17:13.539 bond, but it paid like a much riskier corporate 00:17:13.539 --> 00:17:15.619 bond. It was the perfect product for that exact 00:17:15.619 --> 00:17:18.400 moment in time. And that insane global demand 00:17:18.400 --> 00:17:21.099 drove the machine. Wall Street had to feed the 00:17:21.099 --> 00:17:24.480 beast. But then. They hit a problem, a supply 00:17:24.480 --> 00:17:26.359 problem. This is where the story turns. This 00:17:26.359 --> 00:17:28.759 is the big shift to subprime. Because eventually, 00:17:28.900 --> 00:17:30.759 you just run out of good collateral. You run 00:17:30.759 --> 00:17:33.140 out of aircraft leases and high -quality corporate 00:17:33.140 --> 00:17:37.910 debt. You do. By 2003, 2004, the supply of prime 00:17:37.910 --> 00:17:40.509 mortgages and diverse loans was basically exhausted. 00:17:40.849 --> 00:17:43.569 But the investors, the Chinese sovereign wealth 00:17:43.569 --> 00:17:46.289 funds, the German pension funds, they were still 00:17:46.289 --> 00:17:49.250 screaming for more CDOs. So the bankers on Wall 00:17:49.250 --> 00:17:51.410 Street started to substitute the collateral. 00:17:51.750 --> 00:17:54.289 They started using subprime mortgages. Correct. 00:17:54.529 --> 00:17:57.549 The assets inside the box changed. The recipe 00:17:57.549 --> 00:18:00.829 changed. By 2004, mortgage backed securities 00:18:00.829 --> 00:18:03.460 already accounted for more than half of all CDO 00:18:03.460 --> 00:18:06.559 collateral. By late 2005, some of these portfolios 00:18:06.559 --> 00:18:09.720 were 85 % subprime mortgages. So the diversification, 00:18:09.819 --> 00:18:11.519 that whole safety net we talked about earlier, 00:18:11.559 --> 00:18:13.859 was just gone. It evaporated. It was all betting 00:18:13.859 --> 00:18:16.880 on one thing and one thing only, the U .S. housing 00:18:16.880 --> 00:18:18.480 market. Wait, this is the part that always gets 00:18:18.480 --> 00:18:21.660 me. If you fill a box with really risky subprime 00:18:21.660 --> 00:18:25.180 loans, how do you still get a AAA rating on the 00:18:25.180 --> 00:18:27.720 top slice? I mean, if I fill a box with dynamite, 00:18:27.759 --> 00:18:29.740 I can't write safe on the lid just because I 00:18:29.740 --> 00:18:32.180 packed it really tight. This is the alchemy. 00:18:32.240 --> 00:18:34.980 This is the magic trick. And it involves a very 00:18:34.980 --> 00:18:37.359 specific mechanism called mezzanine recycling. 00:18:38.279 --> 00:18:40.740 This is arguably the single most important concept 00:18:40.740 --> 00:18:42.279 to understand, and it's the one that usually 00:18:42.279 --> 00:18:44.099 gets skipped over in the movies. Okay, let's 00:18:44.099 --> 00:18:46.500 break this down slowly. Mezzanine recycling. 00:18:47.259 --> 00:18:49.400 This sounds like some eco -friendly initiative 00:18:49.400 --> 00:18:51.180 at the office, but I have a feeling it's not. 00:18:51.359 --> 00:18:55.839 Far from it. Okay, so think about a regular mortgage 00:18:55.839 --> 00:18:59.839 -backed security. an MBS, not even a CDO yet, 00:18:59.940 --> 00:19:03.259 just a plain vanilla pool of mortgages. OK. When 00:19:03.259 --> 00:19:05.779 you slice up that pool of mortgages, it's actually 00:19:05.779 --> 00:19:08.940 pretty easy to sell the top AAA slice. Everyone 00:19:08.940 --> 00:19:11.519 wants that safety. Right. It's also surprisingly 00:19:11.519 --> 00:19:14.759 easy to sell the bottom super risky equity slice 00:19:14.759 --> 00:19:17.740 because hedge funds love the gamble. They love 00:19:17.740 --> 00:19:20.740 the high potential return. But the middle. What 00:19:20.740 --> 00:19:22.519 about the slices in the middle? The mezzanine 00:19:22.519 --> 00:19:25.579 slices, the stuff that's rated BBB. It's the 00:19:25.579 --> 00:19:27.900 unloved middle child of the capital structure. 00:19:28.019 --> 00:19:30.119 It's too risky for the conservative pension funds. 