loan sales and accounting conservatism · secondary loan market have drawn much interest in...
TRANSCRIPT
Loan Sales and Accounting Conservatism
Abstract
We examine whether the onset of loan trading in the secondary loan market affects
borrowing firms’ accounting conservatism. We find that borrowers experience a significant
decline in accounting conservatism after initial loan sales. We also document that firms
borrowing from reputable or relationship lenders experience a less pronounced decline in
accounting conservatism after loan sales, whereas firms with higher loan trading liquidity,
firms that borrow loans with financial covenants and collateral requirements, and firms
without credit ratings experience a more pronounced decline in accounting conservatism
after loan sales. Collectively, we provide evidence that loan sales dilute lenders’ monitoring
incentive, which in turn lowers lenders’ demand for conservative reporting. Our primary
results are robust to controlling for self-selection bias of loan trading using various
techniques, and remain qualitatively similar after several sensitivity checks.
JEL: G21, G30, L14
Keywords: Loan Sales; Accounting Conservatism; Monitoring Incentives
1
Loan Sales and Accounting Conservatism
1. Introduction
We examine whether firms’ financial reporting exhibits a lower level of conservatism
following the onset of loan sales in the secondary loan market. Watts (2003a, b) indicates that
because lenders have a fixed claim on a borrower’s assets, they are more sensitive to the
borrowing firm’s bad news than to its good news. As a result, to protect themselves from
downside risk, lenders have a strong preference for accounting conservatism which involves
more timely recognition of bad news than good news. Empirical evidence suggests that lenders
reward firms that exhibit a higher level of accounting conservatism with a lower average cost of
debt (Ahmed, Billings, Morton, and Stanford 2002; Zhang 2008), and reduced collateral
requirements (Chen et al., 2013). In addition, Wittenberg-Moerman (2008) documents that the
average bid-ask spread for loans traded on the secondary loan market is lower for firms with
more conservative reporting. Research also documents how lenders influence firms’ conservative
reporting. For example, Tan (2013) finds that borrower’s financial reporting becomes more
conservative after covenant violations, consistent with the notion that lenders increase demand
for accounting conservatism when they obtain control rights following borrowers’ covenant
violations. Frankel, Kim, Ma, and Martin (2013) document that lenders impose contracting terms
requiring the submission of accounts receivable aging reports, leading to more timely recognition
of bad news and, therefore, to more conservative accounting reports after bank loan initiation.
In this study, we investigate how the onset of loan trading on the secondary loan market
affects lenders’ demand for conservative reporting. According to Thomson Reuters LPC, the U.S.
secondary loan market trading volume increased from $8 billion in 1991 to $ 409 billion in 2011,
representing a compound annual growth rate of 21.62%. The median proportion of loans being
2
sold relative to total assets for our sample firms (the universe of firms) is 26% (30%), which
suggests that loan sales is a significant economic event for the firms. Traditional financial
intermediation theories suggest that because bank loans are illiquid assets, banks are expected to
hold their loans until maturity and conduct due diligence through continued monitoring during
the tenure of the loan contract. Because of the illiquid nature of the loan assets, banks are more
concerned about the potential deterioration of loan quality, and therefore have stronger
incentives to monitor borrowers (Diamond 1984, 1991; Boyd and Prescott, 1986; among others).
However, the increased the liquidity of these once illiquid loan assets, resulting from the
development and the rapid growth of the secondary loan market over the past two decades, has
important consequences for firms’ behavior. The economic consequences of loan sales in the
secondary loan market have drawn much interest in empirical research. For example, Dahiya,
Puri, and Saunders (2003) find that the announcement of loan sales is associated with negative
stock returns. Drucker and Puri (2009) find that, after loan sales, borrowers have a larger credit
availability and maintain a more durable lending relationship with the original lenders (i.e., to
borrow from the original lead lenders). Güner (2006) finds that interest spread on loans
originated by active loan sellers is significantly lower than that on loans originated by moderate
loan sellers, which is consistent with the conjecture that borrowers may receive a price
concession in exchange for bearing potential costs of loan sales such as higher renegotiation
costs. However, prior research has not examined whether the onset of loan sales affects borrower
financial reporting.
We hypothesize that loan sales reduce lenders’ incentives to monitor borrowers, which in
turn lowers lenders’ demand for accounting conservatism. There are two reasons why lenders’
monitoring incentives might be diluted after the onset of the loan trading on the secondary loan
3
market. First, the “originate-to-distribute” model allows banks with liquidity needs to make loans
and then sell those loans in the secondary loan market, resulting in a separation of loan
origination from funding. Winton and Yerramilli (2012) and Parlour and Plantin (2008) develop
models showing that banks have little incentive to monitor borrowers after a loan has been sold,
because they can easily offload loans in the secondary loan market. This theoretical prediction
has some empirical support. For example, Gande and Saunders (2012) show that the
announcement of loan sales results in negative bond price reaction, implying a reduction in
monitoring incentives from creditors in general. Second, anecdotal evidence indicates that buyers
of the resold loans are more likely to be institutional lenders, such as hedge funds, prime funds,
finance companies, and insurance companies, which are among the most active participants in
the secondary loan market. These institutional investors have weaker incentives to monitor
borrowers because, unlike commercial banks, institutional lenders primarily focus on seeking
returns and liquidity (Nandy and Shao, 2011).1 This evidence implies that the onset of loan
trading on the secondary loan market is likely to alter the composition of the original loan
syndication to one that includes more lenders with less monitoring incentives. This, in turn, will
lead to a reduction in demand for accounting conservatism.
Using Dealscan and the secondary mark-to-market loan pricing data both provided by
Thomson-Reuters Loan Pricing Corporation (LPC), we examine whether and how the onset of
loan trading affects borrowing firms’ accounting conservatism. Specifically, we employ a sample
of firms with traded loans to investigate the changes in firms’ financial reporting following the
onset of loan trading. We use C_Score, cumulative non-operating accruals, and skewnesss of
1 In the investment policy statement, hedge funds or prime funds normally specify an objective of a high rate of
return combined with a strict requirement on the liquidity of the assets under their management. Therefore, their
primary goal is to identify investment opportunities that can meet these requirements. And, in a typical loan
syndicate, the institutional investors often serve as syndicate participants and rely on lead arrangers to monitor the
borrowers (See Nandy and Shao, 2011).
4
earnings relative to skewness of cash flow, as well as a composite measure constructed from
these three individual measures, as alternative measures of accounting conservatism. We find
that borrowing firms’ accounting conservatism declines significantly after initial loan trading. To
check the robustness of our baseline results, we use propensity score matching to select a group
of control firms with syndicated loans and similar firm and loan characteristics, but not traded
within five years after loan issuance. Using a difference-in-differences analysis, we find that
while accounting conservatism does not change much for borrowing firms of non-traded loans, it
declines significantly for firms with traded loans in the post-loan trading period.
Next, we investigate whether the impact of loan sales on accounting conservatism is
mitigated by factors that are associated with lenders’ incentives to monitor loans, including
lender reputation and prior lending relationship. Theoretical literature suggests that lender
reputation can serve as a bonding mechanism to induce lead banks to conduct effective screening
and monitoring before and after loan syndication (e.g., Booth and Smith, 1986; Chemmanur and
Fulghieri, 1994). Highly reputable lenders are more skilled and specialized in screening and
monitoring borrowers. To preserve their reputation, reputable lenders are more likely to monitor
borrowers after loan sales. Bushman and Wittenberg-Moerman (2009) present empirical
evidence that sold loans syndicated by reputable lenders experience a smaller decline in
profitability after loan sales, suggesting that reputable lenders preserve monitoring incentives
after loan sales. Relationship lending enables lenders to collect more borrower-specific
information and lower monitoring costs (Boot, 2000). Bharath, Dahiya, Saunders, and Srinivasan
(2011) document that relationship lenders have stronger incentives to monitor borrowers.
Drucker and Puri (2009) show that borrowers are more likely to borrow from their original
lenders after loan sales, resulting in a more durable lending relationship and continued
5
monitoring from relationship lenders.2 Accordingly, we predict that the decline in conservative
reporting is smaller for loans syndicated by reputable or relationship lenders. Consistent with our
predictions, we find that firms borrowing from reputable lenders or relationship lenders
experience a smaller decline in accounting conservatism.
In addition, because higher loan liquidity makes it easier for lenders to sell the loan when
borrowers perform poorly, we predict that the decline in monitoring incentives and thus the
decrease in demand for conservative reporting are more pronounced for more liquid loans. We
use loan bid-ask spread and average number of quotes per traded loan during the loan trading
year as proxies for loan trading liquidity to assess this effect. We find that firms with higher
trading liquidity experience a significant decline in accounting conservatism after initial loan
sales in the secondary market. By contrast, firms with lower trading liquidity do not exhibit a
significant change in accounting conservatism after initial loan sale.
We conduct several additional cross-sectional analyses based on other loan and firm
characteristics. In general, we find that loan sales are associated with a more pronounced decline
in accounting conservatism for firms with loans that have financial covenants and collateral
requirements, and for firms without credit ratings. These results are consistent with loan sales
having a bigger impact on firms that require more bank monitoring, i.e., risky firms with no
credit ratings or with collateral requirements for borrowing.
Our results are also robust to controlling for endogeniety of traded loans. Employing a
two-stage least squares (2SLS) instrumental variables approach with average market bid price
and presence of financial covenants as instruments for loan trading, we find that our baseline
results still hold, i.e., firms with traded loans experience a significant decline in accounting
2 Drucker and Puri (2009) do not distinguish between the types of original loan lenders, i.e., relationship vs. non-
relationship leaders. According to Boot (2000) and Bharath et al. (2011), if relationship lending is beneficial for
borrowers, it is more likely that borrowers will borrow from relationship lenders after loan sales.
