Time variation of second moments from a noise trader infection model

This paper considers the dynamics of conditional second moments resulting from a statistical description of agents' attitudes and behaviour in a noise trader/infection framework. In general, the dynamics of second moments of state variables (e.g. prices) can be seen to exhibit autoregressive structures. Conforming to well-known empirical findings, volatility can change over time, shocks to volatility have lasting effects and mean-reversion in second moments is observed. Thus, this kind of market microstructure modelling can contribute to explain the regularities in variances found in data from asset markets and foreign exchange markets.