To prevent the potential misidentification of lender and borrower names, this thesis examines only the actual lenders and borrowers (Qian and Strahan, 2007; Goss and Roberts, 2011; Li et al., 2011; Santos, 2011) rather than their parents (e.g., Do and Vu, 2010; Lin, Ma, Malatesta, and Xuan, 2011; Murfin, 2012). Although DealScan’s custom bank inventory reports help identify parent bank names, they
should not be used without double-checking. This is because after a merger (or an acquisition), DealScan overwrites the old parent bank’s history and attributes its loans to the most current ultimate parent bank (Cai et al., 2010; McCahery and Schwienbacher, 2010; Erkens, Subramanyam, and Zhang, 2012).
For instance, on 13 November 2004 Bank One, NA, known as the First National Bank of Chicago (First Chicago) before 13 September 1999, was acquired by JPMorgan Chase Bank (JPMorgan).23 DealScan, however, reassigned all First Chicago’s loans to JPMorgan, including those made before 13 November 2004.24 The same approach is applied to borrowers: DealScan overwrites the historical information about a borrower’s ultimate parents after any borrower mergers or acquisitions (Jiang, Li, and Shao, 2010, footnote 17). Misidentification problems can therefore result if the parent lender and parent borrower names are simply obtained directly from DealScan. This thesis employs the actual lenders and borrowers to avoid these problems.
3.2.3. Revolving and Term Loans
Revolving and term loans are the two main categories of bank loans,25 each with their own unique features. This section contrasts them with respect to bank fund access, risk, lending techniques, loan purposes, and borrower types and highlights the need to treat them separately.
23 From the NIC website of the FFIEC at
http://www.ffiec.gov/nicpubweb/nicweb/InstitutionHistory.aspx?parID_RSSD=173333&parDT_END
=99991231 (accessed 21 September 2011).
24 Extracted from DealScan.
25 These two types of loans comprise more than 94 per cent of all loan types in this thesis’s initial sample. Other types include letters of credit, demand loans, bridge loans, etc., which are excluded from the final sample since their structures are uncommon (Strahan, 1999).
With regard to bank fund access, a revolving loan offers the borrower a credit line with a maximum limit. Within this amount, the borrower can withdraw, repay, and re-borrow money throughout the contract (Ho and Saunders, 1983; Angbazo et al., 1998). If the amount borrowed is less than the credit line, the borrower must pay a commitment fee on the unused portion (Gottesman and Roberts, 2004). In contrast, a term loan is usually lent in full at the start and then repaid in regular instalments and the repaid funds cannot be re-borrowed (Angbazo et al., 1998).
Regarding risk, the ways in which customers access their funds expose lenders to different risks. With term loans, the main risk for lenders is credit risk.
With revolving loans, lenders face liquidity risk as well. (Ho and Saunders, 1983).
In terms of lending techniques, revolving loans are typically relationship driven, while term loans are more transactional based (Berger and Udell, 1995).
With relationship-driven loans, the loan approval decision is made based partly on information produced from lending relationships or soft information, whereas transaction-driven loans are based on hard information (Berger and Udell, 2006;
Brick and Palia, 2007).
In relation to loan purposes, revolving loans are primarily used for working capital purposes, whereas term loans are for capital goods (Blackwell, Noland, and Winters, 1998). For the former, lenders must forecast their borrowers’ working capital requirements, potential changes in expected usage, loan repayments, and further drawdowns. For term loans, the funds obtained are a function of borrowers’
long-term capital expenditures and the payment depends on surplus cash flows due to the borrowers’ long-term profitability.
With respect to borrowers, revolving loans are typically allocated to larger, more established borrowers, while term loans are usually allocated to small, young ones (Liu, 2006). This is because larger, more established borrowers can readily access other sources of funds, such as securities and commercial paper markets.
They may therefore need bank loans only for potential investment opportunities suitable to revolving loans (Strahan, 1999; Liu, 2006).
Given their distinguishing characteristics, revolving and loan terms should be examined separately,26 but this is not the case for many studies (for instance, Demiroglu and James, 2010a; Do and Vu, 2010; Bharath et al., 2011). Some studies acknowledge this but then focus only on revolving loans (e.g., Dennis et al., 2000).
Only a few studies examine both types of loans (Strahan, 1999; Coleman et al., 2002; Gottesman and Roberts, 2004, 2007), but they do not cover loan term jointness, borrower information asymmetries, or lending relationships. The four thesis research questions (stated in Chapter 1) are therefore examined across revolving and term loans separately. The sampling procedures discussed next, in Section 3.2.4, cover these two distinct loan samples.
3.2.4. Sample Units, Sampling Procedures, and Composition
The final sample consists of revolving and term loans distributed under sole lender or one-lead lender syndication27 made by banks operating in the United States to U.S. non-financial firms with loan active dates from 1 January 1994 to 31
26 Revolving and term loans’ differences are found to be statistically significant in this thesis’ sample (as shown in Chapter 4, Table 4.3 in Section 4.3.1).
27 Only syndicated loans with one lead lender are considered to facilitate a clear interpretation of lender impact on loan contract terms (Güner, 2006).
December 2009. The thesis emphasizes revolving and term loans since they are the two main loan types, comprising 94 per cent of all loans in the initial sample. Other loan types, such as letters of credit, demand loans, and bridge loans, have less common structures and are therefore excluded (Strahan, 1999).28 The focus on U.S.
borrowers and lenders operating in the United States helps reduce cross-country differences in regulation and funding practices (Coleman et al., 2006; Qian and Strahan, 2007) while providing the best data accessibility. Financial borrowers are excluded because their financing and investment activities differ from those of non- financial borrowers (Dahiya, John, Puri, and Ramírez, 2003). The sample period, 1994–2009, is a function of data availability since loan covenant information prior to 1994 is limited.29 Syndication and sole lender loans are the most common methods of distribution.30 The sampling procedures and composition are detailed next.
As shown in Table 3.3, the initial sample from DealScan comprises all loans made by banks operating in the United States to U.S. non-financial firms with deal active dates from 1 January 1987 to 31 December 2009.31 This results in 52,960 deals and 77,373 facilities. On average, each deal contains 1.46 loans.
28 See Appendix A, Panel A of Table A1 for their annual distribution.
29 Chava and Roberts (2008) also highlight this issue. For further covenant information, see the yearly distribution of covenants in Appendix A, Table A2.
30 Other loan distribution methods include bilateral and club deals and some loans remain unspecified.
Such loans comprise only 5 per cent of this thesis’ initial sample. Since they are not common, they are excluded, following Champagne and Kryzanowski (2007, 2008).
31 Since the DealScan sample starts in 1 January 1987, this database was accessed in January 2010.
Table 3.3: Sample selection process
This table summarizes the sampling procedure results.
Steps Filter No. of Loans
Excluded Remained 1 DealScan is searched for all loans that banks operating in the United
States made to U.S. non-financial firms with deal active dates occurring from 1 January 1987 to 31 December 2009.
77,373
2 Loans without borrower GVKEY or lead lender information. (31,311)
46,062 3 Loans for which borrower financial information from Compustat is not
available.
(4,443)
41,619 4 Bilateral and club deals and unspecified loans. (2,046)
39,573 5 Syndicated loans with more than one lead lender. (7,565)
32,008 6 Loans with active date prior to 1994. (6,372)
25,636 7 Loan types other than revolving or term. (1,375)
24,261
Revolving loans 17,636
Term loans 6,625
Source: Prepared by the author.
In the second step, 31,311 loans without borrower GVKEY or lead lender information are excluded (Sufi, 2007; Chava and Roberts, 2008; Drucker and Puri, 2009; Bharath et al., 2011; Valta, 2012), because without a GVKEY (as explained in Section 3.2.2.3), borrower information cannot be obtained from Compustat. In addition, loans without a borrower GVKEY or lead lender information cannot be traced for prior lending relationships. Of the remaining 46,062 loans, only 41,619 have borrower information available in Compustat.32 All bilateral loans, club deals,
32 Appendix A, Table A1, documents the distribution of these loan characteristics over this initial sample period, 1987–2009.
and unspecified loans are then removed33 (Strahan, 1999), resulting in a sample of 39,573 syndication and sole lender loans. Of these, in a fifth step, the 7,565 syndication loans that have more than one lead lender are excluded (Güner, 2006).34 The sample now has 32,008 loans.
As shown in Appendix A, Table A2, loan covenant information was very limited before 1994. In a sixth step, 6,372 loans made prior to 1994 are excluded (Chava and Roberts, 2008), leaving a sample of 25,636 loans. Finally, those 1,375 loans other than revolving and term loans are removed (Strahan, 1999; Gottesman and Roberts, 2004).35 This results in a final sample of 24,261 loans, consisting of 17,636 revolving and 6,625 term loans.36 This final sample is larger than those of many prior U.S. bank lending studies.37