6
conservatism after initial loan sales. Our results are also robust to other sensitivity checks such
as employing a constant sample requiring firms to have data for the entire post-sales periods
(three-year) and deleting loans beginning trading after 2006 to avoid the potential confounding
effects of the recent 2007~2009 financial crisis.
Our study adds to the body of research investigating the forces that motivate the use of
accounting conservatism. Although it has long been argued that debt holders have significant
influence on firms’ financial reporting, extant research mainly focuses on the relationship
between accounting conservatism and contract design (e.g., Ahmed et al., 2002; Zhang, 2008;
Nikolaev, 2010). An exception is Tan (2013), who provides evidence that lenders exercise
greater influence on firms’ financial reporting when borrowers violate loan covenants. Using the
onset of loan sales as a unique setting in which original lenders have weakened monitoring
incentives and loan buyers may lack monitoring expertise and incentive, we provide robust
evidence that lenders’ monitoring incentives have a significant influence on firms’ financial
reporting in that loan sales are associated with a significant decline in accounting conservatism. 3
We also document that firms borrowing from reputable lenders and relationship lenders
experience a smaller decline in accounting conservatism after initial loan sales.
We also contribute to the loan sales literature by identifying another consequence of
secondary loan market trading, i.e., reduced accounting conservatism. Our results suggest that
loan sales could affect borrowers’ financial reporting in that loan sales dilute lenders’
monitoring incentives, which in turn reduce their demand for conservative reporting. Thus, our
3 Gong, Martin, and Roychowdhury (2012) find that the onset of credit default swaps (CDS) is associated with lower
borrower accounting conservatism due to the lack of lender monitoring incentives. While both loan sales and CDS
are credit risk transfer mechanisms, an important difference between loan sales and CDS is that for loan sales, both
credit risk and control rights over the loans are transferred to loan buyers, whereas for CDS, only credit risk is
transferred to CDS seller, with the originating lender retaining the control rights over the loans. We expect that loan
sales are associated with more dilution of lenders’ monitoring incentive, compared to CDS. Therefore loan sales
may provide a more powerful setting to examine whether lenders’ monitoring incentive has an impact on accounting
conservatism.
7
study complements existing studies that focus on the costs and benefits of loan sales to
borrowers (Dahiya, Puri, and Saunders, 2003; Güner, 2006; Drucker and Puri, 2009; Santos and
Nigro, 2009; Kamstra, Roberts, and Shao, 2013).4 Our study also complements research that
documents loan securitization decreases lender monitoring, leading to higher default probability
and greater asset volatility (e.g., Wang and Xia, 2013).
The remainder of the paper is organized as follows: Section 2 reviews the related
literature and develops the hypotheses. Section 3 presents the sample selection and research
design. Section 4 discusses the results of the primary analyses as well as the results of robustness
checks, and Section 5 concludes the study.
2. Related literature and hypothesis development
Basu (1997) defines accounting conservatism as ‘‘the accountants’ tendency to require a
higher degree of verification to recognize good news as gains than to recognize bad news as
losses’’. Watts (2003) suggests that conservative financial reporting can arise as a result of
lenders’ demand for timely recognition of bad news. This is because lenders have a fixed claim
on firm value, and the downward bias due to timely bad news recognition would provide
assurance that the minimum amount of borrowers’ net assets is greater than the lenders’ claim on
the borrowing firms. In addition, timely recognition of bad news can also serve as an early
warning signal for poor firm performance and allow lenders to obtain creditor rights early by
triggering quicker violations of accounting ratio-based covenants. Consequently, conservative
reporting reduces lenders’ downside risk. Consistent with this argument, recent studies provide
empirical evidence that accounting conservatism is associated with a lower cost of debt (e.g.,
4 Some papers argue that certain mechanisms, including stringent covenants (Drucker and Puri, 2009) and lender
reputation concern (Gande and Saunders, 2012; Yerramilli and Winton, 2012), could mitigate the lack of monitoring
incentive effect.
8
Ahmed et al., 2002; Zhang, 2008) and lower loan bid-ask spread in the secondary loan market
(Wittenberg-Moerman, 2008). Chen, Chen, Lobo and Wang (2010) find that firms that borrow
more from state-owned banks display less accounting conservatism, as state-owned banks have
weaker demand for conservatism. Erkens, Subramanyam, and Zhang (2012) report that having
an affiliated banker on the Board of Directors reduces demand for accounting conservatism
because the affiliated banker’s access to private information from board representation facilitates
monitoring and strengthens the influence of the lender. A recent paper by Tan (2013) shows that
lenders exhibit strong preference for conservative reporting when borrowers violate covenants,
suggesting that lenders increase their monitoring intensity by demanding higher accounting
conservatism after covenant violations.
Unlike the setting in Tan (2013), we examine a unique and significant economic event in
lending, i.e., the onset of loan trading on the secondary loan market, in which lenders may lose
monitoring incentive and lower their demand for conservative reporting. The secondary loan
market facilitates banks’ credit risk management and allows non-banking firms to invest in loans.
However, loan sales separate loan origination, servicing, and funding which may lead to
information and agency related problems. Gorton and Pennacchi (1995) and Pennacchi (1988)
argue that, while the secondary loan market development generates liquidity for banks’ assets
and facilitates their portfolio and risk management, it may also dilute the monitoring incentives
of originating banks because these banks can more easily offload loans to secondary loan market
investors. Parlour and Plantin (2008) present a theoretical model in which banks have lower
incentives to monitor a borrower and to foster a relationship with the borrower, when there is an
active secondary loan market, because the presence of such a market enables banks to unbundle
asset-liability management from borrower relationship management. Existing empirical evidence
9
suggests that loan sales reduce banks’ monitoring incentives (Gande and Saunders, 2012).
Furthermore, loan sales change the composition of the original loan syndicate to one with more
institutional lenders, who have less incentive to monitor borrowers. As loan buyers typically are
non-bank financial institutions such as finance companies, insurance companies, investment
banks, and hedge funds, they lack the information advantage and monitoring skills of originating
banks to effectively monitor borrowers (Drucker and Puri, 2009). Given that lenders’ monitoring
incentives generate demand for accounting conservatism (e.g., Watts, 2003a; Zhang, 2008;
Gormley, Kim, and Martin, 2012), the lack of monitoring incentives by the originating banks due
to loan sales and the lack of monitoring skills by the new lenders (loan buyers) could directly
lead to lower accounting conservatism after loan sales. Given the above reasoning, we present
our hypothesis as follows5:
H1: Accounting conservatism decreases after the initial loan sale.
We also investigate whether the impact of loan sales on accounting conservatism is
mitigated by factors that are related to lenders’ incentives to monitor borrowers, including lender
reputation and prior lending relationship. Lender reputation is an important mechanism for
mitigating information asymmetry and incentive problems in syndicated loans. Highly reputable
lenders are more skilled and specialized in screening and monitoring borrowers. Furthermore,
reputable lenders are more committed against opportunistic behavior to preserve their reputation.
Gopalan, Nanda, and Yerramilli (2010) document that loss of reputation by the lead arranger has
a significant, negative impact on the lead arranger’s ability to obtain a loan syndication mandate,
5 We note that this hypothesis is refutable for the following reason. Loan trading in the secondary loan market may
facilitate dissemination to loan market participants of non-public information about borrowers that was previously
held by the originating lenders. As a result, information transparency between borrowers and loan market
participants improves. This increased information transparency could potentially strengthen market discipline and
increase borrower accounting conservatism.
10
and to attract participant lenders. Pichler and Wilhelm (2001) and Sufi (2007) document that lead
arranger reputation can serve as an effective mechanism in reducing the moral hazard problem.
Gande and Saunders (2012) argue that lenders may continue to monitor a borrower after loan
sale to preserve their reputation, and to manage the credit risk associated with any retained
proportion of the original loan. Bushman and Wittenberg-Moerman (2009) show that borrowers
of traded loans syndicated by reputable lenders experience smaller declines in profitability and
credit rating. Drawing on these findings, we propose the following hypothesis:
H2: The decrease in accounting conservatism after initial loan sale is
smaller for firms borrowing from more reputable lenders.
Banks are special in that they can mitigate the information frictions due to adverse
selection and moral hazard (Diamond, 1984; Fama, 1985). Past lending relationship is valuable
as banks can collect more borrower-specific information over the course of repeated lending,
information that can be used to better monitor borrowers and assess borrower credit risk (Boot,
2000). As a result, relationship lenders can monitor borrowers more effectively and at a lower
cost. Drucker and Puri (2009) document that lenders continue to lend to borrowers after loan
sales, resulting in a more durable lending relationship with borrowers. Accordingly, we
hypothesize the following:
H3: The decrease in accounting conservatism after initial loan sale is
smaller for firms borrowing from relationship lenders.
Once the loan starts trading, loan investors will solicit trading quotes from secondary loan
market makers. The frequency of trading and the bid-ask spread reflect the ease with which a
trade can be made and the liquidity of the traded loan (Bhasin and Carey, 1999). We argue that
11
the onset of loan trading increases loan liquidity, and higher loan trading liquidity further reduces
lenders’ monitoring incentives after loan sales. Therefore, we posit the following:
H4: The decrease in accounting conservatism after initial loan sale is
larger for firms with higher loan trading liquidity.
3. Research Design and Sample Selection
3.1 Research Design
Using both the Dealscan and Mark-to-market databases, we identify a sample of firms
with bank loans that are traded in the secondary loan market.6 We use the initial loan trading date
for each loan to define a dummy variable “After” which equals one after the initial loan sale date,
and zero otherwise. To investigate the effect of loan sales on accounting conservatism, we
employ the following model which controls for firm-specific variables that may affect borrower
accounting conservatism. We include industry and year fixed effects in all models, and winsorize
all variables at the 1% and 99% level.
