|
Article Excerpt
This article examines the cross-sectional and time-series determinants of commercial mortgage credit spreads as well as the terms of the mortgages. Consistent with theory, our empirical evidence indicates that mortgages on property types that tend to be riskier and have greater investment flexibility exhibit higher spreads. The relationship between the loan-to-value (LTV) ratio and spreads is relatively weak, which is probably due to the endogeneity of the LTV choice. However, the average LTV ratio per lender has a strong positive relation with credit spreads, which is consistent with the idea that lenders specialize in mortgages with either high or low levels of risk, and that high LTV mortgages require substantially higher spreads. Finally, we observe that spreads widen and mortgage terms become stricter after periods of poor performance of the real estate markets and after periods of greater default rates of outstanding real estate loans.
**********
Commercial mortgages provide perhaps the best setting for examining default spreads in the fixed income market. In most cases, commercial properties have only one outstanding loan, the loans generally are not prepayable without substantial penalties and assets that are relatively easy to evaluate collateralize the loans. There is currently more than a trillion dollars of commercial mortgages outstanding and the market is growing, both in the United States and around the world.
This article empirically examines the determinants of credit spreads for commercial mortgages, that is, differences between mortgage rates and Treasury bond rates with the same maturities. Using a data set of 26,000 individual commercial mortgages that were originated between 1992 and 2002, with the intent of being included in a commercial mortgage-backed security (CMBS), (1) we examine cross-sectional differences in mortgage spreads, as well as time-series fluctuations in average spreads.
Our cross-sectional tests are motivated by theoretical pricing models developed by Titman and Torous (1989), Kau et al. (1990) and Titman, Tompaidis and Tsyplakov (2004). The earlier articles present models that indicate that mortgages on properties that are more volatile and that have higher payouts tend to have higher spreads. The more recent Titman, Tompaidis and Tsyplakov (2004) model shows that mortgages on properties with more investment flexibility, that is, properties that can be expanded or renovated, should also have higher spreads. (2,3)
Our empirical results are largely consistent with these theoretical predictions. In particular, properties like hotels, which are likely to be both riskier and have the greatest investment flexibility, have significantly higher spreads than warehouses and multifamily housing, which are likely to be less risky and have less investment flexibility. In addition, credit spreads are positively related to the ratio of net operating income to property value (NOI/Value), which is also consistent with the models if we assume that a higher NOI/Value ratio is indicative of higher payouts.
The observed evidence on the relation between mortgage characteristics and spreads is somewhat less straightforward to interpret. Most notably, the loan-to-value (LTV) ratio of a mortgage is expected to be positively related to mortgage spreads, but our evidence on this is mixed. Similarly, we expect from theory that mortgage maturity should be positively related to mortgage spreads, but we empirically find the opposite. These violations of the theoretical predictions are likely to be due to the endogenous choice of mortgage characteristics. Specifically, lenders are likely to require mortgages with higher downpayments, that is, lower LTV ratios and shorter maturities on properties that are likely to be riskier. (4)
To learn more about the endogeneity of the mortgage contract we examine the choices of individual originators. Our results indicate that different originators have different risk preferences; some originators attract riskier clienteles, attracting mortgages with higher LTV ratios as well as mortgages on properties that are riskier. Our analysis suggests that the above-mentioned endogeneity problem is not nearly as severe when we examine average LTV ratios and average spreads across originators. Specifically, we find that the average LTV of the mortgages provided by originators is very strongly related to the spreads on those mortgages, which is consistent with the idea that spreads are strongly influenced by LTV ratios.
We also study the determinants of mortgage characteristics, such as the LTV ratio, the mortgage amortization rate and mortgage maturity. Our results indicate that an important determinant of the LTV ratio and the amortization rate is the NOI/Value ratio. We find that properties with higher NOI/Value ratios have mortgages with higher LTV ratios and higher amortization rates. One explanation for the first observation is that a higher NOI/Value ratio permits the borrower to satisfy debt coverage ratios with mortgages with higher LTV ratios. The higher amortization rate can be explained by the fact that properties with higher NOI/Value ratios are likely to experience less income growth and may be riskier. In addition, we find that relatively safe property types, such as multifamily apartment complexes and anchored retail properties, have higher LTV ratios and lower amortization rates, while riskier properties, such as limited- and full-service hotels, have lower LTV ratios and higher amortization rates.
