The evolution of derivative use by community banks from 1995-2003.

Robicheaux, Sara H.

INTRODUCTION

It goes without saying that the banking industry has changed over the last decade. There have been several seminal events to cause the changes, notably the easing of Depression era regulations and as a result the ever-increasing penetration of the large "mega-banks" into smaller and smaller markets. Additionally, and likely a result of the increased competition caused by the aforementioned events, banks have been forced to dramatically increase their product offerings. It is now common for the smallest of banks to offer such products as home equity lines of credit, certificates of deposit with the interest rate tied to the performance of an equity index, and even interest-only home mortgages.

The increased competition from the mega-banks has also forced the small bank to manage itself in ways that they did not need to in the past. For example, it is not uncommon to see smaller banks using more sophisticated derivative securities as a vital tool in Asset/Liability management. Clearly, the use of these complex financial tools introduces additional accounting and regulatory risks to the small bank. It is our goal in this paper to determine how this subset of banks has evolved since 1995 in their use of derivative securities. A second question that we will attempt to answer is whether or not derivative use affects the performance of the banks in our sample, and whether the results are driven by the largest banks.

The firms we examine are U.S. commercial banks with total assets between $100 million and $1 billion, which we will refer to as community banks. We choose these firms because they are small enough so that derivatives should be a relatively new tool for them as compared to the larger national banks. Yet, they are large enough that they should have enough interest rate risk on their balance sheet to warrant hedging with derivatives. Also, as pointed out by Carter and Sinkey (1998) this subset of banks should be using derivatives for hedging purposes rather than for dealer activity.

The paper proceeds as follows. Section 2 reviews the existing literature. Section 3 describes the data used in our study and our methodology. Section 4 presents the results and Section 5 concludes.

RELATED LITERATURE

There are several studies that have examined the effect of derivative use on the firm. Smith and Stulz (1985) were the first to discuss how hedging with derivatives can be used to offset interest-rate risk and decrease the probability of insolvency. Carter and Sinkey (2000) use year-end data from 1996 and find approximately 5% of community banks use derivatives. They examine the user versus nonuser banks and find usage related to riskier capital structure, larger maturity mismatches between assets and liabilities, greater net-loan charge offs, and lower net interest margins. An additional set of researchers are studying swap positions of U.S. commercial banks. Gorton and Rosen (1995) find that interest rate risk from swap positions has had very little effect on the systematic risk of the banking industry as a whole. Boukrami (2003) finds larger banks with better asset quality and higher capitalization use swaps more intensely.

One study which focuses exclusively on community banks is Deyoung, Hunter, and Udell's (2004) examinion of the past, present, and future of community banks. They show that although regulatory and technological changes have created increased competition for community banks, it has also left well-managed community banks with a potential exploitable strategic position in the industry. Carter and Sinkey's (1998) evaluation of derivative use by community banks from 1990-1993 shows that cost-related incentives motivate interest-rate derivative use. Specifically, they show that community banks use derivatives to control interest-rate risk, with swaps being used in conjunction with credit-risk minimization mechanisms. We will build on the existing literature in two ways. First, we look at an extended time period, and one in which increased competition has forced community banks to use more complex management techniques. Second, we will examine whether the change in derivative use and their effect on performance is driven by the largest quartile of banks.

Although previous research has looked at the use of derivatives by commercial and community banks, this paper will be the first to look specifically at derivative use over such a long time period by so many banks. We are examining the change in derivative use by 4,397 banks over a 9 year period, 1995-2003. This paper is the first to look at how derivative use has evolved among community banks over an extended time frame.

DATA, VARIABLE SPECIFICATION, AND METHODOLOGY

We gather our community bank derivative data from the Annual Reports of Condition and Income (Call Report) required by the Federal Deposit Insurance Corporation. We use the definition of community banks as defined by Carter and Sinkey (1998). We examine US commercial banks with Total Assets between $100 million and $1 billion. There are 4,397 banks and/or bank holding companies in this category between 1995 and 2003. Our sample cannot start earlier than 1995 because there is a lack of derivative data available prior to 1995. The call reports include the notional amount of derivative contracts and the type of derivatives (e.g. option, swap, futures contract) used for hedging purposes.

