This post is similar to our post of June 21 in that it defends the Actuarial Approach advocated in this website against those who find the approach either too complicated or too simple. In this post, we will once again defend it against those who question its effectiveness because it doesn't employ Monte Carlo modeling.
In his Retirement Café post of July 11, Dirk Cotton argues for the use of the Monte Carlo modeling approach when planning for retirement. He says, "many planners and nearly all academics prefer to use Monte Carlo simulation, instead. Monte Carlo generates many outcomes and, unlike the spreadsheet approach, shows the distribution of outcomes."
While Mr. Cotton refers to the benefits of using Monte Carlo modeling over spreadsheet (or deterministic) approaches when planning for retirement, and not specifically for determining one's spending budget, his view is somewhat typical of academics and financial planning experts who believe that Monte Carlo modeling is the best thing since sliced bread and clearly superior to deterministic modeling.
The Actuarial Approach involves using a simple deterministic spreadsheet with recommended assumptions to model future experience (Future annual spending budgets and remaining accumulated assets are shown in the run-out tabs assuming experience exactly follows the assumptions and the annual budget is spent each year). Yes, we know that actual experience will deviate from assumed experience, which is why the Actuarial Approach employs annual valuations and a smoothing algorithm to adjust the spending budget for actual investment return, inflation, spending as well as any changes in assumptions.
No matter how many simulations are examined in a Monte Carlo model, it will not be able to accurately predict the future, and its projections of future experience will be just as wrong as under the deterministic model. As indicated in our post of August 13 of last year, incorrect projections of future experience (whether our "inferior" deterministic ones or the "more sophisticated" ones developed from tens of thousands of Monte Carlo simulations) will need to be adjusted to reflect actual experience as it emerges.
Don't get me wrong. Monte Carlo modeling can be useful for certain purposes. For example, it can be very useful for testing how successful two or more different withdrawal strategies might be in meeting client objectives or even in testing how different smoothing algorithms might work for the Actuarial Approach. But, for the purpose of determining this year's spending budget, it is not necessarily superior to the simple spreadsheet recommended on this website.
