For many small cap assets, using historical data to estimate the beta, results in a beta close to zero. Why is this? Briefly explain, how a fund manager should deal with this issue if (s)he plans to use a single index model to estimate a group of assets variance-covariance matrix.