Troutman, Michael. "Risk Control Techniques for Social Investing." Journal of Investing, Winter 2001.
Clear and useful rundown of the risk impact of commonly-applied social screens on diversified plan sponsor portfolios.
1) A "weapons & harmful products" screen eliminated 135 companies from the Wilshire 5000. A portfolio composed of the remainder had a beta of 0.998 and an r2 of 0.995 vs. the benchmark. The "energy & environment" screen excluded 80 companies, and resulted in a portfolio with a beta of 1.01 and r2 of 0.994. Combining the screens, 204 companies were excluded and the resulting portfolio had a beta of 1.015, and an r2 of 0.993.
2) The Wilshire multiple factor equity model forecasted that removing the 204 stocks would create an 89 basis point annual tracking error relative to the Wilshire 5000.
3) Reports that one large institutional portfolio, socially screened and passively managed to match the Russell 300 index for the 1991-2000 time period, returned 17.43% per annum (standard deviation of 13.31%) vs. 17.38% (standard deviation of 13.39%) for the unscreened benchmark. Tracking error was consistent with the factor model forecast: "The standard deviation of the return difference between the actual portfolio and the index was 98 basis points."
LK comment: At the time this paper was written the author served in a senior investment role for the Evangelical Lutheran Church in America.