تثبیت نرخ ارز در ERM: شناسایی واکنش سیاست پولی اروپا

We develop the structural VAR models for European countries (France, Denmark, and Germany) during the ERM period, to examine the monetary policy reactions, especially the within-ERM exchange rate stabilization. In addition to analyzing the impulse responses and variance decomposition, we present the full monetary reaction function in an interpretable manner, without introducing additional assumptions. The results suggest some asymmetry in exchange rate stabilization (non-German exchange rate stabilization is stronger than German exchange rate stabilization), but do not support the hypothesis that German monetary policy was independent of the ERM.

This paper develops the structural VAR models for European countries (France, Denmark, and Germany)1 during the ERM period to examine the monetary policy reactions, especially the stabilization of within-ERM exchange rates. We improve on the previous studies on identifying monetary policy actions for European countries (Sims, 1992, Grilli and Roubini, 1994, Kim, 1999 and Kim and Roubini, 2000) by considering monetary policy interactions within the ERM based on the asymmetry hypothesis; the Bundesbank, as the leader of the ERM, set its own monetary policy, possibly by reacting to the exchange rate against the other ERM currencies as a whole, while each non-German central bank pegged its exchange rate against the D-Mark or reacted to the exchange rate against the D-Mark. Based on this asymmetry, we develop different models for Germany and non-German countries to identify monetary policy actions. We include the exchange rate against the D-Mark and the German interest rate explicitly for the non-German model, while we include the exchange rate against the ECU for the German model. We allow the contemporaneous monetary policy reactions to the exchange rate for the non-German model, while we do not for the German model. But we do not impose any further restrictions on the lagged policy reactions in both models. These models deliver reasonable dynamic responses to the monetary policy shocks. Using the models, we also identify systematic monetary policy reactions (in addition to monetary policy shocks) to answer some interesting questions on the European monetary policy reactions or interactions. The first question is the general evidence on the within-ERM exchange rate stabilization of each European country. Did each country stabilize the within-ERM exchange rate? In particular, did Germany stabilize the within-ERM exchange rate? Did the monetary authority of each country react to the shocks destabilizing exchange rates? Did non-German countries react to the German interest rate shocks? The second question is the relative size, speed, and the persistency of the stabilization among countries. Does each European country stabilize the exchange rate to the same degree? In particular, did Germany and non-German countries stabilize the exchange rate to the same degrees? How quickly did each monetary authority react to stabilize the exchange rate? Did each monetary authority also stabilize domestic variables? In particular, did non-German monetary authorities stabilize domestic variables? Was the stabilization persistent? To infer the monetary policy reactions from the VAR models, first, we analyze impulse responses and variance decomposition of the monetary instrument, and the exchange rate to the shocks destabilizing the exchange rate. Second, we present the full monetary policy reaction function in an interpretable manner. From the model, we recover the monetary authority’s reactions of the monetary instrument to changes in each variable over time, especially to changes in the exchange rate (and the German interest rate for the non-German countries). Many previous studies which attempted to answer similar questions examined simple relations among a few monetary variables (such as Granger–Causality or/and Cointegration), for example, Cohen and Wyplosz, 1989, DeGrauwe, 1989, Karfakis and Moschos, 1990 and Katsimbris and Miller, 1993, and Hassapis et al. (1999). However, those studies are limited in answering all these questions, since they did not present the full monetary policy reaction function formally. There are some past studies that presented the monetary reaction function, for example, Artus et al., 1991 and Fratianni and von Hagen, 1990, and von Hagen and Fratianni (1990). However, as in most past studies on monetary reaction function, those studies assumed that the monetary instrument is exogenous to all other domestic variables, which is difficult to justify. In this paper, we recover monetary reactions from the model, allowing interactions between monetary instrument and other domestic variables (in addition to exchange rate and the foreign monetary instrument) by connecting to recent VAR literature on identifying monetary policy actions.2 From the viewpoint of the reaction function literature, this paper presents an interesting way to recover the monetary reaction function from the structural VAR models. A recent study by Clarida et al. (1997) suggested a method of recovering the reaction function from the structural VAR models, but introduced additional assumptions (postulating a Taylor type rule based on the forecasts from the underlying structural VAR model). Another important study by Clarida et al. (1997) estimated the monetary reaction function using the GMM method, but it is subject to the simultaneity problem since they used a single equation method. In contrast to these studies, we directly recover the monetary policy reaction function from the structural VAR model (which is not subject to the simultaneity problem as in the single equation estimation) without introducing additional assumptions.3 Even though our model has a built-in asymmetry, it encompasses different degrees of the asymmetry in exchange rate stabilization which have been suggested by past studies, since the built-in asymmetry does not necessarily imply the asymmetry in the degree of exchange rate stabilization. Some studies, for example, Giavazzi and Giovannini, 1987, Wyplosz, 1989 and Artus et al., 1991, and Uctum (1999), supported the ‘strong’ asymmetry hypothesis that the Bundesbank set monetary policy independently of the within-ERM exchange rate and the others pegged their exchange rates against the D-Mark. But others, for example, Fratianni and von Hagen, 1990, von Hagen and Fratianni, 1990 and Cohen and Wyplosz, 1989, (8), and Hassapis et al. (1999), supported the ‘weak’ asymmetry hypothesis that the Bundesbank also stabilized the within-ERM exchange rate, though other central banks stabilized the within-ERM exchange rate more actively. Our model comprises both types of asymmetries, and can be used to examine the degree of the asymmetry, or the degree of the exchange rate stabilization, provided by the Bundesbank and other monetary authorities in the ERM, since most parameters of the monetary reaction function in our model are not constrained.4 Section 2 summarizes the structural VAR modeling. Section 3 presents the basic models and confirms the plausibility of the model by examining the impulse responses to monetary policy shocks. Section 4 infers monetary policy reactions based on variance decomposition and impulse responses. Section 5 presents the full monetary policy reaction function in an interpretable manner. Section 6 develops extended models and examines robustness of the results. Section 7 summarizes the findings.

We develop the models for European monetary policy actions based on the asymmetry within the ERM, which comprises different degrees of the asymmetry in exchange rate stabilization, suggested by past studies. Based on the models, we examine monetary policy reactions, especially within-ERM exchange rate stabilization of France, Denmark, and Germany. The impulse responses to shocks destabilizing exchange rate suggest that all countries stabilized the within-ERM exchange rate. They stabilized the exchange rate against the disturbances in the foreign currency market. In addition, within several months, France and Denmark almost fully stabilized the exchange rate in response to the German interest rate shocks, which represent the German monetary policy actions (both systematic and non-systematic). The estimated full monetary reaction function also suggests that all countries changed the monetary instrument to stabilize the exchange rate, though it was stronger in Denmark and France than in Germany. In addition, Denmark and France fully followed the German interest rate movements within a short period. The monetary reaction function also reveals that Germany stabilized domestic variables, such as output and price, but France did not. From these results, we can draw a general picture of exchange rate stabilization within the ERM. First of all, as consistent with the conventional view, the asymmetry of the exchange rate stabilization was present in the ERM. 1) The monetary reaction to exchange rate was stronger in the non-German countries than in Germany. 2) Non-German countries (especially, France) did not stabilize domestic variables significantly, while Germany actively stabilized them. 3) Non-German monetary authorities fully followed the Bundesbank’s interest rate changes. However, Germany also stabilized the within-ERM exchange rate, though the degree of stabilization was weaker than non-German countries. It suggests that the German side of the ‘strong’ asymmetry hypothesis does not hold, that is, German monetary policy was not independent of the ERM.