Factor strategies, such as value, low volatility and quality, often end up with industry tilts that can be an unintended source of risk. In March 2017, Morningstar published a study that found that value and low volatility investment strategies exhibited consistent industry tilts that did not significantly contribute to their success. This suggests that investors could reduce active risk without materially sacrificing returns by constraining industry weightings among factor strategies with persistent industry tilts.
To investigate this further, I compared the performance of MSCI's sector-neutral and unadjusted quality and value indexes across several markets. While quality was not covered in the previous study, it exhibits similar persistent sector tilts to value, so it would be reasonable to expect similar results. These indexes represent investable strategies, though much of their data is back-tested.
In addition to constraining their sector weightings, these sector-neutral indexes select stocks based on their factor characteristics relative to their sector peers. Arguably, this approach could lead to better performance than only constraining the sector tilts. Valuations and quality characteristics, like profitability, tend to be more comparable within industries than across them, as firms in the same industry tend to have more-similar balance sheets and business models. Therefore, sector-relative factor characteristics may contain greater information about future expected returns than unadjusted factor data.
Quality
MSCI's sector-neutral and unadjusted quality indexes facilitate a clean comparison that isolates the impact of sector adjustments. They both target stocks with high return on equity, low debt/capital and low earnings growth variability over the past five years. The only difference between them is that one does not make any adjustments for sector membership, while the other measures each stock's quality characteristics relative to its sector peers and sets its sector weightings equal to that of its parent index.
Exhibit 1 shows the differences in the performance statistics between MSCI's sector-neutral and unadjusted quality indexes across four regions and three countries where both versions of the indexes were available. The data is from December 1998 through June 2017.
Surprisingly, this data suggests that the sector-neutral quality indexes have not offered a clear performance advantage over their unconstrained counterparts. None of the sector-neutral quality indexes generated higher returns than the corresponding unadjusted indexes. And five of the seven sector-neutral indexes exhibited slightly higher volatility than their unadjusted counterparts. This is likely because sector adjustments cause the indexes to hold stocks with lower absolute quality characteristics than the unconstrained versions, which may be a little less defensive.
Five of the sector-neutral indexes exhibited lower tracking error against their respective market benchmarks than the unadjusted indexes. The sector-neutral versions should exhibit lower tracking error because their sector weightings mirror those of the market. This largely eliminates a source of active risk that the unadjusted indexes have. So, it is a little surprising that the Japan and UK sector-neutral indexes had slightly higher tracking error than their unadjusted counterparts, though this may be due to the limited size of these markets. Even though most of the sector-neutral indexes exhibited lower tracking error, only one (covering the U.S. market) exhibited a more favourable active risk/reward trade-off (measured by the information ratio) than the unadjusted version.
A factor regression analysis offers greater insight into the drivers of each index's returns. I regressed each quality index's excess returns over Treasuries against the market-risk premium, small-cap, value and profitability factors from their respective markets. The differences between the regression coefficients for the two index suites are presented in Exhibit 2.
There wasn't a clear difference in the market risk (beta) or size exposures between the two versions of the quality indexes. Interestingly, the sector-neutral indexes tended to have slightly greater exposure to the value factor, though some had much lower exposure to the profitability factor--arguably one of the most important dimensions of quality. This is likely because these indexes had to reduce their exposure to stocks with the highest absolute quality characteristics to maintain sector neutrality.
After adjusting for differences in factor exposures between the sector-neutral and unconstrained versions of the quality indexes, only one sector-neutral index (covering the United Kingdom) generated better excess returns (alpha). This data further suggests that this sector-neutral quality approach does not offer a clear performance improvement over the unadjusted approach.
Value
MSCI's sector-neutral value indexes (which are branded "Enhanced Value") target stocks that are cheap relative to their sector peers and match the sector weightings of the broad market. However, it does not have a matched version that uses the same approach without the sector adjustments. Instead, I compared the performance of the enhanced value indexes against MSCI's market-cap-weighted value indexes drawn from the same universe.
