Credit market disequilibrium in Poland: Can we find what we expect?: Non-stationarity and the short-side rule

Abstract

This paper presents an empirical investigation of the disequilibrium hypothesis on the Polish loan market in the 1990s. Using data over this period of rapid and sustained transition, we estimate a disequilibrium model with a standard maximum likelihood method. However, the estimates are highly counter-intuitive as regards the timing of the identified regimes. We show that the gap between the econometric evidence and the expected results may stem from the phenomenon of stochastic non-stationarity in a disequilibrium setting based on the short-side rule. We find that the omission of one non-stationary variable of the cointegrating space or the absence of a “structural” cointegrating relationship in one or both regimes lead to a spurious configuration. In such a case, the wrong use of the standard likelihood function, derived under the stationarity assumption, may lead to non-convergent estimates of structural parameters and, hence, to an erroneous identification of the regimes. Therefore, as a first approach to this problem, we estimate a disequilibrium model with stationary data, and identify the disequilibrium as an excess of quantities of new loans supplied (or demanded) on the market at time t. The empirical results are then robust and economically founded and correspond to the set and the timing of expected regimes.