|
Article Excerpt
Using a vector autoregressive model with monthly data from 1988 through 2001, this study investigates the factors that drive the excess returns on a widely followed mortgage-backed securities (MBS) index. We find that eight important economic variables (industrial productions, new home sales, bond horizon premium, bond quality premium, mortgage rate, refinancing proxy, general stock market index and world bond market index) appear to move the excess returns on MBS. Impulse response analysis and variance decomposition further indicate a strong dynamic relationship between MBS excess returns and changes in these economic variables. Additional analysis of Freddie Mac and Fannie Mae MBS also indicates that the risk of the MBS guarantor is an important determinant of the MBS return dynamics after the creation of the Office of Federal Housing Enterprise Oversight.
**********
The mortgage debt market has become an increasingly important component of the U.S. capital market in the past two decades. Mortgage-backed securities (MBS) in particular, which are created through securitization of mortgage loans made by financial institutions such as commercial banks, savings and loans and mortgage companies, have come to dominate the mortgage debt markets in recent years.
The MBS market has grown at a much faster pace than the overall mortgage market because an increasing proportion of mortgage originations are now securitized. As of 2001, the value of U.S. MBS outstanding amounts to about $3.7 trillion, about half the size of the total U.S. mortgage debt market and equivalent to 36% of U.S. gross domestic product. Advances in financial engineering and structured finance techniques, increased availability of consumer credit information and standardization and credit support from government agencies are credited for the success of the U.S. MBS market. Table 1 and Figures 1 and 2 illustrate the dramatic development of the MBS market over the past 25 years.
This study uses a vector autoregressive (VAR) model to examine the economic factors that are important for the excess returns on a widely tracked and followed MBS index (i.e., the total return on MBS index minus the 3-month risk-free Treasury bill rate). This is a new approach applied to the MBS research, although the VAR model has been used in the literature related to traditional financial and real estate market analysis (see, e.g., Hasbrouck 1991, Lee 1992, Liu and Mei 1992, Campbell and Ammer 1993).
[FIGURE 1 OMITTED]
Most MBS are in the form of mortgage pass-through securities. A mortgage pass-through security is created when mortgages with similar loan types, coupons and maturities are pooled and when participations in the pool are sold. (1)
[FIGURE 2 OMITTED]
The great majority of mortgage pass-through securities have been issued or guaranteed by three government agencies created by the U.S. Congress to enhance liquidity in the secondary mortgage markets. The three agencies are commonly called Ginnie Mae (Government National Mortgage Association, GNMA), Fannie Mae (Federal National Mortgage Association, FNMA) and Freddie Mac (Federal Home Loan Mortgage Corporation, FHLMC). Ginnie Mae is a government agency within the Department of Housing and Urban Development (HUD), whose guarantee assumes the full faith and credit of the U.S. government. Fannie Mae and Freddie Mac are New York Stock
Exchange-listed "government-sponsored enterprises" (GSEs), which do not represent the explicit guarantee of the U.S. government but are believed by market participants to convey the government's implicit guarantee. The explicit (for Ginnie Mae MBS) or implicit (for Fannie Mae or Freddie Mac MBS) guarantee of the three federal agencies has made investments in MBS more attractive, as credit risk is either eliminated (for Ginnie Mae MBS) or substantially reduced (for Fannie Mac or Freddie Mac MBS). The Office of Housing Enterprise Oversight (OFHEO) was established in 1992 to "promote the housing sector and a strong economy by ensuring the safety and soundness of Fannie Mae and Freddie Mac and fostering the vitality of the nation's housing finance system." (2) How the risk of the guarantor affects MBS return dynamics, especially after the creation of OFHEO, should be of interest to investors and policy makers.
A critical aspect of the MBS is the exposure of investors to prepayment risk--the premature or unscheduled payment of principal to investors when homeowners refinance, relocate or default. Refinancing typically occurs when the market mortgage rate falls far enough below the weighted average coupon rate (WAC) on the securitized mortgage pool and when homeowners can reduce their monthly mortgage payments significantly. Refinancing, triggered by a drop in mortgage rates, is an undesirable event for MBS investors who must reinvest the proceeds at lower market rates. The amount and the timing of cash flows received from a mortgage pass-through are largely affected by the prepayment of the mortgage pool. Because the timing and speed of repayment may vary, cash flows received by MBS investors are highly irregular.
A change in mortgage rate affects the MBS returns through both the discount-rate effect (i.e., the prevailing mortgage rate and the present value of the mortgage cash flows are negatively related) and the prepayment effect. This study explicitly examines how mortgage rates affect the MBS returns to shed light on both the discount-rate effect and the prepayment effect. Consistent with expectations, our analysis documents negative discount-rate effect of mortgage interest rates and positive prepayment effect.
