We recently published a piece called A Guide to ESG Portfolio Construction, which discussed six different ways to build portfolios for ESG (environmental, social, and governance) investors based on the type of beliefs a client holds and the type of data that is available. In that piece, we reference the concept that if the type of belief (either a binary yes/no or a score on a continuum) is the same, then the portfolio construction will be the same regardless of whether the investor is motivated by values or by alpha seeking. As one of the authors of the paper, I want to extend this idea to a consideration that didn’t make it into the final version of the paper. Namely, if a values investor has a portfolio constructed with data and a portfolio construction process that matches that individual’s values, then the investor is likely to be happy. An alpha-seeking ESG investor who has the same portfolio from the same data and portfolio construction process is quite likely to experience disappointment. At first glance, this may seem contradictory, but it is not.
For the purely values-motivated investor, the creation of an optimal portfolio is an end in itself. If the portfolio is able to stay within its forecast tracking error and avoid ‘bad’ companies, then the investor should be happy. For the purely alpha-motivated investor, the portfolio is a means to achieve the end goal of outperforming the market, and outperforming the market is more difficult than constructing an optimal portfolio. We can see this in the copious data on mutual fund managers underperforming their benchmarks, or we can take a more theoretical approach and illustrate the potential pitfalls of traveling the difficult road from concept to alpha.
(This is as good a time as any to mention that while the remainder of this piece uses potential ESG alpha forecasting as an example, all forecasting of financial markets has a fairly dismal record, as evidenced by the previously cited mutual fund data.)
One of the reasons moving from concept to alpha is so difficult is that generating alpha as a result of skill in forecasting requires that a number of steps be done successfully in tandem.
In sketch outline, here is one take on the sequential steps needed to make a successful forecast to generate alpha:
- Make a forecast about something.
- Translate that forecast into an efficient (or optimal) portfolio.
- Determine the degree to which the idea is already priced into the price of a security.
- Forecast something that is important to future returns.
- Monitor when the portfolio should be adjusted because one of the above steps changes.
- Ensure that pre-tax alpha exceeds the capital gains costs of rebalancing the portfolio at the end of the forecast.
The final generation of alpha for a forecast results only when a portfolio has been moved back to a benchmark position, so the last step should ultimately lead to a situation where something in the forecasting process is no longer valid. This is a complicated process that can go wrong in a myriad of ways. Using an example similar to the soft drink cascade in the Guide to ESG Portfolio Construction, we illustrate how each of these points can go wrong.
Our hypothetical investor is a political consultant who believes that soft drink producers will be blamed for associated rises in health care costs. This consultant has been working to get a tax implemented for the past 15 years and finally believes that legislation will be passed that imposes a tax on soft drink companies with revenues above $3 billion per annum. He does not believe the proposed “soda tax” is priced into the market and wants to profit from the change in government policy. The investor intends to implement this portfolio until the policy changes or until the end of the next legislative session. He is excited to finally have a portfolio that supports his work over the past 15 years and will outperform the market.
In terms of the steps described above, this hypothetical investor has done a lot of things right. First, he has a forecast that is specific and precise. This is about legislation being passed to impose a tax on a specific set of public companies. Second, the consultant can use an Optimized Exclusion to construct a portfolio because his forecast identifies exactly which companies will be impacted. (Other companies that are correlated, such as smaller soft drink companies, will not be impacted.) Third, he has assessed whether the market has priced his idea into current market values The process here is opaque, and the investor’s confidence seems much higher than is probably justified, but at least he thought about it. Fourth, he has a set of conditions that would trigger when the portfolio would be changed back to an index portfolio. (Note the fact that he wouldn’t implement the portfolio for purely value reasons, which indicates that this is really an alpha-motivated portfolio.) The consultant tried to follow some good forecasting practices with analysis and precise thinking; however, he did miss an analysis of whether this tax would be a big enough contributor to returns that even being right might not give the desired results.
What are typical things that go wrong from here?
The forecast could be wrong.
Suppose … at the last, minute a backroom deal emerges whereby all soft drink revenue is exposed to a higher-than-expected tax.
- The optimization results in above-benchmark holdings of smaller soft drink manufacturers, which have moderate negative returns while the large manufacturers have positive returns. The portfolio underperforms the benchmark.
Or … the market already priced the legislation into current market values and something else impacts prices.
The tax was implemented, but the market had already priced it in. Meanwhile, larger soft drink manufacturers gained an upper hand in negotiating prices with suppliers.
- The full impact of the tax was already in the securities prices, while changes in the supply pricing were not. Large manufacturers end up outperforming small manufacturers, which are overweight in the portfolio. The portfolio again underperforms the market.
Or … the forecast is wrong, and something else impacts prices.
Tax legislation is not passed, but there is a buying spree in which large soft drink manufacturers start acquiring small manufacturers.
- The portfolio does very well on the large increase in multiples for the small manufacturers, which are overweight in the portfolio. The alpha is positive due to the M&A activity, not as a result of good forecasting on legislation. This is luck.
Or … the forecast is slightly wrong, but relying on correlations in the portfolio construction undermines the alpha.
The tax law passes and soft drink manufacturers have negative returns. However, this event sparks a large fall in consumer confidence, which results in reduced consumer spending across the economy.
- The portfolio does poorly because the stocks used by the optimization to replace the soft drink stocks perform even more poorly than the soft drink stocks that have been excluded from the portfolio.
Or … the investor fails to exit the position after the event, and realized capital gains result in negative post-tax alpha.
The tax legislation is passed, but the investor now has embedded gains in the portfolio and is busy, so he doesn’t change the portfolio. One year later, midterm elections change control of Congress, and the tax is expected to be repealed.
- The portfolio generates alpha, but then the alpha is reversed over the next year. The net alpha of the portfolio is 0% when it is moved back to benchmark, and there is a realization of capital gains, resulting in negative post-tax performance.
The good news is that a lot of alpha isn’t necessarily generated in this manner of making a forecast with good portfolio construction, identifying an important source of returns, and executing a timed exit. It is generated through a process where one or more of these steps fails but then alpha is created anyway. This is great, but it is a result of luck—and woe betide investors who believe luck will consistently generate alpha.
Send questions or comments to firstname.lastname@example.org.