Estimating Time-varying Fiscal Multipliers in South Africa Using Bayesian Dynamic Linear Model

Abstract

This study investigates the evolution and state-dependence of fiscal multipliers in South Africa from 1961Q1 to 2025Q1 using quarterly data on real GDP, government expenditure, inflation, interest rates, and the ZAR/USD exchange rate. Inflation and interest rate data were sourced from the Federal Reserve Economic Data (FRED) database, while the South African Reserve Bank (SARB) provided GDP, fiscal expenditure, and exchange rate series. Employing a Bayesian Dynamic Linear Model (BDLM), the analysis captures both gradual structural changes and differences in fiscal policy effectiveness across economic growth regimes. The results reveal a marked increase in fiscal multipliers over time, from negative or near-zero values in the early 1960s to approximately 0.08–0.09 in recent years. This upward trend reflects structural shifts such as the democratic transition, expanded public service delivery, infrastructure investment, and improved macroeconomic coordination. However, the estimates also show strong state dependence: the multiplier is significantly larger during high-growth periods (0.0509) than in low-growth periods (0.0006), where it is statistically indistinguishable from zero. This pattern contrasts with advanced economy evidence, where multipliers are often stronger in recessions, and suggests that structural bottlenecks—such as energy shortages, rigid labour markets, and weak investor confidence—limit fiscal effectiveness during downturns in South Africa. The findings imply that fiscal policy has historically reinforced growth momentum rather than reversing contractions. Enhancing its countercyclical potency will require structural reforms alongside targeted public investment. The study underscores the value of state-contingent fiscal planning, supported by real-time monitoring of multiplier dynamics, as a means to maximise growth impacts while maintaining fiscal sustainability.