Evaluating the Impact of Borrower Characteristics, Loan Specific Parameters,  and Property Conditions on Mortgage Default Risk

Ali Mahmoud ANTAR

doi:10.14505/tpref.v15.2(30).24

Ali Mahmoud ANTAR Department of Economics, Faculty of Business Administration, Beirut Arab University, Lebanon https://orcid.org/0009-0001-2807-1543

DOI: https://doi.org/10.14505/tpref.v15.2(30).24

Abstract

This study empirically examines the impact of borrower, loan, and mortgage parameters on default risk in residential mortgage loans. Using 6743 individual housing loan accounts data from Housing Finance Institutions in Lebanon, we develop a comprehensive model using the multivariable binary logistic regression, best subset logistic regression, and stepwise regression analysis procedures to investigate the impact of 21 predictors and 29 sub-predictor parameters on log odds of default risk. In addition, the study conducted a model diagnosis using the Hosmer - Lemeshow Goodness of fit test, Likelihood Ratio Test, Model accuracy- Classification Table, Statistically Significant Test- ROC curve, and Pregibon Delta Beta Statistics. The study aims to assist financial institutions in managing and assessing the default risk more effectively and develop effective strategies to mitigate this risk. The empirical results suggest that the estimated probability of defaulting on a housing loan is approximately 3.8% when all predicted variables are set at their lowest value. In addition, marital status and the existence of dependence have a positive impact on default risk. The higher the number of dependents is, the higher the risk of default. Moreover, a widowed borrower has a higher log odd of default compared to single, married, and divorced borrowers. Furthermore, the results revealed that self-employed borrowers positively impact the risk of default due to the absence of a steady flow of income. In addition, there is an inverse relationship between the market price-to loan ratio and the log odds of default since the borrower’s equity will increase when the house price increases. However, log odds of default will increase when the loan value is higher than the mortgage market price. Moreover, the result shows that the nature of the borrower’s occupation has a positive relationship with log odds of default where borrowers working in real estate and construction sectors have lower default rates than borrowers working in other industries. In addition, a high interest rate increases the loan's monthly payment and therefore increases the probability of default. Furthermore, the loans granted for purchase and renovation purposes have a lower risk of default than the ones given for construction and under-construction. In addition, the model's overall accuracy was demonstrated by a 91.61 percent visible correct classification rate.

References

[1] Bandyopadhyay and A. Saha (2009). Factors driving demand and default risk in residential housing loans: Indian evidence. MPRA Paper 14352, University Library of Munich, Germany.
[2] Bendel R. B. and Afifi A. A. (1977). Comparison of stopping rules in forward ‘Stepwise’ regression. Journal of the American Statistical Association, 72(357): 46–53. DOI: https://doi.org/10.2307/2286904
[3] Canepa and F. Khaled (2018). Housing, housing finance and credit risk. International Journal of Financial Studies, 6(2): 50. DOI: https://doi.org/10.3390/ijfs6020050
[4] Foote L. et al. (2008). Just the facts: An initial analysis of subprime’s role in the housing crisis. Journal of Housing Economics, 17(4): 291-305. DOI: https://doi.org/10.1016/j.jhe.2008.09.001
[5] Gerardi, K. et al. (2008). Making sense of the subprime crisis. Brooking Papers on Economic Activity, Fall.
[6] Hosmer Jr, D. W. et al. (2013). Applied logistic regression. John Wiley and Sons.
[7] Lacour-Little, M. et al. (2016). What role did piggyback lending play in the housing bubble and mortgage collapse? Journal of Housing Economics, 34: 94-117. DOI: https://doi.org/10.1016/j.jhe.2016.08.003
[8] Levy D.S. and Lee C. Kwai‐Choi (2004). The influence of family members on housing purchase decisions. Journal of Property Investment and Finance, 22(4): 320-338. DOI:https://doi.org/10.1108/14635780410550885
[9] Li Y.and J. Yang (2018). Default risk prediction and feature analysis: A comprehensive study. Journal of Finance, 72(3): 456-473.
[10] Mayer, Christopher et al. (2009). The rise in mortgage defaults. Journal of Economic Perspectives, 23 (1): 27-50. DOI: https://doi.org/10.1257/jep.23.1.27
[11] Mickey, R. M. and S. Greenland (1989). The impact of confounder selection criteria on effect estimation. American Journal of Epidemiology, 129(1): 125-137. DOI: https://doi.org/10.1093/oxfordjournals.aje.a115101
[12] Moore et al. (2018). Introduction to the practice of statistics. Publisher: W. H. Freeman Publication.
[13] Neyman, J. (2023). Unit 7: Logistic regression. Available at: https://people.umass.edu/biep640w/pdf/7.%20%20Logistic%20Regression%202023.pdf
[14] Rosengren E. S. (2008). The mortgage meltdown - Implications for credit availability. Discussion presented at the U.S. Monetary Policy Forum, New York, NY. Federal Reserve Bank of Boston. Available at: https://fraser.stlouisfed.org/title/statements-speeches-eric-rosengren-9015/mortgage-meltdown-implications-credit-availability-667430
[15] Sandar, T.et al. (2010). Factor affecting the borrowers' repayment performance of microfinance in sub-urban areas of Yangon, Myanmar. Journal of Economics, 14(2): 125.