Evaluation of yields for insured tax-exempt hospital bonds.

Chief Financial Officers (CFOs) consider access to capital as a critical component of the overall financial stability of their hospitals. A 2004 survey of CFOs revealed that hospitals were expected to increase their investment in plant and equipment by 14 percent over the next five years (HFMA, 2004). This expected rise in capital investment was precipitated by an underinvestment in plant and equipment by hospitals from 1997 to 2001. Recent data reconfirms this CFO survey as hospitals experienced an upward trend in their capital expenditures relative to depreciation expense, which grew in 2003 to 143 percent from 125 percent in 2001 (Fitch Ratings, 2006). (1)

To finance these future capital needs, hospitals and health systems consider taxexempt debt as one of the primary sources of capital (HFMA, 2004). When issuing taxexempt debt, one of the mechanisms assisting hospitals in acquiring the lowest cost of debt capital is bond insurance. Bond insurance enables hospitals and health systems to raise their bond rating, to increase the marketability of their issue, and more

importantly, to lower their cost of debt.

In 1997, 60 percent of the healthcare issues were insured; however, by 2000, the percentage of newly issued healthcare bonds that were insured declined to 28 percent (Fitch Ratings, 2006). This dramatic drop in insured issues was primarily attributed to the 1998 bankruptcy of the Allegheny Health, Education and Research Foundation (AHERF)--one of the largest hospital bond defaults in the United States (Jaklevic, 2004). This $353 million default on an insured bond issue increased uncertainty in the municipal bond markets and caused insurers to tighten their bond quality underwriting standards for healthcare issues (Carpenter et al., 2003).

As indicated by the Fitch Rating Agency (2006), the number of healthcare issues with bond insurance has shown a gradual increase from 35 percent in 2002 to 45 percent in 2005. In terms of volume, the Bond Buyer trade publication indicated that newly issued healthcare bonds with insurance grew by more than 50 percent--from $11.3 billion in 2003 to $17.2 billion in 2005 (Bond Buyer, 2005). However, Haugh (2006) notes that the rise in insured bonds stems not from relaxing of credit quality standards by the bond insurers, but from the financial and operational improvement of hospitals that met the credit standards of the bond insurers.

From a corporate finance standpoint, two theories--market segmentation and information asymmetry--explain why municipal bond issuers consider bond insurance. Market segmentation is a result of investor demand for AAA-rated taxexempt bonds (Dension, 2001); however, the greater credit risk of the hospital industry generally precludes hospitals from achieving this top rating on their own. As a result, hospitals issuing tax-exempt bonds often purchase bond insurance to improve their credit standing and marketability to potential investors (Wareham & Majka, 2003). Information asymmetry is a result of the vast number of municipal and tax-exempt organizations issuing exempt bonds--which makes it difficult for investors to accurately assess and monitor the credit risks. Investors are attracted to hospital bonds with insurance because it consolidates the task of evaluating and monitoring the credit risk of many hospital bonds down to evaluating just a few bond insurance companies. Several studies (Peng 2002; Denison, 2001, 2003) have tested these theories on the municipal bond market in general. However, a gap in the literature exists in testing market segmentation and information asymmetry within the context of tax-exempt hospital and healthcare system bond issues. Therefore, the purpose of this study is to test these theories by examining the association of bond yield difference between insured and uninsured hospital and healthcare system bonds (from 2000 to 2005) and examining the underlying (i.e., non-insured) bond ratings on the yields of insured bonds.

This study has five sections: the next section discusses the literature, theory and hypotheses; the following section describes the methodology, including the sampling process, empirical model, and variable selection; then the results from examining bond yield with insured bonds without underlying ratings are presented. The last two sections include results from examining bond yield of insured bonds with underlying ratings; and finally a discussion of these findings.

LITERATURE REVIEW, THEORY AND HYPOTHESES

Hospital bond issues that have obtained insurance gain the benefit of a guarantee by the insurer that it will meet timely payments of principal and interest. The primary reason investors prefer tax-exempt bond issues with bond insurance is because this enhancement protects them from loss of interest and principal payments. However, bond insurance does not protect the investors from risk of default by the issuer. Bond insurers typically target hospitals that possess sufficient credit strength to potentially qualify for the highest credit rating of AAA--which is typically the rating of the insurer (Dension, 2001).

