This paper has two purposes. The first is to introduce an econometric model of export pricing and sales behaviour, the parameters of which vary with some indicator of excess productive capacity. For the individual firm this is achieved simply by proposing that it follows a demand-constrained regime when levels of working are low, and a supplyconstrained regime when full capacity is approached. Application of this sort of model to the whole economy would lead to implausible discontinuities; as overall capacity utilization rose a switching point would be reached where factors influencing demand-world trade, competitor prices-became suddenly unimportant and those influencing supply-investment, profitability-became suddenly all-important. The second purpose of this paper is to offer a resolution of this conceptual inelegance by postulating that at any one time firms experience a variety of capacity utilization conditions distributed around means which are generally high at peaks of the domestic business cycle and low in troughs. The resulting aggregate equations exhibit no unwonted discontinuities. The potential area of application of this device is large, as all multi-regime models involve some abrupt changes in the behaviour patterns of the agents they describe. It is, however, circumscribed by the technical problems of combining the aggregation procedure with constraints on the agents' behaviour which are essentially stochastic, and by the need to fabricate data on the way the switching variable (capacity utilization) is distributed across individual agents (firms). We shall look at two versions of the export model, the first resting on the assumption that the constraints on firm behaviour share a common stochastic term, thus effectively reducing these technical problems to those of pure aggregation, the second on the more general premise that such random elements are independent. The first suffices to illustrate the important points that weighting or sample-splitting schemes are inappropriate to its estimation, and that there is a real danger of incorrectly rejecting the hypothesis that two regimes are at work if firms differ in their experience of the switching variable. The second model is developed partly to reveal a further source of bias in this direction, but mainly to catalogue the practical problems of obtaining a manageable estimating equation for a model in which independent stochastic constraints coexist with an aggregation problem. This is in fact achieved only by confining all random variables to a non-normal class of distribution functions which is tailored to produce relatively simple expressions for total exports while making full use of the meagre data available on intraindustry variations in capacity utilization. In the first section below, the basic model of the exporting firm is formalized and set in the context of current research on models with varying parameters. In the second section, aggregate equations for the economy's exports are derived and the problems of estimation and measurement posed by them are discussed. Finally, estimated equations for the volume and price of UK exports of manufactured goods are presented with special emphasis on the distinct effects of cyclical variations in the degree of capacity utilization on export demand and supply.