"Using excessive return assumptions destroyed retirement planning for lots of clients. If those advisors had used reasonable assumptions, then they would have been acknowledging the fact that these bad times could happen, and their Monte Carlo simulations would have been incorporating these more reasonable return assumptions to their clients' benefits."
Indefensibly optimistic return expectations
Can you trust your glossy financial planning report? Perhaps you should take a look under the hood of the financial planning Monte Carlo engine your financial planner uses.
Emerging Information System (EISI) commands the biggest share of the financial advisory market with its NaviPlan* (advisor subscription cost: $800/year and up) and Profiles* software systems. EISI's website provides a sample "Comparative Analysis" report, dated July 2008, for two clients with the following return assumptions for some asset classes:
|Asset Class||EISI's forward-looking
annual return assumption
(converted to annualized)
|Appx "Equity Risk
Premium" over "Long Term Bonds"
(forecast annualized 3.9% return)
Note: TIPS not treated as an asset class
|Large Cap Value||10.7%||6.8%|
PIETech's Money Guide Pro boasts over 25,000 subscribers. On page 15 of this sample report, PIETech assumes an average 10.7% return for an investor's portfolio, asserting that this "is a good starting point," and uses this astonishingly optimistic return assumption as the basis of a Monte Carlo simulation.
Another major vendor, FinanceWare* (advisor subscription cost: $90 per client per year) recently published a white paper (dated April 2009) listing the equity return assumptions that the Monte Carlo engine in its "Wealthcare Capital Management" software uses for different asset classes. Here are some examples:
annual return assumption
(converted to annualized)
|Appx "Equity Risk
Premium" over TIPS
(forecast annualized 5.2% return)
|Total Domestic Equity||11.3%||6.1%|
|Small Cap Value||12.3%||7.1%|
A 2002 Financeware* white paper reveals that its forecasts were about equally optimistic in 2002.
Financeware* is commendable for having the transparency to publish their assumptions. But it is only fair to ask: what fundamental basis does Financeware and the other major vendors use to justify their optimistic equity return forecasts?
Are their forward-looking equity return assumptions defensible?
That's doubtful. In the early 2000s, there was a lively academic debate between Robert Arnott and Roger Ibbotson about the "equity risk premium." Robert Arnott took a pessimistic view. Roger Ibbotson had a glowingly optimistic view. But even Roger Ibbotson forecast an equity risk premium of only 4%. And his forecast depended critically on the extraordinary hope that the U.S. economy would grow, indefinitely, at an inflation-adjusted rate of 4% per year (When reading Ibbotson's paper, it's easy to miss his underlying turbo-charged growth assumption: but see his paper, at pages 13-14, 24).
Even Dr. Ibbotson's equity return forecasts, which are far more sobering than NaviPlan's* or Financeware's* equity return forecasts, provoke incredulity. Between 1820 and 2000, the United States, the world's most successful economy over that period, grew at an inflation-adjusted per-capita rate of only about 1.8% from 1820 to 2000. And this was a time of comparably explosive population growth. America's demographics are aging. Its population growth is slowing, and in a few decades will go negative. Yet Dr. Ibbotson's 4% real growth forecast (which he published in 2002) assumed that the best times for the American Empire were still ahead.
According to http://www.measuringworth.com, between 2002 and 2008, real U.S. GDP grew at a rate of about 2.5%/year, considerably less than Ibbotson's 4%+ forecast.
Some of the most respected research on equity premiums come from Elroy Dimson, Paul Marsh, and Mike Staunton of the London Business School.* Every year, they author the content for the Credit Suisse Global Returns Yearbook. In the Credit Suisse Global Investment Returns Yearbook 2009*, the authors forecast a global equity risk premium as compared to T-Bills of 3 to 3.5%. T-bills have historically yielded about 1% above inflation. Which means the Yearbook's implied forward-looking real equity return is about 4 to 4.5%, which (as of late July 2009) was only about a 2% premium to long-term TIPS!
The London Business School* provides a cogent explanation for their estimated 3 to 3.5% expected equity risk premium over T-bills. It's time for the leading commercial financial planning software vendors to defend their extraordinarily higher expected equity risk premiums over the same instruments.
If any of the leading commercial providers of financial planning software have published a cogent, analytical defense of their hyper-optimistic return expectations, please report it to Prospercuity.
One of the most influential theories in finance is Modern Portfolio Theory. In the 1950s, Harry Markowitz analyzed historical correlations between different asset classes and noticed that by diversifying one's assets between multiple negatively-correlated, non-correlated, or weakly correlated asset classes, one could dramatically reduce one's risk while still achieving an attractive rate of return. But Dr. Markowitz didn't stop there. He developed an algorithm for identifying the so-called "Efficient Frontier" the set of optimal mean-variance portfolios that delivered the maximum return for each given level of risk.
In 1990, Dr. Markowitz won a Nobel Prize for his research. Since that time, so-called "mean variance optimizers" have become a standard tool of financial planning.
The lessons of diversification are still valuable today. But mean-variance optimization (MVO) is considerably less useful and, in the wrong hands, downright dangerous.
MVO is backward-looking. Often, researchers calculate the correlations between different asset classes over some historical time period and hoping that those correlations will persist forever plug those same correlations into the portfolio-modeling program. Mean-variance optimization algorithms are, in fact, guilty of data mining finding that past set of historical equity components that would have generated the highest risk-adjusted return.
But correlations between asset allocations are not persistent. Over time, historically uncorrelated asset classes have gotten more and more positively correlated. And different asset classes tend to get highly correlated at the worst possible time in the midst of a financial panic. (The website http://www.assetallocation.com generates insightful graphs illustrating historical cross-correlations between, and the increasingly lock-step behavior of, major asset classes).
What makes MVO dangerous is that portfolio modeling programs that simply use historical correlation values generate expected return and standard deviation parameters for those portfolios that are unrealistically optimistic. Last generation's optimal portfolio is unlikely to be the next generation's optimal portfolio. Indeed, last generation's optimal portfolio is likely because of increasing correlations between asset classes to be considerably more volatile than a backtesting MVO would project. This, in turn, leads to the development and implementation of financial plans that are riskier than many investors can tolerate.
TIP$TER avoids both of the dangerous modeling flaws discussed above. TIP$TER forces you to think in terms of the equity risk premium, and its documentation encourages you to project it on the basis of expected future economic growth, rather than on historical returns. TIP$TER also avoids the fantasy of assuming that an investor can eliminate most of the risk by finding a mean-variant optimal portfolio. Instead, TIP$TER takes a more conservative course, modeling the equity portion of a portfolio as a single asset class.
That does not, however, make TIP$TER incompatible with MVO or Modern Portfolio Theory. You can plug the expected return and standard deviation values for a mean-variance-optimized portfolio into TIP$TER and let TIP$TER simulate its performance. But given the increasing correlations between asset classes, and the diminishing benefits of asset class diversification, Prospercuity strongly advises users to adjust the expected return downward, and the expected standard deviation upward, of any MVO-optimized portfolio, before running those values through a TIP$TER simulation.
Money Guide Pro is a trademark of PIETech, with which Prospercuity claims no sponsorship, connection, affiliation, or association.
Financeware, Inc., is a trademark of Financeware, Inc., which which Prospercuity claims no sponsorship, connection, affiliation, or association.
Credit Suisse is a trademark of Credit Suisse Group Société, with which Prospercuity claims no association, connection, affiliation, or sponsorship
The London Business School is a trademark of the London Business School, with which Prospercuity claims no association, connection, affiliation, or sponsorship