Amid Covid-19, some sectors are misbehaving — with a big impact on turnover for sentiment-aware investors

The Covid-19 pandemic has clearly affected some sectors more than others. This large and mostly binary impact was immediately reflected in changing exposures to three of the style factors in Qontigo’s fundamental multi-factor risk models, namely market sensitivity, volatility, and dividend yield. All three are part of the sector-based ROOF Score methodology used to classify sectors as having either a risk-tolerant or a risk-averse personality. These large swings in exposures have led to dramatic changes in the ROOF personalities of six[1] of the 11 sectors in the US market. ROOF Scores quantify investor sentiment by aggregating a sector’s personality (i.e., style factor exposures) with its popularity (i.e., active return) to determine an investor’s risk appetite (i.e., do they seek greater exposure to the risk-tolerant or risk-averse sectors?).

The tables below show the changes in exposures to each of the eight style factors included in the ROOF Score methodology between February 19, 2020 (the peak of market), and May 8, 2020 for the US market. These changes are significant and in several cases were large enough to cause a change in sign for the sector on that factor. Such swings have a direct effect on the sector’s ROOF personality, as the methodology is based on the sign of the exposure to the predefined sets of factors defining risk tolerance or risk aversion (see methodology document for details).

Dividend yield, and the ability of companies to pay it, is one of the fundamental attributes that investors have called into question in the current crisis. We were therefore not surprised to see large swings in that exposure. Market sensitivity (or Beta) has also seen its volatility shoot up, while experiencing swings in its correlation with certain other factors. In addition, many individual company exposures to Market Sensitivity have changed substantially. We were therefore not surprised to see it have a large impact on sector personality. The volatility factor has also played a big part in the current crisis, especially through its changing correlation with the leverage and profitability factors[2]. Volatility also had a relative impact on sector personality, especially Financials.

The chart below shows the corresponding impact these changes in exposures had on our sectors’ ROOF personalities since February 19, 2020. Recall from the methodology that a negative ROOF value means risk averse, and a positive one is risk tolerant. In most cases, these changes in personality would necessitate portfolio rebalancing for those investors wishing to retain a risk-tolerant or risk-averse portfolio in both periods.

Some observations on these changes:

  • Consumer Staples, a stalwart of the risk-averse portfolio, has lost some of its defensive qualities, but remains firmly in the risk-averse camp.
  • Utilities, the other pillar of defensive portfolios, has lost all its risk-averse character and is now neutral.
  • Financials have become less-than half as risk averse as they were prior to the market rout.
  • The Tech sector has lost the bulk of its risk-tolerant character and is mostly neutral now.
  • Real estate has experienced the most dramatic swing of all sectors, going from firmly risk averse before the crisis, to strongly risk tolerant now.
  • Communication Services went from neutral to risk averse.

These personality swings are all related to the changes in the underlying style factors, but their size is substantial and, in most cases, unprecedented in such a short time. The charts below show the personalities of the six sectors under review since January 2012 and the recent mood swings.

The size and speed of these personality swings created considerable turnover for investors trying to maintain a constant personality for their sentiment-aware portfolios. The table below shows the ROOF personality scores for all 11 sectors in the US market, as well as their weight in the US-LMS Axioma Market Portfolio as of both February 19 and May 8, 2020. Using the sector’s ROOF personality sub-score for risk tolerance (RT) and risk aversion (RA), we created two sentiment-aware portfolios, Risk Tolerant (RT Port.) and Risk Averse (RA Port.), where each sector is over-weight or under-weight proportionally to its corresponding sub-scores. For example, Consumer Discretionary as of February 19, 2020, had a risk tolerance (RT) sub-score of 0.76 and a risk aversion (RA) sub-score of 0.24, giving it a ROOF personality of +0.52 (i.e., ROOF = RT – RA), which is strongly risk tolerant. Consumer Discretionary is therefore over-weight in the RT Port (i.e., 13.3% vs. 10.1% in the benchmark), and under-weight in the RA Port. (i.e. 5.7% vs. 10.1% in the benchmark). The six sectors that have experienced large swings in their ROOF personality scores because of shifting factor exposures are highlighted in green.

