Household Choice of Fixed Versus Floating Rate Debt: A Binomial Probit Model With Correction For Classification Error

The paper presents an econometric analysis of the choice by households in the United Kingdom, of ®xed rate mortgage debt covering the years 1991 to 1994 inclusive. Fixed rate mortgage instruments soon became a numerically signi®cant mortgage choice. For example the proportion of new mortgages that were ®xed rate was 46 percent in the ®rst quarter of 1993 (Coles, 1993) and 60 percent in the third quarter of 1993 (Housing Finance, 1994). The paper identi®es cross-section and interest rate effects on this mortgage choice. Data are obtained from the British Household Panel Survey (BHPS). There are important macro economic implications from the growth in the use of ®xed rate debt (Coles, 1993; Miles, 1994). Lack of ®xed interest rate and mortgage choice data has resulted in a dearth of econometric research in this area for the United Kingdom (Coles, 1993). Empirical work on the choice between ®xed rate mortgages (FRMs) and adjustable rate mortgages (ARMs) is evident in the United States (see Bruekner, 1986; Alm and Follain, 1987; Dhillon et al, 1987; Bruekner and Follain, 1988). The present research is differentiated by a focus on the choice of short run ®xed rate contracts (FRM) and a variable rate mortgage (VRM). North American contracts tend to be ®xed for 15 or 30 years. The paper also considers classi®cation error, a speci®cation problem neglected in previous work. The choice set for consumers in the mortgage market is heterogeneous and extensive, increasing the likelihood of classi®cation problems. The study begins (Section II) with a discussion of ®xed rate mortgage contracts from 1991 (Q.1) to 1995 (Q.1) and includes a review of the OXFORD BULLETIN OF ECONOMICS AND STATISTICS, 62, 1 (2000) 0305-9049