Regardless of where I work, simulation has crept into my financial career. After
nearly a decade of working with it in many capacities I’ve found it to be
a mixed blessing. In many investment companies when the term simulation is
simply brought up there are a variety of reactions. The two most visible camps of
thought seem to be the utilizers, who think the results of a simulation have value
and the skeptics, who think simulation overcomplicates analyses.
The utilizers believe that when a concept or instrument is researched correctly,
information parsed and calculated properly, and a simulation constructed in a
statistically correct manner, the results can be used to make decisions. I tend
to fall into this camp, with a few caveats I will mention later, because I have
seen its utility in a variety of settings. Infrastructure deals that I saw early in my
career that involved vehicular traffic, trade, or passenger flows, made more sense
through simulation results given the wide variety of scenarios that could play
out over time. A commodity company investment that I worked on at Citigroup
involving soybeans seemed more appropriate after seeing the historic volatility of
soybean prices and how their expected evolution might affect our exposure. In
my structured finance career, the value of simulation on a very granular level for
distressed mortgage-backed securities provided insight into obligor delinquency,
default, and eventually expected security value loss. More recently, as I moved into
private equity, simulating pools of corporate exposures and fund performance has
become an important tool in assessing portfolio risk.
With all of these positives, there are some valid criticisms of simulation that
are espoused by the skeptics. Relating to the overcomplication arguments is the
thought that simulation is complex and that many mistakes can be made. I agree
with this criticism, and one of the caveats that I alluded to earlier is that simulation
must be implemented correctly for it to be useful and productive. I have
seen simulations fail for a number of reasons, but most relate to poor implementation.
In one transaction that I saw taken to a credit committee, the simulation
implemented was purely derived from Excel’s random number generator creating
numbers based on a uniform distribution. No analysis was done around the appropriate
distribution, and the CEO, who had an actuary background, instantly
criticized the presentation.