| Zusammenfassung: | abstract (introduction): recent years have witnessed a remarkable growth of interest in testing the assumptions underlying the econometric model building and estimation process. the present paper summarizes our effort to incorporate these ideas intothe ias-system. the ias-system is a dialog-oriented computer software package for econometric modelling and corporate planning. it is being developed since the mid 70's in the department of mathematics and computer science at the vienna institute for advanced studies, in close cooperation with the institute's economics department and external users. the main features of the ias-system are easy data handling and manipulation and the simulation of (possibly nonlinear) econometric models comprising several hundred equations. in addition, the ias-system provides all standard econometric estimation procedures, linear programming and seasonal adjustment techniques, and a broad selection of time series methods such as power spectra and proceduresfor the estimation and forecasting of arma-models. this paper gives an overview of econometric tests and diagnostic checking techniques that have recently been added to the system. among the procedures now available are tests for higher order serialcorrelation in the context of lagged endogenous variables and simultaneous equations, and tests to detect nonlinearity, structural breaks and errors in the variables. the particular selection of procedures reflects the needs of users of the ias-system, and is not intended to be representative of ongoing research on testing econometric hypotheses. section 2 below presents the theoretical framework of the tests. this is meant to give a rough idea of the procedures rather than to provide a substitute for the original literature. for more detailed surveys, see hausman (1978), judge et al. (1980), breusch and pagan (1980), kin (1983) or engle (1982). some problems associated with the implementation of various tests are also discussed in havlik and sonnberger (1983) and maurer (1983) . for a comprehensive user-guide, see maurer et al. (1983). since the material surveyed is rather heterogenous, we do not suggest a unique testing strategy to guide the user in empirical applications. some hintson how to proceed in practice may be taken from the examples that are discussed in section 3, however. a particular strategy to discriminate between autocorrelation and misspecification is found in thursby (1981), and the complications that arise when various departures from the model assumptions occur at the same time are discussed in epps and epps (1977) and thursby (1983). following the survey of the tests, section 3 exemplifies in some details how they might be used in empirical applications. the data and some relevant statistical tables are reproduced in the appendix.;
|
|---|