Could smart beta be the Holy Grail of investing? Much has been written about this latest index innovation, which allows investors to pinpoint exposure to a particular market or select factors. These alternatives purport to offer a smarter way to gain exposure to the drivers of investment returns and an improvement over the commonplace strategy of putting money in index funds. Although smart beta investing has its critics, the strategy is gaining in popularity among institutional investors. Research firm Morningstar reported in May that assets in these strategies increased by 59 percent in 2013 and now account for nearly $300 billion. A January 2014 survey by asset manager Russell Investments found that 32 percent of institutional investors had a smart beta allocation, although it was less than 10 percent of the average investor’s equity portfolio. In another January report Cogent Research, a division of Market Strategies International, said that nearly half of the institutional investors it surveyed expected to invest in smart beta strategies within the next three years.
Smart beta — also known as factor investing, a label that better captures the rationale behind the strategy — promises indexlike management in terms of costs and transparency while affording exposure to more than just the static market factor. The claimed benefit is that by using a set of rules to manage exposures to one or more dynamic factors (things like value, momentum and volatility), investors can achieve a better risk-return profile, essentially outsmarting the static beta that is captured in a traditional index fund.
But I say not so fast. If factor exposures drive portfolio returns, is the time-honored asset-class-driven allocation misguided? Wouldn’t a smart beta approach demand agreement on which factors to include and exclude? And what about unintended tilts? Would factor exposure in a smart beta portfolio be constant? From a theoretical standpoint it’s hard not to be a fan of factor-based investing — until one realizes the complexity surrounding this type of strategy.
Traditionally, investment returns have fallen into one of two categories: alpha or beta. Ever-elusive alpha is the risk-adjusted return above the return of the overall market; it represents manager skill. Beta is the return of the broad market, typically represented by a benchmark like the S&P 500 index. Smart beta — a hybrid product crafted by financial engineers — is meant to deliver the best of both worlds: market-beating returns at indexlike prices.
Despite their great promise, smart beta strategies have their limitations. To begin, it’s much harder to allocate assets along factor lines than by asset class, especially without the use of short positions (a constraint for many investors). If investors are going to move from asset-class allocations to factor allocations, they will need access to sophisticated risk models, which could be a challenge for smaller institutions. In addition, the rules-based approach used by smart beta strategies makes them vulnerable to front running. Still, smart beta is likely to offer a net benefit to investors, if for no other reason than that it will make them smarter about the factors that drive returns.
The idea of building an index to capture the systematic component associated with a particular type of investment is hardly new. The first factor-based index arguably was the Dow Jones Industrial Average, which was launched in 1896. Investors who observed its performance vis-à-vis their own portfolios would have been struck by the degree to which their portfolios’ returns were correlated with the returns of the index. In modern vernacular the portfolio returns could be deconstructed into two pieces: the first a function of the return of the market as measured by the DJIA and the second a function of the particular stocks they owned. If the portfolio mirrored the constituents of the DJIA, the majority of the return and the volatility of the portfolio could be attributed to the same systematic sources that drove the returns of the DJIA.
There is now substantially more understanding about the drivers of returns in various markets, including equities, fixed income and currencies. Obviously, decomposing the return of an equity portfolio requires knowledge of considerably more than just its exposure to the overall market. Within the equity markets, for example, the initial realization that small-cap stocks have a different return distribution from large-cap ones compelled investors to evaluate the small-cap exposure of their portfolios differently — and, more important, to make this decision part of their risk management process. This led to the now-common practice of viewing a portfolio’s return in the context of the market returns as well as the style of the underlying portfolio.
As we have gained a greater understanding of the different types of systematic factors that affect returns, we are increasingly moving to a world in which the portfolio returns are attributed to multiple sources. In the case of equity portfolios, as illustrated in the chart below, investors are increasingly able to single out the portion of a portfolio’s return that is attributable to systematic factors (beta) from that which is idiosyncratic to the portfolio (alpha). This latter component — which is, one hopes, positive — represents the true value of active management.
