THE ECONOMICS OF THE PRIVATE MARKET
Appendix G. Estimates of Issue-Size and Maturity Distributions
This appendix explains how the issue-size and maturity distributions described in part 1, section 2, were computed and presents distributions for all private placements (those in the text are for placements by only nonfinancial corporations). Distributions for commercial and industrial loans made by U.S. commercial banks, private placements issued by nonfinancial corporations, and public debt issues by nonfinancial corporations are shown in the text for 1989. We chose this year because it was the last year before the credit crunch in the below-investment-grade segment of the private placement market. The data and distributions for each of the three types of instrument are described separately.
Both issue-size and maturity distributions for new issues of private placements were produced from a database of 1989 private placement issues obtained from IDD Information Services. This firm is associated with the publisher of Investment Dealers Digest. 206 This database includes information about new private issues of both debt and equity, including issue size (amount); issue date; the name of the issuer; the name(s) of the agent(s) involved; the nationality of the issuer; the type of security involved; an industry code for the issuer (an IDD designation, not an SIC code); a maturity date where applicable; an indication of whether the issue was junk, lease-related, or part of an acquisition-related financing; and, in some cases, a coupon rate or a number of shares issued and a few words of descriptive comments. Collected primarily from reports that agents sent to IDD, the data include few or no deals not involving an agent and omit many agent-assisted deals that were unreported. The accuracy of the data has not been verified, and we have reason to believe that at least some of the data are unreliable (for example, the junk designation is usually ''no'' for issues assisted by Drexel, which does not square with Drexel's reputation for specializing in below-investment-grade issues).
Issue-size distributions were produced using the issue-size (amount) field in the database. Maturity distributions were produced by computing maturity at issue from the issue date and maturity date appearing in the database. This computation often involved some approximation, since in many cases only the month and year (not the day) of issuance and maturity appear in the database, and in some cases only the year of maturity is specified. In these cases we assumed that the month and day of maturity were the same as the month and day of issuance; in effect, we may have rounded up some maturities to the next year.
Both issue-size and maturity distributions were produced from a sample limited to 1989 issues of bonds by U.S. issuers. Two subsamples were analyzed. The one shown in the text (charts 4, 5, 10, and 11) was for issues by nonfinancial corporate borrowers and excluded medium-term notes, convertible or exchangeable debt, and mortgagebacked securities. 207 This subsample totaled 1,020 issues (maturity dates were available for only 901).
The second subsample excluded only mediumterm notes and contained 1,620 issues (maturity dates were available for only 1,373 of them). Charts G.1 through G.4 display the distributions for this subsample. These distributions are similar to those for the other subsample.
The distributions for new issues shown here are not necessarily representative of the distributions for outstanding private placements. Adequate data on outstandings are not available. Kwan and Carleton (1993) report that the average term to maturity for a sample of 563 private placements issued between 1985 and 1992 was 11.12 years, a finding that is in rough agreement with the distributions displayed here. 208
Publicly Issued Bonds
Issue-size and maturity distributions for new issues of public bonds by U.S. issuers were produced from a sample issued during 1989 by nonfinancial corporations and collected by the Federal Reserve Board from public announcements, proxy statements, and other sources. The subsample analyzed here included no government issues, no mediumterm notes, no convertible or exchangeable debt, and no mortgage-backed or other asset-backed securities. An unusual issue related to the RJR-Nabisco merger was also omitted. This sample included 297 issues.
Commercial and Industrial Bank Loans
Issue-size and maturity distributions for new or renewed bank loans were produced using data from the Quarterly Survey of Terms of Bank Lending to Business. This survey is of a stratified random sample of approximately 350 banks representing insured commercial banks in the United States. Large banks are oversampled. Participating banks report data on all loans to businesses made during the first full week in February, May, August, and November (larger banks report data only for loans made on two or three days of those weeks and only for a subset of branches). Individual survey responses are confidential and include for each loan an amount and maturity date as well as other information. No maturity date is recorded for demand loans with no specific maturity date, and rules for reporting loans made under a revolving credit agreement are complicated, in many cases requiring that no maturity date be reported. Construction and land development loans were omitted from the sample we analyzed.
Several steps were required to arrive at distributions representative of commercial and industrial loans by all banks. First, subsamples of loans by banks reporting for less than five days in a survey week or for less than all branches were blown up to make them comparable to subsamples of five-day, all-branch reporters. Then sample banks were stratified according to size, using their four-quarter averages of total C&I loans outstanding at the end of each quarter (computed from Call Report data). We pooled loans for banks in each stratum and computed size and maturity distributions for each stratum. Essentially, we assumed that the sample of loans made by survey banks in a stratum was representative of the population of 1989 C&I loans made by all banks in that stratum. Distributions were computed for each bank size stratum because loan sizes and loan maturities tend to be related to the size of the lending bank, with larger banks making more large and more long-term loans.
We arrived at loan-size and maturity distributions for all bank loans by taking a weighted average of the distributions for the different bank-size strata. The weighting variables were the fractions of all C&I loans outstanding that all banks in a stratum showed on end-of-quarter Call Reports (averaging these outstandings for all of 1989). That is, having achieved (we hope) representative distributions for each size class of banks (as described in the previous paragraph), we weighted the distributions according to the share of total outstanding loans accounted for by each size class. Because the largest banks make a disproportionate share of C&I loans, the distributions for the large-bank strata had more influence on the final, representative distributions than did the distributions for small-bank strata.
As noted, the distributions are for newly originated or renewed loans, not for the population of loans outstanding. Distributions for outstandings may differ substantially from those for originations, for two principal reasons. First, short-term loans may essentially be overweighted by our method of constructing distributions although a sensitivity analysis indicated that any such overweighting probably has little effect on the estimated distributions. Second, bank loans are often prepaid.
Our distributions have other drawbacks. Most notably, the data available from the survey for revolving credit agreements are limited, and these constitute a significant share of all bank loans. However, the proper maturity date definition to apply to revolvers in comparing them to private placements and public bonds is not clear.