The Call Report breaks down the total notional amount of derivative contracts into the following categories: futures contracts, forward contracts, option contracts and swaps. Option contracts are further broken down into written or purchased options and exchange trades or over the counter (OTC). To evaluate the evolution of derivative use by community banks we divide our sample into banks who use derivatives and those who do not use derivatives. For each sample we compute and compare summary statistics. We also compute summary statistics on the number of users each year and the type of derivative use. The variables that we will use in our analysis are defined as follows:

TA = Log of Total Assets

EC = Equity Capital divided by Total Assets

AL = Liquid Assets divided by Total Assets

PS = Dollar amount of Preferred Stock divided by Total Assets

DIV = Dividends divided by Total Assets

NIM = Net Interest Income divided by Total Assets

NLCO = Net Loan Charge-Offs divided by Total Assets

GAP = 12 month Rate Sensitive Assets--12 month Rate Sensitive Liabilities

SWAP = Notional Amount of Interest-Rate Swaps divided by Total Assets

F&F = Notional Amount of Interest-Rate Futures and Forwards divided by Total Assets

OPTION = Notional Amount of Interest-Rate Options divided by Total Assets

SDUMMY (FDUMMY, ODUMMY, DDUMMY) = Dummy variables in which users are assigned a value of 1 and non-users a value of 0 for Swaps, Futures & Forwards, Options, and Derivatives respectively.

LIBOR = The 3-month London Interbank Offer Rate

S&P500 = The return on the S&P 500 stock index

ROA = Net Income divided by Total Assets

ROE = Net Income Divided by Total Equity

To answer our second question we break our sample into quartiles and examine the differences between how derivative use has evolved among the largest quartile compared to the smallest three. We use regression analysis to determine whether firm performance is better for users versus non-users of derivatives in each sub-sample. The regression equation we use for this analysis is as follows:

Performance Measure (ROA or ROE) = [a.sub.0] + [a.sub.1]ln(TA) + [a.sub.2]EC + [a.sub.3]AL + [a.sub.4]PS + [a.sub.5]DIV + [a.sub.6] NIM + [a.sub.7]GAP + [a.sub.8]NLCO + [a.sub.9]F&F + [a.sub.10]OPTION + [a.sub.11]SWAP + [a.sub.13]S&P500 + [a.sub.14]LIBOR

We use the log of total assets to capture the effect of firm size on derivative use. Equity capital is included to control for a firms ability to use derivatives as allowed by regulatory agencies. Asset liquidity is included to control for the bank's on-balance sheet assets and to test whether asset liquidity could represent an alternative to hedging as previous researchers have proposed (Nance, Smith & Smithson, 1993). Preferred stock is included to control for the possibility that it may be viewed as an alternative to hedging. The amount that a firm pays in dividends can indicate whether the firm has sufficient funds to enable them to be approved by creditors to use derivatives for hedging purposes. Previous researchers have shown a positive relationship between derivative use and dividend payouts indicating that if the firm has the cash to pay dividends then it has the ability to use derivatives. Net interest margin is expected to be positively related with banks attempting to protect their spread between interest income and interest expense through the use of derivatives. Net loan charge-offs is used as a proxy for credit risk and tests whether joint management of credit risk and interest rate risk occurs. Finally, we use twelve month GAP because it is the best measure available to proxy for interest-rate risk exposure. Twelve month GAP (GAP) is measured as the dollar gap between interest sensitive assets and liabilities in the twelve-month maturity range. F&F, Option, and Swap denote the ratio of each type of derivative's notional amount divided by total assets. These variables are used to test whether the level of derivative use enhances firm performance. The annual return on the S&P500 and LIBOR is included to control for changing market conditions.

EMPIRICAL RESULTS

Table 1 presents the summary statistics for the community banks in our sample, and reveals several differences between users and non-users of derivatives. In each year, the size of firms that use derivatives is much larger than the group of firms that are non-users. Net interest margin and net loan charge-offs are also larger in the user group in each year. To determine whether or not these differences are statistically significant we compare these variables for the users of derivatives versus nonusers using a simple t-test. We find a significant difference at the 5% level in the size, asset liquidity, and net-loan charge-offs of the users versus non-users. We can conclude from this simple analysis that the derivative users are larger banks with less liquid assets and more net-loan charge-offs than the non-users.