This realization of the magnitude of returns stemming from systematic factors has led to the desire to harvest these sources of return in a cost-effective manner. This approach is based on a sound theoretical foundation — namely, that individual securities are expected to earn a return through their exposure to systematic factors. These factors are rewarded, on average, but they will perform poorly in bad times and more than compensate for this poor performance in good times. The goal of a portfolio manager amounts to controlling these systematic factor exposures and diversifying away those risks that are not systematically rewarded. The management of the systematic risks in a portfolio can be active — requiring forecasts of the different factors’ returns — or passive, in which the exposure of the portfolio to the various factors is kept static.
The recognition that some or all of an active manager’s returns may emanate from systematic factors is undoubtedly one of the reasons for the current enthusiasm for smart beta. Why bother hiring an active manager if investors can simply allocate some of their investments directly to these systematic sources of return? (See also, “Will Smart Betas Make Hedge Fund Managers Obsolete?”) In addition, if returns are driven by exposure to certain factors, shouldn’t investors understand and manage their overall allocations to these same risk factors? Moreover, if security returns are driven by more than a single market factor, wouldn’t it make good financial sense to diversify one’s portfolio across these different types of systematic risk? In a nutshell, these arguments serve as the rationale for the host of smart beta exchange-traded funds and portfolio solutions that are being offered today. Although the discussion around smart beta tends to be equity focused, the theory is applicable to other asset classes, including currencies, fixed income and commodities.
The idea that factor exposures drive a portfolio’s returns more than asset classes leads one to the realization that focusing on asset allocation is misguided. Instead, investors should concentrate on how they allocate to the key factors that drive asset returns. This requires a dramatic recasting of our view of the world — and obviously some agreement on those key factors, or betas. Ideally, the key factors should be uncorrelated with one another; each should capture a unique type of systematic risk or investor behavior.
Risk factors can be static or dynamic. A static factor is one that does not require much active management, such as exposure to the overall equity or fixed-income market. The returns for most asset classes can be thought of as static risk factors. A static diversified capitalization-weighted portfolio, even absent rebalancing, will do a reasonable job of capturing the overall return of an asset class.
A dynamic factor would require at least some modicum of active management — and consequently some associated transaction costs. One example involves investing in cheaper-than-average securities. In equity markets this value trade could be accomplished by investing in securities with lower-than-average price-earnings ratios and shorting ones with higher-than-average P/Es. In the currency market the carry trade, as it is known, could be achieved by borrowing in low-interest-rate currencies and investing in high-interest-rate ones. Similarly, in fixed income an investor could hold long positions in higher-interest-rate parts of the yield curve and short the lower-interest-rate parts — this is sometimes referred to as riding the yield curve. Regardless of the asset class, maintaining exposure to this type of factor requires that the portfolio be continually rebalanced. Dynamic factors often go beyond asset classes and involve active management to maintain exposure to the factor. The process can involve human judgment or a set of rules. Call it what you may, capturing a dynamic factor involves a set of decisions unlike those required for investing in a static factor.
Other examples of dynamic factors that go beyond asset classes include volatility: the tendency for low-volatility securities to outperform high-volatility ones. This effect has been noted in equities, fixed income and currencies. The momentum, or trend, factor — which captures the propensity of securities that have strong relative performance over the past six to 12 months to continue to outperform — also works across equities, fixed income, currencies and commodities.
When it comes to the static factors, there is some agreement that definitions are based on the broad asset classes. There is much less consensus concerning the precise number and construction of the dynamic factors. For example, though most market pundits would agree on value as a dynamic factor, there is no agreement, even among equity analysts, on whether it should be defined based on P/E, price-to-book or dividend yield. Some providers of risk models — which are designed to use these systematic factors to describe the returns of an equity portfolio — often include all three of them as factors, in addition to several others.
The first major hurdle to moving from asset classes to a factor-based approach is agreeing on the number and definition of the factors that are to be used in the allocation process. Once an agreement has been reached, the standard asset allocation tools — mean-variance optimization, scenario analysis, risk allocation — can be implemented to determine the desired factor allocation. One of the difficulties in using a risk factor allocation stems from the fact that most investors face a leverage constraint. In a typical asset allocation exercise, investors avoid using leverage by limiting the total asset exposure to 100 percent of capital; when it comes to risk factor allocation, in which leverage restrictions vary across asset classes, the process is more complex.
