Sunday, June 17, 2012

Risk

We are now 3-4 years into our current economic problems.  I have read several books on the subject.  I think I have come up with a simple explanation (e.g. one much less than book length) of how we got into this mess.  I think the whole thing can be explained by focusing on a single word:  risk.  Now the more enlightened among you know Risk as a board game.  But I am talking about "risk" with a lower case "r".  At its most basic, risk is about whether or not things are going to go badly wrong.

Businesses don't like things to go badly wrong.  So they have long sought out "risk mitigation" strategies.  Let's say a business is worried about a large investment going bad.  And let's say further that the business thinks that the chances of this happening are 1 in 100.  Say this business can pay a premium of 2% of the size of the investment for someone else to cover the loss if the investment goes bad.  On paper this might seem like a bad investment as in a probabilistic sense the cost of avoiding the problem is twice the "expected cost" of the problem.  But the business might decide it's willing to pay the 2% anyhow just to avoid having to worry about the investment going bad.  If this sounds like insurance that's because it is.

There are many complexities in actual specific situations but this "what will it cost and what is the probability it will happen" captures the core concept of risk.  And businesses engage in risky behavior as a normal part of doing business.  Any business deal, investment, loan, etc. can go bad.  So one of the critical skills of a good businessman is to be able to properly manage risk.  If a business engages in a lot of low cost high risk transactions then the business most likely self insures.  They build expected losses into the cost of doing business and eat the loss directly.  This saves them the additional cost of farming the risk out.  If they have judged the risk correctly then the mark up on the successful transactions will more than pay for the losses on the ones that go bad.  Businesses often put a lot of thought into situations where the risk is very low but the cost of failure is very high.  These are the situations where they are likely to buy some form of insurance.  But the key to doing this sort of thing correctly is to be able to accurately judge risk.  If something is low risk but you think it is high risk then you waste money on insurance.  If something is high risk but you think it is low risk then you may choose to under-insure or get no insurance at all.  These situations can lead to very bad things as we all now know only too well.  So let's take a look at events of the past few years from the perspective of risk.

For most of my lifetime a mortgage was a low risk investment for a company.  But this was not always so.  The house my father grew up in was built in 1910.  For various reasons my father took control of the family's finances about 25 years later.  The house originally cost $5,000.  He was shocked to learn that after payments had been made for 25 years $5,000 was still owed on the mortgage.  And this being the middle of the great depression, the house was still worth only about $5,000.  His parents had taken out an "interest only" mortgage.  He quickly moved to start paying down the principal and had the house fully paid off a few years later.  If my father's family had been a little less lucky they would have ended up with nothing.  Many depression era families did.  As a result of the experience of those other families FDR was able to introduce mortgage regulation which banned interest only mortgages.  For many years a 30 year 20% down mortgage was the standard.

And a 30 year 20% down mortgage is a very safe investment for the investor.  If something goes wrong the investor can repossess the house and sell it.  They will likely get enough to cover the mortgage balance and any additional costs.  And there was a good chance the home owner who got in trouble would sell the house and use the proceeds to pay off the mortgage in full before the mortgagor even knows there's a problem.  In only a few rare cases did the investor take a loss on the mortgage.  So the risk of any loss was low.  And even in the case of a loss the investor was likely to get most of his money back.  An investor was fully justified in assigning a very low "risk premium" (the amount he needed to mark the transaction up by) to a mortgage transaction.  This situation continued for 40 years or so.

Then people looked at how to make more money in the mortgage business.  The first thing they asked themselves was "what if we reduced the down payment minimum below 20%?"  The answer turned out to be "not much".  Few mortgages went into foreclosure.  When they did the potential loss was usually well below 20%.  So by changing the down payment requirement from 20% to say 15% or 10% nothing much really happened on the risk side.  The number of mortgages that went bad stayed low and the losses, when there was a loss, stayed small.  But all of a sudden a lot more people could afford a mortgage.  So costs went up but not by much and volume went up a lot.  And down payments went down some more to 5% or 3%.  And even more people flooded into the market.



