The Intelligent Portfolio is written as a broad overview of the key principles of modern portfolio theory and to demonstrate how investors can use portfolio theory effectively. The author, Christopher Jones, is the Chief Investment Officer at Financial Engines. Financial Engines has been a leading force in bringing portfolio analysis and Monte Carlo Simulation [MCS] to a broad retail audience. Mr. Jones has been with Financial Engines since its early days, and has worked closely with Dr, Bill Sharpe, Nobel Laureate in Economics and a founder of the firm.
Before going into the review (and in the spirit of full disclosure), I note that my company, Quantext, also produces Monte Carlo portfolio planning tools.
Given the author of the book, the foreword by Bill Sharpe, and an endorsement by Peter Bernstein, I had high expectations of this book. Much of the content of the book is eloquent, and the writing style is clear and easy to follow. The book provides a valuable summary of the how you would go about planning a portfolio if you believe that markets are perfectly efficient—something that I do not believe—and this is the source of my primary disagreements with the book’s conclusions.
Better Tools for the Masses
The use of Monte Carlo Simulation for portfolio management is a standard of practice among institutional investors. Mr. Jones states that a primary goal of Financial Engines is to bring “institutional quality investment advice to investors of all wealth levels.” A key theme of the early chapters of the book is that technology can help to bridge the enormous knowledge gap that many investors face, having been handed responsibility for their own retirement planning. Having worked for a number of years on Monte Carlo models for large-scale applications, I agree with this argument: there is no reason that individual investors should not have the benefits of portfolio management models available to them. Oddly, over the course of the book, the author promotes a number of portfolio management themes that are opposed to some common opinions in the institutional research on portfolio management—and the source of these disagreements is due to the assumption of perfectly efficient markets.
Forward Looking Models
Some of the strongest sections of this book are those that discuss why it is important to use forward-looking models rather than simply looking at historical data for portfolio planning. The authors explain some basic statistical arguments here that really help to motivate the need for a simulation of the future (forward-looking analysis) rather than developing a portfolio based on some number of years of historical performance data. The historical record represents one realization of a series of market outcomes. In any given historical period, one asset class may dominate all the others in terms of delivering a high return over the period.
This does not mean that this asset class will continue to dominate forever. If you allocate your portfolio based on an historical period, you will end up over-emphasizing the asset classes that have dominated that period—but this has little bearing on the future. The recent bubble in Real Estate is a good example. Prior to the crash in real estate, REITs and many financial stocks had delivered very high returns over a period of years. An asset allocation based on this historical period would have suggested that an ‘optimal portfolio’ should have heavy allocation to these sectors. Forward-looking models use a variety of methods to discount recent performance (good or bad) and determine better measures of long-term expected return.
If we accept that forward-looking analysis is the right way to solve this problem, it is a natural step to then look at how the forward-looking projections generated by Financial Engines have performed. Financial Engines has been generating suggested portfolio allocations for more than ten years now (the company was founded in 1996). I really wanted to see some examples of portfolios that Financial Engines generated over this period for users and how these portfolios have subsequently performed. This is the crux of the problem. Investors need to build portfolios that are likely to generate the most return for the risk that they bear. Forward-looking models like Financial Engines are intended to provide estimates of future return and risk for a portfolio that are better than looking at trailing performance.
This is not hard to test. The decade or so since Financial Engines was founded has been a very interesting one for financial markets: we have seen substantial runs in technology stocks, value stocks, real estate, certain emerging markets, and financial stocks, not to mention commodities and precious metals. Over this same period, the average return from the S&P500 has actually been below than of bonds. These market conditions have posed enormous challenges for investors (and many pension plans for that matter). Have investors who have used Financial Engines for planning fared better than some benchmark? The book gives no evidence to demonstrate that this has been the case.
Efficient Markets and All That
There are a number of different ways to build a forward-looking model and Financial Engines starts from a strong assumption of market efficiency. There is a good discussion of the implications of efficient markets in the book. In the Financial Engines world view, there is no way to determine a price or value for an asset that is going to be consistently more accurate than the market price. In other words, the current price of an asset in the market is the best estimate of ‘fair value’ there is. If markets are this efficient then there is no such thing as a “bubble” in which prices simply depart from a rational basis. Similarly, there are no momentum effects. There is also no way to profitably buy under-valued assets. In a market as efficient as those that Financial Engines assumes, any time is as good as any other to buy into an asset.
A crucial result of the market efficiency assumptions in Financial Engines’ world view is that markets are so efficient that there is very limited value in looking for assets for your portfolio that are really good diversifiers. This can lead to some fairly paradoxical results. Commodities end up looking utterly pointless as investments in this world view. Emerging markets end up with a lower expected return than domestic stocks. Both of these points are made explicitly in the book.
