MODELING AND FORECASTING INTRADAY VOLATILITY OF NIGERIA INSURANCE STOCK, USING ASYMMETRIC GARCH MODELS

The study involved a detailed examination of insurance stock price and returns data, revealing consistent returns and non-normal distribution typical of financial data. Stationarity of the series was confirmed via the Augmented Dickey-Fuller test, essential for Generalize Autoregressive Conditional Heteroskedascity (GARCH)-type models. Various GARCH models, including asymmetric types like Exponential GARCH, Threshold GARCH, and Power GARCH, were employed to capture the leverage effect, where negative shocks impact volatility differently than positive ones. The Exponential GARCH and Threshold GARCH models showed significant coefficients for their components, effectively capturing volatility clustering and leverage effects, validated by diagnostic tests showing no residual ARCH effects. The Power GARCH model also demonstrated strong performance, validated by high log-likelihood values and good AIC and SIC criteria. Comparative analysis indicated that the Exponential GARCH and Power GARCH models provided the best fit for the data, though the high Mean Absolute Percentage Error values in both models indicated considerable relative errors. The study underscores the importance of using multiple criteria for evaluating model performance and highlights the practical implications for risk management, derivative pricing, and portfolio optimization in the Nigerian insurance sector. By accurately modeling and forecasting volatility, the findings support better-informed decision-making, enhancing financial market efficiency and stability. These models not only emphasize the persistent nature of volatility but also provide methods to discern short-term swings from longer-term patterns in insurance stock returns. Overall, the use of these models enriches our understanding of volatility dynamics in the Nigerian insurance market, having practical implications for risk assessment and strategic decision-making within financial markets.