00:19:30.319 --> 00:19:32.200 Their rules say they aren't allowed to buy it. 00:19:32.259 --> 00:19:34.759 But the yield isn't high enough for the aggressive 00:19:34.759 --> 00:19:36.619 hedge funds. They can't be bothered with it for, 00:19:36.720 --> 00:19:38.740 you know, 8 % return when they're looking for 00:19:38.740 --> 00:19:42.140 30%. So Wall Street had this glut. They had a 00:19:42.140 --> 00:19:45.220 problem where these BBB rated mortgage bonds 00:19:45.220 --> 00:19:47.440 were just piling up in their warehouses. Exactly. 00:19:47.440 --> 00:19:49.920 It was a law jam in the system. And if you... 00:19:49.930 --> 00:19:51.930 can't sell the middle part of the deal. You can't 00:19:51.930 --> 00:19:54.710 do the deal at all. The entire assembly line 00:19:54.710 --> 00:19:56.930 was jamming up. So what did they do? They got 00:19:56.930 --> 00:19:59.210 creative. They gathered up all these unwanted 00:19:59.210 --> 00:20:02.190 BBB rated mortgage bonds from dozens of different 00:20:02.190 --> 00:20:05.430 deals and they put them into a brand new CDO. 00:20:05.549 --> 00:20:08.869 A CDO made entirely of the leftovers of the stuff 00:20:08.869 --> 00:20:11.089 nobody else wanted. Entirely. It's what was sometimes 00:20:11.089 --> 00:20:13.509 called a CDO squared because it was a derivative 00:20:13.509 --> 00:20:16.289 of a derivative. But here's the magic. They applied 00:20:16.289 --> 00:20:20.430 that same waterfall logic to this new pool of 00:20:20.430 --> 00:20:22.349 junk. Even though it was all junk. Even though 00:20:22.349 --> 00:20:24.299 it was all junk. They said, OK, even though these 00:20:24.299 --> 00:20:27.000 are all BBB rated bonds, surely they won't all 00:20:27.000 --> 00:20:28.940 default at the exact same time. They're from 00:20:28.940 --> 00:20:31.279 different parts of the country. A default in 00:20:31.279 --> 00:20:33.519 Florida doesn't automatically mean a default 00:20:33.519 --> 00:20:35.779 in Ohio. So they took this pool of leftovers 00:20:35.779 --> 00:20:38.700 and they sliced it up again. Yes. They created 00:20:38.700 --> 00:20:42.319 a new senior tranche on top of this pool of BBB 00:20:42.319 --> 00:20:45.619 rated bonds. And the rating agencies, they looked 00:20:45.619 --> 00:20:47.480 at that new structure, ran their models, and 00:20:47.480 --> 00:20:49.920 they said. That's AAA. Correct. That is just, 00:20:49.960 --> 00:20:54.049 it's insane. You take a pile of BBB bonds stuff 00:20:54.049 --> 00:20:56.869 that was explicitly defined as medium risk that 00:20:56.869 --> 00:20:59.029 nobody wanted. You put them in a box and then 00:20:59.029 --> 00:21:01.410 you claim that 70 or 80 percent of that box is 00:21:01.410 --> 00:21:03.829 now the safest investment on Earth. That is exactly 00:21:03.829 --> 00:21:05.970 what happened. It was called ratings arbitrage. 00:21:06.909 --> 00:21:09.369 Gretchen Morgenson, a fantastic journalist who 00:21:09.369 --> 00:21:12.430 covered this, called it turning gross into gold. 00:21:12.769 --> 00:21:15.730 They were literally manufacturing AAA assets 00:21:15.730 --> 00:21:17.910 out of subprime risk. And this is where the math 00:21:17.910 --> 00:21:19.910 has to come in, right? Yeah. Because they couldn't 00:21:19.910 --> 00:21:21.809 just do this on a whim. They had to have a model, 00:21:21.869 --> 00:21:25.329 a formula to justify this to investors, to regulators. 00:21:25.450 --> 00:21:27.170 You can't just wave a wand and say it's safe. 00:21:27.289 --> 00:21:29.589 You need a spreadsheet. And they have one. Enter 00:21:29.589 --> 00:21:32.950 David X. Lee. In 2001, he published a paper introducing 00:21:32.950 --> 00:21:35.750 the Gaussian copula model. That sounds like something 00:21:35.750 --> 00:21:38.670 out of a sci -fi movie. The Gaussian Copula. 00:21:38.809 --> 00:21:41.970 It was, in its field, a massive statistical breakthrough. 00:21:42.269 --> 00:21:45.630 Before this model, pricing these complex correlations, 00:21:45.769 --> 00:21:48.150 trying to figure out how likely it was that two 00:21:48.150 --> 00:21:50.289 different loans would default at the same time, 00:21:50.410 --> 00:21:53.970 was incredibly difficult and slow. It took massive 00:21:53.970 --> 00:21:56.289 computing power. And Laya's formula simplified 00:21:56.289 --> 00:22:00.039 it. Drastically. It used a single elegant parameter, 00:22:00.380 --> 00:22:03.819 the price of a credit default swap, a CDS, as 00:22:03.819 --> 00:22:06.400 a proxy for the correlation between any two assets. 