The ability of three factors to explain cross-sectional excess returns
Laukkanen, Oskari (2015)
Laukkanen, Oskari
2015
Kuvaus
Opinnäytetyö kokotekstinä PDF-muodossa.
Tiivistelmä
The study challenges the prevailing asset class focus in portfolio allocation by investigating the ability of three factors to explain cross-sectional excess returns. The three factors are value, momentum and carry. The three factors are used to explain returns of equities, bonds, commodities and currencies.
The study covers 60 return series and 192 monthly excess returns per series. The created excess return series’ are used both, as dependent variables and as building blocks for the explanatory factor portfolios. From the created factor portfolios, three linear models are built. Each excess return series then is regressed against the three models in R. The used data, consisting both futures and spot contracts, is exported from Bloomberg.
The study has two parts. The first part creates and tests two models in each asset class to find an optimal factor set and model fit. In the second part, a universal model is created from one cross-asset class sample and each return series is regressed against this universal model. The universal model emphasizes the three factors’ explanatory abilities across asset class boundaries.
The three factors explain excess returns in the four asset classes under study. There is also a more universal price underreaction (momentum) and overreaction (value) that explains returns beyond traditional asset class boundaries. Supporting evidence for universal yield underreaction (carry) is less consistent however. The found evidence highlights two things: The statistical importance and independence value, momentum and carry; The need to shift portfolio allocation focus from asset classes to factors since factors drive returns in every asset class under study.
The three factors’ distribution properties suggest that momentum and carry factor premiums are compensation against rare disaster risk while the risk is unable to explain value premium. All three premiums could however exist in the presence of market frictions.
The study covers 60 return series and 192 monthly excess returns per series. The created excess return series’ are used both, as dependent variables and as building blocks for the explanatory factor portfolios. From the created factor portfolios, three linear models are built. Each excess return series then is regressed against the three models in R. The used data, consisting both futures and spot contracts, is exported from Bloomberg.
The study has two parts. The first part creates and tests two models in each asset class to find an optimal factor set and model fit. In the second part, a universal model is created from one cross-asset class sample and each return series is regressed against this universal model. The universal model emphasizes the three factors’ explanatory abilities across asset class boundaries.
The three factors explain excess returns in the four asset classes under study. There is also a more universal price underreaction (momentum) and overreaction (value) that explains returns beyond traditional asset class boundaries. Supporting evidence for universal yield underreaction (carry) is less consistent however. The found evidence highlights two things: The statistical importance and independence value, momentum and carry; The need to shift portfolio allocation focus from asset classes to factors since factors drive returns in every asset class under study.
The three factors’ distribution properties suggest that momentum and carry factor premiums are compensation against rare disaster risk while the risk is unable to explain value premium. All three premiums could however exist in the presence of market frictions.