Advisor Perspectives welcomes guest contributions. The views presented here do not necessarily represent those of Advisor Perspectives.
Far too often, advisors accept beliefs and practices that are detrimental to the financial wellbeing of clients. By reexamining them, you can achieve better outcomes for your clients.
Let’s jump right in.
The MPT problem
Modern Portfolio Theory (MPT) is an elegantly wonderful confluence of insight and mathematics, worthy of the Nobel Prize that was awarded to its developer, Harry Markowitz. Efforts to implement MPT, unfortunately, have not been as fruitful as the theory.
The reason is simple. To build an “optimal” portfolio using MPT, you must know three things – the future expected returns, volatilities and correlations of the asset classes you use in the portfolio. The problem is no one knows what those numbers are.
This has not stopped highly paid analysts from developing surrogates for the real thing. They go by the name of “capital market assumptions.” This moniker has a reassuring ring to it, which masks what they really are: guesses about the future.
The problem, of course, is that these numbers are not fixed, but change every year, and their historical patterns and sequences are not likely to be repeated in the future. They just keep changing and rarely, if ever, line up with capital market assumptions.
You can look at historical averages, but these numbers change over time, too. In his original MPT paper, Markowitz said that history might be a good starting place, but that ultimately the “judgement of men” would be required to refine the estimates.
Whatever methodology is used, it better be spot on. Small differences in optimizer inputs can produce dramatically different outcomes. And the methodology needs to work equally well in estimating returns, volatilities and correlations across a variety of dissimilar asset classes.
Optimal portfolios are like unicorns – they don’t exist in real life. That wouldn’t be a big deal if we all could simply accept that fact and move on. Instead, we act as though our capital market assumptions had a magical predictive quality. If our models tell us to trade, we trade, thus incurring transaction and, possibly, tax costs.
The rebalancing problem
Even if our expectations about the future have not changed, we still feel compelled to tweak our portfolios to bring them back to our “optimal” mix. This process we call rebalancing. We may rebalance periodically – quarterly for example – or we may set percentage boundaries around each asset class and rebalance when they are exceeded. Either way, the underlying assumption is that our target allocation is better than the allocation the markets have given us.
Research on the value of rebalancing suggests that it has little ability to increase returns or decrease risk. Whatever utility exists depends on factors such as time period, the direction of the market and the relative future expected returns of the asset classes being rebalanced. Yet few, if any, of us take these factors into account in developing our rebalancing strategies. Instead, we employ simple, mechanical rebalancing strategies that add little or no value and may even detract from long-term performance.
The only thing we can be sure about is that our rebalancing strategies result in transaction and tax costs.
The style-drift problem
Just to make sure everyone knows how much we love our optimal mix, we punish active managers who commit the sin of “style drift.” Style drift occurs when a manager’s portfolio does not ape the performance of the asset class that it has been assigned by us to represent.
Forget the fact that the manager was led astray by a perceived opportunity to make money. We want them to strictly adhere to their mandate. Haven’t they read the Brinson, Hood and Beebower research? Don’t they know asset allocation is far more important than securities selection?
Even though we hired them for their skill, we want them to be closet indexers.
The asset class selection problem
And how did we develop our mandates for the managers, anyway? To put it another way, how did we decide which asset classes to use in our portfolios? Was there a scientific process or did we just keep adding asset classes until it felt about right? Did we consider that the addition of each asset class brings with it additional transaction and tax costs?
Is every asset class doomed to a life sentence in our portfolios – once they are added can they ever leave? Aren’t there some supposed “asset classes” like commodities or managed futures that have no real long-term expected return? Wouldn’t it be better to hold them at certain times, but not at others? “Yes,” you say, “but it is too hard to tell when to hold and when to fold, and, anyway, they are in my portfolio as a diversifier to reduce volatility.”
Ah, volatility reduction. Is that a good thing, or a bad thing, or does it depend? Certainly, all things being equal, we’d like less volatility rather than more. No one likes big swings in their portfolio value – at least not the downward swings. But all things are never equal. We rarely have the choice of getting the same return with less risk if Mr. Market is being reasonably efficient. As a practical matter, if we shave off some volatility we also shave off some return.
Shouldn’t the composition of our portfolios be determined by the objectives of the client? Sometimes it makes sense to dial the same set of risky/not-so-risky asset classes up or down depending on the client’s goals. But why construct all of our portfolios using this cookie-cutter approach? Shouldn’t we question why each asset class is in the portfolio? Why build a conservative portfolio for someone in retirement using the exact same asset classes as a portfolio built for an aggressive investor in their 30s? In one case, controlling volatility is important. In the other, it gets in the way of fully achieving the client’s long-term goals.
The data-mining problem
Instead of doing the hard work of solving these problems, we have defaulted to using our massive computing power to mine the data in search of winning patterns in the historical tea leaves. There we seek to unlock secrets buried in the sands of time.
Some of these secrets take the form of “factors” that claim to allow us to tilt our portfolios in one direction or another to give us an edge over traditional passive strategies. These quasi-active strategies are then packaged and promoted as being “smart” by the same people who tell us that skill does not exist in the world of investing.
Part of the problem is that these strategies are backward-looking, and the future is never the same as the past. Yesterday’s smart could be tomorrow’s dumb. It often is. Then there is the “observer effect.” Just by identifying a winning “factor” we may change how it performs in the future. A winner may become a loser simply by being discovered.
But more importantly, the past often doesn’t look like we say it does. The data-miners tell us that value beats growth and small beats large. There are certainly periods in the past when this was true, but there are also many long periods when it was not.
Identifying “winners” depends on what lens you use and when you look through it. In investing, there is nothing that always wins, and we shouldn’t invest our clients’ money as though there is. That would be like determining that there are more sunny days than rainy days and then wearing shorts, a t-shirt and flip-flops every day of the year.
When we tell clients that we are “smart” because we have perma-tilted their portfolios in one direction or another they may be justifiably perplexed. Those tilts often do not pay off within timeframes that are meaningful to our clients. To a client who bought a value/small tilted portfolio at the beginning of 2014, it doesn’t look smart at all.
We need to think more about how the portfolios we build and the labels we put on them affect our clients. Clients can easily lose confidence in our ability to guide them through an uncertain future if our statements about the past don’t line up with their experiences or with reality as they perceive it. We need to spend as much time building trust as we do crunching numbers.
Conclusion
Every portfolio we manage has a client attached to it. We must examine our beliefs and practices to make sure they are consistent with the best interests of those clients.
Start by acknowledging that you know less about the future than some of your practices would suggest and that many of your tools are less precise than you would like to admit. Every transaction you make has a cost that is certain and a potential benefit that is not. The costs and benefits can be measured in both dollars and in trust. We should consider both.
Scott MacKillop is a 40-year veteran of the financial services industry. He can be reached at [email protected].
Read more articles by Scott MacKillop