00:22:06.779 --> 00:22:10.059 It made the math clean. It allowed banks to price 00:22:10.059 --> 00:22:12.359 and structure these complex CDOs with incredible 00:22:12.359 --> 00:22:15.440 speed. It became the industry standard almost 00:22:15.440 --> 00:22:17.619 overnight. The traders must have loved it. They 00:22:17.619 --> 00:22:19.859 adored it. They could value a complex deal in 00:22:19.859 --> 00:22:22.140 minutes instead of days. But there was a flaw 00:22:22.140 --> 00:22:25.720 in the model. A fatal flaw. A huge one. The model, 00:22:25.839 --> 00:22:28.259 like all financial models, relied heavily on 00:22:28.259 --> 00:22:31.240 historical data to predict correlation. And here 00:22:31.240 --> 00:22:32.640 is where we need to talk about the difference 00:22:32.640 --> 00:22:35.359 between a house fire and a forest fire. I love 00:22:35.359 --> 00:22:38.180 this analogy. Let's walk through it. Okay. If 00:22:38.180 --> 00:22:40.940 I'm an insurance company and I insure your house 00:22:40.940 --> 00:22:44.180 against fire, the risk is unique to you. It's 00:22:44.180 --> 00:22:47.440 uncorrelated. Just because your kitchen catches 00:22:47.440 --> 00:22:49.279 fire doesn't mean your neighbor's kitchen will 00:22:49.279 --> 00:22:51.819 suddenly catch fire. Right. So it's safe for 00:22:51.819 --> 00:22:53.859 the insurance company to insure the whole neighborhood 00:22:53.859 --> 00:22:56.339 because the odds of every single house burning 00:22:56.339 --> 00:22:59.240 down at the same time are practically zero. Unless 00:22:59.240 --> 00:23:01.940 there is a forest fire. An external systemic 00:23:01.940 --> 00:23:06.259 event. A strong wind. A drought. The Gaussian 00:23:06.259 --> 00:23:09.299 Copula assumed that mortgage defaults were like 00:23:09.299 --> 00:23:12.420 individual kitchen fires. It assumed that housing 00:23:12.420 --> 00:23:14.380 markets in different cities were largely independent 00:23:14.380 --> 00:23:16.500 of each other. It looked at all the historical 00:23:16.500 --> 00:23:19.180 data and said, we have never had a nationwide 00:23:19.180 --> 00:23:21.180 decline in U .S. housing prices, not since the 00:23:21.180 --> 00:23:23.059 Great Depression. It assumed the forest fire 00:23:23.059 --> 00:23:25.640 was impossible. Exactly. But when lending standards 00:23:25.640 --> 00:23:28.220 collapsed nationwide and when teaser interest 00:23:28.220 --> 00:23:30.640 rates all reset higher at the same time nationwide. 00:23:31.710 --> 00:23:33.410 Correlation went to one. It was a forest fire. 00:23:33.509 --> 00:23:35.710 Everyone defaulted at once. And the model just 00:23:35.710 --> 00:23:37.710 couldn't handle that scenario. It wasn't programmed 00:23:37.710 --> 00:23:40.089 for it. Okay, but what about the rating agencies? 00:23:40.690 --> 00:23:42.890 Why didn't they catch this? Why were they handing 00:23:42.890 --> 00:23:46.210 out AAAs like candy if the model was so obviously 00:23:46.210 --> 00:23:49.289 fragile? This was their only job. Well, this 00:23:49.289 --> 00:23:51.009 is where it gets really ugly. It's about the 00:23:51.009 --> 00:23:53.130 conflicts of interest. We have to look at the 00:23:53.130 --> 00:23:56.069 fee structure. The rating agencies, Moody's, 00:23:56.170 --> 00:24:00.150 S &P, Fitch, they are paid by the issuers. So 00:24:00.150 --> 00:24:02.230 they're paid by the banks creating the CDOs. 00:24:02.349 --> 00:24:05.490 Yes. So if I'm a bank and I want my brand new 00:24:05.490 --> 00:24:08.390 CDO to be rated, I go to Moody's and I pay them 00:24:08.390 --> 00:24:10.549 a fee. And if Moody's looks at my deal and says, 00:24:10.670 --> 00:24:12.690 hmm, this is only worth a double A. I just take 00:24:12.690 --> 00:24:14.450 my business down the street to S &P and see if 00:24:14.450 --> 00:24:16.089 they'll give me the triple A I want. And if they 00:24:16.089 --> 00:24:19.289 do. They get the fee. It was a race to the bottom. 00:24:19.470 --> 00:24:21.930 And it was an incredibly profitable business. 