On the Modelling of Quantity Constraints: An Empirical Point of View

Business surveys suggest that firms are more often than not quantity-constrained, either by the demand for their output or by the availability of the inputs. Until the late seventies, macro-modellers used to put most of the emphasis on sales constraints, dealing with supply-side quantity constraints in a fairly ad-hoc fashion, simply adding "tension variables" in various parts of an otherwise demand-driven (at least in the short run) model. Over the last ten years however, models have been developed that allow an explicit and rigorous representation of these phenomena. In their most sophisticated versions, these models allow time-varying proportions of sales-, capacity- and labor-supply-constrained firms. Although from a theoretical viewpoint this last approach is clearly superior to the initial one, one may wonder whether empirically the initial procedure does not provide a convenient and much more tractable approximation to the "true" model. This paper aims at answering this question by using Monte-Carlo experiments. The answer is that not introducing quantity constraints in a rigorous way may well alter significantly the parameter estimates and the dynamic properties of the model when the proportion of demand-constrained firms is low, as it was during most of the sixties and early seventies. When the proportion of demand-constrained firms is high (as observed after 1975), the simplified model provides quite a correct approximation.