Some observations on this table:

  • As a result of these style factor exposure shifts between the two dates, Energy, Financials, Real Estate, and Utilities saw their active weight in the Risk Averse portfolio (RA Port.) drop by almost half.
  • Health Care saw its large under-weight position in the Risk Averse portfolio (RA Port.) on February 19 turn to an almost neutral position (13% vs 15.4% in the benchmark) on May 8, 2020.
  • Real Estate went from underweight in the Risk Tolerance portfolio (RT Port.) on February 19 to over-weight on May 8, 2020. And vice-versa in the Risk Averse portfolio (RA Port.).
  • Communication Services saw its active weights increase between the two dates, becoming more under-weight in the Risk Tolerant portfolio (RT Port.) and more over-weight in the Risk Averse portfolio (RA Port.).
  • Utilities saw its large over-weight position (almost 2X) in the Risk Averse portfolio (RA Port.) on February 19, become a neutral position (3.5% vs. 3.2% in the benchmark) on May 8, 2020.

The large impacts from these changing factor exposures and corresponding changes in ROOF personality scores would have caused substantial sector allocation changes for any portfolio manager trying to maintain a sentiment-aware portfolio on both dates. The table below shows the change in sector allocations for a variety of sentiment-aware strategies, as well as the market portfolio (US-LMS) itself. Note that these allocation changes do not represent the actual turnover for these portfolios, not can all of the allocation change be assigned to the sentiment-aware strategy since the market will have altered the sector’s weight between the two dates organically. We are simply showing the path from one date to the other without breaking down the portion of the change that is market related and the portion that is sentiment-tracking related.

The organic sector allocation changes for the market portfolio (US-LMS) between February 19 and May 8, 2020 totaled 9.2%. A manager who constructed a risk-tolerant portfolio on February 19 and wished to maintain that characteristic on May 8, would have seen an aggregate allocation change of 15%, driven mostly by the Health Care, Financials, Communication Services, and Utilities sectors re-allocations.

A manager with a risk-averse portfolio on February 19, rebalancing to another risk-averse portfolio on May 8, would have had a total allocation change of 35%, almost seven times that of the market portfolio. The biggest change would have been experienced by managers wanting to swap the personality of their portfolio, from risk tolerant on February 19 to risk averse on May 8, or vice versa. Those managers would have seen aggregate sector allocation changes of 52% and 71%, respectively. Managers who started with the benchmark portfolio on February 19 and wanted to either rebalance to a risk-tolerant or risk-averse portfolio on May 8, would have to make sector allocation changes totaling 27% and 21%, respectively[3].

Unlike Vegas, what goes on in the style factor world does not stay in the style factor world—it affects the ROOF personalities of sectors. And for sentiment-aware investors wishing to maintain the alignment between a portfolio’s personality and their risk appetite, these personality swings have direct consequences on sector allocations.

Those consequences can be even more dire for those managers wanting to change the personality of their portfolio at the next rebalancing. For them, not only will they incur the cost of trading risk-tolerant sectors for risk-averse ones (e.g. selling consumer discretionary and buying consumer staples), or vice versa, but during the recent style factor disruption, they would have incurred the additional turnover from the rapidly changing personalities on six of the 11 sectors.

In summary, sentiment-aware portfolios can be, well, a bit emotional. At times, they can experience greater mood swings than their sentiment-agnostic counterpart (a.k.a. cap-weighted market portfolios). Sentiment-aware portfolio managers may well discover that there is a higher cost to keeping up with the Joneses during times of factor dislocation and high market volatility. In their defense, though, the higher turnover is similar in scale to most risk-constrained or smart beta strategies targeting the same misbehaving factors.


[1] Seven initially, but the swing in Industrials’ personality from risk tolerant to neutral reverted to risk tolerant in subsequent weeks.

[2] More details on these changes are available from our in-house ‘Portfolio Doctor’, Melissa Brown, here.

[3] Note that most risk-constrained portfolio strategies, especially ones designed to capture a style factor premium, would have experienced similar levels of turnover during this period so this is not unique to sentiment-aware portfolios.

Olivier d'Assier

Olivier d'Assier is Head of Applied Research, APAC, for Axioma and is responsible for generating unique regional insights into risk trends by leveraging and analyzing Axioma's vast data on market and portfolio risk. d'Assier's research helps clients and prospects better understand and adapt to the evolving risk environment in the Asia Pacific.