Additional summary statistics are shown in Table 2, which includes the number of banks and notional amounts of futures, forwards, options and swaps in each year. The percentage of derivative users decreased from 7.35% in 1995 to 4.15% in 2001 before increasing to 6.57% in 2003. At first it appears puzzling as to why derivative usage decreases so much from 1995 to 2001. We propose that the decrease is related to the anticipation of a change in the accounting standards. In order to make financial reporting of derivative securities more transparent, the Financial Accounting Standards Board enacted FAS 133 "Accounting for Derivative Instruments and Hedging Activities" in June 2000. FAS 133 remains one of the most controversial standards ever released by the FASB, mainly because of its complexity and extensive areas left for interpretation by the Finance and Accounting industry. Since the effective date, the FASB and accounting industry in general have attempted to simplify and clarify FAS 133. We propose that this change in the required reporting of derivatives has discouraged community banks from engaging in potentially profit stabilizing hedging activities.

Our analysis thus far ignores the possibility that our results could be driven by a small subset of our sample. Therefore, we break the sample into quartiles based on the banks' total assets. Chart A shows the number of users of derivatives by quartile. Chart B shows the percentage of users each quartile represents. These charts emphasize the fact that size matters in determining derivative use by banks. The firms in the largest quartile account for over half of all users, and derivative users in the top two size quartiles account for eighty percent of all derivative users. In 2002 and 2003 they account for approximately 90% of users. This could also indicate that FAS 133 has had the largest negative effect on the smallest community banks. Chart C emphasizes that the two smallest quartiles have very minimal derivative usage in terms of the notional value of the derivative contracts used.

The last chart we want to examine is Chart D, which shows derivative use by type each year based on the notional values of swaps, options, futures and forwards. It is clear that swaps are the dominate type of derivative used by these banks. This is not surprising since a large amount of simple hedging is done by swapping floating rate assets/liabilities to fixed and vice versa. The fact that the greatest use of derivatives is swaps will be relevant to our findings in regards to the effect of derivative use on firm performance.

[GRAPHIC A OMITTED]

[GRAPHIC B OMITTED]

[GRAPHIC C OMITTED]

[GRAPHIC D OMITTED]

Last, we use Tobit regression analysis to examine how the use of derivatives affects the performance of the community banks in our sample. Since very few of the community banks in our sample are publicly traded we cannot use stock returns as a measure of performance, so we evaluate firm performance by comparing return on assets (ROA) and return on equity (ROE) for each firm in our sample.

The results of the regression analysis are presented in Tables 3 and 4. Table 3 presents the firm performance regression for our entire sample over the entire period (1995-2003) and investigates whether the amount of derivative use scaled by Total Assets has an impact on performance. The results show that derivative use by community banks does impact performance as measured by ROA and ROE. Firms that use more swaps as a percentage of Total Assets to hedge tend to have higher ROA's and ROE's. This result is significant at the 1% level for ROA and at the 10% level for ROE. It is important to note that we do not believe these results indicate that community banks should enter as many swap contracts as possible to enhance performance. Rather, they should enter enough swaps as a percentage of Total Assets so as to have a meaningful impact on the interest-rate risk on the balance sheet.

Referring back to Chart D, recall that swaps are the largest type of derivative used by community banks and that the level of usage of swaps versus options, forwards or futures decreased the most during the 1995 to 2000 time period. This leads us to an interesting question. If bank performance is enhanced through an appropriate level of swap usage, then why over these six years did we see a decline? It seems to indicate that our earlier hypothesis that the implementation of FAS 133 has deterred banks from engaging in performance enhancing hedging is worth investigating. However, that topic is beyond the scope of this paper.

Table 4 presents the results for our second set of regressions. Here, we have replicated the tests presented in Table 3 except that we break the sample into two parts. The first sub-sample consists of the first three quartiles (the smallest banks) while the second sub-sample consists of the fourth quartile (the largest banks). As we expected, the results indicate that the largest banks do in fact drive the results presented in Table 3 and previously in the literature. In both the ROA and ROE regressions on the Quartile 1-3 sub-sample, none of the derivative variables are significant factors in determining firm performance. However, when the largest community banks are tested in isolation, we see that the level of swap usage is significant at the 1% level in determining ROA and at the 10% level in determining ROE. Given the clear evidence presented in Table 4 and graphically in Charts A-D, we can confidently conclude that the largest community banks are driving the results.

