The Software Cure For A Financial Meltdown

In my role as amateur financial analyst I like to read the publications taking an in depth look at the current financial turmoil. I then try to imagine the software tool that could have prevented the turmoil and saved the world millions billions trillions gazillions of dollars. This seems like no small feat for software considering the intricate world built by modern financial engineering - a world where terms like “collateralized debt obligation” are tossed around like dice at a craps table. However, I never imagined the answer could be so simple …

First, some background. I find the current situation interesting because ten years ago I joined an Internet startup by the name of Ultraprise.com. At Ultraprise, we tried to create an online exchange for the secondary mortgage market. Let’s say you have a bit of money saved and want to buy a house, so you find a bank that will loan you the rest of the money you need to buy the house.The bank could then wait for your to repay your loan over the next 30 years and slowly make money on the interest they charge you, but many banks just want to sell your loan to someone else and receive cash in return. Wikipedia has solid coverage of this process in the Securitization entry. The fresh cash received from the sale of your loan allows the bank to stay liquid and fund more loans. But, to sell the loans the bank needs to find a buyer, and this is where Ultraprise came in.

Fannie and Freddie pick up roughly half of the mortgages in the U.S – that is their job, but the rest of the pie is still a substantial amount of debt for sale. At Ultraprise we built a web application for banks to upload pools of loans they wanted to sell. Buyers could then login and bid on the pools. The idea was to cut out the brokers who typically sit in the middle of these transactions and take a cut. Economies of scale meant our fees would be substantially less than the typical fees.

Fail

Ultraprise was out of business after three years, despite having a substantial number of loans posted in the system. We couldn’t sell loans. Part of the problem was, I think, that the people who buy loans really like the human touch that a broker adds. Rib-eye steaks and martinis sell more products than a dull web page full of numbers.

Another problem was getting mortgage data in and out of the system. There was no standard format for pushing mortgage data. Every bank with a loan for sale required a custom import, and every bank looking to buy loans wanted to download data into their custom software for analysis and due diligence.

My experience at Ultraprise led me to believe that banks were exceedingly risk averse and highly analytical, and that they required lots of data before making decisions involving millions of dollars. Thus, I was quite surprised to learn that today’s banks have no clue about the cards they are holding.

“ … a major stumbling block for banks is having the right data on hand …”  - American Banker

“…the reason that banks don’t want to lend to each other anymore is that they don’t trust that the other banks really know the value of their mortgage-backed securities … because they themselves don’t trust the value of their own.” – Aster Data

All this secrecy comes in the wake of an economic crisis brought about in part by a proliferation of financial instruments so opaque that virtually no one understood the risks. – USA Today

Epic Fail

There have been documented changes in the mortgage industry since Ultraprise closed. For example - the lowering of lending standards allowed banks to give more money to people with lower credit scores. However, the outstanding change from my perspective was how those loans were sold to investors. What follows is an extremely gross simplification.

As subprime lending increased, the banks found it harder to sell pools of loans. The risks were too high for the big-money conservative investors like pension funds, and the possible rewards were too little for the hedge funds addicted to double digit growth. Thus, the rise of the aforementioned collateralized debt obligation, a.k.a the CDO.

The investment banks use CDOs to package mortgages, subprime loans, home equity loans, automobile loans, credit card debt, corporate debt, and used cat litter into products they slice up and sell to investors in private offerings (with the help of strippers, martinis, and appalling judgments by our trusted credit rating agencies). 

However, not all of the slices from a CDO are an easy sell (particularly the ones filled with risky loans and cat liter). So … firms will create a CDO squared (CDO^2) from the undesirable slices of multiple CDOs – essentially dressing up pigs with lipstick for the next big investors ball. You can be assured that a CDO cubed (CDO^4) will then arise from the dumping grounds of multiple CDO^2s, and then … well … CDO^N should give you an appreciation of how deep into the rabbit hole an investor can fall.

Investment banks issued hundreds of billions of dollars in CDOs over the last 5 years. Why did the risk averse investors, like the banks and the pension funds, stand in line to buy these CDOs and the other credit derivatives that Warren Buffet labeled “financial weapons of mass destruction”?

In part because a CDO does a good job of obscuring it’s underlying qualities– the loans, the assets, but most of all the risk. It was all a sales job, and something that an unknown software company could never pull off. 

Then How Could Software Help?

The $765 million “Mantoloking CDO 2006-1” was underwritten by Merrill Lynch and is the prime example of a “toxic asset”. The Mantoloking, a CDO squared, was built from the unwanted slices of 126 other CDOs, and is perhaps the most infamous CDO because its spectacular losses facilitated the disappearance of at least two hedge funds. You can find the 200 page prospectus online. It’s a lot to digest. In his paper “The Credit Crunch of 2007: What went wrong? Why? What lessons can be learned?”, esteemed financial engineer John C. Hull says transparency is needed.

[CDOs] … are arguably the most complex credit derivatives that are traded. Lawyers should move with the times and define these instruments using software rather than words. 

Sound familiar? To me it sounds like the software practice of using executable specifications - TDD and BDD for the financial world. No one can hide behind wordy documents and inflated credit ratings. They have to look at real numbers. The only question is - do we write these specifications in C#? Ruby? Haskell?

It turns out that Mr. Hull already had a language in mind. You can find it in his paper as footnote #8.

Given its widespread use in the financial community VBA is a natural choice for the programming language. 

What? A Microsoft Excel spreadsheet with macros might have saved our banks, our brokerage accounts, and our retirement funds?

It’s a stretch, but with the proper models in place it’s certainly a step in the right direction. Mr. Hull’s paper has other prescriptions for the finance world, too, as software is only part of the solution. You can never stop anyone who wants to skip due diligence and go directly to short-sighted greed, but I’d like to think that if our industry could make good software more readily available to the business world, the problems we are experiencing today wouldn’t be quite so bad.