Some notes on problematic issues in DSGE models

Abstract

We review some of the problematic issues in DSGE models, which are currently much discussed in the economics profession. All of these issues are concerned with the DSGE models’ (in)ability to match aspects of macroeconomic variables’ observed behaviour. The optimizing agents framework implies that Ricardian equivalence typically holds, which is clearly at odds with the empirical evidence. A distinguishing feature of DSGE models is the assumption that structural parameters are invariant to policy changes. We argue that not all of them can be considered independent from economic policy. It is typical for DSGE models that agents form rational expectations, which can be considered unrealistic. The typical procedure for estimating a DSGE model is to use revised data. As some empirical studies suggest, a model’s behaviour may be different if real-time data are considered. It is also usually assumed that the monetary authority uses the interest rate as a tool of monetary policy. Nowadays, nominal interest rates are close to zero in many economies and cannot be lowered further.

Keywords

• DSGE model
• fiscal policy
• structural parameters
• rational expectations
• learning algorithm
• real-time data
• zero lower bound