Accepted Manuscript Title: Interest-Rate Uncertainty, Derivatives Usage, and Loan Growth in Bank Holding Companies Author: Elijah Brewer III Sanjay Deshmukh Timothy P. Opiela PII: DOI: Reference:
S1572-3089(14)00108-9 http://dx.doi.org/doi:10.1016/j.jfs.2014.10.003 JFS 324
To appear in:
Journal of Financial Stability
Received date: Revised date: Accepted date:
10-4-2014 10-4-2014 15-10-2014
Please cite this article as: Brewer III, E., Deshmukh, S., Opiela, T.P.,Interest-Rate Uncertainty, Derivatives Usage, and Loan Growth in Bank Holding Companies, Journal of Financial Stability (2014), http://dx.doi.org/10.1016/j.jfs.2014.10.003 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Cover Letter, including Author Details
ip t
Interest-Rate Uncertainty, Derivatives Usage, and Loan Growth in Bank Holding Companies
cr
Elijah Brewer IIIa, Sanjay Deshmukhb, and Timothy P. Opielac a
M
an
us
Department of Finance, Driehaus College of Business, DePaul University, 1 East Jackson Boulevard, Chicago, IL 60604, U.S.A. E-mail:
[email protected] b Corresponding Author. Department of Finance, Driehaus College of Business, DePaul University, 1 East Jackson Boulevard, Chicago, IL 60604, U.S.A. Tel: 312.362.8472. E-mail:
[email protected] c Department of Economics, Driehaus College of Business, DePaul University, 1 East Jackson Boulevard, Chicago, IL 60604, U.S.A. E-mail:
[email protected]
d
Abstract
Ac ce p
te
We explore one channel through which interest-rate derivatives usage affects loan growth positively in bank holding companies (BHCs). If interest-rate derivatives usage allows a BHC to substitute more freely among sources of funds, then its reliance on less interest-rate-sensitive sources such as core deposits should be lower. We test the hypothesis that if BHCs use interest-rate derivatives to reduce the adverse effects of interest-rate uncertainty on lending, then their loan growth should be less sensitive to core deposit growth. We find that loan growth is less sensitive to core deposit growth for interest-rate derivatives users than for non-users and that this sensitivity is lower when the extent of derivatives usage is higher. One important implication is that the funding flexibility enjoyed by BHCs using interest-rate derivatives should allow these BHCs to provide a smoother and higher level of intermediation, leading to more stable loan growth and greater economic stability. Keywords: Interest-rate uncertainty; interest-rate derivatives usage; loan growth; financial stability
JEL Classification: G21, G32
Page 1 of 30
*Manuscript, excluding Author Details
1. Introduction In this paper, we consider an implication of the model in Deshmukh et al. (1983) to further understand the relationship between interest-rate derivatives usage and bank lending. Specifically, we explore one channel through which interest-rate derivatives usage affects loan growth
ip t
positively. In the presence of interest-rate uncertainty, a bank is less likely to access funds through sources that are relatively more interest-rate sensitive (such as federal funds borrowing, large
cr
partially-federally-insured certificates of deposits, and Eurodollar borrowing). In such cases, a bank's loan growth will be predominantly funded by sources that are less interest-rate sensitive
us
(such as core deposits), causing its loan growth to be more dependent on less interest-rate-sensitive sources.
an
Prior research identifies lending services as the fundamental basis for the role played by banks in the financial services industry (Bernanke and Lown, 1991; Sharpe and Acharya, 1992; Kashyap et al., 1993; and Brewer et al., 2001). As Diamond (1984) suggests, derivatives
M
contracting and bank lending can complement each other. By reducing exposure to systematic risk and hedging the adverse effects of interest-rate uncertainty, derivatives usage allows banks to
d
intermediate more effectively.
Brewer et al. (2001) argue that banks intermediate by engaging in two complementary
te
activities - they offer debt contracts to depositors and accept debt contracts from borrowers.
Ac ce p
Consequently, the lending specialization that banks achieve in the course of their business allows them to be cost-effective when monitoring borrowers' credit worthiness. Depositors face two choices. They can incur monitoring costs by lending directly to borrowers. Alternatively, they can supply funds to banks, delegate monitoring to them, and benefit from banks’ monitoring specialization in return. As Brewer et al. (2001) indicate, the delegation of monitoring duties creates incentive problems resulting in delegation costs, which banks can reduce by diversifying their assets. However, asset diversification is likely to be imperfect, causing banks to face some systematic risk.
Diamond's (1984) framework suggests that banks can use derivatives contracts to lower their exposure to systematic risk. Brewer et al. (2001) argue that when banks use derivatives contracts to mitigate the mismatch in the interest-rate sensitivities of assets and liabilities, they experience lower delegation costs resulting in more effective intermediation. Brewer et al. (2001) provide support for the prediction in Diamond's (1984) model that interest-rate derivatives activity 1 Page 2 of 30
and lending activity complement each other, causing derivatives usage and lending to be positively correlated. Indeed, Brewer et al. (2000) show that commercial banks have become active participants, both as end-users and/or intermediaries, in the market for interest-rate derivatives instruments and document that loan growth is higher among derivatives users.1
ip t
Deshmukh et al. (1983) develop a model of the effect of interest-rate uncertainty on a financial intermediary's choice of interest-rate exposure. Their model implies that the extent of
cr
intermediation, as measured by lending as opposed to pure brokerage activity, should be inversely related to interest-rate uncertainty (or volatility). Interest-rate uncertainty, combined with the
us
imbalance between the interest-rate sensitivity of a bank's assets and liabilities, has a significant impact on both the volatility of earnings and the common stock returns of depository institutions
an
(see Flannery and James, 1984b; Scott and Peterson, 1986; Kane and Unal, 1988, 1990; and Kwan, 1991). A banking organization can use interest-rate derivatives to lessen the adverse effects of interest-rate uncertainty, thus allowing it to increase its level of intermediation as measured by
derivatives users, supports this prediction.
M
lending. The evidence in Brewer et al. (2000), that loan growth is higher among interest-rate
d
One important implication of the model in Deshmukh et al. (1983) is that a bank that manages interest-rate uncertainty via interest-rate derivatives might find it optimal to access funds
te
from sources that are relatively more interest-rate sensitive. This greater overall access to funds is
Ac ce p
likely to allow a banking organization to move more freely among various sources of funds thereby reducing its dependence on less interest-rate-sensitive sources such as core deposits. In contrast, non-users of interest-rate derivatives, other things equal, are more likely to rely on sources of funds that are less interest-rate sensitive, such as core deposits, to fund their loan growth. Therefore, the ability to substitute more freely among sources of funds provides a potential channel through which interest-rate derivatives usage affects lending positively. As our proxy for a source of funds that is less interest-rate sensitive, we use core deposits because prior research suggests that core deposits are less interest-rate sensitive relative to other 1
Brewer et al. (2000) explore the (direct) relationship between interest-rate derivatives usage and commercial and industrial (C&I) loan growth to determine whether interest-rate derivatives and C&I loan growth serve as complements or substitutes. They, however, do not consider core deposit growth and its relationship to loan growth. Our paper differs fundamentally from theirs in that we investigate one important channel via which interest-rate derivatives usage affects (total) loan growth by examining the effect of interest-rate derivatives usage on the relationship between (total) loan growth and core deposit growth.
2 Page 3 of 30
sources of funds. For example, the evidence in Flannery and James (1984a) indicates that demand, savings, and retail time deposits, all of which are elements of core deposits, are imperfectly responsive to changes in market interest rates. Hutchison and Pennacchi (1996) find that retail deposits have a low duration implying low interest-rate sensitivity. Berlin and Mester (1999) argue
ip t
that core deposits are largely interest-rate inelastic. Hughes et al. (1996) argue that local deposits are not interest-rate sensitive and are mainly subject to geographic market shocks.
cr
In addition to exhibiting low interest-rate sensitivity, core deposits represent an important source of funding for banks. For instance, Jayaratne and Morgan (2000) document that, for banks
us
of all sizes, a significant portion of bank assets is financed by insured deposits. They find that banks in the highest (lowest) size quartile fund about 72% (83%) of their assets with insured
an
deposits. Purnanandam (2007) also finds that deposits represent the main source of funding for banks. Based on Deshmukh et al. (1983), our main prediction is that the sensitivity of loan growth to core deposit growth should be lower for interest-rate derivatives users than for non-users.
