تولید بالقوه نامشخص : پیامدهایی برای سیاست های پولی

A small forward-looking model of the euro-area economy is used to investigate the implications of incomplete information about potential output for the conduct of monetary policy. Three results emerge. First, under optimal monetary policy, output gap uncertainty leads to persistent deviations between the actual and the perceived output gap in response to supply and cost-push shocks. Second, in first-difference form, a simple policy rule such as the Taylor rule continues to perform relatively well as long as the output gap is optimally estimated. Third, incomplete information implies that it is optimal to appoint a more “hawkish” central bank.

In most macro-models that are currently being used for policy analysis the degree of capacity utilisation as, for example, measured by an output gap plays an important role in the determination of inflation. Yet, there are no direct measures of the aggregate supply side of the economy or the extent to which the resources in an economy are fully used. A wide variety of both conceptual and empirical methods have been proposed to estimate potential output and make the notion of an output gap operational.1 In this paper we analyse the implications of imperfect information on potential output for the conduct of monetary policy.2 Only recently this issue has been studied more formally in quantitative models of optimal monetary policy. Most of this analysis takes place in a linear–quadratic framework.3Estrella and Mishkin (1999) emphasised that uncertainty about the NAIRU would not affect the optimal monetary policy in such a framework because certainty equivalence holds with respect to shocks that enter the model additively. Smets (2002) integrates the estimation of the output trend in a backward-looking policy evaluation model, and illustrates the effect of estimation errors on the choice of efficient simple policy rules. He shows that higher uncertainty leads to some policy attenuation in simple Taylor rules. These results are generally confirmed by Tetlow (2000), who extends the analysis to a forward-looking model and also examines inflation-forecast-based rules. Using estimates of measurement error derived from real-time estimates of the output gap, Orphanides (2001) shows that such errors lead to a significant deterioration of feasible policy outcomes and cause efficient policies to be less activist. Rudebusch (2001) shows that these considerations are essential for reconciling estimated policy reaction functions and optimal policy. However, Svensson and Woodford (2000) and Swanson (2000) argue that these results are due to the fact that the central bank does not use its best estimate of the output gap. While certainty equivalence holds when the optimal policy is expressed in terms of the best estimate of the state variables of the economy, the weights put on various indicators used in deriving such an estimate will depend on how noisy these indicators are. In this paper, we illustrate and extend these results using a simple, backward/forward looking, calibrated model of the euro area economy. We chose to look at the euro area rather than the US for two reasons. First, the European Central Bank (ECB) conducts monetary policy for an entirely new currency area, which implies that the problem of uncertainty is even more pressing than for other central banks which operate in a familiar environment. It is therefore useful to analyse these issues in a European context. Second, there has been considerable discussion on the design of a suitable monetary policy strategy for the ECB under this situation of elevated uncertainty. Our analysis can shed some light on this issue, by discussing the importance that should be given to indicator variables in monetary policy making, and by investigating the optimal degree of conservativeness of a central banker in this context. We use the calibrated euro area model to systematically compare the monetary policy implications of two assumptions regarding the information available to the agents in the economy. Under one assumption the central bank and the private sector can perfectly derive the nature of the shocks (including those to potential output) that hit the economy. In the other case, the central bank and the other agents in the economy do not observe potential output and need to estimate it on the basis of current output and inflation. Following most of the literature we perform the analysis in a linear–quadratic framework where the central bank and the private sector know the structure of the model economy. As a result, certainty equivalence (optimal policies are independent of additive uncertainty) and the separation principle (the estimation problem can be separated from the control problem) hold.4 Our first finding is that the welfare loss due to incomplete information about potential output is substantial. This loss is mainly due to a significant increase in the variability of the output gap. Even if the central bank continuously updates its estimate of potential output, ex post perception errors of the output gap are substantial and very persistent in particular in response to a shock to potential output. A second finding is that simple policy rules (like the Taylor rule) that respond to the estimated output gap continue to perform relatively well compared with the commitment equilibrium as long as the output gap is optimally estimated. A third finding is that incomplete information about potential output reinforces the case for appointing a weight-conservative central banker (i.e. a central banker with a smaller weight on output gap stabilisation in his/her loss function). The rest of the paper is organised as follows. In Section 2 we lay out the model and discuss the calibration of the parameters. Careful calibration is necessary because some of the results are likely to be driven by the structure of the economy and in particular the covariance matrix of the shocks. In Section 3 we then characterise the certainty-equivalent optimal monetary policy under both commitment and discretion using the results derived by Svensson and Woodford (2000). Section 4 analyses the implications of incomplete information for a simple policy rule such as the Taylor rule. In Section 5 we examine whether output gap uncertainty affects the optimal delegation to a weight-conservative central bank. Finally, in Section 6 we examine the robustness of the results with respect to changes in the model parameters and analyse the case of asymmetric information. Section 7 reviews the most important conclusions.

