Firm Opacity Lies in the Eye of the Beholder

AuthorMarco Navone,Sandeep Dahiya,Giuliano Iannotta
Published date01 September 2017
DOIhttp://doi.org/10.1111/fima.12168
Date01 September 2017
Firm Opacity Lies in the Eye of the
Beholder
Sandeep Dahiya, Giuliano Iannotta, and Marco Navone
We classify and test empirical measures of firm opacity and document theoretical and empirical
inconsistencies across these proxiesby testing the relative opacity of banks versus non-banks. We
evaluate the effectiveness of these proxies by observing the effectof two cleanly identified shocks
to firm-specific information: credit rating initiation and inclusion in the S&P 500 index. Using
a difference-in-difference approach, we compare firms that are newly rated and firms that are
included in the S&P 500 index with a propensity matched sample of “unchanged” firms. We find
that only the number of analysts and Amihud’s illiquidity ratio provideconsistent patterns across
different estimation specifications and differenteconometric settings. These two proxies show that
banks are more opaque than non-banks. Based on our tests, we recommendthat these proxies be
used as the primary measures of firm opacity.
Information asymmetry between contracting agents has been a rich area of research for
economists. Theories of asymmetric information form key elements of seminal frameworks
about capital structure and corporate governance.1The central theme in this body of research is
that outside investors do not enjoyperfect transparency about the inner workings of a fir m, which
in turn creates a wedge between investors (outsiders) and the manager (insider). “Firm opacity”
is usually meant to reflect the level of information asymmetry between insiders and outsiders.
Firm opacity has been used as a motivating factor for a wide variety of economic outcomes in
recent studies.2
Firm opacity cannot be observed directly.Empirical researchers typically use a measurable firm
characteristic as a proxy for firm opacity to test their hypotheses. A quick survey of the recent
literature in which firm opacity is either the main variable or one of the control variables shows
there is little consensus on what is (are) the best proxy(ies) for firm opacity. Some authors choose
proxies based on information production by third parties such as analysts or rating agencies. The
idea is that the existence of a credit rating reduces a firm’s opacity as an independent agency
We are grateful to Raghu Rau (Editor) and an anonymous referee for their valuable comments and suggestions that
significantly improved the article. We thank Renee Adams, Clifton Green,Ronald Masulis, George Pennacchi, Fernando
Zapatero, and participants at the FinancialManagement Association 2012 conference and the seminar at UTS Business
School for helpful comments. All errors are our responsibility. Dahiya acknowledges the support of Stallkamp faculty
fellowship grantprovided by Georgetown University’sMcDonough School of Business. Iannotta and Navone are grateful
to the CAREFIN programat Bocconi University for financial assistance.
Sandeep Dahiya is an Associate Professor of Financeat the McDonough School of Business, Georgetown University in
Washington, DC. Giuliano Iannotta is a Professor of Finance in the Department of Economics and Business Adminis-
tration, Universita Cattolica in Milano, Italy.Marco Navone is a Senior Lecturer in Finance in the Finance Discipline
Group, UTS Business School, University of Technology in Sydney, Australia.
1For example, the pecking order theory of capital structure (Myers, 1984) uses the information asymmetry between
outside investors and inside managers as its motivation.
2Sufi (2007) shows that bank loan syndicates tend to be more concentrated for opaque borrowers to ensure tighter lender
monitoring. Gomes and Phillips (2012) show that as the level of information asymmetry increases, a firm is more likely
to raise capital from private markets.
Financial Management Fall 2017 pages 553 – 592
554 Financial Management rFall 2017
produces and disseminates firm-specif ic information. This reduces information asymmetry faced
by potential investors. Recent studies that use this approach include Faulkender and Petersen
(2006), Sufi (2007), and Bharath et al. (2011). Similarly, the quantity and quality of equity
analysts who follow a particular firm are also frequently used as measures of firm opacity.
Recent studies that employ this strategy include Leary and Roberts (2010), Mehran and Peristiani
(2010), and Gomes and Phillips (2012). Other studies examine stock market price behavior.
Jin and Myers (2006) argue that firm opacity is correlated with negatively skewed returns and
higher comovements with the market index, and the same intuition is shared by Hutton, Marcus,
and Tehranian (2009) and Haggard, Martin, and Pereira (2008). Finally, Flannery, Kwan, and
Nimalendran (2004) argue that opacity increases stock illiquidity, and thus measures of price
impact or transaction costs are viable proxies for firm opacity.
This wide range of firm opacity proxy variables creates challenges for empiricists seeking to
examine the impact of firm opacity on economic decisions (e.g., capital structure). The problem
is especially acute when considering the theoretical and empirical inconsistencies across these
measures. Consider, for example, the following set of findings reported in recent papers. A high
level of comovementbetween a fir m’s stock price and the broad marketindex (typically estimated
as R2of the regression of firm stock return against the market index return) is considered a sign
of high opacity for that firm by some researchers (see Morck, Yeung, and Yu, 2000; Jin and
Myers, 2006). In contrast, an increase in analyst coverage is typically assumed to be a sign of
increasing transparency (Mehran and Peristiani, 2010). However, Chan and Hameed (2006) show
that an increase in analyst coverage (predicting lower firm opacity) is associated with higher R2
(increase in firm opacity).
