Does accuracy improve the information value of trials?

"We develop a model where products liability trials provide information to consumers who are not parties to the litigation. Consumers use this information to take precautions against dangerous products. A critical assumption is that consumers cannot differentiate between firms that have never been sued and firms that have been sued but settled out of court. In this framework, we show that perfectly accurate courts do not maximize information to consumers and thus welfare, contrary to Kaplow and Shavell (1994). More accurate courts provide more information only if producers go to trial. Greater accuracy, however, encourages producers of dangerous products to settle and hide their type. When courts are perfectly accurate, all low quality producers settle. And given the lack of any information from trials about bad types, consumers (rationally) fail to take precautions. If consumer precautions are relatively more efficient than producer precautions, our conclusion stands even when firms can invest in improving the safety of their products"--National Bureau of Economic Research web site.