A Case Study of Portfolio Construction Using the USER Data and the Barra Aegis System

In this chapter, we estimate a set of monthly regression models to create monthly expected returns and demonstrate the effectiveness of the Barra Aegis system. The Aegis system creates and tests investment management strategies, producing portfolios and attributing portfolio returns according to the Barra multifactor risk model. We find support with the Barra Aegis for the composite modeling, the United States Expected Returns (USER), developed and estimated in Chap. 4, using fundamental, expectations, and momentum-based data for the US equities during the December 1979–December 2009 period. To measure risk, one can vary the period of volatility calculation, such as using 5 years of monthly data in calculating the covariance matrix, as was done in Bloch et al. (1993), or 1 year of daily returns to calculate a covariance matrix, as was done in Guerard et al. (1993), or 2–5 years of data to calculate factor returns as in the Barra system, discussed in Menchero et al. (2010). The Capital Asset Pricing Model, the CAPM, holds that the return to a security is a function of the security beta: $$ {R_{jt }} = {R_{\mathrm{ F}}} + {\beta_j}\left[ {E({R_{\mathrm{ Mt}}}) - {R_F}} \right] + {e_{jt }}, $$