Adjusting for the intervalling effect bias in beta: A Test using Paris Bourse Data

This paper uses a sample of daily returns data from the Paris Bourse to test the predictions of and adjustment procedures implied by the Cohen-Hawawini-Maier-Schwartz-Whitcomb (1983a,b) model of the intervalling effect bias in OLS beta estimates. CHMSW show that the bias diminishes asymptotically to zero as the differencing interval increases and that the sign of the bias depends on a stock's relative value of shares outstanding. We employ the three-pass regression procedure of CHMSW (1983b), but we test a broader set of second pass functional forms and use the Box-Cox (1964) transformation model to verify our chosen form. The CHMSW adjustment procedures are compared with that of Scholes-Williams (1977), and the latter is shown to contain a significant intervalling effect bias.