There is little empirical work that compares the effectiveness of these opacity proxies and
their relation to one another. Our first contribution is to catalog and compare opacity measures
used in previous empirical studies. This is an important issue because if the observed correlation
across proxies is high, it suggests they are all measuring similar sources of information opacity.
This should mitigate any concerns about the results being driven by the opacity proxy being
used. However, if these proxies capture unrelated information about firms, the correlation across
measures should be low. In that case, it is possible that empirical tests seeking to explain variation
in firm opacity may be driven, at least in part, by the opacity proxy chosen by the researcher
rather than by any true variations in underlying firm opacity.3
Previous studies have employed a large number of opacity proxies. Because it is unwieldy
to catalog every measure, we focus on a subset of nine commonly used measures. Our choice
of information opacity measures have been widely used in the literature. We use the banking
industry as a special case to test opacity proxies. Although the question regarding relative opacity
of banks versus non-banks is an important one, we are primarily interested in examining the
consistency across opacity proxies. Thus, if we find that all opacity proxies suggest higher (or
lower) opacity for banks, we can conclude that the relative merits of employing one proxy versus
another is likely to be insignificant. However, our analysis of banks versus non-banks shows that
these proxies produce inconsistent results: compared to non-banking firms, some proxies show
that banks are more opaque whereas others show that banks are less opaque.4
3If there are different types of opacity,the different proxies may capture these different information opacities. However,
there is little theoretical work to guide an empiricist interested in designing tests to examine such differences.
4Our conflicting evidence on bank opacity suggests that the results are highly dependent on the proxy used. Wewant to
stress that the tests of bank opacity suffer from a joint hypothesis problem. A failure to find higher (lower) opacity of
banks could be due to either poor choice of opacity proxy or lack of a significant difference in the opacity of banks and
non-banks.
Dahiya, Iannotta, & Navone rFirm Opacity Lies in the Eye of the Beholder 555
Although the tests of bank opacity highlight the potential pitfall of using a single class of
information asymmetry measure to draw broad conclusions, our final set of tests attempts to
identify opacity proxies that can be used across all industries. Our second major contribution is
to develop a “horse race” to measure the effectiveness of various opacity proxies not just for the
banking industry but for all industries. It is impossible to measure firm opacity directly and we
lack a “true benchmark” against which to calibrate the measures used in prior research. We use
a quasi-experimental approach of identifying firms that are subject to a cleanly identified f irm-
specific information shock (the treatment sample) and estimate the change in pre- and postshock
levels of various opacity measures for this group. Using the propensity score matching (PSM)
methodology, we identify a control group of firms that had the same likelihood of experiencing
the shock and estimate the change in pre- and postshock level of opacity measures for this group.
We examine the difference-in-difference (DID) of these opacity measures for the two groups to
see whether the information shock is indeed accompanied by a significant change in the opacity
proxies.
The two information events we use are the initiation of a credit rating and the inclusion in the
Standard & Poor’s(S&P) 500 index. We arguethat both these events constitute an empirically clean
and identifiable shock to the quality of the information environment faced by outside investors.
Both events are likely to increase the investor base of a firm. Initiation of a rating is likely to
attract new bond investorsand inclusion in the S&P 500 index is frequently associated with higher
institutional ownership. Thus, these events are likelyto be accompanied by a subsequent decrease
in information opacity. Our justification for using these events does not depend on causation.
Our choice of these events should be acceptable as long as there is a significant and positive
association between rating initiation/index inclusion and reduction in opacity. We acknowledge
that obtaining a credit rating is an endogenous firm decision and is likely driven by many
underlying firm characteristics that may also affect the information asymmetry between the firm
and potential outside investors. Still, it is plausible to argue that a firm receiving a credit rating
for the first time experiences a significant reduction in infor mation opacity, as ratings constitute
a major information gathering and dissemination event. For example, Faulkender and Petersen
(2006) show that having a credit rating is almost always associated with the presence of public
debt in a firm’s capital structure. Inclusion in the S&P 500 index appears to be exogenous, but the
probability of being included is not random. S&P has a well-defined process for index selection.
We mitigate these concerns by creating a matched sample of control firms that had a similar
likelihood of experiencing these two events. We exploit this discontinuity in the information
environment to test the effectiveness of various opacity proxies. As mentioned earlier, we adopt
a DID approach. For example, we compare each newly rated firm with a group of matched
unrated firms that were as likely (based on firm-level and other contemporaneous factors) to have
received a rating at the same time. We argue that for newly rated firms, opacity proxies should
differ significantly for the pre- and postrating periods. Our results suggest that the number of
analysts following a firm and the price impact, as measured by Amihud’s (2002) ratio, appear to
be the most consistent across multiple specifications. We repeat these tests for inclusion in the
S&P 500 index and again find that analyst coverage and Amihud’s (2002) measure are significant.
Our article provides additional empirical evidence that stock return synchronicity is not a
robust proxy for information opacity. This comovement is typically measured by the R2derived
from regression of firm stock returns against overall market returns. How to interpret R2as a
measure of firm opacity has been debated in recent studies. Morck et al. (2000) and Jin and
Myers (2006) argue that a firm with a high R2is likely to have less firm-specific infor mation
in its stock price. This line of argument suggests that higher R2implies higher firm opacity.
However, Kelly (2014) finds that low R2stocks appear to havethe g reatest degree of information

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT