A priori knowledge usually enables us to decide that some coefficients must be zero in the particular equation, while they assume non-zero values in other equations of the system. We said that identification of an equation is based on variables not included (not appearing) in it. To be identifiable an equation must be independent of one or more important variables, which are included in other equations of the system. Such excluded variables, if operative during the sample period, will generate shifts in the other equations of the model, which will in turn identify the particular equation from which they are absent (i.e. in which they appear with zero coefficient).
Based on the a priori information a list can be prepared, which should be as complete as possible, of the factors which are relevant to the phenomenon being studied. The list can help us decide which of these factors would normally appear in each relationship. For example, assume that we want to study the demand for an agricultural product. The demand equation belongs to a system of equations describing the market mechanism.