Fachartikel
Estimation of short- and long-term irreversible Relations

Estimation of short- and long-term irreversible Relations

In practice, long-term and short-term irreversible reactions for example of demand due to changes in prices, advertising expenditures, etc. are more important than is commonly assumed in economic theory. The main aspects of this paper are:

- Most empirical analysis do not consider that irreversible reactions of decision makers can be based on quite different behaviour patterns. The nonselective appliance to estimate irreversibility resp. asymmetry inheres the risk of estimating extremely distorted parameters.

- The methodical differences between the five approaches compared in this paper depend on whether in each step the changes in the dependent variable result mathematically from one or both components of the split independent variable. Statistical significance-tests do not enable to decide, which of the five approaches is suited best to estimate the underlying form of irreversibility in a dependent variable. Therefore, a particular method of testing is proposed in this paper.

- An approach to quantify asymmetric relations by splitting the error correction term, proposed by Granger and Lee (1989), has been proved to be not practicable in model analysis as well as in empirical analysis.