When Does Agreement to Disagree Generate Positive Return Autocorrelation?

Banerjee, Kaniel, and Kremer (2009) claim that specific models of disagreement (their Examples 3 and 4) illustrate positive return autocorrelation (price drift). Based on a formal definition of equilibrium with heterogeneous beliefs, we prove that these examples do not actually generate price drift. Although there is aggregate underconfidence and the law of iterated expectations does not apply to average expectations in these examples, traders do not disagree about the dynamics of future prices. We present examples which illustrate how price drift can result from disagreement about the dynamics of future valuations, even when traders’ beliefs are correct on average