M.Comm. (Financial Economics) / The term structure of interest rates, particularly the term spread determined from the difference between ten-year government bond yields and three-month Treasury bill yields, has received increased attention as a valuable forecasting tool for the purposes of monetary policy and recession forecasting. This is on the back of the observed positive relationship between term spread and economic activity. Moreover, the term spread has been observed to invert prior to the occurrence of economic recessions both in developed and developing countries. This study investigated the forecasting ability of the South African (S.A.) term spread in predicting S.A. recessions, taking into account the recent global economic recession. The reason behind the investigation is due to the forecasting consistencies illustrated by the term spread in providing statistically incorrect signals of recession in 2003, which did not transit into reality. It implied a weak relationship between the S.A. term spread and economic activity. Moreover, based on observations from the literature that term spreads and economic activities across countries are correlated, the term spreads of China, United States (U.S.) and Germany were investigated and compared to the S.A. term spread, to determine which better forecasts S.A. recessions. The study employed the Dynamic Probit Model, since it is considered to provide a better predictive edge over the Traditional Static Probit model. The findings revealed that the S.A. term spread accurately predicted all the S.A recessions since 1980; Chinese term spread accurately predicted the 1996 and 2008 S.A recessions; U.S. term spread predicted some recessions; while German term spread predictions were countercyclical.
| Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:uj/uj:7692 |
| Date | 24 July 2013 |
| Creators | Mohapi, Alphons |
| Source Sets | South African National ETD Portal |
| Detected Language | English |
| Type | Thesis |
| Rights | University of Johannesburg |
Page generated in 0.0024 seconds