Leverage and asymmetric volatility: The firm-level evidence *

• Investigate financial leverage and asymmetric volatility with dynamic panel VAR
• Use firm level data for all firms in CRSP and COMPUSTAT from 1971 to 2013
• Allow leverage, volatility and risk premia to dynamically influence each other
• A change in leverage has a prolonged effect on volatility.
• The leverage effect is up to five times larger than a static model would predict.

The relative statistical and economic significance of the leverage and feedback effects on firm-level equity volatility remain an open issue. We use a dynamic panel vector autoregression framework to investigate both effects simultaneously for all firms in CRSP and COMPUSTAT from 1971 to 2013. Crucially, we allow financial leverage, volatility and risk premia to influence each other over time. We find a much larger leverage effect than reported in Christie (1982). Importantly, we find that a change in leverage has a prolonged effect on volatility. The cumulative leverage effect is up to five times larger in twelve quarters than a static model would predict for one quarter.