A brief introduction to long-short tax-aware optimization

The most standard use of long-short tax-aware optimization is the management of concentrated positions. Typically, accounts new to a firm have unrealized gains and carry significant company-specific risk. Common strategies for managing such positions include crafting a “tax overlay” portfolio around existing holdings or gradually liquidating large positions. Both methods require portfolio managers to effectively trade off tax liability with risk.

Long-short tax-aware optimization can assist with this trade off. By using a long-short framework alongside a factor risk model, portfolio managers can use shorts to hedge factor exposures, minimizing the risk of the concentrated positions and creating a more effective overlay portfolio. At the same time, these strategies yield increased opportunities for tax-loss harvesting. As losses are realized over time, they offset realized gains in the concentrated positions, and the proceeds are used to further diversify, bringing the portfolio closer to the model. 

Tax-aware optimization can offer increased flexibility to manage significant taxable events as well. For example, consider an individual preparing to sell a family business or perhaps complete a home sale. The income earned by these and other one-time events results in additional tax burden. Aggressive tax-loss harvesting can help mitigate the impact of these events by reducing an individual’s overall tax liability.

Through an optimization framework, portfolio managers can target a specific level of net gains or simply choose to minimize realized net gains as much as possible. A strategy's net gains budget can be zero or even negative, indicating tax neutral or excess losses realized. Crucially, a manager can incorporate previously realized or anticipated gains into the current rebalancing decision. For example, suppose a client expects a taxable event three months into the future. Minimizing the net tax gain today will encourage the optimizer to offset as much of that future gain as possible.

While the situations above can be handled within a long-only framework, adding a short book increases the latitude given to the optimizer, leading to less net tax gains since short positions create more potential loss harvesting opportunities. In an up-trending market, short positions generate additional losses when harvesting opportunities are scarce on the long side of the book. 

One of the most popular uses of long-short tax aware optimization is a long-short factor tilt strategy, which attempts to improve upon long-only direct indexing. In a bull market, direct indexing portfolios will tend toward holding lots with embedded gains. These portfolios experience diminishing tax alphas over time, because fewer new tax-lots result in fewer loss harvesting opportunities. To avoid this issue, PMs can create a long-short mean-variance optimization case where the expected return is defined as either a proprietary alpha factor model or factor beta tilt using Axioma Risk Model style factors.

Because the resulting portfolio has both long and short positions, it will have more opportunities for tax loss harvesting regardless of the market condition. It is important to note that while both long-short tax aware and direct indexing minimize tracking error to a benchmark, the long-short portfolio has increased ability to trade more generally, because portfolio weights are not zero bounded.  

We see long-short portfolios implemented in a few different ways. Some PMs exclusively hold common stock and allow the optimizer to trade off expected return and active risk across the entire portfolio value. Others will own 100% of the long positions in an ETF that tracks the benchmark and then run a long-short common stock overlay focused on managing tax gains and losses. Clients can use different levels of leverage, but the historical prevalence of 130/30 institutional strategies and mutual funds leads many firms to choose 130% long, 30% short portfolio construction.