Empirically, the healthcare literature includes only a few studies that have focused specifically on the relationship of bond yield and bond insurance within hospital and healthcare systems. Analyzing bond data in the early 1980's, Carpenter (1991) found that insured bonds had a yield 87 percent lower than uninsured bonds. From 1990 through 1993, McCue (1997) assessed a large sample of hospital bond issues and found that enhanced bonds, including bonds with insurance and letters of credit, incurred yields that were 51 basis points lower than issues without these enhancements. Another study (Carpenter et al., 2003) analyzed access to bond insurance and the yields of insured hospital bonds before and after the AHERF default. They found lower yields for insured bonds after the AHERF default but they also found a wider spread between the insured and uninsured bond yields and a lower proportion of insured bonds. These researchers concluded that insurers were only offering bond insurance to hospital issues with the highest bond quality.

Within corporate finance, two theories provide the underlying reasons why investors demand bond insurance in the municipal bond market: market segmentation and information asymmetry (Denison, 2003). Market segmentation is a result of municipal bond market investors' demand for high quality bond issues in a market with a limited supply of AAA-rated securities. The supply of underlying AAA-rated bonds is virtually non-existent in hospital and healthcare system since the overall bond risk of the hospital industry is too high to warrant an actual AAA-rating (HCIA, 2000). High demand for strong quality municipal bonds is a result of institutional investors (corporations, mutual funds, banks and insurance companies) with policies that restrict their portfolio managers to high grade bonds. This limited supply, and greater demand among risk-averse institutional investors may drive down the yields on low-risk securities (Denison, 2001). To overcome this segmentation, researchers (Hsueh & Chandy, 1989; Hsueh & Liu, 1990a; Below, 1994; Peng, 2002) found that companies used bond insurance as a means of reducing the market segmentation between high and low quality bond issues. However, these studies show that the yields on AAA insured bonds were higher than non-insured AAA-rated bonds and thus insured bonds were not a perfect substitute for non-insured AAA bonds. More importantly, these studies suggest that insured bonds had a higher default risk--therefore investors would expect a higher yield on these insured bonds.

Market inefficiency or information asymmetry is another theory regarding the demand for bond insurance (Denison, 2003). This inefficiency in the tax-exempt bond market is a result of the vast numbers of issuers that are local hospitals and municipalities and are generally unknown to the market investors. In addition, taxexempt bond issuers report their bond risk through lengthy and detailed disclosure reports, known as Official Statements, which require accounting and finance experts to provide a reliable interpretation of the bond's riskiness (Denison, 2003). Researchers note that bond insurance companies can reduce this information inefficiency by applying their expertise to track and assess the bond risk of these issuers and provide a broader and more accessible perspective on the bond quality information released to the market (Thakor, 1982; Diamond, 1984).

As previously noted, bond insurance does not protect investors from default risk. As a result, information asymmetry may also influence the yield of the bond due to a lack of information related to the underlying default risk of the insured issuer. A study by Peng (2002) notes that as more municipal bonds are insured, investors become concerned that during a major economic downturn bond insurers would have difficulty meeting the debt service payment guarantees on these insured bonds--a reality that has been painfully born out in the recent economic collapse. Bond investors, who only know the insured rating and not the true underlying rating of the bond issue, are more likely to assign the lowest rating that an insurer would assign to an insured bond, which is BBB. As a result, there is an incentive for issuers with an underlying credit rating greater than BBB to release their underlying bond rating to investors. Such action will allow investors to gain additional information on the issuer's underlying credit risk, and potentially lower the cost of debt demanded from them.

A study by Peng (2002) conducted a comparative analysis between insured issuers that released their underlying bond ratings to the market and insured issuers that did not release this information. The study found statistically significantly lower yields--by four basis points--among insured bond issuers with A-rated or higher underlying ratings. In addition, the study found insured bonds not releasing their underlying rating incurred yields similar to uninsured BBB-rated bonds.

Given that hospital and healthcare tax-exempt organizations generally do not have any underlying AAA-rated bonds (i.e., without insurance), this study compared each insured and uninsured rated issue to BBB-rated uninsured bonds (Moody's, 2005). The BBB-rated issues were selected as the comparison group since this category had the largest number of issues in the study sample.

The following hypotheses were developed from the two theories discussed above. Hypothesis one is derived from market segmentation and the high demand for AAA-rated insured issues:

[H.sub.1]: Hospital and healthcare systems with insured bond issues rated AAA are expected to have lower yields than uninsured AA-rated bonds, when both are compared to uninsured BBB bonds.

Hypothesis two is developed from information asymmetry or the lack of information about the underlying bond risk of the issuer.

[H.sub.2]: Hospital and healthcare systems with disclosed underlying ratings for insured bond issues are expected to have lower yields than uninsured bonds.

METHODOLOGY

This section includes discussion of the study sample, empirical model, and variable selection.

