Econometrics is a method now widely used in economic research. It consists in the application of modern statistical procedures to theoretical models, which have been formulated in mathematical terms. These methods are of interest in connection with the verification of economic laws and also are potentially useful for economic policy.
Part 1 of the present book gives a non-technical introduction to econometrics. Certain problems of methodology are considered, and various fields of applications are discussed with the help of examples.
Part 2 deals with various statistical procedures which come under the general heading of multivariate analysis. They may all be considered as generalizations of the classical method of least squares. These methods appear very promising for certain econometric problems and are illustrated by applications to economic data. The time series nature of the data is, however, neglected; there is no consideration of the mutual interdependence of subsequent terms of the series analyzed.
Part 3 deals with this very difficult and challenging problem. It should be realized that the methods given there are tentative and sometimes not very satisfactory from the point of view of modern statistics.
Some subjects which are connected with econometrics have been omitted. There is only incidental discussion of mathematical economics, which forms the theoretical framework of econometrics. The main emphasis is on the analysis of data which come in the form of time series. There is not much on cross-section studies, which are also potentially useful in econometrics (see, however, sections 3.4, 3.5, 3.6). The theoretical discussion of the aggregation problem is omitted, but a statistical solution is tentatively offered (section 6.3). The analysis of income distributions (Pareto distribution) is omitted. There is also no discussion of sample surveys and related purely statistical matters.
Econometrics is only one of a variety of possible approaches to the study of economics. This approach appears promising, but it is important that one realize the limitations of the method: (1) Our mathematical models are still inadequate. (2) The statistical methods are frequently based upon assumptions which are not strictly true for the date analyzed. Methods which can deal with the type of situations met with . . .