M
We draw on the empirical framework in Jayaratne and Morgan (2000) to examine the effect of interest-rate derivatives usage by bank holding companies (BHCs) on the relationship
d
between loan growth and core deposit growth.2 Jayaratne and Morgan (2000) use core deposit growth as a measure of internal funds and document that loan growth is positively related to core
te
deposit growth. We argue that interest-rate uncertainty can cause loan growth and core deposit
Ac ce p
growth to be positively related and that the strength of this positive relationship depends on the level of interest-rate uncertainty. A banking organization can use interest-rate derivatives to manage interest-rate uncertainty thereby reducing the dependence of loan growth on core deposit growth. We focus on BHCs, rather than on individual banks, as derivatives usage is more likely to 2
This empirical framework is commonly used in the literature in banking that focuses on the wedge between the costs of internal and external funds and its effect on lending. The studies in this genre use various a priori measures of the cost of external funds, such as bank asset size, bank capitalization, and affiliation with a bank holding company, to separate banks into groups. These studies then examine the relationship between loan growth and internal funds for the various groups and hypothesize the sensitivity of loan growth to internal funds to be higher for bank groups that face a higher cost of external funds. In general, this research finds that small banks, which are thought to face more information problems relative to large banks, exhibit a stronger positive relationship between loan growth and internal funds. Other factors such as bank capitalization and the affiliation with a bank holding company appear to weaken the positive relationship between loan growth and internal funds (see Peek and Rosengren, 1995; Houston et al., 1997; Jayaratne and Morgan, 2000; Kashyap and Stein, 2000; Kishan and Opiela, 2000; and Campello, 2002). These studies use loan growth as a proxy for investment spending but differ in terms of the measure they use for internal funds. For example, Houston et al. (1997) use a measure of cash flow based on net income; Kashyap and Stein (2000) use liquidity; Jayaratne and Morgan (2000) use core deposits; and Campello (2002) uses net income.
3 Page 4 of 30
be undertaken at the holding company level. We test our main prediction by examining the sensitivity of loan growth to core deposit growth for a sample of both users and non-users of interest-rate derivatives. Specifically, we examine the following hypothesis: If interest-rate derivatives usage lessens the negative effect of
ip t
interest-rate uncertainty on lending, then we should observe a lower sensitivity of loan growth to core deposit growth for derivative users relative to that for non-users, other things equal. We also
cr
consider a refinement of this hypothesis based on Diamond (1984). Diamond's model implies that a bank's ability to intermediate effectively should rise with the extent of its risk-management
us
activities. Therefore, among interest-rate derivatives users, we expect the sensitivity of loan growth to core deposit growth to vary negatively with the extent of derivatives usage. We use two
continuous measure (extent of derivatives usage).
an
measures of interest-rate derivatives usage: a discrete measure (users versus non-users) and a
We examine the effect of interest-rate derivatives usage on the relationship between loan
M
growth and core deposit growth across different size groups. We do so to ensure that the results we document on the effect of interest-rate derivatives usage do not serve as a proxy for bank size. We
d
adopt this approach because previous research suggests that asset size might affect the relationship between loan growth and core deposit growth and that size is positively related to derivatives
te
usage.3 In addition, within each size group, we control for both BHC size and factors that might
Ac ce p
affect loan growth. Our objective is to identify an effect of derivatives usage that is independent of BHC size.
Our results indicate that, for each of the different size groups, the sensitivity of loan growth to core deposit growth is lower for users of interest-rate derivatives. We also document that the higher is the extent of derivatives usage, the lower is the sensitivity of loan growth to core deposit growth. This difference in the sensitivity between users and non-users is both statistically and economically significant. As this result holds true across the different size groups, our evidence indicates that derivatives usage is not a proxy for bank size and appears to have an independent effect on the relationship between loan growth and core deposit growth. We also estimate our models across the three main loan categories and find that our results remain qualitatively the same. The results continue to hold in the presence of several robustness checks and do not appear 3
See Mian, 1996; Tufano, 1996; Geczy et al., 1997; Gay and Nam, 1998; Jayaratne and Morgan, 2000; Kashyap and Stein, 2000; Campello, 2002; and Purnanandam, 2007.
4 Page 5 of 30
to be affected by potential endogeneity concerns. We do not explore the specifics of how derivatives usage might allow a BHC to substitute among various sources of funds. This analysis is beyond the scope of our study and we leave it for future research. Our overall findings indicate that interest-rate derivatives usage appears to lessen a BHC's
ip t
systematic exposure to interest-rate uncertainty, causing the sensitivity of loan growth to core deposit growth to be lower. This result is an important contribution to the literature because we
cr
have identified one potential channel through which interest-rate derivatives affect lending. To our knowledge, this channel has not been examined previously. Our results are consistent with the
us
notion that interest-rate derivatives usage allows a banking organization to move more freely among various sources of funds thereby reducing its dependence on less interest-rate-sensitive
an
sources of funds. This ability to substitute more freely among sources of funds provides a potential channel though which interest-rate derivatives usage affects lending positively. Our results provide important implications for financial stability. For example, banking
M
organizations that use interest-rate derivatives are likely to enjoy greater funding flexibility as reflected in their lower sensitivity of loan growth to core deposit growth. A bank with greater
d
funding flexibility should be able to weather exogenous shocks to funding (especially core deposits) more effectively, resulting in less disruption to its intermediation activity. This greater
te
funding flexibility resulting in fewer funding constraints for banking organizations, should, in
Ac ce p
turn, allow for a smoother and higher level of intermediation. Therefore, greater funding flexibility should lead to more stable loan growth, a desirable outcome from an economic stability standpoint. On the receiving end, borrowers should benefit greatly from both higher and more stable lending, resulting in greater stability in the real economy. Therefore, any activity that is conducive to providing a more stable economic environment should lessen the likelihood of distress for both the bank and its borrowers, resulting in greater financial and economic stability. Another interesting implication of our results is that a wider use of interest-rate derivatives by banks to combat interest-rate uncertainty should result in a more stable banking system, especially during highly unstable periods such as during financial crises. In general, a more stable financial system implies a lower likelihood of periods of distress and hence greater overall stability. Our implications for financial stability are consistent with both Kaufman (2004) and Jokipii and Monnin (2013). For instance, Kaufman (2004) argues that macroeconomic instability 5 Page 6 of 30
is associated with bank instability and that the two phenomena often take place concurrently. Jokipii and Monnin (2013) document that banking sector stability is an important determinant of future growth in the gross domestic product. They also find that higher uncertainty about output growth follows instability in the banking sector.
ip t
We organize the paper as follows. Section 2 describes our data and the variables. Section 3
cr
discusses the empirical results. Section 4 concludes the paper with a discussion of the findings.
2. Data and Variables
us
As we note earlier, we focus on BHCs to perform our empirical analysis, rather than on individual banks, as derivatives usage is more likely to be undertaken at the holding company
an
level. Consequently, it is important to study the effect of interest-rate derivatives usage on the relationship between loan growth and core deposit growth at the holding company level. This reduces the possibility of a misclassification of banks based on interest-rate derivatives usage,
M
which may induce a bias in the ensuing empirical analysis.
We focus on interest-rate derivatives usage and not on total derivatives usage. The reason
d
is that the assets and liabilities of an intermediary (such as a BHC) are more likely to respond to unanticipated changes in interest rates (than, say, to changes in foreign exchange rates) given a
te
bank's primary focus on intermediation activities. Therefore, interest-rate derivatives provide the
Ac ce p
most effective way to manage this risk. Purnanandam (2007) finds that interest-rate derivatives represent about 90% of a bank's total derivatives usage. Interest-rate risk management is beneficial for a bank because its intermediation activities are likely to result in a mismatch of its assets and liabilities in terms of interest-rate-sensitivity and thus expose it to unanticipated changes in interest rates. Flannery and James (1984b) examine how the gap in a bank's interest-rate sensitivity of its assets and liabilities can influence its common stock returns for a given unanticipated change in interest rates.
