This section provides an overview of the literature and develops the hypotheses associated with RQ1, are loan contract terms (i.e., price, collateral, maturity, covenants, and size) simultaneously determined? Section 2.2.1 outlines studies on the relations between each pair of loan contract terms. Section 2.2.2 then considers the literature on the jointness of these terms to build the appropriate hypotheses.
2.2.1. Relations between Loan Contract Terms
The Arrow and Debreu (1954) market is perfect; it has no information asymmetries or other frictions, so financial intermediations are unnecessary and therefore do not exist (Santos, 2001). In an imperfect market, these institutions’ role
is to produce information and reduce information asymmetries (Leland and Pyle, 1977; Campbell and Kracaw, 1980; Diamond, 1984; Best and Zhang, 1993). Given this role, loan contracts are very important: They convey borrower information to the market (Best and Zhang, 1993), help banks monitor borrowers (Diamond, 1984), and enable borrowers to signal information about themselves through choosing specific loan terms (Bester, 1985), as well as trading off between these terms to optimize their benefits (Melnik and Plaut, 1986).
Prior studies on loan term determinants note the relations between pairs of loan terms (e.g., Merton, 1974; Myers, 1977; Smith and Warner, 1979b; Milde and Riley, 1988). Although they offer mixed results on how borrowers and lenders negotiate loan contract terms (Pham and Lensink, 2008), they generally refer to the paradigms of ‘trade-off’ and ‘sorting by observed risk’ (Dennis et al., 2000;
Gottesman, 2006). Trade-off involves a non-favourable loan term being accepted by a borrower in return for a better one from the lender. In contrast, sorting by observed risk involves the two loan terms being made stricter due to high borrower risk (Dennis et al., 2000).
Although these two views do not explain how all the key loan terms are negotiated, they do suggest some possible interrelations within a loan contract. The following outlines the relations between each pair of loan terms (as summarized in Table 2.1), including those between loan price and the non-price terms of collateral, maturity, covenants, and loan size; then those between the non-price terms of collateral with maturity, covenants, and loan size; those between maturity and covenants and between maturity and loan size; and finally those between covenants and loan size.
Table 2.1: Relations between the pairs of loan terms
This table summarizes the relations between each pair of loan terms across loan price, collateral, maturity, covenants, and size.
Price Collateral Maturity Covenants
Trade-Off Paradigm
Sorting by Observed Risk Paradigm
Trade-Off Paradigm
Sorting by Observed Risk Paradigm
Trade-Off Paradigm
Sorting by Observed Risk Paradigm
Trade-Off Paradigm
Sorting by Observed Risk Paradigm
Collateral Negative Positive
Maturity Positive Negative Positive Negative
Covenants Negative Positive Negative Positive Positive Negative
Size Positive Negative Positive Negative Negative Positive Positive Negative Source: Summarized by the author.
The loan price and collateral trade-off paradigm suggests that these terms are negatively related (Bester, 1985; Chan and Kanatas, 1985; Chan, Greenbaum, and Thakor, 1986; Besanko and Thakor, 1987b, 1987a). This negative relation between the loan price (interest rate) and collateral may be the result of a borrower signalling their creditworthiness (Bester, 1985). High-quality borrowers with low probabilities of default tend to pledge more collateral in return for lower loan prices. High-risk borrowers, however, may prefer loan contracts with lower collateral and thus accept higher prices. In contrast, from the perspective of sorting by observed risk, ex ante high-risk borrowers are more likely to pledge collateral but still pay high prices because their collateral may not fully offset their risk (Pozzolo, 2002). Prior studies find mixed results. While many support trade-offs (Berger and Udell, 1995; Strahan, 1999; Dennis et al., 2000; Pozzolo, 2002), others confirm sorting by observed risk (e.g., Berger and Udell, 1990; John, Lynch, and Puri, 2003).