)1(76
543210
mmiesIndustryDusYearDummieLitigationAge
rBlockholdeMBLeverageSizeAftersmConservati
The dependent variable is accounting conservatism. We measure accounting
conservatism using cumulative non-operating accruals (Non_Acc), C_Score, skewness of
earnings relative to cash flow (Skew), and a composite measure of the three individual measures
(Avg_Rank_Con). Specifically, we measure accounting conservatism by (1) the negative of
cumulative non-operating accruals scaled by total assets (Non_Acc) over the previous three years
( Givoly and Hayn , 2000; Beatty, Webber and Yu 2008); (2) the C_score measure developed by
Khan and Watts (2009), which is based on an augmented Basu (1997) model; (3) the difference
6 If a firm has multiple traded loans, we include all of the firm-traded -loans in the sample. We define “After” by the
initial trading date of each loan.
12
between the skewness of cash flows from operations and the skewness of earnings before
extraordinary items (Skew) as used by Givoly and Hayn (2000) and Beatty et al. (2008); and (4)
a composite measure based on the above three metrics (Avg_Rank_Con). For the composite
measure, we first rank the sample firms annually based on each of the three individual measures
of conservatism, and then divide each individual measure’s rank by its largest rank. We then add
the three relative ranks of each firm-year observation to obtain the composite rank of
conservatism.
The independent variable of interest, After, equals one for the years after the initial loan
sales year, and zero for the initial loan sales year. If loan sales reduce lenders’ incentives to
monitor borrowers and result in lower accounting conservatism, we expect a negative coefficient
on After ( 1 ) in the above model.
Following Ramalingegowda and Yu (2012), and Kim, Li, Pan, and Zuo (2013), we
include the following firm-specific variables as control variables: firm size (Size), leverage
(Leverage), market-to-book ratio (MB), block holder ownership (Blockholder), firm age
(FirmAge), litigation risk (Litigation). We measure firm size (Size) as the natural logarithm of the
book value of total assets, firm leverage (Leverage) as the book value of total debt divided by the
book value of total assets, market-to-book ratio (MB) as the market value of equity divided by its
book value of equity, block holder ownership (Blockholder) as the log of one plus the number of
block holders with 5% or greater holdings in a firm, and firm age (FirmAge) as the log of
number of years a firm is listed on CRSP. We proxy for a firm’s litigation risk (Litigation) by a
dummy variable that is equal to one if the firm operates in an industry with high litigation risk,
including Biotechnology (SIC codes 2833–2836 and 8731-8734), Computers (SIC codes 3570–
3577 and 7370-7374), Electronics ( SIC codes 3600–3674), and Retailing (SIC codes 5200–
13
5961) ; and zero otherwise. We also include year and industry dummies to control for fixed time
period and industry effects. We provide detailed definitions of all variables in appendix A.
To test whether the impact of loan sales on accounting conservatism varies with factors
associated with lenders’ incentives to monitor borrowers (i.e., H2 and H3), we partition the
sample into two subsamples based on lead lender reputation and relationship lender, respectively.
Following Drucker and Puri (2009), we represent reputable lenders by the variable Reputation,
which equals one if the lead arranger is a top 10 lender based on market share, and zero
otherwise. Following Bharath et al. (2007), we define lending relationship using the variable
Relationship, which equals one if a lead lender has lent to the same borrower during the five
years prior to the current loan, and zero otherwise. We expect the coefficients on After ( 1 ) to be
less negative in the subsample of loans syndicated by reputable and relationship lenders
To test H4, we partition the sample into two subgroups based on loan trading liquidity.
We use the annual average of bid-ask spread and number of quotes for each loan in the
secondary loan market as proxies for loan trading liquidity. We then estimate model (1)
separately for each of these subsamples. We expect the coefficients on After ( 1 ) to be more
negative in the subsample of loans with high trading liquidity.
3.2. Sample Selection
We use four different databases to construct our sample. We use the Mark-to-Market
Pricing database provided by Loan Syndication and Trading Association (LSTA) and Loan
Pricing Corporation (LPC) to identify loans that are traded in the secondary market. This
database provides daily secondary market loan quotations such as borrower name, number of
quotes, average bid quote, average ask quote, and mean of average bid and average ask quote,
etc.
14
We construct lead arranger reputation and prior lending relationship from Dealscan. This
database contains detailed information on borrower and lender identities, loan amounts, LIBOR
spread, issue and maturity dates, financial and general covenants, etc. We construct borrower-
specific variables such as accounting conservatism, firm size, leverage, etc. from Compustat, and
obtain block-holder ownership information from Thomson Reuters’ 13f institutional holdings.
We begin the sample selection procedure by first merging the Mark-to-Market Pricing
facility dataset with the Dealscan facility dataset by Facilityid, and then merge the intersection of
these two datasets with the Mark-to-Market Pricing historical pricing dataset by Facility
Number. In step two, we link the dataset obtained in the first step with the Compustat database
using the link file compiled by Chava and Roberts (2008).7 Lastly, we merge the dataset obtained
in step two with the 13f institutional holdings by Cusip and Year.
The sample period covers 1999 to 2009. We begin the sample period in 1999, because the
Mark-to-Market Pricing database is relatively incomplete prior to this date, and end in 2009,
because we require data coverage for three years after the initial loan trading year. The final
samples differ across conservatism measures because of differences in data availability. For
example, when we use the composite measure of accounting conservatism (Avg_rank_con), the
sample contains 1,390 unique firms and 3,646 firm-year observations with traded loans and
financial variables available, including observations for loan trading years (After = 0) and the
three years after loan trading ( After = 1).
4. Empirical Results
4.1. Sample Descriptive Statistics
Table 1 Panel A reports the yearly distribution of firms with traded loans, with event year
as the initial loan trading year. The number of firms with traded loans is relatively evenly
7 We thank Michael R. Roberts for kindly providing us the Dealscan-Compustat link file.
15
distributed in the years prior to the recent 2007–2009 financial crisis, during which the number
of firms with traded loans declines sharply.
Table 1 Panel B presents the firm characteristics and loan characteristics in the initial
loan trading year. The average (median) size, measured as log of total assets, is 7.668 (7.521);
the corresponding statistics are 2.178 (1.733) for market-to-book ratio, 2.675(0.828) for leverage
ratio, 1.876 (2) for number of block-holders, and 8.397(5) years for firm age are. On average
9.6% of firms have an investment grade credit rating, and 13.4% firms are in industries that have
high litigation risk. In terms of loan characteristics, the average (median) loan size is $507.652
($250) million, and average (median) maturity is 63.844 (62.5) months. 99.4% of the loans are
syndicated loans, 77.1% are issued by reputable lenders, 80.9% have financial covenants, 48.1%
have institutional investors as lead arrangers, 78.9% have collateral requirements, and 69.9% are
issued by relationship lenders.
Table 2 displays the level of accounting conservatism for the sample firms in the year of
trading and the following three years after initial loan trading. Column (1) shows the statistics for
the alternative conservatism measures in the year of initial loan trading (t = 0). The average
(median) of negative non-operating accruals (Non_Acc) is − 0.038 (− 0.042), the average
(median) of C-Score is 0.125 (0.116), the average (median) of skewness (Skew) is 1.065(1.086),
and the average (median) average of composite rank (Avg_Rank_Con) is 0.506 (0.496).
Columns (2) – (4) report conservatism measures one, two, and three years after initial
trading, respectively. It is a clear that accounting conservatism is declining after the initial loan
trading year, and this effect persists throughout the three-year period. For example, C_score
decreases to 0.09 in year t + 3 from 0.125 in year t, and Avg_Rank_Con declines to 0.472 in year
t + 3 from 0.506 in year t. These results are corroborated by Figure 1 which shows that firms
16
with traded loans experience a sharp decline in accounting conservatism (proxied by
Avg_Rank_Con) in the years following initial loan sales.
4.2. Test of H1: Effect of loan sales on accounting conservatism
4.2.1 Within-sample analysis
We estimate model (1) with standard errors clustered by firm, and also include year and
industry fixed effects. The coefficient of interest is the coefficient on After. We estimate model
(1) with accounting conservatism measured as C_score, Non_Acc, Skew, and Avg_Rank_Con.
The regression results are presented in Table 3 Panel A. The coefficients on After are negative
and statistically significant for all four alternative conservatism measures, as shown in columns
(1) – (4). These results indicate that conservatism declines significantly following the initial loan
trading year. This finding is consistent with H1, which posits a reduction in demand for
accounting conservatism as lenders’ monitoring incentives decline after initial loan sales. Most
of the control variables exhibit the expected signs and are consistent with those reported in prior
studies.
To examine the dynamic effect of loan sales and accounting conservatism, we also
construct three dummies, After_1, After_2, and After_3, to indicate 1, 2, 3 years after the initial
loan sale year, respectively, and include them in the conservatism model. Results presented in
Panel B of Table 3. After_1 and After_2 are both negative and significant in all models except the
Non_Acc model, and After_3 is negative and significant in all models. The results suggest that
firms exhibit a lower level of accounting conservatism following initial loan sale, and this effect
persists throughout the three-year period after initial loan sale.
17
4.2.2. Difference-in-differences analysis
To alleviate the concern that the above results might be driven by unobservable time
series changes, such as macroeconomic or industry shocks, we explore the impact of loan sales
on accounting conservatism using a matched sample, difference-in-differences design. To
identify potential matched control firms, we restrict our focus to all firms with available
Compustat financial and Dealscan loan data, but the loans are not traded in the secondary loan
market within 5 years after loan issuance.8 We employ the following reduced form of Drucker
and Puri’s (2009) probit model to examine the probability of loan sales (Trade). Then we employ
the propensity score matching methodology to select the control firms.