The core of the argument is this: an asset class with low correlation to the overall market is a powerful diversifier because it can increase portfolio return without increasing risk. In a totally efficient market, investors will bid up the price of such an asset class so that the potential future earnings from that asset class are lower as a standalone (i.e. the value of diversification effects has been ‘priced in’). This is why emerging market stocks end up with lower returns than domestic stocks and commodities have zero real return in Financial Engines projections. Commodities have a very low correlation to broad stock indices, thus they must have a very low expected return going forward if markets are to be perfectly efficient. If you have not previously encountered Modern Portfolio Theory (MPT), this may all sound paradoxical---and you would be in good company.
If markets were totally efficient, rational investors would not bother looking for ‘better investments.’ If rational investors don’t compete in the marketplace, the markets would not be efficient. This is called the Grossman-Stiglitz paradox (Stiglitz shared the 2001 Nobel Prize in Economics, in part for the work on this issue).
I do not want to beat a dead horse here, but this is really important. The heart of the Financial Engines world view (explained in this book and expressed in their model) is that markets are perfectly efficient. There is, however, abundant evidence that this is not the case.
Another implication of the strong version of efficient markets that forms the core of Financial Engines is that it simply does not make much sense to have significant allocations to asset classes other than broad market-cap weighted indexes. Mr. Jones argues that REITs, commodities, and even TIPS (inflation indexed bonds) should be limited to small allocations—if any allocation at all. If markets are perfectly efficient, these asset classes should have very low expected returns because they would confer substantial portfolio benefits otherwise.
While it is not actually noted in the discussion, the efficient market assumption of Financial Engines also means that investing should be done on the basis of market capitalization and that any other approach to weighting assets is sub-optimal. Rob Arnott’s research on fundamental indices, for example, is not consistent with this world view.
We have now reached one of those places where the Financial Engines world view is, in fact, very different from that of many institutional investors. David Swensen, the Chief Investment Officer of Yale’s endowment, and one of the most successful institutional managers on record, invests heavily beyond the core asset classes of global equities and bonds—and largely because of their powerful diversification effects. He is betting that the markets are not perfectly efficient in this regard. Many of the largest institutional managers in the world believe that commodities play an important role in portfolio management.
The strong view of efficient markets in Financial Engines has enormous implications that are reflected in the investing logic presented in this book. I feel that this world view is too restrictive. Several years ago, for example, I looked at adding some exposure in my portfolio to electrical utilities. I found that the very low correlation of these stocks to the broader market (the S&P500) provided my portfolio with strong diversification benefits. Further, utilities were somewhat depressed in terms of their price-to-earnings. I added a concentrated position to utilities to my portfolio. If I were a believer in the perfect efficient markets hypothesized in MPT, I would have decided that these attractive portfolio qualities must already be priced in, so why bother? The problem with this strong view of efficient markets is that they would be inherently paradoxical.
Arbitrary Rejection of Individual Stocks
One of my least favorite sections of this book is the chapter that attempts to convince investors that they should buy funds and avoid individual stocks. The author actually tries to bolster his argument by showing that individual stocks are more risky, in general, than a diverse index. There are tables that compare the volatility in individual risky stocks (like Tivo) to the S&P500 and use this as evidence that owning stocks rather than funds is too risky. Seriously.
The author does, ultimately, concede that combining individual stocks in a portfolio provides diversification benefits and reduces risk, but then goes to considerable lengths to generate a fairly bizarre outcome. In the final case, the author creates a single model portfolio of ten stocks, arbitrarily selected and equally weighted. How were these stocks chosen? There is no discussion of why anyone would actually build this portfolio. When I analyzed this portfolio, I found that it had 40%-50% more risk than the S&P500, depending on whether you looked backwards or forwards. This was not a well-designed portfolio.
The author then shows that this portfolio of individual stocks is really risky as compared to the S&P500, and uses this result to bolster his argument that owning individual stocks is “too risky.” He never notes that it is possible to create a portfolio of ten stocks that would have equal or less risk than the S&P500. He never discusses the possibility that one might choose allocations in a more thoughtful way or that his results is very sensitive to exactly what he puts in the portfolio or how he weights each holdings. This is far from a compelling case against owning a portfolio that contains individual stocks. In fact, it reads more like a contrived example to show what the author wanted to show. There is no attempt here to generate more than this single anecdotal example, but the author attempts to motivate over-arching conclusions.
The Policy Portfolio
An additional rather odd outcome occurred to me as I read this book. As I noted at the start of this article, Peter Bernstein gave the book a nice review on the book jacket. Mr. Bernstein is famous for his outspoken criticism of the idea that institutional investors should buy and stick to a standard “policy portfolio,” which is basically a static asset allocation plan. The strong version of efficient markets upon which Financial Engines is based supports the idea of the policy portfolio. To read Mr. Bernstein’s thoughts on this issue, see this article.