Now what if I write a mortgage to a bad person?  This is someone who is unable or unwilling to pay the mortgage.  What's my risk?  Well, even if the bad person is in the house for only a year the investor will get all his money back if housing prices rise enough.  So why bother with credit checks if you will not lose money if the credit is bad?  In this environment an accurate credit check is not worth much.  If the mortgagee is a good person it's a waste of money.  If the mortgagee is a bad person you still get your money back eventually.  So a entire segment of the mortgage business appeared that specialized in writing mortgages for bad (only in the sense of credit risk) people.  And investors made a lot of money.

It's not just that they failed to lose money.  They actually made money, more money.  This is because of a perverse situation.  If a person has a bad credit rating then you are justified in charging them a higher interest rate (see "risk premium" above).  Investors love getting a higher rate of return (e.g. the higher interest rate) for the same risk (e.g. chance of a loss).  So investors encouraged mortgagors to find problem mortgagees because they could be talked into a higher interest rate.  In fact, it got so bad that many people who could qualify for a low risk low interest rate mortgage were sold a high risk high interest rate mortgage because that was he kind of mortgage investors wanted to buy (and were willing to pay a higher commission on).  Then there was the balloon.

Mortgages began to be structured with a low "teaser" interest rate for the first few years (typically three years).  Then the interest rate would "balloon" up to a much higher rate.  From a mortgagee's point of view this was not as bad as it seemed.  They could refinance, presumably into a new mortgage with another 3 year teaser rate.  If they did this often enough they'd never get to the part where the interest rate ballooned.  If that didn't work they could just sell the house and pay off the mortgage.  So it looked like the mortgagee always had an out, if the mortgagee was smart enough to figure all this out.  If not then he was in for a big surprise but that was his problem.  And on the other side of the deal, the investor side, there was a con job going on.  Mortgages were rated on their return.  Since a typical balloon mortgage would have a low interest rate for 3 years but a high interest rate for 27 years the average interest rate was pretty close to the high interest rate so the mortgage looked like a pretty good investment.  Now this ignores the whole "refinance or sell" thing but investors went with the "ignorance is bliss" strategy in large numbers and pretended the "refinance or sell" option did not exist.

The mortgage industry evolved quickly.  Down payments went down quickly.  Teaser rates and other gimmicks appeared quickly.  So there wasn't a lot of history that accurately represented the current market.  Wall Street took advantage of this.  They would trot out lots of statistics about how slowly mortgages turned over and that only a small percentage of mortgages defaulted.  But most of these statistics covered a different market, a market where down payments were higher and most mortgages did not contain gimmicks.  Averaging a bunch of good old mortgages with some bad new mortgages gives a distorted picture of what is likely to happen with the new mortgages.  But investors in general ignored all this and the marker for bad mortgages was hot, hot, hot.

Now let's step back a little.  For more than 40 years mortgages were a modest boring business.  Volume was relatively low because of the 20% down requirement and the fact that mortgagors usually did a thorough credit check and did not loan to problematic potential customers.  As a result mortgages were in fact very low risk.  Then the market started evolving.  Each evolution was in the direction of higher risk.  The early changes (down payment requirement reduced to 15% or 10%) increased risk but by only a small amount.  But they increased the volume by a lot.  Wall Street loved this.  More transactions meant more profit.  So Wall Street pushed for even more loosening of mortgage standards.  During this period risk increased more quickly than before but the increase in risk was still relatively small and by adding more risk premium (e.g. higher interest rates) loses could be managed and volume increased even more.

This led to a vicious cycle.  Wall Street pushed for more loosening of now already loose standards and volume increased.  And by adding gimmicks the apparent profit margin increased.  Default risk was still small because by this time home prices were increasing by leaps and bounds.  By the end of the process anyone could get a mortgage.  In one book the author met a Los Vegas stripper who owned four houses as investments.  In many cases gardeners and dishwashers were buying McMansions.  There was no way these people could keep up on their mortgage payments even though the vast majority wanted to.  But it didn't matter whether a mortgagee couldn't or wouldn't keep their payments up because any problem could be fixed by flipping the house.  So it appeared if you didn't look closely at what was going on that mortgages were still a low risk investment.  In fact the risk associated with the mortgage market had climbed to the point that it was very high.

