Improving Tax-Aware Performance Through Better Substitutions
Regius Magazine
© 2022 Regius Magazine
Tax-aware strategies - used for a tax-paying investor - often face implementation restrictions imposed by the tax code. Investment managers usually cope by substituting: positioning for the desired exposure by holding alternative assets. However, if investment managers do not execute such substitutions thoughtfully, they can contribute a meaningful tracking error and, at the same time, be unnecessarily costly. Analytical portfolio construction tools supported by advancements in AI and machine learning can enable efficient implementation.
Regius: Let's set the stage. What is tax- aware investing and why should I care about it?
Omega Point: Tax-aware investing is a large and ever-evolving field of investment management. Many institutional investors, such as sovereign wealth funds and pension funds, do not pay taxes on their capital gains. However, other investors, for example, high- net-worth individuals and family offices, are subject to tax, and they should make investment decisions based on after-tax performance metrics.
For example, an individual offered to choose between a tax-agnostic expected return of 6% and a tax-aware alternative of 5% return that provides a 4% 'loss for tax' should prefer the latter. At a 40% tax rate, the 4% tax loss gives them an immediate 1.6% savings.
How is this ‘loss for tax’ generated? Let's illustrate it using a popular tax-aware strategy . The strategy invests in single-name stocks and takes advantage of the difference between the 20% 'Long Term' (LT) capital gains tax rate and the 40% 'Short Term' (ST) rate. Stocks held for less than 12 months pay the 40% ST rate; stocks held for more than a year are subject to the lower LT rate of 20%. Thus, a portfolio manager can consider each stock’s P&L since purchase when trading: sell losers before the 12 months mark but hold winners longer and only sell them after the anniversary.
Example: consider the case where overall the winners generate a $10 LT gain (subject to the 20% tax rate) while the losers experience a $5 ST loss (subject to 40%). Pre-tax, the strategy ends up with a $5 net gain. Calculating the tax liability, 20% of the $10 LT gain and 40% of the $5 ST loss net out to $0 gain for tax (we assume that the investor has other ST gains to offset the ST capital losses generated here). In this example, the $5 gain comes with no incremental tax liability!
Regius: Excellent - it seems like any tax paying investor should consider these strategies. So, where is the catch?
Omega Point: Tax-aware investing is hard to implement, as the tax savings compete with the potential drag on real-world performance. The excess turnover and trading costs are minor considerations here. A more significant concern is the 'wash sale' rule: if the manager sells a stock at a loss, they should not buy the same stock for 30 days. If they buy, the IRS bridges the time gap, and the ST loss is not recognized. As a result, a tax-aware strategy constantly finds itself with a heavily populated list of 'can't trade' names. Such a list is ever- changing and somewhat arbitrary in terms of its composition. Effective stock substitutions are essential to managing the portfolio under such constraints, otherwise the tax-aware strategy will underperform the tax-agnostic version
Regius: Ok...seems like a lot to unpack here. Let’s start with the idea of stock substitutions. What are those and how are they used?
Omega Point: Substitutions are common in many cases; they are not unique to tax-aware portfolios. Some of the drivers include regulation, guidelines, liquidity, or holidays.
For instance, the emergence of environmental, social, and governance (ESG) aware investing has created many strategies that aim to reconstruct global indices, such as S&P500 or MSCI World, without stocks that represent companies with poor ESG characteristics. These strategies need to constantly find substitutions in order to replicate the performance of the indices.
Let's add tax awareness to the mix. We mentioned the wash sale restriction above. Constructing a portfolio that aligns with the strategy while maintaining a very dynamic restricted list of forbidden assets requires the careful practice of the timeless art of portfolio construction.
Regius: Now let’s expand into effective substitution. What does that entail?
Omega Point: To build an effective substitution the manager needs to replicate the performance characteristics of the restricted stocks with the minimum tracking error possible.
Let's focus on the tax-aware example: we have a desired portfolio of holdings, yet a forbidden sub-portfolio that the wash sale rule does
not allow trading into it all. A manager can calculate the Market Beta of this sub-portfolio and substitute it with S&P 500 futures. This straightforward substitution is better than nothing, simple, but not great. Each name in the forbidden sub-portfolio has an industry affiliation, exposure to style factors such as momentum or leverage, exposure to specific commodities, and more. A substitution based on just the Market Beta is likely to have a sizable unintended exposure to industries, styles, and commodities relative to the desired exposure. Such exposure would fluctuate as the restricted list swings and would create a potentially large performance tracking error to the model portfolio.
Regius: Seems like the understanding of ‘factors’ is critical to effective stock substitution. Can you elaborate?
Omega Point: Most people have heard of 'value investing,' buying cheap stocks and selling expensive ones. The factor playing field is much richer than that. About 60 years ago, we started considering a factor called the Market Beta. Then researchers found 'value', 'momentum', 'size', and a few other styles. Factor models today consider hundreds of factors and are intended to capture a smaller subset representing those most persistent in describing market movements.
To illustrate how this is relevant for substitutions, consider two industrial companies of the same 'sector.’ We wish to substitute one with the other. They differ in their dividend policy and leverage appetite. Their commodities exposure is also different due to their product mix or even to their hedging policy. Furthermore, looking at their trailing year performance, one is down and now cheap while the other is up and currently relatively expensive.
