Regulatory Scrutiny, Labor Unions, and Pension Freezes

Editor and Co-authors' note: Upon review and with minor revisions, this research paper was accepted for publication soon before the premature death of Professor John Kim in April 2015. We dedicate this paper to his honor.INTRODUCTIONIn recent years, there has been a shift in the private pension system from the traditional defined benefit pension (hereafter DBP) plan to more volatile markto-market pension accounting. DBP plans guarantee defined benefit, which means that even if market fluctuations cause the value of plan assets to fall short of the pension obligation, the firm is financially responsible for the shortfall, or underfunded amount. Recent pension rules SFAS 158 and the Pension Protection Act (hereafter PPA) of 2006 require that companies with underfunded plans increase their pension contributions and disclose the pension liability on the balance sheet. Since the value of equities in private DBP plans decreased by nearly $1 trillion during 2007-2008 (Munnell et. al., 2008), many companies are likely grappling with underfunded pensions and, after the 2006 pension rules, increased contributions and increased disclosure.Prior studies have examined the freeze decision around regulation changes. Comprix and Muller (2011) find that prior to Sarbanes-Oxley Act of 2002, managers opportunistically bias pension estimates to obtain labor concessions by exaggerating the economic burden of their benefit plans. On the other hand, Beaudoin, et. al. (2010) examine whether freeze announcements of DBP plans made during 2001-2006 were motivated by accounting concerns due to the pending adoption of SFAS 158. They suggest that firms' freeze decisions are significantly associated with the potential SFAS 158 impact.Although these studies provide useful insights into the factors affecting the freezing decision of defined benefit plans, none of them examines the absence of any incentives to adopt downward biased pension estimate for unionized firms when freezing their DBP plans after the adoption of SFAS 158 and the PPA of 2006.Our overall goal is to compare firms' freeze choices before and after the new pension rules of 2006. We obtained data from the 5500-CRR database (Form 5500 Annual Reports Data Base compiled by Center for Retirement Research in Boston College) and Compustat to identify firms with DBP plans with labor unions (i.e., collective bargaining agreements) during 2004 to 2008. Note that 2008 is the most recent year with available data (as of 2014) from the Center for Retirement Research at Boston College, which is attributed to a two-year lag between the firm's fiscal year and the year the 1RS filings are accepted and an additional lag for the data to be compiled by the center.Our study makes several contributions to the literature on pension plans. First, we provide evidence on how firms respond to the adoption of SFAS 158 and the PPA of 2006. Second, we are the first to document that although unionized firms are overall less likely to freeze pension plans, these firms will cancel plans that are underfunded. Last, in contrast to the findings of previous research, we show that that financial health is not the sole determinant of the freeze decision.The remainder of this study contains the following: Section 2 consists of motivation and hypotheses development; Section 3 presents data sources and sample description; Section 4 describes the empirical model and results; Section 5 offers a summary and conclusion.MOTIVATION AND HYPOTHESES DEVELOPMENTIn this section, we discuss how DBP plans have slowly disappeared over time, the decision to freeze DBP plans, and how firms' labor union status potentially affects these DBP events.Two recent empirical studies have focused on DBP plans. Comprix and Muller (2011) examine pension estimates and the decision to freeze DBP plans around The Sarbanes-Oxley Act of 2002. The authors suggest that, when freezing their DBP plans, employers select downward biased accounting assumptions to exaggerate the economic burden of their benefit plans. …