Testing structural stability in macroeconometric models

Since the earliest days of macroeconometric analysis, researchers have been concerned about the appropriateness of the assumption that model parameters remain constant over long periods of time; for example see Tinbergen (1939). This concern is also central to the so-called Lucas (1976) critique which has played a central role in shaping macroeconometric analysis in the last thirty years. Lucas (1976) emphasizes the fact that the decision models of economic agents are hard to describe in terms of stable parameterizations, simply because changes in policy may change these decision models and their respective parameterization. These arguments underscore the importance of using structural stability tests as diagnostic checks for macroeconometric models. A large body of empirical macroeconomic studies provides evidence for parameter instability in a variety of macroeconomic models. For example, considerable evidence exists that the New Keynesian Phillips curve has become flat and/or less persistent in recent years; see for example Alogoskoufis and Smith (1991), Cogley and Sargent (2001), Zhang et al. (2008), Kang et al. (2009). Similarly, there is evidence that the interest rate reaction function is asymmetric over the business cycle; see for example Boivin and Giannoni (2006), Surico (2007), Benati and Surico (2008), Liu et al. (2009).