can local long-term institutional ownership alleviate...
TRANSCRIPT
Can Local Long-term Institutional Ownership Alleviate
Information Asymmetry in Bank Loan Pricing?*
Kiyoung Chang
Ying Li
Ha-Chin Yi
Abstract
We use local institutional ownership as a proxy for ownership that is informed and exogenous
and show that local long-term institutional ownership (LLTIO) is negatively associated with the
spread charged by lenders. LLTIO features geographic proximity and long investment horizons
that alleviate information asymmetry between the borrower and syndicated lenders. The negative
relation between LLTIO and loan spread is salient only when a gap for geography-related soft
information exists and when conflicts of interest between equity and debt holders are unlikely.
We show that better monitoring through LLTIO’s long-term equity commitment is more likely
the reason for the LLTIO effect.
JEL classifications: G14, G21, G32
* Chang, [email protected], College of Business, University of South Florida Sarasota-Manatee, Sarasota, FL
34343; Li (corresponding author), [email protected], School of Business, University of Washington, Bothell, WA
98011; Yi, [email protected], McCoy College of Business Administration, Texas State University, San Marcos, TX
78666. We thank Kee-Hong Bae, Soku Byoun, Hinh Khieu, Jin-Mo Kim, Yong-Cheol Kim, Jieun Lee, Deming Wu,
seminar participants at University of Washington, Bothell, and conference participants at the FMA 2014 Annual
Meeting for comments and suggestions. We thank the review committee of the Joint Conference and Symposium of
All Five Finance-Related Korean Academic Associations Annual Meeting for recognizing our work with a best
paper award. All remaining errors are our own.
I. Introduction
Banking theories posit the presence of information asymmetry between borrowers and
lenders, which may cause adverse selection and moral hazard problems (Diamond, 1984). They
also suggest that informed ownership can serve as a signal (Leland and Pyle, 1977) and that an
informative signal is a valuable indicator of due diligence and monitoring (Holmstrom, 1979;
Holmstrom and Tirole, 1997). As informed ownership of the borrower is associated with
improved due diligence and monitoring, it can serve as a positive signal to the lender in the
process of evaluating the borrower’s credit worthiness and therefore influence the loan pricing
terms. In this paper, we show that the borrower’s long-term informed equity ownership reduces
the loan spread in the syndication process by alleviating information asymmetry between the
lead lender and the borrower. We also show that the local long-term institutional ownership
(LLTIO, hereafter) at the borrowing firm is likely to influence the loan spread through its
monitoring function.
It is difficult to evaluate the price effect of information asymmetry in the syndicated loan
market due to a host of simultaneity and endogeneity problems (Sufi, 2007). Ownership effect
on mitigating information asymmetry is also difficult to show because ownership is usually
endogenous (Demsetz and Lehn, 1985). We use ownership by institutional shareholders whose
headquarters are geographically proximate to the headquarters of the borrowing firm as a proxy
for ownership that carries information on the borrower.1 The major advantage of our proxy of
informed ownership is that the main determinant of proximity, namely, the location of the
1 Coval and Moskowitz (2001) and Baik, Kang, and Kim (2010) find that local institutional investors earn
substantial abnormal returns in their nearby equity holdings and that the amount of local equity investment is
positively correlated with that stock’s expected return. Malloy (2005) shows that analysts possess an informational
advantage on local stocks. Finally, Gaspar and Massa (2007) use local ownership as a proxy for private information
to demonstrate the trade-off between monitoring and liquidity effects.
investor, is reasonably exogenous (Gaspar and Massa, 2007). 2
Our formal endogeneity test using
instrumental variable regressions also suggests that LLTIO is exogenous.3 The presence of local
institutional ownership could serve as a signal to lenders, which is informative and influences
loan pricing. Since local institutional equity ownership exists outside of the loan syndicate, its
effect on loan pricing does not involve simultaneity problems. In this study, we examine the
effect of the ownership by institutional investors that are both geographically close and belong to
the ten largest shareholders of a borrowing firm (Top10LIO) on loan spreads. We focus on the
top10 local shareholders as informed owners with large stakes are subject to great under-
diversification risks (Leland and Pyle, 1977). Top10LIO thus provides a convincing signal to
lenders.
Whereas the local institutional ownership possesses information that entails either
improved due diligence or monitoring (Coval and Moskowitz, 2001; Gaspar and Massa, 2007),
or both, investment horizons play an important role in how institutions exert their influences
(Harford, Kecskes, and Mansi, 2014). Institutional ownerships with a long-term investment
horizon are associated with stronger monitoring effects (Gaspar, Massa, and Matos, 2005) than
short-term ownerships. We therefore differentiate local institutional ownerships with a long-term
and short-term investment horizon and examine their differential effects on loan spreads to gain
insights on the mechanism through which informed equity ownership affects the loan spread.
Using a sample of borrowing firms from the U.S. syndicated loan market over the 1995 –
2009 period, we find the borrower’s local long-term institutional owners (LLTIOs) who belong
to the ten largest shareholders (Top10LLTIO) to be negatively associated with the loan spread
charged by lenders. The local short-term institutional owners (LSTIOs) with large stakes
2 Kang and Kim (2008) make a similar argument. Our empirical results also support the exogeneity of local
ownership. 3 Our results using only index funds also remain unchanged.
(Top10LSTIO) are not associated with lower loan spreads, suggesting LLTIOs’ monitoring
functions are more likely the driving force behind reduced information asymmetry. We refer to
the negative relation between the LLTIOs and loan spreads as the “LLTIO effect”.
The LLTIO effect is salient only under certain conditions: 1) when conflicts of interest
are unlikely between creditors and shareholders, such as in the case of investment-grade
borrowers and during out-of-crisis periods and 2) when the necessary hard information is in
place, for example, when the borrowers have credit rating, which belongs to hard information,
yet there is a gap for geography-related soft information, for another example, when the
borrower actively invest in research and development (R&D). If the need for geography-related
soft information is not present as lenders obtain it through alternative channels, the LLTIO effect
does not exist. For example, we do not observe the LLTIO effect when the lead lender is
geographically proximate to the borrower, or when the borrowers are close to urban cities. These
findings suggest that the LLTIO effect indeed becomes salient only when the presence of the
LLTIOs addresses geography-related information asymmetry between the borrower and lenders.
They also suggest that the possible dual role of LLTIOs, that is, LLTIOs being both shareholder
and creditor of the local firm, is not the likely reason for our findings. When exploring the
syndicate structure, however, we find no evidence to support the existence of the LLTIO effect
within the loan syndicate.
We also use the implementation of Regulation Fair Disclosure (Regulation FD) in 2000
and of the Sarbanes-Oxley Act (SOX) in 2002 as two natural experimental settings to investigate
the mechanism(s) through which the LLTIO effect functions. Our findings suggest that the
LLTIO effect is associated with improved monitoring, as it becomes statistically and
economically more significant after the SOX implementation. Further exploration suggests that
when the LLTIOs are present, the likelihood of observing managerial misbehavior such as
“lucky” CEO (Bebchuk, Grinstein, and Peyer, 2010) drops significantly. We also find a negative
relation between the level of the entrenchment index (E-index) (Bebchuk, Cohen, and Ferrell,
2009) and the proportion of the LLTIOs. Both findings are consistent with the LLTIOs’
monitoring function.
Our findings contribute to the existing literature in several ways. First, although past
studies have investigated the effect of a geographically proximate lender has on loan pricing
(Peterson and Rajan, 2002), the follow-up studies find mixed results. The net effect of a
geographically proximate bank on loan pricing could be driven by either a lower premium due to
less information asymmetry (Knyazeva and Knyazeva, 2012), or a higher premium due to spatial
discrimination (Degryse and Ongena, 2005; Agarwal and Hauswald, 2010). In our study, by
focusing on the informational effect of local institutional equity ownership which resides outside
of the loan syndicate, we avoid the confounding effects of the hold-up problem and conduct a
clean test on how the LLTIO conducive to less information asymmetry between the borrower
and lender can be an important factor in bank loan contracting.4
Second, to the best of our knowledge, our paper is the first to demonstrate that the
presence of the borrower’s geographically proximate long-term institutional shareholders
reduces information asymmetry between the lead bank and the borrowing firm in the context of
loan pricing. Even though monitoring is unobservable, we show that lenders give more favorable
loan terms when borrowers have observable characteristics (Top10LLTIOs in our study) that
provide a signal suggesting improved monitoring.
4 A hold-up problem arises as banks exploit their informational advantage at the borrower’s expense (Rajan, 1992;
Sharpe, 1990).
Finally, whereas past studies show the importance of geography on collection cost of soft
information (Knyazeva and Knyazeva, 2012) and the benefits of soft information (Peterson and
Rajan, 1994; Berger and Udell, 1995), the mechanism through which geography-related soft
information enters into the loan syndication process is less clear. Our paper suggests that
monitoring function is the main mechanism for the LLTIO effect that we document. In summary,
our findings contribute to the literature on loan pricing and shed light on the monitoring role that
Top10LLTIOs play and the channels through which such monitoring works in the context of
bank loans.
The rest of the paper is organized as follows. Section 2 describes the syndicated loan
market and the features of LLTIOs that could address information asymmetry in a syndicated
loan setting. Section 3 presents the data and summary statistics. Section 4 contains our empirical
results and Section 5 concludes.
2. LLTIO and Bank Loan Pricing
Internationally, syndicated loans represent an important and fast-growing source of
financing for corporations, with $1.8 trillion in such loans issued in 2009, more than the total
value of corporate borrowing in the global bond markets (Chui et al., 2010). According to
Thomson Reuters (the provider of the DealScan data used in this paper), the U.S. leveraged loan
issue, a subset of all syndicated loans issued in the U.S., reached $664 billion in 2013. Secondary
loan trading in the country also exceeded $600 billion in 2014. Of the 500 largest Compustat
firms, 90% have obtained syndicated loans (Sufi, 2007), and 51% of all U.S. corporate financing
is in the form of syndicated loans (Weidner, 2000).
Loan quality and pricing are functions of both hard and soft information about the
borrower (Stein, 2002). Hard information, such as a borrower’s credit rating, whether it pays
dividends, and whether it is a component of a major stock index, is easy to collect and verify. On
the contrary, soft information, such as the harmonic ongoing relationship among the top
management team, middle management, and labor, the prospect for potential of research and
development (R&D), and the cultural compatibility between a new CEO and other top
executives, is difficult to collect and verify (Agarwal and Hauswald, 2010; Liberti and Mian,
2009), as many involve a dynamic process. Geographic proximity enables easier access to
certain soft information (Rajan, Seru, and Vig, 2015) and facilitates monitoring of the borrower
(Sufi, 2007).
Prior to loan syndication, the lead bank conducts due diligence on the borrower before
reaching agreement with it on a target spread range over the London Interbank Offered Rate
(LIBOR) (Ivashina, 2009). Although the lead bank possesses a large amount of information on
the borrower, much of it could take the form of hard information, leaving significant gaps in soft
information, which is harder to evaluate or transfer from a distance (Stein, 2002; Peterson, 2004).
A lead bank may also possess soft information from a previous banking relationship with the
borrowing firm, but geography-related and previous relationship-related soft information are not
necessarily identical. For example, although a distant lead bank may have a good understanding
of the business and operations of a borrowing firm based on its previous relationship with that
firm, it is likely to be less costly for a local lead bank to evaluate the firm’s current status and
future prospects.
The composition of ownership may send a signal to outsiders (Fombrun and Shanley,
1990). A high degree of institutional ownership is believed to be associated with a high level of
information gathering effort and low level of information assessment errors (Sias, 1996), and
firms with a high degree of institutional ownership are therefore likely to be viewed favorably.
The largest institutional shareholders are also highly visible to lenders. As LLTIOs possess
superior access to geography-related soft information, their presence can serve as a favorable
signal in loan pricing. Such a favorable signal from the local institutions with the largest stakes
in the borrower is particularly convincing because of the greater risk of under-diversification
(Leland and Pyle, 1977).
Furthermore, LLTIOs with large stakes can serve as an external monitor for the borrower,
alleviating moral hazard problems over the long term. Geographic proximity provides a cost-
benefit justification for monitoring, and facilitates intense monitoring through frequent
interactions with local firms and other stakeholders. For example, Gaspar and Massa (2007) and
Kang and Kim (2008) show such proximity to reduce both transportation and communication
costs and encourage local investors to get involved. Concentrated ownership and a long-term
investment horizon are additional characteristics that contribute to lower monitoring costs
(Hartzell and Starks, 2003; Gaspar, Massa, and Matos, 2005). Therefore, LLTIOs with large
stakes serve as an external monitoring mechanism for the borrower and alleviate moral hazard
problems over the long term. Indeed, Gaspar and Massa (2007) and Chhaochharia, Kumar, and
Niessen-Ruenzi (2012) show that local institutional ownership is associated with improved
corporate governance, which is documented to lead to a lower cost of capital (Stulz, 1999).
Whereas local short-term institutional owners (LSTIOs) may possess private information
on the borrowing firm ex ante, their short-term investment horizon suggests that they will have
little monitoring incentive (Gaspar, Massa and Matos, 2005; Chen, Harford, and Li, 2007).
Frequent trading by LSTIOs leads to more uncertainty, resulting in an offsetting effect on the
loan spreads that will confound the study of the local institutions’ information effect. To explore
the effect of informed equity ownership of the borrower on information asymmetry, we focus on
the LLTIOs. We hypothesize that, ceteris paribus, LLTIOs reduce premiums lenders charge for
loans by serving as a signal that fills the gap in geography-related soft information. We also
expect the LLTIO effect on loan pricing to be more salient when conflicts of interest between
creditors and shareholders are unlikely. Otherwise, a gain to the latter could be synonymous with
a loss to the former in case of elevated conflicts of interest between two stakeholders, such as in
a high level of financial distress, because shareholders selfishly exert their influence on corporate
decisions at the expense of creditors (Jensen and Meckling, 1976; Myers, 1977). Furthermore, in
the absence of information asymmetry or a need for geography-related soft information, either
because the loans are secured or the lenders are geographically proximate, we expect the LLTIO
effect on loan pricing to vanish.
3. Data and Summary Statistics
Our bank loan data come from the Thomson Reuters LPC DealScan database, and our
information on financial characteristics and stock returns from Compustat and the Center for
Research in Security Prices (CRSP), respectively. We also obtain institutional ownership data
from the Thomson Reuters 13F database. We match the DealScan dataset with the Compustat
dataset using the list of identifiers constructed by Chava and Roberts (2008). Our sample
excludes financial and regulated utility industry borrowers and non-U.S. borrowers. The final
sample includes 17,308 loan deals with financial and stock information on 3,810 unique
borrowing firms, which has non-zero institutional ownership over 1995-2009. We report the
variable definitions in Appendix A, and summary statistics in Table 1.
