Third-party certification in new issues of corporate tax-exempt bonds: standby letter of credit and bond rating interaction

Financial Management (Financial Management Association), Spring, 1996 by Roger D. Stover

Figure 2 presents the results from fitting the pricing model to the industrial revenue and pollution control bonds that were supported by standby letters of credit. It assumes that the bank quality variable can influence interest cost either directly or indirectly through the bond rating. With the exception of CALL, all of the coefficients are identical in sign to those for the total sample in Figure 1. While the magnitudes of the paths are less than in Figure 1, the difference between the direct and indirect effects of INT on TIC are very similar to those for the total sample.

The latent variable, QUALITY, is depicted in the model for the LOC banks. This variable, being latent, is unobserved, but it will produce values for different indicator variables. Here I use the variables BANKRAT, ASSET, and RANK as indicators of QUALITY. Factor loadings for these indicators are statistically significant. To interpret these relationships, one must consider how changes in the latent variable result in changes for the indicator variables. Thus, banks having higher levels of QUALITY also have higher levels for BANKRAT and ASSET and higher profitability as evidenced by a lower RANK coefficient. Since BANKRAT and RAT came from the same source (Moody's), the residuals for them were allowed to correlate (r = 0.53; t = 5.63). By using the latent variable and modeling the common measurement error variance, I obtain a clearer understanding of the relationships within the model, one in which measurement error in the path coefficients of interest has been disattenuated. Based upon the signs of the coefficients, higher quality banks are significantly larger and more profitable, thus indicating a better reputation in this debt market. The higher quality banks also exhibit higher bond ratings on their unsecured debt, thus confirming the higher credit quality. Therefore, the latent variable accurately reflects the reputational and credit [TABULAR DATA FOR TABLE 2 OMITTED] quality characteristics expected from observing certifying agents in this debt market.

In estimating the model for the LOC banks, the scale for QUALITY is set using the BANKRAT indicator as the basis for measurement. As shown in Figure 2, the standardized path coefficient from QUALITY to RAT (0.48) indicates that a one-standard-deviation change in QUALITY results in a 0.48-standard-deviation change in RAT.(6) The direct effect of bank quality on bond issue cost is not significant, which indicates that this debt market does not react directly to variations in the quality of the bank issuing the letter of credit. However, the ratings for this sample of credit-enhanced bonds are significantly affected by bank credit quality. Similar to the results in Table 2, in which standby letters of credit impact interest cost indirectly through the influence on the bonds' ratings, the effect of variations in the quality of the bank is correspondingly transmitted. Thus, this market does respond to the quality of the bank issuing the standby letter of credit. This is consistent with other evidence sampling the certification hypothesis furnished by previous studies of initial public offerings. However, as Megginson and Weiss (1991) found for multiple certifying agents in the equity markets, the rating agency plays a substantive mediating role in the corporate tax-exempt debt market. Part of the reason for the mediating role of bond ratings may be the lack of alternative sources of information regarding the risks inherent in the issuer of the credit-enhanced bond and also the quality of the bank issuing the standby letter of credit.