Econometric Modelling: A Likelihood Approach and Introduction to Econometrics

These introductory texts take very different approaches in what is a competitive market.

Gary Koop's book begins with the basics and then discusses serial correlation, heteroscedasticity and the instrumental variables estimator. The second half looks at more specialist topics: time series including stationarity, cointegration and VAR modelling; panel data; limited dependent variables; and Bayesian econometrics. It is well laid out, with chapter summaries, technical appendices and worked examples, such as a limited-dependent variable analysis on "choosing to have an affair".

David Hendry and Bent Nielsen's book takes a likelihood approach, beginning with the Bernoulli model, followed by the standard and logit models. This is odd, as it deals with extensions to the basic approach before it has been fully covered. But the basic material is subsequently explained well, and one third of the way through the matrix algebra of regression, analysis is introduced. There is extensive analysis of time series subjects, but scant coverage of issues such as panel data and limited-dependent variables.

The approach is unusual in that many of the chapters are focused on a specific piece of research. For example, chapter 12, which introduces us to the topics of autoregression and stationarity, is based on an analysis of the price of fish in a New York market. This approach places more emphasis on economic theory than is usual. It is thus econometrics in the round, a fundamental part of which is the formalisation of a model.

The books are very different. Koop's text is the more introductory but offers a more comprehensive analysis of topics. Although reasonably technical, it is more user-friendly, with highlighted sections, chapter summaries and better questions. Its main fault lies in the non-matrix algebra treatment. Matrix algebra is not only the basic language of econometrics, it also lends itself to a simpler exposition than the algebraic alternative.

Hendry and Nielsen's somewhat unusual data-driven approach works well. It is limited in its topic coverage, but what it does, it does well, providing genuine insights at a reasonably advanced level.

Who is it for? Koop's is primarily a final-year undergraduate text, Hendry and Nielsen's is suitable for both the advanced undergraduate and graduate markets.

Presentation: Koop's is a well laid-out textbook, but Hendry and Nielsen's is less well assembled.

Any extras? Koop's has a website with data sets, problems and PowerPoint slides. Hendry and Nielsen's website offers data sets suitable for use with econometrics software such as STATA, as well as the Ox-metrics suite.

Would you recommend it? Neither of these will be on my references as a main text, but both will figure prominently on the reading list, albeit for different reasons.