This finally became apparently when housing prices stalled out.  They stopped rising quickly.  Then they stopped rising at all.  Finally, they started falling.  If someone has made a 20% down payment and the value of the house drops by 10% the investor will still come out OK.  But if the assessment on the house was inflated (as too many were) and the mortgagee paid 0% down and the value of the home drops then the investor is going to lose a serious amount of money.  And that's what happened.  And that fed into a substantial downturn in the economy.  So that people who could normally have afforded their mortgage payments lost their jobs and defaulted.  And this put a lot of distressed houses into the market, which further depressed housing prices.  And many of the people who were employed in the construction business or the appliance business or many other businesses that saw sales drop off dramatically were let go.  So more mortgagees got in trouble and housing prices got depressed some more.

At the peak most mortgage backed securities were rated AAA.  This means they are very low risk investments.  And right up to the end they behaved like they were very low risk.  Losses up to the peak were very small.  There is a branch of mathematics that describes these kinds of situations.  It is called "catastrophe theory".  Imagine a Popsicle stick.  It is placed on the edge of a table so that half of it is sticking out past the edge of the table.  Now imagine holding the stick down on the table and pushing gently on the other end.  The stick will bend slightly.  Push harder and it will bend some more. Let go and it will straighten out.  Now push much harder.  The stick will bend even more and, if you push hard enough, it will break.  Now stop pushing at all after the stick has broken.  What happens?  The stick will not return to being straight.  It will stay bent at the broken spot.  Catastrophe theory deals with these "bend till it breaks" situations.  Fortunately, we don't need to be catastrophe theory experts.  The broken Popsicle stick tells it all.  After the stick breaks a little change like not pushing on the stick any more does not bring the stick back to being straight.

The mortgage business ended up like the broken Popsicle stick.  Once it broke small fixes like lower interest rates did not put it back to where it had been.  And before it broke a number of people made conscious decisions to push the mortgage market harder and harder.  There justification was "well, it hasn't broken yet".  But they kept pushing harder and harder toward higher and higher risk behavior.  On the front lines were the mortgage sellers.  Once they sold a mortgage to a customer they wholesaled it out to Wall Street.  They made their money on volume and retained no risk once the mortgage was sold off.  And they were pushed by Wall Street to make more and more and riskier and riskier mortgages.  As long as nothing went wrong Wall Street made more and more money.  Much of the mortgage origination market is unregulated.  To the extent that it is regulated some regulators tried to push back.  But they were opposed by Wall Street and their powerful lobbying operation.  The regulated mortgage originators also opposed the regulators because they were losing business to unregulated originators.  They added their lobbying muscle to Wall Street's.

A partner in crime were the securities rating firms like Moody's and S&P.  They rated these investments AAA right up until the end.  But they were captured by Wall Street.  If one firm gave a security a bad rating then Wall Street would hire a different firm if the new firm would promise to give the same security a good rating.  Everyone knew how the game worked.  So the ratings agencies would build a paper trail to justify their rating, a paper trail based on the bogus (see above) historical data and other "analysis" Wall Street provided.  The idea was to have plausible deniability.  "We followed accepted industry practises.  We had no idea, honest!"  This, their own lobbying operation, and a "White Shoe" Wall Street law firm on retainer, was judged to be sufficient cover.  And so far their defensive strategy has worked. No one is in jail.  All the firms are still in business and no individuals have lost big law suits.

An argument can be made that mortgage originators are not all that smart.  I don't believe it, but let's just say.  And similarly an argument can be made that the regulators and the ratings agencies are not that smart either.  Again, I don't believe it but let's just say.  Wall Street prides itself on having lots and lots of "smartest guy in the room" types.  But if one of these Wall Street smart guys had applied their intelligence and pointed out the problems in the mortgage industry, what would have happened?  They would been chastised for not being a team player.  If their firm acted on their conclusions they would have stopped making the kind of money other firms were making.  Instead what the smart guys on Wall Street adopted (or in some cases tried to adopt) a different strategy, the "musical chairs" strategy.