There are some important differences between these two index suites, unrelated to sector adjustments, that could affect the results. The enhanced value indexes cover a smaller portion of the market and use slightly different valuation. They also weight their holdings based on both the strength of their value characteristics and their market capitalization, which should strengthen their value orientation. That said, a factor regression analysis should help disentangle these effects.
Exhibit 3 illustrates the performance differences between the enhanced value and market-cap-weighted value indexes across the same markets as the quality indexes, with one exception. Here, the All Country Asia Pacific ex Japan region (which adds Australia and New Zealand to the mix) replaces the All Country Asia ex Japan region because there isn't an enhanced value index available for that region. This data covers the period from December 1997 through June 2017.
Unlike the results for the quality indexes, this data suggests that the sector-neutral enhanced value indexes offered a performance advantage over their unconstrained market-cap-weighted counterparts. Since the sector adjustments aren't the only differences between these indexes, it is harder to gauge the impact of those adjustments. But the results still imply that it is possible for a value strategy to achieve better results while constraining its sector weightings.
Every sector-neutral enhanced value index generated higher returns than its market-cap-weighted value counterpart, but with greater volatility. This higher volatility is likely a result of the enhanced value indexes' narrower market coverage. These indexes also tend to have a smaller market-cap orientation because of their weighting approach, and this can also increase volatility. Despite their higher volatility, all the enhanced value indexes posted better risk-adjusted performance (measured by their Sharpe ratios) than the market-cap-weighted value indexes. The enhanced value indexes also all offered a better active risk/reward trade-off (as measured by their information ratios), despite exhibiting greater tracking error.
While there are important differences between these sets of indexes, a factor regression analysis suggests that the sector-neutral indexes still outperformed after controlling for their style differences. I regressed the excess returns of each index over Treasuries against their exposure to the market risk, small size, value and momentum factors from the French Data Library. The differences between the regression coefficients for the two index suites are shown in Exhibit 4.
All the sector-neutral enhanced value indexes exhibited greater sensitivity to market fluctuations than their unconstrained market-cap-weighted counterparts. This is likely because stocks that are cheap relative to their sector peers tend to be riskier than average and the enhanced value indexes have fewer moderate value stocks to mitigate this risk. Not surprisingly, the enhanced value indexes had greater exposure to the small-cap factor than the market-cap-weighted value indexes. In a couple of cases, the enhanced value indexes had considerably greater exposure to the value factor than their counterparts, but there wasn't a big difference among the other indexes. Similarly, while the enhanced value indexes tended to have slightly greater negative exposure to momentum, these differences generally weren't very big.
After controlling for these stylistic differences, the sector-neutral enhanced value indexes tended to outperform (as their larger alphas indicate) the unconstrained market-cap-weighted value indexes. These results suggest that a sector-neutral value strategy can, in fact, offer a performance advantage over an unconstrained value strategy. This is consistent with Morningstar's previous study on the impact of industry tilts on the performance of the value factor.
Unintended risk
It is a bit of a puzzle that sector adjustments led to better performance for the value factor but not for the quality factor. The arguments for applying sector adjustments to value also apply to quality. Just as some industries consistently trade at lower valuations than others, some industries consistently favour higher-quality stocks. For instance, technology stocks consistently enjoy higher average return on equity (profitability) and lower debt/capital ratios than utilities stocks. So, there should be better comparability of both valuations and quality characteristics within sectors than across sectors.
Both strategies exhibit persistent sector tilts that can be an unintended source of risk. That said, quality's sector tilts may be less persistent than a typical value strategy's because profitability leadership shifts from cyclical stocks to defensive stocks at different points in the business cycle. The unadjusted quality indexes may have better captured these shifts, which may have helped their performance. Although it doesn't always lead to better performance, limiting sector tilts among both value and quality strategies is prudent because it allows investors to mitigate unintended bets and better target the factor of interest.