Sample

Because of the inability to identify yield information on variable-rate issues, the study accessed the Securities Data Corporation (SDC) database and selected fixed-rate, tax-exempt healthcare bond issues from January 1, 2000 to December 31, 2005. The study identified a sample of 199 fixed-rate bond issues. To account for the underlying risk of the insured issues, the study was able to identify the underlying rating of 25 insured bonds. Exhibit 1 presents the number of insured and non-insured bonds as well as the underlying ratings of both groups. Out of the 199 fixed-rate issues, 69 were insured while 130 were non-insured. The underlying ratings of the insured bonds were as follows: five AA-rated, 12 A-rated, and eight BBB-rated issues. Given the limited number of AA-rated issues, the study combined insured bonds with underlying ratings of AA and A into one category. (2) Out of the remaining 44 insured issues, eight were reported as non-rated, and 36 had no underlying rating.

Non-insured bonds totaled 130 issues--104 with underlying ratings and 26 with no rating. The underlying rated issues included the following: 12 AA-rated, 38 A-rated, and 54 BBB-rated issues. Out of the 26 issues with no rating, 23 were listed as having no underlying rating, while three were reported as being non-rated.

With 97 percent of the bonds being negotiated issues, the true interest cost of borrowing could not be used to measure the bond yields. Instead, the bond yields were measured by first identifying issues that had yield data on the last maturing bond; 184 out of 199 issues were identified using this criterion. (3) For the remaining issues, it appears that the bonds were issued at par value, and the yield was equal to the coupon rate on the last maturing bond.

Variables

Equation 1 below defines bond yield as a function of three major constructs: issue attributes, market conditions, and issuer traits and is expressed as:

Bond Yield = f(Issue attributes, Market conditions, Issuer traits) (1)

Issue attributes

The Issue maturity is related to the yield because a bond with a longer maturity is less liquid and is exposed longer to interest rate risk (Denison, 2001). Issue size relates to the ability to achieve economies of scale with issuance costs, which one expects to gain from larger bond issues since issuance costs have a substantial proportion of fixed costs. However, prior empirical findings show either a positive or negative coefficient for issue size, reflecting both economies and diseconomies of scale (Braswell et al., 1982; Kidwell et al., 1987). In addition, very large issues are sometimes difficult to market, forcing the issuer to raise the offered yield on the bond (Peng, 2002).

One would expect that the maturity of the issue would have a positive relationship with bond yield, depending upon the slope of the yield curve at the time of the issuance. If the yield curve is upward sloping, which compensates investors for the higher risk of longer maturity issues, one would expect longer maturity bonds to generate higher yields (Braswell et al., 1982; Hsueh & Liu, 1990b). Bonds with a Call option are expected to generate higher yields to compensate investors for the risk involved in a bond being called prior to maturity.

Market conditions

The Revenue Bond index for municipal bonds serves as a proxy for the market, and is composed of 20 fixed-rate municipal bonds maturing in 30 years. Measured at the time of the bond issuance, these bonds have an average Standard & Poor's rating of A+. The purpose of this variable is to control for prevailing market conditions. One would expect that during periods of high market interest rates the yield on a specific hospital issue will also be high.

The Moody's yield spreads, measured on a monthly basis, between lower quality Baa and higher quality Aaa bonds, may influence bond yields as well. Higher spreads may reflect a greater risk between lower and higher quality bonds. In addition, larger spreads may provide a greater incentive to purchase bond insurance to achieve the benefits of interest rate savings (Denison, 2001). The Variable vs. Fixed-rate spread may provide an incentive to issue a specific type of debt. Large spreads between variable and fixed-rate issues typically provide an incentive to issue variable-rate debt vs. fixed-rate debt. At the date of issuance, the study computed the spread between a fixed-rate index (measured by the revenue bond index) and a variable-rate index (measured by the seven-day, bond market index) for tax-exempt variable-rate demand obligations (Faulkender, 2005).

Issuer traits

To control for organizational size and local market effects, the study included a multi-hospital system variable. One would expect large, multi-hospital systems with geographically dispersed markets, to be viewed by the rating agencies as a lower risk than individual free-standing hospitals (Fitch Ratings, 2005). Since healthcare facilities compete on a local level, the study attempted to control for these market effects by including four geographic Region variables. The Southern region was treated as the omitted reference group.

To measure the default risk of uninsured bonds, the study included the Bond rating assigned by the rating agency. Hospitals or healthcare systems with higher bond ratings have a lower probability of default and are expected to incur lower bond yields. The study developed dummy variables for each of the rating categories, which are listed and defined below. Since these bonds are not insured, the assigned ratings reflect their underlying natural or intrinsic bond risk. Uninsured bonds assigned an underlying AA rating were considered the highest quality within the hospital and healthcare system industry.