In musical chairs there are a number of people walking around a circle of chairs with one less chair than people.  When the music stops everybody tries to sit down.  Whoever does not make it safely to a chair is the loser.  Many people on Wall Street knew there were problems.  But they also knew that were many players involved.  So it was like musical chairs with lots of people but not enough chairs.  As a Wall Street smart guy the strategy I (and pretty much everybody did this) adopted was to make sure I (or my company) always had a chair I could definitely make it to when the music stopped.  Since I'm the smartest guy in the room some other schmuck will get stuck without a chair.  The problem, of course, turned out to be that the whole roof caved in and there were no chairs left for anyone.  For many on Wall Street AIG was the designated schmuck.  Unfortunately no one thought it was their job to make sure AIG had enough money to fulfill its role.  They didn't.  A couple of beats after AIG went under the music was stopped by the roof falling in.

"No chairs" was not a possibility that anyone had considered.  And it turned out that the government and its middle class taxpayers ended up having to come in and bail Wall Street out.  So for the most part it didn't matter to Wall Street that it had screwed up.  A couple of firms went under and a lot of employees got laid off but the system as a whole survived just fine.

So Wall Street didn't know (or pretended it didn't know) that mortgage risk was not being calculated correctly.  Some regulators got it right or at least came closer than any other group but an aggressive lobbying campaign in public and behind the scenes caused them to be ignored when they weren't silenced outright.  The ratings firms got it completely wrong.  And the mortgage origination industry got it completely wrong.  But several of the largest mortgage origination firms (e.g. Countrywide) sold out to Wall Street at very high prices a year or so before the collapse. So the senior executives of these firms did very well.  And ask yourself why so many sold out when things were going so well?  Maybe some people in the mortgage origination business did have a clue.  They differed from Wall Street only in adopting a "sell out at the top" strategy instead of a "musical chairs" strategy.

If it had just been the mortgage meltdown things would have been bad enough.  But unfortunately, it wasn't.  Wall Street has always made money by selling advice.  This is a steady business with a nice profit margin but there isn't a big enough market for it to make the kind of out sized profits Wall Street craves.  Wall Street's main moneymaker used to be buying and selling securities.  And by "securities" I mean Stocks and Bonds.  That used to be a very lucrative business.  But deregulation set in a few years ago and the amount of money a firm can make per transaction plunged.  The fee on a Stock or Bond transaction used to be enough to buy a dinner at a fancy restaurant.  Now it would be lucky to cover the cost of a small Coke at McDonald's.  Squeezing a half a cent out of the cost of a transaction is just not what a Wall Street Master of the Universe dreams of.  They looked around and found derivatives, the big moneymaker for Wall Street for some time now.

A derivative buy or sell transaction is like a Stock or Bond buy or sell transaction from an execution point of view.  So Wall Street didn't have to invent anything new to move into this business.  Stocks and Bonds are traded on exchanges.  The commissions are negotiable and everyone knows what is going on. So the market is fiercely competitive and the fee for executing the transaction is tiny because people can shop around for the best deal.  Derivatives historically have not been traded on exchanges.  For a specific derivative typically only one firm knows what the security is really worth so that firm can add a nice markup into the transaction.  So the profit per transaction can be like the old days with Stocks before deregulation.  In some cases it can even be much better.  The profit potential is awesome with derivatives.  So what is a derivative?

The answer turns out to be pretty much anything.  All you have to do is find two people, one for each side of a bet and you can package it up as a derivative.  In practise Wall Street deals in money.  So derivatives are generally about something financial.  As an example let's talk about mortgages.  As I said above, Wall Street bought bails of mortgages.  Now Wall Street was only interested in the money part of the mortgage.  So the mortgage was split.  A "mortgage servicer" would worry about collecting the payments and all that nitty gritty stuff.  Wall Street left this part alone and concerned itself only with the cash flow.  They would provide the money to buy the house.  Then they would get the cash flow generated by the mortgage payments from the servicer and marry it back to the mortgage.  And a single mortgage for a single house, even if it was a million dollar McMansion, was too small to interest Wall Street.  Typically thousands of mortgages were bundled into a single "mortgage backed security".

