University of Twente Student Theses

Login

The effects of changes in macroeconomic factors for risk parameters on the bank's mortgage portfolio

Meel, Daniël J. (2011) The effects of changes in macroeconomic factors for risk parameters on the bank's mortgage portfolio.

[img] PDF
1MB
Abstract:The goal of this study is to forecast the risk parameters of the portfolios by linking them to macroeconomic factors, such as unemployment and interest rates, to estimate changes in credit losses under macroeconomic scenarios. The advantage of this approach is the possibility to investigate expected portfolio consequences of changes in the macroeconomic environment for the near and middle long future. In case of successful model building, macroeconomic scenarios can be used as input to predict the default fraction of the portfolio and potential losses (called risk parameters). The main research question reflects this goal: What is the influence of macroeconomic factors on the risk parameters for the mortgage portfolio? Macroeconomic factors that might influence the risk parameters of the mortgage (default) portfolio are derived from a brief literature study and as a starting point scenarios are selected. The relation between macroeconomic factors and the default rates are observed by correlation studies to determine the best time lags. By use of logistic linear regression, with regard to time lags of the macroeconomic factors, the best combination of factors that estimates the number of defaults in a financial period of a month is observed and corresponding parameters are calculated. This is called the default rate, the probability of getting in default. The loss rate (LR) is connected to macroeconomic factors by using microeconomic factors, such as Loan-to-Value (LTV) and Loan-to-Income (LTI) ratios, as intermediate step. A cross-table with LTV- and LTI-classes shows the relationship between the loss rate and both explanatory variables; higher classes correspond with higher losses. The LTV and LTI are linked to the house prices and unemployment, respectively. Because of the lack of information about the applicants, especially about their employment status, the loss rate is directly derived from the LTV-ratio. This approach is time dependent and therefore favored to the cross-table. The only step that has to be taken to collect all information for predicting the future credit losses is to estimate the portfolio value. There are several ways to make a useful estimation, but a macroeconomic link is hard to defend. Therefore the current trend is extrapolated to complete the credit loss estimation. The total credit losses for the portfolios in scope, (1) Intermediary Channel, (2) White Label and (3) a Consolidated Portfolio including (1) and (2) and two more small passive labels, is the multiplication of the probability to get into default (default rate), the 3 fraction the financial institution will lose in case of default (loss rate) and the total value of the default portfolio (exposure value). Because the default rate and the loss rate could co-operate, especially by including the same input factors, it can be assumed that a correlation between the default rate and loss rate is included. Therefore a covariance analysis was performed to eventually correct the multiplication for over- or underestimation of the credit losses. The study gives insight in future default rates, loss rates and credit losses based on selected scenarios and extended with a time series scenario. The Time Series Scenario is constructed by developing time series models for each underlying macroeconomic factor and forecasts of the risk parameters are made by using these time series forecasts as input. In other scenarios the end value of the input factor is known and a straight line from now till the end value over the forecast period is assumed. The individual factors are brought together with a regression analysis for each rate. The observed parameters are used for the forecasts. All default rates models are calculated based on macroeconomic factors and an autoregressive term, sometimes extended with a constant value. Loss rates are based on house prices and a constant by deriving from the LTV-ratios. Covariance between the default rate and loss rate is estimated on the aggregated level and a corrected multiplication is used to estimate the expected losses on a loan as presented in Figure II for the Consolidated Portfolio. A result of the analyses is the huge impact on the default rate of eliminating the mortgage interest deduction (MID). Most of the scenarios are estimating the default rate on the middle long run between 0,25 and 0,30 percent. Increases of yield, unemployment or the abolishing of the MID are affecting the DR clearly. Stress scenarios (Adverse and Benchmark) are obviously resulting in worse default rates (and calculated on a shorter time horizon). The covariance between the DR and LR is negligible and therefore hardly not affecting the results. The expected loss on a loan is expected to stabilize around 50 Euros. On the short term the elimination of the MID will increase the loss, but in the last forecasted year the unemployment scenario is performing worse. Although most rates are hard to predict by macroeconomic input, this approach is favored for the DR in the Intermediary Channel and Consolidated Portfolio compared to an approach only depending on the history of the rate. For the White Label, the macroeconomic inputs are not improving the model.
Item Type:Essay (Master)
Faculty:BMS: Behavioural, Management and Social Sciences
Subject:85 business administration, organizational science
Programme:Industrial Engineering and Management MSc (60029)
Link to this item:https://purl.utwente.nl/essays/62991
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page