To isolate the effect of Bond insurance on yields, the study developed two binary measures. The bond insurance AAA variable measures those insured bonds that were assigned an enhanced (insured) rating of AAA. Since 20 percent of the bond issues received an enhanced (insured) rating of AA, the study developed another binary variable to account for these issues.

Since the underlying bond risk of the issuer is considered the primary determinant of obtaining bond insurance, the study also evaluated the relationship of the underlying bond rating on the yield of the bond. Therefore, the study re-adjusted the bond rating to reflect the underlying risk. Twenty-five insured bonds disclosed their underlying bond rating. In contrast, 36 remaining insured issues released no underlying rating information to the market, while eight were reported as non-rated issues. Prior studies (Hsueh & Chandy, 1989; Peng, 2002) found that investors accounted for the underlying rating of insured bonds and demanded higher yields on AAA insured bonds with lower underlying ratings relative to uninsured or natural AAA-rated bonds. In addition, as previously noted by Peng (2002), insured bonds with a disclosed underlying rating can help investors assess the default risk of both the insurer and issuer and act as a mechanism to help separate high and low quality insured bonds. Following Peng's (2002) work, one would expect only issuers with higher underlying ratings to disclose this information to the market. However, this prior study also found that some issuers with higher underlying bond ratings did not release their underlying ratings to the public. Peng's (2002) explanation for the issuers' decisions not to report this information was fear that the market would still view them in the same light as bonds with lower underlying ratings.

Given that uninsured BBB-rated bonds had the largest number of observations for the non-insured group, the study used this group as the omitted reference group when analyzing the relationship between bond ratings and bond yields. Since insured bonds with an underlying AA-rating totaled only five issues, the study combined these issues with insured bonds having an underlying A-rating, and treated them as one variable.

Model

Two versions of the model depicted in Equation 4 above are evaluated. The first version assesses the relationship between bond yield and the insured issues without an underlying rating. The second version evaluates the association of bond yield and the underlying rating of the insured issues. An ordinary least squares regression model is estimated using the yield of the last maturing bond for an issue as the dependent variable. Several independent variables were included based on prior literature. The operational definition of each variable is as follows. Exhibit 2 presents descriptive statistics for the variables.

RESULTS

Insured Bonds without Underlying Ratings

Exhibit 3 presents the ordinary least squared (OLS) regression results of the first version of the model, focusing on the association between bond yield and issues with a bond insurance rating of either AAA or AA. For this model, the following variables were statistically significant at the p [less than or equal to] 0.05 level: issue size, maturity, revenue bond index, variable vs. fixed spread, AA-rated uninsured issues, A-rated uninsured issues, bond insured AAA-rated issues, and bond insured AA-rated issues. The study found that larger bond issues had a positive coefficient, indicating larger bond issues are difficult to market and offer a higher yield in order to attract investors. As expected, the maturity of the issue had positive coefficient indicating longer maturity issues are compensated with higher yield.

As expected, the fixed-rate revenue bond index had a positive coefficient with hospital bond yields. This finding indicates that when the long-term market index for fixed-rates increases by one percent, the yields on long-term, tax-exempt hospital bonds increases by 1.25 percent. The variable and fixed-rate spread variable had a negative coefficient suggesting that during market conditions of wider spreads, hospitals and health systems were issuing bonds at lower yields. During the period of 2002 through 2004, the average spread between variable-rate debt and fixed debt was 3.6 percent, but the market yields on the fixed-rate revenue bond index declined from 5.02 percent to 4.78 percent.

Given the omitted reference group of BBB-rated uninsured bonds, the yields of AA-rated uninsured bonds were 73 basis points lower, while A-rated uninsured bonds were 71 basis points lower. The non-rated uninsured bonds were insignificant, which suggests that these bonds had a similar yield to the reference group of uninsured BBB-rated bonds. Insured AAA-rated bonds had yields that were 109 basis points lower than the BBB-rated uninsured bonds; while insured AA-rated bonds were 74 basis points lower. This finding suggests that insured AAA-rated bond yields were lower than insured AA-rated issues. Finally, insured AA-rated bond yields were similar to AA-rated uninsured issues. (4)

Insured Bonds with Underlying Ratings

Exhibit 4 presents the OLS regression results with the underlying rating of the insured bonds. The bond traits and market conditions had the same significant variables as the Exhibit 3 model. All the rated and non-rated issues, except non-rated uninsured bonds, were statistically significant. Compared to uninsured BBB-rated bonds, the reference group, AA-rated uninsured bonds and A-rated uninsured bonds were statistically significant and had yields that were more than 70 basis points lower. Insured bonds with an underlying rating of AA and A were 101 basis points lower than the omitted uninsured BBB-rated bonds. Insured bonds with underlying rating of BBB were 65 basis points lower. Finally, insured bonds reporting no rating or not-rated were 108 basis points lower than the reference BBB group.

