The best selling Animal Spirits, by Akerlof and Shiller (creator of the Shiller home price index), explores how ignorance of the effects of human psychology on macroeconomic activity has led to periodic crises such as the one we are experiencing. It’s been reviewed by many (Google it), and provides a conceptual framework for rethinking economic and financial theory, which has become highly mathematical and often inaccurate.
The authors make passing reference to another author, Frank Knight, who in 1921 wrote an important book discussing the difference between risk and uncertainty. I just finished reading it as well and found it interesting as well as illuminating.
Knight draws a critical distinction between “risk” and “uncertainty”, two terms which we tend to use interchangeably. Risk is present when future events occur with measurable probabilities, while uncertainty is present when their probabilities are unknown. In other words, risk can be quantified, either based on historical results or observational testing. A fair coin, if tossed, will come up heads 50% of the time (historical results). The 20-year Treasury bond yields 4.36% today (June 24 2009) based on numerous (observed) transactions.
If we know the possible outcomes of a choice (such as an investment, in terms of its economic benefit or cost) and their probabilities, we can compute a probability-weighted expected value and decide rationally whether to invest. We may also be able to insure against a bad outcome (e.g. by hedging). This is a key difference between risk and uncertainty: risk, being fully quantifiable, is insurable because it can be valued or priced, while uncertainty is not. Uncertainty cannot be measured on historical bases, because it is irregular, and it cannot be observed because it is unique. Uncertainty cannot be insured against or eliminated because it cannot be quantified or priced.
One of the causes of the housing crisis was the assumption that house prices would rise indefinitely (and that there would always be buyers). The inventors of mortgage-backed securities considered this a certainty, when in fact it was an uncertainty, as proven by the catastrophic events of the last few years. It was not a risk – nobody can assess the probable trajectories of home prices – it was an uncertainty that was ignored.
It strikes me that most business value drivers – and the valuation assumptions we make – are uncertainties, not risks. We can rarely assess probabilities with confidence. Neither can business managers or owners. We are often confronted with situations in which assumptions and results can fall within reasonable ranges (e.g., a business might grow at from 3 to 6% per year), but we cannot be more precise than that, and we have no idea about the relative probabilities of different growth rates.
Yet much of the advanced valuation technology being developed, primarily in academia, such as option-based models – does NOT distinguish between risk and uncertainty. For example, we are permitted in financial reporting fair value assignments to use an industry volatility index as a proxy for the volatility of a private company. This assumes that (1) historical industry volatility results are applicable to the subject company and that (2) observed volatilities for the industry components are also relevant. These are the only two ways to assess probabilities, and they are assumed, not proven, in this case! These models interchange risk and uncertainty, and this can be fatal!