Food policy planning in Bangladesh has consistently ignored the problem of output fluctuations, despite the fact that this has led to considerable dislocation to overall economic performance. A necessary first step therefore, is to attempt an assessment of the broad magnitudes involved, as well as their dispersion and trend. Three specific questions are investigated in this context : (a) What is the magnitude of instability and has there been a tendency for it to increase over time? (b) Are certain areas (i.e., districts) and certain crops or seasons particularly vulnerable it? (c) Is foodgrain production non-random in character?

Aggregate instability levels in Bangladesh is small, with no evidence of non-randomness or increase over time. Considerable variations by crops and districts noted, localised shortages is nevertheless high. Fluctuations in one crop are also found to be correlated with another (e.g., Aus and Aman, Aman and Boro), suggesting the operation of critical variables whose impact is not limited to a particular crop alone. Some evidence of non-randomness, along with a tendency for fluctuations to increase, is also noted for Boro and Wheat (HYV). If these tendencies persist, the benefits of a strategy to raise productivity through rapid HYV diffusion may need to be viewed against the costs of adjustments to heightened instability.