Customer Reviews

Econometrics

5 star:		(18)
4 star:		(0)
3 star:		(0)
2 star:		(0)
1 star:		(1)

 

 

The title is accurate. This is 'Econometrics' done properly.

Unlike the vast majority of econometrics texts, this book combines solid economic theory with a thorough grounding in basic mathematical statistics. The worked practical examples introduce some classic empirical papers and provide an excellent motivation to study the theory. The little bit of effort required to put on the GMM-tinged glasses makes everything look so much clearer.

The incidental treatment of linear algebra is better than that of any 'Math. for Economists' book I have ever come across.

Anyone at the graduate level of study should own this book. It would also be of value to advanced undergraduates, and out-of-touch academics.

It should be noted that an important aspect of statistical analysis is the treatment of 'outliers', such as the 'review' presented by Mr. Brian J. Phillips.

I have a love/hate relationship with this book. Perhaps I should state as a precursor that I was never formally trained in economics before learning econometrics. And, that the last time I'd done matrix algebra or calculus was some 15 years prior.

We used this book as part of a taught graduate course. It took half a semester to go through the first two chapters - an investment of time that proved well worth it for the rest of the topics which were covered in the remainder of the semester.

Basically, if you can understand the first two chapters on ordinary least square regression for finite and large samples, the required assumptions and properties, then the rest of the chapters are a piece of cake:
- generalized method of moments for single and multiple equations
- panel data
- time series analysis (including unit root analysis)
- extremum estimators
- maximum likelihood
- cointegration.

In short, the book covers all major econometrics topics and does so in a succinct, clear manner. The way in which Hayashi builds on each topic, showing that all models are basically different versions of the same method, with slightly different assumptions is just brilliant. It put statistics in a different light for me, and gave me a much deeper, intuitive understanding of it than any other book or class had done before.

There is a caveat however. This book assumes that you have substantial mathematical grounding. In particular, I found the succinct use of notation, without any verbal explanation, irritating at first. I invested quite some time in a mathematical economics book reminding myself what sets were, rules of matrices, calculus functions, expectations and probability.

Without the support and input of our brilliant teacher who (very patiently) took us through the end of chapter exercises step-by-step, I would never have managed to successfully read this book on my own! While those exercises honed my skills and deepened my understanding, I relied heavily on Hayashi's home page notes and hints to complete them.

For those of you that have strong mathematical skills and an economic background, this book is probably one of the best introductions to econometrics. For those of you who do not, it will prove to be a difficult read at best.

What's certain is that after succesfully completing it, your econometrics and statistical skills will provide a solid enough basis for any graduate program.

I think Hayashi is the best econometrics textbook to come along in a long time. The treatment has that rare quality of being simultaneously sophisticated yet very easy to follow. In that sense, this book is much different than Greene - whereas Greene is (I think) much more of a reference, you can actually sit down and learn a lot of econometrics with this book. Hayashi not only takes the time to explain key concepts in good prose, but in some cases even writes down step-by-step instructions. All this while not compromising the material.

The treatment is also slightly different in that GMM is a central theme instead of something off to the side, which is very nice. There are plenty of empirical examples - these are somewhat helpful, and the exercises are fairly easy but still illustrative.

Two downsides - it would have been nice to see some treatment of Bayesian econometrics, since this appears to be used much more widely (Lancaster is a good supplement). Second, either I got a faulty book, or there are no tables of critical values. This is ultimately a minor gripe since just about every other book has tables (and you really don't even need them these days with packages and such), but it can be annoying.

Ultimately, the combination of sophistication and readability of this book is what sets it apart from all others. If you're looking to learn econometrics, buy this book.

This book has excellent qualities:
1) Its clarity and concision, in exposition and proofs.
2) The modern approach and well structured and complete contents
3) The empirical exercises

But, for pure theory I prefer Davidson & McKinnon's "Estimation and inference", and for an empirical approach Berndt's "Practice of Econometrics".

Hayashi's is a good intermediate alternative, with great theory exposition and good empirical exercises.

One of the best options for a graduate student.

This is a fine book, but probably not the one you want to buy if you are looking for ONE all-encompassing reference. The approach is interesting, but unusual, with all the pros and cons that come with originality. It is for graduate students, or very advanced undergraduates, as it requires quite a lot of previous knowledge of linear algebra and statistics.
What is unusual about this book is that it covers most topics within a unifying Generalized Method of Moments (GMM) framework. Many many estimators are treated as special cases of GMM. The book is clear, and the notation is mostly OK, even if the chapters on panel data and systems of simultaneous equations are a notational nightmare, partly because of the choice of treating everything in a GMM framework. Another unusual aspect of this book is the emphasis on certain regularity conditions (such as ergodicity) that are usually used in a time-series framework, but are not commonly seen in cross-section analysis. I studied (also) on this book as a graduate student, and overall I liked it. The only real minus are the exercises, which contain so many hints that they become trivial (really, I am not a genius...). Worse, they only require mindless application of linear algebra.

