Forecasting nonperforming loans (NPLs) is a primary objective for credit providers. NPL forecasts assist in financial budgeting and provisioning for bad debts. The difficulty in accurately identifying the determinants of domestic NPLs has led to a review of time series forecasting techniques. This dissertation explores whether a forecasting model combining a traditional time series approach with a Fourier series residual modification technique performs well in projecting NPLs. It also seeks to establish if selecting an adequate time series model before modifying its residual terms is of benefit. Using the data of an unsecured consumer credit provider in South Africa, the in-sample and out-of-sample performance for a seasonal time series model and residual modified model were evaluated. The results demonstrate that a time series model performs well but the out-of-sample forecasting errors may be reduced by including the lowest Fourier frequencies to modify the residual terms.
| Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:uct/oai:localhost:11427/22856 |
| Date | January 2016 |
| Creators | Luckan, Pranisha |
| Contributors | Huang, Chun-Sung |
| Publisher | University of Cape Town, Faculty of Commerce, Department of Finance and Tax |
| Source Sets | South African National ETD Portal |
| Language | English |
| Detected Language | English |
| Type | Master Thesis, Masters, MCom |
| Format | application/pdf |
Page generated in 0.0013 seconds