00:24:22.130 --> 00:24:25.210 At its peak, Moody's structured finance division, 00:24:25.430 --> 00:24:27.869 the part that rated these CDOs, accounted for 00:24:27.869 --> 00:24:30.970 44 percent of all their sales. Their profit margins 00:24:30.970 --> 00:24:33.509 were over 50 percent. That's higher than Exxon 00:24:33.509 --> 00:24:35.789 or Microsoft at the time. It was one of the most 00:24:35.789 --> 00:24:37.769 profitable businesses on the planet, so they 00:24:37.769 --> 00:24:40.809 had zero incentive to be tough. In fact, they 00:24:40.809 --> 00:24:42.650 had every incentive to look the other way, to 00:24:42.650 --> 00:24:45.509 be flexible with their models. If they started 00:24:45.509 --> 00:24:47.869 saying no to the banks, they would lose market 00:24:47.869 --> 00:24:50.630 share and their stock price would fall. Joseph 00:24:50.630 --> 00:24:53.109 Stiglitz, the Nobel winning economist, he later 00:24:53.109 --> 00:24:56.150 blamed them for performing alchemy. He said they 00:24:56.150 --> 00:24:58.930 converted F -rated risk into A -rated paper. 00:24:59.130 --> 00:25:01.690 That's a perfect summary. And what happened was. 00:25:02.319 --> 00:25:05.119 investors, even sophisticated ones, they stopped 00:25:05.119 --> 00:25:06.819 doing their own homework. They just bought the 00:25:06.819 --> 00:25:09.180 rating. They saw the AAA stamp and bought it 00:25:09.180 --> 00:25:12.119 blindly. They completely outsourced their due 00:25:12.119 --> 00:25:14.440 diligence to the very agencies that were being 00:25:14.440 --> 00:25:16.920 paid by the people selling the product. It seems 00:25:16.920 --> 00:25:19.059 like we have all the ingredients for a complete 00:25:19.059 --> 00:25:22.160 disaster. We've got bad loans, a flawed structure, 00:25:22.339 --> 00:25:25.700 broken models, and conflicted referees. But wait, 00:25:25.740 --> 00:25:27.839 there's more. We haven't even talked about synthetics 00:25:27.839 --> 00:25:31.059 yet. This is where it goes from a bad idea to... 00:25:32.319 --> 00:25:35.039 Total insanity. Synthetic CDOs. Yeah. This is 00:25:35.039 --> 00:25:36.940 where we really leave the realm of actual lending 00:25:36.940 --> 00:25:39.619 and we enter the realm of pure betting. So what 00:25:39.619 --> 00:25:41.400 is the fundamental difference between a cash 00:25:41.400 --> 00:25:46.000 CDO and a synthetic one? A cash CDO owns actual 00:25:46.000 --> 00:25:48.759 things. It goes out and buys real bonds, which 00:25:48.759 --> 00:25:51.559 are backed by real mortgages on real houses owned 00:25:51.559 --> 00:25:54.440 by real people. You buy a cash CDO, you are, 00:25:54.599 --> 00:25:57.299 in a very indirect way, funding a homeowner's 00:25:57.299 --> 00:26:00.109 loan. Okay. And a synthetic. A synthetic CDO 00:26:00.109 --> 00:26:02.470 doesn't own anything. It doesn't buy a single 00:26:02.470 --> 00:26:05.349 bond. It just references those bonds. And it 00:26:05.349 --> 00:26:08.430 does this using credit default swaps or CDS. 00:26:08.549 --> 00:26:11.390 OK. How does that work in simple terms? A CDS 00:26:11.390 --> 00:26:13.930 is basically an insurance contract. Let's say 00:26:13.930 --> 00:26:16.250 I'm an investor and I sell you a CDS on a particular 00:26:16.250 --> 00:26:19.190 mortgage bond. I'm selling you insurance. I am 00:26:19.190 --> 00:26:21.650 betting that bond will not default. Okay. In 00:26:21.650 --> 00:26:23.890 exchange for taking that risk, you pay me a premium, 00:26:23.970 --> 00:26:26.269 like an insurance premium, every quarter. If 00:26:26.269 --> 00:26:28.349 the bond pays out fine, I just collect the premiums 00:26:28.349 --> 00:26:30.990 and I'm happy. If the bond defaults, I have to 00:26:30.990 --> 00:26:33.450 pay you the full face value of the bond. So in 00:26:33.450 --> 00:26:36.759 a synthetic CDO, the assets aren't bonds. The 00:26:36.759 --> 00:26:39.400 assets are these insurance bets. Exactly. The 00:26:39.400 --> 00:26:41.480 investors are selling insurance, betting the 00:26:41.480 --> 00:26:44.420 bonds won't default, and the cash flow is the 00:26:44.420 --> 00:26:47.200 premiums they collect. And here's why that distinction 00:26:47.200 --> 00:26:51.240 is so unbelievably important. Go on. In the real 00:26:51.240 --> 00:26:54.000 world, there is a limit. There is a finite number 00:26:54.000 --> 00:26:57.220 of bad mortgages that exist in America. There 00:26:57.220 --> 00:26:59.799 are only so many subprime borrowers. You can 00:26:59.799 --> 00:27:02.579 only package the actual loans once in a cash 00:27:02.579 --> 00:27:05.839 CDO. But with synthetics. There is no limit. 00:27:06.190 --> 00:27:08.609 You can create a thousand different synthetic 00:27:08.609 --> 00:27:12.190 CDOs all referencing the exact same pool of bad 00:27:12.190 --> 00:27:13.990 mortgages. It's like fantasy football. There's 00:27:13.990 --> 00:27:16.230 only one Patrick Mahomes. But you could have 00:27:16.230 --> 00:27:18.150 a million different fantasy leagues where people 00:27:18.150 --> 00:27:20.430 are all betting on his performance. If he breaks 00:27:20.430 --> 00:27:23.029 his leg, the Chiefs only lose one quarterback, 00:27:23.210 --> 00:27:25.369 but a million fantasy owners lose their bet. 00:27:25.680 --> 00:27:28.440 That is the perfect analogy. It allowed the amount 00:27:28.440 --> 00:27:30.480 of money moving in the system, the amount of 00:27:30.480 --> 00:27:34.220 risk, to vastly, vastly exceed the actual value 00:27:34.220 --> 00:27:36.819 of the underlying real estate market. It multiplied 00:27:36.819 --> 00:27:38.720 the risk exponentially. And it was faster and 00:27:38.720 --> 00:27:40.480 cheaper to create, right? Yeah. You didn't have 00:27:40.480 --> 00:27:42.519 to go out and actually find real borrowers and 00:27:42.519 --> 00:27:45.339 originate real loans. Much faster. You could 00:27:45.339 --> 00:27:47.500 create these things out of thin air in a conference 00:27:47.500 --> 00:27:49.759 room in an afternoon. You didn't need to wait 00:27:49.759 --> 00:27:51.859 for a bank in California to make a loan. You 00:27:51.859 --> 00:27:54.099 just needed two guys on Wall Street willing to 00:27:54.099 --> 00:27:57.470 take opposite sides. And as the supply of actual 00:27:57.470 --> 00:28:00.309 subprime borrowers started to dry up, synthetics 00:28:00.309 --> 00:28:02.750 took over to keep the fee machine running. OK, 00:28:02.869 --> 00:28:05.509 so we have the broken structure, the bad incentives, 00:28:05.710 --> 00:28:08.869 the flawed math, and now the infinite multiplier 00:28:08.869 --> 00:28:11.970 of synthetics. Now comes the crash. The year 00:28:11.970 --> 00:28:15.250 is 2006. Housing prices, which had been on this 00:28:15.250 --> 00:28:18.190 incredible tear, they peak. They level off and 00:28:18.190 --> 00:28:20.430 in some places they start to fall. The engine 00:28:20.430 --> 00:28:22.539 is starting to sputter. And we start hearing 00:28:22.539 --> 00:28:24.519 these incredible stories about the underwriting 00:28:24.519 --> 00:28:26.680 standards. I mean, the warning signs were there 00:28:26.680 --> 00:28:28.640 for anyone who bothered to look. Oh, they were 00:28:28.640 --> 00:28:31.720 everywhere. Michael Lewis tells the famous anecdote 00:28:31.720 --> 00:28:34.140 in his book of the Mexican strawberry picker 00:28:34.140 --> 00:28:37.019 in Bakersfield, California. An individual with 00:28:37.019 --> 00:28:40.299 a stated income of $14 ,000 a year who didn't 00:28:40.299 --> 00:28:44.440 speak a word of English was lent $724 ,000 to 00:28:44.440 --> 00:28:46.599 buy a house. And I'm guessing that was with a 00:28:46.599 --> 00:28:51.039 teaser rate. Of course it was. A 228 or 327 adjustable 00:28:51.039 --> 00:28:53.220 rate mortgage. You get a very low affordable 00:28:53.220 --> 00:28:55.980 payment for two or three years. But when those 00:28:55.980 --> 00:28:58.380 two years are up, the rate resets to a much higher 00:28:58.380 --> 00:29:01.299 level. The monthly payment skyrockets. And in 00:29:01.299 --> 00:29:03.299 a normal market, you would just refinance into 00:29:03.299 --> 00:29:05.599 a new loan. But you couldn't because house prices 00:29:05.599 --> 00:29:07.799 were now falling. Your house was worth less than 00:29:07.799 --> 00:29:10.440 your loan. You were trapped. So the default start. 