How and Why Do Small Firms Manage Interest Rate Risk? Evidence from Commercial Loans pot

Federal Reserve Bank of New York Staff Reports How and Why Do Small Firms Manage Interest Rate Risk? Evidence from Commercial Loans James Vickery Staff Report no. 215 August 2005 Revised September 2006 This paper presents preliminary findings and is being distributed to economists and other interested readers solely to stimulate discussion and elicit comments. The views expressed in the paper are those of the author and are not necessarily reflective of views at the Federal Reserve Bank of New York or the Federal Reserve System. Any errors or omissions are the responsibility of the author. How and Why Do Small Firms Manage Interest Rate Risk? Evidence from Commercial Loans James Vickery Federal Reserve Bank of New York Staff Reports, no. 215 August 2005; revised September 2006 JEL classification: G21, G30 Although small firms are particularly sensitive to interest rates and other external shocks, empirical work on corporate risk management has focused instead on large public companies. This paper studies fixed-rate and adjustable-rate loans to see how small firms manage their exposure to interest rate risk. Credit-constrained firms are found to match significantly more often with fixed-rate loans, consistent with prior research showing that the supply of internal and external finance shrinks during periods of rising interest rates. Banks originate a higher share of adjustable-rate loans than other lender types, ameliorating maturity mismatch and exposure to the lending channel of monetary policy. Time-series patterns in the share of fixed-rate commercial loans are consistent with recent evidence on “debt market timing.” Key words: fixed-rate loan, adjustable-rate loan, corporate risk management, interest rate risk Vickery: Federal Reserve Bank of New York (e-mail: james.vickery@ ny.frb.org). This paper is a revised version of part of the author’s doctoral thesis at the Massachusetts Institute of Technology. The author thanks his thesis advisers, Ricardo Caballero, David Scharfstein, and Robert Townsend, for their generous guidance, as well as Allen Berger, Olivier Blanchard, Bengt Holmstrom, Don Morgan, Jeremy Stein, David Smith, John Wolken, and seminar participants at the Massachusetts Institute of Technology, the London Business School, the University of Notre Dame Mendoza College of Business, Columbia Business School, the University of Chicago Graduate School of Business, Yale School of Management, Harvard Business School, the Federal Reserve Bank of New York, the Board of Governors of the Federal Reserve System, New York University Stern School of Business, and the Federal Reserve Bank of Richmond for their insightful comments. The views expressed in this paper are those of the author and do not necessarily reflect the position of the Federal Reserve Bank of New York or the Federal Reserve System. 1 1. Introduction Empirical research on corporate risk management has generally focused on large public companies, most often studying firms’ use of financial derivatives. 1 This paper instead examines fixed-rate and adjustable-rate commercial loan contracts to study how small firms adjust their exposure to interest rate risk. Small and medium-sized firms are important to the US economy; firms with less than 500 employees generate half of non-farm private GDP 2 . Small firms are often financially constrained, considered a key theoretical rationale why firms engage in risk management (eg. Froot, Scharfstein and Stein, 1993). Moreover, work on the ‘credit channel’ of monetary policy shows directly that small firms are sensitive to interest rate shocks (eg. Gertler and Gilchrist, 1994; Ehrmann 2000). Although small and medium sized firms make little use of derivatives, they do borrow extensively from financial institutions. In some cases the interest rate on these loans is fixed, while in other cases it adjusts with market interest rates. I study this variation in ‘fixed-versus-adjustable’ outcomes as a window into how small firms adjust their exposure to interest rate risk. I firstly examine the relationship between ‘fixed-versus-adjustable’ outcomes and firm financial constraints. Theoretically, Froot, Scharfstein and Stein (1993, hereafter FSS) shows that optimal risk management policy should aim to generate cash in states of nature where an additional dollar of internal funds is most valuable. Empirically, research on the ‘credit channel’ of monetary policy finds that the availability of finance to bank dependent firms becomes scarcer relative to 1 Cross-industry studies of the determinants of firms’ use of derivatives include Covitz and Sharpe (2005), Purnandanam (2004), Lin and Smith (2003), Rogers (2002), Graham and Rogers (2002), Géczy, Minton and Schrand (1997), Mian (1996) and Fenn, Post and Sharpe (1996). Some studies focus on particular types of derivatives eg. Géczy et al focus on foreign currency derivatives, while Covitz and Sharpe study interest rate contracts. Allayanis and Weston (1998) study the relationship between derivatives use and firm value. Guay (1999) examines how derivatives hedging affects firm risk. Guay and Kothari (2003) examine the quantitative relevance of firms’ derivatives holdings. Other papers take an industry-specific approach; Faulkender (2005) studies chemicals firms, Haushalter (2001) focuses on oil and gas, and Tufano (1996) studies gold mining firms. The literature is also broadening to consider other dimensions of risk management. For example, Bartram, Brown and Minton (2006) and Pantzalis, Simkins and Laux (2001) present evidence that firms use operational hedging (eg. matching foreign sales to foreign production) to manage exchange rate risk. Petersen and Thiagarajan (2000) study two gold mining firms who achieve a similar reduction in exposure to gold price risk, one using derivatives, the other using a combination of operating, financial and accounting decisions. 2 Source: SBA. See http://www.sba.gov/advo/research/rs211tot.pdf. 2 investment opportunities during periods of rising interest rates, causing lower investment and output amongst credit-constrained firms (section 2 reviews this literature in detail). Correspondingly, I test the hypothesis that credit constrained firms match with fixed rate debt, thereby maximizing net cashflows during periods of rising interest rates when the shadow value of internal funds is high. 3 A related implication of FSS is that risk management outcomes should reflect variation across firms in the correlation between interest rates and pre-interest firm cash flows. In sectors where industry output or cashflows covarys positively with interest rates, firms have a partial or complete ‘natural hedge’ against interest rate risk, and thus fixed rate debt is less likely to be optimal. I test the hypothesis that the share of adjustable rate loans is higher in such industries, using an estimated index of industry ‘interest rate procyclicality’. Although plausible, there are several reasons why these two FSS ‘hedging’ hypotheses might fail to hold empirically. One alternative hypothesis is that ‘fixed-versus-adjustable’ outcomes are set by the firm’s banks (eg. perhaps the firm’s relationship lender only originates fixed-rate loans or only adjustable-rate loans, so the firm does not have a choice). Another possibility is that small firms prefer to amplify volatility in the shadow value of internal funds; Adam, Dasgupta and Titman (2006) presents a model where such behavior may be optimal in an imperfectly competitive industry setting. A third possibility is that small firms are financially unsophisticated or the ‘fixed-versus- adjustable’ margin is unimportant, so that there are no systematic correlations in the data. Using data from the Federal Reserve Board’s Survey of Small Business Finance (SBF) I do in fact find evidence consistent with the two FSS ‘hedging’ hypotheses outlined above. First, as predicted, matching with a fixed rate loan is positively correlated with several different proxies for financial constraints. Fixed rate debt is most popular amongst smaller firms, younger firms, firms switching from their primary lender, and firms with low cashflows (measured by current profits) or 3 In the FSS framework, a non-credit-constrained firm would be simply indifferent between fixed and adjustable rate loans, in line with the Modigliani-Miller theorem. However, I argue in the body of the paper that when lenders are also exposed to interest rate risk (consistent with a large body of empirical evidence) unconstrained firms may strictly prefer adjustable-rate debt. 3 high investment opportunities (measured by sales growth). These results are economically as well as statistically significant; for example young, small firms in the SBF are about twice as likely to match with fixed rate debt as old, large firms (69 per cent compared to 38 per cent). Second, fixed rate debt is less prevalent in 2-digit SIC sectors where industry output comoves most positively with interest rates, and thus where firms have a partial natural hedge against interest rate risk. Next, I study how lender characteristics influence ‘fixed-versus-adjustable’ outcomes. Several theoretical papers on loan contract design and bank risk management suggest that the share of interest rate risk in a loan borne by the borrower should depend in part on the lender’s interest rate risk profile (Arvan and Brueckner, 1986; Edelstein and Urosevic, 2003; Froot and Stein, 1998). These models predict that lenders who are exposed ex-ante to rising interest rates will optimally originate a smaller share of fixed-rate loans, since the present value of such loans declines by comparison to adjustable-rate loans when interest rates rise. I test this prediction by comparing bank loans to loans from non-bank institutions. Banks are exposed to rising interest rates in two ways that are specifically tied to their reliance on deposit finance. First, banks are affected by the ‘lending channel’ of monetary policy (Stein, 1998, Kashyap and Stein, 2000, Ashcraft, 2004), in which tight monetary policy reduces the insured deposit base, raising banks’ cost of funds. Second, banks are subject to maturity mismatch, where demand deposits and short-term time deposits fund long-duration assets such as mortgages. Correspondingly, I test the hypothesis that bank loans are more likely to involve an adjustable interest rate than loans from other lender types. This ‘lender risk management’ hypothesis receives strong support in the data; I find that a loan from a commercial or savings bank is 14 percentage points more likely to involve an adjustable interest rate compared to a loan from a non- bank financial institution. Since many small bank-dependent enterprises are closely held and owner-managed, it also seems plausible that owner characteristics play a significant role in ‘fixed-versus-adjustable’ 4 outcomes. Somewhat surprisingly, I find that variables like the owner’s age and the concentration of ownership are nearly uncorrelated with the loan type chosen. I do find some evidence that adjustable rate loans are more commom amongst firms with wealthier owners, consistent with the view that risk aversion is declining in wealth. The last part of the paper studies time-series patterns in the aggregate share of fixed-rate loans. Using data from the Survey of Terms of Business Lending, I construct and study a 28-year quarterly time-series of the fixed rate share for business loans originated by commercial banks. I find that high real interest rates and a steep yield curve are correlated with a lower proportion of fixed rate loans, consistent with previous work on ‘debt market timing’ by Faulkender (2005) and Baker, Greenwood and Wurgler (2003). To my knowledge, this paper is the first to show these results also extend to small, bank dependent firms. Implications of these findings for theoretical explanations of ‘market timing’ patterns are discussed. The rest of this paper proceeds as follows. Section 2 reviews existing literature on the sensitivity of small firms to interest rate shocks. Section 3 describes the Survey of Small Business Finance, and discusses the measures of financial constraints I use. Section 4 presents cross-sectional empirical evidence from the SBF. Section 5 presents time-series evidence on the share of fixed rate commercial loans. Section 6 presents cross-sectional evidence from the STBL. Section 7 concludes. 2. Small firms and interest rate shocks Research on the ‘credit channel’ of monetary policy argues that higher interest rates lead to a decline in the availability of internal and external finance relative to investment opportunities, resulting in lower investment and output amongst credit-constrained firms. This channel is considered to be most important for small, informationally opaque, bank dependent firms, who are most likely to be constrained in their access to finance. Consistent with this view, Gertler and Gilchrist (1994) show that small US manufacturing firms are disproportionately affected during periods of rising interest rates. Small firms reduce external borrowing, shed inventories and experience sharp falls in sales 5 growth. Larger firms maintain debt levels, increase inventories, and experience a substantially smaller decline in sales growth. Ehrmann (2000) finds similar evidence using data on German firms. The broad credit channel can be further decomposed into ‘balance sheet’ and ‘lending’ effects. The ‘balance sheet’ channel is that higher interest rates weaken firm balance sheets, partially by reducing expected future profits, and partially because small firms have long-lived physical assets but mainly short term or adjustable rate liabilities (bank loans, credit lines etc.). This maturity mismatch implies that net current cashflows decline when interest rates increase, and also that the present value of assets declines relative to the present value of liabilities. The latter makes the firm less creditworthy, reducing its ability to raise external finance. Consistent with the balance sheet channel, Bernanke and Gertler (1995) show that firms’ balance sheet strength, proxied by interest coverage, declines during periods of high interest rates. Greenwood (2003) finds firm investment is most sensitive to interest rates when maturity mismatch is high, and that this relationship is most pronounced for financially constrained firms. Ashcraft and Campello (2006) find commercial lending is more sensitive to monetary policy in geographic regions where firm balance sheets are weak. The ‘lending’ channel is that contractionary monetary policy reduces the ability of banks to lend by shrinking the supply of bank deposits. Stein (1998) presents a formal model of the lending channel. Kashyap, Stein and Wilcox (1993) show the bank lending as a share of total debt finance falls during periods of contractionary monetary policy. Kashyap and Stein (1995, 2000) present further empirical evidence on the lending channel for the US. For the purposes of this paper, the key implication of both the ‘lending’ and ‘balance sheet’ channels is that for the average small firm, the supply of internal finance plus external finance declines relative to investment opportunities during periods of rising interest rates. This fall in credit availability will have no real effects if credit constraints are not binding. For firms that are credit constrained, such a shrinkage in the availability of finance induced by rising interest rates will reduce 6 investment and output and raise the shadow value of internal funds. Correspondingly, I test the hypothesis that credit constrained firms match with fixed rate loans rather than adjustable rate loans, maximizing cash flows during periods of rising interest rates, when internal funds are most valuable. What about financially unconstrained firms? In FSS, such firms would be indifferent between the two loan types, in line with the Modigliani-Miller theorem. However, a substantial amount of evidence suggests lenders too are exposed to interest rate risk, via the ‘bank lending’ channel as well as maturity mismatch (Sierra and Yaeger, 2004). In such an environment, the optimal contract will generally involve a financially unconstrained firm bearing the interest rate risk of the loan (Arvan and Brueckner, 1986, Edelstein and Urosevic, 2003, Froot and Stein, 1998). In Section 4.2 I present direct evidence that lender types exposed to rising interest rates originate a higher fraction of adjustable-rate loans, consistent with this view. Alternatively, unconstrained firms may match with adjustable rate debt to signal firm quality, in line with the model of Guedes and Thompson (1995). Even without these effects, it will still hold that unconstrained firms have no explicit incentive to protect cashflows against rising interest rates, unlike constrained firms. 