Executive pension, default risk, and earnings management^*

* Accepted by Jeong-Bon Kim upon recommendation by Cheong H. Yi.

Abstract

The value of executive pension plans depends significantly on the incidence of bankruptcy because executive pension plans have characteristics similar to unsecured debt. These unique characteristics lead us to investigate whether managers change their firms’ accounting policy to protect their pension plans when their firms face imminent default risk. Identifying a firm’s default risk with various proxies, we find that managers with executive pension plans are more likely to engage in income-increasing earnings management during a year of high default risk as compared to managers without such pension plans. The result remains robust with the use of propensity score matching and two-stage least squares regression analysis to alleviate the endogeneity issues in our hypothesized relationships. In addition, we find that the results are more pronounced for ex-post bankrupt firms.

Keywords:

• Executive pensions
• inside debt
• default risk
• earnings management

View correction statement:

Erratum

Corrigendum

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

* Accepted by Jeong-Bon Kim upon recommendation by Cheong H. Yi.

1. Our sample period begins from 2006 because executive pension data on ExecuComp is available from 2006. The Securities and Exchange Commission (SEC) reformed the regulation that forces firms to disclose the present value of pension benefits and other forms of deferred compensation in proxy statements starting in 2006. ExecuComp began to tabulate the information starting in 2006. Also, the Bankruptcy Abuse Prevention and Consumer Protection Act, which severely restricts the payment of executive pension to retired and current managers, was implemented in 2005. Based on these two regulatory changes, we decided to establish the start of our sample period as 2006.

2. Once again, we also use Ohlson’ O-score, two-year consecutive losses, and dividend cuts as alternative measures for financial distress. The results using alternative measures are reported in Table 8.

3. As mentioned in DeAngelo, DeAngelo, and Skinner (1994), this sample restriction may cause selection bias, which limits the generality of our inferences. Specifically, since financially distressed firms are more likely to have negative accruals than others, those firms will naturally tend to engage in income-decreasing earnings management. To alleviate this selection bias issue, we also test the full sample without any restrictions and report the results in Table 4, along with the results for the restricted sample.

4. We selected five-years following García Lara, Osma, and Neophytou (2009). In addition, we alternatively identify a firm’s ex-post bankruptcy using one to four years. These alternatives yield qualitatively similar results as those using five years.

5. As both Chapter 7 and 11 is a way of bankruptcy under the United States Bankruptcy Code, Chapter 7 permits only liquidation to the bankruptcy filing firms while Chapter 11 allows reorganization to them. Whatever code a bankrupt firm is filing for, the firms are prohibited severely from paying executive pension to its manager.

6. As a sensitivity test, we also use discretionary accruals from the typical version of modified Jones model which does not include lagged ROA as our dependent variable. The untabulated regression results using original version of modified Jones model remain qualitatively the same with our main regression results in Table 4. Also, we alternatively use the typical annual cross-sectional Jones model and performance-matched Jones model. The untabulated results are also qualitatively similar to those in Table 4.

7. Additionally, we also control for other types of managerial compensation, such as stocks, options, and cash. Many studies argue that different types of managerial compensation provide managers with incentives to engage in earnings management under certain situations. Since adding this control yields results that are qualitatively equivalent, they are not reported here for brevity.

8. The ratio is smaller than 51%, as reported in Bebchuk and Jackson (2005), and 78%, as reported in Sundaram and Yermack (2007). The difference may arise because we include middle- and small-sized firms from the ExecuComp database, whereas Bebchuk and Jackson (2005) examined S&P 500 and Sundaram and Yermack (2007) examined Fortune 500 firms.

9. To alleviate the concern over this multicollinearity problem, we check the variance-inflation-factors (VIFs) for STDCFO and STDREV in our multivariate tests, respectively. The VIFs for STDCFO and STDREV are all greater than 5.00, which is high enough to raise a serious multicollinearity issue. Therefore, we perform a sensitivity test by dropping either STDCFO or STDREV and find that the main findings in Table 4 still hold in the sensitivity test.

10. A firm can endogenously select whether to provide executive pensions, while a firm’s financial distress is not endogenously selected but ‘exogenously’ determined. Therefore, we assume that financial distress itself does not cause an endogeneity problem.

11. The untabulated results using pension amounts variable are very similar to those using pension dummy variable in Table 6.

12. O-score is calculated from the formula for predicting bankruptcy (Ohlson 1980) at the end of year t as follows: −1.32 − 0.407 (Log Total Assets) + 6.03 (Total Liabilities/Total Assets) – 1.43 (Working Capital/Total Assets) + 0.076 (Current Liabilities/Current Assets) – 1.72 (1 if Total Liabilities > Total Assets, 0 otherwise) – 0.521 ((Net Income − Net Income − 1)/(|Net Income| + |Net Income − 1|)).