It is quarterly reporting season, so our senior consultants have been meeting with clients, delivering results, educating them about fiduciary concerns, and talking about the market environment. In the course of these conversations, we have had ongoing discussions about the role of diversification in client portfolios. Some of our previous articles and blog posts – notably, our most recent article about the apparent failure of diversification – have tried to explain the long term value of diversification, but these short articles have led to follow up questions.
Let’s discuss a recent challenge to diversification. Imagine an investor who learned about investing theory in the 1970’s – let’s call him Mike Brady. Mike buys American. More specifically, Mike invests in U.S. large cap stocks and U.S. bonds exclusively, in a 70% stock /30% bonds blend. In 2014, Mike’s performance was about as good as a blended portfolio could be. Moreover, U.S. securities have dominated most international and alternative asset investments since the financial crisis all the way back in 2008—that’s a good 6 years of U.S. portfolio domination. Mike holds up his recent returns as proof there is no need to diversify investments outside the U.S.
Let’s look at the data. Let’s compare Mike’s 70%/30% U.S.-only mix to a 70%/30% diversified portfolio which includes global bonds, emerging & international developed markets, commodities, real estate, and hedge funds. We will look at performance since a 1999 incept date (the earliest concurrent point of all the underlying investment indices) to the end of 2014. We will indentify the winners – both terms of absolute return and in terms of risk-adjusted returns (Sharpe Ratios & Alpha) for a short term (3 year) and longer term (15 year) time horizon. Winners are marked in green; losers are marked in red.
We know the U.S. only portfolio is going to look great against anything since the 2008 financial crisis. Beyond those six years, it will take several years of international and alternative outperformance to make up the difference. As it happens, by the 10 year mark, the diversified portfolio had caught up and was looking roughly comparable. By longer time measures, the diversified approach is superior both in terms of return and risk adjusted returns. Since the earliest inception points, the diversified approach is better.
So, we know a diversified approach worked in the long-term, but a U.S.-only approach worked best in the short-term. So, some investors have asked us an interesting follow up question: why we can’t we simply use a short-term portfolio that’s designed to do better in the shorter time horizons? They wonder if there are differences between an optimized short term portfolio and a long term optimized portfolio.
It is true some investment shops and brokerage firms have put together capital market assumptions (i.e. target return ranges) for the very long term (say, 50 years) and shorter term (15 years) ranges for the use of tactical reallocation. They really can’t go much lower than that.
Let’s consider two potential flaws with the notion of a short-term optimization:
The first flaw: there may be no material difference between long-term return models and short-term return projections – the primary difference between these projections is the confidence returns will fall within an expected range. For example, over a 30 year time horizon, a financial model might determine, with a 95% confidence level, that U.S. stocks will generate annualized returns between 6% and 10%. In a smaller 3 year time horizon, the model might only be 30% confident of the returns falling within the same range, even though the average return of both projections is the same. Getting the years of exceptionally good and catastrophically bad returns to average out takes time. As an extreme version of this principle, understand the odds of the long-term average return equaling an actual annual return is pretty low for any given year. For example, the S&P 500 has returned 10.44% in the past 25 years (1989 to 2014). However, you have to go back 10 years (all the way to 2004) to get an annual return with 2% points of that long-term average. The interceding years have been quite volatile, ranging from -37% in 2008 to +32% in 2013.
To clarify, institutions often have short-term goals that can be met. Asset allocation studies are driven by institutional goals and tolerances. For instance, if a corporate cash account truly cannot afford to lose any principal within a 3 year investment window, their asset selections are going to be limited to very high quality fixed income. However, there is no such thing as a short term asset allocation optimization because nobody can confidently know how assets are going to outperform in the short term. For institutions with long-term return goals, the basis for prudent investment remains a long-term asset allocation study.
The second flaw: there is an incorrect implication in this question; the implication is short term projections are more likely to be like the past. In this case, the unstated assumption is this: “in the short term, U.S. portfolios do better than diversified portfolios.” It simply isn’t true. By mathematical definition, we can count the number of rolling 3 year periods in which the diversified portfolio outperforms the U.S.-only portfolio; it simply hasn’t happened lately.
In our recent blog post on emotional investing, we point out the recent top-performer – the hot-dot – may actually be less likely to outperform in the future. We understand the frustration with having a diversified portfolio. We understand the regret of not having owned the perfect portfolio that would have worked best in the previous year. However, changing your allocation to align with the previous top-performers is a poor solution which has proven in study-after-study to hurt long term performance. Investors can get whipsawed by investments which quickly appreciate and then just as quickly reorient to their long term, defensible, average rates of return. A model based on very short-term time horizon is either counting on clairvoyance (i.e. knowing which asset classes are going to outperform) on the part of the capital market assumptions or it’s based on the fallacy of assuming yesterday’s outperformers are going to be tomorrow’s outperformers.
We understand—6 years is a long time to have to defend a diversified strategy. If you have quarterly meetings along the way, as many institutions do, that’s 24 meetings. However, a diversified approach means you will have at least some underperforming asset class in every meeting. So, you will always have to defend holding underperforming asset classes within the lineup while other assets look more compelling. Set your expectations according to this principle because, in the long term, there is never any way to know when the winds will change and today’s underperformers again become the top stars.
We look forward to talking with you about this.