Identifying The Monetary Policy Transmission Channels: The Role Of Simultaneity, Model Nonlinearity, Expectation Formation Mechanisms And Policy Rules

The issue of quantifying the empirical relevance of the different channels of transmissions through which monetary policy exerts its influence on demand, output and prices has received wide and increasing attention in recent years. In this paper we examine under which conditions the different channels can be separately appraised in such a way that the sum of their effects coincides with the overall effect of monetary policy on the macroeconomy. While our results can be used to individually assess the empirical relevance of the different mechanisms that are at play, in an econometric model, when a shock is given to a generic exogenous variable, we make explicit reference to the case of a monetary policy shock. In principle, a decomposition of the overall effects of a monetary policy shock into a number of channels (typically, all or a subset of the following channels are considered: cost of capital channel; cash/flow channel; wealth channel; exchange rate channel; credit channel) can be easily carried out with the aid of an econometric model, by means of a set of appropriately designed simulations: in each of them, the monetary policy shock is only allowed to transit through one single channel, while retaining the full simultaneous structure of the model. Several cases can be distinguished, and are first addressed in the case of a generic econometric model. If the policy interest rate is taken to be exogenous, a fairly precise decomposition of the overall effect may be obtained. With a linear model, such a decomposition is precise, in the sense that the sum of all channels' effects equals the overall effect. With a nonlinear model the performance of the approach we propose tends to deteriorate (however, the degree of nonlinearity of econometric models tends to be quite limited, and hence the approach is likely to remain reliable in practice; ways to cope with this issue are also suggested). The goodness of the decomposition does not in any way depend upon the assumptions that one makes concerning the mechanism of expectations formation (i.e., whether expectations are adaptive or model-consistent). If the model also includes a monetary policy rule (so that a monetary policy shock no longer corresponds to a shock to the policy rate, but rather to its discretionary component, i.e., to the error term in the rule), then a nearly-perfect decomposition is, in general, no longer possible. Intuitively, this is so because the channels can no longer be "shut down" one at a time while leaving the model structure unaffected. Conditions under which a decomposition is feasible are provided. Finally, the proposed approach is applied to decompose the effects of a monetary policy shock with a model of the Italian economy.