(2)Dummies
ReRecov
PrLeverageSize
1110987
6543210
mmiesIndustryDuYearesLoanPurposLoanTypes
lationshipputationenantFnloanInstitutioSyndicate
LmatLloansizeeInvestgradyofitabilitTrade
The dependent variable, Trade is a binary variable that equals one if the loan is sold in
the secondary loan market, and zero otherwise. The independent variables include borrower,
loan, and lender characteristics, which are important determinants of loan sale. Borrower
characteristics include logarithm of book assets (Size), debt-to-asset ratio (Leverage),
profitability (Profitability), and an indicator for borrowers that are investment-grade rated
(Investgrade). We expect a higher probability of loan sales for larger borrowers and for more
risky (higher leverage, junk-rated) borrowers. Loan characteristics include the logarithm of loan
size (Lloansize), logarithm of loan maturity in months (Lmat), an indicator for syndicated loans
(Syndicate), an indicator for institutional loans (Institutionloan), and an indicator for loans with
financial covenants (Fcovenant). We expect that larger loans, loans with longer maturity,
8 Drucker and Puri(2009) indicates more than 60% of loans are initially sold within one month of loan origination
and almost 90% are sold within one year. We choose five-year period to be more conservative.
18
syndicated loans, and institutional loans are more likely to be traded. We also include lead
arranger reputation (Reputation) and prior lending relationship (Relationship) in the model. We
expect a higher probability of loan sales for loans issued by more reputable or relationship
lenders. We control for loan type and loan purpose, and for industry and year fixed effects.
Figure 2 shows the changes in accounting conservatism for the treatment sample with
traded loans and the matched control sample without traded loans. For the treatment sample,
there is a sharp decline in accounting conservatism following initial loan sales, consistent with
pattern illustrated in Figure 1. For the matched control sample, accounting conservatism actually
increases for a few years after “the hypothetical trading year” before declining. We use the
estimated coefficients of model (2) to estimate the probability of loan sales for all firms that
borrowed bank loans during 1999~2008.9 We match each firm with traded loans (without
replacement) with a firm with non-traded loans issued in the same year based the closest
estimated probability of loan sales. Then we estimate the following conservatism model to
evaluate the impact of loan sales on conservatism for firms with traded bank loans.
)3(
LeverageSize*
987
6543210
mmiesIndustryDusYearDummieLitigationFirmAgerBlockholde
MBTradeAfterAfterTradesmConservati
The model includes Trade, an indicator for the type of borrower, to differentiate firms
with traded loans from firms with non-traded loans, and After, an indicator for the time period
before versus after the initial loan trading period. For the non-traded firms, we assign the initial
trading year of the matched traded firms as the hypothetical initial trading year. Our variable of
interest is the interaction between these two indicators After*Trade. This research design allows
us to compare the change in accounting conservatism before and after the initial loan trading
9 To alleviate the data truncation problem resulting from loan trading in 2009, we estimate the first stage model
using observations with loan issuance year between 1999 and 2008. Our analysis is robust to using loan issuance
years between 1999 and 2007 and between 1999 and 2006.
19
period for borrowers with traded loans to the corresponding change for borrowers with non-
traded loans. In addition, using firms with non-traded loans as control firms helps us to isolate
the effect of loan sales by excluding possible confounding factors that may change around the
time of loan sales.
The results of the probit model predicting whether the loan is traded or not are presented
in Table 4 Panel A. Consistent with our prediction, loans issued to borrowers that are larger,
have higher leverage, and have speculative grade credit rating (i.e., below BBB) are more likely
to be traded. Loans that are larger, have longer maturity, are originated by more reputable lenders
and relationship lenders, have a larger number of financial covenants and institutional loans are
more likely to be sold. The model correctly predicts 91.3% of the cases, indicating that it is
effective at predicting loan trading.
The regression results for accounting conservatism are presented in Table 4 Panel B. The
coefficient estimates on the interaction term After*Trade are negative and statistically significant
for all four conservatism measures. The results imply that, when compared to the matched
control firms with non-traded bank loans, accounting conservatism declines significantly for
firms with traded loans after the initial loan sales. These findings are consistent with H1. They
indicate that firms reduce conservative reporting after initial loan sales, as lenders’ monitoring
incentive declines, which lowers their demand for conservatism. This decline is economically
significant. Specifically, for the sample of firms with traded loans, firms experience a 5% decline
in accounting conservatism in the years following the year of initial loan trading.10
This decline
is about 7% for the sample of firms with traded loans, compared to the matched sample with
non-traded loans.11
10
This number is calculated as 0.024/0.506 (numbers taken from Panel A of Table 3) 11
This number is calculated as (0.005 – 0.038)/0.506 (numbers taken from Panel B of Table 4)
20
4.3. Tests of H2 and H3: Lender monitoring and the effect of loan sales on accounting
conservatism
Existing studies suggest that reputable lenders and relationship lenders continue to
monitor borrowers after loan sales to preserve reputation and lending relationship (Bushman and
Wittenberg Moerman, 2009; Drucker and Puri, 2009; Gande and Saunders, 2012). Accordingly,
we expect the decline in accounting conservatism to be less pronounced for firms that borrow
from reputable (H2) or relationship lenders (H3).
To test H2, we conduct separate regressions using model (1) for loans syndicated by
reputable and non-reputable lenders. We define reputable lenders as lenders in the top 10 based
on market share each year. If the lead arrangers are not among the top l0 lenders, we classify
them as non-reputable lenders. We report the estimation results for loans syndicated by non-
reputable lenders in column (1) and for loans syndicated by reputable lenders in column (2) in
Table 5. The coefficients on After in columns (1) and (2) are both negative and significant ,
indicating that firms that borrow from both non-reputable and reputable lenders experience a
significant decline in accounting conservatism after initial loan sales. Consistent with H2, the
coefficient on After is more negative for firms that borrow from non-reputable lenders than for
firms that borrow from reputable lenders (coef = –0.045 vs. –0.016; t-value = –3.328 vs. –
2.543), and the difference in coefficients is significant at the 10% level (t-value = 1.75). These
coefficient estimates indicate that firms borrowing from reputable (non-reputable) lenders
experience a 3.16% (8.89%) decline in accounting conservatism after initial loan sales.
To test H3, we divide the sample into relationship loans and non-relationship loans and
estimate model (1), separately for each of these two subsamples. Relationship loans are loans
made by a lender who also made loans to the same borrower in the five-year period prior to the
21
current loan, and non-relationship loans are loans made by a lender who did not make loans to
the same borrower in the five-year period prior to the current loan. The results are shown in
Table 5, columns (3) and (4), respectively. The coefficient on After, in columns (3) and (4),
shows that firms borrowing from both non-relationship and from relationship lenders experience
a significant decline in accounting conservatism following initial loan sales. Consistent with H3,
the coefficient on After for firms that borrow from non-relationship lenders (coef = -0.045; t-
value = -4.099) is more negative than the coefficient on After for firms that borrow from
relationship lenders (coef = - 0.016; t-value = -2.321), with the difference significant at the 5%
level (t-value = 2.04).
4.4. Test of H4: Loan trading liquidity and the effect of loan sales on accounting
conservatism
We also test the effect of loan trading liquidity on accounting conservatism after initial
loan sales. We predict that when loan trading liquidity is higher, it is easier for lenders to offload
poorly-performing loans, which will reduce lenders’ exposure to downside risk and lead to lower
demand for conservative reporting. We use average loan bid-ask spread and average number of
quotes for a traded loan each year as proxies for loan trading liquidity. We divide the sample
into two subsamples based on the sample median of loan bid-ask spread and sample median of
number of quotes in the secondary loan market, and estimate model (1) separately for each
subsample.
We report the results in Table 6. The coefficient on After is negative and significant for
firms with high trading liquidity (coef = -0.049; t - value= -4.833), proxied by low bid-ask
spread in the secondary loan market, whereas it is positive and insignificant for firms with high
bid-ask spread (low trading liquidity) (coef = 0.008; t - value = 0.888). The difference in the
22
coefficients is significant at the 1% level (t –value = 3.54). When using number of quotes as the
alternative measure of loan trading liquidity, we find that firms with high trading liquidity (i.e.,
high number of quotes) experience a significant decline in accounting conservatism (coef = -
0.026; t – value = -2.850). In contrast, for firms with lower number of quotes, the coefficient on
After is not statistically different from zero (coef = 0.002; t –value = 0.224). The difference in
the coefficients is significant at the 10% level (t-value = 1.879). These results provide support
for H4.
4.5. Additional analyses
We perform additional cross-sectional analyses based on other loan- and firm-specific
characteristics. The loan characteristics we examine include financial covenants and collateral,
and the firm characteristics include firm size, age, and credit rating status. The results are
presented in Table 7 Panel A and Panel B, respectively.
Loans with financial covenants and collateral requirements have higher credit risk, and
demand more bank monitoring. As a result, loan sales are expected to have a more negative
impact on conservatism for such loans. The results presented in Panel A of Table 7 are consistent
with our prediction that there is a more pronounced decline in accounting conservatism after
initial loan sales for firms with loans that have financial covenants and collateral requirements.
Smaller firms, younger firms, and firms with no credit ratings have poorer information
environment and higher credit risk. Lenders need to monitor such borrowers more vigilantly to
prevent downside risk. Consequently loan sales are expected to have a more negative impact on
conservatism for such firms. The results presented in Columns (1) and (2) of Table 7 Panel B
show that conservatism declines for both younger and older firms. However, for younger firms,
the coefficient on After is more negative than in the coefficient for older firms (coef = –0.027 vs.