We obtain the data for our initial sample of all publicly and non-publicly traded bank holding companies from the Federal Reserve's Bank Holding Company (FR Y-9C) Reports. Our data cover the period, 1986:Q3 to 2007:Q2. Given the tiered structure of bank holding companies in the U.S., where one BHC may own one or more other BHCs, we include only the highest-tier BHC in our sample to avoid double-counting of data. We drop all BHC quarters with non-positive values for total loans and/or total assets. We also eliminate those BHC quarters with asset growth 6 Page 7 of 30
greater than 50% and/or with a loan-to-asset ratio of less than 10%. In our empirical analysis, we use four quarterly lags of loan growth as control variables. Therefore, we exclude BHCs with fewer than five observations. These data filters result in a final sample of 4,404 bank holding companies with a total of 123,734 BHC quarter observations. We include both traded and
ip t
non-traded BHCs in order to maximize the number of observations in our sample. We use the relevant quarterly balance sheet and income statement data from the FR Y-9C Reports to construct
cr
our variables. We adjust all our variables for inflation and express them in 1986:Q3 dollars. To eliminate the effect of extreme outliers on our empirical results, we winsorize all our variables at
us
the 1st and the 99th percentiles.
We measure the variable Loan Growth as the logarithm of total loans in quarter ‘t’ minus
an
the logarithm of total loans in quarter ‘t-1’. Core Deposit Growth equals the logarithm of core deposits in quarter ‘t’ minus the logarithm of core deposits in quarter ‘t-1’. Log of Assets, a proxy for BHC asset size, equals the logarithm of the level of total book value of assets. Derivatives-User
M
is an indicator variable that equals one if the BHC is an interest-rate derivatives user in the preceding quarter, and zero otherwise. Extent of Derivatives Usage equals the ratio of the total
d
notional value of interest-rate derivatives to total assets. We draw on the extant literature to identify several control variables that are likely to affect
te
loan growth (see Peek and Rosengren, 1995; Houston et al., 1997; Jayaratne and Morgan, 2000;
Ac ce p
Kishan and Opiela, 2000; and Campello, 2002). The various control variables we use are four lags of loan growth along with one-period lagged values of non-performing loans-to-total loans, capital-to-assets ratio, logarithm of total assets, and the logarithm of the level of securities. We use the four lagged values of loan growth to control for a BHC's investment (or growth) opportunities. The lagged values of loan growth also control for the possibility that interest-rate derivatives users might be different from non-users in terms of investment (or growth) opportunities. We calculate the Capital-to-Assets ratio as the ratio of equity to total assets (Jayaratne and Morgan, 2000). We calculate Non-Performing Loans-to-Total Loans as the ratio of non-performing loans to total loans (see Jayaratne and Morgan, 2000; and Campello, 2002). We use the logarithm of the level of securities as a control variable as Kashyap and Stein (2000) document that a bank can draw on its securities to fund its loan growth. These variables also control for loan demand. We use lagged values of these variables to control for a potential endogeneity problem. The derivatives-user indicator variable that we include in the estimation also 7 Page 8 of 30
controls for the possibility that users may have a higher loan growth than non-users (Brewer et al., 2000). For brevity, we do not report the coefficients on the four lags of loan growth and on the (lagged) derivatives-user indicator variable in our tabulated results.4 Table 1 provides summary statistics for the variables used in our empirical analysis. We
ip t
present these statistics separately for users and non-users of interest-rate derivatives and calculate them using available data on each variable. The summary statistics indicate that the values for
cr
most variables are similar for users and non-users. However, there are notable differences between the two groups in that derivatives users are larger, hold more securities, and have a lower
an
[Table 1 Here]
us
core-deposits-to-total assets ratio.
3. Empirical Results
Growth 3.1.1. Derivatives Usage vs. Non-Usage
M
3.1 Effect of Derivatives Usage on the Relationship between Loan Growth and Core Deposit
d
We now examine our hypothesis: If interest-rate derivatives usage lessens the negative effect of interest-rate uncertainty on lending, then we should observe a lower sensitivity of loan
Ac ce p
things equal.
te
growth to core deposit growth for interest-rate derivatives users relative to that for non-users, other
To test whether the difference in the sensitivity of loan growth to core deposit growth is significant across interest-rate derivatives users and non-users, we interact Derivatives-User, our indicator variable, with Core Deposit Growth, which results in an interactive variable that equals core deposit growth for interest-rate derivatives users, and zero otherwise. We include this interactive variable as an independent variable along with Core Deposit Growth, Derivatives-User, and the control variables (discussed in Section 2), and estimate a regression model with Loan Growth as the dependent variable. We control for both bank (i.e., BHC) fixed effects and time fixed effects in our estimation. The coefficient on the interactive variable measures the magnitude of the differential effect (of core deposit growth on loan growth) between interest-rate derivatives users and non-users. We estimate the following regression model:
4
All of the untabulated results in this paper are available from the authors upon request.
8 Page 9 of 30
LGi,t = α0 + αi + αt + α1∗CDGi,t + α2∗CDGi,t∗Derivatives-Useri,t-1 + α3∗ Derivatives-Useri,t-1 + α4∗Wi,t-1 + εi,t
(1)
ip t
where LGi,t is loan growth, CDGi,t is core deposit growth, αi and αt are fixed firm and time effects, Derivatives-User is an indicator variable that equals one if the BHC is an interest-rate
cr
derivatives user and zero otherwise, and Wi,t-1 is a vector of control variables, which includes BHC size.
us
Kashyap and Stein (2000) and Campello (2002) show that the relationship between loan growth and internal funds is affected by bank size. Specifically, a larger bank size might reduce the
an
wedge between the costs of internal and external funds and thus affect the relationship between loan growth and internal funds. Purnanandam (2007) documents that derivatives usage increases with bank size. Research on derivatives usage of non-financial firms also indicates that larger
M
firms are more likely to be derivatives users.5 To ensure that derivatives usage does not proxy for BHC size, we follow Kashyap and Stein (2000) and Campello (2002) and divide our sample into
d
five groups based on the asset size of the BHC: below the 50th percentile, between the 50th and
above the 95th percentile.
te
75th percentile, between the 75th and 90th percentile, between the 90th and 95th percentile, and
Ac ce p
Our main argument is that derivatives usage should allow a BHC to substitute more freely among various sources of funds implying a lower dependence on core deposits. In this paper, we test this implication by examining the effect of derivatives usage on the relationship between loan growth and core deposit growth, both of which measure flows. As a starting point, however, we examine whether interest-rate derivatives users rely less on core deposits to fund their assets. As the summary statistics in Table 1 indicate, the ratio of core deposits to total assets, a stock measure, is lower for interest-rate derivatives users than for non-users. We perform a simple t test (with unequal variances) to determine whether the mean difference (between interest-rate derivatives users and non-users) is significantly different from zero. We also perform the non-parametric Wilcoxon rank-sum test to examine whether the median difference is significantly different from zero. We find that the difference in both the mean and the median ratio of core deposits to total
5
See Mian, 1996; Tufano, 1996; Geczy et al., 1997; and Gay and Nam, 1998.
9 Page 10 of 30
assets is significantly different from zero at the 1% level. Jayaratne and Morgan (2000) document that larger banks rely less on core deposits to fund their assets. Purnanandam (2007) finds that derivatives usage is higher among larger banks. To ensure that our univariate results do not proxy for a bank-size effect, we test for the difference in
ip t
both the mean and the median ratio of core deposits to total assets across the five asset-size groups. We find that, in each of the five groups, the ratio of core deposits to total assets is lower for
cr
interest-rate derivatives users than for non-users. In addition, the differences in both the mean and the median are significantly different from zero (at the 1% level) across these five groups.
us
We now address the main question underlying our study: Does interest-rate derivatives usage affect the relationship between loan growth and core deposit growth? To answer this
an
question, we estimate a regression model (specified in equation (1)) of Loan Growth on Core Deposit Growth, the Derivatives-User indicator variable, Core Deposit Growth interacted with the Derivatives-User indicator variable, size of the BHC (measured by the logarithm of total assets),
M
and the various control variables. We report the results for three size groups: between the 50th and 75th percentile, between the 75th and 90th percentile, and between the 90th and 95th percentile.
d
We ignore the two extreme groups because of a lack of variation in derivatives usage. Specifically, the number of users in “below the 50th percentile” group is very low (about 3.8% of the total
te
observations) and the number of non-users in “above the 95th percentile” group is also very low
Ac ce p
(about 3.1% of the total observations).