DISCUSSION

Protecting against default, achieving a higher bond rating, lowering interest costs, and enhancing marketability are considered several of the underlying reasons why hospitals and healthcare systems purchase bond insurance. Two prior studies (Dension, 2003; Peng, 2002) tested two corporate finance theories behind the demand for bond insurance--market segmentation theory and information asymmetry--by analyzing municipal bond issues in general. The focus of the present study was to test these theories within the context of tax-exempt hospital and healthcare system bond issues.

In evaluating 69 insured bonds, the study found that as a result of this enhancement, more than 80 percent of these bonds were AAA-rated, while the remaining issues received an AA-rating. The findings of this study support the market segmentation theory since AAA-rated insured bonds had the lowest yield relative to uninsured BBB-rated bonds. As previously noted, market segmentation occurs within the tax-exempt debt market because there is a strong demand among institutional investors for higher quality debt. Given the inverse relationship between bond prices and yield, the lower yields for these AAA-rated insured bonds suggests that the institutional investors may have been competing to buy these AAA-rated bonds.

Although insured AAA tax-exempt hospital and healthcare system bonds have no direct comparison to uninsured AAA-rated bonds, the study was able to compare the yields of AA-rated insured bonds and AA-rated uninsured bonds to the reference group of BBB uninsured bonds. In the case of both groups, the study found similar yields, indicating that the market views the default risk of these two groups to be similar. When evaluating the underlying ratings of the insured AA-rated bonds, nine of the issues were either non-rated or reported no rating, while five issues had BBB underlying ratings. Since bond insurers limit their underwriting standards to issues with at least a BBB-rating, these non-rated issues were viewed by the market as having an underlying rating of at least BBB (Peng, 2002). Thus, equivalence of yield between the two types of bonds indicates that bond insurance attracts a high demand for the bonds even though the underlying rating is investment grade for some of these issues.

Overall, these findings underscore the benefits of bond insurance since it helps reduce investors' concerns related to the underlying default risk of the individual issues. A hospital's default risk can be affected by increasing competition, rising managed care penetration, falling local demand for healthcare services, and other factors influencing the facility's business and financial risk. Managers of hospitals, as well as their boards of directors, know that it is difficult to control for uncertainties within the healthcare market place. More importantly, they also recognize that they are competing with lower default risk tax-exempt bond issues backed by the taxing power of a governmental entity (e.g. state and local government bonds) or entities that are natural monopolies facing fewer competitors (e.g. municipal bonds of public utilities, water and sewer plants, etc.). As a result, such governmental tax-exempt bond issuers may be able to achieve a AAA-rating based on their own financial merits without the purchase of bond insurance (Fitch, 2007). Conversely, the findings of this study demonstrated the purchase of bond insurance enabled tax-exempt hospital bond issuers to achieve a lower cost of debt because investors are effectively purchasing the default risk of the insurer and not the issuer, as well as enhancing the liquidity and market value of the bond.

Information asymmetry theory also is supported by the findings of this study. Lower yields among the insured bonds may reflect the enhanced marketability of the bonds attributed to the national reputation of the bond insurer. The study findings support this theory since insured bonds with underlying ratings of AA and A had lower yields than uninsured bonds with similar ratings. With the underlying ratings of the two groups of bonds being similar, bond insurance helps alleviate the risks associated with lack of recognition on a national basis for local hospitals and healthcare systems, and further validates their bond quality.

Finally, insured bonds with no reported underlying rating had yields nearly equivalent to insured bonds with an A and AA underlying rating. An explanation for this finding may relate to Peng's (2002) contention that sometimes insured issuers may have an underlying rating but see no added value in disclosing the rating to the market. The insured issuers did not release this information to the market for fear that the bond markets might group them with issues that had lower underlying ratings and higher yields.

Limitations

This study had several limitations. The first limitation was the small sample size of insured bonds with AA underlying ratings. This included only five issues and they were combined with insured bonds with an A underlying rating. The second limitation was the lack of data on the cost of bond insurance. The yield data of this study did not adjust for the cost of bond insurance, which typically varies by bond quality, issue size, and market conditions. Collecting information on this upfront insurance cost would have allowed the measurement of the true cost of borrowing and would have provided a more valid assessment of the relationship between bond insurance and yield. The third limitation was restricting the sample to only fixed-rate bond issues. The inability to identify yield information from the SDC database precluded the inclusion of variable-rate securities. The final limitation was limiting the analysis to the time periods of 2000 through 2005. The author did not have access to data beyond this time period. However, future studies should focus on the municipal bond insurers' credit crisis commencing in 2007, which occurred from their losses on insuring high risk mortgage backed securities and other factors (Cho, 2007). These losses among bond insurers, coupled with the possibility of losing their AAA-ratings, may reduce the access to bond insurance for hospitals. As a result, hospital bond issuers may face higher yields due to higher credit risk and decreased marketability.

