Equities Risk

More details on statistical factor mimicking portfolios

Synopsis

Here we present more details on the return characteristics of the factor mimicking portfolios designed to capture the influence of statistical variables. The portfolios are formed every December by sorting the universe of All Ordinaries Index stocks into terciles based on their loadings on the variable of interest. Each stock's loadings are calculated via exponentially-weighted moving average time-series regressions over the previous five years with decay parameter λ = 0.95. The factor mimicking portfolio returns are the monthly return spread between the top tercile portfolio and the bottom tercile portfolio. Relative volatility is the ratio of the factor mimicking portfolio volatility to the volatility of the return spread of randomly assigned tercile portfolios.

Table 1. Return characteristics of statistical factor mimicking portfolios.
Factor	Minimum	25th percentile	75th percentile	Maximum	Standard deviation	Relative volatility
Principal component 1	-0.202	-0.035	0.036	0.193	0.062	2.902
Principal component 2	-0.131	-0.039	0.033	0.193	0.054	2.519
Principal component 3	-0.101	-0.025	0.022	0.105	0.035	1.641
Principal component 4	-0.094	-0.017	0.021	0.107	0.032	1.507

Factor mimicking portfolio total return indices

Factor mimicking portfolios based on first principal component — Figure 1. Total return indexes of factor mimicking portfolios based on principal component 1.

Factor mimicking portfolios based on second principal component — Figure 2. Total return indexes of factor mimicking portfolios based on principal component 2.

Factor mimicking portfolios based on third principal component — Figure 3. Total return indexes of factor mimicking portfolios based on principal component 3.

Factor mimicking portfolios based on fourth principal component — Figure 4. Total return indexes of factor mimicking portfolios based on principal component 4.

Principal components 1-4 are the loadings on the first four principal components estimated using the Asymptotic Principal Components method of Connor and Korajczyk (1988), Risk and Return in an Equilibrium APT: Application of a New Test MethodologyJournal of Financial Economics, 21(2), 255–289.