23
–0.017, and t-value = –3.349 vs. –2.071), but the difference is not statistically significant.
Columns (3) and (4) show that conservatism declines for both smaller and larger firms.
Although not reliably different, the decline is greater for smaller firms than for larger firms (coef
on After = –0.028 vs. –0.015, and t-value = –3.634 vs. –2.009). Columns (5) and (6) show that
conservatism declines for firms without credit rating and for rated firms. However, for non-rated
firms, the coefficient on After is more negative in magnitude and more statistically significant
(coef = -0.040; t- value = -3.603) than the corresponding coefficient for rated firms (coef =–
0.012; t-value =–1.754). The difference in coefficients is statistically significant at the 10% level
(t –value = 1.87).
4. 6. Robustness Checks
4.6.1 Self-selection bias of loan trading: instrumental variable approach
To further address the unobservable firm characteristics that may affect both the
probability of loan trading and change in accounting conservatism, we employ an instrumental
variables approach to account for potential selection bias that may arise from the choice of loan
trading.
As in the previous difference-in-differences analysis, we treat Trade and Trade*After as
endogenous variables. We use two market trading indices, market average bid price
(Avg_market_bid) and market average bid-ask spread (market_spread), as instruments for Trade.
Although the probability of a loan being traded in the market increases when the market trading
index is more favorable (i.e., high bid price and low bid-ask spread), these indices are unlikely to
affect individual firms’ conservative reporting. The third instrument is Avg_market_bid*After.
Wooldridge (2010) suggests that when Avg_market_bid is an instrument, the interaction term
24
between Avg_market_bid and an exogenous variable (After) becomes a natural instrument for the
model.
We present the two-stage regression results in Table 8, with the first stage results in
Panel A, and the second stage results in Panel B. The first stage results indicate that market bid-
ask spread (Market_spread) is negatively associated with probability of loan trading (significant
at the 10% level). The second stage regression results show that the coefficient on After*Trade
remains negative and significant (at 1%), indicating that even after controlling for potential
endogeneity, the effect of loan sales on accounting conservatism remains negative and
significant. The weak-identification test (Cragg-Donald Wald F statistic =27.556) rejects the
null that the instruments are weakly correlated with the endogenous variables. The insignificant
Hansen J statistics (J statistics = 0.838; p = 0.360) indicates that the instrumental variables are
valid instruments, i.e., they are not correlated with the error terms in the second stage equation.
4.6.2 Other robustness tests
We conduct two other robustness checks to ensure the validity of our results. First, we
use a constant sample by requiring that the sample firms have available data in each year of the
three-year period after initial loan trading. Table 9 presents the results for this sample. The
coefficients on After are negative and significant for all of the conservatism measures, except for
Skew.
Second, because our sample period overlaps the recent 2007–2009 financial crisis, we re-
estimate the models after deleting observations with loans issued after 2006 to alleviate the
concern that our results may be driven by the financial crisis effect. We present these results in
Table 9. We continue to find that firms experience a significant drop in accounting conservatism
25
after initial loan sale. The coefficients on After are negative and significant for all of the
conservatism measures, except for Non_Acc.
5. Conclusion
We examine the effect of loan sales in the secondary loan market on accounting
conservatism. While the “originate-to-distribute” model allows bank lenders to manage their risk
through loan sales in the secondary loan market, it may dilute the monitoring incentives of
originating banks, as these lenders can easily offload loans on the secondary loan market.
Furthermore, most loan buyers in the secondary loan market are non-bank financial institutions
that do not have information or the skills to monitor the borrower. In this study, we predict that
loan sales reduce lenders’ incentives to monitor borrowers, which in turn lowers lenders’ demand
for accounting conservatism.
The empirical evidence supports our predictions. Specifically, we find that firms with
traded loans experience a significant decline in accounting conservatism after initial loan sales.
These results are robust to a variety of robustness checks, including using propensity score
matching and instrumental variables estimation. We also document that the decline in
accounting conservatism is smaller for firms that borrow from more reputable lenders and also
from relationship lenders. Furthermore, we document that the decline in accounting
conservatism is more pronounced for firms with higher loan trading liquidity, firms with loans
that have financial covenants and collaterals, and firms without credit rating. Collectively, we
provide compelling evidence that loan sales reduce lenders’ monitoring incentives, and hence
lower their demand for accounting conservatism.
Taking the onset of loan trading as a unique event that may affect lenders’ monitoring
incentives, our results show that lenders play an important role in shaping firms’ financial
reporting. We also contribute to the loan sales literature by documenting another consequence
26
of secondary loan market trading, i.e., reduced accounting conservatism. In this regard, our
study complements existing research that focuses on the costs and benefits of loan sales.
27
Appendix A: Variable Definitions
Variables Definition
Dependent variables
Non_Acc A measure of accounting conservatism for each firm-year, calculated as the
accumulated non-operating accruals over three years prior to current year divided by
total assets. Non-operating accruals = -[(net income (Compustat:ni) + depreciation
(Compustat:dp) – cash flows from operation (Compustat: oancf)– changes in
AR(Compustat:rect) –change in inventories(Compustat:invt)+ change in account
payable (Compustat :ap)+ change in tax payable(Compustat:txp)-change in prepaid
expense (Compustat:xpp)) /average assets(Compustat:at)]. When cash flows from
operation is not available, cash flow from operation = funds from operation
(Compustat:fopt)+change in cash (Compustat:che) – change in current assets
(Compustat:act) + change in current liabilities in debt (Compustat:dlc)+change in
current liabilities (Compustat:lct).
Skew A measure of accounting conservatism for each firm-year, calculated as the the
negative of the difference in the skewness of earnings and the skewness of cash flows
over three years prior to current year.
C_Score A measure of accounting conservatism developed by Khan and Watts (2009) (see
appendix B for the estimation procedure).
Avg_Rank_Con The composite measure of accounting conservatism by adding the relative rank of the
three individual conservatism measures (Non_Acc, Skew, C_Score).
Main Independent Variables
Trade An indicator variable equals to 1 if a firm has loans traded on the secondary loan
market and zero otherwise.
After An indicator variable, equals to one if the firm-year is in [t + 1, t + 3] and zero if the
firm-year is in t which is the initial loan sale year.
After*Trade The interaction term between Trade and After defined above.
Reputation An indicator variable, equals to one if at least one of the lead lender in a loan
syndicate is on the top 10 lenders list based on market share, and zero otherwise.
Relationship An indicator variable, equals to one if at least one of the lead lender in a loan
syndicate has lent to the borrower in previous deals within five years, and zero
otherwise.
Average bid-ask spread The average loan bid-ask spread in year [t + 1, t + 3] where t is the initial loan trading
year
Average Number of quotes The average number of loan trading quotes in year [t + 1, t + 3] where t is the initial
loan trading year
Other Independent Variables: Loan Characteristics
Lloansize Natural logarithm of the loan facility amount.
Lmat Natural logarithm of the loan maturity in months.
Fcovenants An indicator variable equals to 1 if there is restrictive financial covenant specified in a
loan contract and zero otherwise.
Institution loan
An indicator variable equals to 1 if a loan facility is a term loan B, C, D, or E, and zero
otherwise.
Secured An indicator variable equals to 1 if a loan facility has collateral requirements, and zero
otherwise.
28
Borrower Characteristics
Size Natural logarithm of the borrower’s total assets.
Leverage The borrower’s book value of total debt over book value of total assets.
MB The borrower’s market-to-book ratio, calculated as (TA+MKVALF-CEQ)/TA, where
TA is the book value of total assets, MKVALF is the market value of equity at the
fiscal year end, and CEQ is the book value of equity.
Investgrade An indicator variable equals to 1 if a borrower has an S&P long term debt rating of
BBB or above, and zero otherwise.
Profitability Calculated as EBITDA over sales.
Litigation An indicator variable, equals to one if the firm is from one of the following industries
Biotechnology (SIC codes 2833–2836 and 8731-8734), Computers (SIC codes 3570–
3577 and 7370-7374), Electronics (SIC codes 3600–3674), and Retailing (SIC codes
5200–5961); and zero otherwise.
Blockholder Natural logarithm of (1 + the number of block holders). Block holders are defined as
the institutional shareholders with the percentage of holdings greater than 5%.
FirmAge The natural logarithm of firm age.
Other variables
Avg_market_bid The average monthly market bid price in the secondary loan market
Market_spread The average monthly bid-ask spread in the secondary loan market
29
Appendix B: Description of C_Score Measure (Khan and Watts, 2009)
Khan and Watts (2009) develop a firm-year specific measure based on Basu’s (1997) notion of asymmetric
timeliness as well as empirical and theoretical evidence that firm size, market to book ratio, and leverage generate
cross-sectional variations in accounting conservatism. The basic Basu (1997) model is the following:
itittittitttP/it DRRRDREt
21011
(A1)
where itE /Pit-1 is earnings per share before extraordinary items deflated by price at the beginning of the fiscal year
(epsfx/prcc_f t-1), Rit is annual returns, itDR is an indicator variable equal to one when returns are negative, and t2
measures the incremental timeliness of earnings loss recognition.
Khan and Watts (2009) extend the Basu (1997) model by incorporating firm size, market-to-book ratio, and leverage
to estimate the following equation:
)/( 432101/ 1 ittittitttititttPit LevBMSizeRDRE
t
itittittittittittitttitit LevBMSizeLevBMSizeDRR )/()/( 3214321 (A2)
where Size is the natural log of market value of equity (log(prcc_f*csho)), M/B is the market to book ratio
(prcc_f*csho/ceq), Lev is the leverage of the firm ((dltt+dlc)/(prcc_f*csho)), and other variables are as defined in the
equation A1.