3.1 Loan Data
DealScan collects loan-level data, mostly on syndicated loans, from various sources,
including annual reports, reports from loan originators, and Securities and Exchange
Commission (SEC) filings. Syndicated loans are medium- or large-sized loans extended to firms
by a group of lenders. In a typical syndicated loan contract, a small number of lenders, called
lead lenders or arrangers, head up a group of participating banks that jointly issue a relatively
large loan package to share the risk and meet capital requirements. The role of the lead lenders is
to serve as a bridge between the borrowers and participating banks. They serve both sides of the
table: for the borrower, the lead bank secures financing, and for the participating banks, it
performs credit-screening on borrowers through due diligence and then offers ex-post
monitoring. Our research variable is the all-in-drawn spread (spread) for syndicated loans,
which, according to the DealScan definition, is the total annual cost in basis points paid over the
London Interbank Offered Rate (LIBOR) for each dollar used under the loan commitment.
3.2 Institutional Ownership Data
Form 13F mandatory institutional reports are filed with the SEC on a calendar quarter
basis, and are compiled by Thomson Reuters (formerly known as the 13F CDS/Spectrum
database). The SEC requires all institutions with more than $100 million under management at
the end of the year to file Form 13F reporting their long positions in equity5 in the next year.
Form 13F filings thus have several limitations: for example, small institutions with less than
$100 million under management are not required to report their positions, smaller holdings that
do not reach the 10,000-share or $200,000 threshold are not included, and short positions are not
reported. Further, Thomson Reuters aggregates holdings reports at the management company
level.6
5 The reported positions are those in which the institution owns more than 10,000 shares or with a market value
greater than $200,000. 6 A given 13F report may include holdings reported by multiple funds/managers that are not necessarily located in
the same area as the headquarters. This problem, which is suffered by most local-related studies using 13F data,
constitutes one of the limitations of our study.
A firm’s local investors are defined as those located within a short distance. As we cannot
differentiate holdings by the local offices of the same institutional investor, we focus on the
location of the corporate headquarters of the management company to identify local institutional
investors, which is similar to the approach used by Gaspar and Massa (2007) and Baik, Kang,
and Kim (2010). Also, similar to Knyazeva, Knyazeva, and Masulis (2013), we obtain corporate
headquarters locations and firm-level financial variables from the Compustat database. If
information on the corporate headquarters location is missing, we obtain it manually. We
identify the institutional location (zip code) by manually searching the SEC EDGAR site for
historical 13F filings.
Consistent with John, Knyazeva, and Knyazeva (2011), we use the distance between the
corporate headquarters of firms and the headquarters of institutional investors to calculate local
institutional ownership. Like Baik, Kang, and Kim (2010), we exclude cases in which either the
firms or institutional investors are located in Alaska, Hawaii, Puerto Rico, or the Virgin Islands.
We first identify the 10 institutional investors with the largest stakes in a firm and calculate the
percentage of shares owned by these top-10 owners (Top10IO). We then calculate the
percentages of shares owned by long-term and short-term investors7 whose headquarters are
located within a 100-mile radius of the firm’s headquarters.8 We use the percentages, including
7 Following Bushee (2001), we categorize ownership by institutional owners who are either dedicated or quasi-index
as long-term institutional ownership. According to Bushee (1998), dedicated institutional investors are characterized
by large average investments in portfolio firms with extremely low turnover ratios, whereas quasi-indexers are
characterized by low turnover and diversified holdings. He argues that both types of investors provide firms with
long-term, stable ownership because they are geared toward longer-term benefits, be those benefits dividend income
or capital appreciation (Bushee, 2001). We thank Brian Bushee for providing institutional investor classification
data (1981–2009) on his website: http://acct3.wharton.upenn.edu/faculty/bushee/. 8 Coval and Moskowitz (1999, 2001) and Gaspar and Massa (2007) use a 100-kilometer radius as a measure of
locality, whereas Ivkovic and Weisbenner (2005) set 250 miles as the maximum radius for local investors, and Baik,
Kang, and Kim (2010) adopt state identifiers to identify local institutional investors. The distance, 𝑑𝑖,𝑗 , between the
headquarters of institutional owner i and firm j is calculated as follows: 𝑑𝑖,𝑗 = arccos(𝑑𝑒𝑔𝑙𝑎𝑡𝑙𝑜𝑛) ×2𝜋𝑟
360, where
𝑑𝑒𝑔𝑙𝑎𝑡𝑙𝑜𝑛 = cos(𝑙𝑎𝑡𝑖) × cos(𝑙𝑜𝑛𝑖) × cos(𝑙𝑎𝑡𝑗) × cos(𝑙𝑜𝑛𝑗) + cos(𝑙𝑎𝑡𝑖) × sin(𝑙𝑜𝑛𝑖) × cos(𝑙𝑎𝑡𝑗) × sin(𝑙𝑜𝑛𝑗) +
those of concentrated overall local, local long-term, and local short-term institutional ownership
(Top10LIO, Top10LLTIO, and Top10LSTIO, respectively), as a proxy for informed equity
ownership. The overall local institutional ownership (Top10LIO) for firm j is calculated as
follows:9
𝑇𝑜𝑝10𝐿𝐼𝑂𝑗 =∑ 𝑉𝑖,𝑗𝑖∈𝐿𝑗
∑ 𝑉𝑖,𝑗𝑖∈𝐼, (1)
where Lj is the set of the ten largest institutions based on shares of firm j owned that are
headquartered within a 100-mile radius of firm j’s headquarters, I is the universe of all ten of the
largest institutions based on their stake in firm j, and Vi,j is the dollar value of institutional owner
i’s stake in firm j.
𝑇𝑜𝑝10𝐿𝐿𝑇𝐼𝑂𝑗 =∑ 𝑉𝑖,𝑗𝑖∈𝐿𝐿𝑇𝑗
∑ 𝑉𝑖,𝑗𝑖∈𝐼, (2)
𝑇𝑜𝑝10𝐿𝑆𝑇𝐼𝑂𝑗 =∑ 𝑉𝑖,𝑗𝑖∈𝐿𝑆𝑇𝑗
∑ 𝑉𝑖,𝑗𝑖∈𝐼, (3)
Top10LLTIO and Top10LSTIO are calculated similarly as described in Equations (2) and (3),
where Top10LLTIOj and Top10LSTIOj are Top10LIOs who have long-term and short-term
investment horizons, respectively, according to Bushee’s categorization: long-term institutional
investors include dedicated and quasi-indexers and short-term include transient institutions.
3.3 Control Variables
We include firm characteristics, loan characteristics, macro-economic variables, and
industry dummies as our control variables. Firm characteristics include firm size, asset
sin(𝑙𝑎𝑡𝑖) × sin(𝑙𝑎𝑡𝑗), lat and lon are the latitudes and longitudes of the institutional owner and firm, and r is the
radius of the earth (approximately 3,959 miles). 9 Coval and Moskowitz (2001) and Gaspar and Massa (2007) define local ownership as the “excess” local ownership
in one firm relative to the benchmark expected for the particular locality in which it is headquartered. We use actual
local institutional ownership out of the top-10 largest shareholders, an approach similar in spirit to that adopted by
Baik, Kang, and Kim (2010). This measure enables us to calculate changes in ownership and assess the effect on
alleviation of information asymmetry.
tangibility, membership of the S&P 500 index, profitability, financial distress (modified Z),
leverage, credit rating, stock volatility, R&D-to-asset ratio, and institutional ownership. Loan
characteristics include whether the loan is secured, loan type, maturity, loan purpose, and
relationship status. The other control variables include term spread, credit spread, and the Fama-
French twelve industry effects. The term spread and credit spread are measured on an annual
basis, and we omit year fixed effects to avoid multicollinearity due to the presence of term
spread and credit spread variables in the model. We use firm size and the average debt issue size
as proxies for economies of scale in flotation costs, following Krishnaswami, Spindt, and
Subramaniam (1999). Also, large public borrowers are usually covered by many analysts, and
accordingly more public information is available on such borrowers. Hence, we expect a
negative relation between firm size and spread. Similarly, information asymmetry is less severe
for S&P 500 index firms, dividend-paying firms, and borrowing firms with a previous banking
relationship with the lead bank (Berger and Udell, 1995; Petersen and Rajan, 1994) and for loans
originating with reputable banks (Ross, 2010; Dennis and Mullineaux, 2000). We thus expect a
negative relation between S&P500 dummy, Div dummy, Relation dummy, Top3 bank, and
spread. Leverage is a proxy variable for the observable default risk, and we expect a positive
relation between it and spread (Merton, 1974; Carey, Post, and Sharpe, 1998). As tangible assets
are easier to value than intangible assets, we expect a negative relation between asset tangibility,
NFA/TA, which is measured as the ratio of net fixed assets to total assets, and spread. Return on
assets (ROA) captures borrower profitability, and is expected to have a negative association with
spread. Finally, top-10 institutional ownership (Top10IO, the ratio of shares owned by the 10
largest institutions to the shares owned by all institutional investors) captures the concentration
level of institutional ownership. Larger, more mature firms are likely to have a large numbers of
institutional owners, and the 10 largest shareholders of such firms are thus likely to be less
representative of overall institutional ownership compared to other firms. We therefore expect a
positive relation between Top10IO and spread.
Table 1 presents summary statistics at the loan level. Syndicated loans are issued as a
package deal, with each deal possibly comprising multiple revolvers (or credit lines) and term
loans (or installment loans). Loan-level presentation provides a good picture of our sample
because revolvers and term loans contain different loan specifications. Our sample comprises
17,308 loans over the 1995–2009 period. The average loan spread is about 182 basis points
above the LIBOR. There is a wide variation in the spreads for our sample, with a minimum
spread of 2.7 basis points and maximum spread of 1,500 basis points.10
Approximately 20% of
the sample loans were obtained by S&P 500 firms, and about half were the outcome of repeat
loans from the same set of lead lenders and borrowers. Approximately 30% of the loans were
issued by the three banks with the largest dollar volume of syndications, namely, JP Morgan
Chase, Bank of America, and Citi Bank. According to Ross (2010), these three banks accounted
for almost half the total syndicated loan volume, measured in dollars, in the 2000–2008 period.
On average, the ten largest institutional investors hold 29% of the equity in a borrowing
firm, with long-term investors constituting the majority. Local owners are a relatively minor
group (approximately 9% of the sample loans), with the long-term investors among them
accounting for roughly 7% of sample loans. About 4% of loans were obtained by borrowing
firms located within 100 miles of the lead syndicate lenders. On average, the book value of the
sample borrowing firms is approximately $786 million, with a leverage level of 31%, but just
over half of all loans (53%) were secured with some form of collateral. Just under half (49%)
10
A closer examination of our sample identifies multiple loans with a spread of more than 1,000 basis points,
suggesting that the wide variation in loan spread is unlikely to be a recording mistake. Our results remain largely the
same after removing the extreme observations as we use the logarithms of spread to minimize the impact of outliers.
were obtained by firms paying dividends. About 24% of loans were obtained by investment-
grade firms (with long-term credit ratings of BBB or above), and the rest were by either
speculative-grade firms (with long-term credit ratings below BBB) or firms that do not have a
credit rating. Roughly half the loans (47%) are obtained by investment-grade firms.
Approximately 56% of loans are revolvers, and 26% are term loans, and the average maturity is
about 48 months.
[Table 1 about here]
Panel A of Table 2 shows the summary statistics for different firm and loan
characteristics with high and low Top10LLTIO and Top10LSTIO, respectively. If a loan is
associated with greater than 5% Top10LLTIO/ Top10LSTIO, it belongs to the high Top10LLTIO/
Top10LSTIO group, and to the low Top10LLTIO/ Top10LSTIO group otherwise. The univariate
statistics show that, relative to those in the low Top10LLTIO group, loans in the high
Top10LLTIO group have a lower loan spread (159 basis points versus 190 basis points, on
average), are obtained by firms with a previous banking relationship with the current lenders,
issued by the top-three reputable banks, from larger, more profitable, more likely dividend-
paying and S&P500 firms, with a lower level of leverage, and are less risky as measured by
credit ratings. With the exception of the long-term revolvers, all of the mean differences are
statistically significant with a confidence level of 1% or better.
Relative to those in the low Top10LSTIO group, loans in the high Top10LSTIO group are
obtained by firms with smaller size, no dividend, higher volatility in operating cash flow, and
higher risk as measured by credit ratings. The loan spread difference for the two groups with
high and low Top10LSTIO is not clear, with the t-statistic on logspread difference being
insignificant. The results from the univariate test confirm our conjecture that due to their
frequent trading and lack of monitoring, LSTIOs’ informational effect is unclear.
Panel B of Table 2 shows the breakdown of our sample into two groups: investment-
grade versus speculative-grade loans. About 77% of our sample loans are speculative loans. As
expected, safe loans tend to be repeat deals, and are more likely to be issued by one of the top-
three banks. Investment-grade firms are more likely to pay dividends, are more profitable, and
tend to have more LLTIOs. The results of univariate analysis are consistent with the literature
reporting that loans issued to borrowing firms with a high credit rating and previous banking
relationship tend to have a lower loan spread (Yi and Mullineaux, 2006; Schenone, 2010).
Panel C of Table 2 shows the comparison of mean spread and logspread between
borrowing firms by their credit rating and the level of Top10LLTIO using a threshold of 5%,
respectively. Whether the borrowing firm has an investment grade or not, the mean differences in
spread and logspread associated with high (≥5%) and low (<5%) Top10LLTIO are similar in
magnitudes and significant at a better than 1% confidence level. Although the univariate result
needs to be verified in a multivariate setting later, it suggests that it is the level of Top10LLTIO is
more likely to drive the result than credit ratings.
[Table 2 about here]
4. Empirical Results
4.1 The LLTIO Effect
Theory tells us that information asymmetry between lenders and borrowers is a key factor
driving the terms of loan contracts, which attempt to deal with adverse selection and moral
hazard problems (Diamond, 1984). Informed ownership can serve as a signal that mitigates the
costs of information asymmetry (Leland and Pyle, 1977), and prior research has demonstrated
that such asymmetry can influence the structure and pricing terms of syndicated loans (Dennis
and Mullineaux, 2000; Sufi, 2007; Ivashina, 2009; Knyazeva and Knyazeva, 2012). We argue
here that because local institutional ownership (LIO) with sizable stakes represents informed
ownership, which can play a credible role in ensuring either due diligence, or monitoring, or both
(see Holmstrom, 1979), the presence of LIOs in a borrowing firm may induce lenders to reduce
the loan spread. Furthermore, as the location of LIOs is exogenous to bank loan contracting
terms, our identification strategy is less troubled by endogenous concerns.
