Authors: Patrick L. Brockett (firstname.lastname@example.org), Hwei-Mei CHEN and James R. GARVEN (James_Garven@baylor.edu) Insurance Mathematics and Economics, Vol. 25, No. 2 (November 1999), pp. 197-216.
Abstract. A number of articles have documented that the classical event study methodology exhibits a bias toward detecting "effects", irrespective of whether such effects actually exist. This paper addresses this bias by presenting a new methodology that explicitly incorporates stochastic behaviors of the market that are documented to exist and which are assumed away by the classical event study methodology. We apply our new methodology to an examination of the effect of the passage of California’s Proposition 103 on the prices of insurance stocks. Proposition 103 was important regulatory event that previously has been investigated using classical event study techniques. We find that the passage of Proposition 103 did not significantly impact the returns on most insurance company stocks, a result that stands in stark contrast to other studies. Consequently, our study suggests that the application of the classical event study methodology, without checking the behavior of security returns for stochastic beta and GARCH effects, may vwery well cause researchers to draw inappropriate conclusions.
Keywords: event study methodology, ARCH, GARCH, cumulative sums, Proposition 103.