Customer Reviews

Empirical Modeling in Economics: Specification and Evaluation

5 star:		(1)
4 star:		(2)
3 star:		(0)
2 star:		(0)
1 star:		(0)

 

 

This is probably the first and the last econometrics book I will have read just for fun! Although the title of the book sounds daunting, I found it very easy to read. If you are interested in intutions behind building econometric models, you should consider reading this book. A word of caution: if you want to find something that treats this subject rigorously, this may not be the book you want to buy. Just to have fun with econometrics is the general idea here.

This a good book for all those who want to understand the relationship between econometrics and economics. Without really getting into the technical description of econometrics, Granger very stylishly describes its use in economic analysis. Through wide ranging examples such as deforectration etc, Granger shows the reader how empirical modelling is achieved. This is tough thing to do as a equation is worth a thousand words, so the author sometimes strugles to keep the pace. But in the end you will definately learn something new about modelling than you have started with. And the price is worth it for you to do so. This book is good for students beginging to understand modelling. In case you are looking at serious time series forecasting and stuff you wont find it here. There are number of books for those topics, but for a non-technical understanding of empirical modelling there are just a few. And this one if one of them, and its good!

In the 1960's and 1970's Hubert Blacock, Otis Dudley Duncan, and a handful of other high-profile social science statisticians sought to introduce sociologists to econometrics. Sociology hasn't been the same since.

Quantitative work was always conspicuous in sociology, but it was more or less accessible to the relatively non-technical reader, and did not dominate upper-drawer journals the way it does today.

In contrast to economics, moreover, sociology was and remains theoretically impoverished. Economic theory may or may not be useful, but it is sophisticated and has been thoroughly mathematicized. This sharp difference between the two disciplines helps us to understand why proper specification of econometric models is even more troublesome in sociology than in economics: for good or ill, economists have formalized theory to refer to, while sociologists rely heavily on a grab bag of trendy concepts, substantive literature produced by others, and commonsense.

A primary virtue of Granger's book is that he makes clear, using brief and accessible explanations and interesting examples, that proper specification of economometric models is sometimes nearly as difficult in economics as in disciplines that are theoretically less well developed. Proper specification, as Granger makes clear, is not is not simply a matter of deduction from mathematical models of economic processes. Instead it requires the researcher to acknowledge inevitable and serious limitations in the application of mathematical models to concrete problems. In many instances these limitations are numerous and hard to recognize. The gap between theory and research, as a result, is wide.

Granger also shares the view that statistical deficiencies of econometric models, such as heteroscedasticity, autocorrelation, and non-linearity, should not be treated first and foremost as inherently statistical problems. Instead, they are often manifestations of specification errors in disguise, and should be investigated as such.

Much of what Granger has to say helps take the mystification out of econometrics. It also serves as a cautionary statement against taking econometric models too seriously, something that researchers in sociology and other disciplines have been slow to learn.

For the most part, this book is easy to read and eminently practical. Its tone is conversational, as if Granger were saying "here are some things I've learned over the years that you may find useful."

I have to admit that this is not the book I expected when I ordered it. Meaning, I suppose, that I, too, expected more of econometric modeling than it really has to offer.