We estimate the regression model in A2 by Ordinary Least Squares regression in each year. All variables in
estimating the coefficients are winsorized at the top and bottom 1%. We calculate the asymmetric timeliness
(C_Score) for each firm-year by using coefficients estimated for that year as follows:
ittittittt LevBMSizeScoreCt
4321 / (A3)
30
References
Ahmed, A., B. Billings, R. Morton, and M. Stanford, 2002. The Role of Accounting
Conservatism in Mitigating Bondholder-Shareholder Conflicts over Dividend Policy and in Reducing Debt Costs.
The Accounting Review 77: 867-890.
Basu, S., 1997. The conservatism principle and the asymmetric timeliness of earnings. Journal of Accounting and
Economics 24: 3–37.
Beatty, A., Weber, J., Yu, J., 2008. Conservatism and debt. Journal of Accounting and Economics 45: 154–74.
Bharath, S., S. Dahiya, A. Saunders, and A. Srinivasan, 2007. So What Do We Get? The Bank's View of Lending-
relationships. Journal of Financial Economics 85: 368-419.
Bharath, S., S. Dahiya, A. Saunders, and A. Srinivasan. 2011. Lending relationships and loan contract terms. Review
of Financial Studies 24 (4), 1141–1203.
Bhasin, V., and M. Carey, 1999. The determinants of corporate loan liquidity. Working Paper, Freddie Mac.
Boot, A. W., 2000. Relationship banking: What do we know? Journal of Financial Intermediation 9:7–25.
Booth, J.R., Smith, R.L., 1986. Capital raising, underwriting, and the certification hypothesis. Journal of Financial
Economics 15, 261–281.
Boyd, J. H., and E.C. Prescott. 1986. Financial intermediary-coalitions. Journal of Economic Theory 38: 211-232.
Chava, S., and M. R. Roberts, 2008. How Does Financing Impact Investment? The Role of Debt Covenants. Journal
of Finance 63: 2085-2121.
Chemmanur, T., Fulghieri, P., 1994. Reputation, renegotiation, and the choice between bank loans and public traded
debt. Review of Financial Studies 7, 475–506.
Chen, H., J. Z. Chen, G. J. Lobo, and Y. Wang, 2010. Association Between Borrower and Lender State Ownership
and Accounting Conservatism. Journal of Accounting Research 48 (5): 973~1014.
Chen, J. Z., G. J. Lobo, Y. Wang, and L. Yu, 2013. Loan collateral and financial reporting conservatism: Chinese
evidence. Journal of Banking and Finance 37: 4989-5006.
Dahiya, S., M.Puri, and A. Saunders, 2003. Bank borrowers and loan sales: New evidence on the uniqueness of bank
loans. Journal of Business 76: 563-582.
Diamond, D. 1984. Financial Intermediation and Delegated Monitoring. Review of Economic Studies 51: 393-414.
Diamond, D. W., 1991. Monitoring and reputation, the choice between bank loans and directly placed debt. Journal
of Political Economy 99: 688–721.
Drucker, S., M. Puri, 2009. On loan sales, loan contracting, and lending relationships. Review of Financial Studies
22 (7): 2835-2872.
Erkens, David, K. R. Subramanyam, Jieying Zhang, 2012. Affiliated banker on board and conservative accounting.
Working Paper, University of Southern California.
Fama, E. F. 1985. What’s Different about Banks? Journal of Monetary Economics 10:10–19.
Frankel, R., Kim, B., Ma, T., X., Martin, 2013. Bank Monitoring and Accounting Recognition:
The case of aging-report requirements. Working Paper, Washington University in St. Louis.
31
Gande, A. and A. Saunders, 2012. Are Banks Still Special When There is a Secondary Market for Loans? Journal of
Finance 67 (5): 1649-1684.
Givoly, D., Hayn, C., 2000. The changing time-series properties of earnings, cash flows and accruals: Has financial
reporting become more conservative? Journal of Accounting and Economics 29: 287–320.
Givoly, D., Hayn, C., Natarajan, A., 2007. Measuring reporting conservatism. The Accounting Review 82: 65–106.
Gong. G., X. Martin, and S. Roychowdhury, 2012. Do financial market developments influence accounting
practices? Credit default swaps and borrowers’ reporting conservatism. Working Paper, Pennsylvania State
University.
Gormley, T., Kim, B., and Martin, X., 2012. Do Firms Adjust Their Timely Loss Recognition in
Response to Changes in the Banking Industry? Journal of Accounting Research 50 (1): 159-
196.
Gorton, G. B. and G. G. Pennacchi, 1995. Banks and loan sales marketing nonmarketable assets. Journal of
Monetary Economics 35: 389-411.
Güner, A. B., 2006. Loan Sales and the Cost of Corporate Borrowing. Review of Financial Studies 19 (2): 687-716.
Kamstra, M. J., G. S. Roberts, and P. Shao, 2013. Does the secondary loan market reduce borrowing costs? Review
of Finance, forthcoming.
Khan, M., Watts, R., 2009. Estimation and empirical properties of a firm-year measure of accounting conservatism.
Journal of Accounting and Economics 48: 132-150.
Kim, Y., S. Li, C. Pan, and L. Zuo, 2013. The role of accounting conservatism in the equity market: evidence from
seasoned equity offerings. The Accounting Review 88 (4): 1327-1356.
Nandy, D., and P. Shao. 2011. Institutional investment in syndicated loans. Working Paper, Brandeis University and
University of Lethbridge.
Nikolaev, V. (2010) Debt covenants and accounting conservatism. Journal of Accounting Research 48, 137-175.
Parlour, C. A., and G. Plantin, 2008. Loan sales and relationship banking. Journal of Finance 63 (3): 1291-1314.
Petersen, M., 2009. Estimating standard errors in finance panel data sets: Comparing approaches. Review of
Financial Studies 22: 435-480.
Pichler, P., and W. Wilhelm, 2001. A theory of the syndicate: Form follows functions. Journal of Finance 56: 2237-
2264.
Pennacchi, G. G., 1988. Loan sales and the cost of bank capital. Journal of Finance 43(2): 375-396.
Ramalingegowda, S., and Y. Yu, 2012. Institutional ownership and conservatism. Journal of Accounting and
Economics 53: 98-114.
Santos, J. A.C., and P. Nigro, 2009. Is the secondary loan market valuable to borrowers? The Quarterly Review of
Economics and Finance 49: 1410–1428.
Saunders, A., P. Shao, and Y. Li. 2013. The monitoring incentives of transactional and relationship lenders:
Evidence from the syndicated loan market. SSRN Working Paper.
32
Sufi, A., 2007. Information asymmetry and financing arrangements: evidence from syndicated loans. Journal of
Finance 62: 629-668.
Wittenberg-Moerman, R., 2008. The Role of information asymmetry and financial reporting quality in debt trading:
Evidence from the secondary loan market. Journal of Accounting and Economics 46 (2-3): 240-260.
Tan, L., 2013. Creditor control rights, state of nature verification, and financial reporting conservatism. Journal of
Accounting and Economics 55 (1): 1-22.
Wang, Y., and H. Xia, 2013. Do lenders still monitor when they can securitize loans? Review of Financial Studies,
forthcoming.
Watts, R., 2003a. Conservatism in accounting Part I: Explanations and implications. Accounting Horizons 17 (3):
207-221.
Watts, R., 2003b. Conservatism in accounting Part II: Evidence and research opportunities. Accounting Horizons 17
(4): 287-301.
Yerramilli, V., A. Winton, 2012. Lender moral hazard and reputation in originate-to-distribute markets. Working
Paper, University of Minnesota.
Zhang, J. , 2008. Efficiency gains from accounting conservatism: Benefits to lenders and borrowers. Journal of
Accounting & Economics 45: 27-54.
33
Figure1. Changes in accounting conservatism for traded loans only This figure shows changes in accounting conservatism around the onset of loan trading (year 0) for the
traded loans only. Avg_Rank_Con is the composite measure of accounting conservatism.
Figure 2. Changes in accounting conservatism for traded and matched non-traded firms
This figure shows the changes in conservatism around the onset of loan trading (year 0) for the
sample with traded loans and matched sample of non-traded loans (Year 0 is the “hypothetical
loan trading year”). Avg_rank_con is the composite measure of accounting conservatism.
.47
.48
.49
.5.5
1
Avg
_Ran
k_C
on
-4 -2 0 2 4Year
Loan Sales and Accounting Conservatism
.46
.48
.5.5
2.5
4
Avg
_Ran
k_C
on
-4 -2 0 2 4Year
Traded Non-Traded
Loan Sales and Accounting Conservatism
34
Table 1: Sample distribution and sample descriptive statistics
This table reports the yearly distribution of the onset of loan trading between 1999-
2009 and sample descriptive statistics in the initial loan trading year (t = 0).
Panel A: Sample distribution
Year Number of Firms Number of Loans
1999 149 278
2000 117 227
2001 103 180
2002 154 253
2003 162 265
2004 169 306
2005 176 295
2006 155 262
2007 134 210
2008 43 59
2009 28 37
Total 1,390 2,372
Panel B: Sample descriptive statistics in the initial loan trading year ( t = 0)
Variable N mean sd p25 p50
Size 1390 7.680 1.344 6.691 7.521
MB 1390 2.178 7.296 0.834 1.733
Leverage 1390 2.675 5.256 0.394 0.828
Litigation 1390 0.134 0.341 0 0
Log (1+ #of blockholders) 1390 1.876 1.760 0 2
FirmAge 1390 8.397 10.035 1 5
Investgrade 1390 0.096 0.295 0 0
Loan Amount (in $millions) 1390 507.652 1187.434 120 250
Maturity (in months) 1368 63.844 20.541 60 62.5
Syndicate 1390 0.994 0.076 1 1
Reputation 1390 0.771 0.421 1 1
Fcovenant 1390 0.809 0.394 1 1
Institutionloan 1390 0.481 0.500 0 0
Secured 1390 0.789 0.408 1 1
Relationship 1390 0.699 0.459 0 1
35
Table 2: Changes in conservatism from the onset of loan trading (t = 0) to subsequent years (t = 1, 2, 3).
