آموزش و استراتژی سیاست های پولی بانک مرکزی اروپا

This paper examines the welfare implications of alternative inflation targeting proposals for the monetary policy of the European Central Bank. We assume that policy makers have to “learn” the laws of motion of inflation in an economy characterized by “stickiness” in domestic price setting behavior and subjected to recurring shocks to productivity, exports and foreign price. We find that a switch from an “asymmetric” inflation targeting strategy to an “symmetric” makes little difference in welfare payoffs, but it comes at a cost of much higher interest-rate variability. We also find that there are practically no welfare gains from switching from an inflation-targeting strategy based on the Harmonized Index of Consumer Prices (HICP) to a strategy based on the domestic price component of the HICP.

This paper examines alternative proposals for the inflation targeting program and the overall monetary strategy of the European Central Bank (ECB). In particular, we compare the current practice of asymmetric inflation targeting with the recent proposal by Svensson (2003a) for a fully symmetric inflation-targeting program. We also compare the current practice of targeting inflation in the Harmonized Index of Consumer Prices (HICP) with that of targeting inflation in just the domestic price component of the HICP. While the European Central Bank has indeed modified its asymmetric strategy to an inflation target of “below but close to 2%.” Svensson (2003b) contends that while this is a “move in the right direction,” it is “not good enough.” Agreeing with de Grauwe (2003), he asserts that the ECB has missed an opportunity “to thoroughly modernize its strategy, remove the ambiguity, and explicitly and transparently adopt flexible inflation targeting” (Svensson, 2003b, p. 2). As for the appropriate index for inflation targeting, Gali and Monacelli (2002) found, for a small open economy with sticky price setting behavior, that domestic inflation targeting dominates, from a welfare point of view, both aggregate CPI inflation targeting and an exchange-rate peg. They base their argument on the “excess smoothness” induced in the exchange rate by CPI targeting or an exchange rate peg. This smoothness, in combination with the assumed stickiness in nominal prices, prevents relative prices from adjusting “sufficiently fast,” thus causing “a significant deviation from the first best allocation” (Gali and Monacelli, 2002, p. 2). In contrast to this point, Svensson (2000) has pointed out that “all real-world inflation targeting economies are quite open economies” and “all inflation targeting economies have chosen to target the CPI inflation” (Svensson, 2000, p.155). More recently, Kara and Nelson (2002) found that for the United Kingdom, CPI inflation in the data “behaves much like domestic-goods price inflation” (Kara and Nelson, 2002, p. 22). They report that models which characterize all imported goods as intermediate goods “provides the most attractive alternatives” for understanding UK data, and argue that their evidence is “consistent with CPI inflation-targeting followed in the UK and other open economies” (Kara and Nelson, 2002, p. 22). In this paper, these alternative targeting strategies are examined using the model put forward by Smets and Wouters (2002) which is calibrated for the Euro area data. In that model, all imported goods are intermediate goods and it has both domestic and import price stickiness. However, our analysis incorporates a learning mechanism for the central bank. While there has been a wide discussion of alternative inflation targeting rules for open and closed economies, learning has, for the most part, been introduced in this literature in only one dimension, as private sector learning of the policy rule of the central bank.1 In contrast, we assume, following Sargent, 1999 and Cogley and Sargent, 2003, that the learning process is on the side of the central bank. The monetary policy authority does not know the “true laws of motion” of inflation generated by the private sector whose behavior can be described by a stochastic dynamic, nonlinear general equilibrium model, with forward-looking rational expectations. Instead the central bank has to learn about the laws of motion of inflation from past data, through continuously updated least squares regression. This information is then used to obtain an optimal interest rate feedback rule based on linear quadratic optimization, using weights in the objective function for inflation which can vary with current conditions. Such a learning framework accords more closely with real life Central Bank policy setting behavior based on approximating models of the true economy. The monetary authority is thus “boundedly rational,” in the sense of Sargent (1999), with “rational” describing the use of least squares, and “bounded” meaning model misspecification. The policy setting framework may also be viewed as an adaptation of the robust optimal control modelling framework of Hansen and Sargent (2002). Of course, this definition of learning is rather specific. Svensson (2003c) characterizes this set-up, with the central bank operating with a quadratic loss function and a view of the transmission mechanism based on linear relations, as “mean forecast targeting.” Information is thus “filtered through the forecasts,” which provides a “safe-guard against overactivism and too much response to incoming information” (Svensson, 2003c, p. 4). In contrast to Smets and Wouters, we also do not simplify the analysis by linearizing the model around steady state values. Rather, we use a nonlinear solution method, and we incorporate shocks from productivity, exports and foreign price. To anticipate results, we find only small welfare differences between an “asymmetric” or “symmetric” targeting strategy, and between overall CPI inflation or domestic inflation targeting. From a welfare point of view, these results suggest that there is little urgency, if at all, for the ECB to “modernize its strategy, remove the ambiguity, and explicitly and transparently adopt flexible inflation targeting.” Moreover, the results show that a switch from asymmetric to symmetric targeting comes at a cost of higher interest rate variability. The paper is organized as follows. The Smets and Wouter model is briefly described in Section 2, together with our model of central bank learning. Section 3 reports on the calibration and solution algorithm. The simulation results are presented in Section 4 and concluding remarks are in Section 5.

This paper has analyzed the welfare implications of one aspect of the current monetary policy strategy of the European Central Bank, namely, the use of asymmetric instead of symmetric response to inflation in a model with learning on the part of the monetary authorities as well as with Calvo-type sticky-price setting behavior for both domestic and foreign goods. We also compare the current practice of targeting inflation in using an aggregate price index with that of targeting inflation in just the domestic price component of the price index. Overall, the results suggest that switching from the practice of asymmetric to symmetric targeting can result in small welfare gains but it comes at a cost of greater variability in interest rate changes.