Can the Balanced Scorecard Help in Designing Conference Calls? The Effect of Balanced Information Composition on the Cost of Capital

Abstract

Most recent studies on conference calls focus on the costs for firms that can arise from the calls’ open nature. We study the benefits of conference calls and hypothesize that firms could use the balanced scorecard concept as a framework for presenting the information (i.e. balanced information composition) in conference calls to lower the cost of capital. Our results show a negative association between a more balanced information composition in conference calls and a firm’s cost of capital. Additional tests substantiate that the effect of such a balanced information composition on the cost of capital is driven by a reduction in information asymmetry. Overall, the findings suggest that firms can benefit from the balanced scorecard concept by using it as a framework for preparing their conference calls.

Keywords:

• Earnings Conference Calls
• Cost of Capital
• Balanced Scorecard
• Textual Analysis

JEL classifications:

• G14
• L2
• M41

Acknowledgments

We thank the editors, Reuven Lehavy and Florin Vasvari, and two anonymous reviewers for their constructive guidance during the review process. We are also grateful for the helpful comments from Nico Lehmann, Jason Chen, Naomi Soderstrom, Anastasiya Zavyalova, Federico Aime, Jonathan Bundy, and Stefan Wagner. Moreover, we would like to thank participants at the Content Analysis Workshop at the Annual Meeting of the Academy of Managament in Anaheim (August 2016), the Finance, Accounting, and Taxes Research Colloquium at University of Goettingen (May 2017), the Annual Meeting of the Academy of Management in Anaheim (August 2016), the Annual Meeting of the European Accounting Association in Valencia (May 2017), the Annual Meeting of the American Accounting Association in San Diego (August 2017), and the Annual Meeting of the German Academic Association for Business Research (VHB) in Magdeburg (May 2018).

Supplemental Data and Research Materials

Supplemental data for this article can be accessed on the Taylor & Francis website, doi:10.1080/09638180.2019.1709523.

Appendix S1. Development of a Word List for Balanced Information Composition

Notes

1 In addition to this, the study by Matsumoto et al. (2011) goes further and demonstrates the informativeness of the content of conference calls beyond accompanying press releases from an investor-focused perspective. In contrast, our study intends to take a managerial perspective by proposing a framework that lays out a specific course of action for firms to prepare their information composition in a beneficial (value-creating) manner.

2 We follow a frequent assumption in the disclosure literature and expect a reduction in information asymmetry to lower a firm’s cost of capital (e.g., Armstrong et al., 2011; Brown et al., 2004; Diamond & Verrecchia, 1991). A lower cost of capital resulting from the presentation of information in conference calls seems plausible as conference calls are meant to help investors and analysts comprehend existing information from the earnings announcement.

3 The Gini coefficient determines the inequality among values of a frequency distribution using the Lorenz curve. It measures the ratio of the area between the Lorenz curve and the egalitarian line (i.e., a 45-degree line) relative to the entire area under the egalitarian line. Larger values of the Gini coefficient imply greater inequality.

4 The literature commonly stresses that the less formal disclosure format, the lower standard of legal liability, and the opportunity for extemporaneous disclosure vest conference calls with a more open nature (e.g., Frankel et al., 1999; Hollander et al., 2010; Matsumoto et al., 2011).

5 Diamond and Verrecchia (1991) suggest that revealing information to lower information asymmetry can reduce firms’ cost of capital.

6 In line with previous studies, we do not explicitly refer to information quality or quantity as the driver of the reduction in information asymmetry as both are empirically hard to disentangle. Instead, we refer to Leuz and Verrecchia (2000, p. 91), who state that ‘theory is sufficiently broad as to allow the notion of ‘increased levels of disclosure’ to be interpreted as either an increase in the quantity of disclosure or an increase in the quality of disclosure (or both).’ For example, in our case, a more balanced information composition could lead to an increase in information quantity, which could then be used to assess information that is already available, thereby improving information quality.