3. Data: Survey of Small Business Finance (SBF) The SBF is a cross-sectional survey conducted approximately every five years by the Federal Reserve Board, containing detailed microeconomic information on firm characteristics and financing behavior for a representative sample of US small and medium sized enterprises, defined as firms with less than 500 employees at the end of the reference year 4 . The SBF provides particularly detailed information on the firm’s most recent loan, including the size of the loan, interest rate and fees paid, category of loan (eg. line of credit, business mortgage etc.), maturity, and collateral posted against the loan. Most importantly for this paper, the SBF also records whether the most recent loan was issued at a fixed or variable interest rate. 4 For the 1987 survey, the relevant population is US firms with less than 500 full time equivalent employees. For the 1993 and 1998 survey the population is firms with less than 500 full-time plus part-time employees. Other things equal, this implies firms in 1993 and 1998 are smaller on average than those in the 1987 survey. 7 I pool data from the 1987, 1993 and 1998 SBF surveys, for a total of 11422 firm observations. Of these, 4000 firms had received a loan within three years of the end of the survey reference year. (I use three years as a cutoff because the 1993 and 1998 surveys do not collect information on the most recent loan if it was originated more than three years ago.) I then drop firms where data is missing for one or more key variables: total assets, firm age, profitability, total debt, sales growth, years with primary lender, or the amount, maturity or ‘fixed or adjustable’ status of the most recent loan. This yields a final sample of 3248 loans matched with firm characteristics. (N.B. the SBF is not a panel dataset, so each of these observations relates to a different firm.) [INSERT TABLE 1 HERE] Table 1 presents descriptive statistics for this final sample of 3248 observations. Since the survey oversamples large firms and minority-owned firms, I present weighted averages based on the SBF sampling weights, as well as unweighted statistics for comparison. 70 per cent of unweighted observations are S- or C-corporations. Average assets are $939 000 ($2.7 million on an unweighted basis). Comparing the last two columns of the Table, firms in the final sample are of similar age although substantially larger than the overall SBF sample. There are relatively fewer observations in the final sample from the 1998 survey, partially due to a change in survey design; in 1998 the survey does not consider renewals of existing credit lines to be ‘new loans’. 38 per cent of loans are lines of credit (43 per cent on an unweighted basis). The distinction between credit lines and other loan types is important for the ‘fixed or adjustable’ dimension of the loan contract. A fixed rate credit line in fact create a potential arbitrage opportunity, since any change in market rates will affect the wedge between market rates and the rate on the commitment. For example if interest rates rise sharply, the firm could potentially aggressively draw down the line of credit, investing the proceeds at the higher market rate. For this reason, only 29 per cent of lines of credit in the SBF are fixed rate, compared to 70 per cent for other loan types. Moreover, most fixed rate credit lines are short term, 70 per cent have a maturity of 1 year or less. For these short term 8 commitments, interest rates are unlikely to shift enough before the credit line is renegotiated for the arbitrage opportunity described above to be profitable after transaction costs. A second point of difference is that the interest rate risk of an adjustable-rate credit line is ‘contingent’, since the firm only faces risk to the extent that it actually draws down the line in the future. Given these differences, but also taking into consideration the moderate sample size, I always present two sets of empirical estimates, one based on the full sample, the other on a subsample excluding credit lines. Loans in the sample have an average maturity of 4 years, and the average weighted loan size is $324 thousand (around one-third of average firm assets). Most importantly, there is substantial variation in firms’ choices between fixed and adjustable rate loans. 52 per cent of loan observations in the sample were drawn at a fixed rate (59 per cent on a weighted basis), the rest at an adjustable rate. 92 per cent of variable rate loans are indexed to a commercial prime lending rate. [INSERT TABLE 2 HERE] Table 2 breaks down the fixed rate share by type and source of loan. Importantly since loan type dummies are included in most regressions, there is a significant share of both fixed- and adjustable-rate contracts within each loan type. At the extremes, credit lines have the lowest fixed rate share (29 per cent), while capital leases and vehicle mortgages are most likely to be fixed (89 per cent and 88 per cent respectively). 