78 K. Li and N.R. Prabhala
unobserved attributes that are negatively related to value. These unobserved attributes
positively impact value if the founder is not the CEO.^29
- Other applications of matching methods
12.1. Bank debt versus bonds:Bharath (2004)
Debt financing by a corporation gives rise to conflicts of interest between creditors and
shareholders that can reduce the value of the firm. Such conflicts are limited more effec-
tively in bank loans than in public debt issues if banks monitor.Bharath (2004)measures
the size of agency costs by calculating the yield spread between corporate bonds and
bank loans (the Bond-Bank spread) of the same firm at the same point in time. To quan-
tify the difference, Bharath needs to match bonds with bank loans of the same firm at the
same point in time and having substantively identical terms. The matching problem is
complicated by the fact that bank loans and public bonds are contractually very different
on multiple dimensions such as credit rating, seniority, maturity, and collateral.
Bharath argues that because bank loans and bonds are matched at the same point of
time and for the same firm, matching based on observables should adequately control for
differences between bank debt and public debt. Thus, propensity score based matching
methods are appropriate tools to control for differences between bank loans and public
debt. Bharath uses the propensity score matched difference between bank and bond
credit spreads as the treatment effect, or the value added by banks. The spread can be
interpreted as the value added by banks in enforcing better investment policies, or more
generally, as the price of the “specialness” of banks due to their ability to monitor,
generate information, or better renegotiate loans, or even perhaps other explanations
such as monopoly rents.
Using a sample of over 15,000 yield observations, Bharath finds that the Bond-Bank
spread is negative for high credit quality firms and positive for low credit quality firms.
He interprets his findings as being consistent with the view that for high quality firms,
the benefits of bank monitoring are outweighed by the costs of bank hold-up. This
causes the spread to be negative, indicating that bank debt offers few benefits for high
quality firms. For low quality firms, the opposite is true, causing the spread to be posi-
tive. The magnitude of the potential agency costs mitigated by banks is more important
for poor quality firms, justifying the decision to borrow from banks.
(^29) An interesting question raised by this study is survivorship (e.g.,Brown, Goetzmann and Ross, 1995). Per-
haps family owned firms that survived and made it to Fortune 500 status are of better quality, and hence these
firms are valued more. This question can perhaps be resolved by looking at broader samples that incorporate
smaller firms outside the Fortune 500 universe.Bennedsen et al. (2006)take a step in this direction.