00:29:10.660 --> 00:29:14.059 And if we go back to our water analogy, the water 00:29:14.059 --> 00:29:16.700 stops flowing. The flow slows to a trickle and 00:29:16.700 --> 00:29:19.299 then it stops. The bottom cups, the equity tranches, 00:29:19.319 --> 00:29:22.259 they dry up immediately. Then the mezzanine cups. 00:29:22.519 --> 00:29:24.539 But because of that mezzanine recycling we talked 00:29:24.539 --> 00:29:26.720 about, where the senior tranches of many CDOs 00:29:26.720 --> 00:29:28.480 were actually made of mezzanine risk from other 00:29:28.480 --> 00:29:31.680 deals, the top cups, the safe AAA cups, started 00:29:31.680 --> 00:29:33.960 to dry up too. The turning point, the moment 00:29:33.960 --> 00:29:36.380 the market really woke up, seems to be the summer 00:29:36.380 --> 00:29:40.680 of 2007. July 2007. Two hedge funds run by the 00:29:40.680 --> 00:29:43.200 investment bank Bear Stearns completely collapsed. 00:29:43.519 --> 00:29:46.079 They were highly leveraged in subprime CDOs. 00:29:46.240 --> 00:29:48.259 They sent a letter to their investors telling 00:29:48.259 --> 00:29:49.980 them they would get little to nothing back on 00:29:49.980 --> 00:29:52.000 their investment. That was the tremor before 00:29:52.000 --> 00:29:53.660 the earthquake. And then the downgrade started. 00:29:53.920 --> 00:29:56.319 The referees finally changed the call. It was 00:29:56.319 --> 00:29:58.599 an avalanche. It was too little, too late. In 00:29:58.599 --> 00:30:01.160 the first quarter of 2008 alone, the rating agencies 00:30:01.160 --> 00:30:05.339 announced 4 ,485 downgrades of mortgage securities. 00:30:05.799 --> 00:30:08.410 4 ,000 in three months. Think about that number. 00:30:08.509 --> 00:30:12.630 4 ,485 securities that were sold as safe investments 00:30:12.630 --> 00:30:15.069 suddenly weren't. It's the forest fire. It's 00:30:15.069 --> 00:30:17.210 the forest fire. And the numbers are just shocking. 00:30:17.390 --> 00:30:20.430 By 2009, more than half of all the AAA -rated 00:30:20.430 --> 00:30:23.190 CDO tranches that were issued between 2005 and 00:30:23.190 --> 00:30:25.930 2007 were either downgraded to junk status or 00:30:25.930 --> 00:30:28.009 had already lost principal. That's $300 billion 00:30:28.009 --> 00:30:31.430 of safe money just gone. There's that famous 00:30:31.430 --> 00:30:33.730 line from the big short movie. They describe 00:30:33.730 --> 00:30:37.670 CDOs as... Can we say it? Dog shit wrapped in 00:30:37.670 --> 00:30:40.769 cat shit. It's a graphic description, but by 00:30:40.769 --> 00:30:43.250 the end, it was an accurate one for how the product 00:30:43.250 --> 00:30:46.390 had evolved. It was toxic waste wrapped in more 00:30:46.390 --> 00:30:49.109 toxic waste and sold with a gold -plated bow 00:30:49.109 --> 00:30:52.890 on top. So the system collapses. Banks fail. 00:30:53.349 --> 00:30:56.329 The government steps in with bailouts. But I 00:30:56.329 --> 00:30:58.609 want to talk about the aftermath. The blame game. 00:30:59.019 --> 00:31:00.779 Yeah. Because it wasn't just that a bunch of 00:31:00.779 --> 00:31:03.700 investors lost money. It was the systemic impact 00:31:03.700 --> 00:31:06.339 of these things. They were the engine. The Financial 00:31:06.339 --> 00:31:08.799 Crisis Inquiry report makes this very clear. 00:31:09.079 --> 00:31:11.859 Without the CDO machine, lenders would not have 00:31:11.859 --> 00:31:15.059 pushed subprime loans so aggressively. They only 00:31:15.059 --> 00:31:17.160 made those bad loans because they knew there 00:31:17.160 --> 00:31:18.980 was a buyer. They knew they could sell them to 00:31:18.980 --> 00:31:20.980 the CDO machine. The demand for the securities 00:31:20.980 --> 00:31:23.880 created the supply of the bad loans. And the 00:31:23.880 --> 00:31:26.140 whole idea of dispersing risk turned out to be 00:31:26.140 --> 00:31:27.920 a fantasy. We didn't disperse it. We concentrated 00:31:27.920 --> 00:31:30.279 it. Exactly. The risk didn't disappear. It just 00:31:30.279 --> 00:31:32.460 got hidden. And where did it end up? It ended 