Should a Fiduciary Use Historic Returns or Economic Forecasts when Making Retirement Return Projections?
What’s more reliable: Historic returns or economic forecasts? A long stagnant economy has placed increasing pressure on financial professionals to ignore historic data. This applies to any number of critical calculations for retirement savers – from long-term asset allocation to retirement withdrawal rates. We all agree we can’t predict the future, so it’s a matter of determining what is more reliable – however imperfect – at guessing the most likely result of any advice.
Therein lies the awful rub in terms of fiduciary liability. Whichever path is chosen requires sound reasoning behind the choice. For a fiduciary, that mean “defensible” reasoning, the kind that can stand up to judicial scrutiny. We’ve asked fiduciaries across the county to look at both alternatives and provide their assessment. Before we look at what they have to say, let’s review what the historical return data tells us.
Important Statistics Regarding Historic Return Data
For the purposes of this article, we analyzed Stern S&P 500 return data (http://pages.stern.nyu.edu/~adamodar/New_Home_Page/datafile/histretSP.html) in various rolling year scenarios from one year to eighty-eight years. The date contains annual returns from 1928 through 2015. Obviously, the number of rolling years decreased as we increased the number of years in the period. The data reveals several interesting statistics. First, with the sole exception of nine and ten-year rolling periods, the median annual return for nearly all rolling periods is solidly between 10% and 11%, i.e., just about where we’d expect it to be. Even for the nine and ten-year rolling periods, although the median annual return is below 10%, it’s close enough where one can justify rounding it up to 10%.
With that track record, why do so many financial professionals express concern over using the historically justified 10%-11% number? The answer is clear if we focus on the long-term returns beginning in 1928 and ending in each of the last five years. (For those of you interested in keeping score, this data represents the median annual returns of 82-year to 88-year annual rolling and reflects period-ending years from 2009 through 2015.) Here we see some of the lowest median return figures and, even worse news, the trend is towards lower returns. We can no longer justify rounding up to 10% as the data is closer to 9%. This appears to support economic forecasts which suggest lower stock returns in the future.
However, as has been repeated ad naseum, there are lies, damned lies, and statistics. A closer inspection of the internals of this data shows, for every rolling year period save one, the lowest annual returns occurred in periods beginning from 1928-1932 – the onset of the Great Depression. The sole exception was the 72-year rolling period ending in, surprise, 2008. While this correctly exposes the sequence of return risk we’ve been hearing about lately, it is also primarily an example of the risks of “snapshot-in-time” anomaly performance reporting. In other words, return periods that either start with significant losses or end with significantly losses will reflect abnormally low annualized returns. That appears to be what we’re seeing here.
To summarize, in looking at rolling periods typical for making long-term retirement projections (i.e., 35-40 rolling year periods), the historic data remains consistent with long-time assumptions. The typical median annual return for these rolling periods ranges from 10.9% – 11.04%. The standard deviation of this return data is from 0.87% to 1.04%. More importantly, the worst performing period ranges from 8.02% – 8.49%. These lows, however, all began with years beginning in the Great Depression. If we remove those years, we’ll get worst case annual returns about 1% higher, ranging from 8.31% to 9.65%. (For comparison, the highest annual returns for this rolling period range goes from 12.23% to 12.72% and are spread out over several decades, although there is a small cluster around the late 1990s, again reflecting the “snapshot-in-time” anomaly. For further analysis of historic returns and their impact on asset allocation, read “The Hows, Whys, and Right and Wrong Way to Use Asset Allocation,” (FiduciaryNews.com, June 30, 2015).
The Use of Historic Return Data to Make Retirement Return Projections
With a firm understanding of what the historic data actually says, practitioners use that information in different ways. On a fundamental basis, it quickly and easily shows investors the different risk/return characteristics of the various asset classes. “I use historic return data as a starting point – a baseline,” says Russell Robertson, founder of Alidade Wealth Partners in Atlanta, Georgia. “Most clients would have no idea what the differences are between say, emerging market equities and US Treasuries from a risk/return standpoint (other than ‘Treasuries are safer’) without seeing the historical data. If the client needs a target portfolio return in order to cover her expenses over the lifetime of the plan, historical data is a good starting point for what that portfolio looks like…or maybe a conversation about cutting back on expenses.”
Even when using the historic data as a baseline and then tweaking it, financial professionals know they are standing on firm, defensible ground when it comes to making projections. “We use historical returns, but we use specialized statistical techniques to account for the potential sample error in historical returns,” says Jim Lear, Chief Investment Officer for Guideway Financial, LLC, of Austin, Texas. “Failing to account for the sample error produces overly optimistic forecasts. The long-term historical data is the best unbiased estimate of the underlying system that we have, and throwing it away and replacing it with someone’s forecasting guess has shaky mathematical basis. As long as we are careful with the knowledge of historical sample inaccuracies, we can make reliable forecasts and risk estimates.”
