This paper uses Kennan's (1988) model to separately identify supply-side and demand-side dynamics in US import data. Dynamics arise from both autocorrelated shocks to supply-and demand-side fundamentals, and from lagged adjustment to these shocks. The model consists of a pair of partial adjustment models in which consumers and producers each attempt to follow a stochastic target level of imports subject to a quadratic adjustment cost. The model is applied to quarterly data on US imports of seven narrowly deยฎned commodities: Autos, Beer, Cameras, Wine, Cigars, Tea, and Soap. Two main results emerge. First, adjustment costs are important on both sides of the market. Second, supply-side adjustment costs are larger than demand-side adjustment costs. # 1998 John Wiley & Sons, Ltd. estimating the lags in international trade (see e.g. Mann, 1989 or Krugman and. Ideally, we would like to say more than that a `structural break' occurred. Doing this, however, requires an understanding of the supply and demand factors that give rise to the dynamics in international trade.
On the supply-side, there have been many papers that apply standard partial adjustment or error correction models to the dynamics in international trade (see e.g. . On the other hand, and develop models of consumer switching costs and brand loyalty in which dynamics arise from a lock-in eect operating on the demand side of the market. While all these papers shed light on the nature of the dynamics in international trade, a troublesome feature of this literature is that, in principle, dynamic factors operate on both sides of the market. This means that models incorporating dynamics on only one side of the market run the risk of falsely attributing to one side of the market dynamics that actually arise from the other side of the market. Clearly, this could seriously distort our inferences about the sources of international price dierentials (i.e.