Policy interactions in dynamic equilibrium models

The proposed project aims at contributing to the newly emerging literature on strategic interactions between monetary and fiscal authorities in dynamic stochastic general equilibrium models. The independence of central banks from fiscal authorities in most developed countries suggests that this line of research is of both theoretical and empirical relevance. Yet, somewhat surprisingly, only very few (recent) papers have set up models in which policy authorities interact strategically in a dynamic non-cooperative game. These papers either use a deterministic framework, or assign relatively little weight to the response of fiscal and monetary policy to shocks hitting the economy. We want to fill this gap and focus our analysis on the optimal policy response to a broad variety of disturbances. We plan to start with a rather simple model and work our way towards a model with a rich economic structure. The final model should include capital accumulation, various shocks, endogenous government spending, distortionary taxes, and government myopia. Within this framework we plan to investigate several interesting topics related to the optimal policy mix over the business cycle: the optimal response to shocks, the connection between monetary policy success and fiscal discipline, the effects of physical capital accumulation on optimal policy, and the roles of monetary conservatism and fiscal myopia. Another contribution of the proposed project will be the provision of computer code that allows solving stochastic dynamic games between monetary and fiscal authorities. In models with many state variables, standard Projection methods are likely to fail due to the curse of dimensionality. We will thus apply alternative computational methods to solve our models. The computer code will be written in the popular programming language MATLAB, such that it will be easy for other researchers to adopt the algorithms to related problems.

The original goal of this project was to study the role of strategic interactions between different macroeconomic policy makers, notably between the government and the central bank of a country. We were planning to analyze the implications of these interactions in a model that includes many realistic features like capital accumulation, imperfect competition, distortionary taxation, various types of stochastic shocks, and discretionary policy making (i.e., the inability of policy makers to make binding commitments for future policy). Although we have not been able to derive results for this general setting, we have obtained a few important insights into the underlying mechanisms and we have developed an efficient algorithm to numerically analyze these and similar questions in medium-scale models. Our first basic insight is that there are significant differences between policy making under commitment and discretionary policy making. For example, in an economy in which policy makers have discretion, the so-called Friedman rule (a monetary policy designed to avoid distortions created by non-zero nominal interest rates) turns out to be optimal provided that prices adjust sufficiently fast. This result holds independently of the degree of competition on product markets. Furthermore, the business cycle properties generated by discretionary policy making are markedly different from those under perfect commitment: whereas taxes and public debt under discretion typically revert back to their respective long-run mean values after a shock, this is hardly observable under commitment. The second basic insight is that in a model where commitment is ruled out, the choice of the monetary policy instrument (nominal interest rate or nominal money growth rate) is of crucial importance. More specifically, our preliminary findings indicate that the two policy instruments reflect differing degrees of `implicit` commitment power, which becomes relevant only if there is strategic interaction between the two authorities. If the central bank chooses the interest rate, the government loses its control over the future public debt level, whereas this is not the case if the central bank chooses the money growth rate. The nature of the models used for our analysis, in particular the strategic interaction of policy makers without commitment power, makes the application of standard numerical algorithms based on local approximations very difficult, if not impossible. Instead, one needs to solve these models globally, that is, the behavior of all agents in the economy (including the policy makers) has to be determined for all conceivable states. The algorithm developed in the course of this project is such a global algorithm which is roughly comparable in speed and accuracy to the best available global methods but it is considerably easier to implement.