With regard to loan price and maturity, trade-off presumes a positive relation between the two (Goss and Roberts, 2011). To avoid borrower agency problems such as asset substitution and underinvestment, lenders tend to make shorter loans (Myers, 1977; Barnea, Haugen, and Senbet, 1980). Because short-term loans present liquidity problems when they become due, borrowers prefer longer-term loans (Gottesman, 2006). Lenders will therefore only lend long term when higher prices offset the risk of the longer maturities (Pham and Lensink, 2008). A low-risk borrower who can afford to roll over short-term debt may nevertheless seek a shorter loan to signal its high quality and in return receive lower prices (Flannery, 1986; Kale and Noe, 1990). In contrast, sorting by observed risk suggests a negative relationship between loan price and maturity (Booth and Booth, 2006; Goss and Roberts, 2011). It argues
that lenders offer short-term loans to riskier borrowers but longer-term loans to less risky ones (Pham and Lensink, 2008). So, riskier borrowers have to borrow short term (Diamond, 1991a). Since less risky borrowers are charged less, loan price and maturity are expected to have a negative relation (Gottesman, 2006). Prior findings are ambiguous: Some studies (e.g., Coleman, Esho, and Sharpe, 2002; Bharath et al., 2011) support the trade-off view, but others support sorting by observed risk (e.g., Strahan, 1999; Dennis et al., 2000; Gottesman and Roberts, 2004; Wittenberg- Moerman, 2009).
Loan prices and covenants are negatively related under the trade-off view.
Covenants could convey borrower information such as borrowers with low risk of shifting opportunities are willing to accept loans with more controlling right covenants to negotiate lower prices (Demiroglu and James, 2010a). Since covenants are costly for borrowers but reduce agency problems for lenders, borrowers can receive lower prices by accepting more covenant restrictions (Smith and Warner, 1979b). In contrast, sorting by observed risk predicts a positive relation between loan prices and covenants. Since agency cost problems relate to poor past borrower performance, both higher prices and covenants may be used (Mattes, Steffen, and Wahrenburg, 2012). Prior findings support both theories, with some studies (e.g., Iskandar-Datta and Emery, 1994; Demiroglu and James, 2010a) supporting the trade- off view and others (e.g., Brav et al., 2009; Mattes et al., 2012) sorting by observed risk.
The relation between loan price and loan size is unclear (Angbazo, Mei, and Saunders, 1998). The trade-off view claims they should be positively related, since
the larger the loan, the greater the lender’s risk exposure (Champagne and Kryzanowski, 2009). Lenders can therefore charge more for larger loans in compensation (Moore and Craigwell, 2003). In addition, borrowers can signal their high quality by accepting smaller loans in return for lower prices (Milde and Riley, 1988). Conversely, sorting by observed risk proposes a negative relation between loan price and loan size. Since large loans are usually granted to larger and well- organized firms, such borrowers should have less risk and information asymmetries (Strahan, 1999). Small loans are granted to smaller firms, which are generally much riskier, with more information asymmetry problems (Fungáþová, Godlewski, and Weill, 2009). Lenders therefore require higher prices on smaller loans to recover their risk and managerial costs. Similarly, the lower economies of scale on small loans can also incur higher prices (Fungáþová et al., 2009). Finally, since smaller firms have fewer alternative sources of financing, they may simply be forced to accept these higher prices (Grunert and Norden, 2012). Most studies find a negative relation between loan price and loan size (see Berger and Udell, 1990; Booth, 1992;
Cressy and Toivanen, 2001; Beatty, Ramesh, and Weber, 2002; Ivashina, 2009;
Wittenberg-Moerman, 2009), but a few shows a positive relation (for example, Melnik and Plaut, 1986).
