Inflation targeting: a view from the ECB

Continued from last page

To exemplify, let us begin by assuming that the only objective of policy is to maintain price stability. If prices move in tandem with the existing tension on employable resources, the policy goal of price stability dictates keeping the economy continuously close to its potential. Under these circumstances, reacting to a pure inflation forecast figure, with no reference to any additional indicator of macroeconomic performance, would be a recipe for policy mismanagement any time the economy is hit by a transitory (say, favorable) supply shock. This type of shock would entail both a downward blip in forecast inflation and an upward movement of output away from its sustainable level. Hence, restricting the central bank's information set solely to the inflation forecast figure--to the exclusion of a broader suite of other indicators, which help discriminate between supply and demand shocks--would, in this situation, call for a policy easing today, which could destabilize the economy by laying the ground for the build-up of inflationary pressures tomorrow.

This is an elementary example of the general proposition that inflation forecasts alone are not sufficient to reveal the nature of the threat to price stability and that it would therefore be misleading to follow a rule that requires setting the policy instrument simply as the function of a forecast. Even in an ideal world in which the models producing the forecast are properly specified, the policymakers are not interested in the result of the forecast per se but instead aim at a consistent economic picture--or, to put it differently, they aim at identifying the relevant shocks underlying the forecasts and how different types of disturbances to the economy imply different kinds of policy responses. The relation between forecasts and underlying shocks is clearly one-to-one in many simple stylized models used in the monetary policy literature, but this relation clearly breaks down once we depart from that simple set-up. So, once again, forecasts of a few macro-variables cannot be sufficient statistics to determine monetary policy action.

Target rules are somehow immune from the above problem, given that they are routinely implemented by producing forecasts of future inflation and output conditional on the path of the policy instrument and searching for the path, which minimizes a proper loss function. Consequently, when evaluating inflation targeting in the context of target rules, the discussion should primarily focus on the structural model used to define the central bank's constrained optimization problem. In other words, an evaluation of the target rule characterization of inflation targeting is largely equivalent to an evaluation of the economic model employed to derive that rule. (13) One criticism of the models underlying most target rule characterizations of inflation targeting is that they neglect any role that might be played by the monetary aggregate or financial frictions in the determination of price developments. This opens the way to a second set of remarks on the issue of model misspecification and the robustness of the rules.

Robustness and Model Misspecification

The possible presence of model misspecification is something that economists and econometricians have some difficulty in acknowledging. However, every model we write down and estimate contains some form of shortcut and approximation. This uncertainty is worsened since economists have not yet agreed upon a proper, commonly accepted approximating model. This implies that the appropriateness of a monetary policy strategy cannot be evaluated only within a particular class of model--rather a good strategy has to perform well across a variety of empirically plausible models.

However, most advocates of inflation targeting--at least those referring to simple rules for monetary policy decisions--ultimately rely on a view of the economy whose essence can be captured by no more than three equations. The defining characteristics of these equations are (i) staggered pricing, (ii) the centrality of the output gap (or Phillips curve), and (iii) the notion that monetary impulses propagate primarily via a price (interest rate) channel, with monetary quantities playing no role.

The presence of only a market for goods and the absence of a fully formalized market for assets whose supply is inelastic in the short run implies that money has no role other than to facilitate the exchange of goods. Decisions about money holdings are not seen as part of a wider portfolio decision that--at times--may lead households to prefer liquidity over risky assets. For example, a positive change in money demand has no counterpart in an excess supply of some other asset. On the contrary, if truly alternative assets were to exist whose issuance was related to their private issuers' investment decisions and capital formation, then generating a higher (or lower) supply of money--at any given interest rate--could become all but inconsequential.

Quoting McCallum (2001 a), there is "nothing fundamentally misguided" about the model used by advocates of inflation targeting. Such a model is internally consistent and elegant. It can also mimic the observed behavior of modern economies in "normal" circumstances. Yet it rests upon what can certainly be regarded as extreme assumptions about the role of money in the economy. A central bank can legitimately question the usefulness of a model for monetary policy-setting in which money has been deprived of its basic liquidity--or, equivalently, its "store-of-value'--function that generations of scholars have recognized and discussed for decades (cf. Hahn, 1990, inter alia).

Levin, Wieland, and Williams (1999, 2001) demonstrate that Taylor-like instrument rules perform quite robustly in a particular set of macroeconomic models. However, this robustness does not survive a broadening of the suite of candidate models beyond those considered in these papers. Suitably parameterized Taylor-like rules appear to work well in stabilizing the economy within the confines of the mainstream New Keynesian paradigm, in which money market equilibrium conditions are redundant. This last assumption, in particular, proves to be absolutely crucial. If financial markets are not free of frictions, then Taylor-like rules often do not prove to be robust and yield suboptimal outcomes.

Examples of financial market frictions are prominent in transmission mechanism literature, e.g., in so-called limited participation models (Christiano and Eichenbaum, 1992) or segmented markets models (Alvarez, Lucas, and Weber 2001). Within the class of limited participation models, Christiano and Gust (1999) show that the set of parameters under which a Taylor-like inflation-targeting rule becomes a source of instability is much broader than for mainstream New Keynesian models.

More recently, Alvarez, Lucas, and Weber (2001) presented some experiments on the stabilization properties of simple Taylor rules within a segmented financial markets model. They conclude that central banks pursuing a Taylor-like interest rate instrument rule--by systematically ignoring money market (velocity) shocks--censor the information set available to policymakers and thereby reduce the effectiveness of their responses to economic shocks by arbitrarily excluding relevant monetary information from the policy decision. In a similar vein, but in a different class of model, Christiano and Rostagno (2001) show that a Taylor-like interest rate instrument rule can generate equilibria with undesirable properties; this outcome could be avoided by a policy rule that takes into account the information provided by monetary aggregates.

It should be clear that, from the view point of a central bank, a serious attempt should be made to construct a model where shocks to velocity are treated appropriately within the context of broader portfolio shifts, possibly in the presence of (changing) risk assessments. Unless disturbances in money holdings are formalized in such a way as to reflect financial decisions, then nothing can be said about the role of money in the business cycle and insufficient policy advice can be drawn from analyses of models that do not properly tackle these problems.

4. THE ECB'S MONETARY POLICY STRATEGY

Let me now turn to the ECB's monetary policy strategy. The ECB started to conduct policy in 1999 with the inception of the euro. While taking stock of the experience of the central banks of the Eurosystem, the ECB was at the time facing a major institutional change. Eleven national economies (14) merged into a unified market almost overnight; in this context, past experience and data might turn out not to be particularly informative with regard to the new economic structure.

Source: Findarticles.com

next>>