Monthly Archives: November 2017

Robert Avery’s Guest Lecture on November 1st

Robert Avery gave a very interesting hour long talk.  He started out by talking about his background…

I was a University of Pennsylvania Economics undergraduate and then I joined a hedge fund.  It seemed like the sexy thing that people did at the time.  It was a lot of fun, I was there for eight years.  When I first got there they gave me five bosses – simultaneously!  The main problem with that was that none of my bosses talked to each other so they all expected me to do only their work.  It was sort of an issue but I managed to handle it.  This happens in small firms, I think I was the 45th employee there.  You have to wear so many hats, they had me in the agency CMO arbitrage group, as well as the STRAITS group (statistical group), the credit sensitive portion of securitized bonds, they had me building a legal database to store all of their legal documents, as well as an accounting database.  So over time I managed to consolidate myself into subprime.

Subprime is a term that people associate with the financial crisis.  Some people say that subprime caused the financial crisis.  That is not really true.  What is subprime?  It is a portion of mortgage backed securities that grew extremely quickly coming off of the dot com bubble.  The reason for this is that they took mortgage rates really low because there was so much Chinese money flowing around the system that just had to go somewhere.  Eventually it made its way out of more regulated areas to less regulated areas and eventually into US housing, because people viewed stocks as not doing very well.  When people wanted an attractive investment vehicle they started looking at houses as an investment vehicle.  These savvy investors continued to renovate their homes and then move into bigger homes.  In fact working as a real estate agent was a really good job from 2003 to 2007.  Real estate prices were going up by 15% per year.  Why invest in the stock market when you can make money in housing?  Well there are some good reasons for not doing that, transacting in houses is more difficult than buying a stock which you can sell in a second.  Whereas selling a house takes a long time.  You have to go through the closing process, you have line up financing, in addition there are fixed costs like you have to pay the brokers 5 to 6 percent of the value of the house and the transaction costs.  So it is not a really great vehicle but people kept doing it because they saw housing prices kept going up and up, and they thought they could never go down.  Why not just keep piling in?  Then the banks decided to do the same thing since the entire country was going home buying crazy.  Why not continue to lend to more and more people because that is what they want?  So a couple of Wall Street banks as well as the so called shadow banks country wide started ramping up their origination and not just to the normal high quality borrowers who get fed into the Fannie Mae and Freddie Mac systems but also lending to riskier borrowers.  People who are first time home owners, or have FICOs less than 620, or who had a loan in the past, or who have other loans but no credit, started being able to get mortgages.  The banks just assumed that prices were just going to keep going up and up at 15% each year, it didn’t matter who the borrower was because you still had the collateral, the value of the house.  This meant you could always just take away the house and sell it.  So they just kept giving riskier and riskier mortgages.  But what were the banks going to do with these mortgages?  They didn’t want to hold them on their books.  First of all that would limit the number of transactions they could carry out, so instead they created securities called residential mortgage backed securities.  This is what we call subprime.  So once all of these subprime instruments were out there, they were defined as fixed income securities.  Fixed income securities are defined as well it is not a stock, it is more like a bond.  It has a payment schedule that is often known in advance, but not really definitely not for a mortgage backed security.  Usually you get a fixed coupon for life, floating off a platform, which is going away, and it gets very complicated.  They also have a payment schedule that is dependent upon rules.  These rules are created by these Wall Street underwriting shops, slicing and dicing the deals to make certain types of risk exposures that the market wants.  I have been using a ton of jargon, please feel free to stop me at any time.  I will try to explain these terms before I use them.

