Deconstructing Smart Beta

February 10, 2015

Traditional approaches to asset allocation are typically static and diversified across asset classes. Arguably, they are constructed with little regard to types of underlying risks. Against this backdrop, there is an emerging trend, especially among institutional investors, towards more dynamic asset allocation that hinges on diversification across risk factors.
As most investment portfolios are still constructed on the basis of direct asset class exposure, it may not be feasible for investors to apply a factor-based framework to implement policy-level decisions. Practical solutions are needed to enable investors to incorporate risk factors in the portfolio construction process while accommodating existing investment processes.
Exactly how risk factors should be included is still a nascent and fiercely debated area of research. Here we look at the practical aspects of implementation, using stylized case studies. First, we examine how returns may be enhanced (or risk reduced) by adopting alternate beta strategies designed to capture both beta exposure from individual asset classes and systematic factors (such as value). Secondly, we assess the feasibility of using risk premia portfolios, which take long-short positions to target systematic factors.

Alternate Beta as Portfolio Building Blocks
Historically, market capitalisation index strategies were a means to capture market beta, while active managers were used to generate alpha. More recently this boundary has blurred. Investors now look at a continuum of options, from traditional market capitalisation weighted strategies at one end of the spectrum, to actively managed strategies at the other.
A significant portion of the alpha delivered by active managers may be attributed to a handful of systematic risk factors]. These risk factors can be implemented within a passive mandate known as “alternate beta” or “smart beta”. Alternate beta is increasingly imposing itself as a credible choice that avoids the inefficiencies of market capitalisation weighted indexes and the higher cost (and at times lesser performance) of active managers (Exhibit 1).
Major asset classes, including equities, fixed income and commodities have seen a surge in interest in alternate beta strategies. Alternate beta typically captures systematic risks. In equities these might include small capitalisation, value, low volatility, momentum and quality. In commodities, these could include curve, value and momentum.
The development of fixed income alternate beta strategies is in its infancy. So far it has centered on fundamental-based indexes that overweight sovereign issuers with better fiscal strength and corporate issuers with lower credit risk. This is by contrast to traditional capitalisation weighted bond indexes, which accord the highest weights to the most indebted issuers.
Most alternate beta strategies aim to achieve enhanced return, reduced risk, or both. In general, investors select their strategy according to
their investment objectives.

PracticalConsiderationsFactor_fig1

Case Study
To demonstrate the potential benefits of using alternate beta strategies as building blocks in portfolios, we constructed a hypothetical portfolio with a 40 percent allocation in low volatility equities with the aim of reducing risk, and 60 percent equally spread across small cap, value, momentum, and quality indexes in order to enhance return. Similarly, we created an alternate commodity beta portfolio with 40 percent weight in the S&P GSCI Risk Weight index, which is an index based on equal risk contribution from five commodity sectors, and 60 percent weight equally spread across curve, value and momentum. All the building blocks here are represented by long-only equity and commodity indexes.
Implementation Issues to Consider
Overall, the development of alternate beta expands the repertoire of possibilities for a portfolio. This is especially true in North America and Europe, where the diversity of investment styles and strategies can be accessed through ETFs, funds, swaps and segregated mandates in a transparent and inexpensive way.
For institutional investors, the process of investing in alternate beta requires setting investment objectives, establishing target factors, selecting index strategies and managers to carry out the implementation, and finally constructing the portfolio while measuring and monitoring performance on an ongoing basis (Exhibit 4). This involves certain challenges.
PracticalConsiderationsFactor_fig2