To investigate the informational effect of LIO on syndicated loan pricing, we estimate the
following multivariate regression, which includes both long- and short-term ownership by the
top-10 (largest) shareholders with headquarters located within a 100-mile radius of the
borrowing firm’s corporate headquarters, as well as the control variables specified in Equation
(2), and report the results in Table 3. We use the logarithm of spread (logspread) as the measure
of loan spread, similar to other studies in the banking literature (for example, Graham, Li, and
Qiu, 2008), in all regression analyses.
Loan spread = f (Top10LLTIO, Top10LSTIO, institutional ownership, loan characteristics, firm
characteristics, macro-economic variables, industry dummies). (4)
Column (1) examines the relation between logspread and overall local ownership with
large equity stakes in the borrowing firms (Top10LocalIO). With the exception of the added
institutional ownership variables, the results in Column (1) are consistent with the findings from
the previous studies of bank loan pricing. That is, a larger, more profitable firm, longer maturity
loan, and firm with a prior relationship with the lender obtain loan rate discounts, whereas a
highly levered, volatile firm with a low credit rating and a loan backed by collateral pays a
higher spread. The coefficient estimates on loan purpose and industry are also generally
significant. The coefficient estimate on Top10IO is positive, as we expected, since Top10IO is
likely to be a higher percentage at smaller firms and smaller firms tend to face higher loan
spreads. It is also significant at the 1% level.
Column (2) examines the relation between logspread and top-10 local institutional
ownership with different investment horizons (Top10LLTIO for long-term and Top10LSTIO for
short-term local institutional ownership, respectively). After controlling for creditworthiness and
other firm and loan characteristics, the coefficient estimate on Top10LLTIO is negative and
significant with a confidence level of 1%, whereas that on Top10LSTIO is positive and
significant with a confidence level of 5%. Short-term investors appear to be better informed and
to trade in a way that exploits their informational advantage (Yan and Zhang, 2009), and local
short-term institutional investors are able to generate superior returns (Baik, Kang, and Kim,
2010). Banks may view short-term (or transitory) investors as harmful to the stability of a firm
due to the higher level of equity trading turnover and volatility they create. Furthermore,
transitory ownership is usually opportunistic and lacks ex-post monitoring incentives (Gaspar,
Massa, and Matos, 2005). These factors may explain the observed positive relation between
Top10LSTIO and logspread. Conversely, the top-10 LLTIOs have incentives to monitor and
access to geography-related soft information obtained through a long-term commitment. These
features of the LLTIOs help to alleviate the information asymmetry between borrowing firms
and lenders and explain the observed negative relation between Top10LLTIO and logspread.
Column (3) is identical to Column (2) except for a slightly different definition of the
credit rating variable, an indicator variable that captures the default risk of the borrowing firm
(Invgrade in Column (2) and Invgrade2 in Column (3)).11
Although the sample size changes, the
11
The dummy variable Invgrade takes a value of one if the S&P rating on a borrowing firm’s long-term debt is
BBB- or above. Because many borrowing firms do not have a long-term debt rating, when Invgrade takes a value of
results in Column (3) are consistent with those reported in Column (2), with the coefficient
estimates on Top10LLTIO negative and significant, both with a confidence level of 1%. When
we include bank fixed effects in the regression, the LLTIO effect remains significant, suggesting
that it is not driven by specific bank characteristics. We report these results in Column (4). 12
To disentangle the roles played by geographic proximity from long-term investment
horizons, as either may be driving the LLTIO effect on logspread, we also report results using
the 10 largest long- and short-term institutional owners (Top10LTIO and Top10STIO) as
explanatory variables. The results, reported in Column (5) show that Top10LTIO presence is not
associated with loan spread, suggesting that geographic proximity is the key element for the
existence of the LLTIO effect.
Urban areas, especially vicinities of New York City and Connecticut are the headquarters
of a large cluster of institutional owners. For example, in results that are not tabulated, we
observe that the average level of Top10LLTIO is above 14% in the vicinity of Connecticut and
0.5% in the vicinity of Texas. The level of Top10LLTIO in urban areas is therefore highly
skewed and the LLTIO effect we observe could be due to a borrowing firm’s central location.
Previous studies (Loughram and Schultz, 2005; John, Knyazeva, and Knyazeva, 2011; Chen,
Gompers, Kovner and Lerner, 2010; Cumming and Dai, 2010) show that urban location does
matter for a firm’s dividend payout policy and for venture capital success. To address this
concern, we exclude firms that are located in one of the ten largest metropolitan statistical areas
(MSAs) and re-estimate Equation (4). The results in Column (6) show that the LLTIO effect
zero, neither non-investment-grade borrowers nor those without a rating are included. We thus define another
dummy variable, Invgrade2, which takes a value of zero only when the borrower has a long-term debt rating and
that rating is below BBB-. The sample size changes depending on whether borrowing firms without a long-term debt
rating are included, and Columns (2) and (3) thus have different numbers of observations. 12
Since our sample is built on loan facilities, the data is not panel. We therefore rely on the ordinary least squares
(OLS) regression instead of panel data techniques for empirical analysis.
remains negative and significant in non-urban areas, with a confidence level better than 1%. To
compare the effect of LLTIO and LSTIO on loan spreads for the same borrowing firm at the
same time, we include firm fixed effects and year fixed effects13
for the non-urban borrowing
firm sample and report the results in Column (7). The LLTIO effect remains significant,
suggesting that it does not exist in the urban areas only.
[Table 3 about here]
4.2 LLTIO as a Proxy for Exogenous Ownership
The validity of our argument depends on whether LLTIO can be considered a form of
exogenous ownership because the effect of endogenous ownership on asymmetric information is
difficult to show (Demsetz and Lehn, 1985). Although following the arguments in Gaspar and
Massa (2007) and Kang and Kim (2008) renders Top10LLTIO reasonably exogenous, we also
adopt an instrumental variable (IV) approach to formally establish causality between
Top10LLTIO and logspread. IV regressions can help alleviate the endogeneity concern, which
stems from certain unobservable firm characteristics being omitted from the model but is related
to both logspread and Top10LLTIO. We introduce the two following IVs for Top10LLTIO.
State Top10LDIO: Annual average of top10 local dedicated institutional owners14
with
the largest stakes for all other firms in the same state but in different industries, as
defined by their 2-digit SIC code.15
Industry Top10LQIO: Annual average of top10 local quasi-index institutional owners16
with the largest stakes for all other firms within the same industry, as defined by their 2-
digit SIC code.17
13
Fixed year effects are subsumed under term spread and credit spread as both variables have annual variation. 14
Following Bushee (1998), we define dedicated institutional ownership as being characterized by large average
investments in portfolio firms with extremely low turnover ratios. It is a component of LLTIO. 15
The IV (State Top10LDIOi) for Top10LLTIOi is constructed by including all other firms in the same state, but not
the same industry, as firm i, identifying the aggregate Top10LDIO level for each, and calculating the annual average
Top10LDIO across firms in a given year. Similarly, we construct our other IV (Industry Top10LQIO) using
information on Top10LQIO for all other firms with the same 2-digit SIC codes to calculate the annual average.
A valid IV needs to satisfy two conditions: relevance and exclusion. We expect that
whether they belong to dedicated (Top10LDIO) or quasi-indexers (Top10LQIO), Top10LLTIO
are likely to be indifferent with their targets, if they choose to monitor due to the same reason,
that is, lower cost of doing so. Therefore, an institutional investor with monitoring motivation
will likely take actions at other firms that are also geographically close. This assumption
suggests that our location-based IV, State Top10LDIO, satisfies the relevance condition. The
exclusion condition requires that State Top10LDIO affect loan spread at the borrowing firm only
through its information asymmetry alleviation effect, not because of other factors that can
influence both LLTIO and loan spreads. For example, State Top10LDIO focuses on the
Top10LDIO of other borrowing firms in different industries, satisfying the exclusion criterion.
Furthermore, in results that are not tabulated here, we find that the LLTIO effect remains
negative and significant in a similar regression to Equation (4) after controlling for added state
fixed effects of the borrowing firms. With state fixed effects, we focus on the within-state
variation of LLTIO and we still find high LLTIO to be associated with lower logspread. This
suggests that location in different states does not have a systematic effect on loan spreads and
therefore is not a factor that drives our results. We also include Industry Top10LQIO as a second
IV to conduct the endogeneity test for Top10LLTIO. Hansen’s J-test confirms that at least one
instrument is valid.
We report the IV regression results for the overall sample and rated borrower-only
sample in Columns (1) – (4) of Table 4. The Chi-square statistics for the endogeneity test are
1.62 and 1.07 for the overall and rated borrower-only samples, with p-values of 0.203 and 0.301,
16
Following Bushee’s (1998) definition, quasi-indexer institutional ownership is characterized by low turnover and
diversified holdings. It is the other component of LLTIO. Details of the variables can be found in Appendix A. 17
We similarly construct IVs based on Top10LLTIO for firms in other industries in the same sate, and obtain similar
results: Top10LLTIO is not endogenous, and the LLTIO effect remains.
respectively, suggesting that Top10LLTIO is, indeed, not endogenous at the conventional
significance level. The t-statistics for both instruments are positive and significant at a 1%
confidence level. The F-statistic of joint significance from adding the two IVs is 141.24 and
108.29 for the overall and rated-only samples, respectively, suggesting that neither IV is weak.
The coefficient estimates from the second stage of the IV regression on the instrumented
Top10LLTIO are -0.255 and -0.360 for the overall and rated-only samples, respectively,
significant at the 5% level. The results from the IV regressions suggest that Top10LLTIO leads to
a lower logspread.
[Table 4 about here]
4.3 LLTIO, Relationship Banking, and Soft Information
A previous banking relationship constitutes a source of soft information for the lead bank,
which mitigates information asymmetry and leads to lower loan spreads (Diamond, 1991; Berger
and Udell, 1995; Bharath, Dahiya, Saunders, and Srinivasan, 2011; Boot, 2000). The adverse
selection and moral hazard problems in the banking relationship can be mitigated through a
strong relationship between the lending bank and borrower based on prior lending experiences or
other business ties such as deposit and working capital management (Boot, 2000). Such a
relationship gives the lending bank access to intimate soft information on the borrower, possibly
including information on its business prospects. To investigate the nature of soft information, we
examine whether the LLTIO effect varies depending on the existence of a previous banking
relationship. If the soft information that LLTIOs possess is identical to the soft information that a
lead bank collects from a previous banking relationship, then the LLTIO effect will vanish in the
presence of such a relationship.
We create an indicator variable, Relation dummy, which takes a value of one if there is a
previous banking relationship between the lead bank and borrowing firm, and zero otherwise.
We then interact Relation dummy with Top10LLTIO, and report the results in Column (1) of
Table 5. We find that both Top10LLTIO and Relation dummy carry a negative coefficient
estimate, significant at the 5% level. The coefficient estimate on the interaction term Relation
dummy ×Top10LLTIO is negative and insignificant, whereas the aggregate LLTIO effect when a
previous banking relationship exists is negative, with a coefficient magnitude of (-0.087−0.089 =
-0.176), and significant at a better than 5% level. Analyses using subsamples of borrowing firms
that do and do not have a previous banking relationship with the lead bank also show the LLTIO
effect to remain significant, with a magnitude of -0.178 and -0.073 for the subsamples with and
without such a relationship, respectively. Both coefficients are significant at a better than 5%
confidence level, and the results are reported in Columns (2) and (3) of Table 5. The existence of
the LLTIO effect irrespective of a previous banking relationship suggests that the soft
information that LLTIOs possess differs from that garnered by lenders from a previous banking
relationship. We refer to the soft information that LLTIOs possess as geography-related soft
information.
[Table 5 about here]
4.4 Conditions that Influence the LLTIO Effect
The nature of geography-related soft information and the conditions that influence the
LLTIO effect require empirical exploration. We suggest that convenient, frequent interactions
between locals help alleviate geography-related soft information gap. Furthermore, because the
LLTIOs are outside of the loan syndicate, we argue that the geography-related soft information is
secondary to hard information. In other words, LLTIO is a signal that “hardens” geography-
related soft information and latches onto existing hard information. In sum, we propose that the
LLTIO effect is only salient for a borrowing firm with the necessary hard information in place,
but still has a need for geography-related soft information.
We use two proxies to capture borrowing firms in which the lender’s dealing of soft
information is non-trivial. First, the R&D-to-asset ratio serves as a proxy for the lender’s need
for geography-related soft information at the borrowing firm level because it is difficult to
discern the future prospects of firms with R&D investments (Lorek, Stone, and Willinger, 1999).
Indeed, Cohen, Diether, and Malloy (2013) suggest that the stock market appears unable to
distinguish between “good” and “bad” R&D investments. The presence of the LLTIOs provides
a better monitoring environment with more frequent updates on R&D investments and helps
address information asymmetry. Second, the intangible-to-asset ratio serves as another proxy for
the lender’s need of geography-related soft information. The economic value of a firm’s
intangible assets such as human capital and customer satisfaction is hard to assess from a
distance (Edmans, Heinle, and Huang, 2015). The LLTIOs possess an advantage in conducting
such assessment due to their geographic proximity. We therefore expect the LLTIO effect to be
more salient in borrowing firms with either high R&D investments or high proportion of
intangible assets, as the need for soft information arises from long-term monitoring of the
borrowing firm in both cases (Sufi, 2007).
Credit rating is the most important piece of information about a potential borrower that a
lender can easily obtain (Sufi, 2009), and contains hard information on the borrower’s
creditworthiness. We use the borrowing firm’s credit rating as another proxy to capture
borrowing firms that have necessary hard information in place so that the addition of soft
information is effective. Since borrowers without long-term bond ratings lack the necessary hard
information for geography-related soft information to latch onto, the LLTIO effect should
disappear.
Finally, secured loans can serve as a proxy for the lack of a need for additional
information, as little uncertainty is involved in the event of loan default. We expect the LLTIO
effect to diminish as the geography-related soft information the LLTIOs offer is not relevant
when an information gap does not exist.
The results for the subsamples constructed using the above proxies to explore the
informational effect of the LLTIOs are reported in Panel A of Table 6. Columns (1) and (2)
examine the R&D-to-asset ratio as a measure of the lender’s need for geography-related soft
information. The LLTIO effect is salient only for the subsample in which the borrowing firms
have a positive R&D ratio, with a coefficient estimate that is negative and statistically different
from zero at a 1% level of confidence. This finding suggests that the presence of LLTIOs leads
to lower loan spreads only for borrowing firms that involve R&D, where geography-related soft
information is helpful. Columns (3) and (4) investigate the intangible-to-asset ratio as another
measure of the lender’s need for geography-related soft information. Similarly, the LLTIO effect
is salient only for the subsample in which the borrowing firms have a positive intangible asset
ratio, with a coefficient estimate that is negative and statistically different from zero at a 1%
level of confidence.
