Problems with retirement calculators
It’s been a while since I’ve discussed their weaknesses and the options. There are a few updates to existing programs along with a new kid on the block from USAA. Before I start naming programs and websites, though, let’s look at the concepts behind the various choices. Maybe by the time we’re done with this post you won’t trust calculators.
Why use a retirement calculator? (If you find a career that you truly enjoy, retirement is irrelevant!) The best reason is so that you know when you’re financially independent. A calculator helps you decide when you can stop working for money. Another important reason to use a retirement calculator is to appreciate how your investment decisions will affect your portfolio, and how your spending plans will be limited by it. The answer to “Am I financially independent?” depends on how much you’re planning to spend and how much you have to support it. Those parameters have a lot of flexibility, and your plans might change during the rest of your life.
Most retirement calculators fall into two broad classes: historical data and simulated data. The former attempt to tell you how your retirement assets and plans would stack up against years of historical data. The software assumes that you’d retired during a certain year and then applies the subsequent historical data against your portfolio and your planned withdrawals. At the end of the run, the calculator starts over and assumes that you’d retired one year later. The runs continue until the calculator reaches the end of its historical data.
Simulated data uses assumptions about market volatility, historical performance, and asset classes to generate sequences of returns. The returns aren’t the actual performance data but rather just a number of sequences that could have occurred within those assumptions. The calculator does as many runs as necessary to generate a statistical analysis of your portfolio’s performance. A popular type of this simulation is called “Monte Carlo” for the methods used by casinos to optimize the probabilities of their gambling games.
Both types of calculators have significant problems. For example, the historical calculator uses roughly 130 years of stock-market data for popular indexes like the Dow-Jones Industrial Average. However a retirement can easily last 40 years, which means that there are fewer than 100 sequences of data running for 40-year periods. There’s even less data for recently created asset classes like small-cap value stocks, international stocks, and Treasury I bonds. It’s hard to pile up a convincing set of statistically valid results from these “small” sample sizes. We’re not going to build up a thousand years of stock-market data, either.
A major drawback to the historical calculators is the assumption that the future will be no worse (and no better!) than the past. Sure, we’ve seen two world wars and a number of other global changes during the 20th century, but “this time is different” happens every day. The calculators don’t handle very high volatility or new asset classes. They don’t account for changes in correlations between asset classes over time.
Simulated data also has problems with the sequences of returns. Markets can steadily rise or fall across the years of bull or bear markets, but simulated data runs choose the years randomly without accounting for this persistence. Simulations suffer from sparse data on new asset classes or may put arbitrary limits on volatility. (They can’t create much variation with only a few decades of history on which to base the their limits.) They also don’t account for changing correlations between asset classes.
Simulated data struggles with catastrophes like the extremes of the Great Depression or the 2008 freeze of the credit markets because the markets are random series of events. Sure, they tend to go up more than they go down, and they usually change within certain limits, but they’re still random events that could just as easily decline over a number of years (1966-82) or drop a thousand points in a day. However the first assumption made by every simulation retirement calculator is that the performance of stock markets fits into a bell curve. That allows the calculator to apply probability and statistical techniques of analyzing the bell curve, even though the “real” stock market can still exceed those parameters. The calculator’s assumptions may handle 99.994% of the possibilities, but that’s not much comfort if your retirement happens to stumble across one of the .006% realities.
Both types of calculators assume that retirement spending is constant (even if it’s adjusted for inflation with “constant dollars”). Yet retiree spending varies. When the market is down, we all tend to defer spending in some categories. When the market is up, we tend to spend more money in almost all categories. Investor psychology has identified and analyzed those behaviors, but retirement calculators don’t do a very good job of simulating them.
