Is there a way for PRC to treat Inflation as a stochastic variable in the Monte Carlo simulations?
From history it seems like long periods of high inflation are the biggest risk to retirement plan success, probably even moreso than bear markets and sequence of returns risk, so I want to make sure that inflation risk is modeled in my Monte Carlo simulations if possible.
I don't see where to input inflation standard deviations or whatnot, and I thought maybe I'm just missing something obvious so wanted to ask.
Many thanks!
@boston-spam-02101gmail-com No, there's no way to do that with MC simulations. But historical simulations do use historical inflation.
Stuart
Maybe Monte Carlo using historical block bootstrap technique could address this?
Thanks!
Maybe Monte Carlo using historical block bootstrap technique could address this?
@boston-spam-02101gmail-com, There is no inherent reason a parametric monte carlo like pralana's couldn't treat inflation as having its own mean and standard deviation. (Portfolio Visualizer's monte carlo does exactly that, for instance.) To your point, Pralana's use of constant inflation underestimates stock and especially bond volatility from inflation shocks. But, on the other hand, "random" inflation around a mean would ignore that bond yields and stock prices get determined based on expected inflation, so that isn't really accurate either. To properly utilize stochastic inflation, Pralana would need to utilize correlation matrices (that is: the individual correlations between each asset to each other, including to inflation)--a big change. (And, to state the obvious, it would require more knowledge and entries by users too.)
Keeping this in mind, it's probably best to leave things as they are, with the "work around" that users might increase their asset volatility a bit more than usual to account for the additional effects of inflation's volatility, especially for bonds. The designers might also consider using inflation-adjusted returns for the monte carlo (which then gets adjusted by constant inflation for "future dollars") rather than using nominal dollars (which then gets adjusted by constant dollars for "today's dollars"). The former incorporates inflation's volatility better than does the former as done now.
Do remember that monte carlo is meant to be used in conjunction with historical analysis to give a full picture. In contrast to monte carlo, Historical analysis includes periods with high inflation of which you speak, including reflecting the subtle effects of inflation on asset returns. At the moment is probably the best way to explore the effects of high inflation towards the beginning of the plan (e.g. 1970s). And, yes, @lancaster22, block bootstrapping would capture the effects of periods of high inflation.
Keeping this in mind, it's probably best to leave things as they are, with the "work around" that users might increase their asset volatility a bit more than usual to account for the additional effects of inflation's volatility, especially for bonds.
Nice reply. I've been thinking about the impact of stochastic inflation as well. Working with the system "as is", I agree with your suggestion above about increasing the volatility as probably the best approximation. It won't get items indexed to inflation (expenses, tax brackets, etc) exactly right, but I think it's better than not adjusting the volatility.
For a GBM model, I believe the volatilities can be combined as follows:
Asset Inflation Adjusted Volatility = SQRT( (Asset Volatility)^2 + (Inflation Volatility)^2)
So, for a few example assets:
1) Nominal Volatility (Annual, 1928-2024)
Inflation = 3.9%
SP500 = 19.5%
3mo Tbill = 3.0%
10yr Tbond = 7.9%
2) Inflation Adjusted Volatility
SP500 = 19.9%
3mo Tbill = 4.9%
10yr Tbond = 8.9%