Investment objective and beliefs:
The adoption of alternate beta often reflects the investment philosophy of an organisation, for example the Yale Model versus the Norwegian Model. The Yale Model is an endowment model of investing, particularly well known for its substantial allocations to alternative asset classes like private equity, real estate and hedge funds. It is based on the assumption that active management is effective in generating returns in these less-efficient asset classes. By contrast, the Norwegian Model is premised on the idea that long-term risk premia can be harvested to achieve long-term returns. Investors who subscribe to this model are often fervent adopters of alternate beta strategies.
Other investors may adopt alternate beta strategies because of their investment objectives and constraints; for example, to better utilise their risk budget, achieve risk diversification, or reduce overall portfolio risk. Meanwhile, investors with a greater risk appetite may wish to employ these strategies as a means to generate higher returns.
However, the adoption of alternate beta is often a strategic rather than a tactical decision and, as a result, requires a high level of commitment. Furthermore, since investing in alternate beta strategies involves taking active investment decisions via passive implementation methods, the investment and governance process is different to the traditional management structure common to most institutions. Successful implementation rests on being able to manage the risk of these strategies in-house because investment decisions are effectively being transferred from active managers to the in-house team.
Selection of factors:
Factors should be chosen on the basis of how they can achieve investment objectives within the confines of constraints such as risk appetite, ESG policies, and so on. Investors will benefit from examining the economic and investment rationale underpinning the candidate factor premia to ascertain whether their returns are derived from market inefficiencies, investor behavioral biases or from a rebalancing premium.
As in asset class-based investing, timing is a challenge. Factors can underperform for long periods of time. For example, value and low volatility factors underperformed during the momentum-driven technology bubble of the late 1990s. Blending factor strategies to exploit their low cross-correlations could potentially alleviate timing difficulties.
Selection of strategies and managers:
Once the right blend of factors has been selected, the next stage is implementation, which involves selecting the best strategies and managers. Broadly speaking, investors can opt for passive implementation that makes use of purely rules-based strategies (such as index strategies), or semi-active implementation, which is systematic but still allows portfolio managers to have some discretion in constructing the portfolio.
In view of the diversity of implementation approaches, the due diligence process is often complex and time-consuming. Investors should be cognizant of the implications associated with different portfolio construction methodologies and their exposure to risk factors.
Figure 2
For a larger view, please click on the image above.
Figure 2
For a larger view, please click on the image above.
As alternate beta strategies move away from their market capitalisation benchmarks, investors need to be aware of sector concentration or secondary factor tilts that may arise. For instance, some value-based strategies may have a momentum bias during certain periods, and low volatility strategies may inadvertently be exposed to value stocks. Investability may also be an issue.
To illustrate the importance of understanding the resulting exposure of investment portfolios, we created a stylised portfolio based on the S&P 500® that blended a 40 percent low volatility strategy with 60 percent equally spread across small-cap, value, momentum and quality strategies. With the help of Northfield risk models, we dis-aggregated the active return of the blended portfolio.
Exhibit 5 shows that, as compared to the benchmark, the blended alternate beta portfolio has more exposure to small-cap stocks, high dividend yield and lower beta. In terms of sector exposure, it has a slight tilt towards utilities and away from technology companies. In addition, it has a bias towards credit risk premium widening, meaning that the strategy would have historically performed well in an economy that was underperforming or growing below trend. These exposures might not be what investors expected at the outset, and highlight the necessity of understanding the characteristics of their investments — in particular, the factor/industry tilts their portfolios have, the return of the factors/industries over time and macroeconomic factors to which they are most exposed.
This analysis can be extended to all asset classes and performed on the overall portfolio level as it allows investors to evaluate the efficacy of their overall portfolios over time, in view of their investment objectives and constraints.
Figure 2
For a larger view, please click on the image above.
It is important to recognise the trade-off between an investor’s willingness to assume risk and the expected portfolio return, net of rebalancing costs. Overall, according to Jaconetti et al (2010), the risk-adjusted returns are not meaningfully different whether a portfolio is rebalanced monthly, quarterly, or annually. However, as portfolio turnover increases, transaction costs can rise significantly. In general, investors may want to select alternate beta strategies that are both simplistic and transparent in their approach and require relatively low turnover.
Portfolio construction:
Implementation costs are key considerations in the portfolio construction stage, and they can vary significantly from investor to investor, as they are dependent on the size of the investment in question and how the strategy is being executed. In general, beyond custody fees, managers’ fees and potential index license fees, two types of transaction costs—both direct and indirect—need to be considered. The direct costs may include commissions, duties and taxes, while the indirect costs may include bid/ask spread, market impact and the opportunity cost of trading.