REFERENCES

Arrick, M.D., Zuckerman, L., & Infranco, S. (2006). U.S. not-for-profit health care 2006 outlook: Emerging evidence of a turning tide. New York, NY: Standard & Poor's.

Below, S. (1994). The existence of municipal bond insurance: Theory and evidence. As cited in Denison, D.V. (2003). An empirical examination of the determinants of insured municipal issues. Public Budgeting & Finance, 23: 96-114.

Braswell, R.C., Nosari, E.J., & Browning, M.A. (1982). The effect of private municipal bond insurance on the cost of the issuer. The Financial Review, 17: 240-251.

Carpenter, C.E. (1991). The marginal effect of bond insurance on hospital, tax-exempt bond yields. Inquiry, 28: 67-73.

Carpenter, C.E., McCue, M.J., & Moon, S. (2003). The hospital bond market and the AHERF bankruptcy. Journal of Health Care Finance, (Summer) 29(4): 17-28.

Cho, D. (2007). Municipal bond deals squeezed by credit crisis. Washington Post, (Nov. 29): A1.

Denison, D.V. (2001). Bond insurance utilization and yield spreads in the municipal bond market. Public Finance Review, 29: 394-411.

Denison, D.V. (2003). An empirical examination of the determinants of insured municipal issues. Public Budgeting & Finance, 23(1): 96-114.

Diamond, D.W. (1984). Financial intermediation and delegated monitoring. Review of Economic Studies, 51: 393-414.

Faulkender, M. (2005). Hedging or market timing? Selecting the interest rate exposure of corporate debt. The Journal of Finance, 60(2): 931-962.

Fitch Ratings. (2005). Investment and Debt Portfolio Trends of Hospitals and Health Care Systems--1995-2003. New York, NY: Fitch, Inc.

Fitch Ratings. (2006). 2006 Nonprofit Hospitals and Health Care Systems Outlook. New York, NY: Fitch, Inc.

Fitch Ratings. (2007). Default Risk and Recovery Rates on U.S. Municipal Bonds. New York, NY: Fitch, Inc.

Haugh, R. (2006). Bond Protection. Hospitals & Health Networks, 80(3): 32-33.

Health Care Investment Analysts (HCIA). (2000). The Comparative Performance of U.S. Hospitals: The Sourcebook. Baltimore, MD: Deloitte & Touche.

Healthcare Financial Management Association (HFMA). (2004). Financing the Future, How are the Hospitals Financing the Future? The Future of Capital Access? Westchester, IL.

Hsueh, L.P., & Chandy P.R. (1989). An examination of the yield spread between insured and uninsured debt. The Journal of Financial Research, 12: 235-244.

Hsueh, L.P., & Liu, Y.A. (1990a). An examination of the biases in estimating the benefit of debt insurance. The Financial Review, 25: 473-486.

Hsueh, L.P., & Liu, Y.A. (1990b). The effectiveness of debt insurance as a valid signal of bond quality. Journal of Risk and Insurance, 57: 691-700.

Jaklevic, M.C. (2004). Back in the game; AAA bond insurer returns to healthcare. Modern Healthcare, (July 26): 16.

Kidwell, D.S., Sorensen, E.H., & Wachowicz, J.M. (1987). Estimating the signaling benefits of debt insurance: the case of municipal bonds. The Journal of Financial & Quantitative Analysis, 22: 299-313.

McCue, M.J. (1997). Association of HMO penetration and other bond quality factors with tax-exempt bond yields. Inquiry, 34(3): 217-227.

Moody's. (2005). Not-for-profit Healthcare: 2005 Outlook and Medians, Moody's Investor Service New York, NY.

Peng, J. (2002). Do investors look beyond insured triple-A rating? An analysis of Standard & Poor's Underlying Ratings. Public Budgeting & Finance, 22(2): 115-131.

Thakor, A.V. (1982). An exploration of competitive signaling equilibria with 'third party' information production: The case of debt insurance. Journal of Finance, 37: 717-739.

The Bond Buyer (2005). 2004: Looking back: Municipal bond insurance hits 54% record in 2004. The Bond Buyer, 351(32,053): 7.

Wareham T.L., & Majka, A.J. (2003). Best practice financing. Kaufman Hall White Paper, Northfield, IL: Kaufman Hall & Associates.