We present the results in Table 2, which indicate that the coefficient on core deposit growth is positive and significant at the 1% level across the three size groups. The results also indicate that the coefficient on the interactive variable is negative and significant at the 1% level across the three size groups.6 This finding indicates that, other things equal, the loan growth of derivatives users is less sensitive to core deposit growth than for non-users and is thus consistent with our hypothesis that derivatives usage lessens the reliance of loan growth on core deposit growth. In addition, our findings indicate that interest-rate derivatives usage is not a proxy for BHC size and appears to have an independent effect on the relationship between loan growth and core deposit
6
For the two excluded size groups (i.e., “below the 50th percentile” and “above the 95th percentile”), the coefficient on core deposit growth is positive and significant at the 1% level. However, as suspected, the coefficient on the interactive variable is not significant.
10 Page 11 of 30
growth.7 Our results remain qualitatively the same when we include an indicator variable (termed as State ID) to identify the state in which the BHC is located. State ID equals one for the state in which the BHC is headquartered and zero otherwise. We include State ID to control for differences in the banking environment (see Stiroh and Rumble, 2006).
ip t
[Table 2 Here]
Our results in Table 2 also indicate that the magnitude of the coefficient on core deposit
cr
growth increases with the size of the BHC. The magnitude of the (negative) coefficient on the interactive variable is substantially higher for the largest size group, which suggests that relatively
us
larger BHCs derive greater benefits from interest-rate derivatives usage. This finding provides a partial explanation for the positive relationship between derivatives usage and size documented in
an
prior research. The coefficient on the lagged derivatives-user indicator variable is not significant for any of the size groups. The coefficients on all the other control variables, however, are significant (with one exception) at the 1% level and are reasonably stable over the different size
M
groups. The signs on these coefficients are consistent with those in prior literature (see Houston et al., 1997; and Jayaratne and Morgan, 2000). Specifically, the coefficient on both the logarithm of
d
total assets and non-performing loans-to-total loans is negative and significant (at the 1% level) while the coefficient on both the capital-to-assets ratio and the logarithm of the level of securities,
te
with one exception, is positive and significant (at the 1% level).
Ac ce p
In our analysis thus far, we identify users and non-users based on whether they use interest-rate derivatives. Beginning in 1995, BHCs began reporting notional values for derivatives in separate trading and non-trading accounts. Purnanandam (2007) finds that a majority of banks uses derivatives for risk-management purposes and only the top 1% of banks (in terms of size) appears to be active traders/dealers. In fact, the top 25 banks in his sample account for about 99% of the trading/dealing activities. The exclusion of the "above the 95th percentile" size group in our empirical analysis suggests that banks likely to be predominantly dealers are not included in our analysis.
Nonetheless, for robustness, we classify users and non-users based on values in the non-trading account and estimate the models in Table 2 on data over the period, 1995:Q1-2007:Q2. Our results remain qualitatively the same across the three size groups. Our 7
We estimate all the models in Table 2 after including the lagged value of core deposits-to-total assets. All of the main results remain robust to the inclusion of the level of core deposits.
11 Page 12 of 30
findings in Table 2 thus indicate the presence of a notable risk-management component in the overall interest-rate derivatives usage of a BHC. If derivatives usage is motivated solely by the BHC acting as a broker/dealer or by proprietary trading, then we should not find any effect of derivatives usage on the relationship between loan growth and core deposit growth. Our robust
ip t
finding of a systematic effect of derivatives usage provides support for a risk-management motivation.
cr
We test for collinearity in our data by computing the variance inflation factors for the independent variables. The mean variance inflation factor for all the explanatory variables varies
us
between 1.16 and 1.57 across the three size groups.8 The low value for the variance inflation
an
factors indicates that collinearity is not a problem in our data.
3.1.2. Extent of Derivatives Usage
We now present results from a refinement of the regression model in equation (1) by
M
utilizing data on the extent of derivatives usage. Specifically, we examine the effect of the extent of interest-rate derivatives usage on the relationship between loan growth and core deposit growth.
d
Diamond's (1984) model implies that a BHC's ability to intermediate effectively should rise with the extent of its risk-management activities. Therefore, among interest-rate derivatives users, we
te
expect the sensitivity of loan growth to core deposit growth to vary negatively with the extent of
Ac ce p
derivatives usage. We calculate the extent of derivatives usage as the ratio of the total notional value of interest-rate derivatives to the total value of assets for derivative users and zero for non-users.
We present results from a regression model of Loan Growth on Core Deposit Growth, Core Deposit Growth interacted with the Extent of Derivatives Usage, the Derivatives-User indicator variable, size of the BHC (measured by the logarithm of total assets), and the various control variables. We also report the results for the "Size Above 95th Percentile" group. As we note in the previous subsection, non-users of derivatives represent only about 3% of the total observations for this group. However, among users, there is notable variation in the extent of interest-rate derivatives usage. Therefore, we include this group in Table 3 as our focus is on the effect of the 8
Collinearity is likely to be a problem if the largest variance inflation factor (VIF) is greater than ten and the mean of the VIFs (across all independent variables) is substantially larger than one. For further details, see Chatterjee et al. (2000).
12 Page 13 of 30
extent of derivatives usage on the relationship between loan growth and core deposit growth. More formally, we estimate the following regression model for each of the four size groups:
ip t
LGi,t = α0 + αi + αt + α1∗CDGi,t + α2∗ CDGi,t∗Extent of Derivatives Usagei,t-1 + α3∗ Derivatives-Useri,t-1 + α4∗Wi,t-1 + εi,t
cr
(2)
where LGi,t is loan growth, CDGi,t is core deposit growth, αi and αt are fixed firm and time
us
effects, Derivatives-User is an indicator variable that equals one if the BHC is an interest-rate derivatives user and zero otherwise, Extent of Derivatives Usage equals the ratio of the total
an
notional value of interest-rate derivatives to the total value of assets for derivatives users and zero for non-users, and Wi,t-1 is the vector of control variables, which includes BHC size. [Table 3 Here]
M
The results in Table 3 indicate that the coefficient on the interactive variable (Core Deposit Growth x Extent of Derivatives Usage) is negative and significant at the 1% level across all the
d
four size groups. This result suggests that, other things equal, the loan growth of derivatives users is less sensitive to core deposit growth than for non-users and that this sensitivity declines as the
te
extent of interest-rate derivatives usage increases.9 This finding is consistent with our hypothesis that the higher the extent of interest-rate derivatives usage, the lower the reliance of loan growth on
Ac ce p
core deposit growth.10 The coefficient on core deposit growth is positive and significant at the 1% level across all the size groups and indicates that, for non-users, when core deposit growth declines (increases) by 1%, loan growth declines (increases) by anywhere between 0.36% and 0.55% (across the four size groups). The coefficient on the interactive variable indicates that the sensitivity of loan growth to core deposit growth for users is lower than that for non-users by anywhere between 29% and 83% across the four size groups. Thus, the difference in the sensitivities between users and non-users indicates that our main finding is also economically significant. Again, the coefficient on the lagged derivatives-user indicator variable is not 9
This finding remains the same (across all the size groups) when we replace the (stand-alone) Derivatives-User indicator variable with the Extent of Derivatives Usage in equation (2). 10
We estimate all the models in Table 3 after including the lagged value of core deposits-to-total assets. All of the main results remain robust to the inclusion of the level of core deposits.
13 Page 14 of 30
significant for any of the size groups. The results with respect to the other control variables are qualitatively similar to those in Table 2 and our overall findings are robust to the inclusion of State ID. Taken together, our findings in Tables 2 and 3 indicate that for banks that use interest-rate
ip t
derivatives, loan growth is less sensitive to core deposit growth. This finding is consistent with our argument that derivatives usage should allow a banking organization to substitute more freely
cr
among alternative sources of funds to support its loan growth. This ability to substitute among various sources of funds, in turn, should allow a banking organization to provide intermediation
us
that is both higher in terms of its level and less volatile.
We also estimate equation (2) for three categories of loans: commercial and industrial
an
(C&I), real estate (RE), and consumer loans (Consumer). These three loan categories (in total) represent a very large fraction of the total loans outstanding for a typical BHC. For example, across all the sample observations, the mean (median) value of the ratio of sum of the three loan
M
categories to total loans is 94.2% (98%). These values are similar for interest-rate derivatives users and non-users. For users, the mean (median) value of the ratio of sum of the three loan categories
d
to total loans is 93.8% (96.2%) and for non-users, the mean (median) value of the ratio of sum of the three loan categories to total loans is 94.3% (98.3%).
te
Our results in Table 3 remain qualitatively the same for each of these three loan categories.
