Defending Monte Carlo Simulation

In spite of the criticism it has received recently, many advisors still see valid reasons for using MCS in clients' forecasts -- and several of them are more than glad to explain why.

It's been interesting to watch the commentary on Monte Carlo simulation (MCS) over the years. Initially, the forecasting technique was praised as an improvement over then-existing forecasting methods that relied on single assumed values for key variables. MCS offered a way to illustrate the range of possible outcomes around a forecast. Assuming the advisor explained and clients understood the underlying assumptions and the range of outcomes around the projected results, planning discussions of key decisions could move to a higher level. If that conversation didn't take place, then MCS was just another black-box forecasting technique.

The most recent bear market has revealed a weakness in using MCS that has been noted in the news media: If clients' retirement income plans don't allow for worst-case scenarios and the advisor fails to update clients' forecasts and strategies regularly, there is a risk the MCS projections will be wildly inaccurate.

In spite of the criticisms, many advisors still see valid reasons for using MCS in clients' forecasts. Tom Nihoul, CFP, is a senior financial advisor with Nihoul and Associates, affiliated with Ameriprise Financial in Spokane, Wash. He has been using MCS for about eight years and currently is using it as part of the NaviPlan software. "We're using it for basically everything: the education goals, retirement goals, if they have an accumulation goal," he says.

Nihoul takes several steps to help avoid the potential concerns being raised about MCS. One step is to regularly update the models with new figures; additionally, he updates each client's financial plan each year. "So we can take into account that two years ago things looked really good, and now when you run it and it's got the updated performance of the last two years, we can ask if we need to make any changes. With Monte Carlo, we can show clients what happens if we're down to the 50 percent probability, or what happens at the 10 percent level, so they say, 'Oh, there are some issues out there.'"

David Hultstrom, CFP, CFA, ChFC and president of Financial Architects LLC in Woodstock, Ga., has been using MCS since 2002. He's developed MCS-based spreadsheets that determine the reasonableness of a client's financial plan and their optimal asset allocation. He points out a potential problem in using MCS. "Many advisors confuse precision with accuracy," he says.

"Financial projections frequently have very high precision -- the numbers are to the dollar and percentages have decimal points -- but low accuracy: We have no idea what the numbers actually are. That doesn't imply it shouldn't be used, just that it should be used intelligently. It is just a tool, and all tools can be misused, particularly by novices," Hulstrom says.

"All of the inputs are put into portfolio optimization software to determine if the model portfolios are forward estimates of the parameters. The output is the input to the MCS. The recent silliness about 'fat tails' (excess kurtosis) applies a great deal to daily returns and barely to monthly returns. It doesn't appear to exist with annual returns. For the 83 years of good market data we have, there have been no three standard deviation events. Now we could wish for more data (and I do), but it looks like short-term momentum (irrationality?) is eliminated over annual periods, and annual periods are what the MCS programs are using," he explains.

Reprints Discuss this story
This is where the comments go.