7 While many empirical studies exploit the assumed link between a reduction in information asymmetry and a lower cost of capital, the theorists’ debate revolves around various positions. The theory widely agrees on the indirect effect of information asymmetry on the cost of capital as information asymmetry among investors poses the problem of adverse selection in capital markets leading to illiquidity (Amihud & Mendelson, 1986; Leuz & Verrecchia, 2000; Welker, 1995). Information asymmetry is linked with price protection by uninformed investors facing informed investors (Welker, 1995). Investors’ behavior then reduces the liquidity in the market and exacerbates selling or buying shares. In turn, illiquidity imposes trading costs on investors, for which they expect to receive compensation linking information asymmetry with the cost of capital (Amihud & Mendelson, 1986; Brennan & Subrahmanyam, 1996). Besides the indirect link, the theory also considers a direct link between information asymmetry and the cost of capital as increased information asymmetry results in higher estimation risks (Lambert, Leuz, & Verrecchia, 2007; Lambert et al., 2012). In order to investigate the effect of balanced information composition on the cost of capital in more detail, we conduct additional tests on other capital market outcomes linked to information asymmetry and liquidity risk in Section 5.

8 Because we exclude firm-quarters without conference calls, a potential selection bias is inherent in this sample based on the firms’ choice regarding whether to hold conference calls. We employ a sample selection correction suggested by Heckman (1979) by including a correction factor derived from a first-stage probit regression in the subsequent regression. In unreported tables, we obtain similar results.

9 The calculation models and their components are described in detail in Appendix A.

10 In unreported regressions, we consider all words spoken by the management including those from the Q&A session and obtain similar results.

11 We also calculate other common inequality measures such as the relative mean deviation, the coefficient of variance, the Mehran measure, Piesch measure, Kakwani measure, and the Theil entropy measure (He & Huang, 2011). The correlation between these measures and the Gini coefficient is mostly above 0.96 and is 0.82 at the lowest, indicating the reliability of the Gini coefficient.

12 The extensive description of the word list development in the online Appendix might prove particularly fruitful for researchers who intend to develop their own dictionaries. Thereby, the applied development procedure in this paper might help to uncover new measurements.

13 The challenge with 2SLS is to find a suitable instrument variable that is correlated with the independent variable but uncorrelated with the error terms (Larcker & Rusticus, 2010). Due to a lack of fitting instruments, the use of industry averages is quite common but far from perfect (Larcker & Rusticus, 2010). The same would be true for the use of 2SLS in this study. In the regression results in the robustness section, we employ the industry average of the balanced information composition as the instrument and find that the results for our hypothesis using 2SLS hold. When considering the advantage that GMM does not require external instruments, we chose GMM over 2SLS.

14 The issue with a potential fixed-effects model in our case is that the independent variable and, even more so, the dependent variable are not independent of past realizations, and thus are relatively sticky over time. This issue is even more pronounced when examining the cost of capital, which is why the majority of studies investigating the cost of capital abstain from using the fixed-effects model.

15 A detailed description of the variables and their sources is provided in Appendix B.

16 In 95% of our sample’s observations, the finance perspective is the most dominant.

17 To obtain the marginal effect, we multiply the coefficient with the standard deviation of the balanced information composition measure (BIC): –2.283 × 0.211 = –0.482.

18 In addition to the prominence of the information dimensions, we also looked for established and readily available word lists to avoid developing new word lists. For Forward-looking to backward-looking, Long-term to short-term, Non-financial to financial, and CSR to other we used the word lists directly from the papers, and for Non-quantitative to quantitative we used the quoted word lists numbers and quantifiers from the software Linguistic Inquiry and Word Count (LIWC). While Brochet et al. (2015) provided two word lists, all other sources provided one word list (e.g., forward), so we calculated the counterpart (e.g., backward) by subtracting the frequency count from all words in the MD session. We then standardized and normalized the values and created the ratio.

19 Kothari et al. (2009) find that positive news reduces uncertainty and thus has a directional link with a lower cost of capital. Thus, balanced information composition could systematically coincide with positive news.

20 Bid–ask spreads are a common proxy for information asymmetry (Lee, 2016; Leuz & Verrecchia, 2000). Some studies, however, have also employed bid–ask spreads as a measure for illiquidity as they are closely linked to ‘price protection that uninformed market participants demand as compensation for the perceived information risk associated with trading in equity markets’ (Welker, 1995, p. 802).

21 In unreported regressions, we test the effect of balanced information composition on analyst forecast errors and obtain similar results.