A client preference about how money is invested arrives at this node as a constraint on the feasible set, which is a form the machinery of constrained mean-variance optimization already handles. That section priced a 50% equity cap in Sharpe-ratio terms; the same method prices a screen.
What are the main ESG investing strategies?
Six approaches are used, and they are not variations on one idea.
Negative or exclusionary screening removes an industry, an activity or a set of issuers. Positive or best-in-class screening keeps every industry but tilts within it toward higher-rated issuers, which leaves the sector allocation roughly intact and therefore costs less. Thematic investing buys an exposure defined by a theme rather than by a sector, and it is a concentrated active bet whether or not it is presented as one. Impact investing targets a measurable non-financial outcome alongside a financial one, usually in private markets, and is priced there. Engagement and stewardship buys the holding and uses the vote and the meeting rather than the sell button, which changes no allocation figure at all. Integration folds material sustainability factors into the security-level forecast, so it changes expected returns rather than the feasible set; EC1.07 owns that.
Only the first three change the frontier. The last three change either governance or the inputs.
Advertisement
How much does an ESG screen cost?
Price a screen on Bharat Family Office's own numbers and the argument stops being abstract. Assume the client excludes an industry group that makes up part of the domestic equity index. The screened equity sleeve is not the index any more, so model it honestly with two effects rather than one: expected return falls by 30 basis points to 11.70%, and volatility rises by one point to 20%, because a screened portfolio is less diversified than the index it came from. Everything else in the capital-market assumptions table is unchanged.
Recomputing the unconstrained tangency portfolio gives a Sharpe ratio of 0.5533, against 0.5780 unscreened, so the constraint costs 0.0247 of Sharpe ratio, or 4.27% of it. Two readings of that number make it useful. At the tangency portfolio's volatility of 6.2966%, a Sharpe fall of 0.0247 is 0.1555 percentage points of expected annual return, which is the form a trustee can compare with a fee. And 0.5533 is almost exactly the 0.5535 that constrained mean-variance optimization computed for the 50% equity cap, so on these inputs this screen costs the client about what the cap costs, which is a comparison worth putting in front of them.
Decompose the 0.0247 and the client can choose which half to negotiate. Cutting the equity return by 30 basis points on its own takes the Sharpe ratio to 0.5665, a cost of 0.0115. Raising equity volatility by one point on its own takes it to 0.5637, a cost of 0.0143. The volatility effect is the larger of the two here, which is the argument for a best-in-class screen over an exclusionary one: keeping every sector at a reduced weight preserves most of the diversification and gives up less of it than dropping a sector entirely.
What are the limits of this cost estimate?
Three honest limits. The 30 basis points and the one point are assumptions, not estimates, and the whole answer scales with them; a real mandate needs them measured on the actual screened universe, which is a tracking-error exercise. The screen may also raise returns if the excluded exposure underperforms, and nothing here forecasts that in either direction. And the cost computed is a cost against an unscreened optimum that the client has already declined, so the right way to present it is as the price of a preference the client holds, not as evidence the preference is a mistake.
Two neighbours own the rest of this. AA1.01's job ends at the allocation-level constraint; PW1.01 owns how a client preference becomes an enforceable line in an investment policy statement and what reporting the client should get, and EC1.07 owns security-level integration and the governance factors that drive it.