This study, which analyzes stock returns associated with the announcement to terminate overfunded defined-benefit pension plans, differs from prior research by considering both the role of the replacement plan and prior anticipation of the termination by shareholders. These innovations enable us to (1) evaluate the use of the reversion amount (plan assets minus termination liabilities) as a measure of wealth transfer from plan participants to shareholders, and (2) resolve inconsistencies across prior studies regarding the existence of a significantly positive market reaction at the termination announcement. This study provides insight into the controversy over the appropriate measure of pension liability: accumulated benefit obligation (ABO) versus projected benefit obligation (PBO). PBO is based on expected future salaries, while ABO is based on current salaries. Barth (1991, 457) suggests that investors consider expected salaries but view PBO as noisy. The current study complements this finding. By examining stock returns at the time of pension plan termination-when any uncertainty regarding the effects of future salaries on pension liability is resolved-we are able to assess whether investors use future salaries to value pension liabilities prior to termination. In addition, the study describes institutional features of pension plans and illustrates the Lanen and Thompson (1988) result that market anticipation may attenuate the relation between the market reaction to and the cash flow effects of an event. We hypothesize that the change in stock price at the time of termination is directly related to the wealth transfer from plan participants to shareholders. If defined-benefit coverage is continued after termination, the primary wealth transfer is from retired participants because annuity contracts purchased from insurance companies typically do not allow for cost-of-living adjustments (COLAs). If defined-benefit coverage is canceled, retired workers lose COLAs, and current workers lose the difference between pension liabilities based on future salaries and those based on current salaries. The results are presented in a descriptive, systematic fashion in order to properly associate any market reaction with underlying features of termination and to ascertain the cause of inconsistencies across previous studies. An event-type methodology is employed by using daily CRSP data. Department of Labor Form 5500 data are used in conjunction with disclosures about the nature of the replacement plan to estimate worker losses at the time of termination. In addition, estimates from a censored bivariate probit model are incorporated to control for market anticipation of the termination. The results suggest that stock returns at the time of termination are affected by the form of the replacement plan and, to a lesser degree, market anticipation. The results are consistent with investors' using future salaries to value pension liabilities prior to termination. Analysis of the replacement plan and comparison of the reversion amount to the theoretically derived wealth transfer suggest that inconsistencies across prior studies principally result from including firms that continue defined-benefit coverage after termination. In addition, the results help explain why previous studies documented small changes in market value relative to the reversion amount.