And Wall Street could stop there and just market the mortgage backed security.  But that was not interesting (read profitable) enough.  Different potential investors had different investment objectives.  Some investors were high risk - high reward types.  Others were low risk - low reward types.  It was just too difficult to build special packages for each investor type.  Then someone came up with a brilliant idea called "tranches".  I have no idea where the word came from but the idea is simple.  Say your bundle of mortgages has 1,000 mortgages in it.  Divide them into 10 tranches.  Tranche 1 would contain the 100 riskiest mortgages.  Tranche 2 would contain the next 100 riskiest mortgages.  Tranche 10 would contain the 100 least risky mortgages.  Now we can sell tranche 1 to a high risk investor and tranche 10 to a low risk investor.  The details of which tranche a specific mortgage ended up in are complex and might vary from security to security.  But this "tranche 5 of mortgage package xxx" is a derivative.  Its value is derived from some package of underlying securities.  And a derivative can be created that is a bundle of other derivatives.  So you can have layer upon layer upon layer.

And now the magic happens.  In a normal market what are the probabilities that 10% of a typical package of mortgages will default?  The answer is practically zero.  So a ratings agency might easily rate tranches 2 - 10 as AAA.  Anyone can buy a AAA investment.  Lots of people (pension funds, insurance companies) are required to only deal in "investment grade" securities.  AAA securities are investment grade.  So we have just created a bunch of AAA securities.  If the interest rate on the tranche 1 mortgages is high enough then someone will take a flier on it.  And let's say you are one of those "AAA only" investors.  If you are given a choice between an investment that returns 5 1/4% and one that returns 5 1/2% what do you do.  They are both AAA so of course you buy the 5 1/2% one.  By bundling and tranching Wall Street was able to create giant piles of AAA investments.  They could get the ratings agencies to rate lots of stuff AAA by saying "there is only a one in a million chance that more than 10% of mortgages will default".

Now this "bundle and tranche" strategy was applied to all kinds of stuff besides mortgages.  If you tranche Credit Card Accounts Receivable (your outstanding balance on your credit card) you can create a bunch of AAA securities.  Everybody wants to buy the AAA security that has a little higher return than the run of the mill AAA security.  And "due diligence" for many investors started and finished with "is it AAA?"  And for a fee Wall Street would customize the security.  Typically the Wall Street firm that created one of these derivative securities was the only one who had full information on it.  So if they bundled in some fees and added some markups they could still make the security look like a good deal.  So they did and made fantastic amounts of money doing so.

One more piece was necessary to make the whole scheme work.  Wall Street firms needed a way to assign a risk rating to each of these derivatives.  Now we can ask how many Boeing 575 airplanes fall out of the sky in a given year.  There are lots of 757's around so we can come up with a pretty good estimate for how much risk is built into a 757.  (Fortunately for all of us the answer in most years is none).  But in the case of derivatives many of these securities are unique.  Even when this is not true there is not much to go on.  Wall Street was saved a few years ago.  A math genius came up with an "algorithm" (essentially a computer program) that would pop out a single magical "risk" number no mater how many moving pieces the derivative contained.

The details are for the most part irrelevant to our discussion because we can understand why things went so wrong by concentrating on something easy, what information went into the algorithm.  The first kind of information was the specific details of each underlying security (e.g. mortgage amount, interest rate, duration, FICA number (credit score), etc.).  The other kind of information was historical information for similar securities. For these securities additional information on default rate, delay till default, loss amount, etc. were added to the basic "security" information.  The algorithm would use the historical data to make an estimate of what would happen to the specific set of securities in the package in question.  It would then boil that down to a single magical "risk" number.  With this algorithm a uniform procedure could be applied to any kind of derivative.  Every derivative could be assigned a risk based on the magic algorithm.  And all the major Wall Street firms bought off on this system.  With it all these derivatives could be boiled down to three numbers.  What's the price of the derivative?  What's the return (interest rate) of the derivative?  What's the risk of the derivative?  Now Wall Street was ready to sell large quantities of derivatives.  All the complexity was safely hidden out of sight and everyone could concentrate on "sell, sell, sell".  It looked like a good deal to customers because Wall Street was offering all these AAA securities with these wonderful returns.  What's not to like?