A securitization is called a deal, within a deal there are certain tranches.  So the simplest way to think about it is if you take a tranche of mortgages say 100 of them and then you slice it in half.  The first loans that pay off in full go in the A tranche so all of the bottom bonds go to the lower tranches.  If the A tranche pays off at 90% and the B tranche pays off 80% then the C tranche is going to take losses.  If you are an investor these losses are not a good thing.  So what this is doing is tranching out the risk of the securities in ways that generate very highly rated bonds that have low yields, and very lowly rated bonds that have high yields.  When people want to generate a very high return they are going to go into the riskier bonds.  Usually with very lowly rated bonds.  So with subprime Wall Street has figured out how to take all of these loans that are created by them and by their shadow banks and put them in these deals and sell them to investors in bond form.  Who are these investors?  Well there are tons of investors.  The banks were investors themselves, hedge funds were all over this, and German banks were investors.  This kept going on and in retrospect everyone says they knew it was a bubble, but often when you are in the midst of a bubble you don’t know it is a bubble.  After the fact they say, “Hindsight is 20 20.”  So this bubble will pop and it did in 2008 in rather spectacular form.  You could see it starting in late 2005, borrowers start defaulting, and there was no real market for these bonds in 2007.  In 2007 in February ADX cracked.  ADX is the Average Direction Index tracker and that’s when we knew that things were really bad but the stock market didn’t care, the stock market kept going up.  It went all the way up until October 2007.  At that point it started to decline and you see it crack in 2008.  There were tons of people losing money, hedge funds starting blowing up.  Instead of a virtuous cycle with prices going up, now there is a situation where Wall Street banks were no longer able to place these bonds in investor hands.  Then the Wall Street banks stopped buying these bonds from the shadow banks.  The problem is that once you stop originating these loans then the borrowers who had taken on riskier products, like instead of a 30 year fixed mortgage which is the traditional market standard, it had evolved to 2/28 structure this is where you have a fixed teaser for two years then you float based on a platform thereafter.  The problem is that the borrowers are only good for two years once the two years expired they start defaulting unless they refinance the loan.  Which they usually did until the banks stopped securitizing the loans.  So once you couldn’t make any more new deals the market just blew up.  Many of the people who had these mortgages defaulted.  Many of them had gotten loans for modifications from the government or private servicers and a lot has happened in that market.  So I am happy to talk about that part but that is why people say that fixed income caused the financial crisis.

I did that at Ellington from 2003 to 2011 after I originally fixed the five boss situation.  I was focused on subprime, I was responsible for loading deals.  That means that whenever a securitization comes out I had to look at the collateral, run it, figure out what the bonds were worth, tell the senior portfolio manager what I think and then he made an investment decision.  It was pretty boring.  It involved opening a lot of PDFs, parsing the PDFs to get the tables to see what it looked like, then jam it into the model which was a the horrible thing I had to deal with.  Since there were always four or five deals happening every day it was just a churn.  I didn’t particularly like that so I decided to use VBA.  Excel is extremely useful but even more useful after I started programming.  After I had automated this to a large degree I got to hire four guys to take over the aspects of it that I didn’t really want.  I needed to understand risk management positions.  In the context of 2005 to understand risk management positions, I needed to run them on an automated basis.  So I went through several iterations of batch processes and I realized that Excel was ideal not only to view the deal but to see the requests.  So what I did is I paired up with another guy, a CIS genius.  He created a calculation cluster so it was all these C plus plus programs connecting to the ADX API running on a bunch of Linux boxes.  My job was to connect that in order to get the results, but I wanted maximum flexibility.  I wanted to set it up so that, “OK this is going to be the screen that the traders use, this is going to be screen that risk management is going to use,” I wanted them to all use the same interface I didn’t want to rewrite any software or to have to deal with any technology issues.  Eventually the way it worked was I defined a lot of functions and these functions were – “I want to get the name of these inputs, I want to get these outputs.”  So I would just give it to the technician cluster guy and he would go implement it on his side.  Then I built a sequel server to be an intermediary between myself and the cluster server.  Basically when you hit F9 in Excel all of these custom functions in Excel were automatically generated.  So with the push of a button I could generate one thousand tranches, I could get one thousand custom PDFs to get whatever you wanted.  So it generated the PDFs, you can use in in Excel, it works immediately, hit F9 it will then send the asynchronous request to the sequel server where it is going to store what question I am asking.  Meanwhile the cluster is pinging that looking for answers to the questions I am asking and if you hit F9 you are going to get an answer eventually.  It is an extremely flexible tool that traders use.  It is actually quite convenient.  This is at the advent of single link CDS one of the other magical inventions of the time.  This is where we have this enormous issuance of subprime and it wasn’t enough you still wanted more and the way you get more was that you would synthetically grade it.  How would you synthetically grade it?  Well the CDS (Credit Default Swap) existed for corporations for quite some time.  If IBM issues a bond you can buy CDS on a bond to protect you in case that bond defaults.  That hadn’t existed in securitized products because there are a variety of nuances that it is based upon.  But they figured it out in spring of 2005 which was very convenient because by then I had built the Excel platform and that enabled me to run all the bonds in the universe to determine what is cheap, what is rich and we became the first guys to put out a major CDS list.  So literally we would put out a list, got bids in from the street and executed it and we got 750 million back in one day.  So for context 750 million of foreign exchange risk maybe your daily var might be .1% but in structured products it could literally be enormous, a 75 million dollar swing either way based on the forecast.  [Structured products have longer tenors and so to cover the risk the cost must be higher]  Since I was the guy who built the tool I knew how to use it.