(1) However, Standard & Poors' rating agency claims that these higher capital expenditures are occurring with providers with strong credit quality. Standard & Poors' data indicate in 2005 that providers with bonds rated at investment grade of BBB- had capital expenditure to depreciation expense ratios of 100 percent, compared to 150 percent for providers with A+ rated bonds (Arrick et al., 2006).

(2) Out of the 69 insured issues, 78 percent were rated by more than one credit rating agency and only four had split ratings.

(3) Because the SDC database does not include the cost of bond insurance, the study is unable to develop a valid measure of bond yield that accounts for this cost. The yield data from this study does account for gross spread, which includes the underwriting fee, takedown, management fee and other expenses. By excluding the cost of bond insurance, the yield measure for insured bonds underestimates the true bond yield. However, a consultant for a national healthcare capital advisory firm indicated that the yield measure used in this study would reflect the effects of bond insurance through lower underwriting costs. Specifically, the yield on insured bonds would be lower because bond insurance would increase marketability, which in turn, would lower underwriting costs and therefore decrease yield.

(4) By eliminating insured bond issues that were either non-rated or non-reporting, the sample decreased to 155 issues. The findings show AA-rated uninsured issues with a significant coefficient of -0.723; A-rated uninsured issues with a significant coefficient of -0.671; Non-rated uninsured issues with a significant coefficient of -0.237; Insured AAA-rated issues with significant coefficient of -0.98, and Insured AA-rated issues with a marginally significant, (p [less than or equal to] 0.10 level) coefficient of -0.62.

Address for correspondence: Michael J. McCue, Department of Health Administration, Virginia Commonwealth University, PO Box 980203, Richmond, VA 23298-0203 USA, [e-mail].

Michael J. McCue

Virginia Commonwealth University

EXHIBIT 1
ISSUE SAMPLE SIZE BY
INSURED VS. NON-INSURED AND BY
DISCLOSED UNDERLYING RATING

Total Sample Size                                  199
Insured                                             69
Non-insured                                        130
  Total                                            199
Insured, with Disclosed Underlying Rating           25
Insured, with No Rating                             44
  Total Insured                                     69
Non-Insured, with Underlying Rating                104
Non-Insured, with No Rating                         26
  Total Non-Insured                                130
Underlying Ratings of Insured Issues
  AA                                                 5
  A                                                 12
  BBB                                                8
  Total Underlying Ratings of Insured Issues        25
Underlying Ratings of Non-insured Issues
  AA                                                12
  A                                                 38
  BBB                                               54
  Total Underlying Ratings of Non-insured Issues   104

Dependent Variable:
Bond Yield               = Yield on last maturing bond in an issue

Independent Variables:
Issue Attributes:
Issue size               = Total dollar size of the bond issue
Issue maturity           = Maturity date of the bond issue
Call option              = 1 for bonds with a callable option, 0
                           otherwise

Market Conditions:
Revenue Bond index       = Fixed-rate index for revenue bonds measured
                           on the date of issuance
Moody's spread Aaa       = Spread between lower quality Baa and higher
vs. Baa                    quality Aaa bonds, measured monthly
Variable vs.             = Spread between variable-rate and fixed-rate
Fixed-rate spread          revenue bonds

Issuer Traits:
Issued by System         = 1 if bond issued by healthcare system, 0 if
                           issued by a free-standing hospital
Northeast region         = 1 if issuer is located in CT, ME, MA, NY,
                           or NH, 0 otherwise
Mid-Atlantic region      = 1 if issuer is located in DE, MD, PA, NJ,
                           VA, or WV, 0 otherwise
Mid-West region          = 1 if issuer is located in IA, IL, IN, KS,
                           MI, MN, MO, NE, OH, SD, or WI, 0 otherwise
South region             = 1 if issuer is located in AL AR, FL, GA,
                           KY, LA, MS, NC, OK, or TX, 0 otherwise
                           (omitted reference group)
West region              = 1 if issuer is located in CA, CO, HI ID,
                           MT, NM, NV, OR, WA, or WY, 0 otherwise
AA-rated uninsured       = 1 if bond issue is AA-rated and uninsured,
                           0 otherwise
A-rated uninsured        = 1 if bond issue is A-rated and uninsured,
                           0 otherwise
BBB-rated uninsured      = bond issue is BBB-rated and uninsured
                           (omitted reference group)
Non-rated uninsured      = 1 if bond issue is uninsured and no reported
                           rating or non-rated, 0 otherwise
Insured underlying       = 1 if bond issue is insured with underlying
AA or A-rated              AA or A-rating, 0 otherwise
Insured underlying       = 1 if bond issue is insured with underlying
BBB-rated                  BBB-rating, 0 otherwise
Insured underlying       = 1 if bond issue is insured with no
Non-rated                  underlying rating or non-rated, 0 otherwise
Bond insured AAA-        = 1 if bond issue is insured with AAA insured
rated                      rating, 0 otherwise
Bond insured AA-         = 1 if bond issue is insured with AA insured
rated                      rating, 0 otherwise

EXHIBIT 2
DESCRIPTIVE STATISTICS

Variable                   Mean           S.D.