Ac ce p
We present the results for C&I loans in Table 4, for RE loans in Table 5, and for consumer loans in Table 6. These findings for the various loan categories indicate that interest-rate risk management affects lending in each of the three main loan categories. We also estimate equation (1) for the three loan categories where we use the discrete derivatives-usage variable. The results in Table 2 hold for these three loan categories. The untabulated results indicate that, for each loan category, the interactive variable is negative and significant for each of the three asset-size groups. [Table 4 Here] [Table 5 Here] [Table 6 Here]
3.2 Endogeneity Issues We now address potential endogeneity concerns arising from reverse causality and omitted-variable bias as well as the endogeneity of derivatives usage. 14 Page 15 of 30
3.2.1. Endogeneity arising from reverse causality To determine the effect of interest-rate derivatives usage on BHC lending, we examine the sensitivity of loan growth to core deposit growth across users and non-users of derivatives. One
ip t
potential problem with our approach is that demand for loans determines the level of deposits chosen by a BHC causing core deposit growth to be endogenous. As Jayaratne and Morgan (2000)
cr
illustrate, this reverse causality between loan growth and core deposit growth is actually not a problem if the goal is to examine the effect of another variable on this relationship. Therefore, the
us
direction of causality between loan growth and core deposit growth is not important for our tests given that we are examining the effect of derivatives usage on this relationship.
an
Jayaratne and Morgan (2000) also suggest that a positive relationship between loan growth and core deposit growth would exist only when a BHC faces financing frictions in the market for external funds. They argue that a positive relationship between loan growth and core deposit
M
growth is consistent with the presence of liquidity constraints. In the context of the model in Deshmukh et al. (1983), interest-rate uncertainty would cause a bank to increase its reliance on less
d
interest-rate-sensitive sources (such as core deposits) to fund its loan growth. In this sense, a bank would face "liquidity constraints" arising from interest-rate uncertainty thus resulting in the
te
positive relationship between loan growth and core deposit growth.
Ac ce p
We elaborate on the above reasoning to hypothesize that interest-rate derivatives usage allows a BHC to substitute more freely among various sources of funds causing the positive relationship between loan growth and core deposit growth to be weaker. Our results do show a positive relationship between loan growth and core deposit growth for BHCs, indicating the presence of liquidity constraints arising from interest-rate uncertainty. Our results also show that this positive relationship is weaker for users of interest-rate derivatives than for non-users. Reverse causality cannot explain this difference in the sensitivity of loan growth to core deposit growth between users and non-users and therefore cannot represent a problem in our study (Jayaratne and Morgan, 2000).
3.2.2. Endogeneity arising from an omitted-variable bias Another potential problem with our approach is that the positive relationship between loan growth and core deposit growth arises when a higher deposit growth reflects a higher demand for 15 Page 16 of 30
loans. This problem can arise from an omitted-variable bias. We address this potential omitted-variable problem by controlling for loan demand as effectively as possible by drawing on the vast literature to identify our control variables. In addition, we split the sample into several groups based on BHC-asset size, where a larger size is associated with fewer financing frictions.
ip t
Further, within each group, we examine whether the relationship between loan growth and core deposit growth is weaker for derivatives users. Omitted-variable bias might explain the positive
cr
association between loan growth and core deposit growth across all BHCs. However, it is not clear why this bias should be different for interest-rate derivatives users. Hence, omitted-variable bias
us
cannot explain the results we document on the effect of derivatives usage on the relationship
an
between loan growth and core deposit growth.
3.2.3. Endogeneity of derivatives usage
In our empirical analysis of the effect of interest-rate derivatives usage on the relationship
M
between loan growth and core deposit growth, we control for loan demand with several variables. Some of these control variables might also affect the derivatives usage of BHCs. For example,
d
Purnanandam (2007) finds that derivatives usage is more likely the higher are the costs of financial distress, the larger is the bank size, and the lower is the level of liquid assets. He also finds that
te
banks are more likely to experience financial distress, the higher is the amount of non-performing
Ac ce p
loans. We use the capital-to-assets ratio, the logarithm of total assets, the logarithm of the level of securities, and non-performing loans-to-total loans as control variables in our empirical analysis. In Table 3, we document that the sensitivity of loan growth to core deposit growth is lower the higher is the extent of interest-rate derivatives usage. To ensure that this result does not proxy for any of the other determinants of loan demand (i.e., the various control variables), we interact each of our independent variables with core deposit growth. Specifically, we interact each of the lagged values of non-performing loans-to-total loans, capital-to-assets ratio, logarithm of total assets, and the logarithm of the level of securities with core deposit growth. We include these interactive variables along with all the variables from Table 3 in our empirical specification for loan growth. The negative effect of the extent of derivatives usage on the relationship between loan growth and core deposit growth continues to hold. In other words, we still find that, across the various size groups, the sensitivity of loan growth to core deposit growth decreases as the extent of interest-rate derivatives usage increases. This robustness check indicates that derivatives usage has 16 Page 17 of 30
an independent and a significant incremental effect on the relationship between loan growth and core deposit growth. The inclusion of the various interactive variables in the above robustness check also rules out other explanations of our results in Table 3. For instance, Jayaratne and Morgan (2000)
ip t
document the relationship between loan growth and core deposit growth to be weaker in a bank with a higher capital ratio. Peek and Rosengren (1996) and Kishan and Opiela (2000) provide
cr
evidence consistent with the notion that bank capital can reduce the cost of external funds arising from asymmetric information. Our inclusion of the variable representing the interaction of core
us
deposit growth and the capital-to-assets ratio controls for a potential capital-based explanation of our results. Kashyap and Stein (2000) find that the relationship between loan growth and internal
an
funds is weaker when a bank holds a relatively higher level of securities in its asset portfolio. Our inclusion of the variable representing the interaction of core deposit growth and the logarithm of the level of securities controls for this potential explanation of our findings. The main result, that
M
the sensitivity of loan growth to core deposit growth is lower for interest-rate derivative users than for non-users, remains after controlling for other potential explanations of the relationship between
te
4. Summary and Conclusion
d
loan growth and core deposit growth.
Ac ce p
In this paper, we shed further light on the documented positive relationship between interest-rate derivatives usage and loan growth for banks. We draw on the model in Deshmukh et al. (1983) to explore one channel through which interest-rate derivatives usage affects loan growth. In the presence of interest-rate uncertainty, a bank is less likely to access funds through sources that are relatively more interest-rate sensitive to fund its loan growth and instead rely on less interest-rate-sensitive sources. However, a bank that manages interest-rate uncertainty via derivatives might find it optimal to access funds from sources that are relatively more interest-rate sensitive. Consequently, interest-rate derivatives usage can potentially allow a banking organization to move more freely among various sources of funds thereby reducing their reliance on less interest-rate-sensitive sources. In our empirical analysis, we use core deposits as a proxy for a source of funds that is less sensitive to interest rates. As previous research indicates, core deposits also represent a predominant source of funding for banks. A greater reliance on less interest-rate-sensitive sources might limit a bank's loan growth. 17 Page 18 of 30
We test the hypothesis that if interest-rate derivatives usage allows a bank to mitigate the adverse effects of interest-rate uncertainty, then the sensitivity of loan growth to core deposit growth should be lower for interest-rate derivatives users than for non-users. Based on Diamond (1984), we also test a follow-up hypothesis: Among interest-rate derivatives users, the sensitivity
ip t
of loan growth to core deposit growth will be negatively related to the extent of derivatives usage. To control for confounding effects related to bank size, we examine the effect of derivatives usage
groups. We draw on previous research to form these size groups.
cr
on the relationship between loan growth and core deposit growth across different asset-size
us
We find that, within each asset size group, the sensitivity of loan growth to core deposit growth is lower for users of interest-rate derivatives after controlling for both BHC size and
an
several other factors that might affect loan growth. We also find that, within each size group, the sensitivity of loan growth to core deposit growth is lower, the higher is the extent of derivatives usage. The difference in the sensitivity between users and non-users is both statistically and
M
economically significant. These results also indicate that derivatives usage does not proxy for BHC asset size given that we estimate our models across different size groups and control for size
d
explicitly within each group. Our results remain qualitatively the same in the presence of several robustness checks and do not appear to be affected by endogeneity concerns. We thus document an
te
independent and incremental effect of interest-rate derivatives usage on the relationship between
Ac ce p
loan growth and core deposit growth.
