Abstract
The study forecasts interest rates amid economic and financial instabilities using data sourced from South Africa. Interest rates are a crucial financial instrument used by central banks to manage the economy. The South African Reserve Bank (SARB) sets the repo rate when providing credit to commercial banks and this rate influences the overall interest rates and other lending rates in the economy. The ability to control gives the central bank the ability to contain inflation, stabilize prices, induce investment, and boost economic growth. Forecasting interest rates is challenging due to their daily fluctuations, which makes it difficult to link them to economic fundamentals like monetary or fiscal policies. Nevertheless, macroeconomics can assist since the long-term trend of interest rates is influenced by inflation trends and the equilibrium of real interest rates. Economic instabilities involve a shock to the traditional operations of the economy leading to lower investment, lower spending, and lower growth.
This study aims to cover the gap in studies in the South African context when it comes to the forecast of interest rates by accounting for instabilities. The country has experienced high inflation rates over the years, which have impacted interest rates. Additionally, global economic conditions, such as the COVID-19 pandemic, have had a significant impact on the South African economy, resulting in fluctuating interest rates. Instabilities such as political uncertainty and inflation can lead to inaccurate forecasting outcomes and result in suboptimal policy decisions. The study aims to investigate whether interest rates rapidly change in the presence of economic instabilities. This is to add to the body of knowledge in this under-researched area. Accurate interest rate forecasts can help policymakers make informed decisions regarding monetary policy, while investors can use them to make investment decisions.
The study follows a positivist philosophy, with a quantitative approach utilizing high-frequency time series regression techniques of autoregressive conditionally heteroscedastic to measure volatility and be able to test forecasting power. The Generalized Autoregressive Conditional Heteroskedasticity (GARCH) model and the Threshold Generalized Autoregressive Conditional Heteroskedasticity (TGARCH) have been employed to cater to symmetry and asymmetry assumptions. Monthly interest rate data from the South African Reserve Bank online, ranging from 2000 to 2022 were used. As the repo rate changes only quarterly and less frequently in stable periods, two other series were included, namely, treasury bill rate and government bonds rate. The preliminary test and post-test in estimating GARCH
v
models, for example, unit root test, cointegration test, and diagnostic tests were applied. In-sample forecasting was used, with the actual data ranging from 2000-2022, in forecasting sample of 2018-2022 was used. The advantage of in-sample forecasting is the ability to compare with the real values that are part of the study sample.
The results show that the normal error distribution assumption is the best performing assumption, given better forecasting than the counterparts, across all three series. It was evident that instabilities are inherent within the series, more pronounced around 2009, and between 2016-2020. The results point to the need for better macroeconomic policy management to reduce instabilities and have better predictable trends to allow investors time for planning. Future students can consider other variations in GARCH models such as EGARCH, IGARCH, and APARCH to create a wide range of models for comparison.