Columns (5) and (6) consider a borrowing firm’s long-term credit rating as a measure of
hard information on its creditworthiness. It can be seen that the LLTIO effect is salient only for
the subsample of borrowing firms with a long-term debt rating (rated), with a coefficient
estimate that is negative and statistically different from zero at a 1% level of confidence,
suggesting that the LLTIO effect based on geography-related soft information is secondary to
credit ratings. Columns (7) and (8) examine secured loans as a measure of the need for
geography-related soft information revealing that the LLTIO effect is salient only for the
subsample with unsecured loans, as indicated by the negative coefficient estimate that is
statistically different from zero at a 1% level of confidence. Secured loans remove uncertainty,
and the LLTIO effect exists only when there is information asymmetry.
If the observed LLTIO effect is the result of geography-related soft information, it may
lose its salience when (1) the lead bank is close to the borrowing firm and (2) when the
borrowing firm has an urban location because (1) a lead bank that is geographically proximate
has easy access to the geography-related soft information that LLTIOs possess and (2) a
borrowing firm located in an urban location is subject to greater scrutiny and is better governed,
as managerial investment decisions are easily observable (John, Knyazeva, and Knyazeva,
2011). To examine these two conditions, we create two indicator variables, Close Bank and
Urban10, which takes a value of one if the headquarters of the lead bank that issued the loan is
within a 100-mile radius of the borrowing firm’s corporate headquarters and if the borrowing
firm is located in one of the 10 largest MSAs in the U.S., respectively, and zero otherwise. We
then interact Top10LLTIO with Close Bank and Urban10, and examine the LLTIO effect for
loans (1) with a geographically proximate lead bank and (2) a borrower that is located in one of
the 10 largest MSAs by testing whether the respective sum of coefficient estimates, namely,
(Top10LLTIO + Close Bank × Top10LLTIO) and (Top10LLTIO + Urban10×Top10LLTIO), is
statistically different from zero. We report the results in Columns (1) and (2) of Panel B in Table
6. The coefficient estimates on Top10LLTIO are negative and significant at a better than 5%
level of confidence in both columns, suggesting that the LLTIO effect is salient when the lead
bank is not geographically close to the borrowing firm and when the borrowing firm does not
have an urban location. The coefficient sums, (Top10LLTIO + Close Bank × Top10LLTIO) and
(Top10LLTIO + Urban10 × Top10LLTIO), are both insignificantly different from zero (Chi-
square statistics of 0.49 and 0.88, respectively), suggesting that the LLTIO effect vanishes when
the loan is issued by a lead bank located close to the borrowing firm or when the borrowing firm
is located in a large urban area.
[Table 6 about here]
As LLTIO is a type of equity ownership, we expect the salience of the LLTIO effect to
vary with the likelihood of conflicts of interest between creditors and shareholders. Conflicts of
interest arise when there is a risk of default. Myers (2001) states: “If debt is totally free of default
risk, debtholders have no interest in the income, value or risk of the firm. But if there is a chance
of default, then shareholders can gain at the expense of debt investors. Equity is a residual claim,
so shareholders gain when the value of existing debt falls, even when the value of the firm is
constant” (p. 96). To explore how the LLTIO effect varies with the likelihood of a conflict of
interest, which is driven largely by default risk, we employ two proxies for default risk: whether
the borrowing firm has an investment-grade rating on its long-term debt and whether the loan
was syndicated in the midst of a financial crisis. As the default risk is lower for investment-grade
borrowing firms than for their non-investment-grade counterparts, and lower during non-
financial crisis periods than crisis periods, we expect the LLTIO effect to be more salient in
subsamples with a lower default risk in which risk-driven conflicts of interest are also less likely.
The results for subsamples constructed using the two foregoing proxies are reported in
Table 7. Columns (1)–(2) examine logspread in subsamples of borrowing firms with and without
investment-grade long-term debt, with the LLTIO effect salient only in the former. Columns (3)–
(4) consider logspread in subsamples with loans issued in financial and non-financial crisis
periods, with crisis periods defined as 2000–2002 and 2007–2009. The LLTIO effect is salient
only in the non-crisis subsample. In addition, the control variables for a previous banking
relationship (Relation dummy) and bank reputation (Top3Bank) exert different influences on
logspread in the crisis and non-crisis subsamples.18
Prior research has documented an association
between a lower loan spread and a previous banking relationship (Berger and Udell, 1995;
Petersen and Rajan, 1994) and lender certification (Ross, 2010), whereas here both Relation
dummy and Top3Bank are (negatively) significant only in non-crisis periods. Our conjecture
concerning the differential effects of Relation dummy and Top3Bank on logspread is that loans
are in greater demand during crisis periods, as concerned borrowers prepare for a potential
liquidity squeeze and lenders systematically increase loan spreads to compensate for the greater
default risk (Santos, 2011). In this extreme environment, the benefits of lender certification and
relationship lending diminish, putting pressure on prospective borrowers. The differential LLTIO
effect that we report in Tables 6 and 7 also suggests that it is unlikely due to the dual role of the
institutional investor as both a shareholder and a creditor.
[Table 7 about here]
4.5 Within-Syndicate LLTIO Effect
We also investigate whether the LLTIO effect exists within a loan syndicate, as the lead
bank in a syndicate essentially serves as a half-agent for the other participating banks, and there
is thus information asymmetry between the lead and participating banks (Sufi, 2007). As
monitoring is not observable, lead banks can shirk from their duties, with the other banks
possibly bearing the full cost of such shirking. To explore whether the presence of LLTIOs at the
borrower level alleviates the severity of within-syndicate information asymmetry, we investigate
18
Because our Investment Grade dummy variable is defined in a way that includes non-rated firms and below BBB-
rated firms, the differential effects of pos_relation1 and Top3Bank with respect to Investment Grade are unclear.
whether the syndicate structure changes with such presence. As Sufi (2007) shows, more severe
information asymmetry problems force a lead bank to take a larger stake in a loan. If the LLTIO
effect alleviates information asymmetry within the syndicate, we expect a negative relation
between Top10LLTIO and LeadShare, that is, the stake held by the lead bank. In unreported
results with LeadShare as the dependent variable, and after controlling for firm-, loan-, and
macroeconomy-level characteristics and industry effects, we find LLTIOs to have a negative,
albeit non-significant, effect on LeadShare. We therefore do not find empirical support for the
existence of a within-syndicate LLTIO effect.
4.6 Mechanisms of the LLTIO Effect
As information asymmetry leads to both adverse selection and moral hazard problems,
the mechanisms driving the LLTIO effect could serve to (1) alleviate the ex ante information
asymmetry associated with adverse selection problems, (2) alleviate the ex post information
asymmetry associated with moral hazard problems, or (3) alleviate both. Even though we argue
that better monitoring is likely the reason for the LLTIO effect, we use the implementation of
Regulation FD and SOX as natural experiments to examine whether the LLTIO effect is
associated with less severe adverse selection problems or moral hazard problems, respectively.
The use of natural experiments helps mitigate concerns over the results being driven by
endogenous factors. By promoting the full and fair disclosure at public companies, Regulation
FD has greatly reduced the informational advantage of institutional investors (Cornett,
Tehranian, and Yalcin, 2007). By imposing higher standards of corporate governance, SOX has
encouraged institutional activism through changes in legal and regulatory standards (Gillan,
2006).
Although we are aware that the implementation of both Regulation FD and SOX has
affected all institutional investors, we argue that their marginal effects are greater for local
institutions. First, the information advantage enjoyed by local institutions is widely documented
(see, for example, Malloy, 2005) but yet to be established for general institutions.19
Second, we
demonstrate in Section 4.1 that it is geographic proximity that drives the LLTIO effect. When
reacting to a stronger institutional monitoring environment, however, local institutions are more
likely to monitor because the cost of doing so is lower (Kang and Kim, 2008).
We adopt a difference-in-difference (DiD) approach that captures the incremental effect
of Regulation FD and SOX implementation to examine the mechanisms through which the
LLTIO effect operates. If it relies on the mechanism of alleviating adverse selection problems,
we expect it to weaken and possibly disappear following Regulation FD implementation. If, in
contrast, the LLTIO effect relies on the mechanism of alleviating moral hazard problems, we
expect it to strengthen in the post-SOX implementation period.
We define an indicator variable, post Regulation FD, which is set to one for the years
after 1999 (as Regulation FD was implemented in October 2000), and zero otherwise. Our other
indicator variable, post SOX, is set to one for the years after 2001 (as SOX was implemented in
July 2002), and zero otherwise. We also include the interaction terms Top10LLTIO×post
Regulation FD and Top10LLTIO×post SOX as additional variables in the respective baseline
specifications. After controlling for firm-, loan-, and macroeconomic characteristics and industry
effects, we compare the LLTIO effect over the two-year subsamples before and after Regulation
FD implementation (1998–1999 to 2000–2001), over the two-year subsamples before and after
19
As Coval and Moskowitz (2001) point out, prior studies of mutual fund managers and pension fund managers
report that, if anything, their investors consistently underperform the market and other passive benchmark portfolios.
See, for example, Carhart (1997), Chevalier and Ellison (1999), and Lakonishok, Shleifer, and Vishny (1992), to
name a few.
SOX implementation (2000–2001 to 2002–2003), and over 1998–1999 to 2002–2003. Our
expectation is that if a due diligence mechanism that alleviates adverse selection problems exists,
the LLTIO effect will be weaker over the 2000–2001 period than over the 1998–1999 period,
whereas if a monitoring mechanism that alleviates moral hazard problems exists, that effect will
be weaker over 2000–2001 than over 2002–2003. Finally, comparison of the LLTIO effect over
1998–1999 and 2002–2003 is informative for comparing the due diligence and monitoring
mechanisms.
The estimation results are reported in Table 8, with Column (1) showing the effect of
Regulation FD implementation by comparing 1998–1999 and 2000–2001. The coefficient
estimate on Top10LLTIO is negative and significant with a magnitude of -0.1, whereas that on
Top10LLTIO×post Regulation FD is positive and significant, with a magnitude of 0.14, both at a
confidence level of 1%, suggesting that the LLTIO effect vanished after the implementation of
Regulation FD. This provides evidence for the existence of an adverse selection-alleviating
mechanism (due diligence mechanism).
Column (2) of Table 8 examines the effect of SOX implementation by comparing 2000–
2001 and 2002–2003. The coefficient estimate on the interaction term Top10LLTIO ×Post SOX is
negative and significant, with a confidence level of 1%, suggesting that the LLTIO effect became
more salient after the implementation of SOX. This provides evidence for the existence of a
moral hazard-alleviating mechanism (monitoring mechanism). Finally, Column (3) examines the
relative importance of the two mechanisms by comparing 2002–2003 and 1998–1999. The
coefficient estimates for both Top10LLTIO and the interaction term Top10LLTIO ×Post SOX are
negative and significant, suggesting that the monitoring mechanism is stronger than the due
diligence mechanism.
To check for the robustness of both mechanisms, we vary the two-year period before and
after the implementation of Regulation FD and SOX used in the DiD regression, respectively.
We do not observe the same significant results when the two-year period before and after
Regulation FD changes to 2001-2002. The significant decrease in logspread associated with the
interaction term Top10LLTIO ×Post SOX remains as we vary the cutoff months of the two-year
period in different ways. The highly robust results on the monitoring mechanism suggest that the
monitoring function of the LLTIOs is more likely the factor that drives the LLTIO effect.20
[Table 8 about here]
4.7 LLTIO’s Monitoring Role
We next explore further evidence for the monitoring role of the LLTIOs and the possible
channels for such monitoring function. As Chhaochharia, Kumar and Niessen-Ruenzi (2012)
document a number of empirical evidence suggesting that geographic proximity of the
institutional investors is associated with improved corporate governance, we investigate direct
evidence to establish the causal link between LLTIO and corporate governance using our loan
sample. Our proxy for internal governance weakness is “lucky” option grants to CEOs and
directors, where lucky option grants are defined as options granted at the lowest stock price of
the month. Bebchuk, Grinstein, and Peyer (2010) show that the opportunistic timing of option
grants reflects internal governance weakness and that lucky option grants to CEOs and directors
suggest weak monitoring. If the LLTIOs monitor, we expect to observe a lower likelihood of
lucky CEO option grant. We run a pooled regression with clustered standard errors using the data
on lucky CEO option granting21
over the sample period of 1996-2005 and find that this is indeed
20
We also vary the two-year periods to ensure that our results are not sensitive to exclusion of the years in which the
two regulations took effect. These results are available upon request. 21
We thank Professor Bebchuk for providing the data on his website at
http://www.law.harvard.edu/faculty/bebchuk/data.shtml.
the case. The results are displayed in Column (1) of Panel A, Table 9. Here the dependent
variable is a dummy variable indicating whether a CEO grant event was lucky and Top10LLTIO
is the independent variable of interest. We control for Lucky director, which is a dummy variable
indicating whether an independent director grant event was lucky, and other variables like
Top10IO, Top10LSTIO, as well as a number of firm characteristics, including firm size, S&P 500
membership, leverage, profitability, R&D, etc. The coefficient estimate on Top10LLTIO is
negative and significant with a confidence level of 5%, consistent with our conjecture that the
LLTIOs monitor and reduce the likelihood of a lucky CEO grant event.
Bebchuk, Cohen, and Ferrell (2009) also show that the entrenchment index (E-index),
which is based on six out of the twenty-four provisions that are included in the G-index
(Gompers, Isshi and Metrick, 2003), is monotonically associated with economically significant
reduction in firm value. If LLTIOs monitor, we expect to observe a negative relation between
LLTIO and the level of E-index. To test our conjecture, we run three pooled regressions, with E-
index and G-index22
as the dependent variable, respectively, and with Top10LLTIO as the
independent variable of interest. Whether we use OLS or ordered Probit models, the coefficient
estimates of Top10LLTIO are negative and significant, with a confidence level of at least 5%,
when the dependent variable is E-index, as shown in Columns (2)-(3) of Table 9. The coefficient
estimate of Top10LLTIO is insignificant, as shown in Column (4) of Table 9, when the
dependent variable is G-index. These results suggest that the LLTIOs monitor material
governance provisions that matter for firm valuation.
[Table 9 about here]
4.8 Propensity Score Matching Analysis
22
E-index and G-index are constructed following Bebchuk, Cohen, and Ferrell (2009) and Gompers, Isshi, and
Metrick (2003), respectively. We thank Professor Bebchuk for providing the E-index on his website at:
http://www.law.harvard.edu/faculty/bebchuk/data.shtml.
