Behavioral Finance, Spreadsheets, and Bad decisions

Posted by Peter Orr on Jan 04, 2009

"The whole secret of mysticism is this: that man can understand everything by the help of what he does not understand. The morbid logician seeks to make everything lucid, and succeeds in making everything mysterious."

- G.K. Chesterton

Behavioral finance has been heralded as at once a new sunrise and false dawn in the annals of financial economics. However, behavioral finance has no unifying theory at this point though it has exposed a number of "cognitive illusions" which we human types tend to display when making financial decisions. And as often as those would claim that insights from behavioral finance sound the death knell for the efficient market hypothesis, others say it's impossible to determine whether the market is truly inefficient or that the market model being tested is wrong. Since behavioral finance offers no model of its own, it's impossible to test market efficiency under its finding. I wouldn't presume to add any real insight into this debate; I say let the debate rage on and a thousand more PhDs be granted. That said, I do question how or if the financial technology we surround ourselves with has been a contributor to our current situation…

Some behavioral finance findings relate to heuristic decision-making, the "rules of thumb" or educated guesses that we make in the face of complicated problems and uncertainty. For example, availability bias is the tendency towards overweighting information that is easily attained. Anchoring is the tendency towards extrapolating recent trends, possibly leading to an under-reaction to changing conditions. Overconfidence leads people towards over-estimating their predictive skills. Each of these three phenomena has been studied and documented as common in the human condition; even evolutionary psychologists have reasonable theories for some of these behaviors. But so what?

Consider these findings as they interrelate with our technology. The one nearly ubiquitous tool available to the masses in finance is the spreadsheet. I love spreadsheets. Spreadsheets can't be beat for certain purposes. However, I find that for measuring potential variability in a multi-factor risk setting, unenhanced spreadsheets display some pretty major shortcomings which I won't belabor here. Suffice it to say that in the absence of any other tool to more powerfully process information, if a spreadsheet is all that's available, an analyst will use a spreadsheet. People are forced to make a decision with what they've got, so often the only information that goes in is the stuff that can be reasonably quickly generated in a spreadsheet. Further, our natural inclination towards anchoring with recent data, as well as natural overconfidence in forecasts makes the spreadsheet the ideal medium for us to completely delude ourselves.

A MAD Example

Let me give you just one (of many) examples that frankly doesn't make any sense to me, particularly in this modern financial era. One liability related metric understandably deemed important by many analysts and certainly the rating agencies is MADS, or Maximum Annual Debt Service. It is supposed to represent the maximum of principal and interest payments that might be made by an issuer of debt over an annual fiscal cycle. It is one of those metrics that can be easily calculated in a spreadsheet by a user with only modest skills. However, this same user and likely worse, the audience for her/his analysis may be suffering badly from the cognitive issues described above.

The number of misunderstood, under-appreciated, and heroic assumptions that go into calculating MADS can be staggering. What assumption was used for calculating possible debt service on variable rate bonds or commercial paper? How about the performance of hedges such as interest rate swaps? Any basis variability? How about the likelihood of debt acceleration? Or perhaps a liquidity crunch which leads to either expensive or unavailable letters or lines of credit? What happens to debt service and MADS then? What does the "maximum" in MADS mean when all of the market assumptions that go into it are based upon some 10 or 20 year average, whose sole redeeming feature is that it's easily entered in a spreadsheet?

And MADS is an important number because it often is the denominator for statistics like debt service coverage ratios which are relied upon by investors, rating agencies, and even bond trustees. These are legally required calculations, and yet the amount of time and energy that goes into understanding their potential variability is often next to nothing.

IMHO, this is low hanging fruit which must be changed if we're going to improve disclosure, increase the value of our information and the density of its content, and ultimately enable people to make better decisions. What am I missing? What do you think?