One UNimporant cons of this book is the fact that (cover aside) it is... ugly! How could the publisher choose the boring "Times New Roman" font for this book!? But this, of course, does not really matter....

Overall, a useful and good book, but if you are looking for ONE textbook in cross-section econometrics Wooldridge is probably a better choice, and if what you are looking for is ONE book in time series, Hamilton is likely to be what you want on your shelf.

Yes, indeed I also think this one is the best around. Some points I'd add are:

Hayashi's book is the only econometrics textbook I am aware of (IMHO, these are certainly not all, but quite a few) that is truly complete. Not only from a material point of view - it covers both time series and cross section material in a unified framework (as opposed to Wooldridge). The empirical exercises are useful (outperforming, say, Davidson/MacKinnon) and yet you can also go on to read theoretical papers or Amemiya after having read Hayashi's coverage of asymptotic properties of GMM and M estimators. Finally, also more towards the theoretical side, it provides some first training in hands-on programming, as opposed to some books that tell you where to click in eViews.

Furthermore:

- A corollary of the above comments is of course that it's not very detailed in each of the fields covered. E.g., all you find on limited dependent variables is Probit, Logit, Tobit.
- It's well managed: the typo list (which isn't very long, I should emphasize, especially for a 1st ed.) is always up to date
- The typsetting really is debatable. I find it unpleasant to have italics, boldface, boldface italics, verbatim environments etc all on one page. Less is more.

A nice review of basic econometrics with detailed explanation of the GMM. Clearly the other reviewer does not know anything about what he is talking about.

This book will definitely be a classic. It's clearly written, well structured and covers a very modern approach to econometrics. In my opinion it's by far the best graduate econometrics text available.

Seems to me that Brian Phillips wouldn't know a good econometrics book if you hit him over the head with it. I have no hesitations in recommending this text.

What a great book it is! I saw two undergraduate book such as gujarati, johnston. But I can't understand what econometrics is.Hayashi put me on the right road to Econometrics.
I think it suffices to kwow some matrix and algebra concepts for reading Hayashi's econometrics! He explains every concepts clearly in plain english.
Thanks to Joon-yong Park(SNU) and Hayashi(U of Tokyo) ^^

Over the past four or five decades, econometric methods have been borrowed and used more or less effectively by social scientists in a broad range of disciplines. Generally, though certainly not in ever case, those who use econometric methods in other social sciences are not as well trained in mathematics as economists, and they have little or no knowledge of economic theory.

To meet the demand for accessible econometric literature in other disciplines, authors and publishers have produced textbooks that are much less mathematically demanding than the staple sources. Examples include Wooldridge's Introductory Econometrics, Gujarati's Essentials of Econometrics, Stockman and Watson's Introductory Econometrics, and Mirer's Economic Statistics and Econometrics. Also, Peter Kennedy's Guide to Econometrics is an accessible catalog of tests and correctives for violation of assumptions, provided the non-specialists stay out of the technical appendices.

However, Hayashi's Econometrics clearly does not belong in the category of textbooks that appeal to a broad-based audience of social scientists. Hayashi, quite rightly, has a different audience in mind, and he assumes that the reader knows and has facility in applying the mathematics that is legitimately expected of economists. He also liberally incorporates economic theory into his presentation. While the econometric texts mentioned above lean heavily on OLS estimators, Hayashi treats OLS as just a special case of the generalized method of moments, a concept that is entirely alien to most students and practitioners who are not well schooled in the mathematical methods of economics.

One consequence is that social scientists who are not cautious in selection of self-instructional materials order Hayashi's text and are then dumbfounded when they encounter mathematics and economic theory that they are not prepared to handle. Sure, they could take the time and effort needed to get up to speed, but most folks are in a hurry and have more pressing concerns for released time and sabbaticals.

I was introduced to econometrics in an independent study thirty-five years ago. I was assigned sections from the first edition of Johnston's Econometric Methods and Kmenta's Elements of Econometrics, as well as few journal articles, and was then on my own. As a result, I can sympathize with other non-economists' frustration. Johnston and Kmenta may seem pretty accessible now, but at the time they were regarded as difficult. I initially preferred Kmenta because his first edition (as well as his second) did not require practiced facility with matrix algebra. I'm still embarrassed to recall, however, how mystified I was when Kmenta began introducing alternatives to OLS estimators. I didn't know there were any! I eventually learned a lot, but making sense of Johnston forced me teach myself the rudiments of matrix algebra.

I know that textbook publishing is a racket, just look at the prices. But I think that publishers have an obligation to very conspicuously apprise prospective readers that a text like Hayashi's understandably makes demands on the reader that non-economists typically will not meet. Otherwise, social scientists who might use applied econometrics in their less mathematically-intensive disciplines will waste money and be deprived of statistical tools they might put to good use.