Plan of the chapter

This chapter naturally splits into two parts, which address related but different concerns.

In the first part we look at the assignment of the expected returns obtainable from the spliced distribution of the various risk factors. Assigning this input is delicate and affects greatly the optimal allocation, as we discuss in Chapter 28. We provide a ‘model-assisted’ answer to this problem. We say ‘model-assisted’ instead of ‘model-based’, because the expected returns implied by our subjective views will be guided, but not dictated, by an asset-pricing model.

The model we use is the CAPM. We choose to work with it not because we believe that it is a normatively ‘perfect’ model, but because that it can provide a useful sanity check that our return expectations are at least consistent (say, in terms of ranking) and that they reflect a plausible trade-off between return and risk (with risk understood as variance and covariance).

The treatment in the first part of the chapter almost completely disregards information from market prices. What we are trying to achieve is a degree of model-guided internal self-consistency.

Observed market prices come to the fore in the second part of the chapter (Sections 21.6–21.9). Here we assume that the consistency checks in the first part of the chapter have been carried out to our satisfaction, and we try to ascertain whether, given our views, a security is ‘cheap’ or ‘dear’.