Greg Smith, and PERA’s Survivability Analysis

Last Thursday, Colorado PERA Executive Director Greg Smith gave his annual SMART Act testimony to the Joint Finance Committee of the Colorado General Assembly.  Since 2012, all departments of the Colorado state government have been required to testify during the interim concerning their operations, their efficiency, and the degree to which they are fulfilling their mission.

PERA recently changed its assumed long-term rate of return from 8% to 7.5%.  In light of that change, its amortization period – the time when PERA will have no unfunded liability, assuming a constant rate of return – has grown from 30 years, probably to something near 40 years.

During the Q&A session, State Rep. Lori Saine (R-Dacono) asked Smith about using a Monte Carlo simulation to test PERA’s long-term soundness.  As described in more detail here, averaging a given return over a period of time isn’t the same thing as getting that rate of return every year.  PERA’s portfolio might well average 7.5% over 40 years, and still go bust because its returns in the next few years are below average, leaving to try to make up the difference from a lower balance.  Rep. Saine was asking if PERA did or could run a simulated 40-year set of returns, and see how often the fund went bust and how often it stayed solvent at the end of that 40-year window.

Mr. Smith replied that yes, they do Monte Carlo simulations, and then proceeded to describe not those, but instead a sensitivity analysis available in the Comprehensive Annual Financial Report (CAFR).  The sensitivity analysis is something else altogether.  It looks as what happens to PERA if the returns over 40 years are 6.5%, 7.5%, 8%, up to 9.5%, but it still assumes a constant rate of return.  This is a completely different analysis, one that admits the possibility of lower-than-expected long-term returns, but ignores the danger in a few years of very low short-term returns, or even a couple of years of severe losses.

Monte Carlo simulations model the year-to-year variability in return that is an inherent function of risk.  As a result, even funds that appear to be well-funded, or that appear to have a long-term path to being fully-funded, can show low likelihoods of staying solvent through their amortization periods.  Doing such an analysis helps to prevent unpleasant surprises, and is to be preferred to a simple sensitivity analysis such as the type PERA performs and publishes.

Regardless of one’s preference, the two shouldn’t be confused for each other.

You can hear the entire exchange here:

Comments are closed.