As can be observed from Panel A of Table 2, there are significant differences in most
firm and loan characteristics between loans with high and low levels of Top10LLTIO. To ensure
that the LLTIO effect that we have documented is not driven by firm or loan characteristics, we
match loans using borrowing firm characteristics and loan characteristics based on propensity
scores, and then compare the loan spreads based only on one variable for the matched samples:
whether Top10LLTIO is higher than 5% (high Top10LLTIO) or not (low Top10LLTIO). The
results indicate that the LLTIO effect continues to hold for these matched loans.
We conduct propensity matching using a logit model with the following borrowing firm
characteristics: S&P 500 index membership dummy, overall institutional ownership, stake held
by the 10 largest institutional owners, firm size, leverage usage, Tobin’s Q, R&D-to-asset ratio,
ROA, dividend dummy, asset tangibility, and cash flow volatility. The model also incorporates
the following loan characteristics: previous banking relationship with lead bank, loan originated
by one of the top-three banks in loan syndication, secured loan dummy, short-term revolver
dummy, long-term revolver dummy, term loan dummy, other loan dummy, loan maturity,
investment grade dummy, term spread, credit spread, various dummies for loan purposes, and
Fama-French twelve-industry categorization. Based on the closeness of their propensity scores,
we select the nearest syndicated loan-firm observation with similar (matched) characteristics and
compare mean spread and logspread based on one variable: whether the level of Top10LLTIO is
above 5% or not. We then conduct the same exercise choosing from the three nearest syndicated
loan-firm observations, and compare the difference in spread and logspread with respect to
Top10LLTIO. The propensity matching results reported in Table 10 show that we are able to
match a group of firms than resemble one another within an allowed error margin (caliper) of
0.05. The spread for the high Top10LLTIO (Top10LLTIO ≥5%) group is 31.285 basis points
lower than that for the low Top10LLTO (Top10LLTIO <5%) group before matching. The
magnitude of the difference in spread after matching using different criteria ranges from -6.650
to -7.437 basis points, remaining negative and significant at a confidence level of 1%. The
magnitude of the difference in logspread after matching using different criteria is about 1.06
basis points, remaining negative and significant at a confidence level of 1%. The difference in
economic significance with spread and logspread may be due to outlier effect, which is more
drastic with spread.
[Table 10 about here]
5. Conclusion
We propose geographically proximate institutional ownership with a large stake in a
borrowing firm as a proxy for exogenous, informed ownership that serves as an informative
signal. Using this proxy, we test the theoretical arguments in Leland and Pyle (1977) and
Holmstrom (1979), and provide empirical support for the proposition that informed ownership
provides an informative and valuable signal in the context of syndicated loan contracting. We
find that long-term and not short-term institutional ownership is associated with lower loan
spreads at local borrowing firms. We show that the LLTIO effect is only salient when
geography-related soft information helps to reduce information asymmetry in the syndication
process and when there is a lack of conflicts of interest between creditors and shareholders. This
result is robust to controlling for firm characteristics and loan contracting terms.
We also show that the monitoring function is the main mechanism that drives the LLTIO
effect. We provide empirical evidence for the LLTIO’s monitoring role, which explains the
spread discount that the lead lender is willing to give. Future studies could examine how the
LLTIOs exercise their monitoring function and how geographic proximity changes the cost-
benefit analysis for these institutions in more detail.
References
Agarwal, S. and R. Hauswald, 2010, “Distance and Private Information in Lending,” Review of
Financial Studies 23(7), 2757-2788.
Baik, B., J. Kang, and J. Kim, 2010, “Local Institutional Investors, Information Asymmetries,
and Equity Returns,” Journal of Financial Economics 97, 81-106.
Bebchuk, L.A., A. Cohen, and A. Ferrell, 2009, “What Matters in Corporate Governance,”
Review of Financial Studies 22, 783-827.
Bebchuk, L.A., Y. Grinstein, and U. Peyer, 2010, “Lucky CEOs and Lucky Directors,” Journal
of Finance 65, 2363-2401.
Berger, A. and G. Udell, 1995, “Relationship Lending and Lines of Credit in Small Firm
Finance,” Journal of Financial and Quantitative Analysis 26, 83-95.
Bharath, S.T., S. Dahiya, A. Saunders, and A. Srinivasan, 2011, “Lending Relationships and
Loan Contract Terms,” Review of Financial Studies 24, 1141-1203.
Boot, A. 2000. “Relationship Banking: What Do We Know?” Journal of Financial
Intermediation 9, 7–25.
Bushee, B. J., 1998, “The Influence of Institutional Investors on Myopic R&D Investment
Behavior,” Accounting Review 73, 305-333.
Bushee, B. J., 2001, “Do Institutional Investors Prefer Near-Term Earnings over Long-Run
Value?” Contemporary Accounting Research 18, 207-246.
Carey, M., M. Post, and S. Sharpe, 1998, “Does Corporate Lending by Banks and Finance
Companies Differ? Evidence on Specialization in Private Debt Contracting,” Journal of Finance
53, 845-878.
Carhart, M., 1997, “On Persistence in Mutual Fund Performance,” Journal of Finance 52,
57–82.
Chava, S. and M.R. Roberts, 2008, “How Does Financing Impact Investment? The Role of Debt
Covenants,” Journal of Finance 63, 2085-2121.
Chevalier, J. and G. Ellison, 1999, “Are Some Mutual Fund Managers Better than Others?
Cross-Sectional Patterns in Behavior and Performance,” Journal of Finance 54, 875–99.
Chen, X., J. Harford, and K. Li, 2007, “Monitoring: Which Institutions Matter?” Journal of
Financial Economics 86, 279-305.
Chen, H., P. Gompers, A. Kovner, and J. Lerner, 2010, “Buy Local? The Geography of Venture
Capital.” Journal of Urban Economics 67 (2010), 90-102.
Chhaochharia, V., A. Kumar, and A. Niessen-Ruenzi, 2012, “Local Investors and Corporate
Governance,” Journal of Accounting and Economics 54, 42-67.
Chui, M., D. Domanski, P. Kugler, and J. Shek, 2010. “The Collapse of International Bank
Finance During the Crisis: Evidence from Syndicated Loan Markets,” BIS Quarterly Review,
September, 39–49.
Cohen, L., K. Diether, and C. Malloy, 2013, “Misvaluing Innovation,” Review of Financial
Studies 26, 635-666.
Cornett, M. M., H. Tehranian, and A. Yalcin, 2007, “Regulation Fair Disclosure and the
Market’s Reaction to Analyst Investment Recommendation Changes,” Journal of Banking &
Finance 31, 567-588.
Coval, D. J., and T. J. Moskowitz, 2001, “The Geography of Investment: Informed Trading and
Asset Prices,” Journal of Political Economy 109, 811-841.
Coval, J. D., and T. J. Moskowitz, 1999, “Home Bias at Home: Local Equity Preference in
Domestic Portfolios,” Journal of Finance 54, 2045-2073.
Cumming, D. and N. Dai, 2010, “Local bias in venture capital investments,” Journal of
Empirical Finance 17, 362 – 280.
Degryse, H. and S. Ongena, 2005, “Distance, Lending Relationships, and Competition,” Journal
of Finance 60, 231-266.
Demsetz, H., and K. Lehn, 1985, “The Structure of Corporate Ownership: Causes and
Consequences,” Journal of Political Economy 93, 1155-1177.
Dennis, S. and D. J. Mullineaux, 2000, “Syndicated Loans,” Journal of Financial Intermediation
9, 404-426.
Diamond, D., 1984, “Financial Intermediation and Delegated Monitoring,” Review of
Economics Studies 51, 393-414.
Diamond, D., 1991, “Monitoring and Reputation: The Choice between Bank Loans and Directly
Placed Debt,” Journal of Political Economy 99, 689-721.
Edmans, A., M. Heinle, and C. Huang, 2015, “The Real Cost of Financial Efficiency When
Some Information is Soft,” working paper, University of Pennsylvania.
Fombrun, C. and M. Shanley, 1990, “What’s in a Name? Reputation Building and Corporate
Strategy,” The Academy of Management Journal 33, 233-258.
Gaspar, J., and M. Massa, 2007, “Local Ownership as Private Information: Evidence on the
Monitoring-Liquidity Trade-Off,” Journal of Financial Economics 83, 751-792.
Gaspar, J., M. Massa, and P. Matos, 2005, “Shareholder Investment Horizons and the Market for
Corporate Control,” Journal of Financial Economics 76, 135-165.
Gillan, S. L., 2006, “Recent Development in Corporate Governance: an Overview,” Journal of
Corporate Finance 12, 381-402.
Gompers, P, J. Ishii, and A. Metrick, 2003, “Corporate Governance and Equity Prices,”
Quarterly Journal of Economics 118, 107-155.
Graham, J. R., S. Li, and J. Qiu, 2008, “Corporate Misreporting and bank loan contracting,”
Journal of Financial Economics 89, 44-61.
Harford, J., A. Kecskes, and S. Mansi, “Do Long-term Investors Improve Corporate Decision
Making?” Working paper, University of Washington.
Hartzell, J., and L. Starks, 2003, “Institutional Investors and Executive Compensation,” Journal
of Finance 58, 2351-2374.
Holmstrom, B., 1979, “Moral Hazard and Observability,” The Bell Journal of Economics 10(1),
74-91.
Holmstrom, B., and J. Tirole, 1997, “Financial intermediation, loanable funds, and the real
sector,” Quarterly Journal of Economics 112, 663–691.
Ivashina, V. 2009, “Asymmetric information effects on loan spreads,” Journal of Financial
Economics 92, 300-319.
Ivkovic, Z. and S. Weisbenner, 2005, “Local does as local is: information content of the
geography of individual investors’ common stock investments,” Journal of Finance 60, 267-306.
Jensen, M. and W. H. Meckling, 1976, “Theory of the Firm: Managerial Behavior, Agency Costs
and Ownership Structure,” Journal of Financial Economics 3(4), 305-360.
John, K., A. Knyazeva, and D. Knyazeva, 2011, “Does Geography Matter? Firm Location and
Corporate Payout Policy,” Journal of Financial Economics 101, 533-551.
Kang, J., and J. Kim, 2008, “The Geography of Block Acquisitions,” Journal of Finance 63,
2817-2858.
Knyazeva, A.and D. Knyazeva, 2012, “Does Being Your Bank’s Neighbor Matter?” Journal of
Banking and Finance 36, 1194-1209.
Knyazeva, A., D. Knyazeva, D., and R.W. Masulis, 2013, “The supply of corporate directors and
board independence,” Review of Financial Studies 26, 1561-1605.
Krishnaswami, S., P.A. Spindt, and V. Subramaniam, 1999, “Information Asymmetry,
Monitoring, and the Placement Structure of Corporate Debt,” Journal of Financial Economics 51,
407-434.
Lakonishok, J., A. Shleifer, and R. W. Vishny, 1992, “The Structure and Performance of the
Money Management Industry,” Brookings Papers Econ. Activity: Microeconomics, 339–79.
Leland, H.E. and D. H. Pyle, 1977, “Informational Asymmetries, Financial Structure, and
Financial Intermediation,” Journal of Finance 32, 371-387.
Lorek, K.S., M. S. Stone, and G. L. Willinger,1999, “The Differential Predictive Ability of
Opaque and Transparent Firms’ Earnings Numbers,” Quarterly Journal of Business and
Economics 38, 3-19.
Liberti, J., A. Mian, 2009, “Estimating the Effect of Hierarchies on Information Use,” Review of
Financial Studies 22, 4057-4090
Loughran, T. and P. Schultz, 2005, “Liquidity: Urban versus Rural Firms,” Journal of Financial
Economics 78, 341-374.
Malloy, C., 2005, “The Geography of Equity Analysis,” Journal of Finance 60(2), 719-755.
Merton, R. C., 1974, “On the Pricing of Corporate Debt: The Risk Structure of Interest Rates,”
Journal of Finance 2, 449-470.
Myers, S. C., 1977, “Determinants of Corporate Borrowing,” Journal of Financial Economics 5,
147-175.
Myers, S. C., 2001, “Capital Structure,” Journal of Economic Perspectives 15, 81-102.
Petersen, M., 2004. “Information: Hard and Soft,” Working paper, Northwestern University and
NBER.
Petersen, M.A., R. Rajan, 1994. The benefits of lending relationships: evidence from small
business data. Journal of Finance 49, 3–37.
Petersen, M., and R. Rajan, 2002, “Does Distance Still Matter? The Information Revolution in
Small Business Lending,” Journal of Finance 57, 2533–2570.
Rajan, R., 1992, “Insiders and Outsiders: The Choice between Informed and Arm’s-Length
Debt,” Journal of Finance 47, 1367–1400.
Rajan, U., A. Seru, and V. Vig, 2015, “The Failure of Models that Predict Failure: Distance,
Incentives, and Defaults,” Journal of Financial Economics 115, 237-260.
Ross, D., 2010, “The ‘Dominant Bank Effect’: How High Lender Reputation Affects the
Information Content and Terms of Bank Loans”, Review of Financial Studies 23, 2730-2756.
Santos, 2011, “Bank Loan Pricing Following the Subprime Crisis,” Review of Financial
Studies 24, 1916-1943.
Schenone, C., 2010, “Lending Relationships and Information Rents: Do Banks Exploit Their
Information Advantages?”, Review of Financial Studies 23, 1149-1199.
Sharpe, S. A., 1990, “Asymmetric Information, Bank Lending, and Implicit Contracts: A
Stylized Model of Customer Relationships,” Journal of Finance 45, 1069-1087.
Sias, R.W., 1996, “Volatility and the Institutional Investor”, Financial Analysts Journal 52, 13-
20.
Stein, J., 2002, “Information Production and Capital Allocation: Decentralized versus
Hierarchical Firms,” Journal of Finance 57, 1891-1921.
Stulz, R.M., 1999, “Globalization, Corporate Finance, and the Cost of Capital”, Journal of
Applied Corporate Finance 12, 8-25.
Sufi, A., 2007, “Information Asymmetry and Financing Arrangements: Evidence from
Syndicated Loans,” Journal of Finance 62, 629-668.
Sufi, A., 2009, “Bank Lines of Credit in Corporate Finance: An Empirical Analysis,” Review of
Financial Studies 22, 1057-1088
Weidner, D., 2000, “Syndicated Lending Closes Out ’90s on a Tear,” The American Banker, Jan.
10.
Yan, X., and Z. Zhang, 2009, “Institutional Investors and Equity Returns: Are Short-term
Institutions Better Informed?” Review of Financial Studies 22, 832-924.
Yi, H. and D. Mullineaux, 2006, “The Informational Role of Bank Loan Ratings,” Journal of
Financial Research 29, 481 - 501.
