Political Uncertainty and Accounting Conservatism

Abstract

Political uncertainty leads to greater information asymmetry among contracting parties to the firm, resulting in an increased demand for accounting conservatism. Exploiting the exogenous variation in political uncertainty induced by the U.S. gubernatorial election cycle over the period 1963–2016, we find that the asymmetric timeliness of news recognition increases with political uncertainty. Our political uncertainty hypothesis operates through the contracting demand channel. Accordingly, we find that the political uncertainty effect is more pronounced for firms in states with lower electoral participation, for firms with greater industry exposures to contracting needs, for firms with higher leverage and lower managerial ownership, and for firms with stronger internal corporate governance mechanisms.

Keywords:

• Accounting conservatism
• Political uncertainty
• Contracting demand

JEL codes:

• G38
• M41

Acknowledgements

We thank two anonymous referees and Beatriz Garcia Osma (editor) whose constructive comments have substantially improved this paper. We are also grateful for helpful comments from Douglas DeJong, Gilles Hilary, Yaxuan Qi, Wim Van der Stede, David Veenman, Terry Walter, Mark Wilson, and participants at the Australian National University accounting seminar, and at the 2013 American Accounting Association Conference in Anaheim CA. All errors are our own.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplemental Data and Research Materials

Supplemental data for this article can be accessed at https://doi.org/10.1080/09638180.2020.1760117.

This Internet Appendix (IA) provides additional tables for “Political Uncertainty and Accounting Conservatism.”

We summarize the content as follows:

• Table IA1: Variable definitions and data sources in the Internet appendix

• Table IA2: Additional tests controlling for industry effects on conditional accounting conservatism

• Table IA3: Additional tests based on other earnings components

• Table IA4: Election cycle and conditional accounting conservatism

• Table IA5: The effect of political uncertainty on conditional accounting conservatism and exposures to labor and international trade

• Table IA6: Alternative model specifications of conditional accounting conservatism

• Table IA7: Political uncertainty and conservatism proxies capturing both unconditional and conditional accounting conservatism

• Table IA8: Political uncertainty and conditional accounting conservatism measured by spreads in conditional variance of earnings and accruals

• Table IA9: Alternative measurements of political uncertainty

Notes

1 Prior literature on how political forces shape financial reporting, for example, includes Bird et al. (2017), Boone et al. (2019), Kido et al. (2012), Jiang et al. (2019), Nagar et al. (2019), Ramanna (2008), and Ramanna and Roychowdhury (2010).

2 Jagolinzer et al. (2020) document that managers’ access to politicians increases their insider trading profits. Also, see Political Risk Can’t Be Avoided, But It Can Be Managed, Forbes, by Steve Culp, August 27, 2012, for a discussion about how political risks can be identified, measured and controlled by managers (Accessed on February 5, 2020): https://www.forbes.com/sites/steveculp/2012/08/27/political-risk-cant-be-avoided-but-it-can-be-managed.

3 We examine conditional conservatism, because it helps to incorporate new information into financial reporting, improves contracting efficiency, and reduces agency costs, while unconditional conservatism might be inefficient, or at best, neutral in the contracting process (Ball & Shivakumar, 2005). Moreover, unconditional conservatism does not necessarily imply conditionally conservative reporting. For example, Qiang (2007) finds that unconditional conservatism reduces conditional conservatism and concludes that the two forms meet distinct needs, and it is necessary for firms to trade them off.

4 When demanding greater conservatism, investors may also strive to gain more information from managers via other sources (e.g., textual disclosures and voluntary disclosures) when political uncertainty increases. Relating to this argument, Jiang et al. (2019) show that political uncertainty leads managers to make longer and less readable textual disclosures in their 10-K and 10-Q filings, increasing the cost for investors to digest and acquire information. Moreover, Nagar et al. (2019) find that even though managers increase voluntary disclosures in response to the rising political uncertainty, they document that the increased disclosures only partly mitigate the increased information asymmetry.

5 We find the results are significant in these firms with strong contracting demand, but insignificant in those with weak contracting demand. The differences between subsamples are statistically significant in some of the analyses with regard to industry exposures, leverage, and corporate governance mechanisms, but insignificant in other analyses, such as the analyses in relation to electoral participation and lower managerial ownership.

