Leary and Roberts, JF (2014)
Many CFOs responding to Graham and Harvey’s survey have indicated that peer firm financial policy plays an important role in their decision-making. A number of empirical papers have shown that average industry leverage seems to be an important determinant of individual firm leverage. Intuitively, peer firm leverage might be a signal of optimal capital structure and/or investment opportunities. If a CFO does not have all the answers, he may try to infer them from peer firms’ decisions.
There are two general reasons why a firm’s capital structure decisions might appear to be related to the decisions of peer firms. The first reason is that the firms may be subject to common factors. Exposure to these common factors might have led to the firms selecting to become “peer firms” in the first place. The second reason is that the firm may respond to the characteristics and/or the actions of its peer firms in a causal manner. The confoundedness of these two reasons lead to what Manski (1993) calls the “reflection problem.” This is an endogeneity problem that arises when trying to determine the impact of group behavior on the behavior of individuals within the group.
This paper asks, in addition to the title question, whether firms respond to the actions or to the characteristics of peer firms.
The basic idea is to regress firm outcome on peer firm characteristics, perfectly controlling for common variation; or, to measure the impact on firm outcome of an exogenous shock to peer firms. This paper regresses peer firm stock returns on market excess returns and on firm fixed effects over rolling 5-year windows, and uses the residuals as the exogenous shock. They use (lagged) idiosyncratic stock return as the exogenous shock because it is easy to calculate and can be measured each period, unlike other plausibly exogenous shocks such as natural disasters or CEO death. The inclusion of industry fixed effects is supposed to account for any common factors, rather than for priced return determinants.
Table 3, Panel A shows the results from regressing several capital structure outcomes on own-firm idiosyncratic stock return and characteristics, average peer-firm idiosyncratic stock return and characteristics, and industry and year fixed effects.
- Columns 1-2 use market leverage and book leverage as dependent variables
- Columns 3-4 use first-differences of market and book leverage, with no industry fixed effects, to control for unobserved own-firm characteristics in another way.
- Columns 5-7 use different types of security issuance as dependent variables.
Table 3, Panel B conducts a number of robustness tests to address a number of concerns.
- Column 1 replaces lagged own-firm idiosyncratic stock return with lagged and contemporaneous own-firm total stock return. The results are similar. This alleviates the concerns that (1) lagged peer-firm are correlated with own-firm idiosyncratic shock and this is what drives the results, and (2) the asset pricing model is misspecified so that the residuals are biased in some way.
- Column 2 controls for a bevy of additional variables that the previous literature has presented as determinants of capital structure.
- Column 3 controls for bank fixed effects.
- Column 4 includes own-firm lagged leverage ratio to account for possible leverage targeting.
- Column 5 replaces lagged controls with contemporaneous controls to verify that the results are not due to the idiosyncratic shocks affecting characteristics with a lag.
- Column 6 adds quadratic and cubic polynomials of the peer-firm and own-firm characteristics to control for misspecification of functional form in the baseline regression.
The authors then redefine peer groups, and use shocks to peer-firm customers, that are not own-firm customers and that are in a different industry, as the exogenous shock.
Finally, they perform a double 5×5 sort on peer-firm idiosyncratic equity shock and peer-firm actions (changes in leverage). For each group, they calculate the own-firm change in leverage. The differential changes in own-firm leverage across the quintiles of one variable, controlling for the other variable, shed light on whether firms respond to the characteristics or to the actions of their peers.
- Firms respond to peer-firms’ financing decisions.
- Firms respond to peer-firms’ characteristics, but to a lesser degree than to their decisions.
- Peer firm behavior is a more important determinant than the observable determinants from prior literature.
- Smaller, younger, and less-successful firms tend to be more influenced by peer firms, while industry leaders appear to be less influenced.
I love that the main findings of the paper can be presented in just one table (table 3)! The authors use an interesting type of exogenous shock and draw on a variety of finance research to motivate and defend it. I also love the literature review section(s) of this paper. It is very well written…or maybe it’s just personal preference for the subject.
This paper is in line with Lemmon, Roberts, and Zender (2008) in questioning the relevance of several decades of capital structure. If firm peer groups are relatively static, then this could explain the just-mentioned paper’s findings that firm fixed effects are the best determinant of firm capital structure.