At this point we all know this story did not turn out well.  After the fact the magic algorithm was roundly criticized.  A lot of the criticism was fairly technical (e.g. "tail effect").  But most of the criticism ignored the question of whether the algorithm was properly applied.  Frequently the answer was no.  The original work was done using mortgage data.  The author was only interested in proving that the algorithm worked.  So he grabbed the data that was most readily available.  This consisted of a few years of mortgage data from the boom times.  There was no "bad things happening" data in the historical data he used.  This weakness was unimportant in terms of proving that the algorithm was mathematically correct.  But it was critical in real world applications.  Wall Street never expanded the historical data they used to included a broader sample.  The algorithm with the crippled historical data gave them the answer that allowed them to sell the security so they were happy.  But it gets worse.  The magic algorithm was applied to all kinds of stuff, not just mortgages. Was a custom set of historical data developed for these other types of securities?  No!  The mortgage data was handy.  It had a proven track record of generating a low risk number so why mess with success?  In fact the whole magic formula procedure as implemented on Wall Street was one giant con.  There was nothing wrong with the underlying math.  But the algorithm has to be applied properly and frequently it was not.

If you talk to Wall Street today they claim that they have fixed everything.  They still use the magic algorithm but they now claim that they are using it correctly.  But I see no reason to believe this.  Wall Street went wrong in the first place because it was too hard to make the kind of money they wanted to make without getting creative.  If anything, it is now even harder to make money the old fashioned way.  Margins in traditional lines of business are tighter than ever.  Competition is fiercer than ever.  So there is still a strong incentive to get creative.  And there is still money to be made by selling something that is risky but claiming it is not.  We have just seen clear evidence of this.  A few weeks ago JP Morgan Chase announced that they had taken a 2 billion dollar trading loss.  Many news reports claim the actual figure is 3 billion.  I have even seen one claim that it is 7 billion.  Jamie Dimon, the CEO claims that he now has things under control and that JP Morgan can absorb the loss without significant harm to the company or to the economy at large.  But it highlights the fact that a fundamental problem still exists.

The working assumption of all these Wall Street firms is that all these derivatives are "liquid".  In short this means if you decide to sell the instrument you can easily find a buyer.  And the price you get will be predictable.  But there is no reason to believe this.  In 2008 during the worst of it there were lots of derivatives that could not be sold for any price.  People came to believe they did not know what was in the underlying bundle of securities and, to the extent that they could tell what the underlying securities were, they could not accurately estimate their value.  So they said "I'm staying on the sidelines while I wait for things to shake out".  And the "value" of many derivatives went from 100% of purchase price to 90% or 70% or 50% or less, sometimes in a few days.  Some of these "AAA" rated derivatives ended up being worth 30% or less of face value.  The very fact that the content of these derivative bundles varies from instrument to instrument means that each one is unique or nearly unique.

This has been a problem for JP Morgan.  They have not released details on just what went wrong with which securities.  But traders think they have a pretty good idea.  And if a trader knows (or suspects) that a particular security being offered for sale is a JP Morgan problem child he knows JP Morgan is in a bind.  And that means he can low ball his price knowing JP Morgan might have to accept it anyhow.  Outsiders speculate that traders taking advantage of JP Morgan's bind is why the estimates for how much this will eventually cost JP Morgan vary so much.

In some sense the JP Morgan case is a special case.  But in another sense it is not.  Traders might chose to avoid a particular derivative or class of derivatives at any time for any reason or no reason.  Mostly what they care about is getting the best return for the least risk.  The fact that derivatives are generic because mostly all people care about is price, return, and risk but are custom in the sense that each derivative has a different set of underlying securities makes any specific "risk" number unreliable.  There are always other derivatives that have a similar price, return, and risk but a different set of underlying securities.  And this means that all derivatives are risky all the time.

In summary, we got here by doing a terrible job of estimating risk.  All this was started by various groups working together to drive risk in the mortgage market through the roof, all the while loudly claiming that no such thing was happening.  The market was driven so far from stability that it broke in a catastrophic (in the mathematical sense and in its impact on society) manner.  This firestorm swept through the markets on Wall Street and exposed more and larger failures to properly calculate risk.  We are probably doing a better job now, both in the mortgage market and on Wall Street.  If nothing else, I think customers now take Wall Street "risk" estimates with a 5 pound bag of salt.  The question is are we, and here I primarily mean Wall Street, now doing a good enough job of correctly calculating risk.  The JP Morgan "problem" argues strongly that we are not.