This table presents accounting conservatism from year t = 0 to t = 3 in columns (1) - (4), where t is the first year of loan trading. Accounting
conservatism is measured by accumulated non-operating accruals (Non_Acc), C_Score, Skewness of earnings (Skew), and the average rank of the
three individual conservatism measures (Avg_Rank_Con). The variable definitions can be found in Appendix A.
(1) (2) (3) (4)
t = 0 t = 1 t = 2 t = 3
Variable mean sd p50 N mean sd p50 N mean sd p50 N mean sd p50
Non_Acc 1284 -0.038 0.102 -0.042 1265 -0.039 0.094 -0.043 1252 -0.046 -0.096 -0.051 1172 -0.053 0.094 -0.057
C_Score 952 0.125 0.136 0.116 952 0.109 0.139 0.104 933 0.101 0.135 0.090 885 0.090 0.139 0.076
Skew 1290 1.065 1.756 1.086 1266 0.991 1.805 1.035 1255 0.874 1.903 0.868 1178 0.894 1.929 0.847
Avg_Rank_Con 921 0.506 0.193 0.496 923 0.498 0.195 0.485 923 0.484 0.191 0.473 879 0.472 0.196 0.457
36
Table 3: Effect of loan sales on accounting conservatism
This table presents the estimation results of the effects of loan sales on firms’ accounting conservatism. The dependent variables are measures of
accounting conservatism: accumulated non-operating accruals (Non_Acc), C_Score, Skewness of earnings (Skew), and the average rank of the
three individual conservatism measures (Avg_Rank_Con). Panel A presents the results showing the cumulative effects of loan trading in year t + 1
to t +3, where t is the first year of loan trading. After is an indicator variable that equals 1 if the firm-year is in [t + 1, t + 3] and 0 if the firm year
is t. Panel B shows the individual year effect of loan trading on accounting conservatism with after_1, after_2, and after_3 representing one-year,
two-years, and three-years after loan trading, respectively. The definition of the regression variables can be found in Appendix A. Superscripts
***, **, * correspond to statistical significance at the 1%, 5%, and 10% levels, respectively, with standard errors clustered at firm level.
Panel A: Loan sales and accounting conservatism
(1) (2) (3) (4)
VARIABLES Non_Acc C_Score Skew Avg_Rank_Con
After -0.007*** [-2.996] -0.021*** [-5.597] -0.123** [-2.476] -0.024*** [-4.286]
Size -0.001 [-0.250] -0.049*** [-24.043] 0.009 [0.197] -0.045*** [-8.412]
Leverage 0.006*** [10.463] 0.015*** [4.942] 0.055*** [6.345] 0.021*** [8.205]
MB -0.000 [-0.644] -0.002** [-2.380] -0.009** [-2.015] -0.003**
Blockholder -0.004 [-1.136] -0.000 [-0.075] 0.074 [0.911] 0.023**
FirmAge -0.000 [-0.256] 0.000 [0.608] -0.013* [-1.792] -0.001
Litigation 0.002 [0.242] -0.020*** [-3.205] -0.033 [-0.194] -0.031 [-1.447]
Constant -0.041** [-1.964] 0.559*** [11.265] 1.437** [2.463] 0.909*** [16.078]
Year fixed effects Yes
Yes
Yes
Yes
Industry fixed effects Yes
Yes
Yes
Yes
Observations 4,973 3,722 4,989 3,646
Adjusted R-square 0.159 0.413 0.061 0.210
37
Panel B: Dynamic effect of loan sales on accounting conservatism
(1) (2) (3) (4)
VARIABLES Non_Acc C_Score Skew Avg_Rank_Con
After_1 0.000 [0.114] -0.016*** [-4.448] -0.069* [-1.769] -0.013*** [-2.827]
After_2 -0.004 [-1.401] -0.020*** [-4.377] -0.170*** [-2.968] -0.024*** [-3.624]
After_3 -0.011** [-2.567] -0.028*** [-5.388] -0.145* [-1.864] -0.040*** [-4.798]
Size -0.001 [-0.269] -0.049*** [-24.032] 0.008 [0.189] -0.045*** [-8.428]
Leverage 0.006*** [10.170] 0.015*** [4.933] 0.055*** [6.340] 0.021*** [8.166]
MB -0.000 [-0.736] -0.002** [-2.383] -0.009** [-2.005] -0.003**
Blockholder -0.004 [-0.976] -0.001 [-0.129] 0.074 [0.912] 0.024**
FirmAge -0.000 [-0.136] 0.000 [0.705] -0.013* [-1.783] -0.001
Litigation 0.002 [0.246] -0.020*** [-3.186] -0.034 [-0.196] -0.031 [-1.456]
Constant -0.026 [-1.221] 0.560*** [11.296] 1.445** [2.474] 0.912*** [16.144]
Year fixed effects Yes
Yes
Yes
Yes
Industry fixed effects Yes
Yes
Yes
Yes
Observations 4,973 3,722 4,989 3,646
Adjusted R-square 0.161 0.414 0.062 0.212
38
Table 4: Effect of loan sales on accounting conservatism: Difference-in-differences approach
This table presents the estimation results of the effects of loan trading on firms’ accounting conservatism based on a matched control sample.
Panel A reports the estimation results from a probit model predicting the probability of loan sales. The dependent variable, Trade, equals 1 if
a firm-loan is traded in the secondary loan market, and zero otherwise. We include several loan and firm characteristics in the probit model.
Panel B presents the estimation results of the effects of loan sales on accounting conservatism. The dependent variables are measures of
accounting conservatism: accumulated non-operating accruals (Non_Acc), C_Score, Skewness of earnings (Skew), and the average rank of
the three individual conservatism measures (Avg_Rank_Con). Trade is an indicator variable that equals 1 for traded loans and zero for
matched non-traded loans. After is an indicator variable that equals 1 if the firm-year is in [t + 1, t + 3] and 0 if the firm year is t. For traded
firm-loans, t is the first year of loan trading; for non-traded loans, t is the hypothetical trading year, the first trading year of the matched
traded firms. After*Trade is an interaction term between After and Trade. We select non-traded firm-loans as the matching sample based on
loan characteristics and firm characteristics in the year of loan issuance and the propensity matching score, presented in Panel A of this table.
The definition of the regression variables can be found in Appendix A. Superscripts ***, **, * correspond to statistical significance at the
1%, 5%, and 10% levels, respectively, with standard error clustered at firm level.
Panel A: Probit regression results of likelihood of loan sales
Variables Trade
Size 0.436*** [11.886]
Leverage 0.059*** [5.765]
Profitability -0.160 [-1.369]
Investgrade -1.463*** [-13.235]
Lloansize 0.175*** [5.429]
Lmat 0.728*** [14.518]
Syndicate 0.626** [2.084]
Institution loan 1.345*** [18.452]
Fcovenant 0.611*** [8.422]
Reputation 0.603*** [4.249]
Relationship 0.151** [2.528]
Constant -15.053*** [-19.694]
Loan Purposes and Loan Types fixed effects Yes
Industry and year fixed effects Yes
Observations 10,959
Pseudo R2 0.551
39
Panel B: Effect of loan sales on accounting conservatism
(1)
(2)
(3)
(4)
VARIABLES Non_Acc C_Score Skew Avg_Rank_Con
Trade 0.010* [1.746] 0.036*** [6.507] 0.332*** [3.142] 0.072*** [5.955]
After 0.004 [1.180] -0.007* [-1.947] 0.013 [0.207] 0.005 [0.832]
After*Trade -0.009* [-1.873] -0.026*** [-4.576] -0.152* [-1.778] -0.038*** [-4.373]
Size -0.007*** [-3.733] -0.048*** [-36.998] -0.096*** [-2.888] -0.055*** [-16.037]
Leverage 0.006*** [9.900] 0.017*** [6.427] 0.047*** [5.469] 0.021*** [8.265]
MB -0.001* [-1.746] -0.002*** [-3.199] -0.011** [-2.475] -0.005***
Blockholder -0.003 [-0.689] 0.003 [1.192] 0.142** [2.129] 0.030***
FirmAge -0.000 [-1.591] 0.000 [1.510] 0.001 [0.158] -0.000
Litigation -0.001 [-0.119] -0.008 [-1.477] -0.121 [-0.871] -0.032** [-2.110]
Constant -0.007 [-0.295] 0.441*** [13.763] 2.737*** [5.532] 0.888*** [12.173]
Year fixed effects Yes
Yes
Yes
Yes
Industry fixed effects Yes
Yes
Yes
Yes
Observations 5,471 4,599 5,432 4,498
Adjusted R-square 0.117 0.503 0.0585 0.300
40
Table 5: Effect of reputable and relationship lenders on accounting conservatism after loan sales
This table presents the estimation results of the differential effects of loans syndicated by reputable and relationship lead
lenders on firms’ accounting conservatism after loan trading. The dependent variables are the average rank of the three
individual conservatism measures (Avg_Rank_Con). After is an indicator variable, equals to 1 if the firm-year is in [t + 1, t
+ 3] and 0 if the firm year is year t, where t is the first year of loan trading. Reputable lead lenders is an indicator variable
that equals 1 if a loan syndicate involves at least one reputable lender (the top 10 lenders in each year) and 0 otherwise.