For larger funds, while the direct costs (especially commissions) are not immaterial, they pale in significance when compared with the indirect costs, such as market impact costs that traders can incur as a result of slippage arising from insufficient investment capacity. Moreover, indirect costs are remarkably difficult to model. This may suggest why larger investors in alternate beta strategies tend to prefer weighting schemes that retain some kind of link to the market capitalisation of the individual stocks.
Furthermore, there is a trade-off between investment capacity and the degree of factor exposure. A weighting scheme that is designed to get as much exposure as possible to a certain factor is unlikely to have a high investment capacity. Precisely what weighting scheme investors should choose should depend on the purpose of the allocation; a tactical allocation involving a small amount of money may call for a different weighting scheme than a strategic/core allocation.
Figure 2
For a larger view, please click on the image above.
Performance measurement and monitoring:
Ongoing performance monitoring is indispensable to conform the alternate beta strategies chosen with investor expectations and objectives. This would involve examining the portfolio’s overall factor exposure, sector biases and whether it has any secondary exposures to macroeconomic factors.

Risk Premia as Portfolio Building Blocks
There is a growing interest in investing in risk premia directly. The difference between alternate beta and risk premia strategies is market exposure. Essentially, alternate beta are long-only strategies that have both market and factor exposures. Portfolios based on these strategies are primarily dominated by market risk. In contrast, the risk premia strategies that we refer to in this paper are long-short strategies that aim to separate systematic risk factors from overall market risk. Bender et al (2010) note that correlations between many risk premia have been historically low, and a portfolio of risk premia may represent a new approach to portfolio diversification.

Case Study
To demonstrate this, we constructed a portfolio consisting of 10 liquid risk premia by taking long positions on different alternate beta strategies and offsetting them against their corresponding benchmark in order to attempt to isolate the factor. A long-short portfolio constructed in this manner does not capture beta-neutral exposures, but it could be easier and cheaper to implement.
For equities, we took long positions on the small-cap, low volatility, value, momentum and quality indexes and, simultaneously, took a short position on the benchmark. Similarly, for commodities, we took long positions on three risk premia—curve, value and momentum—through long-short commodity indexes. Finally, for fixed income, we proxy the credit risk premium by going long on U.S. corporate high yield bonds and short on the U.S. Treasury bills. The term premium was isolated by taking a long position in long-duration U.S. Treasury bonds and a short position in U.S. Treasury bills. These risk premia are well understood and relatively easy to implement.

Exhibit 6 indicates that, on average, these risk premia yielded strong returns historically. However, despite removing significant market exposure their volatilities and maximum drawdowns were relatively high. Moreover, they were susceptible to long periods of underperformance. As an example, the maximum drawdowns of the small-cap and low volatility factors were as high as 42.4 percent and 44.8 percent, respectively, during the examined period.
Figure 2
For a larger view, please click on the image above.
On a more positive note, the correlation between risk premia tended to be low historically (Exhibit 7) and the average pairwise correlation between risk premia was almost zero. These results support the findings of Bender et al (2010). Of particular interest, is the fact that they stayed low during the last financial crisis. This is in stark contrast with the average pairwise correlation between asset classes, which was about 0.25 for the whole period, but increased to 0.35 during the financial crisis.
Central to the risk premia portfolio construction process is the exploitation of low correlated risk factors to reduce overall portfolio risk. To illustrate this, we put together a hypothetical portfolio consisting of these 10 liquid risk premia based on a risk parity methodology and compared it with a traditional balanced portfolio comprising 50 percent equities, 40 percent fixed income and 10 percent commodities.
Figure 2
For a larger view, please click on the image above.
PracticalConsiderations__Fig_8b
For a larger view, please click on the image above.

Figure 2
For a larger view, please click on the image above.
Figure 2
For a larger view, please click on the image above.
Exhibit 8 shows that the volatility of the risk premia portfolio was very low during the entire period. Moreover, the maximum drawdown was 2.9 percent.
This compares favorably with the balanced portfolio, which was 3.5 times more volatile and had a drawdown of over 32 percent. If the volatility of the risk premia portfolio were to be scaled to the same level as that of the balanced portfolio, the leveraged risk premia portfolio would have achieved a much higher excess return (15.1 percent p.a. vs. 4.5 percent p.a.).
It is interesting to note that the correlation of the risk premia portfolio with equities was -0.1 over the entire period and this may be the consequence of having a low exposure to traditional market beta risks. Some institutional investors have already started making allocations in risk premia strategies as a low-cost alternative to absolute return strategies.