One strand of MBS research focuses primarily on the theoretical valuation and pricing issues of MBS that relate to prepayment risk resulting from interest rate movements (Dunn and McConnell 1981, Schwartz and Torous 1989, 1992 and Stanton 1995). Bennett, Peach and Peristiani (2001) analyze the impact of structural changes in the mortgage market on homeowners' prepayment patterns. Other studies examine the impact of MBS issuance on the yield or interest rate of the mortgage debt, demonstrating the relevance and contribution of MBS in the financial market (Black, Garbade and Silber 1981, Kolari, Fraser and Anari 1998).
We conduct an empirical analysis of the key factors that affect MBS returns, on which limited academic research is available. Although pricing and yield questions have been widely researched, investors often find yield-to-maturity (YTM) an unsatisfactory indicator of total return on the MBS investments. This is because YTM assumes that the security is held until maturity and that all the cash flows received prior to maturity are reinvested at the YTM.
We use total returns, instead of YTM, in our analysis to avoid the confounding problems of reinvestment and prepayment. Total return, also called the holding-period return, measures what investors can earn from a security over a specified holding period. It is the most commonly used measure of return for securities in general (e.g., stocks, bonds). Total return on mortgage-backed securities includes price return, coupon return and paydown return. (3)
Our results are important for several reasons. First, despite the economic importance of the MBS market, very little attention has been paid to empirical economic factors underlying returns. Although fundamental economic variables have been identified for excess returns on stocks and bonds (Chen, Roll and Ross 1986, Campbell and Ammer 1993, Elton, Gruber and Blake 1995), little is known for the type of economic variables relevant for MBS excess returns.
Specific economic factors related to MBS are examined and included in our analysis as follows. The real activity has been shown to affect stock returns in the finance literature (Lee 1992). Similarly, it should also be expected to affect the MBS returns. We use two variables--industrial productions and new home sales--to proxy for the real activity. (4) That is, as the economy expands, these two variables will show signs of growth in real activity for the overall economy and the housing sector. The MBS returns are expected to be lower during economic expansion because higher economic growth often leads to a higher return in real sector investments and a higher real interest rate.
The MBS, which are risky debt-related instruments, are expected to be sensitive to credit, liquidity and call risk premiums in the fixed income market. Thus, the bond quality premium proxy (BQP, the difference in total returns between long-term AAA corporate bonds and long-term Treasury bonds) for credit, liquidity and call risk premiums should be positively related to MBS returns. Similarly, shifts in interest rate yield curve may affect MBS returns given that MBS are long-term fixed-income securities. As a result, we use the change in bond horizon premiums (BHP, difference in total returns between long-term Treasury bonds and 30-day Treasury bills) to capture the term structure risk.
As argued earlier, MBS returns are subject to the discount-rate effect and prepayment effect as a result of interest rate movements. To capture the discount-rate effects, we use the change in the 30-year mortgage rate. To capture the prepayment effect, we use a variable that reflects the refinancing effect when the prevailing mortgage rate is lower than the weighted average coupon rate of the MBS mortgage pool. As a result, we are able to separate these two confounding effects caused by interest rate volatility.
Several studies on MBS have demonstrated that MBS are well integrated into the capital market (Hendershott and Van Order 1989, Devaney, Pickerill and Krause 1992, Goebel and Ma 1993), but the relationship of MBS with the stock market is still not well researched. We provide a useful piece of information on how investors select MBS over stocks as an alternative asset class in their investment portfolio. The analysis provides information helpful to government-sponsored agencies (such as FHLMC and FNMA) for improving the working of the secondary mortgage markets and the risk management of their mortgage security portfolios. To this end, we include stock market return in our model, whereby we view MBS as part of the investors' investment portfolio.
We also analyze the impact of the world bond market index on MBS returns. Barr and Priestley (2004) and Ilmanen (1995) indicate that the world bond market affects domestic bond market returns. As the MBS market is an important investment vehicle for both domestic and foreign investors, it is worthwhile to investigate to what extent the MBS returns are affected by the world bond market risk.
In our model, we identify eight economic (real and financial) variables that are shown to relate significantly with MBS excess returns. They are growth in industrial productions and new home sales, which are used to capture the real activity of the economy and the housing sector; changes in bond quality premium (differences in total returns between long-term AAA corporate bonds and long-term Treasury bonds) to capture credit, liquidity and call risks; bond horizon premium to capture the term structure risk; change in mortgage rate to shed light on the discount rate effect; refinancing variable to reflect the prepayment effect; stock market excess returns to analyze the substitution effect of the equity market on MBS; and world bond market excess returns to examine the impact of common world market risk factor.
In our analysis, we also try to analyze two other related issues. First, we investigate whether there is a seasonality pattern in the MBS excess returns by incorporating an exogenous variable for the summer period (i.e., May, June and July) in the VAR model. (5)
Second, we apply a similar VAR model to analyze mortgage REIT returns to see if the MBS and Mortgage REITs are subject to...
|