CONCLUSION

This paper examines the use of derivatives by U.S. commercial banks with total assets between $100 million and $1 billion. We limit our study to these "community banks" because we can be reasonably certain that derivative use will be limited to hedging purposes instead of dealer activity. The purpose of this paper is to provide researchers with an understanding of how derivative use has changed over the time period of 1995 to 2003. Instead of increasing as we initially hypothesized, derivative use by community banks decreased significantly from 1995 to 2000 before increasing slightly from 2000 to 2003. One reason we propose for this decrease in derivative use is an accounting standard change, FAS 133. Although it did not become effective until June 2000, it was initially scheduled to become effective in 1999, and banks were undoubtedly anticipating it for a few years prior. Although it is beyond the scope of this paper, the next research question to study is whether FAS 133 has led community banks to use less derivatives, and further has this harmed their performance.

Our performance regressions show that the amount of derivative use as a percentage of firm size does matter in predicting ROA and ROE. Firms that use enough swaps to have an effect on their balance sheet do experience a significant increase in firm performance. Our results support the view that although derivative use has decreased since 1995, some community banks still engage in the use of derivatives to hedge interest rate risk. In fact, those that have used derivatives to hedge their interest rate risk have experienced a statistically significant increase in performance. We further show that these results are driven by the largest community banks in our sample.

REFERENCES

Boukrami, L. (2003). The use of interest rate swaps by commercial banks. Working paper, Manchester Metropolitan University.

Carter, D., and J. Sinkey (1998). The use of interest rate derivatives by end-users: The case of large community banks. Journal of Financial Services Research, 14(1), 17-34.

Deyoung, R., W. Hunter & G. Udell (2004). The past, present, and probable future of community banks. Journal of Financial Services Research 25(1), 85-133.

Gorton, G. & R. Rosen (April 1995). Banks and derivatives. NBER Macroeconomics Annual, 299-339.

Nance, D.R., C.W. Smith, & C.W. Smithson (1993). On the determinants of corporate hedging. Journal of Finance, 48(1), 267-284.

Sinkey, J. & D. Carter (2000). Evidence on the financial characteristics of banks that do and do not use derivatives. The Quarterly Review of Economics and Finance, 40(4), 431-449.

Smith, C.W., & R. Stulz (1985) The determinants of firms' hedging policies. Journal of Financial and Quantitative Analysis, 20(4), 391-406.

Sara H. Robicheaux, Birmingham-Southern College

TABLE 1: SUMMARY STATISTICS OF ANNUAL DERIVATIVE USERS VERSES
NON-USERS.

 1995 1995 1996 1996 1997

Variable User Nonuser User Nonuser User

N 191 2600 148 2709 165
TA * 408858 227887 406135 231584 413879
EC 0.0925 0.0955 0.0961 0.0951 0.0949
AL * 0.1021 0.0902 0.0922 0.0849 0.0941
PS 0.0008 0.0002 0.0011 0.0003 0.0012
DIV 0.0071 0.0066 0.0070 0.0063 0.0081
NIM 0.0419 0.0415 0.0425 0.0413 0.0428
NLCO * 0.0035 0.0025 0.0043 0.0029 0.0042
SWAP 0.1325 0.0961 0.0494
F&F 0.0066 0.0027 0.0115
OPTION 0.0564 0.0542 0.0487

 1997 1998 1998 1999 1999

Variable Nonuser User Nonuser User Nonuser

N 2695 147 2764 131 2838
TA * 235590 406776 232610 411289 238781
EC 0.0969 0.0890 0.0962 0.0864 0.0936
AL * 0.0866 0.0944 0.0934 0.0670 0.0735
PS 0.0003 0.0002 0.0002 0.0001 0.0002
DIV 0.0064 0.0071 0.0066 0.0068 0.0069
NIM 0.0410 0.0422 0.0397 0.0417 0.0399
NLCO * 0.0025 0.0032 0.0026 0.0026 0.0024
SWAP 0.0524 0.0595
F&F 0.0097 0.0033
OPTION 0.0388 0.0346