At the very least, historic returns can be used to show “real-life” consequences of taking certain investment strategies. Nick Vail, co-founder and financial advisor with Integrity Wealth Advisors in Indianapolis, Indiana, says, “We use historical returns only when showing our clients some stress tests that will show how their portfolio would have reacted during different market downturns in the past.”
The practical usefulness of conducting a “stress test” cannot be understated. “Historical returns are useful in determining risk tolerance and setting client expectations,” Matt Hylland, an investment adviser for Hylland Capital Management in Virginia Beach, Virginia. “Historically, 10% corrections have occurred once every year or so, and 20% corrections roughly every 3 years. I’m comfortable telling clients to expect that trend to continue in the future to make sure they do not panic when the inevitable decline happens again. If I show a client what their portfolio would look like after a 20% decline and they go pale, I can be pretty sure they need a less volatile portfolio.”
Perhaps the most often used application of historic return data is to take those numbers and then apply a “haircut” to make the projections more conservative. “I explain to clients past performance of different asset classes, then I generally use a 4-5% return when doing retirement calculations based on long term historical returns of a balanced portfolio net of taxes,” says Ilene Davis, a registered rep at Financial Independence Services located in Cocoa, Florida.
Historic return data, therefore, represents a convenient and inarguable “comfort” zone for the typical fiduciary. “I use historical return data to project a client’s rate of return for the future,” says Gilbert Cerda of Cerda Munoz Advisors, Inc. located in Downey, California. “I take it a step further and incorporate a range of +/- 2% from such data since, as you said, no one knows what the future holds. It’s a good tool to discuss clients’ goals, investment objectives, fears, etc… I am comfortable using historical return data because, as Mark Twain said, history doesn’t repeat itself but it sure does rhyme. When you study the history of the markets, they tend to work in cycles.”
But, as we have seen, historic returns have their limits, and the fear of continued and long-term economic stagnation has people rethinking how they use that data. “I used to use historical data to stress-test financial plans,” says Paul A. Ruedi, CEO of Ruedi Wealth Management, Inc., Champaign, Illinois. “However, the downside is that if you use only historical returns, or sequences of returns, you are operating on the assumption that you won’t have worse markets than have ever existed. This is not a healthy idea.”
The Use of Economic Forecasts to Make Retirement Return Projections
As with historic returns, using economic forecasts to generate projected returns can be used for the purposes of a stress test. “We use projected returns to help us gauge the amount of volatility that could be expected should we see a market correction or large uptick,” says Alex Sylvester, Retirement Plan Advisor at Assured Partners in Indianapolis, Indiana. “I believe this is one of the many evaluations that must be completed in order to provide the due diligence for selecting and monitoring a retirement plan’s investment lineup.”
The use of economic forecasts provides another source of return data that can broaden the understanding of the variance of potential returns. Robertson says, “I use other return projections in order to set expectations for clients. I explain that long-term averages are simply that – long-term averages. It is highly unlikely that they will experience the exact long-term average over their (much shorter) retirement period. Additionally, even if the average return ends up being the same, the sequence of returns could vary enough year by year that it still has an effect on their overall portfolio. So for example, I will tell prospective clients that the long-term average return of the US stock market has been about 10%, but for X, Y, and Z reasons, I expect it will be closer to 3% for the next decade.”
Then there’s the hybrid approach. Economic forecasts can help determine the appropriate haircut to apply to the historic data. “We use projected returns that are around 1.5% lower on average than historical returns have been,” says Vail. “We use these because many economists believe that future returns will be less robust than historical returns. If returns end up being closer than historical, great. Our clients will be even better off in that case. However, we would rather be cautious and plan for lower returns.”
Another way to employ economic forecasts is to better assess the sequence of return risks. The historic data provides examples of what happens if you retirement at the wrong time, but it’s not helpful on a forward looking basis. That’s we’re economic forecasts might offer their greatest value. “I think what any great adviser is doing is to first, recognize that not only do we not know what our rate of return will be, but we do not know what the sequence of those returns will be,” says Ruedi says. “Therefore, it is critical to create simulations that have economic reality built into them, i.e., your capital market assumptions should create a distribution of returns that are somewhat realistic. For example, when using my capital market assumptions for the broad US market in my simulator, at the median I get a return of about 11% which stacks up well with the actual median return. At around the 75th percentile, I get a return of approximately 8.6% which is close to the worst “annual” return for the broad US market. Close to the 90th percentile of returns, I get a return of around 7.2%, which is close to the worst 30 year return for the broad US market using monthly data.” [Editor’s Note: the actual worst 30 year return is 7.97%.] “In essence, around 25% of my distributions for returns for the broad US stock market using a 30-year period are worse than we have ever experienced. I think that is more than reasonable.”