00:31:32.460 --> 00:31:34.500 up right back on the balance sheets of the biggest 00:31:34.500 --> 00:31:36.779 banks in insurance. companies in the world. People 00:31:36.779 --> 00:31:39.740 like AIG, Citigroup, Merrill Lynch. Why did they 00:31:39.740 --> 00:31:41.700 end up holding it? Because they would often hold 00:31:41.700 --> 00:31:44.480 on to the super senior tranches, the absolute 00:31:44.480 --> 00:31:47.579 safest top of the waterfall slice. They thought 00:31:47.579 --> 00:31:50.740 it was basically risk free and they could earn 00:31:50.740 --> 00:31:53.579 a tiny bit of extra yield by keeping it for themselves 00:31:53.579 --> 00:31:56.319 instead of selling it. When those supposedly 00:31:56.319 --> 00:31:59.380 risk free tranches failed, the institutions themselves 00:31:59.380 --> 00:32:02.160 failed. They choked on their own cooking. We 00:32:02.160 --> 00:32:03.960 have to mention some of the specific villains, 00:32:04.200 --> 00:32:06.720 or at least the specific examples of excess that 00:32:06.720 --> 00:32:09.480 came out in the investigations. Because there 00:32:09.480 --> 00:32:11.799 are stories here that sound completely illegal, 00:32:12.119 --> 00:32:14.380 even if they were technically just business. 00:32:14.799 --> 00:32:17.779 The story of Magnetar is probably the most damning. 00:32:17.819 --> 00:32:19.640 They were a hedge fund out of Illinois. They 00:32:19.640 --> 00:32:21.299 figured out early on that the market was going 00:32:21.299 --> 00:32:23.599 to crash. But instead of just shorting the market, 00:32:23.720 --> 00:32:26.420 they actively helped create more bad CDOs just 00:32:26.420 --> 00:32:28.660 so they could bet against them. This is the infamous 00:32:28.660 --> 00:32:32.049 Norma CDO. That's the one. They worked with Merrill 00:32:32.049 --> 00:32:35.089 Lynch to create a CDO called Norma. Magnetar 00:32:35.089 --> 00:32:37.049 helped select the assets that went into it. And 00:32:37.049 --> 00:32:40.009 guess what? They picked the absolute worst, most 00:32:40.009 --> 00:32:42.950 toxic stuff they could find. One consultant called 00:32:42.950 --> 00:32:45.329 it a hairball of risk. So they built a bomb, 00:32:45.470 --> 00:32:47.769 helped Merrill Lynch sell it to unwitting investors 00:32:47.769 --> 00:32:50.289 as a safe house. And then they went and took 00:32:50.289 --> 00:32:52.609 out a massive insurance policy on the house blowing 00:32:52.609 --> 00:32:55.509 up. That is exactly what they did. They would 00:32:55.509 --> 00:32:58.390 buy the riskiest equity slice, the bottom cup. 00:32:58.779 --> 00:33:00.299 just to get the deal off the ground that would 00:33:00.299 --> 00:33:02.160 cost them a little money but then they would 00:33:02.160 --> 00:33:04.960 take out huge bets against the upper slices of 00:33:04.960 --> 00:33:07.720 the deal using credit default swaps when the 00:33:07.720 --> 00:33:10.579 whole thing inevitably blew up Their small loss 00:33:10.579 --> 00:33:13.079 on the equity was dwarfed by their massive gains 00:33:13.079 --> 00:33:15.200 on the shorts. And Goldman Sachs had a similar 00:33:15.200 --> 00:33:18.099 situation with their Abacus CDO deal. Right. 00:33:18.460 --> 00:33:21.220 That led to the 2010 settlement with the SEC. 00:33:21.460 --> 00:33:24.460 They paid half a billion dollars. They were accused 00:33:24.460 --> 00:33:26.779 of not disclosing to the investors who bought 00:33:26.779 --> 00:33:30.559 the Abacus CDO that a famous hedge fund, Paulson 00:33:30.559 --> 00:33:32.700 & Co., had not only helped select the assets, 00:33:32.779 --> 00:33:35.420 but was also simultaneously making a massive 00:33:35.420 --> 00:33:37.880 bet that it would fail. It really paints this 00:33:37.880 --> 00:33:40.730 picture of a rigged game. It wasn't just bad 00:33:40.730 --> 00:33:43.990 luck or a flawed model. In some cases, it was 00:33:43.990 --> 00:33:46.789 engineered failure. It does. Patrick Parkinson, 00:33:47.049 --> 00:33:49.750 who is a senior official at the Fed, later called 00:33:49.750 --> 00:33:52.849 the whole concept of these subprime -backed CDOs, 00:33:52.910 --> 00:33:56.369 these EBS CDOs, an abomination. It wasn't just 00:33:56.369 --> 00:33:59.589 a bad trade. It was a structurally flawed concept 00:33:59.589 --> 00:34:02.450 from the beginning. It relied entirely on the 00:34:02.450 --> 00:34:04.670 one assumption that could never break. that house 00:34:04.670 --> 00:34:07.329 prices never go down. And yet people made fortunes. 