Andrew Lo Counts Uncertainties

Chidem Kurdas

If risk measures misled investors about mortgage-linked securities, what is wrong with the measures? Professor Andrew Lo of MIT offered a list of what’s missing from quantitative risk models. The root cause of the trouble, he suggests, is physics envy!

Speaking at a conference last month at New York University’s Stern School of Business, Mr. Lo presented a paper he’s written with a physicist, Mark Mueller. It shows that quantitative methods don’t catch the full spectrum of uncertainty. But because physics has been very successful, economists try to emulate physics by expressing everything mathematically.

As a result, only quantifiable risks got attention, while hazards that can be understood qualitatively but not mathematically were ignored. Physics envy can be hazardous to your wealth, says Mr. Lo. “We move from one level to another as we get to deeper levels of understanding,” he said.

He presented a taxonomy that starts with complete certainty and goes to irreducible uncertainty. An in-between category, risk without uncertainty, is amenable to quantitative measurement—these are situations where the unknowns, that is, possible outcomes and their probabilities, are known. By contrast, where there is irreducible uncertainty, the unknowns are, well, unknown.

What price will oil trade at when the leak from the exploded BP well stops?  That’s uncertain. But the level of uncertainty is not set in stone, Mr. Lo says. As knowledge accumulates, uncertainty decreases.

Messrs. Lo and Mueller applied the taxonomy to an equity market neutral hedge fund, analyzing various issues from trading to legal. This does not result in a single numerical measure of risk, but the checklist gives a more comprehensive understanding of the different types of potential problem.

It is certainly an inexact science. For one thing, people’s perceptions vary—estimates of odds are subjective. Some individuals’ subjective probabilities are more accurate than others’. For instance, John Paulson’s subjective probability for mortgage securities was pretty good, Mr. Lo quipped, as he gave the presentation at Paulson auditorium.

The hedge fund manager, an NYU alumnus, has donated to the school. For other presentations from this conference, click Edward Altman or Gary Gorton and Andrew Metrick.


%d bloggers like this: