THE ECONOMICS OF THE PRIVATE MARKET
Differences among Firms Issuing in the Public, Private, and Bank Loan Markets
To summarize the preceding discussion, borrowers' access to debt markets is apparently closely related to firm size, with size mainly a proxy for the degree of information problems that borrowers pose for lenders. Broadly speaking, very information-problematic companies without collateral may be unable to borrow even from an information-intensive lender. 53 Such companies, which are typically small, may be forced to rely on venture capital or on other forms of equity finance. Small firms that are less information problematic or those that can provide collateral are confined largely to the bank and finance company loan markets for debt financing. Even less problematic firms, which are typically medium-sized, also have access to the private placement market. Large corporations can borrow in any of these markets and in the public bond market. Besides size of the firm, other characteristics, especially those related to the nature and size of the financing, are important in determining a firm's choice of credit market.
Empirical evidence supports these assertions. We analyzed the characteristics of firms classified according to a hierarchy of access to the public, private, and bank loan markets and found a pattern of firm sizes and other characteristics consistent with the explanation of borrowers' choice of market that focuses on the different information problems posed by different firms. However, borrower size is also correlated with issue size and with observable borrower risk, so the observed difference in sizes of firms with different levels of access is also potentially consistent with explanations based on issuance costs or risk. To evaluate the relative importance of the three explanations, we looked at several other firm characteristics that are plausibly correlated either with the degree of information problems or with observable risk.
We employed an indirect approach in identifying the access of actual firms to the three markets, since access is not directly observable. We combined information on corporations in COMPUSTAT with data on private placements from the IDD database. 54 Corporations in COMPUSTAT with a long-term credit rating are assumed to have access to the public bond market, inasmuch as they must have issued corporate bonds at some time to have received a bond rating; those without a rating are assumed to lack access to the public market. 55 In 1989, 1,149 corporations in COMPUSTAT had ratings and thus constitute the public market group, that is, those corporations with the ability to raise funds in public debt markets. 56 To form a group of firms with access to the private placement market but not to the public market, companies listed in the IDD private placement database as issuing in 1989 were matched with those in COMPUSTAT that had no credit rating. The cross-matching of the two databases yielded a total of 113 such companies, which make up what is called the private market group. Those firms in COMPUSTAT that in 1989 had neither a credit rating nor outstanding long-term debt but that did have some short-term debt outstanding were assumed to be constrained to borrow only from banks (or other, bank-like intermediaries such as finance companies); this collection of firms is called the bank group and contains 472 members. Finally, those firms in COMPUSTAT that had neither a credit rating nor any outstanding debt (short or long term, except for trade debt) in 1989 were assumed to be shut out of all three debt markets. This collection of firms is called the equity group and consists of 613 firms.
This method of classifying firms is far from perfect for various reasons. First, and perhaps most important, implicit in the definition of each group is an assumption that a company cannot tap a particular debt market if it has not actually done so. This assumption is clearly not correct in all cases. For example, several firms classified in the bank group probably could have issued in the private or public bond markets on standard terms but simply chose not to do so. Firms that issued private placements before 1989 but not in 1989 are less likely to fall in the bank group because such firms probably still showed long-term debt on their balance sheets in 1989. Second, according to the bank group definition, the presence of shortterm debt on the balance sheet indicates the firm's ability to tap the bank loan market. However, COMPUSTAT's definition of short-term debt includes loans from various lenders: loans payable to stockholders, officers of the company, parents, subsidiaries, and brokerage companies as well as loans payable to banks, finance companies, and other intermediaries. Our aim is to include in the bank group all firms that have access to banks or bank-like intermediaries, but several firms without such access were probably misclassified (they should be in the equity group) because they had loans outstanding from stockholders or other non-intermediary sources. Third, many equity group firms may have had bank lines of credit that were simply unused at the end of their 1989 fiscal years. 57 Fourth, the presence of a credit rating in COMPUSTAT implies only that a firm once had access to the public bond market, not that it had access in 1989.
The private market, bank, and equity groups are also undoubtedly biased selections of firms because only those firms that appear on the COMPUSTAT tapes have been selected. COMPUSTAT's bias toward large firms means that the firms in these three groups are likely larger on average than corresponding groups of firms for the economy as a whole. Other characteristics may show some bias as well. However, the bias probably makes observed differences across groups less dramatic. Consequently, any differences found in the analysis are unlikely to be the result of this sampling bias.
Finally, the criteria used to define the four groups focus on the characteristics of the firm, not on the characteristics of the debt issue. As mentioned earlier, some firms that could readily issue straight debt in the public market may be constrained to the private market for more complicated issues such as some leases or project financings. We address this issue later in this section.
Despite these classification problems and biases, we believe our method of classifying firms is on the whole roughly accurate and that the distinctions that are revealed are economically meaningful.
The firm characteristics examined include the size of the firm, measured by total assets, sales, and market value of equity. We also looked at the three-year growth rate of sales, return on assets, (measured by operating income before depreciation divided by total assets), research and development (R&D) expenditures as a percentage of sales, the fixed-asset ratio, the ratio of total debt to assets, and the interest coverage ratio.
