Pension Deficits and Corporate Financial Policy: Does Accounting Transparency Matter?

Abstract

We study changes in financial policies following a regulatory shock to the accounting transparency of defined benefit pension plans. We estimate the hidden pension deficits of French companies subject to mandatory IAS 19 adoption in 2005 using disclosures of early adopters of IAS 19. We find that financially risky companies reporting unexpectedly high pension deficits on first-time IAS 19 adoption subsequently reduce leverage and incur higher cost of debt. Our results suggest that in the absence of transparency the credit market anticipates off-balance sheet pension deficits. However, the introduction of the more transparent IAS 19 regime allows the credit market to correct estimation errors. Our study is one of the first to show that the greater transparency offered by IFRS has negative economic consequences for some companies.

Keywords:

• Accounting transparency
• Defined benefit pension plans
• Capital structure
• Off-balance sheet liabilities

JEL codes:

• G32
• M41

Acknowledgements

We are grateful to Paul Zarowin (associate editor) and two anonymous reviewers for valuable feedback and suggestions. We also thank Ulf Brüggemann, Stefano Cascino, Mark Clatworthy, Panagiotis Couzoff, Christina Dargenidou, Joanne Horton, Ursa Kosi, Bart Lambrecht, Christian Leuz, Gilad Livne, Tim Marklew, Paul Metcalf, Giovanna Michelon, John O’Hanlon, Per Olsson, Bill Rees, Steven Young and participants at Lancaster University and ESSEC Business School seminars, Varna INTACCT workshop, 27th EAA Doctoral Colloquium, and 34th EAA Annual Congress for helpful comments on previous versions of this paper and Mahmoud El Haj and Azeddine Elhanaoui for providing excellent research assistance in relation to extracting pension disclosures from French annual reports. Paraskevi Vicky Kiosse gratefully acknowledges financial support from Exeter University Business School. Fani Kalogirou gratefully acknowledges the support from FCT – Portuguese Foundation of Science and Technology for the project ‘UID/GES/00407/20013’.

Supplemental Data and Research Materials

Supplemental data for this article can be accessed at https://doi.org/10.1080/09638180.2020.1792321

Appendix A – Regulatory Setting and Pension Disclosures in France.

Appendix B – Additional Analysis.

Notes

1 Based on our original sample of non-financial, non-utility companies, before deleting observations for missing non-pension related variables.

2 A study showing greater transparency has negative consequences is that by Callahan et al. (2012) who find the mandating of Variable Interest Entities (VIEs) consolidation resulted in an increase in the cost of capital of affected companies.

3 Age is used as a proxy for the age of the pension plan, in the absence of more specific information to calculate the maturity of pension plans.

4 Source: Institut National de la Statistique et des études économiques, Annule et remplace le N° 1272 – Décembre 2009.

5 Entropy balancing has certain advantages over one-to-one propensity score matching, including a higher degree of covariate balance and the prevention of information loss owing to the retention of all observations (Hainmueller, 2012).

6 Note that in the transition year, mandatory adopters must restate their financial statements from the previous fiscal year (i.e., 2004) to reflect the change in accounting standards.

7 Only a small sub-set of the companies in our sample had traded bonds and hence a meaningful test using market data is not feasible. Thus, we follow prior literature, which in the absence of market data uses this variable to proxy for the corporate cost of debt (see for example Pittman & Fortin, 2004).

8 We focus on sales in order to proxy for the marginal tax-shield benefit before pension contributions and debt interest.

9 We thank the anonymous reviewer for suggesting to include this variable.

10 The results are similar when using industry median or sample mean/median. In addition, the results are similar when using dividend levels and the Whited-Wu index to classify companies. The Whited-Wu index is calculated based on the following model: −0.091CF_it−0.062DIVPOS_it + 0.021TLTD_it−0.044LNTA_it+ 0.102ISG_it−0.053SG_itwhere CF_it is the ratio of cash flows to total assets, DIVPOS_it is an indicator variable taking the value of 1 if the company pays cash dividend, TLTD_it is the ratio of long-term debt to total assets, LNTA_it is the natural logarithm of total assets, ISG_it is the industry sales growth, and SG_it is the company’s sales growth (Whited & Wu, 2006).

11 This includes companies that do not have DB plans and companies that do not publish consolidated accounts and thus are not required to apply IFRS.

12 While relevant pension data are available in the Worldscope database, coverage is limited. We therefore hand-collect the data required, which results in a larger sample size. For example, the study by Bartram (2016) examining a different research question includes 154 French companies based on a Worldscope sample, whereas we are able to collect data for 348 companies.

13 Our final sample is statistically similar to the non-financial French listed companies dropped from the sample in terms of leverage (0.22 versus 0.21) and total book value of assets (2.9 billion versus 2.5 billion). However, the companies in our sample are larger in terms of market capitalization (2 billion versus 1.16) and more profitable (0.05 versus 0.03).

14 Of these 177 companies, 123 already reported information under IAS 19 in fiscal year 2003 and the remaining 54 companies were first-time voluntary adopters in 2004.

15 We replace missing R&D data with zero, based on the requirement in IAS 38 that companies should report material R&D expenditure and the discussion in Heitzman et al. (2010).

16 Unfortunately, even after the mandatory adoption of IAS 19 and the recognition of the pension plan deficit on the balance sheet, many companies continue not to disclose the amount of their projected pension obligation. Hence, while the net amount is reported there is not always information about gross pension assets and liabilities. We thank an anonymous reviewer for prompting us to clarify this point. For the sub-set of companies providing this information, we find that on average the level of unfunded pension liability relative to the projected benefit obligation (Unfunded%) is economically and statistically different between mandatory and early adopters. More specifically, our univariate comparison of the two groups suggests that companies with relatively well-funded pension schemes may opt to adopt IAS 19 and disclose information about pension obligations early (0.65 vs 0.73). Alternatively, this difference could be attributed to companies feeling under pressure to better fund their pension plans once they adopt IAS 19.

17 For example, Carrefour S.A. in its 2004 French GAAP balance sheet recognized a provision for pension liabilities of 179.1 million euro. In its restated 2004 IFRS balance sheet, it recognized a provision for pension liabilities of 732 million euro. In this case, the TRSurprise is 552.9 (732–179.1) million euro scaled by 2004 French GAAP total assets of 39,000 million euro, equal to 1.4%. Note that our measure of Surprise is equal to −1.6% (equivalent to an expected deficit of 1,100 million euro) suggesting that creditors were expecting an even higher IFRS pension deficit, based on the company’s characteristics and the information provided by early adopters.

18 Schneider Electric SA is one such case. Schneider Electric SA reported a pension deficit of 660.9 million euro (5.1% of total assets) in 2004 under French GAAP. In its IFRS adjusted statements however the pension deficit is reported to be 1,030 million euro (7.9% of total assets). Our prediction model yields an expected (Predicted) pension deficit of 470.8 million euro (3.6% of total assets). This Predicted value while lower than the reported deficit, it is still significantly higher compared to the industry (sample) average of 1.8% (1.5%). In this scenario, the Surprise variable will take the value of 4.3% (7.9%–3.6%) and RRSurprise will take the value of 0% instead of −0.8% ([7.9%–5.1%]–3.6%).

Additional information

Funding

This work was supported by the EU-funded INTACCT programme - The European IFRS Revolution: Compliance, Consequences and Policy Lessons (Marie Curie Actions - European Commission) (contract number MRTN-CT-2006-035850).