Bond Yield                 5.64%          1.05%
Issue size               $68.5 million   80.01 million
Issue maturity            24.7 yrs.       7.3 yrs.
Call option                0.924          0.264
Revenue Bond index         4.950%         0.421%
Moody's spread Aaa         0.983%         0.234%
vs. Baa
Variable vs. Fixed-        2.908%         1.031%
rate spread
Issued by System           0.376          0.485
Northeast region           0.160          0.368
Mid-Atlantic region        0.180          0.385
Mid-West region            0.256          0.437
South region               0.249          0.411
West region                0.155          0.363
AA-rated uninsured         0.060          0.238
A-rated uninsured          0.191          0.394
BBB-rated uninsured        0.271          0.445
Non-rated uninsured        0.130          0.337
Insured underlying AA      0.085          0.280
or A-rated
Insured underlying         0.040          0.196
BBB-rated
Insured underlying         0.221          0.416
Non-rated
Bond insured AAA-rated     0.276          0.448
Bond insured AA-rated      0.070          0.256

EXHIBIT 3
OLS REGRESSION
BOND INSURANCE
(Dep. Var. = Bond Yield)

Independent           Expectation     Coef   Std. Error     p-Value
Variable

Constant                            -0.507        0.730   0.494
Issue size            +/-            0.002        0.007   0.004 ***
Issue maturity        +              0.018        0.008   0.021 **
Call feature          +              0.292        0.212   0.171
Revenue Bond index    +              1.25         0.128   0.001 ***
Moody's spread Aaa    C              0.164        0.267   0.541
vs. Baa
Variable vs. Fixed    C             -0.171        0.058   0.004 ***
Spread
Issued by System      -             -0.137        0.113   0.228
Northeast region      C              0.196        0.165   0.235
Middle Atlantic       C              0.003        0.162   0.984
region
Mid-West region       C             -0.039        0.152   0.798
West region           C             -0.200        0.174   0.254
AA-rated uninsured    -             -0.737        0.233   0.002 ***
A-rated uninsured     +             -0.712        0.154   0.001 ***
Non-rated uninsured   +             -0.157        0.190   0.411
Bond insured AAA-     -             -1.09         0.142   0.001 ***
rated
Bond insured AA-      -             -0.742        0.221   0.001 ***
rated

Note: C = control variable.
Adjusted [R.sup.2] = 55%
*** significant at the p [less than or equal to] 0.01 level
** significant at the p [less than or equal to] 0.05 level
* significant at the p [less than or equal to] 0.10 level

EXHIBIT 4
OLS REGRESSION
INSURED UNDERLYING RATINGS
(Dep. Var. = Bond Yield)

Independent Variables    Expec-   Coef     Std. Error   p-Value
                         tation
Constant                          -0.817   0.748        0.276
Issue size               + /-      0.002   0.007        0.005 ***
Issue maturity           +         0.019   0.008        0.016 **
Call feature             +         0.257   0.213        0.230
Revenue Bond index       +         1.30    0.132        0.001 ***
Moody's spread Aaa       C         0.269   0.266        0.314
  vs. Baa
Variable vs. Fixed       C        -0.185   0.058        0.002 ***
  Spread
Issued by System         -        -0.138   0.113        0.226
Northeast region         C         0.202   0.165        0.224
Middle Atlantic region   C         0.018   0.163        0.912
Mid-West region          C        -0.024   0.152        0.873
West region              C        -0.202   0.165        0.276
AA-rated uninsured       -        -0.724   0.234        0.002 ***
A-rated uninsured        +        -0.719   0.155        0.001 ***
Non-rated or not         +        -0.162   0.191        0.396
  reported uninsured
Insured underlying AA    -        -1.01    0.204        0.001 ***
  or A-rated
Insured underlying       +        -0.662   0.279        0.019 **
  BBB-rated
Insured underlying       +        -1.086   0.150        0.001 ***
  Non-rated
or not reported

Note: C = control variable.

Adjusted RZ = 55%

***significant at the p [less than or equal to] 0.01 level

** significant at the p [less than or equal to] 0.05 level

* significant at the p [less than or equal to] 0.10 level