The main contribution of our paper is that we identify one potential channel through which interest-rate derivatives affect bank lending. Our results provide support for the idea that interest-rate derivatives usage allows a banking organization to move more freely among various sources of funds thereby reducing their reliance on less interest-rate-sensitive sources. Hence, the ability to substitute more freely among sources of funds provides a potential channel though which interest-rate derivatives usage has a positive effect on bank lending. Our paper also provides two important implications. First, the use of interest-rate derivatives by banking organizations, to mitigate the adverse effects of interest-rate uncertainty on their lending, should result in a positive externality in terms of its effect on both financial and economic stability. Banking organizations that use interest-rate derivatives are likely to enjoy greater funding flexibility and consequently fewer funding constraints. This funding flexibility, as reflected in a lower sensitivity of loan growth to core-deposit growth, should allow these 18 Page 19 of 30
organizations to provide a smoother and higher level of intermediation. On the receiving end, borrowers should benefit greatly from both higher and more stable lending, leading to greater stability in the real economy. The use of interest-rate derivatives thus provides a bank with more funding flexibility by making it less susceptible to liquidity constraints associated with funding
ip t
shocks. This flexibility should reduce the variability in funding. To the extent that the variability in loan growth depends on the variability in funding, the use of interest-rate derivatives could reduce
cr
loan variability and have positive implications for macroeconomic stability. Second, a wider use of interest-rate derivatives by banks to combat interest-rate uncertainty should result in a more stable
us
banking system, especially during financial crises. Therefore, regulation that hampers both the use of derivatives and the cost of using them could have negative implications for both financial
an
stability and broader resource allocation in the economy.
Acknowledgements
M
We are grateful for the valuable suggestions and guidance provided by the editor, Iftekhar Hasan, and an anonymous referee. We are also grateful to Lamont Black, Robert DeYoung, and Philip
Ac ce p
te
d
Strahan for their comments.
19 Page 20 of 30
References Berlin, M. and L. J. Mester, 1999, "Deposits and relationship lending," Review of Financial Studies, 12(3), 579-607.
ip t
Bernanke, B. and C. Lown, 1991, "The credit crunch," Brookings Papers on Economic Activity, 2, 205-247.
cr
Brewer III, E., B. A. Minton, and J. T. Moser, 2000, "Interest-rate derivatives and bank lending," Journal of Banking and Finance, 24(3), 353-379.
us
Brewer III, E., W. Jackson III, and J. T. Moser, 2001, "The value of using interest rate derivatives to manage risk at U.S. banking organizations," Economic Perspectives, Federal Reserve Bank of Chicago 3Q, 49-66.
an
Campello, M., 2002, "International capital markets in financial conglomerates: evidence from small bank responses to monetary policy," Journal of Finance, 57(6), 2773-2805.
M
Chatterjee, S., A. S. Hadi, and B. Price, 2000, Regression Analysis by Example, 3rd ed, (John Wiley & Sons).
d
Deshmukh, S. D., S. I. Greenbaum, and G. Kanatas, 1983, "Interest rate uncertainty and the financial intermediary's choice of exposure," Journal of Finance, 38(1), 141-147.
te
Diamond, D., 1984, "Financial intermediation and delegated monitoring," The Review of Economic Studies, 51(3), 393-414.
Ac ce p
Flannery, M. J. and C. James, 1984a, "Market evidence on the effective maturity of bank assets and liabilities," Journal of Money, Credit, and Banking, 16(4), 435-445. Flannery, M. J. and C. James, 1984b, "The effect of interest rate changes on the common stock returns of financial institutions," Journal of Finance, 39(4), 1141-1153. Gay, G. D. and J. Nam, 1998, "The underinvestment problem and corporate derivatives use," Financial Management, 27(4), 53-69. Geczy, C., B. A. Minton, and C. Schrand, 1997, "Why firms use currency derivatives," Journal of Finance, 52(4), 1323-1354. Houston, J., C. James, and D. Marcus, 1997 "Capital market frictions and the role of internal capital markets in banking," Journal of Financial Economics, 46(2), 135-164. Hughes, J. P., W. W. Lang, L. J. Mester, and C. Moon, 1996, "Efficient banking under interstate branching," Working Paper 96-9, Federal Reserve Bank of Philadelphia.
Page 21 of 30
Hutchison, D. E. and G. G. Pennacchi, 1996, "Measuring rents and interest rate risk in imperfect financial markets: The case of retail bank deposits," Journal of Financial and Quantitative Analysis, 31(3), 399-417.
ip t
Jayaratne, J. and D. P. Morgan, 2000, "Capital market frictions and deposit constraints at banks," Journal of Money, Credit, and Banking, 32(1), 74-92.
cr
Jokipii, T. and P. Monnin, 2013, "The impact of banking sector stability on the real economy," Journal of International Money and Finance, 32, 1-16.
us
Kane, E. J. and H. Unal, 1988, "Change in market assessments of deposit-institution riskiness," Journal of Financial Services Research, 1, 207-229. Kane, E. J. and H. Unal, 1990, "Modeling structural and temporal variation in the market's valuation of banking firms," Journal of Finance, 45, 113-136.
an
Kashyap, A., J. Stein, and D. Wilcox, 1993, "Monetary policy and credit conditions: Evidence from the composition of external finance," American Economic Review, 83(1), 78-98.
M
Kashyap, A. and J. C. Stein, 2000, "What do a million observations on banks say about the transmission of monetary policy?," American Economic Review, 90(3), 407-428.
d
Kaufman, G., 2004, "Macroeconomic stability, bank soundness, and designing optimum regulatory structures," Multinational Finance Journal, 8 (3 & 4), 141-171.
Ac ce p
te
Kishan, R. P. and T. P. Opiela, 2000, "Bank size, bank capital and the bank lending channel," Journal of Money, Credit, and Banking, 32(1), 129-141. Kwan, S. H., 1991, "Reexamination of interest rate sensitivity of commercial bank stock returns using a random coefficient model," Journal of Financial Services Research, 5, 61-76. Mian, S. L., 1996, "Evidence on corporate hedging policy," Journal of Financial and Quantitative Analysis, 31(3), 419-439. Peek, J. and E. Rosengren (1995), "The capital crunch: Neither a borrower nor lender be," Journal of Money, Credit, and Banking, 27(3), 626-38. Purnanandam, A., 2007, "Interest rate derivatives at commercial banks: An empirical investigation, Journal of Monetary Economics, 54(6), 1769-1808. Scott, W. L. and R. L. Peterson, 1986, "Interest rate risk and equity values of hedged and unhedged financial intermediaries," Journal of Financial Research, 9, 325-329. Sharpe, S. and S. Acharya, 1992, "Loan losses, bank capital and the credit crunch," Working Paper, Federal Reserve Board of Governors, Washington D.C.
Page 22 of 30
Stiroh, K. J. and A. Rumble, 2006, "The dark side of diversification: The case of US financial holding companies," Journal of Banking and Finance, 30, 2131-2161.
Ac ce p
te
d
M
an
us
cr
ip t
Tufano, P., 1996, "Who manages risk? An empirical investigation of risk management in the gold mining industry," Journal of Finance, 51(4), 1097-1137.
Page 23 of 30
ip t
Table 1 Summary Statistics The sample consists of 4,404 bank holding companies (BHCs) drawn from the Federal Reserve's Bank Holding Company (FR Y-9C) Reports. We base our summary statistics on quarterly data over the period, 1986:Q3-2007:Q2, and calculate them by using all available BHC-quarter observations on each variable. We calculate Loan Growth and Core Deposit Growth over two consecutive quarters. Log of Securities equals the logarithm of the level of securities held by the BHC. Log of Assets equals the logarithm of the level of total assets. Capital-to-Assets equals the ratio of equity to total assets. We calculate Non-Performing Loans-to-Total Loans as the ratio of non-performing loans to total loans. Extent of Derivatives Usage equals the ratio of the total notional value of interest-rate derivatives to total assets. Core-Deposits-to-Total Assets equals the ratio of core deposits to total assets.