Implementation Issues to Consider
In addition to the implementation issues previously identified, the following points should be taken into account.
Short selling:
Long-short risk premia strategies make extensive use of shorting and leverage. However, it may become prohibitive or even impossible to short securities in times of crisis, as illustrated in the lending spreads of the S&P 500, S&P MidCap 400® and S&P SmallCap 600® as shown in Exhibit 9. In general, the securities’ lending spreads aim to reflect the difference between the funding rate and the average securities’ lending rate for the reference equity index.
They are used to approximate the true cost to borrow. Evidently, the less liquid a stock is, the higher the potential cost of borrowing. In addition, some investors in certain countries may not be permitted to use derivatives to take short positions. Without derivatives, implementing a short position may be costly and impractical.
High transaction costs:
The extensive use of short selling, leverage, the need for regular rebalancing given the volatile nature of the factors, and the low capacity of some factors, may lead to high transaction costs that may erode the returns of these risk premia strategies in practice.
Unstable correlations between factors:
One of the cornerstones underlying the concept of risk premia strategies is the low correlation between factors. However, factor correlation can be volatile and unstable. Exhibit 10 shows the rolling three-year correlations of four factors. For example, the correlation between small cap and equity value ranged from -0.13 to 0.66.
Many systematic risk premia strategies follow simple weighting schemes, such as equal weight, volatility weight or risk parity weight. This may result in overweighting underperforming factors and underweighting outperforming factors.

Conclusion
In this paper, we explored two approaches to incorporating risk factors into asset allocation and portfolio construction. The first approach involved enhancing returns and reducing risk using alternate beta building blocks; and the second examined constructing an absolute return portfolio using risk premia building blocks.
Alternate beta or factor-based investing is becoming a viable way of constructing institutional portfolios. The approach can either seek to enhance return or reduce risk, or both. In this paper, the challenges institutional investors face in their decision-making and implementation were highlighted.

Finally, we reviewed the concept of long-short risk premia strategies, which some investors use as a low-cost alternative to absolute return strategies.
While the concept may have a strong theoretical underpinning, its more complex nature and relatively lower capacity mean that implementing it in large institutional portfolios will be a significant challenge.
In summary, investors are increasingly adopting alternate beta and risk premia as building blocks for asset allocation and portfolio construction. While the implementation of these strategies is passive, selecting the right blend of factors and implementation strategies is an active decision-making process.
Nevertheless, the continued development and adoption of these tools may help to increase transparency of investment processes and reduce costs in the asset management industry.