A few retirement analyzers attempt to sidestep these drawbacks by iteration. Instead of simulating an entire retirement, retirees use their methods to assess a portfolio’s performance each year and decide how much spending the portfolio can handle next year. When the markets are down, they’ll recommend spending cutbacks. When the markets are up, they’ll support a full spending budget but suggest that some excess be set aside for a potential future bear market. The system is similar to “negative feedback” used in cruise controls and autopilots. However it’s not much help in predicting when you’ll be financially independent.
After decades of perpetual debate about the “best” retirement calculators, what retirees really want is some sort of insurance against the (very few) situations that nobody could predict. What if you don’t want to trust your life to a retirement calculator, let alone attempt to explain it to your spouse? What happens when the calculator doesn’t predict the right future? What’s your “Plan B”?
One extreme plan would be to work until you have too much money to spend. Groucho Marx famously claimed that he invested all of his retirement money in Treasuries. When a reporter said “Nobody can live on Treasuries!”, Groucho’s answer was “You can if you have enough of them.” But if you’re working until you join the asset ranks of Warren Buffett or Oprah Winfrey, then you’re probably not reading a blog about financial independence. You’re having too much fun to ever retire. If you find yourself in that situation then keep on doing it!
Another option is “Oh, well, I’ll get a job”, even if it’s a Wal-Mart greeter or neighborhood handyman. This is a great assumption for early retirees in their 30s or 40s. The reality is more difficult for retirees in their 50s, especially after they’ve been out of the job market for a few years. It’s even less appealing for retirees in their 60s or 70s who could have very real physical limitations.
Another choice is a “bare bones” budget for bad times. Again you’ll assume that the worst is unlikely to happen, but you’ll have a contingency plan. This might be a reserve fund that you set aside from your portfolio to tap into once a decade. It might be severe spending cutbacks to the point where you’re sharing group housing with several roommates and growing your own food. It might be attempting to live within your Social Security annuity, or purchasing additional annuities before retirement to guarantee an absolute minimum level of income… as long as the insurance company (and the federal government) pay out.
One popular compromise is “living off the portfolio”. This is a plan to spend only the dividends without ever touching the principal. This works quite well for private universities and inherited family wealth, where each generation is just a custodian of the assets while living off their income. Of course those situations aren’t much help to workers who’d like to know the earliest date they’re eligible to join this club.
Portfolios have survived for centuries and you can’t lose your wealth if you’re only limited to spending the dividends. However your spending limits can be quite volatile unless you pick the right diversified combination of dividends. Another “worst case” is when inflation eats into your dividend returns. A final issue is that (like Groucho’s portfolio) today’s dividends aren’t as good as they used to be. Even broad equity indexes are only returning about 2% in dividends, and dividend funds (or individual dividend stocks) may achieve higher rates only with much higher volatility and risk of loss. This historical deviation has only been around for a few decades so it might not be a trend, but this is scant comfort if you’re only going to be around for a few more decades.
With all these flaws, why use a calculator at all? Well, some of us investors enjoy the analysis process almost as much as the results. (Guilty!) If you’re worried about your retirement, then plugging numbers into a calculator helps you worry constructively. A few investors try to figure out whether they can maintain their lifestyle or whether they need to make major changes– either to get to retirement or to stay there. Best of all, a calculator acts as a warning system: you’re reassured that you’ve covering all the bases, or you learn that you need to reconsider your plan.
In a future post I’ll discuss the most popular calculators– and the most useful ones. If you can’t wait to dive in, though, consider a good hard workout of the free FIRECalc.com, a paid subscription to FinancialEngines.com, and an overview of Bud Hebeler’s “Analyze Now!” system.
That retirement calculator may be lying to you
Details of the 4% Safe Withdrawal Rate
“Present value” estimate of a military pension
Effect of inflation on a dollar
During retirement: take small financial steps (part 1 of 2)
How many years does it take to become financially independent?
Retiring on multiple streams of income
Military retirement: how much can I really spend?
Military retirement spending: how much will I need?
Does this post help? Sign up for more free military retirement tips via e-mail, Facebook, or Twitter!