Cutting debt into little bits

Comprehending what caused the recent credit-crunch/liquidity-crisis is much easier after reading Tom Cunningham's explanation:

But wait, there's a loose end, what about all the risky leftover pieces of investment? Who's going to buy those?

This is the clever part. Suppose you're offered the chance to flip a coin, where you'll get £100 if it's heads, but nothing if it's tails. On average you get £50, but it's risky. Now instead say you pool your winnings with 100 other people who are in the same situation and you all share the proceeds. Again you'll get £50 on average, but it's now a much surer prospect.

This is the trick to deal with the risky parts of the investments: if one firm can buy up a huge number of these separate risky investments then the risks start to cancel out, and as a whole it becomes a fairly safe investment.

It all makes sense. Sort of.

Cutting up debt

The thing investors really don't like is uncertainty. And what the financial engineers in the city have been trying to do is "smooth out" risk. Risk remains, but it is well managed *in theory*.

The logical conclusion would have been one enormous world investment, stitched together from all of the billions of separate individual investments. But the financial engineers stumbled before they could construct that one.

Cunningham's conclusion is that "no one has found any fundamental problem with the principle of sharing risk."

This raises the interesting possibility that once the smoke has cleared financiers might start doing it right.

[image from SqueakyMarmot]

Finance and computers: an analogy

A computer takes digital data and transforms it in fairly basic ways. Adding, subtracting, logical statements etc.

Computers are powerful tools because we can make them perform more complex tasks than these basic functions by building new layers of abstraction on top of these basic processes.

Machine code can be used to make a compiler for the C programming language, which can in turn be used to make an operating system, on top of which can run applications, which can be used in ever more complex and elaborate ways.

Abstraction and "higher order" properties are important in computing, but when the same ideas are applied to finance things go all gooey.

The basic unit of economic interaction is not a bit or a logic gate; it is a human being, an absurdly complicated thing, and one that we don't fully understand.

In computing, logic gates and machine code are fairly simple. Because they are simple and well understood, we can build a layer of abstraction on top of them and rely on them to function correctly whilst we pursue higher order things (like writing and reading blogs).

This is what SF writer Iain Banks calls the "dependency principle" - complex software is based on a simpler layer beneath it, which is in turn based on an even simpler layer beneath it, until you get out of software and into the bare metal.

In the recent economic troubles - the credit crunch caused by the insidious spread of bad mortgage debt and the fact that banks now don't know how much these assets are worth (if anything) - can be thought of in similar terms to the structure of software.

A layer of abstraction is based on a simpler substrate: derivatives that are based on the risk of a given mortgage defaulting.

The difference between finance and computers is the layer beneath the abstraction isn't straightforward and predictable.