ECONOMIC SENTIMENT AND STOCK MARKET VOLATILITY: EVIDENCE FROM THE NIGERIAN EXCHANGE GROUP

Abstract

This study examined the effects of direct investor sentiment, indirect investor sentiment, and selected macroeconomic variables on stock market volatility in the Nigerian Exchange Group using 120 monthly observations from January 2016 to December 2025 within a multiple regression framework. Direct sentiment was proxied by the Consumer Confidence Index and indirect sentiment by a composite index of trading volume and turnover ratio; macroeconomic controls included inflation, exchange rate, oil price, and stock market development. Direct investor sentiment had a negative and statistically significant effect on stock market volatility (Beta = -0.528, p < 0.001), with the model explaining 27.9 per cent of variance in volatility — a finding that departs from the conventional expectation that rising optimism fuels speculative excess and is interpreted through the Adaptive Market Hypothesis as a context-dependent sentiment effect in a persistently pessimistic frontier market. Indirect investor sentiment produced no significant effect on volatility in either the bivariate or full model, indicating that volume-based proxies capture liquidity dynamics rather than behavioural sentiment in the Nigerian market microstructure. In the full model, inflation was the only macroeconomic variable to reach significance, confirming that price instability is a direct transmitter of uncertainty into equity market risk; the combined model explained 34.4 per cent of variance (Adjusted R² = 0.309, F = 9.864, p < 0.001). Severe autocorrelation was present across all models, qualifying the reported significance levels and motivating re-estimation through GARCH or Newey-West HAC approaches. The findings contribute updated evidence on the sentiment-volatility relationship in a frontier market and recommend that regulators incorporate consumer confidence indicators into market surveillance frameworks and that future research adopt volatility-specific time-series estimation methods.