 2000 2000 2001 2001 2002

Variable User Nonuser User Nonuser User

N 128 2888 130 3001 152
TA * 361440 243475 388328 248979 438948
EC 0.0951 0.0968 0.0972 0.0971 0.1003
AL * 0.0687 0.0765 0.0645 0.0858 0.0730
PS 0.0002 0.0002 0.0002 0.0002 0.0003
DIV 0.0068 0.0068 0.0049 0.0061 0.0073
NIM 0.0400 0.0392 0.0381 0.0374 0.0388
NLCO * 0.0032 0.0023 0.0043 0.0026 0.0051
SWAP 0.0398 0.0379 0.0394
F&F 0.0048 0.0153 0.0166
OPTION 0.0349 0.0267 0.0497

 2002 2003 2003 95-03 95-03

Variable Nonuser User Nonuser User Nonuser

N 3109 222 3155 1414 25759
TA * 251991 422674 252299 408392 240893
EC 0.0999 0.0964 0.0996 0.0943 0.0968
AL * 0.0848 0.0678 0.0752 0.0811 0.0832
PS 0.0002 0.0004 0.0002 0.0005 0.0002
DIV 0.0059 0.0063 0.0064 0.0069 0.0064
NIM 0.0380 0.0356 0.0367 0.0402 0.0393
NLCO * 0.0026 0.0044 0.0027 0.0039 0.0026
SWAP 0.0312 0.0609
F&F 0.0101 0.0090
OPTION 0.0144 0.0394

All numbers are reported in thousands. The variables we use in our
analysis are defined as follows. The log of total assets measures firm
size (TA), Equity Capital (EC), Asset Liquidity (AL), Preferred stock
(PS), Dividend usage (DIV), Net interest margin (NIM), Net loan
charge-offs (NLCO), Swap notional value divided by TA (SWAP), Forwards
and future notional divided by total assets (F&F), Option notional
divided by total assets (OPTION). The variables with * indicates a
significant difference between the user versus non-user banks in
regards to that variable at a 5% level of significance using a t-test.

TABLE 2: SUMMARY STATISTICS OF DERIVATIVE USA GE BY YEARS

 1995 1996 1997

Number 2791 2857 2860
Swaps 144 108 88
Futures/
 Forwards 15 10 41
Written
 Options 11 19 11
Purchased
 Options 95 53 64
Total
 User Banks 191 148 165
Total Nonuser
 Banks 2600 2709 2695
Percent
 of Users 7.35% 5.46% 6.12%

Notional Value (Thousands)

Swaps 9,896,888 6,510,937 3,216,026
Futures/
 Forwards 413,809 95,346 500,552
Options 3,464,680 2,474,917 3,066,249

Total 13,775,377 9,081,200 6,782,827

 1998 1999 2000

Number 2911 2969 3016
Swaps 84 75 65
Futures/
 Forwards 27 22 26
Written
 Options 8 8 9
Purchased
 Options 61 49 47
Total
 User Banks 147 131 128
Total Nonuser
 Banks 2764 2838 2888
Percent
 of Users 5.05% 4.41% 4.24%

Notional Value (Thousands)

Swaps 3,093,971 3,396,296 1,901,933
Futures/
 Forwards 538,134 187,185 156,592
Options 2,129,787 1,915,017 1,412,505

Total 5,761,892 5,498,498 3,471,030

 2001 2002 2003

Number 3131 3261 3377
Swaps 72 85 95
Futures/
 Forwards 30 35 78
Written
 Options 8 30 96
Purchased
 Options 46 35 26
Total
 User Banks 130 152 222
Total Nonuser
 Banks 3001 3109 3155
Percent
 of Users 4.15% 4.66% 6.57%

Notional Value (Thousands)

Swaps 2,087,951 2,856,130 3,303,288
Futures/
 Forwards 699,603 977,444 686,173
Options 1,253,720 2,619,738 976,090

Total 4,041,274 6,453,312 4,965,551

TABLE 3: FIRM PERFORMANCE--ENTIRE SAMPLE

 Return on Assets

Variable Coefficient Chi-Square

Intercept -0.0174 162.79 ***
TA 1.0459 96.42 ***
EC 0.0479 1024.62 ***
AL -0.0011 1.36
PS -0.0145 0.47
DIV 0.4835 9133.10 ***
NIM 0.2455 1797.57 ***
GAP 0.0003 0.2039
NLCO -0.3439 1580.22 ***
F&F 0.0030 0.31
SWAP 0.0258 139.20 ***
OPTION 0.0005 0.07
SP500 0.0007 5.89 **
LIBOR -0.0050 2.55

