Econometrica: Apr, 1967, Volume 35, Issue 2

Long-Run Coefficients and Distributed Lag Analysis: A Reformulation

https://doi.org/0012-9682(196704)35:2<278:LCADLA>2.0.CO;2-P
p. 278-293

Yair Mundlak

The difficulty of obtaining empirical coefficients of long-run economic relations stems from the fact that long-run equilibrium values of economic variables are unobservable. To overcome this difficulty, assumptions are usually made about the relationships between observed and the unobserved long-run variables. Such assumptions have led to distributed lag analysis. In an earlier paper I have shown that such an approach is likely to give erroneous results, for the adjustment process may not be consistent with the comparative statics solutions for the short-run variables. In the present paper, the earlier analysis is supplemented and it is also shown how the problem can be overcome and how the long-run coefficients can be estimated even though the long-run equilibrium values are unobserved. The basic deviation from other formulations of this subject is that adjustment equations are used to described only those movements of variables about which comparative statics has nothing to say. In this way, contradictions between comparative statics and ad hoc dynamic formulations are avoided.