The relation between collateral and maturity is ambiguous. The trade-off view predicts it should be positive (Dennis et al., 2000) for two reasons (Leeth and Scott, 1989): First, for lenders, short-term loans are better than long-term ones because they reduce borrower asset substitution. Long-term loans are therefore more likely to require collateral. In addition, the longer the loan maturity, the riskier is the loan. Even with a low-risk project, a long-term loan gives the borrower opportunities
to transform the project into a suboptimal or even riskier project (Jensen and Meckling, 1976). Long-term loans are therefore more likely to be secured. In contrast, sorting by observed risk suggests that collateral and maturity may be negatively related since both shorter maturity and collateral could be used to address borrower risk (Gottesman, 2006). Moreover, collateral value can decrease over time and thus become useless for longer loans and short term loans are therefore more likely to require collateral (Stulz and Johnson, 1985). Empirical studies are inconclusive, with some positive findings (Harhoff and Kửrting, 1998; Degryse and van Cayseele, 2000; Dennis et al., 2000; Voordeckers and Steijvers, 2006) and some negative (Scott and Smith, 1986; Boot et al., 1991).
The relation between collateral and covenants is unclear. Covenants should motivate lenders to monitor borrowers (Rajan and Winton, 1995) and could substitute for collateral (Steijvers and Voordeckers, 2009). In contrast, covenants cannot guarantee good monitoring (Rajan and Winton, 1995) and thus perhaps complement collateral (Strahan, 1999). Empirical studies, however, show an inverse relation (Schwartz, 1989; Citron, 1992; Citron, Robbie, and Wright, 1997; Niskanen and Niskanen, 2004).
Collateral and loan size may have a positive or negative relation. The trade- off view argues it should be positive. Good borrowers signal their high quality by pledging collateral to borrow larger amounts (Schwartz, 1989). In addition, a larger loan size can affect the borrower’s leverage (Leeth and Scott, 1989; Jaffee and Stiglitz, 1990). Since this increases the possibility of default (Steijvers and Voordeckers, 2009), collateral is required (Degryse and van Cayseele, 2000).
Moreover, since formalized collateral contracts involve certain costs, they may prove too expensive for small loans. Therefore collateral is used more for larger loans (Menkhoff, Neuberger, and Rungruxsirivorn, 2012). The negative relation between these loan terms, however, could be explained by the fact that larger loans are often made to large firms with lower risk exposure (Boot et al., 1991). Empirical studies are also inconclusive, with some indicating a positive relation between collateral and loan size (Leeth and Scott, 1989) and some a negative one (Scott and Smith, 1986).
The maturity and covenants relation is also unclear. Since agency costs increase with debt maturity, longer term loans are likely to have more covenants (Kare, 1996), since they help reduce asymmetric information problems between lenders and borrowers (Wittenberg-Moerman, 2009). Covenants function as an early warning device such that lenders can demand borrowers that breach the covenants to repay their loans immediately (Dichev and Skinner, 2002; Bradley and Roberts, 2004). Hence, long-term loans are usually associated with covenants. This relation, nevertheless, may also be negative, since collateral can required if a shorter maturity is not enough to minimize lender risk (Strahan, 1999). Some empirical findings, nevertheless, show a positive relation between maturity and covenants (Myers, 1977;
Berger and Udell, 2005; Billett, King, and Mauer, 2007).
The relation between maturity and loan size is also mixed. Some studies argue that larger loans are safer than smaller ones and therefore can have longer maturities (Boot et al., 1991; Lee, 2004). Others, however, claim that maturity and loan size are complementary in addressing information asymmetries and credit risk
problems and should therefore be positively related (Strahan, 1999; Cressy and Toivanen, 2001).
The relation between covenants and loan size has not been confirmed. Since large loans are usually made to large, well-established firms, covenants may be less common (Apitado and Millington, 1992), but this notion is inconsistent with the findings of Citron et al. (1997).
In summary, prior prediction and empirical evidence on the relations between each loan term pair are mixed, as shown in Table 2.1. While the literature does show that these terms should be in a loan contract, it does not show how they are negotiated or decided. Section 2.2.2 discusses these issues.
2.2.2. Jointness of Loan Contract Terms (Hypotheses for RQ1)
The jointness of loan terms is mentioned in many credit rationing studies (e.g., Harris, 1974; Azzi and Cox, 1976; Arzac, Schwartz, and Whitcomb, 1981;
Bester, 1985). The equilibrium of the loan market, for instance, is achieved due to loan price and non-price terms being used simultaneously (Harris, 1974). In addition, since loan prices and collateral are jointly decided, good borrowers signal themselves by offering collateral to obtain lower interest rates (Bester, 1985). These studies, however, do not explain how lenders and borrowers negotiate loan terms or establish loan contracts. Such negotiations were first explained by the Melnik and Plaut (1986) argument that a contract’s loan terms are in a ‘bundle’ that cannot be negotiated separately. Borrowers, furthermore, can obtain better terms in exchange for less favourable ones.
