Mortgage termination behaviour and alternative mortgage instruments

Previous chapters noted that economic behaviour might differ according to the type of mortgage instrument. This involves aspects of both signalling and selectivity. For example, it was suggested that households choosing an ARM might be more mobile and more inclined to default. Prepayment behaviour was generally considered for samples of fixed rate mortgage holders. However, the presence of interest rate floors and caps and the use of 'teaser rates' has meant that adjustable rate mortgages also have option theoretic features. Empirical research into mortgage default and prepayment has increasingly recognised the importance of the choice of mortgage instrument (Cunningham & Capone 1990; Capone & Cunningham 1992; Phillips et al. 1996; Vanderhoff 1996; Green & Shilling 1997; Ambrose & LaCour Little 2001; Calhoun & Deng 2002).

The essence of the ARM is that interest rate risk is passed on to the borrower. However, the default risk that lenders then face may have a significant impact upon their cash flows. This risk of default might arise out of the characteristics of the borrower (e.g. credit worthiness), or from the specific features of the mortgage instrument (e.g. periodic caps and their adjustment frequencies). Determining the relative contribution of these factors is important for the lender's cash and risk management, and for MBS valuation. Prepayment is also an issue insofar as the cap and adjustment terms establish boundary conditions that create a call option on the adjustable rate mortgage debt. There is evidence that the rates of prepayment on ARM mortgages in the US during the 1990s, have significantly exceeded the rates for FRM borrowers (Ambrose & LaCour Little 2001).

The methods and development of research in this area reflects mortgage termination work in general. For example, there is research covering aggregate ARM termination from mortgage pools (Huang & Xia 1996; Sanyal 1994). There are also studies using loan level data focusing upon prepayment (Lea & Zorn 1986; Zorn & Lea 1989; Cunningham & Capone 1990; Phillips et al. 1996; Vanderhoff 1996; Green & Shilling 1997); or default (SA-Aadu 1988; Cunningham & Capone 1990). Recent work has adopted the competing risk perspective (Ambrose & LaCour Little 2001; Calhoun & Deng 2002). Of course, a relevant issue not always considered is the extent to which the choice of mortgage instrument creates a selectivity issue for studies focusing upon a single mortgage type.

A concern of early studies using ARM data was the relative importance of the characteristics of the borrower and the features of the ARM contract (SA-Aadu 1988; Cunningham & Capone 1990). The research also used data from a single lender. The results of studies differ with SA-Aadu finding variables reflecting the borrower's credit worthiness generally statistically significant, and Cunningham & Capone finding only the net worth of the borrower to be significant. Cunningham & Capone conclude that differences in default rates between FRM and ARM contracts 'result from the contractual provisions of ARMs . . . and not from borrower clientele effects'.15 The Cunningham & Capone study did use a multinomial logit model to account for prepayment, and utilised current data on house prices. A finding emphasising the importance of the features of the contract generally supports the option theoretic approach to mortgage de-fault,16 and variables indicating the extent to which the option was 'in the money' (e.g. equity net of moving costs) were also found to be statistically significant.

The option theoretic approach can be tested by analysing a household's response to interactions between key state variables and contract features (boundary conditions). The analysis can be extended by determining if households respond in the same manner to contract characteristics, regardless of whether they hold an FRM or an ARM. Calhoun & Deng (2002)

separately estimated a multinomial logit, competing risk, model on ARM and FRM contracts. Though the research found similar responses to option theoretic variables for both ARM and FRM holders, there was some evidence that borrowers might self-select between these two forms of contract. This could reflect unobserved heterogeneity, or some other selectivity mechanism such as potential household mobility. A further indicator of selectivity was the lesser sensitivity of ARM holders to interest rate terms, possibly indicating a shorter expected holding period.

