Investors have become more sensitive to measuring factor exposures in their portfolios; however, this measurement is not easy. There are two major approaches to measuring factor exposures, time series analysis and analysis based on firm characteristics (cross-sectional). Both have benefits and drawbacks; however, focusing on firm characteristics can provide more timely and detailed insights.
The times series approach has been extensively researched and used by investors. As developed by Fama and French, the times series approach regresses portfolio returns onto risk factor returns to measure the statistical sensitivity of a portfolio to well-defined risk factors like size, value, quality and momentum. These well-defined risk factors can be easily found in the Ken French website.
The alternative approach is to look deeply into the firm characteristics or firm descriptors which proxy for style factors that can be further grouped into more traditional factors like value, size, and volatility. This cross sectional approach has been extensively developed into well-defined frameworks, (see, for example, MSCI "Measuring Factor Exposures"). In the characteristic approach, the focus is on current holding exposures that translate into factor risks.
A portfolio can be decomposed into the descriptors for a factor which provide details on the composition of the portfolio. This fundamental composition can be compared with the times series factor exposures and the factor exposures associated with these descriptors. A close examination shows that the risk factors from cross-sectional analysis map better than times series with the actual descriptors of the portfolio. A cross-sectional style factor for value will do better than a times series measure at representing actual real time risks.
The time series approach is easy to implement; however, there are some problems that are hard to solve. If there are large and sudden shifts in the portfolio exposures, the time series approach will not pick them up. Time series risk measures will be sensitive to set-up specifications. For example, result will change with using weekly or monthly data and whether measurement is done over one year or three years. There is the potential for spurious correlations from these regressions.
Characteristics, on the other hand, need detailed information on the portfolio and a large database by which to measure risks. If an investor does not have fully transparency of portfolio holdings, it will be hard to obtain a good measure of risks. For internal risk management, the characteristics approach is far superior to the alternative.
The characteristic approach is foundational for forming an alternative risk premia index. For example, a quality portfolio can be formed through forming a weighting scheme of quality firm characteristics. Differences in quality portfolios can be measured through different characteristic weights.
The Fama French time series changed thinking about on a number of levels, but as our ability to gather and analyze data has increased, alternative approaches have been commoditized and made easier to employ.
The Fama French time series changed thinking about on a number of levels, but as our ability to gather and analyze data has increased, alternative approaches have been commoditized and made easier to employ.
No comments:
Post a Comment