Even so, some bank lending studies emphasize the determinants of loan terms using single equations (Leeth and Scott, 1989; Petersen and Rajan, 1994;
Guedes and Opler, 1996). These studies may therefore suffer from simultaneity problems if the other loan terms are jointly determined (Dennis et al., 2000). Others studies avoid this problem by using a reduced-form regression13 (for example, Berger and Udell, 1995), but this approach does not allow for any trade-offs between loan terms (Dennis et al., 2000).
Addressing the problem of simultaneity with Melnik and Plaut (1986) idea, Dennis et al. (2000) perform their econometric estimates through a two-stage regression. Loan prices (loan spreads and commitment fees) and two non-price terms (maturity and collateral) are first examined. The simultaneous equation results are then compared with those from the single-equation regressions. The authors’
findings suggest that the results will be biased if simultaneous equation estimation techniques are not applied. Their simultaneous equation model has since been adopted by many subsequent studies (for instance, Brick and Palia, 2007; Brav et al., 2009; Bharath et al., 2011; Li et al., 2011).
A standard loan contract contains five key terms: loan price, collateral, maturity, covenants, and loan size (Lummer and McConnell, 1989; Strahan, 1999;
Gottesman, 2006; Ge, Kim, and Song, 2012). While no single study examines the jointness of all five terms, a few consider that of at least some. Collectively, these empirical findings suggest that these loan terms may be simultaneously determined.
For instance, the evidence on loan price and non-price terms indicates that loan price
13 See Section 3.3.2 for the reduced-form explanation.
is simultaneously determined with collateral and maturity (Dennis et al., 2000;
Bharath et al., 2011; Li et al., 2011). Loan price is also found to be jointly determined with maturity and covenants (Brav et al., 2009) and with maturity and loan size (Champagne and Kryzanowski, 2009). These studies collectively show that all key non-price terms (i.e., collateral, maturity, covenants, and size) are jointly determined (Dennis et al., 2000; Champagne and Kryzanowski, 2008; Pham and Lensink, 2008; Vasvari, 2008; Brav et al., 2009; Alexandre et al., 2011; Bharath et al., 2011).
As the prior literature suggests, all five loan terms may be simultaneously determined. The first hypothesis for RQ1 is therefore stated as follows.
H1.1: Loan contract terms (i.e., price, collateral, maturity, covenants, and size) are simultaneously determined.
In practice, lenders may set their loan prices based on the other loan terms negotiated when finalizing a loan contract (Dennis et al., 2000; Bharath et al., 2011;
Li et al., 2011). Prior research (e.g., Strahan, 1999) finds that lenders alleviate information asymmetry problems and control for borrower risk through the non-price terms and then price for the remaining risks. This initial trade-off between non-price terms and the subsequent determination of loan price is also documented in Standard
& Poor's (2011a) guide to U.S syndicated loans. In other words, the loan price is decided from non-price terms that are traded off through negotiations between lenders and borrowers. Accordingly, the relations between loan price and non-price terms are unidirectional, whereas those between non-price ones are bidirectional (Dennis et al., 2000; Bharath et al., 2011). Therefore, two additional hypotheses are more specifically worded as follows.
H1.2: Loan prices have unidirectional relations with non-price terms.
H1.3: Each non-price term has a bidirectional relation with the other non-price terms.
As discussed in Section 3.3.2, if the five loan terms are not empirically endogenous (i.e., H1.1 is not supported), then H1.2 and H1.3 will not be tested.14 Since revolving and term loans differ in their characteristics, these hypotheses are tested for the two loan types separately (as discussed in Section 3.3.1). There are no specific expectations for their test results.