| Summary: | ABSTRACTS
This paper attempts to discuss the coefficient of determination (R2) in time series econometric analysis. The coefficient is the most commonly used measure of the goodness of fit of a regression line. With time series data high R2 values can be obtained if the linear regressions are estimated with the level. Therefore, there is a strong tendency to estimate the time series model in levels rather than, for example, in first difkrences. In general, researchers associate a high R2 with a good fit and it can be considered as indicative of a strong ability of the independent variables to "explain" the dependent variable. In this case, investigators may face the "R2 syndrome " and the "spurious regression" problems.
However, the le statistic is only valid if the proposed model must be linear and include a constant or intercept term and be estimated using ordinary least squares' (OLS). Furthermore, it must be noted that the R2 may not be directly comparable if the dependent variables of the models under consideration are not the same.
Key words: Coefficient of determination, goodness of fit, le syndrome, stasionarity and spurious regression.
|
|---|