I continued to meander on my way and did a lot of interesting things.  So our equity portfolio managers came in one day and said, “Peace out, we are going to Moscow and we are never going to see you again.”  So the managers of the Hedge Fund took me and one other guy, and put us on it, and we had to learn in two hours how their eTrading platform worked, what their algorithms were, and what their fundamental ideas were.  I maintained it for a year before I determined that it had some magic numbers that didn’t make any sense, and I didn’t believe in their hedging methodology either.  So I decided to just shut it down.  So I did that.  I eventually I became the guy who was trying to stay ahead of the entire market, after the financial crisis there were a lot of interesting trades to do because there were mechanisms within these securitizations which literally had never been triggered before.

Someone asks a question but I don’t have a good enough recording of that person speaking

The problem with that is Bitcoin gets to the topic of the FinTech revolution.  Most prices are not linked to gold anymore.  They devalued it a while back, they have a fiat currency so you control your own destiny by printing as much as you want, this is very helpful for getting you out of recessions.  Now they tried it with the GFC (Great Financial Crisis), it basically worked and it never created the inflation that was feared.  One of the supposed benefits of Bitcoin is that it is not controlled by any government and so nobody can go and inflate Bitcoin away.  There are only a fixed number of coins that you can get and as you approach that and the miners continue to mine that you are not going to get anymore.  So what is probably going to happen is that if more people want to use Bitcoin they will put more money in it. Then Bitcoin will only go up so you get a weird kind of inflation.  As more people want to use it, especially people who want to take money out of totalitarian regimes, or out of China where they have all kinds of national controls to prohibit people from translating their fortunes.  That is like ICOs, I have read about them and they seem like total scams.  It seems like the most ridiculous thing, basically with an ICO somebody gets on a platform like Ethereum.  Ethereum is the big one where Ethereum is kinda like Bitcoin except they allow other things like encoding legal contracts so you can theoretically automate payment mechanisms.  This eliminates counter party risk, if you provide someone a service, as soon as the network detects that there has been a service money is going to be automatically deducted from their account, and put into yours.  You don’t have to worry about their willingness to pay.  The thing is you can make an app on top of Ethereum for example a messaging app, let’s say I want to make the next FaceBook Messenger.  In order to send a message you have to have enough coins to send a message.  What I can do is sell all of these initial investors coins, so I could sell one coin for one penny.   Then I sell a billion of them so you have all of these people lining up to buy coin, and they paid you 10 million bucks initially.  So the people who are going to build this get the 10 million bucks, they probably have not even built the software yet.  So the coins are not really useful yet, but theoretically they are going to be useful in the future.  So people are probably going to keep buying in because they are assuming that just like Ethereum or Bitcoin, the prices are probably going to continue to go up.  These people actually become less interested in the package they initially bought the bitcoin for.  In a couple of these cases they were just total frauds. They just absconded with a hundred million of investor’s money and it is not even clear if they are obliged to produce the software anyway.  I would advise you to stay away from ICOs.