We measure the nature and severity of a variety of belief distortions in market reactions to hundreds of economic news events by synthesizing structural estimation with algorithmic machine learning to quantify bias. We find that investors systematically overreact to perceptions about multiple fundamental shocks, a phenomenon we show often dampens rather than amplifies market volatility via a “shock composition effect.” Such effects imply that the stock market can underreact to news, even when investors overreact to all shocks.

We examine the role of insurance companies in supporting resiliency in the corporate bond market. We show that during the COVID-19 liquidity crisis, insurers increased their corporate bond positions, particularly in bonds facing fire sales by mutual funds. Insurers with more stable funding were more likely to buy, and they bought more from dealers with whom they had prior trading relationships. Dealers improved their bond liquidity provision when they had trading relationships with insurers with more stable funding. Our work demonstrates that insurers can play an important role in supporting bond market resiliency during times of stress.