In his recent blog, "The Trouble with Retirement Planning Calculators," Mark Miller references a report by the Society of Actuaries (John Turner is the author) on 12 calculators that "had a host of problems." One of these calculators is ESPlanner (Economic Security Planner), which provides economics-based planning. I know this calculator very well since I developed it through my company. None of the problems usefully referenced by Mark pertain to our software. And Mark makes this clear in his piece, although in less detail than I would have preferred.
In an earlier version of this blog, I criticized Mark very strongly for not exempting, more precisely, our software, but my criticism was overblown and unfair to him and I apologize to Mark for flying off the handle.
Let me consider the six problems Mark correctly identifies.
1. Social Security Projections. Most retirees get a third or more of retirement income from Social Security. Yet many retirement calculators don't gather the detailed information needed to project these benefits accurately, Turner says. "They often project Social Security income using a bare minimum of information: typically your current earnings, your age, and the year you expect to retire," he says. The Social Security Administration offers the best projection tool, customized to your actual earnings history.
What Turner said is correct. But I differ with Mark on his faith in Social Security's projections.
Turner is right that every calculator, but one (namely ESPlanner), makes incredibly crude Social Security benefit calculations because they don't ask users to input their precise past covered earnings history or their projected future covered earnings. ESPlanner does precisely this and then proceeds to calculate survivor, child, retiree, divorcee, mother and father, and spousal benefits taking into account earnings reductions, early retirement reductions, delayed retirement credits, family benefit maximums, recomputations of benefits, windfall elimination provisions, offset provisions, and the list goes on.
In contrast, Social Security doesn't calculate for you your spousal, child, retiree, divorce, mother or father, or survivor benefits. It only calculates your retirement benefit. And when it does this, either online or in the annual benefit statement it sends us, Social Security projects zero economy-wide real wage growth and zero inflation in all future years. This is clearly a highly unrealistic assumption. Social Security makes this assumption because it's afraid that people will compare their future benefit with their current pay, infering a higher replacement rate than will actually end up being the case because their earnings will also likely grow through retirement. So for younger people, in particular, Social Security is significantly understating their likely future benefit.
2. Rate-of-Return Assumptions. Three of the free calculators used pre-set future investment rate-of-return assumptions that you can't change, and their percentages varied widely. One, created by the U.S. Department of Labor's Employee Benefits Security Administration, assumed a 5 percent average annual return from 401(k)s; several others assumed 10 percent. If a calculator won't let you choose your anticipated rate of return, either be sure you're comfortable with its assumption or walk away.
ESPlanner lets you set your annual return and also lets you change it. And if you run the Monte Carlo simulations, the mean return can change every year based on what you said you'll be holding.
3. Life Expectancy. It's impossible to know how long you'll live, of course. On average, 65-year-old men can expect to live another 17 years, and women another 20 years. Some calculators, the study found, automatically input life expectancy figures. But they fail to account for differences by race, income, and gender. And they also don't take into consideration that you or your spouse might live longer than the averages.
I differ with Mark on the length of the planning horizon. Economics suggests that the maximum, not the expected remaining length of life is the right planning horizon. You have to plan to live to your maximum age for the simple reason that you might. Maximum age of life is what ESPlanner uses.
4. Housing. The calculators make very different assumptions about what you'll do with your house at retirement. "Some assume you won't liquidate your home; others assume you will sell and downsize," Turner says. Very few of the tools analyze the impact on your finances of carrying a mortgage into retirement.
ESPlanner lets you do change your regular and vacation homes twice in the future and to specify mortgages that continue through retirement.
Among the free calculators reviewed, only the U.S. Department of Labor calculator lets you plug in home equity when calculating your retirement assets.
We now have a free version of ESPlanner available at www.eplanner.com/basic that lets you enter your home equity for both your primary and vacation homes.
5. Inflation. None of the free calculators -- and few of the professional tools -- listed inflation as a retirement-planning risk. Some of the tools let you plug in just one percentage forecast, even though inflation can fluctuate widely over time. Others put in their own default inflation rate, ranging from 2.3 to 4.6 percent. That spread can make a huge difference in how much the purchasing power of your assets will shrink over a 25-year retirement.
ESPlanner lets you enter changes in future inflation rates and encourages you to do What Ifs.
6. Spouses. Few of the free calculators helped couples forecast retirement income for a surviving spouse. They rarely let users enter separate information for both spouses and run numbers with differing life expectancies for them, for example. When the calculators recommended annuities for retirement income (most didn't), none suggested buying one with a survivor's benefit.
ESPlanner makes extremely precise calculations for survivors. Indeed, you can kill off your spouse at any age and see in very fine detail how you will do in every future year after the murder. An example of our being hyper anal here is calculating federal and state taxes in extremely fine detail separately for each year the survivor might be alive, conditional on the age the other spouse dies (or is murdered).
Laurence Kotlikoff is a professor of economics at Boston University, President of Economic Security Planning, Inc. and the author of Jimmy Stewart is Dead: Ending the World's Ongoing Financial Plague with Limited Purpose Banking.
Where I say:
The report analyzed 12 retirement calculators created by financial services firms, software companies, nonprofits, and government for consumers and financial planning pros. All but one of the six consumer calculators were free: the Fidelity Retirement Income Planner, the AARP retirement planning calculator, MetLife calculator, U.S. Department of Labor and the T. Rowe Price Retirement Income Calculator. ESPlanner, created by Boston University economics professor Larry Kotlikoff, starts at $149 per year. Unlike the freebies, ESPlanner gathers more detailed data, making its forecasts more reliable.
There is no calculation that predicts the future.
Thanks for your comment. Much appreciated.
We've spent 15 years at this and just getting Social Security's provisions right was torture.
Social Security is the primary retirement savings program for the entire population. It has 2728 rules in it's Handbook and thousands or rules in its Program Operating Manual System designed to explain the 2728 rules in the Handbook. On some issues, that should be remarkable simple like "If my husband is retired and I take spousal benefits early, am I forced to take my own retirement benefits early?" you'll go nuts trying to find the answer in any of the rules or the rules about the rules.
In the end, you're forced to talk directly to Social Security's actuaries, who do, fortunately, provide excellent assistance.
Anyway, you see where my overreaction to Mark is coming from.
Re the future, economics says that you need to do what's called Expected Utility Maximization that formally incorporates every type of uncertainty you face. This is conceptually straightforward, but impossible for any computer, either now developed or that could possible be developed, to do.
So our method in ESPlanner of dealing with this issue is to do dynamic programming with respect to one form of uncertainty -- returns on your assets -- and then let you very quickly run What Ifs? E.g., we let you raise future taxes or cut future Social Security benefits or raise future inflation rates or change your future earnings or change homes ...
All best, Larry