That is the case in our model. However, I’ve been, if anything, a bit skeptical about that. But when you reduce any component—exogenous government spending, exogenous investment, a change in the saving rate—you create an opening in the demand–supply balance that is then going to be filled by something, and you want it filled by exports. That would be the pro-adjustment response.
Our model suggests that, as monetary policy eases in response to flagging demand, we’re more likely to get interest-sensitive components of U.S. domestic demand stepping in than to induce a change in the exchange rate and get higher exports. But that is all conditional on two things, one of which is how interest-sensitive that demand is. And if the interest-sensitivity isn’t as high as we have it in the model or if those components suffer some of the residual problems of the sort they have suffered lately because of confidence or whatever, maybe U.S. interest- sensitive demand won’t come rushing in. That’s one possibility. The other is that, as far as the model is concerned, that relationship depends on the model’s ability to tell us what the exchange rate is going to do because the induced effect on the exchange rate is how that adjustment works. I always start and stop these conversations by saying that we don’t have a good model of the exchange rate and that’s why we make all our statements conditional. In order to close models—most models, including ours—we use uncovered interest parity because the model has to have an exchange rate equation. There has to be something that does that, and that’s what economic theory says to put in there, so that’s what we put in. But uncovered interest parity has never succeeded empirically, ever. So that conclusion rests on these two properties: that we don’t really know what the exchange rate response would be and that we don’t know if our interest sensitivity is as high as the model says it is.