Research | Projects

VAR-BASED ANALYSIS OF DGSE MODELS FEATURING EQUILIBRIUM INDETERMINACY

Consider an econometrician who observes a time series for some economic variables of interest (e.g. inflation rate and real output), which are believed to be consistent with a given set of assumptions about the underlying generation process - i.e. a theoretical model. In general, the analyst is unaware of the nature as well as the exact sequence of the exogenous impulses (i.e. preference or technology shocks) which drive the observed series, or even their associated times. A current change in an observable might be caused by a contemporaneous impulse, but which one if there are many competitors? Also, the observed change might be due to a delayed response to shock that occurred in the past, so how to identify the timing of the impulse responses? The project aims at exploring the interplay between indeterminacy and the potential for nonfundamental equilibrium representations of DSGE models. Specifically, our research question is twofold: First, under what conditions (if any) are VARs informative about how DSGE model's dynamics respond to fundamental shocks in the presence of equilibrium indeterminacy? Second, under what conditions (if any) are checks and tests for nonfundamentalness informative about determinacy versus indeterminacy? Answering this question is relevant for both our understanding of the effects of changes in fundamentals (e.g. technology), as well as for policy prescriptions aimed at preventing the occurrence of equilibrium indeterminacy - e.g. the well-known Taylor principle in New Keynesian frameworks - as the latter is typically associated with excessive volatility in macroeconomic aggregates (e.g. output).Operationally, we aim at contributing to the extant literature on the applied economics of DSGE models by exploring conditions for the emergence of nonfundamentalness in standard frameworks featuring unanticipated shocks vis-à-vis a more recent generation of models in which anticipated or "news" shocks are found to be a major source of aggregate fluctuations. The main idea is that advance information on (or expectations about) future shifts in fundamentals can lead economic agents to adjust their intertemporal decisions, hence triggering boom and bust cycles absent any change in current fundamentals. As emphasized in the new shocks literature (e.g. Sims, 2012), shock anticipation typically involves nonfundamental reduced form representations, as it generates a critical hidden state problem. Our original goal is to show, either analytically or through simulations, that in the presence of news shocks, an indeterminate equilibrium can be consistent with a fundamental representation of the data, even when the corresponding determinate equilibrium reduced form always proves nonfundamental. As a result, statistical inference on the effects of news shocks as drawn from structural VARs need not be invalidated if the actual data generating process belongs to the indeterminate solution set of a given DSGE model.Sims, E. (2012), News, Non-Invertibility, and Structural VARs, Advances in Econometrics 28, 81-136.