Using Stock Return Data to Measure the Wealth Effects of Regulation: Additional Evidence from California's Proposition 103

Article excerpt

Introduction

On November 8, 1988, despite intense insurance industry lobbying efforts, California voters approved Proposition 103, radically reforming the state's insurance laws. Among the referendum's more controversial provisions was a 20 percent reduction in property-liability (notably auto) insurance rates. Proposition 103 outlined which risk factors could and could not be used to underwrite auto insurance. Geography could no longer be used as a primary rating factor.(1) Instead, drivers' records, miles driven, and years of driving experience became the primary determinants of insurance premium rates. Insurers were also required to lower insurance premium rates an additional 20 percent for good drivers. No other lines of insurance received such specific attention.

Within two days of the vote, the California State Supreme Court issued a temporary staying order, delaying implementation of the measure. This stay was ultimately overturned, and the elected California insurance commissioner directed automobile insurers to pay rebates in 1993. Consequently, this act has had an impact on the operating profits of California's insurers. This article examines whether the market correctly anticipated Proposition 103's impact on California's property-liability insurers. We also examine whether Proposition 103's impact varies by firm-specific characteristics.

Previous stock market-based research by Fields et al. (1990), Szewczyk and Varma (1990), and Shelor and Cross (1990) on the wealth effects of Proposition 103 report conflicting conclusions. Both Fields et al. and Szewczyk and Varma find a significant negative impact on insurers with high California exposure. In addition to a time-series event methodology, Fields et al. employ cross-sectional regression and find that insurer negative abnormal stock returns are proportional to California exposure. Both Fields et al. and Szewczyk and Varma also conclude that the California reform represents a leading indicator of other states' likely reforms, supported by stock market evidence. By comparison, Shelor and Cross report no significant negative impact of Proposition 103 on California insurers but a significant negative impact (-2.87 percent) on non-California insurers on election day and the following day. These authors offer size effects as an explanation; large insurers were able to diversify away adverse effects of regulation while small non-California insurers were unable to do so.

This article uses the stock market to measure the wealth effects of California Proposition 103. The article differs from previous research in several respects. First, we examine the abnormal performance of both equally-weighted and value-weighted portfolios of exposed and nonexposed securities.(2) Portfolios are drawn from the sample of firms used by Fields et al. and also from a somewhat different sample. Results from the value-weighted portfolios indicate that Proposition 103 had little impact on the aggregate (dollar) value of the insurance industry.

Second, we compare the security return behavior of five firms (Chubb, Fireman's Fund, Ohio Casualty, Safeco, and Zenith National) mentioned in the Wall Street Journal's Heard on the Street (HOTS) column on the day before the election to the stock return behavior of four other firms not mentioned in HOTS but that had roughly comparable California exposure. Thus, the relative performance of these stocks can shed some light on market efficiency.

Third, to gauge the impact of this referendum and the subsequent court decision to delay implementation, we estimate a cross-sectional regression in value terms; that is, we regress the abnormal change in (dollar) value of the firm's equity against direct premiums written in automobile and other insurance lines in estimating relative exposure to Proposition 103.

Theory and Hypotheses

We test three hypotheses concerning the stock market reaction to Proposition 103: the contagion effect hypothesis, the information effect hypothesis, and the dribs and drabs hypothesis. …