The theoretical work of Kau et al. (1993) discussed in Chapter 9 pointed to the strong impact of 'teaser rates' on prepayment behaviour, mortgage value and mortgage pricing. Any study of household behaviour of ARM holders must account for 'teaser rates'. However, some research has found that 'teaser rates' do not have a significant effect on, or discourage, prepayment (Green & Shilling 1997; Phillips et al. 1996; Vanderhoff 1996). More recent work by Ambrose & LaCour Little (2001) employing the competing risk framework of Deng et al. (2000), found a positive relationship between the size of discount and the likelihood of prepayment. Also ARM holders were found to be more likely to prepay than borrowers holding an FRM. Ambrose & LaCour Little (2001) provide empirical confirmation of the theoretical predictions and numerical simulations of Kau et al.

Though the more sophisticated recent research has detected the expected effects of 'teaser rates' on prepayment behaviour it is useful to consider why the results of studies might differ. Most studies of ARM prepayment behaviour have utilised sometimes limited data from individual lenders covering short time periods, thus inhibiting generalisation. The chosen period of the business or interest rate cycle could also significantly influence results (Phillips et al. 1996). Ambrose & LaCour Little (2001) note that Vanderhoff (1996) finds a lower risk of prepayment for ARMs during a period of recession when prepayment might have been inhibited, perhaps by default, again emphasising the importance of a competing risk perspective. The question of significant clientele/signalling effects in mortgage instrument choice and subsequent termination behaviour remains a crucial one for research.

Given the significant differences between the UK and US mortgage markets discussed in Chapter 1, and in other chapters, then there are also important variations in the features of mortgage contracts across housing finance systems. Unfortunately, there are few UK studies of mortgage termination behaviour available. Theoretical work on the valuation of UK mortgage instruments has been presented by Pereira et al. (2002, 2003). There is also empirical research on time to default Lambrecht et al. (1997), and default risk (Chinloy 1995). There is currently only one study of UK prepayment behaviour (Institute of Actuaries 2002). However, UK research does offer examples of both the application and the limitations of the option theoretic approach to the analysis of mortgage termination.

Lambrecht et al. note that in the UK 'ruthless' default is less likely because the liability for the outstanding mortgage balance remains with the borrower.17 Using a duration analysis on a sample of defaulters, from 1987-1991, the researchers found that ability to pay variables exceeded equity variables in importance.18 Chinloy also found in favour of the importance of liquidity constraints for default in the UK mortgage market, reflecting the liquidity squeezes that can arise with a variable (bullet) rate of interest. The Institute of Actuaries examined UK fixed rate contracts finding the age of the debt, house price inflation, interest rate differentials and prepayment charges effect the likelihood of prepayment. Especially given the theoretical lead of Pereira et al. there is much more scope for research into mortgage termination behaviour in the UK mortgage market.

Mortgage termination studies that have considered mortgage instruments other than the US standard FRM have also found for the importance of the option theoretic approach. The competing risk methodology has proved effective in analysing ARM loan performance. However, there has been some evidence of clientele, or selection effects. Research needs to account for potentially important selectivity biases arising from the choice of mortgage instrument. This has implications for mortgage security valuation and the balance sheet management of lenders. 'Teaser rates' on the ARM, or the VRM in the UK, appear to have a significant impact upon prepayment behaviour. There is no econometric study as yet of prepayment that reflects consumer search and switching between providers, a matter related to the heavy discounting of mortgage debt driven by mortgage market competition.

Another issue for further attention is the known lags in the adjustment of ARM rates to changes in indices. Simulations show that these different reactions can have significant effects on the sensitivity of mortgage price to interest rate variables (Boudoukh et al. 1997; Stanton & Wallace 1999). This may also be an imperative for UK research, where mortgage interest rates have demonstrated a variable lag with respect to base rate changes (Miles 1994; Leece 2000a).

Emergency Money

Emergency Money

Within my 50 page guide, "101 Ways To Raise Emergency Money," you'll learn all the tricks, tips, and cost-cutting strategies you need to start saving and earning emergency funds. The report downloads straight to your computer and you can get started raising that cash in less than an hour from now!

Get My Free Ebook


  • senja
    What is instrument level data in mortgage?
    6 years ago
  • shishay
    What type of mortgage instrument?
    6 years ago

Post a comment