Leta i den här bloggen


Banks, computer models and Risk, Brady

Härom skrev Nicholas Brady, former US Treasury secretary, i Financial Times, August 26, 2012

Han har givit namn åt Brady Bonds, dollar-denominated bonds, issued mostly by Latin American countries in the late 1980s, a novel debt-reduction agreement for developing countries.  

LDC Debt Crisis, Argentina, Brazil, Mexico etc, remember?

Jag skrev en bok om det, Den Stora bankkraschen, Timbro, 1983. Det vore synd, skrev jag, om västvärlden skulle drabbas av en djup ekonomisk kris just när Sovjetväldet håller på att vittra sönder ekonomiskt, politiskt och moraliskt.

Brady:  The days when a large multinational bank could borrow money from depositors – backed by a government insurance guarantee – and then lend it out to customers at a reasonable spread are gone, at least as their main source of revenues. 

With net margins squeezed, banks have had to reach for profits in increasingly esoteric areas, such as derivatives trading, which rely on complicated computer models and mathematical algorithms few people understand.

Most of these activities boost profits by cranking up leverage, which the banks feel comfortable using because of confidence in their computer models.

This computer modelling is impressive stuff. However, while these models create the appearance of mathematical certainty about the relationships between markets and the way world events will affect prices, it is essential to recognise that, at their root, these models rely on man-made assumptions about human behaviour – not iron-bound laws of nature.

In addition, the behaviour of derivative markets can be episodic and illiquid at precisely the times we most need greater liquidity and confidence. 

No matter how sophisticated the maths or how large the data base supporting a model, no one can predict behaviour – human or market – with certainty. Inevitably, this means the formulas break down at the most critical times.

Mer här

Inga kommentarer: