Beta estimation in the European network regulation context: what matters, what doesn't, and what is indispensable

Most studies on beta estimation look at the whole universe of stocks. We focus on a small subset that consists of stocks of companies which are subject to European network regulation. This allows us to examine beta time series of individual stocks and small peer groups in great detail. Our most important conclusions are: (1) Sudden beta increases or decreases occur that often last only short periods of time and may therefore cause a significant misestimation of the future beta. (2) Three- and especially five-year betas are much more stable than one-year betas. (3) The choice between purely local, European or global betas may matter considerably. (4) Weekly or daily betas seem to be better than monthly ones. (5) Vasicek and Blume adjustments towards one lead to beta predictions that are too high.