Were harvests in pre-industrial England "self-contained"?
Data on seed yields and grain prices are the primary sources of information regarding the productivity of pre-Industrial English agriculture. It has long been known that grain prices display runs, with high and low prices persisting year to year. The Hoskins hypothesis argues that the persistence of grain prices reflects underlying harvest dynamics. In contrast, a more recent empirical literature contends that grain stores produced the runs in grain prices and that grain harvests were random self-contained events. A model of harvests incorporating variable sow rates, grain stores, and a subsistence constraint is developed. The model suggests that; (i) exceptionally poor harvests may persist; and, (ii) existing tests for harvest persistence are flawed. The model's predictions are tested using 13th century seed yields from the Winchester manors. Two empirical innovations contrast with standard approaches in the literature: (i) seed yields are not aggregated but analyzed as a balanced panel by year, manor, and crop; and (ii) only subperiods with continuous annual observations by vill are analyzed (no interpolation is employed). Evidence is presented consistent with persistent harvests at the level of the manor. While we fail to reject a version of the Hoskins hypothesis, we do reject its application to grain price dynamics.