6 Early work in this area referred to political or policy uncertainty as regime uncertainty. For example, Higgs (1997) argues that policy changes are much more uncertain under certain political regimes.

7 One exception is the state of Louisiana, where election timing can differ because of the open primary system applied to gubernatorial elections.

8 In extenuating circumstances, there may be special elections with irregular timing. In our sample period, there are four such gubernatorial elections: (1) the California gubernatorial recall election in 2003; (2) a Utah special election was conducted in 2010; (3) a West Virginia special gubernatorial election was conducted in 2011; and (4) Oregon held a special election in 2016.

9 We verify the election data using internet sources such as www.ourcampaigns.com. For robustness tests, we also collect the presidential data from the U.S. Census Bureau and the economic policy uncertainty index data from www.policyuncertainty.com.

10 We identify firm’s headquarters state, using two Compustat files, ‘Company Header History’ and ‘Company.’ We start with the ‘Company Header History’ file, which allows us to track the relocations of headquarters based on the effective dates. We complement it with the headquarters information in the ‘Company’ file when such information is missing in the ‘Company Header History’ file.

11 See, for example, Ahmed and Duellman (2013), Ball et al. (2013a), LaFond and Roychowdhury (2008), LaFond and Watts (2008), Lawrence et al. (2013), Ramalingegowda and Yu (2012), and Roychowdhury and Martin (2013).

12 Prior studies using the approach proposed by Collins et al. (2014) include, for example, Banker et al. (2016), Bens et al. (2018), Kim (2018), Basu and Liang (2019), and Khurana and Wang (2019).

13 We acknowledge that our main findings based on the asymmetric timeliness might be confounded by the effects of non-accounting factors, including expected returns, cash flow persistence, and asymmetry in return distribution (Dutta & Patatoukas, 2017). To alleviate this concern, in the subsequent sections, we perform cross-sectional analyses to examine whether the change in conservative accounting plays a role through the contracting demand channel and find supportive evidence to our hypotheses, though we cannot completely rule out the confounding effects suggested by Dutta and Patatoukas (2017).

14 It is a judgment call to assess the economic magnitude of our baseline findings. To evaluate our results, we review the literature and find nine papers relating to inter-temporal changes in conditional accounting conservatism for U.S. firms from the Top 6 Accounting Journals and Management Science in the period 2014–2018. The median value of the absolute effect of an inter-temporal shock to conservatism documented in these papers is 35.2%, and 4 out of 9 papers have economic magnitudes within 10% of our main effect (17%), ranging from 11.0% to 25.5%. This comparison analysis suggests that our findings are comparable to prior studies in terms of economic magnitude.

15 We are grateful for the voter turnout data provided by Professor Michael P. McDonald.

16 The enforcement sensitivity is equal to one minus the Herfindahl index of an industry’s input shares from other industries and varies from zero to one. A zero value of this measure indicates that an industry j only uses the inputs from one other industry i, i.e., industry j has the lowest sensitivity to contract enforcement.

17 We focus on the role of internal governance mechanisms because the prior literature has documented mixed evidence on the relation between external corporate governance mechanisms (such as anti-takeover provisions) and accounting conservatism (e.g., García Lara et al., 2009; Jayaraman & Shivakumar, 2013).

18 BoardEx and MSCI GMI Ratings (formerly known as Corporate Library, used in Ahmed and Duellman (2007)) both provide the information of board composition and ownership, while in our sample, the numbers of observations for non-missing values of Board Insider Percentage_it-1 (Board Outsider Ownership_it-1) are 54,529 and 36,173 (10,193 and 36,173) in BoardEx and MSCI GMI Ratings, respectively. Therefore, we choose BoardEx to construct Board Insider Percentage_it-1 and use MSCI GMI Ratings to estimate Board Outsider Ownership_it-1.

19 Although the timing of presidential elections is exogenous, this measure is subject to a lack of cross-sectional variation and has fewer election observations – our sample period covers 14 presidential elections, compared to 699 gubernatorial elections.

20 There are two drawbacks of using these PU _EPU measures to investigate our research question: (1) there is little cross-sectional variation in the measures, since they are based on country-level indices; and (2) they may not be purely exogenous – that is, collectively, firm’s accounting and financial decisions could impact news coverage, government policy, and economic forecast.