Where theory and reality meet

As is well known (and often frowned upon), the work of theoretical economists can be very abstract and, as a consequence, it is felt, rather removed from reality (whatever that may mean). Furthermore, economic theory in its purest and most general form does not give much by way of exact predictions about what happens in the real and complex world. The conclusions it reaches are not always informative regarding what one may want to know or a very helpful guide for policy. Indeed, except in very specific cases, most of the interesting questions that a government, a firm or an individual may ask an economist will be answered with the famous "it depends". Even relatively simple questions such as "what is the effect of the introduction of a minimum wage?" or "what is the average effect of some direct or indirect tax change?" have been answered with "it depends". This answer should not be interpreted as evidence of a lack of explanatory power of mainstream economics. In fact, this type of answer can be very informative. Often, the process of going through all of the conditions of an answer that starts with "it depends" will reveal a range of possibilities that may not have been present at the outset in the questioner's mind. The root cause of all these possible regimes lies in the uncertainty and difficulties economists have with the quantitative measurement of relevant economic quantities and concepts. In this context, the raison d'être of applied econometrics is to provide reliable and theoretically sound measurement methods and measures so that theoretical and empirical debates within the profession and (mostly) policy debates outside the profession are enlightened about the empirical nature of the "world" in an economic sense.

Applied econometrics is a difficult discipline. Being between theory and the real world, one is faced with many difficulties, all of them seemingly insurmountable. Even when an area is identified as worthy of study (and to do so is not as straightforward as it sounds), many aspects conspire to make the applied econometrician's work quite a struggle. The nature of the data and the nature of theoretical economic models provide ample opposition to a successful investigation, and both need to be properly recognised and addressed. In my experience, the reason why any applied work fails to provide useful answers, even to perfectly legitimate questions, lies in the failure to observe any of the latter. Hence, two volumes dedicated to applied econometrics and its models, methods and difficulties are a godsend. If I properly understand the intentions of the editors, more will follow. If you are an applied econometrician, this is good news.

The first volume of the Handbook of Applied Econometrics surveys mostly macro-economic applications, while the second volume is concerned with micro-economic applications. Both are collections of work by recognised academics. All the topics relate to the various activities an applied economist has to master: modelling and definition of the parameter of interest; identification of the parameters per se, given the limitations imposed by the data; the choice of the appropriate estimation method; and inference and, when relevant, forecast or simulation of economic policy. The material in both volumes demands a good command of modern analytical methods and the ideas of (mainstream) economics and econometrics, and the willingness to read with pen and paper nearby. In the main, the presentation is technical and does not allow for a cursory read. So far as I can make out, neither data nor computer code is provided with any of the chapters. Clearly, the books are addressed to well-trained graduate students and professional economists; amateurs should abstain.

A leading theme is the importance of economic theory throughout the practice of applied econometrics, first to set up the model and think of the data requirements, and second to understand and interpret the results. For example, the theoretical methods of the rational expectations (RE) revolution can be seen to have had a large effect on the modern applied econometrics of the dynamic behaviour of individuals and/or economies. M. Binder and M. H. Pesaran dedicate a chapter to multivariate rational expectations and macro-economic modelling. It does not stop there. The idea of RE appears either as a modelling tool, an interpretation device or a source of criticism in most of the chapters that are concerned with the introduction of dynamic elements in an uncertain world. This amounts to more than half of the themes discussed in the two volumes. Theoretical concerns are present in more static contexts too. A good understanding of theoretical arguments stemming from micro-economic theory is essential if the reader is fully to appreciate what is involved when the object of the applied work is to study firm technologies or individual preferences. For example, the discussion by A. Lewbel on the specification of consumer-demand systems relies only on results from micro-economic theory, some relatively recent. Economic theory does not only play a role in determining what is of interest, it also plays a role in determining what cannot be meaningfully measured. Follow, for example, Lewbel's account of the difficulties with identification of household-equivalence scales.

Obviously, since the objective of the Handbook is to report on the state of the art in applied econometrics, many of the discussions revolve around the several stages necessary for an empirical piece of work to reach a satisfactory conclusion. The choice of an estimation method (full information likelihood, generalised method of moments, calibration or non-parametric procedures), the potential numerical difficulties one is likely to encounter - in some cases how to extend the model and account for unobserved heterogeneity and/or measurement error - and how to evaluate the precision of the estimates, are all discussed with varying levels of detail, but extensive reference to the relevant literature is always available. For example, K. West's chapter on inventories, R. Miller's chapter on models of dynamic optimisation with micro-economic data, or G. Neumann on search models and duration data are very thorough accounts of what is involved in modern empirical work. Although the classical linear model is still an important tool for applied econometricians, what is shown here is that, depending on the context, more appropriate techniques and models are available and should be used. However, their use does not guarantee success; making sense of data is a lot more difficult than that. New PhD students would benefit from reading and pondering on Miller's concluding comments.

J. Muellbauer and R. Lattimore's excellent piece on the consumption function provides a good example of the contribution of applied econometrics to a fundamental debate. This chapter starts with the microeconomic foundations and slowly builds a more complex model of consumption behaviour over the life cycle (model with credit constraints, model without the separability of consumption and leisure, model without time separable preferences, and so on). The presentation then considers the issue of aggregation and hence discusses issues of a more macro-economic nature. Many testable hypotheses are surveyed throughout, and possible testing strategies are presented. The chapter concludes with a summary of the findings so far and presents possible interesting developments for future research. Clearly this chapter alone could be the basis for a challenging and interesting graduate course in applied econometrics.

On the whole, this is a very useful (and, in paperback, relatively inexpensive) collection of survey articles that should figure on the bookshelves of any applied econometrician or applied economist. Of course many topics absent in the present selection deserve at least one chapter. Personally I would quite like to read several contributions on empirical labour economics. In any case, if more volumes are to follow, I would like to have the chance to try my skills on the exact data that illustrate each chapter. A collection of Handbooks of Applied Econometrics would seem pointless otherwise.

Gauthier Lanot is lecturer in economics, University of Keele.