Appendix A. Variable definitions
Variable Name Definitions and measurements Source
Bank loan spread Spread Initial all-in-drawn spread over LIBOR DealScan
Log (loan spread) Logspread Log of Initial all-in-drawn spread over LIBOR DealScan
S&P500 dummy S&P500 Takes 1 if a firm belongs to S&P500, else 0 Compustat
Urban10 Urban10 Takes 1 if a belongs to the top 10 urban areas, else 0
Close bank Close bank
Takes 1 if lenders and borrowers are located within 100
miles, else 0
Compustat,
DealScan
Relation Relation Number of loan experiences from the same bank DealScan
Relation dummy Relation dummy
Takes 1 if a borrower has borrowed from the same bank
before, else 0 DealScan
Top10 Institutional Ownership Top10 IO
# of shares held by largest ten (measured by stakes in the
borrowing firm) institutional investors / # of total shares
outstanding
Thompson
Reuters 13F
Top10 Local Institutional
Ownership Top10 Local IO
Ownership by the largest ten (measured by stakes in the
borrowing firm) institutional investors that have
headquarters within 100 miles from headquarters of the
borrowing firm
Thompson
Reuters 13F
Top10 short-term institutional
ownership Top10 STIO
Ownership by the largest ten (measured by stakes in the
borrowing firm) institutional investors that are identified
as belonging to the transient type by Bushee (1998).
Thompson
Reuters 13F
Top10 long-term institutional
ownership Top10 LTIO
Ownership by the largest ten (measured by stakes in the
borrowing firm) institutional investors that are identified
as belonging to either the dedicated or quasi-indexer type
by Bushee (1998).
Thompson
Reuters 13F
Top10 local short-term
institutional ownership Top10 LSTIO
Ownership by the largest ten (measured by stakes in the
borrowing firm) institutional investors that are identified
as belonging to the transient type by Bushee (1998) and
have headquarters within 100 miles from headquarters of
the borrowing firm
Thompson
Reuters 13F
Top10 local dedicated
institutional ownership Top10LDIO
Ownership by the largest ten (measured by stakes in the
borrowing firm) institutional investors that are identified
as belonging to the dedicated type by Bushee (1998) and
have headquarters within 100 miles from headquarters of
the borrowing firm
Thompson
Reuters 13F
Top10 local quasi-indexer
institutional ownership Top10LQIO
Ownership by the largest ten (measured by stakes in the
borrowing firm) institutional investors that are identified
as belonging to quasi-indexer type by Bushee (1998) and
have headquarters within 100 miles from headquarters of
the borrowing firm
Thompson
Reuters 13F
Top10 local long-term
institutional ownership Top10 LLTIO Top10LLTIO=Top10LDIO+Top10LQIO
Thompson
Reuters 13F
Total Assets TA at Compustat
Log (Total Assets) LogTA Log(at) Compustat
Leverage Leverage Total debt / TA Compustat
Tobin’s Q Tobin’s Q Market value of assets / Book value of assets Compustat
Return on Assets ROA Net Income / TA, ni/at Compustat
R&D/Total Assets R&D/TA Xrd/at Compustat
Intangible Assets /Total Assets Intangible/TA Intan/at Compustat
Dividend dummy Div dummy Takes 1 if a firm pays dividend, else 0 Compustat
Net Fixed Assets/Total Assets NFA/TA Ppent/at Compustat
Standard deviation of cash
flows STD CF Standard deviation of previous 5 year cash flows Compustat
Secured loan Secured loan Takes 1 if loan is secured, else 0 DealScan
Short-term revolver loan St revolver Takes 1 if loan is short-term revolver, else 0 DealScan
Long-term revolver loan Lt revolver Takes 1 if loan is long-term revolver, else 0 DealScan
Term loan Term loan Takes 1 if loan is term loan, else 0 Deal Scan
Other loan Other loan Takes 1 if loan is other loan, else 0 Deal Scan
Loan Maturity Maturity Maturity of loans, expressed in months Deal Scan
Long-term bond credit rating LT CR rating 1 (CCC-) to 22 (AAA) Compustat
Investment grade Invgrade
Takes 1 if a company’s S&P long-term credit rating is
BBB- and above, else 0 (0 includes not rated firms) Compustat
Investment grade2 Invgrade2
Takes 1 if a company’s S&P long-term credit rating is
BBB- and above and takes 0 if the long-term rating is
below BBB- (0 does not include not rated firms) Compustat
Rated status Rated
Takes 1 if a company has S&P long-term credit rating,
else 0 Compustat
Term spread Term spread Annual term spread (10 year – 1 year Tbond spread) FED
Credit spread Credit spread Annual credit spread (CCC – AAA corporate bond) FED
Lucky CEO Lucky CEO
Takes 1 when options to the CEO are granted at the
lowest stock price of the month, else zero.
Bebchuk,
Grinstein,
and Peyer
(2010)
Lucky director Lucky director
Takes 1 when options to directors are granted at the
lowest stock price of the month, else zero.
Bebchuk,
Grinstein,
and Peyer
(2010)
E-index E-index Governance index (composed of 6 items)
Bebchuk,
Cohen, and
Ferrell
(2009)
G-index G-index Governance index (composed of 24 items)
Gompers,
Ishii, and
Metrick
(2003).
Table 1. Summary Statistics
Table 1 reports summary statistics for our sample over the period of 1995-2009. An institutional owner is defined as
“local” if the headquarters of the institution is within a 100-mile radius of the company’s headquarters. Annual
Compustat data are matched to Thomson Reuters DealScan data using the identifiers of Chava and Roberts (2008).
We exclude securities with share codes different from 10 or 11, financial and utilities companies, borrowers
incorporated or headquartered outside the U.S., loans originated outside of the U.S., loans denominated in foreign
currencies, loans with benchmark rates other than the LIBOR, and observations with missing data. The sample
includes 17,308 firm-loan observations and the list of variable definitions and measurements is shown in Appendix
A.
Variable N Mean Median SD Min Max
Spread 17308 181.509 165.000 131.781 2.700 1500.000
Logspread 17308 4.894 5.106 0.867 0.993 7.313
S&P500 17308 0.201 0.000 0.401 0.000 1.000
Urban10 17308 0.326 0.000 0.469 0.000 1.000
Close bank 6983 0.036 0.000 0.186 0.000 1.000
Relation 17308 0.490 0.000 0.591 0.000 4.000
Relation dummy 17308 0.442 0.000 0.497 0.000 1.000
Top3bank 17308 0.394 0.000 0.489 0.000 1.000
IO 17308 0.577 0.616 0.265 0.003 1.000
Top10 IO 17308 0.288 0.300 0.159 0.000 0.660
Top10 STIO 17308 0.072 0.052 0.072 0.000 0.337
Top10 LTIO 17308 0.215 0.212 0.137 0.000 0.580
Top10LSTIO 17308 0.020 0.000 0.061 0.000 0.390
Top10LLTIO 17308 0.069 0.000 0.144 0.000 0.745
Top10LQIO 17308 0.052 0.000 0.114 0.000 0.615
TA (million U$) 17308 3602 787 7937 24 55272
LogTA 17308 6.764 6.667 1.710 2.827 10.893
Leverage 17308 0.309 0.286 0.212 0.000 1.016
Tobin's Q 17308 1.745 1.438 0.995 0.699 6.565
ROA 17308 0.029 0.040 0.094 -0.445 0.239
R&D/TA 17308 0.015 0.000 0.034 0.000 0.196
Intangible/TA 15471 0.193 0.132 0.195 0.000 0.776
Div dummy 17308 0.486 0.000 0.500 0.000 1.000
NFA/TA 17308 0.325 0.269 0.229 0.013 0.901
STD CF 17308 0.046 0.028 0.058 0.003 0.386
Secured loan 17308 0.532 1.000 0.499 0.000 1.000
St revolver 17308 0.123 0.000 0.329 0.000 1.000
Lt revolver 17308 0.559 1.000 0.497 0.000 1.000
Term loan 17308 0.261 0.000 0.439 0.000 1.000
Maturity 17308 48.226 59.000 24.337 1.000 264.000
Modified Z 17308 1.701 1.740 1.689 -73.295 5.308
Invgrade 17308 0.235 0.000 0.424 0.000 1.000
Invgrade2 8620 0.472 0.000 0.499 0.000 1.000
Rated 17308 0.498 0.000 0.500 0.000 1.000
Term spread 17308 0.811 0.490 0.856 -0.410 2.830
Credit spread 17308 0.882 0.810 0.349 0.550 3.380
Repay purpose 17308 0.195 0.000 0.396 0.000 1.000
CP backup purpose 17308 0.073 0.000 0.260 0.000 1.000
Working capital purpose 17308 0.182 0.000 0.386 0.000 1.000
Buyback purpose 17308 0.011 0.000 0.106 0.000 1.000
Takeover purpose 17308 0.198 0.000 0.399 0.000 1.000
LBO purpose 17308 0.042 0.000 0.201 0.000 1.000
Project purpose 17308 0.006 0.000 0.074 0.000 1.000
Others purpose 17308 0.031 0.000 0.174 0.000 1.000
Table 2. Univariate Tests
Panel A. Loan spread by High vs. Low Top10 Local Long-Term and Short-Term Institutional
Owners (Top10LLTIO and Top10LSTIO)
Panel A reports results from a univariate comparison of firm and loan characteristics between borrowing firms with
high and low Top10LLTIO and Top10LSTIO sub-samples. High Top10LLTIO (Top10LSTIO) is a sample with 5%
and above of Top10LLTIO (Top10LSTIO), else firm belongs to Low Top10LLTIO (Top10LSTIO) sample.
Top10LLTIO Difference Top10LSTIO Difference
Variables High Low (low-high) High Low (low-high)
Spread 158.875 190.160 31.285*** 176.771 182.154 5.383*
Logspread 4.694 4.971 0.277*** 4.909 4.892 -0.017
S&P500 0.285 0.169 -0.116*** 0.179 0.204 0.025***
Urban10 0.440 0.283 -0.157*** 0.478 0.306 -0.173***
Close bank 0.098 0.023 -0.047*** 0.095 0.029 -0.066***
Relation 0.523 0.477 -0.046*** 0.507 0.487 -0.020
Relation dummy 0.463 0.434 -0.029*** 0.448 0.441 -0.007
Top3bank 0.454 0.371 -0.083*** 0.409 0.392 -0.017
IO 0.607 0.566 -0.041*** 0.614 0.572 -0.041***
Top10 IO 0.313 0.279 -0.034*** 0.316 0.284 -0.031***
Top10 STIO 0.067 0.073 0.006*** 0.111 0.066 -0.045***
Top10LTIO 0.245 0.204 -0.041*** 0.202 0.217 0.015***
Top10LSTIO 0.045 0.010 -0.036*** 0.155 0.001 -0.153***
Top10LLTIO 0.245 0.002 -0.244*** 0.164 0.056 -0.108***
LogTA 7.017 6.667 -0.350*** 6.681 6.775 0.094**
Leverage 0.289 0.317 0.029*** 0.296 0.311 0.016***
Tobin's Q 1.883 1.693 -0.190*** 1.937 1.719 -0.218***
ROA 0.035 0.027 -0.008*** 0.033 0.029 -0.004*
R&D/TA 0.019 0.014 -0.005*** 0.019 0.015 -0.004***
Intangible/TA 0.222 0.182 -0.040*** 0.219 0.190 -0.030***
Div dummy 0.543 0.464 -0.079*** 0.394 0.498 0.104***
NFA/TA 0.267 0.348 0.080*** 0.260 0.334 0.074***
STD CF 0.042 0.048 0.006*** 0.053 0.045 -0.007***
Secured loan 0.446 0.564 0.118*** 0.539 0.531 -0.008
St revolver 0.159 0.110 -0.049*** 0.112 0.125 0.012*
Lt revolver 0.560 0.558 -0.002 0.568 0.557 -0.011
Term loan 0.234 0.272 0.038*** 0.263 0.261 -0.002
Maturity 46.276 48.972 2.695*** 48.44 48.20 -0.245
Modified Z 1.770 1.674 -0.095*** 1.607 1.714 0.106**
Invgrade 0.301 0.210 -0.092*** 0.192 0.241 0.049***
Invgrade2 0.586 0.426 -0.160*** 0.408 0.480 0.072***
Rated 0.514 0.492 -0.022*** 0.470 0.502 0.032***
Panel B. Borrowers with Investment Graded- vs. Non-Investment Graded Borrowers
Panel B reports results from a univariate comparison of firm and loan characteristics between borrowing firms with
investment grade and non-investment grade long-term bonds. Borrowing firms with non-investment grade bonds
include non-rated firms.
Variables Investment grade (1) Non-investment grade (0) Difference
(N=4066) (N=13242) ( 0 – 1)
Spread 78.815 213.042 134.227***
Logspread 3.976 5.176 1.200***
S&P500 0.683 0.053 -0.631***
Urban10 0.353 0.318 -0.034***
Close bank 0.044 0.033 -0.010*
Relation 0.672 0.434 -0.238***
Relation dummy 0.581 0.399 -0.182***
Top3bank 0.598 0.332 -0.266***
IO 0.679 0.546 -0.133***
Top10 IO 0.309 0.282 -0.027***
Top10STIO 0.049 0.078 0.029***
Top10LTIO 0.259 0.202 -0.057***
Top10LSTIO 0.012 0.022 0.010***
Top10LLTIO 0.078 0.066 -0.012***
LogTA 8.651 6.184 -2.467***
Leverage 0.273 0.321 0.048***
Tobin's Q 1.885 1.703 -0.182***
ROA 0.056 0.021 -0.035***
R&D/TA 0.017 0.015 -0.003***
Intangible/TA 0.184 0.196 0.012***
Div dummy 0.832 0.380 -0.452***
NFA/TA 0.358 0.315 -0.043***
STD CF 0.026 0.053 0.027***
Secured loan 0.115 0.660 0.548***
St revolver 0.355 0.052 -0.303***
Lt revolver 0.483 0.582 0.988***
Term loan 0.110 0.308 0.198***
Maturity 39.126 51.021 11.895***
Modified Z 1.993 1.611 -0.382***
Rated 1.000 0.344 -0.656***
Panel C. Loan Spread by Top10LLTIO and Credit Ratings
Panel C reports spread and logspread for borrowing firms with Top10 LLTIO classified by 5% threshold and credit
ratings.