Relationship lead lenders is an indicator variable that equals 1 if a loan syndicate involves at least one relationship lead
lender and 0 otherwise. The definition of the regression variables can be found in Appendix A. Superscripts ***, **, *
correspond to statistical significance at the 1%, 5%, and 10% levels, respectively, with standard error clustered at firm level.
VARIABLES Reputation=0 Reputation=1 Relationship=0 Relationship=1
After -0.045*** [-3.328] -0.016** [-2.543] -0.045*** [-4.099] -0.016** [-2.321]
Size -0.043*** [-4.177] -0.044*** [-7.547] -0.053*** [-6.964] -0.045*** [-7.337]
Leverage 0.020*** [5.683] 0.022*** [7.012] 0.018*** [4.301] 0.022*** [7.300]
MB -0.002 [-1.551] -0.004** [-2.313] -0.005*** [-3.193] -0.003** [-1.990]
Blockholder 0.019 [1.162] 0.024** [2.087] -0.001 [-0.051] 0.035*** [2.981]
FirmAge 0.000 [0.197] -0.001 [-1.202] -0.001 [-1.250] -0.000 [-0.557]
Litigation -0.014 [-0.396] -0.039* [-1.660] -0.041 [-1.431] -0.022 [-0.899]
Constant 1.035*** [10.082] 0.853*** [10.849] 1.062*** [15.266] 0.859*** [9.531]
Year fixed effects Yes
Yes
Yes
Yes
Industry fixed effects Yes
Yes
Yes
Yes
Observations 823 2,823 1,114 2,532
Adjusted R-square 0.221 0.214 0.235 0.229
41
Table 6: The Effect of Loan Liquidity on Accounting Conservatism after Loan Sales
This table presents the estimation results of the differential effects of loan liquidity on firms’ accounting conservatism
after loan sales. The dependent variable is the composite measure (Avg_Rank_Con) of the three individual measures of
accounting conservatism: C_Score, Non_Acc, and Skew. We use loan bid-ask spread and average number of trading
quotes as measures of loan liquidity. Low bid-ask spread and greater number of trading quotes are associated with more
liquid loans. After is an indicator variable that equals 1 if the firm-year is in [t + 1, t + 3] and 0 if the firm year is in t,
where t is the first year of loan trading. The definition of the regression variables can be found in Appendix A.
Superscripts ***, **, * correspond to statistical significance at the 1%, 5%, and 10% levels, respectively, with standard
error clustered at firm level.
Bid-ask Spread Number of Quotes
VARIABLES (1) High (2) Low (3) Low (4) High
after 0.008 [0.888] -0.049*** [-4.833] 0.002 [0.224] -0.026*** [-2.850]
mb -0.003*** [-2.976] -0.002 [-1.182] -0.006*** [-5.447] -0.001 [-0.749]
size -0.034*** [-5.136] -0.048*** [-6.718] -0.048*** [-6.757] -0.031*** [-3.964]
leverage 0.018*** [8.555] 0.030*** [2.876] 0.016*** [6.145] 0.028*** [5.757]
litigation 0.007 [0.251] -0.050* [-1.859] -0.029 [-1.103] -0.026 [-0.764]
log (1+block holder) 0.026** [2.051] 0.032** [2.040] 0.024* [1.840] 0.022 [1.647]
log(firm age) -0.001 [-0.765] -0.001 [-0.552] -0.001* [-1.795] -0.000 [-0.008]
Constant 0.825*** [12.924] 0.774*** [12.648] 0.860*** [13.326] 0.709*** [9.782]
Year fixed effects Yes
Yes
Yes
Yes
Industry fixed effects Yes
Yes
Yes
Yes
Observations 1,403 1,396 1,407 1,403
Adjusted R-square 0.209 0.246 0.282 0.203
42
Table 7: Results of additional analysis This table presents the estimation results for the differential effects of firm size, firm age, the availability of rating, presence of financial
covenants and collateral requirements on firms' accounting conservatism after loan sales. The dependent variables are the average rank of the
three individual conservatism measures (Avg_rank_con). After is an indicator variable that equals 1 if the firm-year is in [t + 1, t + 3] and 0 if the
firm year is year t, where t is the first year of loan trading. The definition of the regression variables can be found in Appendix A. Superscripts
***, **, * correspond to statistical significance at the 1%, 5%, and 10% levels with standard error clustered at firm level, respectively.
Panel A
(1) (2) (3) (4) (5) (6)
VARIABLES firm age>median firm age<median size>median size<median notrated=1 notrated=0
after -0.017** [-2.07] -0.027*** [-3.35] -0.015** [-2.01] -0.028*** [-3.63] -0.040*** [-3.60] -0.012* [-1.75]
Control
variables Yes
Yes
Yes
Yes
Yes
Yes
Year fixed
effects Yes
Yes
Yes
Yes
Yes
Yes
Industry fixed
effects Yes
Yes
Yes
Yes
Yes
Yes
Observations 1716 1930 1823 1823
1,163
2,483
Adjusted R-
square 0.204 0.235 0.201 0.199 0.227 0.225
Panel B
(1) (2) (3) (4)
VARIABLES fcovenant=1 fcovenant=0 secured=1 secured=0
after -0.025*** [-3.997] -0.017 [-1.085] -0.028*** [-4.266] -0.007 [-0.542]
Control variables Yes
Yes
Yes
Yes
Year fixed effects Yes
Yes
Yes
Yes
Industry fixed
effects Yes
Yes
Yes
Yes
Observations 2793 673 2844 802
Adjusted R-square 0.203 0.272 0.171 0.305
43
Table 8 Robustness Test: Trade and after*trade as Endogenous Variables
This table reports the regression results using the two-stage least squares regression approach, where "trade" and
"after*trade" are treated as endogenous variables. Columns (1) and (2) report the first stage results, and Column (3)
reports the second stage conservatism regression results. The instrument variables are "Market_spread",
"Avg_market_bid", and "Avg_market_bid*post". In the first stage, we estimate the "Trade" and "After*Trade"
regressions, respectively, and use the predicted value from the first stage regressions in the second-stage
conservatism regression. The definition of the variables can be found in Appendix A. The main variable of interest
is the interaction term "After*Trade". Superscripts ***, **, * correspond to statistical significance at the 1%, 5%,
and 10% levels with standard error clustered at firm level, respectively.
Regression First Stage Reduced Form Estimation
Second stage Estimation (3) Variables Trade (1) after*trade(2)
After 0.533*** [2.61] -1.920*** [-5.62] 0.056*** [4.594]
Trade
0.497*** [5.174]
After*Trade
-0.320*** [-3.918]
Size 0.032*** [8.56] 0.023*** [8.22] -0.070*** [-20.770]
Leverage 0.044*** [9.39] 0.033*** [9.22] -0.007* [-1.761]
MB 0.003* [1.84] 0.002* [1.85] -0.002*** [-3.352]
Litigation -0.044** [-2.53] -0.032** [-2.49] 0.021** [2.361]
Blockholder 0.046*** [5.17] 0.036*** [5.31] 0.007 [1.211]
FirmAge -0.004*** [-7.10] -0.003*** [-7.05] 0.000 [0.555]
Market_spread 0.045* [1.78] -0.342* [-1.75]
Avg_market_ bid 0.052 [0.25] -0.383** [2.24]
Avg_market_ bid*After -0.116*** [-2.58] 0.455*** [6.02]
Weak Identification test
Cragg-Donald Wald F statistic 27.556
Stock-Yogo weak ID test critical values 13.43
Weak-instrument-robust inference
Anderson-Rubin Wald test F statistics 16.500 P-val 0.000
Hansen J statistic
0.838 P-val 0.360
Endogeneity test of endogenous regressors
40.034 P-val 0.000
Observations 20,159 20,159 20,159
44
Table 9: Other sensitivity checks This table presents the estimation results of the effects of loan sales on firms' accounting conservatism for a constant sample (Panel A),
and for a sample after removing loans traded after 2006 to avoid the confounding effect of financial crisis of 2007 – 2009 (Panel B) . In
Panel A, to be included in the constant sample, firms are required to have all variables available from year t to year t + 3, where t is the
first year of loan trading. In Panel B, we remove loans traded after 2006 to avoid the confounding effect of financial crisis of 2007 - 2009.
The dependent variables are measures of accounting conservatism: accumulated non-operating accruals (Non_Acc), C_Score, Skewness of
earnings (Skew), and the average rank of the three individual conservatism measures (Avg_Rank_Con). After is an indicator variable that
equals 1 if the firm-year is in [t + 1, t + 3] and 0 if the firm year is the initial loan sale year t. The definition of the regression variables can
be found in Appendix A. Superscripts ***, **, * correspond to statistical significance at the 1%, 5%, and 10% levels, respectively, with
standard error clustered at firm level.
Panel A Constant sample (1) (2) (3) (4)
VARIABLES Non_Acc C_Score Skew Avg_Rank_Con
After -0.006** [-2.340] -0.021*** [-4.840] -0.085 [-1.609] -0.017** [-2.587]
Control variables Yes
Yes
Yes
Yes
Year fixed effects Yes
Yes
Yes
Yes
Industry fixed
effects Yes
Yes
Yes
Yes
Observations 4,044 2,700 4,064 2,628
Adjusted R-square 0.145 0.459 0.0533 0.232
Panel B Removing loans traded after 2006
(1) (2) (3) (4)
VARIABLES Non_Acc C_Score Skew Avg_Rank_Con
After -0.002 [-0.744] -0.029*** [-8.019] -0.143*** [-2.679] -0.026*** [-4.353]
Control variables Yes
Yes
Yes
Yes
Year fixed effects Yes
Yes
Yes
Yes
Industry fixed effects Yes
Yes
Yes
Yes
Observations 4,278 3,209 4,288 3,141
Adjusted R-square 0.182 0.478 0.068 0.226