The Interaction of Events and Ideas and the Role of Empirical Macroeconomics

One of the reasons for the success of macroeconomics in the past 80 years is that it has always been closely connected to real developments. In short, macroeconomics has always been ultimately empirical, and its progress owes much to the interaction of events and ideas.

The econometric tradition, which initially appeared independently of the General Theory, was extended in various directions by econometricians, such as Tinbergen [1937], Haavelmo [1944], and the Cowles Commission and Burns and Mitchell [1946]. It resulted in the development of early Keynesian macroeconometric models by Klein [1950], Klein and Goldberger [1955], and others. These econometric models were used both for prediction and policy analysis.

In addition, many of the major theoretical developments in macroeconomics (such as the microeconomic foundations of the consumption function, the investment function, money demand, and the Phillips curve) were motivated by empirical as well as theoretical considerations. The breakdown of the original Phillips curve is perhaps the most celebrated example. Theoretical advancements were usually tested econometrically by single-equation methods and eventually incorporated into large-scale econometric models.

Empirical macroeconomics changed dramatically following the Lucas [1976] critique and the gradual emergence of DSGE models. One of the approaches that emerged, based on Sims [1980], was entirely atheoretical and sought to model the data-generating process of macroeconomic variables using vector autoregressions.

The second approach, based on Prescott [1986], was based on the calibration of tightly specified DSGE models, the generation of some statistical moments implied by these models through stochastic simulations, and the comparison of these statistical moments to the statistical moments of actual macroeconomic variables.

These latter models, which combine calibration and Bayesian estimation, are increasingly used by central banks. At the same time, more traditional econometric models, with minor improvements in the treatment of expectations, continued to be used outside academia.⁹

Thus, the modern approach to empirical macroeconomics is a combination of the atheoretical and the structural approaches. The methodology can be briefly described as follows. DSGE models are treated as dynamic systems of equations. Their parameters are nonlinear functions of deep structural parameters. A combination of calibration and Bayesian estimation methods is utilized to derive empirical versions of the models, as maximum likelihood estimation is usually infeasible. VARs are used to describe the properties of the data before a model is estimated and to compare the impulse response functions of the empirical structural model with that of the VAR representation.¹⁰

However, the rules applied in empirical academic research in macroeconomics may be unnecessarily restrictive. As pointed out by Blanchard [2009, p. 225]:

A macroeconomic article today often follows strict, haiku-like rules. It starts from a general equilibrium structure, in which individuals maximize the expected present value of utility, firms maximize their value, and markets clear. Then, it introduces a twist, be it an imperfection or the closing of a particular set of markets, and works out the general equilibrium implications. It then performs a numerical simulation based on calibration, showing that the model performs well. It ends with a welfare assessment.

Blanchard [2018] and others have argued in favor of a partial relaxation of these strict rules in the direction of encouraging more empirical research in a partial equilibrium setting, so as to investigate the relevance of particular hypotheses used as building blocks of DSGE models.

23.3