Differences in firm size across groups, measured by total assets, total sales, or market value of equity, are pronounced (table 5). Firms in the public market group are much larger than firms in the private market group, which in turn are very much larger than firms in the bank or equity groups. For example, mean assets of companies in the public market group are $6.3 billion, considerably larger than the mean of $3.4 billion for firms in the private market group. The means for the bank and equity groups are even smaller at $40 million. These differences in means are all statistically significant at the 1 percent level. The medians have a similar relationship among the three groups. 58
Table 5 presents statistics for three other variables that are plausibly correlated with the degree of information problems posed by firms: the ratio of R&D expenditures to sales, the fixed-asset ratio, and a three-year average growth rate for sales. Many economists have used R&D expenditures as a proxy for the potential severity of agency problems between shareholders and debtholders. 59 The risk implicit in research and development cannot be easily monitored by outsiders, including debtholders, as a firm with large R&D expenditures has wide scope for discretionary behavior. For example, such a firm may require intensive monitoring by debtholders to ensure that it is working on a mundane research project with a moderate but fairly sure payoff rather than a longshot with a high payoff. Intensive monitoring may be required to ensure that the firm is not underinvesting in projects with positive net present values (Myers, 1977). R&D-intensive companies, being inherently more information problematic than other firms, may therefore find banks more receptive to providing financing because banks can monitor more intensively than lenders in the public markets. The evidence provided by this variable on the intensity of monitoring in the private placement market generally conforms with our hypothesis about differences in the degree of information problems across the four groups. Mean R&D intensity is higher in the private placement market than in the public market, although the medians are about the same. The significantly higher R&D intensity for the bank and equity groups than that for the private market group indicates that issuers in the former groups tend to require significantly more monitoring by lenders than do issuers of private placements.
A similar hierarchy of information problems is suggested by the fixed-asset ratios. Firms with a large percentage of fixed assets may have fewer information problems than other firms for two reasons. First, they may be able to offer some of their fixed plant and equipment as collateral to potential creditors. Second, monitoring the sale of fixed assets or their transformation from one use to another may be easier than it is for more liquid assets. The more of a firm's assets that are fixed, therefore, the smaller may be the scope for shareholders to engage in wealth-transferring investment projects.
As one moves from the public to the private to the bank and finally to the equity group, the decline in fixed-asset ratios implies that information problems increase. The higher fixed-asset ratio for the bank group compared with that for the equity group suggests that a small firm's ability to provide fixed assets as collateral may be a factor in its ability to obtain bank loans.
Sales growth rates may also be correlated with information problems in that high growth may be a sign of entry into new lines of business or of being in lines of business that are in rapidly developing markets. Both situations offer more scope for agency problems to surface during the life of a debt contract. The evidence from this variable, however, is weaker than that from R&D intensity and the fixed-asset ratio: The mean is significantly smaller for firms in the public group than for those in the private group, a finding consistent with private issuers requiring more monitoring; the median is smaller as well. Values for the private group do not differ significantly from those for the bank and equity groups, however, and the medians display an uneven pattern.
On the whole, the results for the three variables conform with our hypothesis about the differing degree of information problems posed by the four groups of firms. They also accord with the remarks of market participants, who asserted that buyers of private placements, especially the larger life insurance companies, engage in organized and active monitoring, although their monitoring programs are typically not so intensive as those of banks.
Average return on assets and two measures of leverage, total-debt-to-asset ratios and interestcoverage ratios, are indicators of observable credit risk. As noted in part 1, section 1, information problems and observable credit risk are separate concepts, and in principle there is no reason that the pattern of credit risk should be different in information-intensive and non-informationintensive markets. In practice, however, both are related to borrower size.
Caution should be used in interpreting the differences between the bank and equity groups and the other groups in the measures of leverage, as firms in the former groups either had no long-term debt outstanding or no debt at all on their balance sheets (according to COMPUSTAT and ignoring trade debt). Thus, zeros will appear in either the numerator or the denominator of the ratios for many equity group firms, making the ratios poor measures of the riskiness of these firms and influencing the mean and median values for the groups.
A comparison of ratios for the public and private placement groups indicates that differences in credit risk may not be as great as differences in information problems. Both the mean and median debt-to-asset ratios and the return on assets are similar for the two groups. Median interestcoverage ratios are also similar, but the mean interest-coverage ratio is significantly higher for the public group. The implication is that private placements issuers may be somewhat riskier as a class, but not a great deal riskier, than public bond issuers. Comparing ratios for the private placement and bank groups, the means of the three ratios differ significantly; the medians also differ as predicted except for the debt-to-asset ratio. It appears that members of the bank group pose larger observable credit risks for lenders. 60
On the whole, these results accord well with the remarks of market participants, who often described private issuers as ''solid companies'' that have taken a major step in ''graduating'' from having access only to the bank loan market but that are typically ''not quite ready'' to issue in the public bond market. Some investors also indicated that their historical experience of loss on private placements and public bonds was virtually identical within credit-rating categories. The statistics presented here and the remarks of participants offer little support for a hypothesis that low observable credit risk is the primary requirement for a borrower to have access to the public market, instead of only the private placement and bank loan markets. The existence of the public junk bond market and the fact that contract terms, especially covenants, and lender due diligence and monitoring activities differ across the public and private markets for borrowers with the same bond ratings also imply that information problems are a more important determinant of market access than observable credit risk.
In sum, if the groups of firms analyzed here are representative of borrowers' access to debt markets, then their characteristics are broadly consistent with our explanation of the factors influencing borrowers' choice of debt market. Corporations able to borrow in the public markets tend to be large and to pose relatively few information problems for lenders; thus they can borrow from a wide variety of lenders. Companies issuing in the private but not the public market are smaller and appear to be more information problematic; however, they apparently do not represent substantially greater observable credit risks. Such companies must be served by information-intensive lenders. The companies confined to the bank loan market or to equity markets are much smaller, are more information problematic, and pose larger pure credit risks. Consequently, they require the greatest degree of due diligence and loan monitoring by lenders, or they are unable to issue debt at all. The information problems associated with smaller and medium-sized firms and their increased need for information-intensive lenders appear to be the major reasons for the size pattern observed among the three groups and for the differential access of firms to credit markets.