The Effect of Board Independence on Information Asymmetry

Abstract

Boards have an important role in ensuring that investors’ interests are protected. Our paper first examines whether the independence of a firm's board affects information asymmetry among investors. We provide evidence that greater board independence leads to lower information asymmetry. Next, we provide evidence that more voluntary disclosure and greater analyst coverage are two underlying mechanisms via which greater board independence reduces information asymmetry. Of the two mechanisms, we find that analyst coverage is more significant in influencing how board independence affects information asymmetry. Overall, our paper contributes to a better understanding of the effect of board independence on information asymmetry.

Acknowledgements

We wish to thank Gilles Hilary (editor), an anonymous reviewer, Heidi Vander Bauwhede, Mark Evans, Ole-Kristian Hope, Steve Huddart, John (Xuefeng) Jiang, Suresh Radhakrishnan, Sugata Roychowdhury, Ewa Sletten, Oktay Urcan, Rodrigo Verdi, Ross Watts, Joe Weber, and workshop participants at the 2008 Annual Conference on Financial Economics and Accounting, the 2009 AAA FARS Midyear Meeting, the 2009 Midwest Finance Association Conference, and the 2009 EAA Annual Congress, for their comments.

Funding

We also appreciate the financial support from the School of Accountancy Research Center (SOAR) at Singapore Management University.

Notes

¹The intuition for the first instrument is based on the concept of director social networks (Adams & Ferreria, 2009). Briefly, when directors of a board sit on other boards with more independent directors, they know more independent directors whom they can persuade to join the board and/or are more open to having more independent directorships. The intuition for the second instrument is that when there are more external directors serving in firms in the nearby location, there is a greater supply of independent directors from which the firm can appoint its independent directors (Knyazeva, Knyazeva, & Masulis, 2011). We discuss more about these instruments in Section 4.1.

²There are key differences between both approaches. First, the instrumental variable approach can typically be applied to any time period. In contrast, the use of an exogenous event shock, by construction, requires the empirical analyses to be at a single time point. Second, an exogenous event shock offers a cleaner identification to the extent that it satisfies the requirement that no other confounding events and that it causes a spike in board independence. Instrument variables often face significant validity threats due to imperfect exogeneity (Larcker & Rusticus, 2010). Given the differences in the identification techniques, similar findings help strengthen the conclusion that more board independence leads to an improvement to the information environment.

³In terms of the language used in path analyses, information processing cost has a moderating (as opposed to mediating) effect on the effect of board independence on the information environment.

⁴Bushman and Smith (2003) also highlight other channels through which financial accounting information can affect economic performance: (1) better identification of good vs. bad projects by managers and investors and (2) discipline on project selection and expropriation by managers.

⁵For example, there could be public disclosures that are not (directly) related to earnings but which could reduce information asymmetry, such as discussions about the future direction of the firm in the annual reports, press releases, and conference presentations. Board independence could also affect mechanisms other than public disclosures. For example, with better oversight by independent directors, there might be less private communication between managers and selected stakeholders and less private-information-based insider trading.

⁶Many studies have suggested that the findings on the relation between board independence and information environment suffer from endogeneity (e.g. Armstrong et al., 2010, 2014; Hermalin & Weisbach, 2003; Larcker, Richardson, & Tuna, 2007; Larcker & Rusticus, 2010).

⁷To study how gender-diverse boards affect firm performance, Adams and Ferreira (2009) instrument gender-diverse boards (i.e. the fraction of female directors on the board) using board connections to female directors (i.e. the fraction of male directors on the board who sit on other boards that have female directors). For example, Jackson (2009) state ‘The fact that social networks are an important conduit of information about and access to jobs is evident to anyone who has ever looked for employment in almost any profession.’ In the case of the employment of independent directors to be on the firm's boards, social networks are likely to play an especially important role because there is a small pool of qualified directors to begin with. By definition, independent directors have to be from outside the firm. It is highly unlikely that a complete stranger would be added from the outside to the board because boards are typically quite small and individual board members can have significant influence on the board dynamics.

⁸The unconditional probability of an independent director getting a new directorship the following year is 2.5%, while the same probability for a connected independent director is 4.2%.

⁹The prior literature has used the fraction of a board's directors who sit on at least one additional board as a proxy of ‘busy directors’ (e.g. Fich & Shivdasani, 2006). Note that this proxy does not require the additional board(s) to have a majority of independent directors. Nevertheless, one might still view Board connections as potentially a proxy for busy directors. When directors sit on other boards, there are many outcomes such as (i) they get to know more directors, (ii) they are busier, and (iii) they learn more from other directors and of other firms. While we focus on the first outcome (i.e. the social connections aspect), to the extent that the other outcomes are unlikely to have a direct effect on the information environment, Board connections remains a reasonable instrument.

¹⁰We thank an anonymous referee for suggesting this additional instrument.

¹¹Unlike structural equation modeling, which deals with both measured and latent variables, path analysis is regarded as a special case of structural equation modeling in that the latent variables are explicitly specified.

¹²One standard deviation in PIN is 0.0632, and hence the impact of 0.015 standard deviation decline in PIN is (0.0632 × 0.015) divided by 0.1247 (mean value of PIN) = 5.32%. The other comparative statistics are computed analogously.

¹³We thank the authors for sharing the data on Adj_PIN with us.

¹⁴The order flow of each trade is classified as a buy or a sell using the standard Lee–Ready algorithm (Lee & Ready, 1991), which involves a ‘quote test’ and a ‘tick test’. For the ‘quote test’, any trade that takes place above (below) the midpoint of the current quoted spread is classified as a buy (sell) order because trades originating from buyers (sellers) are most likely to be executed at or near the ask (bid). For trades taking place at the midpoint, a ‘tick test’ is used to classify the trade. This test classifies a trade as a buy (sell) order if the trade price is above (below) the previous price. In the event there is no change in the trade price, the order flow is regarded as indeterminable and the trade is not used in computations. The daily number of buy orders and sell orders is determined by adding up the number of orders in each category for each day for each firm.