00:34:07.349 --> 00:34:09.550 And many of them kept those fortunes. So as we 00:34:09.550 --> 00:34:11.389 wrap this whole thing up, what's the big takeaway? 00:34:11.489 --> 00:34:14.230 If you're listening to this in 2026, why should 00:34:14.230 --> 00:34:16.690 you care about the intricate mechanics of a CDO 00:34:16.690 --> 00:34:19.449 from two decades ago? The takeaway for me is 00:34:19.449 --> 00:34:22.130 about complexity. The CDO started as a genuinely 00:34:22.130 --> 00:34:24.849 useful tool for diversification, a way to fund 00:34:24.849 --> 00:34:27.650 aircraft and student loans. But complexity became 00:34:27.650 --> 00:34:30.349 a hiding place. A way to hide risk. A way to 00:34:30.349 --> 00:34:33.869 hide risk and a way to generate fees. The waterfall 00:34:33.869 --> 00:34:36.349 structure, the mathematical models, it all looks 00:34:36.349 --> 00:34:39.570 so sophisticated. But it only protected the senior 00:34:39.570 --> 00:34:42.469 investors as long as defaults were low and uncorrelated. 00:34:42.730 --> 00:34:45.809 The incentives, the origination fees for banks, 00:34:46.010 --> 00:34:48.130 the desperate search for yield by investors, 00:34:48.349 --> 00:34:57.150 the ratings fees for the... It's that old reminder, 00:34:57.349 --> 00:35:00.190 if you don't understand it, you probably shouldn't 00:35:00.190 --> 00:35:02.670 buy it. And if the person selling it to you relies 00:35:02.670 --> 00:35:04.690 entirely on a model they can't explain without 00:35:04.690 --> 00:35:08.380 a PhD in physics, you should probably run. And 00:35:08.380 --> 00:35:10.900 remember the human element in all of this. These 00:35:10.900 --> 00:35:13.079 weren't faceless machines destroying the world. 00:35:13.139 --> 00:35:15.579 It was people. People at every level who convinced 00:35:15.579 --> 00:35:17.940 themselves that house prices never go down. People 00:35:17.940 --> 00:35:19.559 who didn't want the music to stop because they 00:35:19.559 --> 00:35:21.699 were making too much money. It was a mass collective 00:35:21.699 --> 00:35:25.139 willful blindness. So here's a final provocative 00:35:25.139 --> 00:35:27.519 thought for you to chew on. We've just spent 00:35:27.519 --> 00:35:30.039 this time looking at how CDOs turned into these 00:35:30.039 --> 00:35:32.739 monsters. But financial engineering didn't die 00:35:32.739 --> 00:35:35.119 in 2008. No, it didn't. Wall Street never sleeps. 00:35:35.260 --> 00:35:37.579 It just rebrands. Today, we have things called... 00:35:37.579 --> 00:35:40.380 called structured operating companies. We have 00:35:40.380 --> 00:35:43.139 a boom in new forms of securitization in private 00:35:43.139 --> 00:35:46.980 credit. We have CLOs collateralized loan obligations, 00:35:47.179 --> 00:35:49.900 which are booming and look eerily similar to 00:35:49.900 --> 00:35:52.860 CDOs, but are backed by corporate loans instead 00:35:52.860 --> 00:35:56.300 of mortgages. We changed the acronyms, but have 00:35:56.300 --> 00:35:58.940 we actually fixed the underlying incentive problem? 00:35:59.320 --> 00:36:01.539 That is the multi -trillion dollar question. 00:36:01.699 --> 00:36:04.079 Does financial complexity in the end actually 00:36:04.079 --> 00:36:07.239 create risk rather than dispersing it? By trying 00:36:07.239 --> 00:36:09.659 to engineer perfect safety, do we just end up 00:36:09.659 --> 00:36:11.619 building a bigger, more complicated bomb? Are 00:36:11.619 --> 00:36:13.900 we just inventing new ways to hide the same old 00:36:13.900 --> 00:36:16.159 human greed and fear? Something to think about 00:36:16.159 --> 00:36:18.340 next time you see a product that promises a guaranteed 00:36:18.340 --> 00:36:20.699 high return. Thanks for listening to The Deep 00:36:20.699 --> 00:36:21.639 Dive. Stay curious.