Investment grade Non-investment grade
Top10LLTIO Difference Top10LLTIO Difference
Variables High(≥5%) Low(<5%) (low-high) High (≥5%) Low(<5%) (low-high)
Spread 67.718 84.912 17.815*** 198.183 218.062 19.878***
Logspread 3.804 4.071 0.267*** 5.077 5.209 0.132***
Table 3. Loan Spread and Institutional Ownership
Table 3 reports results from estimating Equation (2), the relation between loan spread and local institutional
ownership after controlling for institutional ownership in general, firm characteristics, loan characteristics,
macroeconomic variables, as well as industry effects over the period of 1995-2009. An institutional owner is defined
as “local” if the headquarters of the institution is within a 100-mile radius of the company’s headquarters. Annual
Compustat data are matched to Thomson Reuters DealScan data using the identifiers of Chava and Roberts (2008).
We exclude securities with share codes different from 10 or 11, financial and utilities companies, borrowers
incorporated or headquartered outside of the U.S., loans originated outside of the U.S., loans denominated in foreign
currencies, loans with benchmark rates other than the LIBOR, and observations with missing data. Robust standard
errors are two-way clustered at the borrowing firm and year levels. ***, **, * denote statistical significance at the
1%, 5%, and 10% levels, respectively. The list of variable definitions and measurements is shown in Appendix A.
(1) (2) (3) (4) (5) (6) (7)
Variables Logspread Logspread Logspread Logspread Logspread Logspread Logspread
Bank fixed Non urban Non urban
Firm fixed
S&P500 -0.165*** -0.163*** -0.156*** -0.152*** -0.163*** -0.183*** -0.085*
(-4.397) (-4.337) (-4.461) (-4.399) (-4.523) (-4.871) (-1.930)
Relation dummy -0.028** -0.029** -0.041*** -0.004 -0.028** -0.028* -0.002
(-2.376) (-2.407) (-2.651) (-0.291) (-2.351) (-1.876) (-0.128)
Top3 bank -0.021 -0.021 -0.012 0.813*** -0.020 -0.031 -0.078***
(-1.327) (-1.330) (-0.735) -4.311 (-1.317) (-1.590) (-3.609)
IO -0.160*** -0.162*** -0.197*** -0.173*** -0.160*** -0.166*** -0.300***
(-3.669) (-3.676) (-2.648) (-3.852) (-3.660) (-2.907) (-3.765)
Top10 IO 0.276*** 0.277*** 0.484*** 0.293*** 0.163 0.291*** 0.203**
(4.479) (4.472) (4.899) (4.657) (0.564) (4.166) (2.316)
Top10 Local IO -0.054
(-1.612)
Top10 LSTIO
0.222** 0.300** 0.155**
0.172 0.048
(2.542) (2.020) (1.969)
(1.459) (0.349)
Top10 LLTIO
-0.122*** -0.224*** -0.130***
-0.156*** -0.162**
(-3.164) (-3.297) (-3.294)
(-2.856) (-2.251)
Top10 STIO
0.235
(0.729)
Top10 LTIO
0.072
(0.271)
LogTA -0.073*** -0.073*** -0.053*** -0.077*** -0.073*** -0.073*** -0.092***
(-5.700) (-5.661) (-3.315) (-6.178) (-5.555) (-4.517) (-3.962)
Leverage 0.463*** 0.463*** 0.435*** 0.431*** 0.460*** 0.447*** 0.296***
(10.850) (10.820) (6.605) (11.150) (10.700) (8.477) (4.248)
Tobin’s Q -0.102*** -0.102*** -0.136*** -0.105*** -0.103*** -0.094*** -0.086***
(-9.918) (-9.944) (-7.558) (-9.242) (-9.238) (-7.779) (-7.258)
ROA_n -0.762*** -0.763*** -0.867*** -0.728*** -0.762*** -0.790*** -0.599***
(-7.104) (-7.161) (-4.326) (-7.259) (-7.208) (-6.513) (-4.740)
R&D/TA -0.351 -0.33 -0.587 -0.363** -0.351 -0.157 -0.253
(-1.638) (-1.545) (-1.124) (-1.977) (-1.616) (-0.612) (-0.364)
Div dummy -0.075*** -0.073*** -0.057*** -0.061*** -0.074*** -0.077*** 0.004
(-6.054) (-5.951) (-3.040) (-5.081) (-5.963) (-5.028) (0.195)
NFA/TA -0.091** -0.092** -0.126** -0.062 -0.086** -0.103** -0.228*
(-2.361) (-2.423) (-2.208) (-1.516) (-2.286) (-2.439) (-2.098)
STD CF 0.802*** 0.789*** 0.488*** 0.725*** 0.796*** 0.699*** 0.340
(6.861) (6.935) (2.843) (7.339) (6.845) (6.143) (1.636)
Secured loan 0.368*** 0.367*** 0.372*** 0.350*** 0.367*** 0.368*** 0.294***
(11.970) (11.990) (9.005) (13.380) (11.690) (12.55) (12.30)
ST revolver -0.480*** -0.480*** -0.520*** -0.443*** -0.481*** -0.507*** -0.339***
(-17.260) (-17.370) (-11.130) (-16.890) (-17.280) (-16.10) (-12.53)
Lt revolver -0.319*** -0.319*** -0.340*** -0.290*** -0.319*** -0.316*** -0.217***
(-18.840) (-18.890) (-13.290) (-17.950) (-18.810) (-17.05) (-15.61)
Other loans -0.090*** -0.090*** -0.059* -0.083*** -0.090*** -0.110*** -0.035
(-3.384) (-3.434) (-1.788) (-3.232) (-3.447) (-3.545) (-1.175)
Maturity -0.003*** -0.003*** -0.003*** -0.002*** -0.003*** -0.003*** -0.002***
(-6.316) (-6.350) (-3.895) (-7.625) (-6.307) (-7.257) (-3.756)
Modified Z -0.017*** -0.016*** -0.030** -0.016*** -0.016*** -0.016** -0.003
(-2.908) (-2.877) (-2.381) (-2.656) (-2.865) (-2.147) (-0.364)
Invgrade -0.448*** -0.447***
-0.431*** -0.445*** -0.411*** -0.146***
(-12.250) (-12.290)
(-12.640) (-11.950) (-11.87) (-3.322)
Invgrade2
-0.452***
(-9.911)
Rated 0.120*** 0.119***
0.125*** 0.119*** 0.123*** 0.047
(6.483) (6.500)
(6.809) (6.304) (6.127) (1.365)
Term spread 0.070** 0.071** 0.086** 0.071*** 0.070** 0.071** 0.119***
(2.268) (2.280) (2.570) (2.700) (2.270) (2.304) (4.651)
Credit_spread 0.389*** 0.389*** 0.432*** 0.342*** 0.390*** 0.369*** 0.191**
(6.471) (6.462) (8.016) (8.134) (6.491) (5.972) (2.883)
Constant 5.383*** 5.382*** 5.464*** 5.027*** 5.376*** 5.395*** 5.335***
(67.050) (66.500) (34.950) (45.050) (65.830) (72.56) (27.36)
Loan purpose Yes Yes Yes Yes Yes Yes Yes
FF12 Industry fixed Yes Yes Yes Yes Yes Yes Yes
Bank fixed No No No Yes No No No
Firm fixed No No No No No No Yes
Two-way clustered Yes Yes Yes Yes Yes Yes Yes
Observations 17308 17308 8620 17206 17308 11600 11600
R-squared 0.657 0.658 0.720 0.695 0.657 0.640 0.822
Table 4. Instrumental Variable (IV) Regressions for Bank Loan Spread
Table 4 reports results from instrumental variable regressions for the relation between natural logarithm of loan
spread and Top10LLTIO. We use two instruments, State Top10LDIO and Industry Top10LQIO for Top10LLTIO.
State Top10LDIO is annual average of top10 local dedicated institutional ownership for all firms in the same state
but in different industries defined by 2-digit SIC code. Industry Top10LQIO is annual average of top10 local quasi
institutional ownership for all other firms within the same industry defined by 2-digit SIC code. Column (1) and (2)
report results from the second stage regressions for overall sample and rated sample only, respectively. An
institutional owner is defined as “local” if the headquarters of the institution is within a 100-mile radius of the
company’s headquarters. Annual Compustat data are matched to Thomson Reuters DealScan data according to
Chava and Roberts (2008). We exclude securities with share codes different from 10 or 11, financial and utilities
companies, borrowers incorporated or headquartered outside of the U.S., loans originated outside of the U.S., loans
denominated in foreign currencies, loans with benchmark rates other than the LIBOR, and observations with missing
data. Robust standard errors are two-way clustered at the borrowing firm and year levels. ***, **, * denote
statistical significance at the 1%, 5%, and 10% levels, respectively. The list of variable definitions and
measurements is shown in Appendix A.
(1) (2) (3) (4)
Variables First Stage
Top10LLTIO
Logspread
Full Sample
(Second stage)
First Stage Top10LLTIO Logspread
Rated only sample
(Second stage)
Top10LLTIO -0.255** -0.360**
(-2.227) (-2.186)
S&P500 0.009 -0.183*** -0.001 -0.165***
(1.08) (-5.722) (-0.05) (-4.605)
Relation dummy -0.003 -0.030*** -0.002 -0.041***
(-1.24) (-2.789) (-0.54) (-2.709)
Top3bank 0.004 -0.020 0.001 -0.009
(1.05) (-1.440) (0.24) (-0.470)
Logat 0.001 -0.070*** 0.006 -0.067***
(0.43) (-9.978) (1.45) (-4.821)
Leverage 0.001 0.501*** -0.006 0.398***
(0.12) (12.67) (-0.28) (6.286)
Tobin’s Q -0.001 -0.074*** 0.003 -0.117***
(-0.76) (-7.407) (1.09) (-5.044)
ROA_n 0.024 -0.216* 0.017 -0.290
(1.65) (-1.941) (0.86) (-1.092)
R&D/TA 0.029 -0.363* 0.104 -0.518
(0.46) (-1.693) (0.79) (-0.958)
Div dummy 0.006 -0.073*** -0.001 -0.059***
(1.33) (-5.055) (-0.21) (-2.861)
NFA/TA -0.042*** -0.122*** -0.028* -0.139**
(-3.84) (-3.178) (-1.95) (-2.295)
STD CF 0.004*** 0.015*** -0.010 0.067
(4.14) (5.402) (-0.21) (0.413)
Secured loan -0.007* 0.402*** -0.004 0.394***
(-1.68) (25.05) (-0.64) (15.03)
ST revolver -0.004 -0.516*** -0.007 -0.568***
(-0.55) (-19.13) (-0.83) (-15.48)
LT revolver 0.001 -0.330*** 0.006 -0.353***
(0.49) (-29.25) (1.59) (-21.04)
Other loans -0.009* -0.095*** -0.005 -0.069*
(-1.67) (-3.871) (-0.79) (-1.915)
Maturity -0.000* -0.003*** -0.000** -0.004***
(-1.74) (-9.001) (-2.00) (-7.364)
Modified Z -0.002 -0.043*** -0.002 -0.050***
(-1.19) (-5.494) (-0.42) (-3.760)
Invgrade -0.007 -0.392*** -0.003 -0.464***
(-0.97) (-13.06) (-0.34) (-13.48)
Term spread -0.002 0.072*** -0.000 0.090***
(-1.00) (11.03) (-0.02) (9.814)
Credit spread 0.001 0.384*** -0.004 0.426***
(0.35) (19.37) (-0.61) (14.26)
State Top10LDIO 1.429*** 1.609***
(14.14) (13.11)
Industry Top10LQIO 0.544*** 0.482***
(8.05) (5.56)
Constant 0.044* 5.465*** -0.004 5.703***
(1.69) (74.59) (-0.09) (41.90)
Loan purpose control Yes Yes Yes Yes
FF12 Industry fixed Yes Yes Yes Yes
Clustered Yes Yes Yes Yes
F-test of excluded
Instruments (p-value)
141.24
(p=0.000)
108.29
(p=0.000)
Endogenous Test 1.620 1.069
(p-value) (p=0.203) (p=0.301)
Hansen’s J-test 0.505 0.731
(p=0.477) (p=0.393)
Observations 16719 16719 8275 8275
R-squared 0.173 0.649 0.236 0.717
Table 5. Loan Spread and Previous Lending Relationship
Table 5 examines whether soft information from previous lending relationship is identical as geographical-related
soft information that the LLTIO possess. An institutional owner is defined as “local” if the headquarters of the
institution is within a 100-mile radius of the company’s headquarters. Annual Compustat data are matched to
Thomson Reuters DealScan data according to Chava and Roberts (2008). We exclude securities with share codes
different from 10 or 11, financial and utilities companies, borrowers incorporated or headquartered outside of the
U.S., loans originated outside of the U.S., loans denominated in foreign currencies, loans with benchmark rates other
than the LIBOR, and observations with missing data. Robust standard errors are two-way clustered at the borrowing
firm and year levels. ***, **, * denote statistical significance at the 1%, 5%, and 10% levels, respectively. The list
of variable definitions and measurements is shown in Appendix A.
(1) (2) (3)
Variables Logspread Logspread Logspread
With Previous Relation
subsample
Without Previous Relation
subsample
S&P500 -0.163*** -0.155*** -0.155***
(-4.321) (-3.449) (-4.425)
Relation dummy -0.023**
(-2.006)
Top3 bank -0.021 0.015 -0.050***
(-1.326) (0.787) (-3.336)
IO -0.162*** -0.206*** -0.138***
(-3.679) (-3.393) (-3.281)
Top10 IO 0.277*** 0.436*** 0.151**
(4.485) (6.289) (2.146)
Top10LSTIO 0.225** 0.362*** 0.110
(2.545) (2.784) (1.085)
Top10LLTIO -0.087** -0.178** -0.073**
(-2.535) (-2.431) (-2.021)
Relation * Top10LLTIO -0.089
(-1.116)
Loan purpose Yes Yes Yes
Loan Type Yes Yes Yes
Loan related variables control Yes Yes Yes
Financial variables control Yes Yes Yes
FF12 Industry fixed Yes Yes Yes
Two-way clustered Yes Yes Yes
Observations 17,308 7,653 9,655
R-squared 0.658 0.708 0.598
Table 6. Top10LLTIO, Loan Spread, and Soft Information
Table 6 examines how the need for geography-related soft information and the value of such information with
respect to evaluate the borrower’s credit worthiness influence the relation between logspread and Top10LLTIO. An
institutional owner is defined as “local” if the headquarters of the institution is within a 100-mile radius of the
company’s headquarters. Annual Compustat data are matched to Thomson Reuters DealScan data according to
Chava and Roberts (2008). We exclude securities with share codes different from 10 or 11, financial and utilities
companies, borrowers incorporated or headquartered outside of the U.S., loans originated outside of the U.S., loans
denominated in foreign currencies, loans with benchmark rates other than the LIBOR, and observations with missing
data. Robust standard errors are two-way clustered at the borrowing firm and year levels. ***, **, * denote
statistical significance at the 1%, 5%, and 10% levels, respectively. The list of variable definitions and
measurements is shown in Appendix A.