 Return on Equity

Variable Coefficient Chi-Square

Intercept -0.1389 19.11 ***
TA 13.2174 28.24 ***
EC -0.4298 151.34 ***
AL 0.0295 1.85
PS -0.1937 0.16
DIV 4.5357 1474.06 ***
NIM 3.0864 520.93 ***
GAP 0.0565 17.78 ***
NLCO -4.5649 510.65 ***
F&F 0.1046 0.70
SWAP 0.0889 3.02 *
OPTION -0.0215 0.22
SP500 0.0020 0.10
LIBOR -0.1212 2.76

The variables we use in our analysis are defined as follows. The log
of total assets measures firm size (TA), Equity Capital (EC), Asset
Liquidity (AL), Preferred stock (PS), Dividend usage (DIV), Net
interest margin (NIM), twelve month dollar Gap (GAP), Net loan
charge-offs (NLCO), Swap notional value divided by TA (SWAP),
Forwards and future notional divided by total assets (F&F), Option
notional divided by total assets (OPTION), SP500 is the S&P500 return,
LIBOR is the London Interbank Offer Rate. The variables with
*** indicate 1% level of significance, ** indicate 5% level of
significance, * indicates 10% level of significance.

TABLE 4: FIRM PERFORMANCE--QUARTILES

 Return on Assets

Quartile 1-3 Coefficient Chi-Square

Intercept -0.0200 62.91 ***
TA 1.3011 39.40 ***
EC 0.0280 285.94 ***
AL -0.0030 8.70 **
PS 0.0152 0.48
DIV 0.4857 6465.10 ***
NIM 0.2834 2095.69 ***
GAP 0.0025 15.52 ***
NLCO -0.2916 745.56 ***
F&F 0.0067 1.46
SWAP -0.0001 0.0002
OPTION -0.0070 1.63
SP500 0.0005 2.74 *
LIBOR -0.0086 6.58

Quartile 4

Intercept -0.0155 8.05 ***
TA 0.9385 5.15 **
EC 0.0905 831.30 ***
AL -0.0017 0.59
PS -0.0387 0.54
DIV 0.4580 2434.40 ***
NIM 0.1336 99.49 ***
GAP -0.0029 6.03 **
NLCO -0.3476 461.67 ***
F&F -0.0104 0.62
SWAP 0.0244 77.02 ***
OPTION 0.0042 3.01 *
SP500 0.0013 4.06 **
LIBOR 0.0059 0.71

 Return on Equity

Quartile 1-3 Coefficient Chi-Square

Intercept -0.1983 7.48 ***
TA 17.3617 8.47 ***
EC -0.4081 73.42 ***
AL 0.0138 0.24
PS 0.0644 0.01
DIV 5.6601 1060.05 ***
NIM 3.2815 339.23 ***
GAP 0.0665 13.14 ***
NLCO -6.3129 421.99 ***
F&F 0.0531 0.11
SWAP 0.0136 0.01
OPTION -0.0563 0.13
SP500 -0.0031 0.13
LIBOR -0.1686 3.07 *

Quartile 4

Intercept -0.0082 0.04
TA 5.5229 2.89 *
EC -0.4512 335.72 ***
AL 0.0869 26.38 ***
PS -1.2611 9.25 ***
DIV 2.7400 1414.92 ***
NIM 2.3482 499.22 ***
GAP 0.0267 8.30 ***
NLCO -1.9568 237.58 ***
F&F 0.3134 9.14 ***
SWAP 0.0706 10.48 ***
OPTION -0.0069 0.13
SP500 0.0141 8.06 ***
LIBOR -0.0208 0.14

The variables we use in our analysis are defined as follows. The log
of total assets measures firm size (TA), Equity Capital (EC), Asset
Liquidity (AL), Preferred stock (PS), Dividend usage (DIV), Net
interest margin (NIM), twelve month dollar Gap (GAP), Net loan
charge-offs (NLCO), Swap notional value divided by TA (SWAP), Forwards
and future notional divided by total assets (F&F), Option notional
divided by total assets (OPTION), SP500 is the S&P500 return, LIBOR is
the London Interbank Offer Rate. The variables with *** indicate 1%
level of significance, ** indicate 5% level of significance,
* indicates 10% level of significance.