N
Loan Growth
0.0138
0.0189
18203
Core Deposit Growth
0.0056
0.0154
18781
Log of Securities
12.6742
12.7862
Log of Assets
14.0671
14.2885
Capital-to-Assets
0.0769
0.0796
Non-Performing Loans-to-Total Loans
0.0082
Extent of Derivatives Usage
0.0372
Core-Deposits-to-Total Assets
0.6398
Median
Mean
N
0.0124
0.0164
94805
0.0049
0.0123
94757
us
Mean
10.8484
10.9183
95823
18968
12.1000
12.2530
95839
18969
0.0831
0.0856
95842
0.0123
17528
0.0081
0.0131
88780
19340
0
0
99363
0.6126
19340
0.7420
0.7180
99363
M
18968
d
an
Median
te
0.1248
Ac ce p
Variable
Derivatives Non-Users
cr
Derivatives Users
Page 24 of 30
cr
ip t
Table 2 Sensitivity of Loan Growth to Core Deposit Growth: Derivatives Users versus Non-Users This table presents results from the estimation of a regression model of Loan Growth on Core Deposit Growth and Core Deposit Growth interacted with an indicator variable, Derivatives-User. We estimate the model on the combined derivatives user and non-user samples with available quarterly data in the Federal Reserve's Bank Holding Company (FR Y-9C) Reports over the period, 1986:Q3-2007:Q2. We calculate Loan Growth (LG) and Core Deposit Growth (CDG) over two consecutive quarters. Derivatives-User equals one if the BHC is an interest-rate derivatives user (in the preceding quarter) and zero otherwise. Log of Securities equals the logarithm of the level of securities held. Log of Assets equals the logarithm of the level of total assets. Capital-to-Assets equals the ratio of equity to total assets. Non-Performing Loans-to-Total Loans equals the ratio of non-performing loans to total loans. Columns 2 through 4 in the table present results from a regression model for a subsample based on the size of the BHC. The t-statistics are in parentheses below the coefficients. ***, **, and * represent significance at the 1%, 5%, and 10% levels, respectively. We estimate the following model where αi and αt are fixed firm and time effects, respectively, and Wi,t-1 is the vector of control variables.
(1)
(4) Size Between th th 90 and 95 Percentile
0.5260*** (10.53)
0.5407*** (7.92)
0.4637*** (62.21)
0.5569*** (29.87)
-0.0830*** (-6.21)
-0.0449*** (-3.64)
-0.1206*** (-5.34)
-0.0588*** (-25.93)
-0.0550*** (-19.84)
-0.0628*** (-11.58)
Capital-to-Assets
0.0700*** (2.64)
0.1513*** (4.25)
0.1186 (1.44)
Log of Securities
0.0270*** (23.75)
0.0262*** (15.54)
0.0314*** (8.87)
-0.6169*** (-20.13)
-0.6112*** (-13.05)
-0.6243*** (-6.71)
Yes Yes
Yes Yes
Yes Yes
122.80*** 25093
98.44*** 13051
36.16*** 3673
0.4034
0.4535
0.4941
Core Deposit Growth
0.3631*** (73.29)
Ac ce p
Log of Assets
te
Core Deposit Growth x Derivatives-User
Non-Performing Loansto-Total Loans
Time Fixed Effects Firm Fixed Effects F Number of Observations 2 Adjusted R
M
0.6660*** (17.48)
d
Constant
an
(3) Size Between th th 75 and 90 Percentile
Variable
(2) Size Between th th 50 and 75 Percentile
us
LGi,t = α0 + αi + αt + α1 ∗ CDGi,t + α2∗ CDGi,t ∗ Derivatives-Useri,t-1 + α3∗ Derivatives-Useri,t-1 + α4 ∗ Wi,t-1 + εi,t
Page 25 of 30
us
cr
ip t
Table 3 Sensitivity of Loan Growth to Core Deposit Growth: Extent of Derivatives Usage This table presents results from the estimation of a regression model of Loan Growth on Core Deposit Growth and Core Deposit Growth interacted with Extent of Derivatives Usage. We estimate the model on the combined derivatives user and non-user samples with available quarterly data in the Federal Reserve's Bank Holding Company (FR Y-9C) Reports over the period, 1986:Q3-2007:Q2. We calculate Loan Growth (LG) and Core Deposit Growth (CDG) over two consecutive quarters. Extent of Derivatives Usage equals the ratio of the total notional value of interest-rate derivatives to total assets for derivatives users and zero for non-users. Derivatives-User equals one if the BHC is an interest-rate derivatives user (in the preceding quarter) and zero otherwise. Log of Securities equals the logarithm of the level of securities held. Log of Assets equals the logarithm of the level of total assets. Capital-to-Assets equals the ratio of equity to total assets. Non-Performing Loans-to-Total Loans equals the ratio of non-performing loans to total loans. Columns 2 through 5 in the table present results from a regression model for a subsample based on the size of the BHC. The t-statistics are in parentheses below the coefficients. ***, **, and * represent significance at the 1%, 5%, and 10% levels, respectively. We estimate the following model where αi and αt are fixed firm and time effects, respectively, and Wi,t-1 is the vector of control variables. LGi,t = α0 + αi + αt + α1∗ CDGi,t + α2∗ CDGi,t ∗ Extent of Derivatives Usagei,t-1 + α3∗ Derivatives-Useri,t-1 + α4∗ Wi,t-1 + εi,t
(1)
(3) Size Between th th 75 and 90 Percentile
(4) Size Between th th 90 and 95 Percentile
(5) Size Above th 95 Percentile
Constant
0.6693*** (17.56)
0.5259*** (10.55)
0.5695*** (8.37)
0.0204 (0.34)
Core Deposit Growth
0.3553*** (75.33)
0.4608*** (72.33)
0.4957*** (40.70)
0.5457*** (36.74)
-0.3821*** (-6.99)
-0.2721*** (-4.04)
-0.1564*** (-5.55)
M
d
-0.2259*** (-3.64)
Ac ce p
Core Deposit Growth x Extent of Derivatives Usage
te
Variable
an
(2) Size Between th th 50 and 75 Percentile
Log of Assets
-0.0592*** (-26.12)
-0.0547*** (-19.78)
-0.0654*** (-12.05)
-0.0109** (-2.48)
Capital-to-Assets
0.0725*** (2.73)
0.1523*** (4.29)
0.1021 (1.24)
0.0873 (1.17)
Log of Securities
0.0271*** (23.84)
0.0260*** (15.44)
0.0325*** (9.09)
0.0105*** (2.90)
Non-Performing Loansto-Total Loans
-0.6187*** (-20.18)
-0.6077*** (-12.99)
-0.6276*** (-6.73)
-0.6696*** (-7.23)
Yes Yes
Yes Yes
Yes Yes
Yes Yes
122.36*** 25093 0.4028
99.16*** 13051 0.4551
35.88*** 3673 0.4923
32.61*** 3584 0.4585
Time Fixed Effects Firm Fixed Effects F Number of Observations 2 Adjusted R
Page 26 of 30
us
cr
ip t
Table 4 Sensitivity of C&I Loan Growth to Core Deposit Growth: Extent of Derivatives Usage This table presents results from the estimation of a regression model of C&I (Commercial and Industrial) Loan Growth on Core Deposit Growth and Core Deposit Growth interacted with Extent of Derivatives Usage. We estimate the model on the combined derivatives user and non-user samples with available quarterly data in the Federal Reserve's Bank Holding Company (FR Y-9C) Reports over the period, 1986:Q3-2007:Q2. We calculate C&I Loan Growth (C&I LG) and Core Deposit Growth (CDG) over two consecutive quarters. Extent of Derivatives Usage equals the ratio of the total notional value of interest-rate derivatives to total assets for derivatives users and zero for non-users. Derivatives-User equals one if the BHC is an interest-rate derivatives user (in the preceding quarter) and zero otherwise. Log of Securities equals the logarithm of the level of securities held. Log of Assets equals the logarithm of the level of total assets. Capital-to-Assets equals the ratio of equity to total assets. Non-Performing Loans-to-Total Loans equals the ratio of non-performing loans to total loans. Columns 2 through 5 in the table present results from a regression model for a subsample based on the size of the BHC. The t-statistics are in parentheses below the coefficients. ***, **, and * represent significance at the 1%, 5%, and 10% levels, respectively. We estimate the following model where αi and αt are fixed firm and time effects, respectively, and Wi,t-1 is the vector of control variables. C&I LGi,t = α0 + αi + αt + α1∗CDGi,t + α2∗ CDGi,t ∗ Extent of Derivatives Usagei,t-1 + α3∗ Derivatives-Useri,t-1 + α4∗Wi,t-1 + εi,t
