I recently received notice that my paper, "Does Fraud Buffer against Fraud Shocks?" was accepted for publication in the Journal of Risk and Financial Management's special issue on "Fixed Income Securities," guest edited by Prof. Dolly King, Ph.D.
This is my fourth publication overall (first solo publication), and largely draws upon my third dissertation chapter. Greatly inspired by the work of my dissertation chair, Dr. Wade Pfau, my article uses data from the Ibbotson's Stocks, Bonds, Bills, and Inflation (SBBI) Yearbook 2019 edition to bring the Trinity Tables up to date through 2019 and to model both single incidence and serial fraud within a retiree's portfolio. I asked how damaging fraud could be to one's retirement portfolio success.
The short answer? About 3% reduced success for every incidence of fraud.
The longer answer is much more complicated. At the outset, it's important to understand that several assumptions were made (as is routinely the case in the retirement income planning literature):
No tax consideration. Most of the literature excludes taxes because it adds another dimension to the calculation. As is, the Trinity Study encapsulates four dimensions: asset allocation, time horizon, withdrawal rate, and retirement success chance;
Nice and neat asset allocation values. The Trinity Studies posit five different asset allocations (the ratio of stocks-to-bonds within the retirement portfolio): 100/0, 75/25, 50/50, 25/75, and 0/100. In reality, many retirees do not structure their portfolios like this. Even target date funds often do not drastically rebalance their portfolios in this way;
Conveniently planned time horizons. This is another departure from reality. While it's important from a modeling standpoint, we should always bear in mind that many retirees do not have the luxury of choosing their life expectancy (i.e. when they will die); and
Steady, "safe" withdrawal rates. Another variable in the Trinity model, withdrawal rates in the retirement literature assumes the retiree picks a percentage of the portfolio's value on day 1 of retirement and then proceeds to withdraw that much on an annual (or broken down into smaller chunks) basis. Again, however, life often works differently. For many, as expenses--particularly healthcare--increase, so to does the draw downs from their nest eggs.
I ran 10,000 Monte Carlo simulations using Octave (the free version of Matlab) to simulate stock market performance using data from 1929 through 2019. I simulated fraud shocks in five different ways including a no-fraud baseline:
No fraud (normal retirement outcome using 1929 - 2019 SBBI data).
Best-case scenario (3% fraud shock, Year 15 of retirement).
Random (between 3% and 10% fraud shock, anywhere from Years 1 to 15 of retirement).
Worst-case (10% fraud shock, Year 1 of retirement).
Serial fraud (randomized magnitude every year of retirement).
I then calculated the difference between (I) and each of (2) through (4) for each combination of the Trinity variables (asset allocation, time horizon, and withdrawal rate). I then averaged each difference, which produced the 3% value. It should be noted that in the case of serial fraud, there was no combination of variables that resulted in an acceptable retirement (defined in the literature as 90% or better probability of success).
I should like to thank Wade Pfau, William Bengen, David Blanchett, and the many others who have substantially contributed to the retirement planning literature and for their inspiration. Many disciplines don't have the luxury of current contributors sharing the same intellectual space at the same time with the field's legends. In that respect, financial planning provides a nice exception.
Now, onward to the next project!