Recently in an article about my take on Early Retirement I talked about sequence of returns and costs as a reason why I am not considering early retirement. Well, it just so happens that there is a crowd sourced tool available that really illustrates this point. Many of you may already be familiar with the tool known as Fire Calc and thus may want to skip down a few paragraphs. For everyone else a brief primer on retirement calculators:
FireCalc.com is a free crowd sourced calculator populated with the returns over every time period starting in 1871. The idea here is the tool gives you a picture of what likelihood, based on all of the prior performances of the US, your current investments would have of surviving the length you specify given the spending you specify. You can do something as simple: I have 1 Million dollars and I spend 50K a year, what is the likelihood of the money lasting 30 years? The tool will show you lines representing the amount the portfolio would have contained every year of that 30 years until expiry based on any starting point from 1871 to present. This calculates the likelihood of failure based on the prior performance of the market.
You can also add things like social security and pensions, adjust how the spending changes over time (based on inflation, based on various reduction studies, and based on a percentage of remaining portfolio), adjust your asset allocation across the US market, or with a consistent annual returns/random performance base on your own market return assumptions. Finally, it will allow you to work backwards from these assumptions to give you portfolio sizes needed based on your desired probabilities of success.
Vanguard Monte Carlo Simulator
Vanguard also has a retirement calculator. This one uses Monte Carlo Simulations. It uses returns in a database just like FireCalc. However, instead of calculating from different starting points it randomly chooses a return to represent each year’s return. This mapping is based on all prior year returns for an investment class. It proceeds through this for the number of years you outline. This becomes one possibility. It proceeds to run through this same simulation 5000 times and the result provides the probability of success.
Mathematical Limitations of Retirement Calculators
Both tests are very cool and a lot of fun to play with. Mathematically, they both have many positives plus two negatives based on assumptions. In the case of FireCalc the system gives you every previously existing scenario of a series of prices for the number of years you pick. The issue is that it assumes the series of prices will follow the same series produced sometime in history. This means there are only about 116 previous outcomes to compare to for a 30 year period, which is admittedly a small sample size.
Vanguard solves this by assuming that the prices of any given year are more representative. Thus they randomly select each years return into a series. It does this for 5000 iterations which is a better statistical approach. The issue here is that each year is treated as an independent random event, which ignores the tendency of the stock market to have momentum (move in the same direction over a series of years).
Retirement Calculator Usage Limitations
Beyond mathematics both have some serious blind spots that require them to be used carefully. First and foremost, prior performance does not necessarily indicate future results. Each one of these calculators assumes that future investment performance will have some commonality with past performance of the US market. This is not necessarily true. In fact, if we look at the Japanese market over the last 25 years we see performance of which the US has never experienced. The Japanese market is still down from it’s peak in 1989. Now, I will admit the likelihood of this with a balanced worldwide portfolio is low, but it is not impossible.
The Nut Behind the Wheel
My bigger fear about these retirement calculators relate to the inputs themselves by the “Nut behind the wheel” so to speak. First, we have the asset side.
When you input your assets into these tools, one thing commonly not accounted for is pre tax and post tax money. This will have a huge impact on your results. 1 million in a Roth IRA at retirement (post tax) is not the same as 1 million in a 401k (pretax). After the tax man’s 25% bite, you’re really looking at an investment of $750K. Furthermore, while we’re on the subject of taxes, they change. There is a possibility somewhere over your retirement, especially for those retiring early, that tax laws could change considerably. One such option often talked about is a wealth tax. Depending on how these are structured your portfolio could take a significant hit. There are some tax mitigating strategies you can take, but it is unlikely your tax rate will be zero.
Retirement Expense Assumptions are Complicated
The other input that concerns me is expenses. Cost prediction based on current spending is common in the personal finance world. However most real world retiree spending studies are based on those retiring at or near 65. These do indicate a reduction in cost with age. There are not very many studies of what happens to spending for someone who retires somewhere between 35 and 55. One would hope it will stay at a given level increasing by inflation only. However, there are many unknowns: future health care costs, insurance costs, inflation, etc. These may lead you to under estimate.
This post is not meant to scare. Chances are high that a prediction from one of these retirement calculators will predict the actual outcome. Also, chances are good that those who are truly committed to cost reduction and early retirement will overcome these risks. I strongly advise if you go this route to build in a cushion to the prediction and a backup plan. Anyone interested in personal finance should take some time to play with these tools. It will give you an eye opener to how much or how little (depending on your perspective) you will need to retire based on past performance.