208 S. Drucker and M. Puri
fees. A number of studies build on this comparison of fees by focusing directly on the
impact of lending relationships on gross spreads. In public debt markets, two studies
find that prior lending relationships reduce the gross spread. Using a regression frame-
work that controls for bond and issuer characteristics,Roten and Mullineaux (2002)find
that prior lending relationships lower debt underwriting fees by 10 basis points during
the 1995 to 1998 period.Yasuda (2005)gathers a sample of 1,535 bond issues from
1993 to 1997 and uses a more advanced methodology to examine the impact of lending
relationships on fees. The author points out that studies that examine the effect of lend-
ing relationships on fees use the equilibrium pricing outcomes that are observed.Yasuda
(2005)argues that the gross spread that is observed is likely to be lower on average than
the unconditional distribution of the gross spread. Therefore, the author imputes the im-
plied gross spread for each of the other underwriters that the firm could have selected
to underwrite the issue. Using the Expectation-Maximization Algorithm that accounts
for this downward bias in observed gross spread,Yasuda (2005)estimates a joint model
of the gross spread and the firm’s selection of underwriter. The gross spread is modeled
as a function of bond and issuer characteristics, as well as if the potential underwriter
was an arranger on any of the firm’s prior loans before 1993. The choice of underwriter
is a function of the implied gross spread, bond and issuer characteristics, and existing
lending relationships.Yasuda (2005)also finds that lending relationships significantly
decrease the gross spread by approximately nine basis points.
As in debt underwriting, the evidence suggests that lending reduces the gross spreads
of equity offerings. Three papers use the framework developed byAltinkilic and Hansen
(2000), who find that gross spreads for seasoned equity offerings are U-shaped with re-
spect to the size of the offering. Theoretically, U-shaped curves can arise because scale
economies cause gross spreads to decline initially, but as issue size increases, higher
placement costs can override the benefits of scale economies, causing gross spreads to
increase.Narayanan, Rangan, and Rangan (2004), using seasoned equity offerings from
1994 to 1997, include a variable that captures if a commercial bank in the underwriting
syndicate has a lending relationship with the issuing firm. They find that the existence
of a lending relationship reduces gross spreads by 46 basis points, which is significant
at the one percent level. This result is consistent with informational economies of scope
from combining lending and underwriting.Drucker and Puri (2005)study “concurrent
lending” and underwriting, which occurs when the underwriter of a seasoned equity
offering provides a loan to the issuer between six months before and six months af-
ter the issuance. As part of their study, the authors examine the impact of concurrent
lending and prior lending on seasoned equity offering gross spreads. The authors argue
that informational economies of scope are likely to be large when issuers receive a loan
concurrently because the information from the lending transaction is directly re-usable
in the equity offering. The authors extend theAltinkilic and Hansen (2000)model to in-
clude variables that control for firm characteristics and prior underwriting relationships
as well as variables that indicate if the lead underwriter provided concurrent loans or
had a prior lending relationship with the issuer. For a sample of 2,301 seasoned equity
offerings from 1996 through 2001, concurrent lending without a prior lending rela-