A. The Market. Unlike for large firms, the information available about small and private firms has historically been limited and difficult to access. With the exception of some in high growth industries - which are a very small portion of our sample - analysts do not follow these firms. Since these firms do not raise capital in public markets, they are not required to disclose much information. The firm’s lenders clearly know about the firms, but these lenders are few in number and did not readily share information in the past. Since information about the firm was not compiled, stored, and distributed by a central bureau, but instead resided in the minds of the firm’s bankers, much of it tended to be soft- whether the firm generally maintained adequate balances, for example, rather than hard information specifying when and to whom it had, or had not, made payments in the past.
B. Information Technology and Small Business Lending. The use of information and communications technology, by which we mean everything from hardware like computers and phones to software like credit scoring and client profitability programs, has transformed the financial sector over the last three decades ( Mishkin and Strahan, 1999). Three aspects are particularly significant to us. First, the ability to collect, store, process, and communicate large amounts of information has expanded tremendously. Second, this has resulted in the expansion of the activities of infomediaries whose sole purpose is to collect, organize, and make available this information to paying customers. Third, the availability of hard, processed information lends itself to cost effective credit appraisal and monitoring techniques. Since the first aspect is fairly uncontroversial, let us examine the latter two in more detail.
1. Expansion of the Activities of Infomediaries. Technological change has resulted in the expansion of the activities of infomediaries such as rating agencies and credit bureaus. Consider, for example, Dun and Bradstreet. It was founded as far back as1841, with the aim of establishing a network of correspondents that would function as a source of reliable, consistent and objective credit information. Big changes in its procedures came in the 1960s and the 1970s. In 1963, the introduction of the Data Universal Numbering System used to identify businesses numerically for data-processing purposes revolutionized the collation and distribution of business information. In the 1970s, a new "Advanced Office System" fully computerized D&B’s data-collection operations, giving it the ability to link and analyze categories of information in entirely new ways, and to deliver information to customers faster and more economically.
The pace of these changes have accelerated with the coming of the Internet (an era that post-dates our sample). Over our sample period, the number of firms on which they have records has grown at 6.3 percent per year, a rate over two and a half times the real growth of the economy. Hundreds of millions of pieces of data, ranging from trade experiences to financial statements, are integrated every day into one file. D&B collects information from millions of on-site and telephone contacts with business owners and managers, as well as from all federal bankruptcy filing locations, from millions of trade and bank experiences, from public utilities, from over 2,500 state filing locations, and from daily newspapers, publications and electronic news services. The data that is entered is automatically checked, and also subject to random verification. Finally, D&B alerts its customers to increases in a business’s risk profile so as to prevent unnecessary losses. Specialized infomediaries like D&B can save on duplication, and amortize the costs of information collection over a larger number of customers than could lenders in the past. As a result, they can distribute more information than ever available to lenders in the past.
2. More Efficient Appraisal and Monitoring. The increased availability of systematic reliable information has allowed loan officers to cut down on their own monitoring. Moreover, the information can now be automatically processed, eliminating many tedious and costly transactions. For example, Automatic Loan Machines (ALM) now offer loans on the spot to individuals who have a reasonable credit history, regardless of who they banked with in the past (see Rajan (1996)). In an ALM, the process of taking the client’s information, checking records, evaluating the expected profitability of the loan, and then making the actual loan has been completely automated. Credit scoring - a process by which a loan applicant’s credit history and characteristics are summarized in a credit score which forms the basis for approval or rejection of most applications -is increasingly used by large banks such as Wells Fargo to make lending decisions even for small businesses (Mishkin and Strahan, 1999). By using financial histories, credit reports, and scoring methods, the banks can dramatically lower the time their loan officers spend on a given application and thus the cost (Padhi, Srinivasan, and Woosley, 1999, Mester, 1997).
Small firms gain substantially by a lowering of the fixed costs in the lending process (Frame, Srinivasan, and Woosley, 1999). The median loan in our sample is for $18,000. Firms in our sample were asked the total fee (not including interest) that they paid to obtain their loan. The level of the fee is uncorrelated with the size of the loan across the sample (correlation = 0.001). Thus, fees as a fraction of the loan size declined with the size of the loan. Every ten percent increase in the loan size, lowered the fees as a percent of the loan amount by 4.8 percent (t=13.4). Given that they are largely fixed, reductions in the cost of loan origination and information collection could therefore produce the largest gain for smallest firms. Additionally, if transactions costs drop sufficiently, the number of lenders that are willing to lend may expand. This has the possibility of not only expanding the supply of finance to small firms, but also reducing the cost of financing to the extent that geographically larger markets are more competitive. We explore this issue in Section V. If in addition to being available at low cost, information is timely, the lender reduces the potential loss from borrower moral hazard. If a borrower, either because of incompetence, or because of malevolence, takes improper actions, the lender can act quickly to stop further lending and can demand repayment. By contrast, if the lender acquires information after a long lag, he may have thrown good money after bad, the borrower’s assets may have deteriorated under poor management, and other lenders may have seized anything of value. Thus timely information thus reduces the costs of lending.
There is a potential downside, however, to information technology. The literature on small firms has stressed the importance of relationship lending for small firms (see the survey by Berger et al. (1998) for references). Historically, part of the incentive for a lender to develop a relationship with a borrower, even if initial loans were not cost effective, was the knowledge that, if successful, the firm would be locked in to a long term relationship because the lender would have monopoly access to information about the borrower. The cost of the initial loan could be amortized over the longer relationship (see Greenbaum, Kanatas and Venezias (1989)). However, when information is widely shared, the credit market becomes more competitive, which could reduce the availability of credit (see Petersen and Rajan (1995) for theory and evidence). Another potential problem is that soft information is difficult if not impossible to incorporate when credit decisions are made by computer models and credit reports. Thus a concern is that small firms who are truly good credit risks - but on paper look like bad credit risks - will find capital more difficult to obtain. As the relative costs of funding such firms rise with advances in information technology, lenders may simply ignore them, preferring to focus attention on the transparent. We will attempt to sort out these explanations in the rest of the paper.
No comments:
Post a Comment