Vicarelli’s Keynes and the New-Keynesian analytical method

ABSTRACT

This article provides a critical review of the foundations of the new Keynesian apparatus, to evaluate the actual strength of the Keynesian inheritance. As a reference of the Keynesian vision, Fausto Vicarelli’s recognition of the Keynesian message is adopted. The critical recognition of the new Keynesian methodology focuses on its analytical foundations, the recent extensions and on the criticality of its empirical performances and controls. The new Keynesian construction is shown to lack truly Keynesian roots in at least three key theoretical features: the role of the microfoundations, the stability of equilibrium, and expectations formation. The directions of an ongoing research narrowing the new Keynesian theoretical fracture from Keynesian economics are addressed.

KEYWORDS:

• Equilibrium
• expectations
• identification
• microfoundation
• new Keynesian economics

JEL CLASSIFICATIONS:

• B31
• B41
• E12
• C18
• C51

Acknowledgments

The author is grateful to Elton Beqiraj, Giuseppe Ciccarone, Pierluigi Ciocca, Claudio Gnesutta, Jan Kregel, Paolo Paesani, and Mario Tiberi. A special thanks to Claudio Gnesutta and the organizers of the Seminar “Keynes nel Pensiero di Fausto Vicarelli,” held at Sapienza University of Rome on November, 25th, 2016. All errors are mine.

Notes

¹For a review of Vicarelli’s contributions, see, among others, Gnesutta (2001), Ciocca et al. (1988), and Gandolfo (1999).

²A structure is in this case forced by modelling the contemporaneous correlation in the data as a system of linear relations between the estimable reduced form errors and their structural counterparts.

³The vector autoregressive component is defined by the state variables of the model and the structure being imposed is the result of the hypotheses defining the micro-foundation, the production technology, the constraints, and intertemporal optimization.

⁴Sims’ suggestion to identify the fundamental shocks through a convenient triangular factorization of the reduced-form covarariance matrix of the VAR does not solve the identification issue. The triangular factorization, needed to identify the shocks through their orthogonalization, forces a recursive contemporaneous structure in the system which, basically, is not an innocuous theoretical hypothesis.

⁵A DSGE model generally does not produce contemporaneous exclusion restrictions.

⁶In this way, the theoretical variable is conceived as a latent variable which is linked to the observed counterpart by means of the measurement error. The approximation of the likelihood function is thus obtained with the Kalman filter.

⁷A further useful characteristic of the Bayesian method is that it allows model comparison of not nested models through the comparison of the posterior marginal likelihoods.

⁸The presence of forward variables makes empirical identification an even harder task.

⁹The problem with the use of dogmatic priors could be of second order if prior constraints are derived from objective extraneous evidence. However, the standard practice is the one of using “conventional” values for some parameters that, in the absence of direct evidence, lack any empirical support (Blanchard, 2016).

¹⁰Iskrev’s (2010) analytic derivatives’ method is one of these checks. The idea is quite simple: Basically, the identification check is based on the evaluation of the model derivatives with respect to the estimated parameters and shocks.

¹¹In models belonging to the SEM tradition, the steady state is not explicitly specified. In general, it is implicit in the dynamic equations estimates, or in the static representation of the long-run equilibrium relations (co-integration).

¹²More recently, based on this approach, Comin et al. (2014) proposed a two-country model of endogenous and slow diffusion of technologies to tackle the spillover effects from developed to developing countries, taking place through new technologies embodied in traded new capital goods.

¹³The choice for the semi-endogenous specification is based on the observation that the Romer’s product variety model encounters some empirical difficulties in replicating the nonlinear relation between resources devoted to the R&D sector and growth (Bottazzi and Peri, 2007).

Additional information

Notes on contributors

Massimiliano Tancioni

Massimiliano Tancioni is with the Department of Economics and Law at Sapienza University of Rome, Rome, Italy.