In this paper we have investigated the implications of incomplete information about potential output for the conduct and design of monetary policy using a small backward/forward looking model calibrated to fit annual euro area data. The model features three shocks: a shock to potential output, a cost-push shock and a demand shock. Using a standard quadratic loss function in inflation and output gap stabilisation, we systematically compare optimal monetary policy under two assumptions regarding the information available to the central bank and the private sector. In the first case the information set contains current output, potential output and inflation. As a result, the central bank can perfectly deduce which of the three shocks is hitting the economy and adjust its policy rate accordingly. In the second case, information is incomplete. The central bank and the private sector do observe current inflation and output, but do not observe potential output. As a result, the central bank faces a signal extraction problem when estimating potential output and setting the optimal interest rate. Our main results can be summarised under three headings. First, we characterise and compare the optimal monetary policy under both commitment and discretion in the two informational cases using recent results of Svensson and Woodford (2000). We find that the loss due to incomplete information is substantial and mainly results from a significant increase in the variability of the output gap. In our benchmark case, the standard deviation of the output gap is about one-third higher under incomplete information. Even if a central bank continuously updates its estimate of potential output and thus ex ante does not make any systematic estimation errors, ex post the misperception of the output gap is substantial and very persistent, in particular, in response to a shock to potential output. This result would suggest that the empirical findings of Orphanides (2000) that the Federal Reserve Board made serially correlated estimation errors in the output gap in the 1970s may not be evidence of systematic ex ante mistakes. We also find that in response to a shock to potential output, such misperception errors lead to a very persistent effect on inflation, which is exacerbated when the central bank optimises under discretion. Nevertheless, the overall variance of inflation is not very much affected, because under incomplete information, the response of inflation to cost-push shocks is mitigated. Second, we investigate the implications of incomplete information for simple policy rules such as a Taylor rule that rely on estimates of the output gap. Imperfect information about potential output leads to some limited attenuation in the optimal response to the estimated output gap. However, those rules continue to perform relatively well compared with the commitment equilibrium as long as the output gap is optimally estimated. Finally, we examine the implications of the unobservability of potential output for the optimal delegation of monetary policy to a discretionary central bank. We find that incomplete information about the output gap implies that it is optimal to appoint a relatively more conservative central bank than in the complete information case. Following Rogoff (1985) and Clarida et al. (1999), we show that society would like to appoint a more weight-conservative central banker (i.e. a central banker with a smaller weight on output gap stabilisation in his/her loss function) when output gap uncertainty is large. Overall, our results suggest that incomplete information about potential output does not significantly overturn results regarding optimal monetary policy that are derived in models with perfect information, as long as the central bank uses its best estimates of the state of the economy and has the true model of the economy at its disposal. Of course, these results very much depend on the linear–quadratic framework used in this paper and are likely to break down when model uncertainty and/or non-linearities are considered. The results do strongly indicate that there are potentially large gains to be made from trying to identify the particular shocks that hit the economy at any point in time. An early identification of the nature of the shocks that affect the economy allows for a more appropriate policy response and improves the stabilisation of the economy significantly.