5 Bank loans are less likely to involve a fixed rate (46 per cent compared to 78 per cent for non-bank loans). 3.1 Measuring financial constraints The SBF contains several potential measures of financial constraints. Below I discuss the measures I use, and briefly review the evidence associated with each of them. 5 Beyond the earlier discussion of credit lines, this paper does not provide a full explanation for why the fixed rate share varies across loan types. Differences in securitization rates provide a potential explanation, however. For example, a vehicle mortgage, backed by a standardized, easy-to-value asset, may be easier to securitize than a business mortgage secured by assets that are difficult for outsiders to value and monitor. (There is an active secondary market for auto loans in the US, consistent with this argument.) This may in turn explain the high share of fixed-rate vehicle mortgages, analogous to the argument that securitization underpins the popularity of the US fixed rate household mortgage (Green and Wachter, 2005). [...]... public firms? Chava and Purnandanam (2006) study the floating-to-fixed ratio of the debt of around 1800 public companies They find firms close to financial distress use a higher share of fixed rate debt, consistent with the results from this paper that credit-constrained firms match with fixed rate loans A notable point of difference, 22 however, is that Chava and Purnandanam find managerial and corporate... matching to a fixed interest rate, even after controlling for bank fixed effects This result confirms that individual banks do offer both types of loans, and also demonstrates that small firms match with fixed rate loans even just by comparison to larger firms who borrow from the same bank A final set of robustness checks are presented in Table 6 The first of these considers the hypothesis that the lender... Eisfeldt and Rampini (2004) show that small firms invest more often in used capital, which they argue is due to credit constraints Evans (1987) finds that small firms have more volatile growth rates Finally, the ‘credit channel’ evidence cited earlier suggests that smaller firms are more sensitive to interest rate shocks (Gertler and Gilchrist 1994, Ehrmann 2000) On the theory side, Albercurque and Hopenhayn... creditworthiness (English and Nelson, 1998) Thus, reverse causality may be an issue; if banks view adjustable rate loans as riskier, they will apply less favorable risk ratings to such loans, leading to a positive correlation between adjustable rates and risk ex post, even if loans and firms are randomly assigned ex-ante 31 7 Conclusions This paper finds evidence that small, bank-dependent firms use loan contracts... hedging and other non-derivatives decisions for firm risk management outcomes (eg Bartram, Brown and Minton, 2006; Chava and Purnandanam, 2006; Petersen and Thiagarajan, 2000) 5 Time-series Patterns in Fixed and Adjustable Rate Lending Recent work by Baker, Greenwood and Wurgler (2003) and Faulkender (2005) shows that the interest rate exposure of firms new debt fundings fluctuates over time in response... size Small firms are generally thought to face more severe financial constraints than large firms, due to scale economies in monitoring and information aquisition Within the class of bank dependent firms, the focus of this paper, Petersen and Rajan (1995) show that smaller firms pay higher interest rates and take lesser advantage of attractive early-payment discounts on trade credit, signs that such firms. .. or cashflows covary positively with interest rates, higher internal cashflows will at least partially offset the effects of interest rates on the supply of credit In these industries, firms have a ‘natural hedge’ against rising interest rates, and fixed rate debt is less likely to be optimal In this section I test the hypothesis that the share of adjustable rate loans is higher in such industries This... section, I show that these patterns also extend to small firms, by analyzing a long (26-year) time-series of the share of fixed -rate business loans Regarding the source of capital, Faulkender and Chava and Purnandanam both note that bank loans are significantly more likely to involve an adjustable interest rate than public debt fundings.13 Neither paper suggests an explanation for this stylized fact, however... loan, I visit the small business websites of 10 major US commercial banks, and study online documentation for two types of loans, unsecured term loans and commercial mortgages In each case I record whether the bank offers firms a choice between a fixed and adjustable loan contract I find that firms are indeed offered this choice in 12 of 15 cases where this information could be determined from the website... term loans from commercial banks This is significant, because Baker et al and Faulkender’s samples include a mix of different types of debt, and, as shown earlier, the source of finance is highly correlated with the debt contract’s final interest rate exposure Secondly, the results show that ‘debt market timing’ patterns extend to small, privately-held firms This is to my knowledge a new result, and . Reserve Bank of New York Staff Reports How and Why Do Small Firms Manage Interest Rate Risk? Evidence from Commercial Loans James Vickery Staff Report no are the responsibility of the author. How and Why Do Small Firms Manage Interest Rate Risk? Evidence from Commercial Loans James Vickery Federal Reserve