(4) Size Between th th 90 and 95 Percentile
(5) Size Above th 95 Percentile
0.6791*** (4.40)
0.8720*** (6.26)
-0.1962** (-2.02)
an
(3) Size Between th th 75 and 90 Percentile
0.4476*** (4.33)
Core Deposit Growth
0.3820*** (30.17)
0.4536*** (29.87)
0.4371*** (16.56)
0.6169*** (24.35)
-0.7457*** (-5.96)
-0.3964*** (-2.69)
-0.2251** (-4.71)
-0.0635*** (-10.61)
-0.0489*** (-7.53)
-0.0737*** (-6.32)
0.0120 (1.60)
Capital-to-Assets
0.0063 (0.09)
0.3440*** (4.07)
-0.0409 (-0.24)
0.2911** (2.26)
Log of Securities
0.0230*** (7.62)
0.0209*** (5.26)
0.0199*** (2.59)
0.0003 (0.04)
-0.8127*** (-10.11)
-0.8495*** (-7.74)
-1.1449*** (-5.86)
-0.9013*** (-5.77)
Yes Yes
Yes Yes
Yes Yes
Yes Yes
27.10*** 25936 0.1131
21.56*** 13634 0.1473
9.00*** 3794 0.2057
16.47*** 3677 0.3244
-0.4921*** (-2.93)
Ac ce p
Core Deposit Growth x Extent of Derivatives Usage
d
Constant
te
Variable
(2) Size Between th th 50 and 75 Percentile
M
(1)
Log of Assets
Non-Performing Loansto-Total Loans
Time Fixed Effects Firm Fixed Effects F Number of Observations 2 Adjusted R
Page 27 of 30
us
cr
ip t
Table 5 Sensitivity of RE Loan Growth to Core Deposit Growth: Extent of Derivatives Usage This table presents results from the estimation of a regression model of RE (Real-Estate) Loan Growth on Core Deposit Growth and Core Deposit Growth interacted with Extent of Derivatives Usage. We estimate the model on the combined derivatives user and non-user samples with available quarterly data in the Federal Reserve's Bank Holding Company (FR Y-9C) Reports over the period, 1986:Q3-2007:Q2. We calculate Real-Estate Loan Growth (RE LG) and Core Deposit Growth (CDG) over two consecutive quarters. Extent of Derivatives Usage equals the ratio of the total notional value of interest-rate derivatives to total assets for derivatives users and zero for non-users. Derivatives-User equals one if the BHC is an interest-rate derivatives user (in the preceding quarter) and zero otherwise. Log of Securities equals the logarithm of the level of securities held. Log of Assets equals the logarithm of the level of total assets. Capital-to-Assets equals the ratio of equity to total assets. Non-Performing Loans-to-Total Loans equals the ratio of non-performing loans to total loans. Columns 2 through 5 in the table present results from a regression model for a subsample based on the size of the BHC. The t-statistics are in parentheses below the coefficients. ***, **, and * represent significance at the 1%, 5%, and 10% levels, respectively. We estimate the following model where αi and αt are fixed firm and time effects, respectively, and Wi,t-1 is the vector of control variables. RE LGi,t = α0 + αi + αt + α1∗CDGi,t + α2∗ CDGi,t ∗ Extent of Derivatives Usagei,t-1 + α3∗ Derivatives-Useri,t-1 + α4∗Wi,t-1 + εi,t
(4) Size Between th th 90 and 95 Percentile
(5) Size Above th 95 Percentile
0.6208*** (7.75)
0.6106*** (7.24)
0.1940** (2.45)
an
(3) Size Between th th 75 and 90 Percentile
0.7758*** (16.21)
Core Deposit Growth
0.3722*** (63.42)
0.5057*** (64.22)
0.5662*** (35.53)
0.6180*** (29.41)
Core Deposit Growth x Extent of Derivatives Usage
-0.1674** (-2.15)
-0.6030*** (-9.29)
-0.4120*** (-4.63)
-0.0925** (-2.33)
-0.0591*** (-21.13)
-0.0601*** (-17.74)
-0.0696*** (-9.87)
-0.0185*** (-3.01)
Capital-to-Assets
0.0497 (1.52)
0.1382*** (3.16)
-0.1111 (-1.06)
0.0480 (0.45)
Log of Securities
0.0242*** (17.17)
0.0247*** (11.96)
0.0354*** (7.64)
0.0105** (2.05)
-0.7116*** (-18.75)
-0.7046*** (-12.20)
-0.8295*** (-6.92)
-0.6505*** (-5.00)
Yes Yes
Yes Yes
Yes Yes
Yes Yes
83.40*** 25936 0.3122
73.21*** 13634 0.3682
24.34*** 3787 0.3821
22.32*** 3676 0.3750
Ac ce p
d
Constant
te
Variable
(2) Size Between th th 50 and 75 Percentile
M
(1)
Log of Assets
Non-Performing Loansto-Total Loans
Time Fixed Effects Firm Fixed Effects F Number of Observations 2 Adjusted R
Page 28 of 30
cr
ip t
Table 6 Sensitivity of Consumer Loan Growth to Core Deposit Growth: Extent of Derivatives Usage This table presents results from a regression model of Consumer Loan Growth on Core Deposit Growth and Core Deposit Growth interacted with Extent of Derivatives Usage. We estimate the model on the combined derivatives user and non-user samples with available quarterly data in the Federal Reserve's BHC (FR Y-9C) Reports over the period, 1986:Q3-2007:Q2. We calculate Consumer Loan Growth (Consumer LG) and Core Deposit Growth (CDG) over two consecutive quarters. Extent of Derivatives Usage equals the ratio of the total notional value of interest-rate derivatives to total assets for derivatives users and zero for non-users. Derivatives-User equals one if the BHC is a derivatives user (in the preceding quarter) and zero otherwise. Log of Securities equals the logarithm of the level of securities. Log of Assets equals the logarithm of the level of total assets. Capital-to-Assets equals the ratio of equity to total assets. Non-Performing Loans-to-Total Loans equals the ratio of non-performing loans to total loans. Columns 2 through 5 in the table present results from a regression model for a subsample based on BHC size. The t-statistics are in parentheses below the coefficients. ***, **, and * represent significance at the 1%, 5%, and 10% levels, respectively. We estimate the following model where αi and αt are fixed firm and time effects, respectively, and Wi,t-1 is the vector of control variables.
Variable
(2) Size Between th th 50 and 75 Percentile
(3) Size Between th th 75 and 90 Percentile
(4) Size Between th th 90 and 95 Percentile
(5) Size Above th 95 Percentile
0.3750*** (2.58)
0.4795*** (3.46)
0.2856** (2.41)
0.4932*** (34.32)
0.5075*** (19.64)
0.5254*** (16.57)
an
(1)
us
Consumer LGi,t = α0 + αi + αt + α1∗CDGi,t + α2∗ CDGi,t ∗ Extent of Derivatives Usagei,t-1 + α3∗ Derivatives-Useri,t-1 + α4∗Wi,t-1 + εi,t
0.7633*** (7.98)
Core Deposit Growth
0.4269*** (36.42)
Core Deposit Growth x Extent of Derivatives Usage
-0.7766*** (-5.03)
-0.7618*** (-6.45)
-0.5054*** (-3.53)
0.0527 (0.89)
-0.0631*** (-11.37)
-0.0464*** (-7.59)
-0.0671*** (-5.73)
-0.0264*** (-2.88)
Capital-to-Assets
-0.1167* (-1.78)
-0.0594 (-0.74)
-0.0253 (-0.15)
-0.0552 (-0.35)
Log of Securities
0.0179*** (6.41)
0.0144*** (3.84)
0.0383*** (5.09)
0.0104 (1.36)
Non-Performing Loansto-Total Loans
-0.6191*** (-8.28)
-0.5250*** (-5.09)
0.0018 (0.01)
-0.5774*** (-3.02)
Yes Yes
Yes Yes
Yes Yes
Yes Yes
31.21*** 25602 0.1436
24.17*** 13480 0.1947
8.99*** 3666 0.2241
9.96*** 3624 0.1927
d
te
Ac ce p
Log of Assets
Time Fixed Effects Firm Fixed Effects
F Number of Observations 2 Adjusted R
M
Constant
Page 29 of 30
*Highlights (for review)
Highlights Explore the effect of interest-rate derivatives usage on loan growth.
Examine whether usage affects the impact of interest-rate uncertainty on lending.
Test whether usage reduces reliance on less interest-rate-sensitive funds.
Loan growth of derivatives users is less sensitive to core deposit growth.
Usage provides funding flexibility, leading to lending/economic stability.
Ac ce p
te
d
M
an
us
cr
ip t
Page 30 of 30