Ch. 2: Self-Selection Models in Corporate Finance 67
i.i.d. for the purposes of estimation. As Song points out, additional data may not help
shed light on interpretation or robustness because there have been structural changes in
the banking industry since 1996, due to several mergers and further relaxation of the
Glass–Steagall Act.
8.3. Underwriter reputation:Fang (2005)
Like the other papers reviewed in this section,Fang (2005)also studies the role of
underwriter choice in explaining at-issue bond spreads. Unlike the other papers in the
section, however, Fang draws on an earlier literature and classifies underwriters by repu-
tation rather than by organization into commercial or investment banks. Fang examines
whether the information in the choice of a reputed underwriter impacts underwriting
spreads and yields.
Fang uses a probit specification to model underwriter-issuer matching. If issueiis
underwritten by a reputed underwriter, the yield isYE,iand if not, the yield isYNE,i.
Yields are specified as a function of regressorsxiwith different regression coefficients
across the two choices. Thus, Fang’s model is exactly the switching regression of Sec-
tion3.1. Fang also estimates an auxiliary regression where the dependent variable is
gross spread rather than offer yield.
Fang finds that reputed underwriters underwrite higher grade, less risky issues of
large and frequent issuers, and are more likely to be associated with longer maturity
callable issues that she interprets as being more complex. The self-selection term in the
yield equation is negative. Thus, the unobserved information that leads firms to choose
reputed underwriters leads to lower bond yields or better offer prices. In the specifi-
cation analyzing gross spreads, Fang finds that issue size increases fees more rapidly
but risk variables matter less for reputed underwriters, indicating greater marginal costs
and superior risk bearing capacity of reputed underwriters. Most importantly, the coef-
ficient for the inverse Mills ratio in the gross spread equation is positive, suggesting that
reputed underwriters charge greater fees to issuers.
Taken together, the yield and gross spread specifications show that reputed underwrit-
ers charge issuers greater fees and lower the offer yields (i.e., increase the offer price)
to borrowers. Fang shows that the benefit of lowered debt yields typically outweighs
the higher commissions paid by issuers. The pattern of results is shown to strengthen
in lower yield bonds, so that reputation matters more for more informationally sensitive
issues.
8.4. Debt covenants:Goyal (2005)
While the papers reviewed in this section study and model information in underwriter
choice,Goyal (2005)examines the information in the choice of covenants attached to
debt issues. Goyal argues that commercial banks often enjoy franchise value because of
regulations that deter free entry. Banks with more valuable franchises are less likely to
engage in excessive risk taking, so they should have less need to include covenants in