Even tools that traditional rely on historic returns can benefit from economic forecasts. “We don’t use non-historical based (NHB) expected return projections, but there is one advantage to using them rather than historical data,” says Lear. “In general, Monte Carlo and other statistical tools generate an optimistic bias when they are fed historical data. Many tools also allow NHB expected return and standard deviation projections to be fed into Monte Carlo. The NHB assumptions are usually more pessimistic that the historical data, potentially helping to reduce the optimistic bias.”
The Problem with Using Historic Data
With all the investment in tools that incorporate historic data and the general availability of such data, the fiduciary runs the risk of running afoul with the Securities and Exchange Commission (SEC). The SEC mandates that all RIA’s employ the disclaimer “Past results do not guarantee future results.” But isn’t that exactly what’s being implied by using historic returns to make retirement projections? And aren’t advisers taking a huge fiduciary risk by continuing to use historic returns?
Why would the SEC impose this restriction? “The SEC is government,” says Davis. “It doesn’t have to have a reason. And as for advisers, it’s the only thing we have to build on. Plus if you look historically, through good and bad times, a 6% gross return doesn’t seem unreasonable if we assume the world doesn’t totally collapse, in which case it won’t matter.”
One advantage of using the disclaimer is to reduce investor expectations to a more realistic level. “The SEC has to demand we disclose that past performance is not a guarantee of future results because clients often have unrealistic expectations,” says Vail. “It is important for us as financial advisers to educate our clients on risk and explain to them how the amount of risk they are taking can impact their portfolio – both positively and negatively.”
That being said, the real fiduciary risk isn’t the disclaimer itself, but violating the spirit of the disclaimer. Lear says, “This doesn’t mean that using historical data is improper. Using historical data improperly is improper. Too often in the past, advisers have recommended that historical average be used as an assumption of future returns. An assumption of average returns has a 50% chance of leaving the investor short of the plan, effectively focusing the client on high returns without little regard for risk. The SEC recognizes that investors must be informed that future returns may be lower than past returns.”
The reality is today’s world is a rapidly changing world. What worked in the past may not work in the future. “In our times, anything can happen, just because an investment has done particularly well in a niche it has doesn’t mean that the next best thing can’t come out the next day and steal its thunder,” says Michael Cirelli, a Financial Advisor at SAI Financial Services, Warrenville, Illinois. “We live in a world of rapidly changing technology and when it comes to investments there is no ‘sure thing’ nor are there any guarantees. I continue to use them because what else can I base a projection off of? Time and time again I’ve had clients ask me after setting up a new allocation model or managed account what this portfolio would have done during the recession of 2008. Past performance data (again, so long as all underlying investments can be tracked) is the most efficient way for me to show my clients details such as those.”
The generally beneficial application of historic returns caused some to question a too broad interpretation of the intention of the SEC’s disclaimer. “It is ironic that the SEC demands that advisers use the ‘past results’ disclaimer because the past is the most reliable source of expected future returns,” says Cerda. “One of the reasons, however, why the disclaimer exists is because the future is uncertain and you are investing in equities. Anytime you invest in equities, there is really no guaranteed that you will get your money back.”
In fact, the word “guaranteed” is probably the main culprit here. While bank deposits can be guaranteed, investments are not. The SEC wants all people to know and understand that. “No investment is 100% guaranteed,” says Sylvester. “Unfortunately, the world we live in is to disclosure any and every thing in order to provide some protection against a less than satisfied investor. Think about an investment like you would think about an employee. If an employee has been a great worker with better than average attitude in the past, they most likely will be a great worker with better than average attitude in the future. This is not foolproof but usually past performance is a good indicator of the performance you would see in the future.”
Indeed, the way one makes projections will say a lot about the extent of the fiduciary liability exposure one assumes. If it’s clearly stated that the intention is to test for a range of possible results, then it’s clear the historic data is not being used to predict future returns. Ruedi says, “The idea is not to predict the future, only reasonable distributions of potential outcomes that the plan can be tested against.”
The True Danger of Economic Forecasts: The Dark Side of Market Timing
Market timing is the practice of utilizing investment strategies not one the basis of fundamental analysis, but on the belief in the existence of some form of timing advantage as it relates to that particular strategy. On the face of it, economic forecasting can be seen as a kind of fundamental analysis. But, is it similar enough to be called “market timing.” “I have no real idea why someone would say they are similar,” says Davis. “Market timing is pure guessing.”