Panel A. Geography-related Soft Information, Hard Information, and the Effect of LLTIO on
Loan Spread
Panel A reports the results based on four proxies of the need for geography-related soft information and necessary
hard information for soft information to latch onto. These subsamples include (a) with and without positive a R&D
to assets ratio, (b) above- (high) and below-and-equal-to median (low) intangible assets, (c) with and without long-
term credit rating, and (d) with and without secured status for the loan.
(1) (2) (3) (4) (5) (6) (7) (8)
Variables Logspread Logspread Logspread Logspread Logspread Logspread Logspread Logspread
Pos R&D No R&D High Low Rated Not Secured Not Intangible intangible Rated Secured
S&P500 -0.167*** -0.144*** -0.222*** -0.116** -0.156*** -0.282*** 0.058 -0.163***
(-2.848) (-3.684) (-4.126) (-2.292) (-4.461) (-4.934) (1.347) (-5.099)
Relation dummy -0.032 -0.025** -0.037** -0.034** -0.041*** -0.005 -0.028** -0.018 (-1.449) (-2.254) (-2.296) (-2.299) (-2.651) (-0.354) (-2.166) (-0.992)
Top3 bank -0.019 -0.020 -0.027 -0.022 -0.012 -0.033 -0.051*** 0.014
(-0.783) (-1.244) (-1.326) (-1.008) (-0.735) (-1.453) (-2.955) (0.827) IO -0.187** -0.159*** -0.187*** -0.102 -0.197*** -0.088* -0.246*** -0.083
(-2.156) (-3.037) (-3.347) (-1.707) (-2.648) (-1.909) (-7.045) (-1.242)
Top10IO 0.437*** 0.196*** 0.289*** 0.244** 0.484*** 0.048 0.101 0.341*** (3.244) (2.846) (3.373) (2.508) (4.899) (0.621) (1.456) (4.455)
Top10 LSTIO 0.260** 0.229** 0.243* 0.259* 0.300** 0.176 -0.042 0.633***
(2.417) (1.988) (2.027) (2.029) (2.020) (1.568) (-0.447) (3.693) Top10 LLTIO -0.199*** -0.055 -0.164*** -0.070 -0.224*** -0.028 -0.002 -0.209***
(-2.855) (-1.068) (-2.970) (-1.083) (-3.297) (-0.578) (-0.0491) (-4.305)
Loan purpose Yes Yes Yes Yes Yes Yes Yes Yes Loan Type Yes Yes Yes Yes Yes Yes Yes Yes
Loan related variables Yes Yes Yes Yes Yes Yes Yes Yes
Control Financial variables control Yes Yes Yes Yes Yes Yes Yes Yes
FF12 Industry fixed Yes Yes Yes Yes Yes Yes Yes Yes
Two-way clustered Yes Yes Yes Yes Yes Yes Yes Yes Observations 6324 10984 7873 7598 8620 8688 9203 8105
R-squared 0.689 0.624 0.692 0.637 0.720 0.499 0.303 0.638
Panel B. Lender and Borrower Location, and Their Effect on Soft Information Production
Panel B examines how urban location and geographically proximate lead bank influences the relation between
natural logarithm of loan spread and Top10LLTIO. Two geography variables are created: Close Bank and Urban10,
which takes a value of one if the headquarters of the lead bank that issued the loan is within a 100-mile radius of the
borrowing firm’s corporate headquarters and if the borrowing firm is located in one of the 10 largest MSAs in the
U.S., respectively, and zero otherwise.
(1) (2)
Variables Logspread Logspread
Close Bank Urban 10
S&P500 -0.162*** -0.176***
(-4.320) (-3.669)
Relation dummy -0.028** -0.032**
(-2.411) (-2.370)
Top3 bank -0.020 -0.077***
(-1.300) (-3.413)
IO -0.164*** -0.155***
(-3.738) (-2.642)
Top10IO 0.276*** 0.307***
(4.454) (3.888)
Top10 LSTIO 0.237*** 0.183
(2.717) (1.345)
Top10 LLTIO -0.160*** -0.169**
(-2.955) (-2.257)
Urban10 -0.034**
(-2.366)
Urban10 * Top10LLTIO 0.108
(1.430)
Close bank -0.101*
(-1.950)
Close bank * Top10LLTIO 0.032
(0.144)
Loan purpose Yes Yes
Loan Type Yes Yes
Loan related variables control Yes Yes
Financial related variables control Yes Yes
FF12 Industry fixed Yes Yes
Two-way clustered Yes Yes
Observations 17308 6983
R-squared 0.658 0.686
Chi square Test: 0.88
Top10LLTIO + Urban10 * Top10LLTIO=0 (0.348)
(p-value)
Chi square Test: 0.49
Top10LLTIO + Close * Top10LLTIO=0 (0.485)
(p-value)
Table 7. Conflict of Interests between Shareholders and Creditors
Table 7 examines how the likelihood of conflict of interests between shareholders and creditors influence the
relation between natural logarithm of loan spread and Top10LLTIO. An institutional owner is defined as “local” if
the headquarters of the institution is within a 100-mile radius of the company’s headquarters. Annual Compustat
data are matched to Thomson Reuters DealScan data according to Chava and Roberts (2008). We exclude securities
with share codes different from 10 or 11, financial and utilities companies, borrowers incorporated or headquartered
outside of the U.S., loans originated outside of the U.S., loans denominated in foreign currencies, loans with
benchmark rates other than the LIBOR, and observations with missing data. Robust standard errors are two-way
clustered at the borrowing firm and year levels. ***, **, * denote statistical significance at the 1%, 5%, and 10%
levels, respectively. The list of variable definitions and measurements is shown in Appendix A.
(1) (2) (3) (4)
Variables Logspread Logspread Logspread Logspread
Inv. grade Non-inv. grade Crisis Non-crisis
S&P500 -0.144*** -0.090* -0.172*** -0.139***
(-3.895) (-1.778) (-3.456) (-2.691)
Relation dummy -0.025 -0.017 0.009 -0.024*
(-0.881) (-1.412) (0.326) (-1.757)
Top3 bank 0.021 -0.027 -0.015 -0.038**
(0.841) (-1.476) (-1.123) (-2.300)
IO 0.135 -0.215*** -0.121* -0.204***
(0.898) (-5.744) (-1.886) (-3.621)
Top10IO 0.396*** 0.147** 0.214** 0.288***
(2.932) (2.147) (2.350) (4.167)
Top10 LSTIO 0.645** 0.135 0.076 0.294***
(2.114) (1.401) (1.459) (2.860)
Top10 LLTIO -0.210** -0.061 0.021 -0.180***
(-2.127) (-1.411) (0.537) (-5.986)
Loan purpose Yes Yes Yes Yes
Loan Type Yes Yes Yes Yes
Loan related variables controls Yes Yes Yes Yes
Financial related variables control Yes Yes Yes Yes
FF12 Industry fixed Yes Yes Yes Yes
Two-way clustered Yes Yes Yes Yes
Observations 4066 13242 4699 12609
R-squared 0.585 0.472 0.669 0.664
Table 8. Mechanisms of the LLTIO Effect: Difference-in-Difference Analyses
Table 8 reports results from three difference-in-difference (DID) analyses on the relation between natural logarithm
of loan spread and Top10LLTIO, using two exogenous shocks: the implementation of Regulation FD (RegFD) and
Sarbanes-Oxley (SOX). By comparing the two-year periods before and after these exogenous shocks, the DID
analyses capture the change in LLTIO effect. An institutional owner is defined as “local” if the headquarters of the
institution is within a 100-mile radius of the company’s headquarters. Annual Compustat data are matched to
Thomson Reuters DealScan data according to Chava and Roberts (2008). We exclude securities with share codes
different from 10 or 11, financial and utilities companies, borrowers incorporated or headquartered outside of the
U.S., loans originated outside of the U.S., loans denominated in foreign currencies, loans with benchmark rates other
than the LIBOR, and observations with missing data. Robust standard errors are two-way clustered at the borrowing
firm and year levels. ***, **, * denote statistical significance at the 1%, 5%, and 10% levels, respectively. The list
of variable definitions and measurements is shown in Appendix A.
(1) (2) (3)
Variables DIF (RegFD) DIF (SOX) DIF (SOX)
98-99 vs 00-01 00-01 vs 02-03 98-99 vs 02-03
S&P500 -0.230*** -0.151*** -0.187***
(-6.387) (-2.848) (-4.079)
Relation dummy 0.024* 0.002 -0.025*
(1.733) (0.0834) (-1.660)
Top3 bank -0.036*** -0.046** -0.056***
(-2.763) (-2.177) (-2.763)
IO -0.155*** -0.183*** -0.125**
(-2.873) (-3.269) (-2.499)
Top10IO 0.260*** 0.440*** 0.285**
(3.236) (5.358) (2.492)
Top10 LSTIO 0.119 0.113* 0.184**
(1.212) (1.651) (2.046)
Top10 LLTIO -0.100*** 0.074* -0.089**
(-2.880) (1.911) (-2.548)
Post RegFD(98-99 vs. 00-01) 0.019
(0.579)
Post RegFD * Top10LLTIO 0.144***
(4.271)
Post SOX(00-01 vs 02-03) -0.045
(-1.535)
Post SOX * Top10LLTIO -0.291***
(-4.321)
Post SOX2(98-99 vs 02-03) -0.156***
(-5.589)
Post SOX2 * Top10LLTIO -0.116*
(-1.835)
Loan purpose Yes Yes Yes
Loan Type Yes Yes Yes
Loan related variables control Yes Yes Yes
Financial related variables control Yes Yes Yes
FF12 Industry fixed Yes Yes Yes
Two-way clustered Yes Yes Yes
Observations 5052 4910 5017
R-squared 0.705 0.701 0.707
Table 9. The LLTIOs’ Monitoring Role
Table 9 examines Top10 LLTIO’s monitoring role. Dependent variable of Column (1) is Lucky of CEO of Bebchuk,
Grinstein, and Peyer (2010). Lucky CEO takes 1 when options are granted at the lowest stock price of the month,
else zero. Dependent variables of Column (2) and (3) are E-index of Bebchuk, Cohen, and Ferrell (2009) and
dependent variable of column (4) is G-index of Gompers, Ishii, and Metrick (2003). Data is from
http://www.law.harvard.edu/faculty/bebchuk/data.shtml. Robust standard errors are clustered at the borrowing firm
level. All the regressions are include year and industry (defined as SIC2 digit level) fixed effects. ***, **, * denote
statistical significance at the 1%, 5%, and 10% levels, respectively. The list of variable definitions and
measurements is shown in Appendix A.
Variables (1) (2) (3) (4)
Lucky CEO E-index E-index G-index
(Logit) (OLS) (Ordered Probit) (OLS)
Lucky director 3.611***
(9.722)
S&P500 -0.616 0.363*** 0.362*** 1.456***
(-1.282) (4.505) (4.135) (7.172)
IO 1.500** 0.588*** 0.602*** 2.107***
(2.508) (2.700) (2.764) (4.098)
Top10IO -0.663 -0.468* -0.480* -1.371**
(-0.656) (-1.851) (-1.836) (-2.333)
Top10 LSTIO 1.263 0.121 0.174 -1.661
(0.860) (0.310) (0.402) (-1.612)
Top10 LLTIO -1.059** -0.557*** -0.602** -0.042
(-2.006) (-2.585) (-2.556) (-0.0990)
LogTA -0.035 -0.130*** -0.130*** -0.251***
(-0.252) (-4.354) (-3.832) (-3.462)
Leverage -0.265 0.105 0.084 0.058
(-0.440) (0.552) (0.446) (0.156)
Tobin’s Q 0.052 -0.070*** -0.070** -0.232***
(0.348) (-2.780) (-2.257) (-3.571)
ROA_n -1.997 -0.410 -0.436 -0.483
(-1.634) (-1.427) (-1.496) (-0.775)
R&D/TA 2.855 -1.262 -1.173 -3.203
(0.843) (-1.497) (-1.268) (-1.479)
Div dummy -0.450** 0.204*** 0.200*** 0.872***
(-2.087) (3.045) (2.971) (5.851)
NFA/TA -1.353* 0.176 0.170 0.282
(-1.714) (0.809) (0.715) (0.575)
STD CF -0.773 -0.986 -0.899 -2.873**
(-0.507) (-1.590) (-1.308) (-2.558)
Post SOX (92-2001 vs. 2002-2009) -0.653
(-1.027)
Constant -14.352*** 1.070 5.126*** 5.524***
(-10.50) (1.155) (5.002) (5.148)
SIC2 Industry fixed Yes Yes Yes Yes
Year fixed Yes Yes Yes Yes
Clustered at the firm level Yes Yes Yes Yes
Observations 2263 8901 8901 8901
(Pseudo) R-squared (0.318) 0.097 (0.035) 0.158
Table 10. Propensity Score Matching Analysis
Table 10 reports results from a two-stage propensity score match. In the first stage, we use a logit model to estimate
propensity scores for each loan observation. We match loan observations which differ in the level of Top10LLTIO,
with high Top10LLTIO (5% and above) and low Top10LLTIO (below 5%), respectively, and which are similar in
size, S&P 500 index membership or not, Tobin’s Q, R&D intensity, dividend-paying or not, ROA, asset tangibility,
cash flow volatility, institutional ownership, secured loan status, type of loan, maturity, investment grade or not,
term spread, credit spread, loan purposes, and the borrowing firms are in the same Fama-French 12 industries. We
report results from the matches using the nearest one observation and the nearest three observations, which is based
on the distance of their propensity scores, as well as requiring the error margin (caliper) to be less than 0.05,
respectively below. NN1 refers to the nearest one neighbor and NN3 refers to the nearest three neighbors in
conducting the matches. An institutional owner is defined as “local” if the headquarters of the institution is within a
100-mile radius of the company’s headquarters. Annual Compustat data are matched to Thomson Reuters DealScan
data according to Chava and Roberts (2008). We exclude securities with share codes different from 10 or 11,
financial and utilities companies, borrowers incorporated or headquartered outside of the U.S., loans originated
outside of the U.S., loans denominated in foreign currencies, loans with benchmark rates other than the LIBOR, and
observations with missing data. Robust standard errors are used. The list of variable definitions and measurements is
shown in Appendix A.
Difference Caliper 0.05
& NN1
Caliper 0.05
& NN3
Spread After matching -7.437*** -6.650***
(High vs. Low Top10LLTIO) (-3.00) (-3.16)
Logspread After matching -0.058*** -0.056***
(High vs. Low Top10LLTIO) (-4.09) (-4.69)