In short, Mr. Bernstein is an out-spoken critic of simply picking an asset allocation and sticking to it for all time.
I have discussed my own position on this issue in this article [pdf file], and how it relates to Mr. Bernstein’s thoughts (among others).
The idea that investors should stick with a simple asset allocation for all time, and add to it regardless of prevailing market conditions and prices, is too simplistic. This decade of anemic returns on a domestic market-cap weighted index is a case in point. There are many domestic sectors that have performed well in the recent decade. Investing in stocks following market-cap weighting has simply not been a successful strategy, and there is, in my opinion, less than compelling evidence for arguing that market-cap-weighted portfolios make sense.
Treatment of Post Retirement
In using Monte Carlo Simulation to calculate the probability that an investor will be able to successfully fund his or her retirement, Financial Engines makes an assumption that deserves further discussion—more than it gets in the book. The Financial Engines approach assumes that retirees will take all of their funds at retirement and purchase an annuity to fund their retirement. The book is not specifically endorsing this approach, but the very fact of its assumption has important implications for the types of portfolios that are endorsed by the methodology.
This issue has come up in the design of target date funds. If you assume that an investor sells his/her portfolio at retirement and buys an annuity, the optimal asset allocation choices will tend to be considerably more conservative (more fixed-income / less equities) than if you assume that an investor draws income from his/her investment portfolio during retirement.
Given the high expenses of annuities, the built-in assumption that all investors will annuitize at retirement age is not ideal. For investors who do not intend to annuitize, the portfolios endorsed by the Financial Engines methodology are not ideal, even within the basic capital market assumptions of the system.
This book is an interesting read and well worth owning, despite having a clear tilt towards promoting the conceptual model of perfectly efficient markets. This book would be far stronger if it acknowledged the limits and counterpoints to its arguments. If you read A Random Walk Down Wall Street (by Burton Malkiel), for example, you will find a far more nuanced and balanced discussion of efficient markets. Malkiel, for example, discusses the evidence in favor of buying stocks at low price-to-earnings ratios. In the perfectly efficient markets discussed by Mr. Jones, and embodied in Financial Engines, P/E ratios don’t matter, for example.
If you are willing to accept the underlying assumptions about the perfection of markets, Financial Engines provides a valuable tool for financial planning. For the large numbers of investors who make rookie mistakes in financial planning like chasing hot asset classes or putting the majority of their holdings into one or two stocks, Financial Engines will be valuable. For both of these classes of investors, simply putting money into a well-designed ‘pie chart’ asset allocation comprised of index funds or ETF’s would, however, probably be at least as good a choice as trying to use Financial Engines---you don’t need all the technology and theory to accomplish this either. In fact, I believe that a well-thought out static asset allocation from any number of sources may be better choices—and certainly have historically been better choices—because they are not fettered by the assumptions of perfectly efficient markets.
One area in which Financial Engines could be very useful is in determining the right risk level to meet your targets, but users will need to remember that the selection of the best risk level determined by Financial Engines is partly determined by the assumption that investors will sell their portfolios and purchase an annuity at the retirement age.
In closing, I want to say a few words about financial models. I developed Quantext Portfolio Planner [QPP], which is a Monte Carlo model that makes very different assumptions than Financial Engines about how the world works. There are limitations and issues with QPP’s underlying models just as there are with Financial Engines. If Bill Sharpe or Christopher Jones reviewed my writings about portfolio analysis using QPP, I am sure that they would also find plenty to disagree with. The choice between financial models (and their underlying assumptions about financial markets) will impact how a portfolio designed using a model performs.
To choose between models and the world views that drive these models, it is a good idea to look at the degree to which the output from these models makes sense and is supported by other expert resources and the evidence from applying these models to real markets. To restrict oneself to investing the types of portfolios that result from the Financial Engines approach means that you must fundamentally disagree with the perspectives of many of the most successful long-term investors of our time (i.e. Warren Buffett, Charlie Munger, Bill Gross, David Swensen), as well as with the viewpoints of many of the most credible institutional advisors (i.e. Roger Ibbotson, Peter Bernstein).
Christopher Jones’ response to this is already in the book: He believes that the rare person who manages to beat the market over the long haul is the fortunate recipient of good luck: if you had an infinite number of monkeys throwing darts at The Wall Street Journal once a year, and rebalanced your portfolio each year on the basis of the stocks that get hit by darts, there is some possibility that a number of monkeys would beat the market as substantially as Warren Buffett and Charlie Munger. Personally, I will continue to listen to Buffett and Munger.