AbstractIn testing moral hazard and tax benefit hypotheses regarding defined benefit plan funding and contribution incentives by incorporating sponsors' bankruptcy risk, the authors proposed that high-bankruptcy-risk sponsors have a strong moral hazard incentive because the put value of the U.S. Pension Benefit Guaranty Corporation guarantee is high. For low-bankruptcy-risk sponsors, the put value is low; maximizing tax benefits associated with pension contributions becomes a powerful incentive. Results based on sponsors' voluntary contributions support both hypotheses. Underfunding of corporate defined benefit (DB) pension plans has become a prevalent issue among U.S. companies amid the recent financial crisis. A key question that has attracted considerable research interest is what determines a DB plan sponsor's decisions on pension funding and contributions. Within one unified framework, the authors tested two hypotheses—one on the moral hazard incentive and the other on the tax benefit incentive—with respect to decisions on DB pension funding and contributions by incorporating sponsors' expected bankruptcy risk, as measured by Moody's EDF (expected default frequency). The incorporation of expected bankruptcy risk is critical in better understanding a sponsor's incentives because it relates directly to the put option value derived from U.S. Pension Benefit Guaranty Corporation (PBGC) insurance. The authors hypothesized that sponsors with high expected bankruptcy risk are dominated by the moral hazard incentive because the put option on the PBGC guarantee has the greatest value. In contrast, for sponsors with low expected bankruptcy risk, the PBGC put option value is low; maximizing tax benefits associated with pension contributions becomes a dictating incentive.Unlike earlier researchers who used total pension contributions, the authors decomposed total pension contributions into mandatory and voluntary contributions. They used voluntary contributions as a major measure for sponsors' incentives regarding pension funding and contributions. Using IRS Form 5500 data for 1990–2010, they calculated sponsors' voluntary pension contributions on the basis of applicable pension laws and regulations. The results based on sponsors' voluntary contributions are consistent with the two hypotheses after controlling for potential endogeneity. In particular, sponsors with high expected bankruptcy risk make low voluntary contributions; for those with low bankruptcy risk, voluntary contributions increase with the marginal tax rate. Using the recent financial crisis as a natural experiment on the effects of sponsors' expected bankruptcy risk, the authors found that both moral hazard and tax benefits have intensified since the crisis.Suggesting that the existing pension regulations do not successfully reduce sponsors' moral hazard incentive, the authors discuss three important policy implications: (1) The PBGC premium structure should fully reflect the bankruptcy risk that a plan sponsor poses to PBGC, (2) the PBGC claims on unfunded pension liabilities during corporate bankruptcy proceedings should be more strictly enforced, and (3) policymakers should address sponsors' moral hazard incentive as a critical issue when designing pension regulations to ensure the soundness of the U.S. private pension system. Yu and Zhang gratefully acknowledge a research grant from Netspar, an independent network for research on pensions, aging, and retirement in the Netherlands. For helpful comments and suggestions, we all thank Monika Bütler, Gaobo Pang, and participants at the 2010 Netspar International Pension Workshop. We thank Moody's KMV for providing the KMV EDF measures and the Center for Retirement Research at Boston College and the U.S. Department of Labor for providing IRS Form 5500 data. We also thank John Graham for sharing data on corporate simulated marginal tax rates and Barry Hirsch and David Macpherson for making the Union Membership and Coverage Database available on their website. We especially thank Chris Baum, Douglas Dwyer, Kelly Haverstick, Beatty Patrick, Mauricio Soto, Vlada Stein, Jing Zhang, and Sue Zhang for their help with the data used in this study.Notes1 Boeing contributed a total of $3.6 billion to its defined benefit pension plans in 2004. If we assume a tax rate of 30%–35%, Boeing received a tax benefit of $1.1 billion to $1.3 billion from its pension contributions; see Karen Richardson, "How Companies Make the Most of Pensions," Wall Street Journal (24 January 2005):R3.2 See www.pionline.com/section/surveys.3 If a sponsor eventually emerges from bankruptcy but ends up with a severely underfunded pension plan by betting on high pension investment risk, it must continue to fund its pension plan with valuable financial resources. Therefore, CitationRauh (2009) posited and reported that sponsors are dominated by the risk management incentive and make less risky investments when their pension plans are severely underfunded.4 EDF is a trademark of Moody's.5 If an insolvent sponsor's pension assets are short of pension obligations, PBGC can take over the pension plan and guarantee benefits up to a certain limit (in 2012, $55,840 for employees retiring at age 65). In addition, PBGC is authorized to place liens on a sponsor's assets up to 30% of net asset value.6 The CRR Form 5500 data end in 2007. We obtained Form 5500 data for 2008–2010 from the Department of Labor.7 Available at http://faculty.fuqua.duke.edu/~jgraham/taxform.html.8 The data are maintained by Barry Hirsch and David Macpherson and are available at www.unionstats.com; see CitationHirsch and Macpherson (2003) for a detailed discussion of the construction of the dataset.9 GMM generates efficient and consistent estimates in the presence of heteroscedasticity (CitationBaum, Schaffer, and Stillman 2003). Heteroscedasticity is a concern especially when the explanatory variables in a regression model are highly correlated. In our regression in EquationEquation 2VCTi,t=α+β1EDFi,t−1+β2TAXi,t−1+β3PROFITi,t−1+β4CFi,t−1+β5FRi,t−1+β6LNASSETi,t−1+β7TOBINi,t−1+β8UNIONi,t−1+ϵi,t.(2) , EDF, PROFIT, CF, and FR are intercorrelated. In untabulated results, we found a negative correlation between EDF and PROFIT (ρ = –0.28), suggesting that high bankruptcy risk is related to poor operating performance. The correlation between PROFIT and FR is 0.29.10 According to a news release on 27 January 2012 from the U.S. Bureau of Labor Statistics, the union membership rate decreased from 20.1% in 1983 to 11.8% in 2011 (www.bls.gov/news.release/union2.nr0.htm).11 In performing the endogeneity test here, instead of separating the total sample into high-, medium-, and low-EDF groups (as we did with EquationEquation 2VCTi,t=α+β1EDFi,t−1+β2TAXi,t−1+β3PROFITi,t−1+β4CFi,t−1+β5FRi,t−1+β6LNASSETi,t−1+β7TOBINi,t−1+β8UNIONi,t−1+ϵi,t.(2) ), we used the dummy variable EDF_LOW so that we would not lose about two-thirds of the observations.12 According to the Congressional Research Service report to Congress on the PPA (2006, p. 2), the worst-case scenario assumes that "(1) the employer is not permitted to use credit balances to reduce its cash contribution and (2) employees will retire at the earliest possible date and will choose to take the most expensive form of benefit. If a plan does not pass this test, it will be deemed to be at-risk unless it is at least 80% funded under standard actuarial assumptions."13 PBGC has recently become more active in filing claims on unfunded pension liabilities and in preventing unnecessary plan terminations in bankruptcy court. According to PBGC's 2011 annual report (www.pbgc.gov/documents/2011-annual-report.pdf), it opened 56 bankruptcy cases in 2011 and worked with 19 companies in bankruptcy to continue their pension plans. A recent example is AMR Corporation (American Airlines' parent company). AMR filed for bankruptcy in November 2011 and announced its intention to terminate its pension plan, shifting more than $10 billion of unfunded liabilities to PBGC. PBGC filed more than 70 liens, totaling $91.7 million, on American Airlines' assets in Texas and outside the United States. AMR reversed course in March 2012 by freezing, rather than terminating, its pension plan at current levels, which means the company remains liable for its pension obligations (CitationWalsh 2012; CitationWalsh and Mouawad 2012).