Microstructure and high-frequency price discovery in the soybean complex

We develop a theoretical framework and propose a relevant empirical analysis of the soybean-complex prices’ cointegration relationships in a high-frequency setting. We allow for heterogeneous expectations among traders on the multi-asset price dynamics and characterize the resulting market behaviour. We demonstrate that the asset prices’ autoregressive matrix rank and the speed of reversion towards the long-term equilibrium are related to the market realized and potential liquidity, unlike the cointegrating vector. Our empirical application to the soybean complex, where we control for volatility, supports our theoretical results when the price idleness of the different assets is properly accounted for. Our analysis further suggests that the presence of cointegration among assets is related to the time of day and the contract maturities traded at a given time.