Learning, noise traders, the volatility and the level of bond spreads

According to various studies, sovereign bond spreads often deviate from any "sensible" perception of default risk. It is usually attributed to behavioral effects (overreaction) or illiquidity. The former explanation imposes some irrationality or bounded rationality on investors; while the latter usually relies on some informational asymmetry or thin markets. The paper presents a different source of liquidity risk: in a Diamond-Dybvig type model, where agents face a liquidity risk (becoming more risk-averse early consumers), changes in the speed of public learning about default risk may increase bond spreads. This effect operates through a link between future volatility and current levels: increased expected future price volatility (a volatility effect) leads to lower prices today (a level effect). Under reasonable parameter values, accelerated information revelation may increase spreads by 50%. I also compare the welfare of the issuer and investors under different speeds of learning: revealing information may be good or bad for the issuer (issue prices may increase or decrease), and also for the investors (ex ante utility might be higher or lower).