Various risk and return models is used to compute the cost of equity but they all assume that the marginal investor is well diversified. If you use these models to estimate cost of equity for private or closely held firms, you are likely to under or over estimate the cost of equity. How would you fix the bias?