A number of empirical studies have focused upon prepayment behaviour alone (Green & LaCour Little 1999; Abrahams 1997; Quigley & Van Order 1990; Bennet et al. 1998, 2000, 2001). This does not mean that default is ignored entirely. For example, Green & LaCour Little and Abrahams, assume that default patterns are modelled in the base hazard of a Cox Proportional Hazards model. Other studies offer some control by including a measure of equity, or the loan-to-value ratio, at origination (Quigley & Van Order 1990; Bennet et al. 1998). The importance of default may depend upon the sampling period and the associated economic environment. There are occasions when default is dominated by prepayment, and so presents a less critical selectivity or specification problem.
Of course, prepayment only studies fall far short of modelling the competing risk of prepayment and default. However, a review of this work is helpful in highlighting key issues in the analysis of prepayment behaviour. In particular, some research has been useful in pointing to the impact of structural changes in the mortgage market (Bennet et al. 1998, 2001). Prepayment specific studies have also stressed the importance of institutional factors and credit market constraints in explaining prepayment rates. There has been an increasing tendency to use loan level data to include personal characteristics including those factors likely to induce an household to move and thus prepay (Quigley 1987; Archer et al. 1996; Archer et al. 1997; LaCour Little 1999; Green & LaCour Little 1999).
One problem with early studies is that they seldom distinguish between the various motives for prepayment. In the case of the analysis of mortgage pools this extends to combining prepayment and default motivated terminations (Peters et al. 1984; Richard & Roll 1989; Schwartz & Torous 1989a, b, 1992; Foster & Van Order 1990). Brady et al. (2000) uses data generated by the University of Michigan Consumer Survey to analyse refinancing motives. They note that 21% of prepayments allowed a switch from an ARM to an FRM and 35% were to liquefy their equity in the property. This indicates the different purposes served by refinancing, some of which may be jointly determined. Caplin et al. (1997a,b) combined refinancing and mobility induced terminations in their data, finding that interest rate volatility, which effects the value of the call option, had no impact upon prepayment behaviour. Studies by Giliberto & Thibodeau (1989) and Bennett et al. (2000) did not confound these separate motives and detected a statistically significant negative impact of volatility on termination.
LaCour Little (1999) controlled for the various motives for refinancing by using a sample of borrowers who remained with a single lender, and only changed their mortgage rate and term with that lender.11 Therefore the sample was standardised by restricting the observed prepayments to refinancing only. A probit was estimated with refinancing expressed as a function of the ratio of the current mortgage rate to the coupon rate, personal characteristics, and the transaction costs of refinancing. The research found that borrower and loan characteristics only have a significant effect when the option to prepay is at or 'near the money', a result consistent with the vega estimates of Bennett et al. (2000). Transaction costs, which differ across individuals create a new threshold before prepayment takes place (Bennett et al. 2000). In another study (Green & LaCour Little 1999) found that falling house prices,12 and hidden transactions costs are not sufficient explanations for non-prepayment.
Transaction costs can include the difficulty of refinancing if a household is liquidity/credit constrained (Archer et al. 1996; Peristiani et al. 1997; Bennet et al. 1998, 2000, 2001). Transaction costs can also change as a consequence of structural changes in mortgage markets (Bennet et al. 1998, 2001; Sanyal 1994). An interesting feature of the US market is prepayment cycles. For example, there is an apparent acceleration of prepayment rates during the 1990s, compared to the 1980s. A number of important structural changes could account for this phenomenon. These include the spread of securitisation, mortgage lending by a wider range of financial institutions, faster processing of mortgage applications and the increased integration of mortgage markets with other capital markets. Such changes may also be responsible for high levels of mortgage refinancing observed in the UK.
Bennet et al. (1998, 2001) estimated a proportional hazard model on 12,835 observations, covering the periods 1984-1990 and 1991-1994. Shifts in the survival curves were interpreted as evidence of a positive impact of structural change on refinancing behaviour. The research confirmed the importance of credit ratings and homeowner equity for mortgage prepayment. The importance of liquidity constraints has been indicated by other work. Peristiani et al. (1997) estimated a logit model and found evidence of credit market constraints. The results highlighted the importance of the homeowner's credit history. Changes in the level of home equity and in the lending environment were also significant.13 Refinancing was less responsive to interest rate falls during the 1990s. This phenomenon might be explained by poor credit histories arising from bankruptcies in the late 1980s.
There is a case for examining liquidity and credit constrained households separately. Archer et al. (1996) used American Housing Survey data to estimate a logistic regression. The estimation included a dummy variable to indicate those borrowers who were 'in the money' and not subject to income and collateral constraints (dummy = 1). The results showed that non-constrained borrowers were more likely to refinance. The research also indicated that demographic characteristics allocated households between the constrained and unconstrained groups. Bennet et al. (1998, 2001) found that credit constrained borrowers were less sensitive to changes in the intrinsic value of the option to prepay.
Prepayment focused studies have demonstrated the importance of identifying the motives for refinancing. They have also suggested the significance of structural change in explaining prepayment cycles. Liquidity and credit market constraints can discourage prepayment and prepayment certainly seems to have become increasingly important post-financial deregulation. The importance of refinancing motives, structural change and credit histories has been established. The negative impact of interest rate volatility upon the likelihood of prepayment suggests the importance of rational calculation (Giliberto & Thibodeau 1989; Bennett et al. 2000). However, if we wish to have efficient and consistent estimates of prepayment behaviour then the competing risks of prepayment and default must be recognised, and incorporated into our methodology.
Was this article helpful?
The Shocking True Story of How I Raised My Credit Score 165 Points in 3 Months and Saved $1,000’sIn Interest. First off I’d like to say I am not a lawyer and this is by no means legal advice. Before implementing the ideas in this report consult with a qualified attorney. This is simply my story on the tactics I used to legally and ethically raise my credit score.