Investing in cap-weighted indexes is commonplace for both institutional and individual investors. Cap-weighted indexes that are used both as benchmarks and as measures of an investor’s opportunity cost have several desirable properties not present in a typical smart beta index. First, by virtue of the fact that the index contains every stock in proportion to its market capitalization, if everyone invested in this index, it would be true that the sum of all the holdings would equal the value of the index. Large-cap stocks would have a larger weight in all investor portfolios, and small-cap stocks would have smaller weights. This is not true for most smart beta indexes, however, as the portfolio weightings are driven by factors other than market capitalization. Even if every investor wanted to invest in the same smart beta allocation, they would be unable to do so, as weights in the smart beta portfolio are not proportional to the relative size of each company. In other words, for one investor to have a holding in a stock that is greater than its weight based on market capitalization, another investor has to be willing to have a holding in that same stock that is less than its cap-based weight.
What this means in static terms is that we cannot all hold the same smart beta portfolio. In fact, for even one investor to hold a smart beta portfolio, another investor has to have an anti–smart beta (which sounds more polite than “dumb beta”) tilt of the same magnitude. Although this is a truism, one should ask whether every investor should have the same smart beta allocation. We have long accepted the idea that asset allocation differs among investors: Some hold more fixed-income securities; others hold more equity. Few, if any, hold fixed income and equity in exact proportion to their market weights.
The market portfolio is held by the average investor. If you are different from the average, there is no reason you should hold the same portfolio as the average investor. Consider an investor that has more risk-bearing capacity than others. Such an investor can afford greater exposure to those investments that have poor returns in bad times. By contrast, an investor that has a below-average ability to bear risk in poor times should have less than average exposure to such an investment. This does not require one investor to be smart and the other dumb, but it does require investors to be cognizant of their relative strengths and weaknesses from an investment perspective.
Exposure to the small-cap factor, for example, is generally considered desirable from a return standpoint. This factor, however, tends to have negative returns in bad times and positive returns in good times. For an investor with a long time horizon, a tilt toward this factor probably makes sense. For an investor with a shorter horizon and less ability to bear risk in bad times, higher-than-average exposure to this factor is probably a bad idea. In this example the relationship between the factor returns and the economic environment is straightforward. For other factors, such as momentum, the relationship can be harder to identify. Regardless, the key message here is that there is no universal factor allocation. Rather, each investor must identify an appropriate factor allocation and then select the appropriate smart beta indexes to replicate it.
The returns for select equity factors in different economic environments are shown in the table below. The market factor is static, represented by the cap-weighted market return. The other factors are dynamic and represent different potential sources of return. It is evident that no single factor provides positive returns in each economic environment — illustrating the value of managing individual factor exposures. In fact, the only security that has such a property would be the quintessential risk-free security. The small-cap factor, for example, has the same return pattern as the market, experiencing its lowest returns in the early part of the recession. The low-volatility factor, represented by the differential in performance between portfolios of low-beta and high-beta stocks, delivers its highest return in the early part of the recession; combining it with the small-cap and overall market factor will create a portfolio with a better return profile than just investing in the static market factor. Adding the momentum and value exposures will further improve the return profile, although the allocation needs to account for the fact that value, momentum and low volatility all have negative returns in the late stage of a recession. In a factor allocation world, the challenge is to build a portfolio that has a diversified exposure to each of the key factors so as to benefit from the lack of perfect correlation among them.
The preceding example considers four distinct economic regimes. The world is considerably more complex, so prudent investors would also take into account the performance of key factors in different regimes of volatility, investor sentiment, liquidity, sunspot activity, hemlines and whatever other measures they believe to have a material impact on asset markets. Like it or not, this involves making assumptions about the relationship among the factors and the future return associated with each — not dissimilar to the job of the traditional active investment manager. The difference between traditional active management and a smart beta strategy is that with the latter the rules on how the exposure is managed are specified in advance.