References and Endnotes
  • Ang, Goetzmann and Schaefer, Evaluation of Active Management of the Norwegian Government Pension Fund – Global, 2009
  • Banerjee and Srivastava, Limiting Risk Exposure with S&P Risk Control Indices, 2012
  • Bhansali et al., The Risk in Risk Parity: A Factor-Based Analysis of Asset-Based Risk Parity, 2012
  • Bender et al., Portfolio of Risk Premia, 2010
  • Jaconetti et al., Best practices for Portfolio Rebalancing, 2010
  • Kang, Evaluating Alternative Beta Strategies, 2012
  • Ung and Kang, Alternative Beta Strategies in Commodities, 2013
  • The S&P 500 Low Volatility Index (“the Index”) was launched on April 4, 2011. All information presented prior to the launch date is back-tested. Back-tested performance is not actual performance, but is hypothetical. The back-test calculations are based on the same methodology that was in effect on the launch date. Complete index methodology details are available at www.spdji.com.
  • The S&P SmallCap 600 (“the Index”) was launched on Oct. 28, 1994. All information presented prior to the launch date is back-tested. Back-tested performance is not actual performance, but is hypothetical. The back-test calculations are based on the same methodology that was in effect on the launch date. Complete index methodology details are available at www.spdji.com.
  • The S&P Global BMI Indices (“the Index”) and its sub-indices were launched on Dec. 31, 1992. All information presented prior to the launch date is back-tested. Back-tested performance is not actual performance, but is hypothetical. The back-test calculations are based on the same methodology that was in effect on the launch date. Complete index methodology details are available at www.spdji.com.
  • The S&P GSCI (the “Index”) was launched on May 1, 1991. All information presented prior to the launch date is back-tested. Back-tested performance is not actual performance, but is hypothetical. The back-test calculations are based on the same methodology that was in effect on the launch date. Complete index methodology details are available at www.spdji.com.
  • The S&P GSCI Dynamic Roll (the “Index”) was launched on Jan. 27, 2011. All information presented prior to the launch date is back-tested. Back-tested performance is not actual performance, but is hypothetical. The back-test calculations are based on the same methodology that was in effect on the launch date. Complete index methodology details are available at www.spdji.com.
  • The Commodities Value Strategy on the S&P GSCI is constructed by going long on 18 commodities with the highest gradient based on their futures curve and is rebalanced monthly.
  • The Quality Strategy on the S&P 500 is constructed by selecting the top quintile of securities that are ranked the highest based on their accruals ratio, return on equity and financial leverage ratio.
  • The Momentum Strategy on the S&P 500 is constructed by selecting the top quintile of securities that are ranked based on their 6-month risk-adjusted return and their 12-momnth risk-adjusted return.
  • S&P Dow Jones Indices defines various dates to assist our clients in providing transparency on their products. The First Value Date is the first day for which there is a calculated value (either live or back-tested) for a given index. The Base Date is the date at which the Index is set at a fixed value for calculation purposes. The Launch Date designates the date upon which the values of an index are first considered live; index values provided for any date or time period prior to the index’s Launch Date are considered back-tested. S&P Dow Jones Indices defines the Launch Date as the date by which the values of an index are known to have been released to the public, for example via the company’s public Web site or its datafeed to external parties. For Dow Jones-branded indices introduced prior to May 31, 2013, the Launch Date (which prior to May 31, 2013, was termed “Date of Introduction”) is set at a date upon which no further changes were permitted to be made to the index methodology, but that may have been prior to the Index’s public release date.
  • Past performance of the Index is not an indication of future results. Prospective application of the methodology used to construct the Index may not result in performance commensurate with the back-test returns shown. The back-test period does not necessarily correspond to the entire available history of the Index. Please refer to the methodology paper for the Index, available at www.spdji.com for more details about the index, including the manner in which it is rebalanced, the timing of such rebalancing, criteria for additions and deletions, as well as all index calculations.
  • Another limitation of using back-tested information is that the back-tested calculation is generally prepared with the benefit of hindsight. Back-tested information reflects the application of the index methodology and selection of index constituents in hindsight. No hypothetical record can completely account for the impact of financial risk in actual trading. For example, there are numerous factors related to the equities (or fixed income, or commodities) markets in general which cannot be, and have not been accounted for in the preparation of the index information set forth, all of which can affect actual performance.
  • Additionally, it is not possible to invest directly in an Index. The Index returns shown do not represent the results of actual trading of investable assets/securities. S&P Dow Jones Indices maintains the Index and calculates the Index levels and performance shown or discussed, but does not manage actual assets. Index returns do not reflect payment of any sales charges or fees an investor may pay to purchase the securities underlying the Index or investment funds that are intended to track the performance of the Index. The imposition of these fees and charges would cause actual and back-tested performance of the securities/fund to be lower than the Index performance shown. For example, if an index returned 10% on a US $100,000 investment for a 12-month period (or US$ 10,000) and an actual asset-based fee of 1.5% was imposed at the end of the period on the investment plus accrued interest (or US$ 1,650), the net return would be 8.35% (or US$ 8,350) for the year. Over a three-year period, an annual 1.5% fee taken at year end with an assumed 10% return per year would result in a cumulative gross return of 33.10%, a total fee of US$ 5,375, and a cumulative net return of 27.2% (or US$ 27,200).
 

Find your next ETF

Reset All