Dilbert - Financial black boxes


Possible the most failure fraught models (models seem to be my theme of the week) are the financial ones. My first two jobs were writing complex financial models. Curiously, my very first assigned was writing one of the first every models of financial risk to mortgage default. So I have a pretty intimate appreciation for how problematic they are in representing reality never mind predicting its future. Models are great tools for building understanding of dynamics – eg. directions, sensitivities, relative impacts. But for pegging absolutes, their weak track records speak for themselves.

Macrobusiness’ article ‘The Insufferable Conceit’ (thanks Chris) articulates the failures inherent in most financial models.

  • “Two problems plague the analysis of the financial system, problems that are related. Let’s call them the twin delusions. One is the persistent use of metaphors to characterise what is happening in the markets by people who do not seem to understand what a metaphor is, so they are seduced by them. The second is the mathematicisation of transactions, which creates the illusion that ‘the system’ can be analysed as if it is like a physical system subject to scientific laws. Let’s go to some basic definitions — to quote Marcus Aurelius: “What is it, what does it do?” Financial and economic markets are transactions. GDP, for instance, is not an indicator of wealth, or consumption, or growth. It is a recording of transactions. If there is a property bubble, for instance, consumption and wealth might actually fall, as people transact more aggressively over an existing stock of dwellings. I can remember an economist was asked by the Nebraska state government how to raise the state’s GDP. His response was: ‘Well, you could burn down the Capitol building.’ He was right, of course. That is why, for instance, GDP in Japan rose after their tsunami. As people cleaned up the mess, they transacted more. This is where the twin delusions hit. To characterise transactions, metaphors are created, or the transactions are turned into a metaphor. So, we talk of national GDP as if it is a metaphor for the nation. ‘Australia’s GDP went up 3%” = ‘Australia is doing well.’

Metaphors are used to help explain one part of a complex thing through simplification. To say a “leader is like a steering wheel” helps to put an abstract concept (“leader”) into concrete terms (“steering wheel”). It is a means of short-hand to describe the dynamic of one piece of a complex mechanism influencing the direction of the rest of the mechanism. But a ‘leader’ is not a ‘steering wheel.’ The whole point of using metaphor is for the sake of simplification. And in that simplification, many other attributes that a “leader” would have are lost. Just like when a summary statistic is entered into a financial model instead of representing every instance of financial activity. The statistic is a simplification, but it is not the activity. As Aurelius would say, it is not ‘what it is, what it does’.