Taking an investor’s circumstances into account relative to factor returns is somewhat different from what is sometimes referred to as liability-driven investing. In a typical LDI approach, the goal is to match the interest rate and inflation sensitivity of the liabilities of an investor. While an investor’s ability to bear certain types of interest rate and inflation risk is an important determinant of factor allocation, there are many other elements to consider, including relative size, investment horizon, liquidity demands and the ability to bear losses in difficult economic environments. Each plan will differ on such dimensions, with resulting implications for the optimal factor allocation. For example, although most pension plans have long horizons, very large ones will naturally find it challenging to gain meaningful exposure to certain factors, such as small capitalization. By contrast, smaller plans may be able to gain as much exposure as they would like to this factor.
Assuming that an investor can determine the optimal allocation to risk factors that makes allowances for its unique circumstances, the next step in the process would be to construct a portfolio that provides exposure to the desired risk factors. In an ideal world, in which there was agreement on the factors and low cost exposure to each of the relevant factors was available, this would be a relatively straightforward task — akin to going to a pharmacy with a prescription. Unfortunately, implementing a desired factor allocation is not as easy as allocating to a few index funds or selecting the best managers in each asset class.
The importance of asset allocation in determining both the risk and the return of a portfolio is well understood by investors. Asset classes are tangibly different; for example, stocks differ from Treasury securities and high-yield bonds. Investment strategies have for years been designed around asset classes, so implementing an asset allocation is much easier than implementing a factor allocation. This is especially true when disagreement exists in determining whether a particular stock has exposure to a certain factor. To illustrate by way of example, consider that the Russell value and growth indexes share approximately 335 stocks; classifying individual stocks that have exposure to something as dynamic as the value factor is not black and white.
Putting the definitional issue aside, there are two additional challenges in implementing a smart beta approach. First, although capturing most dynamic factors requires both long and short positions to efficiently gain exposure, most smart beta products are long only. The absence of short positions dramatically reduces the ability to capture the true factor returns. Based on the fundamental law of active management, one would estimate that the reduction in effectiveness would be as large as 50 percent. This can result in a portfolio that fails to capture the desired factor return and leads to inevitable investor disappointment.
Second, in constructing the risk factor portfolios, attention must be paid to unintended tilts that may creep into the portfolio as a consequence of the portfolio construction process itself. For example, a smart beta equity portfolio attempting to have a value bias also may have a small-cap and low-volatility bias. For the investor seeking to implement a factor allocation, it would be a lot easier if the allocation had constant exposure to certain factors and no exposure to others. If a portfolio has exposure to multiple factors, it should be by design and specified in advance.
For their part, large institutional investors are likely to be overwhelmed by the number of choices from managers or other providers that are eager to offer customized smart beta solutions. Creating a portfolio with the desired factor allocation is not a hugely difficult task for a manager with knowledge of quantitative portfolio management techniques and the ability to manage a short portfolio. Though management of turnover and transactions is critical in the implementation of a factor-based allocation approach, this is well-trodden ground and should not pose a major impediment to implementing such an investment strategy.
On the other hand, a less privileged investor looking to implement a factor portfolio using off-the-shelf products faces a much more difficult task. Most portfolios that fall into the smart beta camp contain long-only positions in the underlying securities. With few exceptions, investors can buy only a prepackaged version of a static beta (the market return) and a dynamic beta (such as the value premium). If investors want additional exposure to the low-volatility factor, for example, they would need to make an allocation to a product that would combine market exposure along with the low-volatility factor exposure. In other words, these investors can only purchase a factor if they are willing to assume the market exposure that comes with the factor exposure. Contrast this with the ideal world, in which investors are given access to a market portfolio, a market-neutral value fund and a market-neutral low-volatility fund — much like the portfolios used in the accompanying table with factor returns. The availability of the funds with long and short positions would increase the investors’ investment opportunity sets, which in turn could produce superior performance. Recall that the theoretical justification for these strategies is that one is able to manage the exposure to different factors. Without being able to purchase each one in isolation, it is difficult, if not impossible, to get close to one’s ideal factor allocation.
In some product offerings the designers have realized the issue associated with long-only attempts to capture a single smart beta and have tried to overcome this by combining more than one smart beta into one index. Although this is a somewhat better solution, the fact that factor components are provided in prespecified proportions is a major hindrance to an investor looking to make a factor allocation. Most smart beta products are sold on the basis of their simulated track records, and though the simulation in itself is not necessarily a negative, all too often there is little description of the precise factor exposures in the portfolio. If investors are going to move from asset-class allocations to factor allocations, they need to be able to easily access the factor exposure of the funds in which they invest. This is a relatively easy task for a larger investor with access to a risk model and portfolio holdings, but it would be a challenging task for the smaller institutional or individual investor.
For active money managers there are two major implications of smart beta development. First, if the manager’s returns are a result of a combination of passive factor tilts, there is little justification for charging a typical active fee for generating the return. Second, even if the manager’s return is not purely attributable to a combination of factor tilts, a smart beta strategy or a combination of smart beta strategies may be a more attractive alternative for an investor.
The first implication is hardly new. Investors have long benchmarked their managers against style-adjusted benchmarks. Often, however, these benchmarks have been based on a static tilt to one factor and as such may not have captured the true factor return associated with a particular strategy. The trend in the late 1980s and early ’90s to create normal portfolios for managers was, in part, a recognition that static one-factor indexes did not capture the full complexity of a particular strategy. This normal-portfolio approach failed to achieve mainstream acceptance partly because of its increased complexity. Even with the current crop of smart beta indexes, it is difficult to measure and evaluate a manager’s factor footprint.
One could sidestep the performance issue by viewing the manager’s single- or multifactor smart beta index as an opportunity cost. In such a framework the returns from a smart beta index reflect a return-and-risk opportunity that is achievable with the investor’s capital, and the manager can be measured relative to this index. In other words, a combination of smart betas would become the alternative to the market benchmark as a way of measuring manager performance.
The first smart beta funds — namely, index funds — proved a boon to investors. They offered low cost and a very high level of predictability. The predictability stems directly from the fact that these funds replicated static factors. There is very little active management involved in re-creating the performance of the overall market. When one is attempting to replicate a dynamic factor, however, for the various reasons discussed earlier, the world is not so static.
The differential performance of funds purporting to capture the value factor in the equity market is a good example of the confusion that besets investors. The lack of transparency regarding which factors are managed in a particular portfolio and which are ignored is also a challenge. Until we reach the point that the disclosure for factor exposure in a portfolio mirrors that on a nutrition label, it will be difficult for investors to make informed decisions and avoid disappointment.
The rules-based approach used by some of these strategies makes them susceptible to front running by informed investors. Much like the effort around gaming additions and subtractions to a static factor fund, like the S&P 500, the more-successful smart beta funds will be the focus of investors who try to take advantage of the particular rebalance rule used by the fund. If there are multiple funds attempting to capture the same dynamic factor exposure — which is highly likely — there is also the possibility that the rebalancing will have a large systematic impact on the returns for a particular factor. The survivors of the quant crisis of 2007 remember well the potential impact of large flows in and out of a particular factor.
From a theoretical standpoint it’s hard not to be a fan of factor-based investing — until one realizes the complexity surrounding this type of strategy. Because of the complexity associated with defining the dynamic factors and monitoring exposures, it’s difficult to see factor-based investing ever completely surpassing the more commonly used asset-class approach. The lack of agreement on factors makes it virtually impossible to achieve the factor-based equivalent of a 60-40 balanced equity–fixed income portfolio.
Lack of wholesale adoption, though, does not necessarily bode ill for all factor-based strategies. Such strategies are an attractive alternative to static factor funds such as traditional index funds. They also raise the bar for active managers in that they will no longer be able to claim skill as a result of a passive tilt to one of the dynamic factors. Going forward, the challenge for active managers will be not only identifying the factor tilts in their portfolios but also demonstrating the value that stems from the tilts. The biggest contribution of smart beta will likely not be the products themselves but their role in making investors smarter about the various factors that drive returns and improving the level of diversification in their portfolios. • •
Harindra de Silva is president and a portfolio manager at Analytic Investors in Los Angeles. He oversees the research and development of the firm’s investment processes.
