Uncertainty, Financial Stress and Monetary Policy in South Africa

Abstract

In this thesis, we examine the role of economic uncertainty and financial market stress in the South African economy and its implications for monetary policy. Financial market disruptions and economic uncertainty are commonly listed as the main sources of economic turmoil and slow recovery experienced by many economies globally following the 2007-08 global financial crisis. Understanding how financial and uncertainty shocks impact the real economy and the ability of financial market information to predict economic conditions is imperative from a policy perspective, potentially informing prudent macroprudential and monetary policy. Against this backdrop, this thesis is organised into three main chapters. In Chapter 2, we develop an index to monitor the intensity of stress in the South African financial sector, and examine the potential non-linearity in the transmission of financial shocks to the real economy. The index (called SAFSI) is constructed using a novel approach that selects and aggregates financial indicators based on their information content i.e. their ability to capture key periods of financial stress in the economy, thereby striking a balance between parsimony and efficacy. Furthermore, the use of time-varying correlations in the aggregation allows the index to capture the interconnectedness of financial markets as well as enabling each indicator to be assessed in terms of its systemic importance. In addition to capturing the benchmark episodes of financial stress in South Africa, the SAFSI successfully captures other global and idiosyncratic risks that affect the financial markets in the country. A regime-switching model reveals non-linearity in the transmission of financial shocks to the real economy. Specifically, financial shocks are more detrimental to the real economy during stressful periods than normal times. An unexpected shock to financial stress conditions during financially vulnerable times is associated with a more prominent contraction in output and higher inflation. However, during normal times, the financial shock has a negligible impact on prices and interest rates, with a small output impact. Chapter 3 examines the predictive ability of financial market information, captured by the SAFSI, by evaluating economic forecasts from three vector autoregression specifications: a mixed-frequency specification which includes quarterly and monthly time series data, a standard linear quarterly frequency model, and a threshold (non-linear) quarterly frequency specification. Out-of-sample forecasts reveal that accounting for intra-quarterly information improves the forecasting performance of financial information in terms of output growth and inflation, which are considered key economic indicators in formulating monetary policy decisions. Finally, in Chapter 4 we examine the connection between economic uncertainty and financial market conditions in South Africa, documenting that the macroeconomic implications of an uncertainty shock differs across financial regimes. A non-linear VAR is estimated where uncertainty is captured by the average volatility of structural shocks in the economy, and the transmission mechanism is characterised by two distinct financial regimes (i.e. financially stressful versus normal periods). We find that while the deterioration of output following an uncertainty shock is much more prominent during normal periods than during stressful periods, it is much more persistent during stressful financial times. The share of output variance explained by the volatility shocks in good financial times is more than double the share in bad times. Uncertainty shocks are found to be inflationary in both regimes, with the impact being larger in the stress regime. While our estimates reveals that financial frictions do not amplify the impact of uncertainty on real output, it does increase the impact on prices.