Asset Allocation

(1) Specify the expectations needed, given the client's horizon and asset classes; (2) research the historical record; (3) specify the methods and models; (4) determine information sources; (5) interpret the current environment; (6) provide the expectations and document conclusions; (7) monitor outcomes and feed lessons back into the process.

Strategic asset allocation needs forward-looking estimates of each asset class's expected return, volatility and correlations; weak or inconsistent CME inputs flow straight into the long-term policy mix, so errors in expectations become errors in the allocation.

Errors introduced when data are gathered, recorded or copied (typos, misclassification, wrong adjustments), so the series used for estimation does not match what actually happened.

Databases that drop failed or closed funds keep only the survivors, which tend to have better records, so average historical returns are biased upward and risk is understated.

Appraisals are infrequent and anchored to prior valuations, which smooths reported returns; the smoothing lowers measured volatility and correlations with public markets, making the asset look less risky and more diversifying than it is.

The series is nonstationary: its mean, volatility or correlations differ across regimes, so pooled estimates may describe no regime well. Using a longer history improves precision only if the underlying parameters were stable.