Estimating How the Macroeconomy Works

By Ray C. Fair
Harvard University Press, 2004
314 pp.

DESCRIPTION

Macroeconomics tries to describe and explain the economywide movement of prices, output, and unemployment. The field has been sharply divided among various schools, including Keynesian, monetarist, new classical, and others. It has also been split between theorists and empiricists. Ray Fair is a resolute empiricist, developing and refining methods for testing theories and models. The field cannot advance without the discipline of testing how well the models approximate the data. Using a multicountry econometric model, he examines several important questions, including what causes inflation, how monetary authorities behave and what are their stabilization limits, how large is the wealth effect on aggregate consumption, whether European monetary policy has been too restrictive, and how large are the stabilization costs to Europe of adopting the euro. He finds, among other things, little evidence for the rational expectations hypothesis and for the so-called non-accelerating inflation rate of unemployment (NAIRU) hypothesis. He also shows that the U.S. economy in the last half of the 1990s was not a "new age" economy.

PREFACE

This book presents my work in macroeconomics from 1994 to the present. It is an extension of the work in Fair (1984, 1994). The period since 1994 contains the U.S. stock market boom and what some consider to be a "new age" of high productivity growth and low inflation. It is also the period that includes the introduction of the euro. A number of chapters are directly concerned with these issues. This period is also one of continuing large advances in computer speeds, which allows much more to be done in Chapters 9–14 than could have been done earlier.

The macro theory that underlies this work is briefly outlined in Section 1.3 and discussed in more detail in Chapter 2. It was first presented in Fair (1974). The theory stresses microfoundations, and in this sense it is consistent with modern macro theory. It does not, however, assume that expectations are rational, which is contrary to much current practice. It makes a big difference whether or not one assumes that expectations are rational. If they are not rational, the Lucas critique is not likely to be a problem, and one can follow the Cowles Commission methodology outlined in Section 1.2.

The rational expectations (RE) assumption is hard to test and work with empirically. The widespread use of this assumption has moved macroeconomics away from standard econometric estimation toward calibration and matching moments. The work in this book follows the Cowles Commission methodology and is thus more empirical than much recent macro research: the data play a larger role here in influencing the specification of the model. The empirical results in this book do not support some current practices. The tests of the RE assumption in Chapter 2 are generally not supportive of it. The results discussed in Chapter 7 do not support some of the key properties of what is called the "modern-view" model. The results in Chapter 4 do not support the dynamics of the NAIRU model.

The advances in computer speeds have greatly expanded the feasibility of using stochastic simulation and bootstrapping. Chapter 9 provides an integration of stochastic simulation in macroeconomics and bootstrapping in statistics. The avail-ability of these techniques allows a way of dealing with possible non stationarity problems. If some variables are not stationary, the standard asymptotic formulas

may be poor approximations of the actual distributions, and in many cases the exact distributions can be estimated. Chapter 4 contains an example of this. The working hypothesis in this book is that variables are stationary around a deterministic trend. This assumption is not tested, but, as just noted, exact distributions are sometimes estimated. Regarding the RE assumption, the increase in computer speeds has made it computationally feasible to analyze even large scale RE models using stochastic simulation and optimal control techniques. This is discussed in Chapter 13, where a large scale RE model is analyzed.

I am indebted to many people for helpful comments on the research covered in this book. These include Don Andrews, Michael Binder, William Brainard, Don Brown, Gregory Chow, Joel Horowitz, Lutz Kilian, Andrew Levin, William Nordhaus, Adrian Pagan, David Reifschneider, Robert Shiller, and James Stock. Sigridur Benediktsdottir, Daniel Mulino, Emi Nakamura, and Jon Steinsson read the entire manuscript and made many useful suggestions.

Ray C. Fair New Haven January 2004

TABLE OF CONTENTS

REVIEW

The Fair model stands out as carefully articulated, carefully estimated, and a precious reference for those of us who like to know how the consumption function, or the investment function looks like in the data.
     — Olivier Blanchard, Department of Economics, Massachusetts Institute of Technology