In South Africa, there is currently little research available on mortgage defaultand the determinants of default. The purpose of this research was to evaluateand distinguish between mortgage specific factors (including borrower and loancharacteristics), driving mortgage default patterns in South Africa.This study intended to provide guidance to the South African banking industry(i.e. banking institutions), the South African Reserve Bank (policy maker andregulator), and the broader South African population in general. In addition, theresults of the study can be utilised by banking institutions as inputs into theirorigination scorecard, policy rules and risk-based pricing models.Due to the sensitivities around financial institution’s data in South Africa, thesample used for this study was all current (on book) mortgage loans from ananonymous (single) financial institution in South Africa. Also, due to thepractical availability of data, the sample period covered all current loans fromsaid financial institution from August 2006 to December 2008, with informationcovering each loans default or non-default status during that time period.Based on the review of the literature surrounding mortgage defaults, fivehypotheses were formulated. Only 1 of the 5 variables was found to be a gooddifferentiator of mortgage default in South Africa. Current Loan-To-Value (LTV)was found to be the most critical and by far the largest contributor to themodel’s ability to differentiate defaults. It was found that the higher the CurrentLTV, the more likely those borrowers would default. This finding is in line withconventional wisdom in the mortgage industry, in that loan-to-value ratios arepositively correlated with mortgage default.The Age of the mortgage (i.e. account age in months since loan origination),Instalment to Income (ITI), Borrower age (years), Borrower Income and BalanceOutstanding were found to be irrelevant in contributing to the model’s ability todifferentiate mortgage defaults. Also, Balance (Loan) Outstanding was found tobe the variable having the least impact on the model’s ability to differentiateiiimortgage defaults.In summary, given that Current LTV (loan characteristic), was found to be theonly suitable variable in differentiating mortgage defaults, the conclusion is thatonly “Loan” characteristics (i.e. Loan to value) impact on mortgage defaults inSouth Africa. Furthermore, it is the “equity” theory that describes mortgagedefaults, and not the “ability to pay” theory of default