Factors such as dividend yield, leverage, commodity sensitivity, value, and momentum, can capture all of the above differences between the stocks, which impact their performance characteristics. Thus, what superficially seems to be a good substitution candidate may justify deeper analysis. If we consider the traditional tax-aware strategy mentioned earlier, where winners are held and losers are sold to avoid incurring taxes, it is sensible that “factor creep” is introduced into the portfolio. Winners are more likely to have a bloated P/E (low value) and recent out-performance (high momentum) and this can translate to unintended factor exposures. In fact, any factor that outperforms is likely to create winners in the stocks that have higher exposure to it. When winners substitute losers, a factor bias is likely.
Regius: It appears that there are numerous factor characteristics to track for an individual manager. Are there any tools or processes that do this in practice?
Omega Point: Substitutions are a general need in portfolio construction and risk management. Managers can use the tools such as factor models, correlation matrices, and optimizers developed initially for risk management to solve tax-aware substitution problems.
A prudent manager should gather their current stock holdings, desired holdings, and allowed investment universe and add investment parameters such as exposure limits to factors. They should also know if (for example) substituting one name with another creates significant, unintended, and undesired exposure to value, momentum, or some other factor.
Portfolio optimizers can find cost-effective trades into a portfolio that meets all those constraints. In addition to Market Beta, the optimizer would consider other style and industry factors and minimize the single name risk by improving diversification.
The bottom line: Investments that are thoughtful about tax introduce restrictions and mandate substitutions. Those substitutions, if not managed diligently, can create significant undesired exposures. Such exposures are measurable and avoidable with the use of risk management and portfolio construction tools.
Regius: So, is this problem solved with today’s risk management technology? What does the future hold?
Omega Point: There are several trends at play: The first is democratization of risk management and portfolio construction tools. Once only available to the PhDs in Finance and Math, these tools are being made increasingly available to the broader investment community ‘as a service’ through platforms such as Omega Point.
The second is the increased use of alternative data sources in defining performance characteristics of securities. The meme-stock craze has highlighted the need to track characteristics related to institutional and retail crowding. Quantitative approaches can transform economic data to offer a better view into macro drivers.
The third trend we see is the utilization of AI and machine learning in the service of factor models. The current factor models are certainly sophisticated but far from perfect. They cannot, by construction, capture temporary factors such as Brexit or COVID. They are also not designed to catch all interactions between factors. For example, we do not have a 'Delta variant recovery factor.'
The recent advances in AI and machine learning for finance provide more comprehensive models of stocks returns. Semi Supervised learning methods such as Reinforcement learning, spectral clustering, and others, offer the unique intersection of modeling and finding nonlinearities in factor models. All the while having an element of interoperability and explainability important for traditional investing strategies. We believe those models will lead to better accuracy when constructing substitutions in the future.
Authors
Omer Cedar – CEO @ Omega Point
Omer is CEO and co-founder of Omega Point. Previously, Omer served as SVP, Research at Two Sigma Investments. In this role, he founded and built Two Sigma's global equity research analyst survey platform ("TAP"). He was responsible for research/modeling, product development, and relationship management. He also managed Two Sigma's quantitative risk arbitrage system, developed models, and conducted classic fundamental Risk Arb research. Omer holds a Bachelor's Degree in Electrical Engineering and Computer Science from UC Berkeley and an MBA from the MIT Sloan School of Management.
Prof Tal Kachman – AI, Machine Learning @ Omega Point
Prof Tal Kachman received his B.Sc (chemistry), B.Sc (physics and mathematics), M.Sc (Engineering) from the Technion I.I.T. He obtained his Ph.D in theoretical physics jointly from the Massachusetts institute of technology and Technion I.I.T, He was a researcher staff member in IBM research, AQR capital management and co-founded Rhizome Works a boutique deep learning consulting firm where he is currently a principal and a Researcher. Currently he is a professor of AI at Radboud university doing working on large scale reinforcement learning and machine learning
Avi Rosenbluth – Portfolio Construction, Tax Aware Strategies @ Omega Point
Avi’s long career in analytical finance has spanned a wide spectrum: banking and investment management, liquid and illiquid assets, equities, FX, derivatives and structured products. During his tenure with Goldman Sachs, Avi advised issuers and investors on portfolio valuation and structuring. Later, on the buy-side, he focused on implementing data-centered investing as a portfolio manager with AQR Capital Management, a large systematic fund.
Chris Martin – Product Manager @ Omega Point
Chris spent over 11 years at Axioma in a variety of senior positions, most recently as Director of Product Management. In this role he worked closely with all members of the Axioma team, including: Product, Research, Content, Sales, and Support. Chris received his Masters in Financial Engineering, a joint degree from the Drucker School of Management and Mathematical Sciences at Claremont Graduate University. He received his bachelor’s degree in General Engineering with a concentration in Aeronautical and Mechanical Engineering and a Minor in Physics from California Polytechnic State University, San Luis Obispo. Chris is a certified Engineer-in-Training in California and is a CAIA and CIPM charterholder.
