The smart beta exchange traded fund universe is quickly expanding as money managers employ various market factors and come up with new ways to potentially enhance returns or diminish risk exposure.
Among the new innovations that could hit the space, the so-called Tranching and Fama-Macbeth OLS Regression Methodology, are two novel features that could change the way international ETFs provide country exposure.
Outlining a potential construction of a multi-factor fund-of-funds diversified international portfolio through single-country ETF components, Sailesh S. Radha, president and chief investment officer at Borealis Global Advisory, wrote a paper found in the summer edition of the Journal of Index Investing, titled “Global Country Allocation Framework: A New Paradigm for Constructing an International Fund-of-Funds ETF,” that two new novel components that go beyond traditional features of smart beta investing include tranching and the Fama-Macbeth (FMB) ordinary least squares (OLS) regression methodology.
Tranching is described as a novel scheme that classifies investable entities of a given type found in a benchmark or index universe into groups by high level investment attributes like market-cap, volatility, degrees of correlation, style and levels of development.
By breaking down the benchmarks into groups of chosen types of investment entities, Borealis Global Advisory are separating them into so-called silos that exhibit certain attributes to more clearly derive factors that drive the returns of the entities in each tranche.
“The investment attributes that are used to carve the benchmark into tranches are elements that portfolio managers want to control in order to model a multi-factor portfolio that achieves risk premiums every stage of an economic or equity life cycle,” Radha said, referring to controlled elements like style, regions and volatility.
By building these tranches, managers are better able to deal with the differing natures of economic and equity cycles across countries, design portfolios by overweighting certain groups with certain attributes that are outperforming and underweight those that are falling behind.
The Fama-Macbeth OLS Regression Methodology refers to periodically running FMB regression across rolling time-horizons of fixed spans to determine factors that drive returns of entities chosen in each tranche. Radha describes the FMB regression as simply the set of cross-sectional OLS regressions run over multiple periods spanning the immediate trailing time horizon for a given point in time.
“The regressions allow for periodic observation of the cyclicality of the factors to ascertain which factor regimes are increasing in dominance, losing dominance or emerging,” Radha said. “This aspect of implementing the FMB regressions helps set factor weights of multi-factor portfolios.”
Borealis Global Advisory would then allocate weights to the factors proportional to the magnitude of the t-statistic, or the probability of difference between populations, from the FMB regressions.