TRAP@NCI

Mortgage Default Rates in Ireland – Descriptive statistics and prediction modelling

Leslie, Aidan (2014) Mortgage Default Rates in Ireland – Descriptive statistics and prediction modelling. Diploma thesis, Dublin, National College of Ireland.

[img]
Preview
PDF (Diploma)
Download (1MB) | Preview

Abstract

The purpose of this dissertation is to analyse default rate trends in the Irish mortgage market using data from one of the two largest retail banks in Ireland. The goal of this analysis is to try to discover what variables are most likely to cause a property loan to fall behind on mortgage repayments. Under the reporting structure of the Central Bank of Ireland, which can be found in most publications on the subject of mortgage arrears, a loan is said to be in default if it is more than 90 days in arrears, but all arrears accounts, including those just one payment in arrears are reported on. This project will analyse the latter, that is all loans that fall behind on payments and any referral to default loans within this paper refer to this cohort.

Once the variables that seem to have the most impact on arrears have been identified, this project will then present whether this data can be used to predict whether a loan will go into default with any degree of certainty. The C5.0 and rpart algorithms were employed to understand if mortgage default rates could be predicted based on the variables identified. What will be presented is that these algorithms failed to identify default accounts, the C5.0 being slightly more successful but still having a high error rate, particularly type I errors, predicting that loans would not default, when in fact they did. The rpart algorithm did not predict any defaults when 12% in fact did go into default.

The overall findings of this analysis supports previous reports on the mortgage arrears crisis (mostly industry papers which have been quicker to respond to the causes and solutions than academic studies on the subject), namely that loans taken out in the peak of property valuations have the highest default rates, loans taken in rural and midland areas have a higher default rates (as a percentage of all loans in that area), and that investment loans have a higher rate of default.

The mortgage book being analysed as part of this dissertation included loans held under separate legal entities, one being the traditional bank (where the majority of loans reside) and the second being a building society owned by the bank. It was found that the default rates were higher for the building society but that this data was not included in the final analysis because the administrative structures in place at this entity were not as well established as the banking entity and as such, the data quality was less reliable, or for many variables was unavailable.

In conducting the analysis on mortgage default rates the variables used focused on the three main aspects of credit risk analysis in mortgage loans, that this, (i) the property, (ii) the loan product and (iii) the borrower. Details on these variables including the reason for their inclusions are explained in greater detail further on.

Item Type: Thesis (Diploma)
Subjects: Q Science > QA Mathematics > Electronic computers. Computer science
T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science
Divisions: School of Computing > Higher Diploma in Science in Data Analytics
Depositing User: CAOIMHE NI MHAICIN
Date Deposited: 12 Dec 2014 14:59
Last Modified: 12 Dec 2014 15:00
URI: http://trap.ncirl.ie/id/eprint/1840

Actions (login required)

View Item View Item