Last week I challenged the notion that financial markets are capable of making effective decisions about lending for investment in productivity, given all the uncertainty that businesses face about the future.
In last week’s post I could have used the word “risk” rather than “uncertainty” and it would not have changed the meaning at all. Unless you’re a mathematician.
In mathematics, risk is something that you can quantify. If you can work out the percentage chance of a particular outcome, for example based on historical data, then that is the risk of that outcome. Uncertainty is something that you cannot quantify – or at least, you cannot do it mathematically based on available data, it’s just your best guess. It’s possible to make an informed guess of the outcome of a football game between Brazil and Germany, but such a guess would be informed by our knowledge of the current teams, the quality of their players and managers, their recent form. If we say Germany have a 2:1 chance of winning, that’s our opinion informed by such analysis, it’s not a mathematical calculation.
Bookies determine odds based on the bets they are receiving, so prices quoted are the aggregate of all the informed opinions of punters, expressed through the prices they are willing to pay and the odds they will accept.
It would be completely irrational to base a prediction on the historical data of the result of every previous game. It would be plainly be ridiculous to believe that because Germany have won 20% of all the games they’ve ever played against Brazil, that their percentage chance of winning the next game is 20%. And no bookie would ever offer odds calculated on that basis.
So isn’t this how financial markets work – doesn’t the price of an asset reflect the aggregate of what all traders are willing to buy and sell for? Yes, but these traders are not looking at the fundamentals of all the businesses funded through sale of these assets. The typical models used to predict future prices of assets basically work on analysis of historical price data. It’s exactly the same as setting the price of a Germany win based on the outcome of all previous matches with Brazil.
In particular, they calculate the risk of default using a normal distribution. A normal distribution looks like this:
It exists everywhere in nature. If you were to plot the height of adult males in the UK, you would end up with a graph that looks like this. You will find endless talk of “tail risk” in finance literature, and of “fat tails” – this is the risk that the tail of the normal distribution is “fatter”, meaning there is a greater chance of extreme events. But financial risk does not follow a normal distribution, there is no logical reason to assume that it would do so, and the historical data does not support this assumption.
And what is happening here is finance economists and professionals are treating uncertainty as if it was quantifiable risk. They are assigning a numeric probability to uncertain future events that cannot be mathematically calculated by any valid statistical method. A brilliant exposition of the difference between uncertainty and risk, and the relevance of normal and other forms of distribution to financial markets, can be found in the book “Economyths” by mathematician David Orrell.
When we look at the future of the economy we face uncertainty, not risk. We can make informed guesses about the future, but we cannot calculate a mathematical probability of likely futures. Once you become aware of this distinction, it will stand out to you when reading economists who clearly don’t understand the difference, and indeed those who do (like Minsky or Mehrling).
Indeed, mainstream economists sweep uncertainty under the carpet. The standard models of national economies used by mainstream macroeconomists (Dynamic Stochastic General Equilibrium models, or DSGE) assume “perfect information”, or that the agents in the models have “all relevant information”. And worse, their models do not include financial markets at all. Financial markets are simply assumed to intermediate saving to investment with perfect efficiency, so no account needs to be taken of them. It’s like assessing the likelihood of a crash on a proposed new road system, and ruling out any chance of human error on the part of drivers.
Since the last crash they have started to tweek their models to incorporate financial markets as “frictions” – adding in a mathematical factor to allow for markets being less than perfect – when what they needed to do was recognise the fundamental flaws in the core assumptions of their models.
Like I said last week, the reality is that those purchasing financial assets simply aren’t looking at the fundamentals of the businesses that invest in the real economy. And beyond that, no-one is looking at the likely possible futures resulting from the myriad financial and economic investment decisions being made. It’s no wonder that “the market” keeps getting it wrong.
The empirical evidence presented in the opening 10 posts of this section of the blog is incontrovertible, that markets in their current form are making these decisions incredibly badly and the entire system is inherently unstable.
And there is one undeniable master of financial instability, who foresaw everything that we are living through now – so next week we will see, with the benefit of hindsight, just how perfectly accurate Hyman Minsky’s “financial instability hypothesis” was.