In addition to our cross-sectional analysis we examine the time-series variation in spreads and mortgage characteristics. Consistent with the analysis in Titman and Torous (1989) we find that mortgage spreads decrease with increases in Treasury bond rates. Moreover, our results indicate that not only do higher interest rates lead to lower spreads, but average LTV ratios decline as well, possibly due to the higher interest payments or to binding debt coverage ratios. We also find that spreads increase following periods when real estate markets perform poorly, which is consistent with the idea that the supply of mortgage capital declines when the financial institutions that provide the mortgages are financially weaker.
Our analysis is closely related to earlier work of Maris and Segal (2002) and Nothaft and Freund (2003) who studied the credit spreads of entire CMBS deals rather than individual commercial mortgages. Similar to our results, they find that CMBS spreads are affected by macroeconomic factors. In particular, Maris and Segal (2002) show that competitive pressure during the 1994-1997 period lowered underwriting standards, while the 1998 Russian default crisis weakened the commercial real estate lending market, leading to higher spreads. Consistent with our findings, Nothaft and Freund (2003) find that average LTV ratios and average term to maturity are not statistically significant determinants of credit spreads, while spreads are negatively related to commercial property appreciation rates. However, in contrast to our results, Maris and Segal (2002) and Nothaft and Freund (2003) find a significantly positive relationship between CMBS spreads and corporate spreads (Nothaft and Freund (2003) studied the A-AAA spread, while Maris and Segal (2002) studied the AAA-risk-free Treasury rate spread). Instead, we find a marginally significant negative relationship between commercial mortgage spreads and the AAA-BBB spread.
The results in this article can also be compared to the recent literature on credit spreads on corporate bonds. For example, Collin-Dufresne, Goldstein and Martin (2001) examine empirically the determinants of changes in credit spreads of corporate bonds. Similar to our work, they investigate the impact of both firm and loan characteristics, as well as macroeconomic variables. In contrast to our findings, they find that changes in leverage are statistically significant determinants of changes in corporate credit spreads. However, the economic significance of leverage is small. Similar to our results, they find that an increase in the risk-free Treasury rate is negatively related to corporate credit spreads. They also find that increases in the average level of market volatility increase corporate credit spreads, and that S & P 500 returns are significantly negatively correlated with corporate credit spreads. While they are able to account for only 25% of the variation of corporate credit spreads, they determine that 75% of the variation of the residuals can be explained by a single factor, which appears to be unrelated to firm characteristics.
The remainder of the article is organized as follows. In the next section we describe the data set. The third section introduces the explanatory variables and discusses the cross-sectional determinants and time-series determinants of spreads of commercial mortgages. It also offers evidence of clientele effects. The fourth section discusses the cross-sectional and time-series determinants of mortgage characteristics such as the LTV ratio, amortization rate and mortgage maturity. The final section summarizes the article and discusses directions for future research.
Data Overview
Our data set, which was provided by Standard & Poor's, includes information on over 26,000 commercial mortgages. The mortgages originated between 1992 and 2002 with most of the originations taking place in the mid to late 1990s. The mortgages were later pooled and used as collateral for CMBS. All the mortgages in our sample were issued specifically for inclusion in a CMBS and are referred to as conduit deals. (5) The value of the commercial properties collateralizing the mortgages varies from $60,000 to $725,000,000, and the aggregate value is approximately $250 billion. The mortgages were originated by more than 130 commercial banks, investment banks, insurance companies and financing arms of large companies. The data set includes detailed information on cross-sectional characteristics of individual properties and their mortgage contract specifications.
The property types in the data set include multifamily apartment complexes, unanchored retail, anchored retail, medical offices, industrial, warehouse, mobile home parks, office buildings, properties of mixed use, limited-service hotels, full-service hotels and self-storage. The most common type is multifamily apartment complexes, which represent 34% of the total number of properties. More than a third of the mortgaged properties in the data set are located in California, Texas and Florida.
Summary statistics are presented in Table 1.
Mortgage Characteristics
The data include the following financial information for individual mortgages: mortgage rate; LTV ratio; origination date; whether the mortgage is balloon, amortizing or semiamortizing; whether the mortgage rate is fixed or adjustable and the maturity of the...
|