Vail says, “We use projected rates of return in our financial planning but that is not the same as market timing. We are still looking to create a strong asset allocation based on risk tolerance and goals, and then holding that policy for the long-term. Market timing is an investors’ worst enemy. We see the DALBAR QAIB study that comes out every year that shows how investors underperform based on market timing. We prefer to manage risk for our clients and take a long-term view.”
Sylvester agrees. He says, “I do not see using non-historical based projections as market timing. I believe in using as much data as possible to make informed decisions. Advisors will continue to use projections because it is our duty to gather the most information possible and create the best output for our clients.”
But there is no consensus on this. Many do see the use of non-historical based return projections as a type of market timing. “Return projections based on things other than historical returns essentially acknowledge the fact that the long-term historical average includes periods where returns were higher than average and periods where returns were lower than average, and try to capitalize on that by investing more in the higher return periods,” says Robertson. “This could absolutely be considered a form of market timing, and I would argue that this type of timing strategy is a necessary tool in managing portfolios in today’s policy-driven markets. Advisors will continue to use other-than-historical return projections because it differentiates them from other advisors and becomes part of the value proposition that they offer to their clients.”
The linkage is clearly apparent to some, and the evidence against it formidable. “Making non-historical based future projections is very similar to market timing and, unfortunately, market timing doesn’t work,” says Cerda. “It is true that you can get lucky and guess correctly a few times, but it’s nearly impossible to consistently time the market correctly. The Dalbar study illustrates this phenomenon. For instance, in 2015, the average equity mutual fund investor underperformed the S&P 500 by a margin of 3.66%, while the same is true for the 20 year period ending in 2015, where the gap was 3.52% (per year). http://www.talkmarkets.com/content/us-markets/dalbar-2016-yes-you-still-suck-at-investing-tips-for-advisors?post=96516. I think advisors continue using non-historical based future projections for several reasons: 1) it’s sexy and cool if you will; 2) it sounds great on paper and humans are driven by fear and greed; 3) it’s a marketing strategy based on emotions, not logic. The sad reality is that people who follow this end up losing thousands (if not hundreds of thousands) of dollars by following such strategy.”
Unlike trying to determine short-term economic forecasts, the use of long-term historic returns avoids the temptation of market timing because the long-term data is fairly consistent. “Non-historical-based projections and market timing are close cousins,” says Lear says. “Using NHB projections to determine how much to save is not market timing. However, using NHB projections to select the investment allocation is effectively market timing. Historical data, especially long data sets from indexes, does not change much from year to year, and so it does not lend itself to market timing.”
There is a plethora of studies which refute the concept and practice of market timing. Cirelli says, “They could be similar in the way that a non-historical based projection could be made for an investment that looks to do well over the coming months, quarters, or even years. Whereas market timing does something similar in that it tries to move client assets in and out of different sectors/industries when the adviser believes (or projects) they will fall in or out of favor. I know Charles Schwab conducted research on market timing where they concluded that, “the cost of waiting for the perfect moment to invest exceeds the benefit of even perfect timing.” (http://www.schwab.com/public/schwab/nn/articles/Does-Market-Timing-Work) I suppose advisors continue to use these projections when they are put in a situation where it is all that they have to go on when delivering the idea to the client. I feel that these projections however, may be ill-guided and misleading which is why other methods should be shown in tandem with these projections to give clients a more thorough look at the investment.”
So, Which to Use?
The consensus among those interviewed it to use all available means to create a range of long-term scenarios, rather than one single prediction. Above all, if it smacked of market timing, nobody liked it. Perhaps Ruedi sums it up best when he says, “I would use any reasonable distribution of returns before I would use market timing. Nobody can consistently gain a timing advantage over the market, at least you can’t pick the one that might do it by pure random chance any more that someone flipping a coin and getting heads ten times in a row. It could happen, you just can’t predict ahead of time which one will.”
Are you interested in discovering more about issues confronting 401k fiduciaries? If you buy Mr. Carosa’s book 401(k) Fiduciary Solutions, you’ll have at your fingertips a valuable reference covering the wide spectrum of How-To’s every 401k plan sponsor and service provider wants and needs to know.
Mr. Carosa is available for keynote speaking engagements, especially in venues located in the Northeast, MidAtantic and Midwestern regions of the United States and in the Toronto region of Canada. His new book Hey! What’s My Number? – How to Increase the Odds You Will Retire in Comfort is available at your favorite bookstore.