Breaking The Banks

I read Geithner's Stress Test, the Koch's end to bank corporate welfare and look back at my own experience working in structured finance (securitization). There are three important parts in getting bonds right. First is deal tranching, the process of layering the mortgage pools into various risk grades as the center piece. Second, bond creation starts after tranching, and third, loss analytics ends at tranching or loss analytics drives tranching which in turn drives bond creation. That is, if you fail the loss analytics, you fail the bonds.

I spent 10 years in loss analytics, built very sophisticated statistical models of mortgage defaults, even figured out how to quantify Nassim Nicholas Taleb's Black Swans for Commercial Mortgage Backed Securities (CMBS) a subset of the Commercial Real Estate (CRE) market, and was surprised that CMBS Black Swans could be as high as 80% of the deal size i.e. those CMBS deals were worthless. And I used to write UMB Bank's US economic report for its Colorado Advisory Board.

The old way of estimating loss was "1x, 2x & 3x" of the mean loss, obviously a carryover from the days when losses were believed to obey a Normal (bell curve) distribution. In a normally functioning economy, we now know that residential mortgages are Log-Normal with tails of 7 sigma (standard deviations) and my own research showed that much of the time commercial property mortgage losses obey the Gamma distribution with tails up to 25 sigma. Residential mortgages are another story for another day.

That is, the "1x, 2x & 3x" is approximately equivalent to three sigma at 3x, when actual losses are between 7 and 25 sigma? Do you think this loss methodology can protect us in a financial meltdown?

Years ago, I had observed something else. I found a CMBS deal with 240 mortgages. I forget the deal name and the banker now. Each mortgage was sliced into a 100 slices. One slice from each of the 240 mortgages was combined into a "bundled" mortgage, creating 100 bundled mortgages. Then all 100 bundled mortgages were placed into a singled CMBS deal for tranching.

Did you get the risk mitigation fiction?

My 1995 master's thesis in finance with the University College Dublin, Ireland, was a 2 volume, 550 page thesis titled "Unsystematic Risk". I had shown that risk had 3 properties, (1) minimization per portfolio diversification, (2) transference per securitization, and (3) elimination per experimental design. That unsystematic risk could not be fully diversified away, in spite of what is taught in colleges today. And when markets homogenize they tend to be asymmetrical, severely negative (crashes) or mildly positive (booms).

The problem with our legal system is that there is too much legalese and not enough prosecution. Imagine, over the last seven years banks paid out in fines, $150 billion of shareholders cash. In my opinion, proving once and for all, that these banks did not suddenly become dishonest.

So how do we manage a system that is not broken but dishonest? We play by their rules.

Mandatory Double Dipping: The bottom most tranche, sometimes named the equity tranche, is the most risky, and therefore has a premium interest rate. For example, 14% to 23% return, while the next lowest tranche has about 8%.

Did you get the sleight of hand?

Neighboring tranches have similar risk profiles, therefore why the sudden departure in the reward rates between lowest and second lowest tranches?

Investment bankers collect fees for securitization, first dip. If they don't have skin in the game, there is no realistic method to determine junk passing as prime until there is a financial crash. In this case, super seniors and AAAs are the wrong type of skin in the game. Therefore, all investment banks should be required to hold the bottom 15%, second dip, of their deal tranches while the bonds are out.

Naked Obligations: The Wall Street Crash of 2008 showed us that most risk mitigation products were not worth the paper they were written on, because when panic sets in on a national scale, investors are not running for their magnifying glasses to read the fine print but stampeding for the exit.

That means that Basel II & III are worthless in the event of a financial crash. They only work in a functioning economy where investors have the ability to estimate asset worth with some degree of accuracy. So how do we do prepare for when pricing is impossible? My recommendation is to start with an understanding of the full naked loss obligations present with all instruments and mortgages, without any credit enhancements or risk mitigation. These should be reported in the financial statements. And then don't game the system.

By virtue of their profit mandates, bankers are always trying to game the system to (1) maximize the amount of funds they have to invest, and (2) maximize the rewards without sufficient consideration for the risks involved. If you have used return on economic capital or some similar risk adjusted return statistic you know that many times this statistic does not make any sense.

Fracturing: Conceptually, the simplest way to prevent a nationwide panic is to fracture the financial system into several parts. Maybe along the lines of the 12 Federal Reserve Banks. Thus a financial collapse in one fractured part cannot spread to the next. Just as materials are stronger because the built-in fractures prevent future cracks from extending across the whole material.

Conceptually this is a very elegant solution, but practically difficult to implement. So the next best equivalent is to break up the large banks into sufficiently capitalized smaller ones.

Eat what they kill: Why should home owners who earn $55,000 a year be more knowledgeable of the economy than bankers who earn $20 million? That is, all owner occupied (first, second & third) mortgages need to be nonrecourse across all states. Bankers need to pay their penalty by absorbing their poor underwriting. This is the only way to ensure their utmost care. Yes, it might slow down the markets for a while, but name me a banker who does not want to make a profit? Unless they are below their pay grade. Sure there will be the rogue who tries to cheat the banks. It is the job of the banks to risk mitigate, but you don't punish 96% of the population to contain the 2% who are cheaters.