This article explores the consequences of various approaches to the conduct of monetary policy. A small, calibrated model of the Norwegian economy is used, which highlights the shortrun tradeoff between stabilising inflation and stabilising output. Some approaches to policy can be shown to be unambiguously better than others. However, when policy is efficient, the central bank must decide how much output variability it is willing to tolerate in order to attain more stable inflation.

Introduction
Since the beginning of the 1990s, several countries have adopted an inflation targeting framework for monetary policy. Since 2001, the operational target of monetary policy in Norway has been annual consumer price inflation of 2.5 per cent over time. Norges Bank operates a flexible inflation targeting regime, so that weight is given to both variability in inflation and variability in output and employment.
Within this type of framework there is considerable leeway regarding how policy is conducted. Subject to maintaining the inflation target in the longrun, the central bank has to decide how closely it will attempt to stabilise inflation around the target, at a cost of higher variability in output. This tradeoff is particularly stark in the case of a shock that causes inflation and output to move in different directions (a costpush or supply shock). The central bank's chosen course of action will depend on the perceived costs of variability in output and inflation respectively.
The aim of this article is to illustrate the consequences of various approaches to the conduct of monetary policy, using a small model for the Norwegian economy. We model different approaches to monetary policy by altering the interest rate response to different signals from the economy and examine the resulting variability in inflation and output. Some of the accepted "stylised facts" regarding inflation targeting monetary policy are illustrated. It is not the goal of this analysis to reach conclusions regarding what objectives the central bank should have, or what manner of conducting monetary policy might be optimal for Norges Bank. Two main points are illustrated:
* A move from flexible towards stricter inflation targeting implies accepting higher variability in output in order to keep inflation closer to the target on average. Stricter inflation targeting is illustrated in three different ways: i) responding relatively more strongly to inflation than to the output gap, ii) responding to nearerterm inflation forecasts, and iii) overall stronger policy responses.
* Some approaches to the conduct of policy are unambiguously more efficient than others, that is, they attain the desirable result of lower variability in both output and inflation. For example, the central bank can generally achieve better outcomes by being forwardlooking in its behaviour.
Section 2 presents the model used in the analysis, while Section 3 discusses the concept of an efficient policy frontier (EPF). Section 4 examines the implications of varying the coefficients in a simple policy rule. Section 5 concludes.

The model
This section describes the small, calibrated macroeconomic model that is used in the analysis. We give only a broad overview here; for a more detailed description of the model and its calibration see Husebo, McCaw, Olsen and Roisland (2004). (2)
2.1 A general overview
The model is highly aggregated, and provides a stylised representation of the key mechanisms in the economy, with a particular emphasis on the transmission mechanisms of monetary policy. It can be viewed as the smallest model necessary to explain the interaction of output, interest rates, exchange rates and inflation, under an inflationtargeting framework. (3) Although very simple and highly aggregated, the model has a considerable theoretical content. Starting with the classic smallscale openeconomy model by Dornbusch (1976), many similar models have been developed both in the academic literature and in central banks around the world. The quarterly model is calibrated to match salient features of the Norwegian economy, drawing on theory and a wide range of empirical estimates to choose parameter values for the model that result in appropriate aggregate properties.
Expectations play an explicit role in the model. First, expectations of future inflation are of importance as they will affect price and wagesetting behaviour today. Second, expectations of future interest rate developments affect today's exchange rate. Finally, expectations of future economic cycles will affect today's spending decisions.
The model aims to explain how deviations from equilibrium develop and dissipate over the medium to long term. (4) There is a clear role for monetary policy in the model: to provide the economy with a nominal anchor, that is, to prevent actual and expected inflation from drifting away from the target. When the central bank fulfils its role, the economy converges to a welldefined equilibrium. The model is designed such that the monetary authorities cannot boost output above its supplydetermined level (5) in the long run. In other words, in the longrun, monetary policy is neutral and there is no tradeoff between the levels of output and inflation.
The model consists of just four key equations:
1) An aggregate demand (IS) equation for an open economy that expresses the dynamic relationship between the output gap (i.e. output relative to its sustainable or trend level), the real interest rate, the real exchange rate and world output;
2) An inflationadjustment equation (Phillips Curve) characterising the dynamic response of inflation to inflation expectations, the output gap and the real exchange rate;
3) An uncovered interest parity (UIP) equation expressing the dynamic relationship between the exchange rate and the spread between domestic and foreign interest rates; (6)
4) A monetary policy rule describing how the central bank sets interest rates in order to balance the shortrun tradeoff between stabilising inflation around target and stabilising developments in the real economy. We discuss a simple rule specification in more detail later.
Each of these equations has a shock term that represents effects on the dependent variable from all sources other than the dynamics of the other variables appearing in the equation. These shocks will be important in our analysis. A demand shock could for example represent changes in tastes and preferences or the effects of fiscal policy. A shock to the Phillips curve could represent the growing importance of cheaper imports from China or stronger competition in the product market. A shock to the UIP equation could represent a change in the risk premium associated with Norwegian financial assets. Finally, there is also the possibility of adding exogenous shocks to the monetary policy rule, representing interest rate responses to changes in variables that are not included in the monetary policy rule.
Even though the model is simple, its strength is the focus on the role of monetary policy, a property that makes it well suited for the analysis carried out in this paper. Monetary policy affects inflation and the real economy through three main channels in the model.
First, there is a traditional demand channel. An increase in the nominal interest rate also increases the real interest rate, due to nominal rigidities. This discourages expenditure. Less demand pressure, in turn, results in lower inflation through both lower wage inflation and profit margins (not modelled explicitly).
Second, there is an exchange rate channel. Higher domestic nominal interest rates relative to those abroad cause the currency to appreciate, all else equal. Imported goods become cheaper and inflation falls. However, a stronger currency also has a negative effect on demand and output, via both an expenditure switching effect towards imports, and reduced competitiveness for industries that compete with firms internationally. Lower demand and output reduce inflation, as above.
Finally, there is the expectations channel. Expectations...