هدف قرار دادن تورم با قوانین پیش بینی بازخورد در اقتصاد کوچک باز
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|25377||2006||21 صفحه PDF||سفارش دهید||8720 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Journal of Economic Dynamics and Control, Volume 30, Issue 3, March 2006, Pages 393–413
We argue that in practice, the inflation-targeting strategy can be approximated by the interest rate responding to the unchanged-interest-rate forecast of inflation. A method is developed to derive unchanged-interest-rate forecasts in forward-looking models and evaluate the performance of the policy rule in an optimizing New Keynesian model due to Monacelli (European Central Bank, Working Paper Series: 227), estimated on UK data. We find that the policy rule is less prone to generate a determinate rational expectations equilibrium if based on an unchanged interest rate, compared to the rule-consistent forecast. Both rules approximate the optimal commitment policy if the central bank attaches sufficient weight to inflation as opposed to output gap stabilization. The optimal forecast-feedback horizon is close to a year and a half and is largely independent of how much the central bank prefers inflation to output gap stability.
The arguably poor performance and robustness of fixed exchange rate systems and monetary targeting have resurrected a belief in more activist policy throughout the 1990s. Such activism is normally associated with the central bank's discretionary use of the interest rate in order to directly steer policy toward price stability, in the sense of low and stable inflation. Such a framework is often referred to as inflation targeting. Inflation targeting has been formally introduced in several countries, e.g., New Zealand, Canada, Sweden, the United Kingdom, Australia, Norway and Iceland, where the central banks have been given explicit targets for inflation and instrument independence to set the interest rate so as to achieve the inflation target. We interpret inflation targeting as adherence to a forecast feedback rule for the interest rate where the deviations of the forecast of inflation from the target level are the prominent indicator. If the inflation forecast is above (below) the inflation target, the central bank sets a contractionary (expansionary) monetary policy stance, i.e., by setting the interest rate above (below) its natural rate or moving the interest rate in steps towards this target rate. This interpretation is in line with the interpretations made by Batini and Haldane (1999), Batini and Nelson (2001), Levin et al. (2003) and others.1 Several central banks state the use of such a procedure to guide policy. Sveriges Riksbank (1999)Inflation Report 3/99, p. 58 states: Monetary policy is sometimes described with a simple rule of thumb: if the overall picture of inflation prospects (based on an unchanged repo rate) indicates that in twelve to twenty-four months’ time inflation will deviate from the target, then the repo rate should normally be adjusted accordingly. (My italics) Jansson and Vredin (2003) interpret the procedure of monetary policymaking at Sveriges Riksbank as the use of forecasts feedback rules. Svein Gjedrem, the Governor of the Central Bank of Norway, states The key rate is set on the basis of an overall assessment of the inflation outlook 2 years ahead. If it appears that inflation will be higher than 2 per cent with unchanged interest rates, the interest rate will be increased. If it appears that inflation will be lower than 2 per cent with unchanged interest rates, the interest rate will be reduced. ( Gjedrem, 2002) (My italics) A representation of such a forecast feedback rule is given by equation(1) View the MathML sourcert=ρrrt-1+(1-ρr)βπ[π¯^t+H-π¯*], Turn MathJax on where r is the policy interest rate, View the MathML sourceπ¯* the annual inflation target and View the MathML sourceπ¯^t+H the HH-period-ahead forecast of the annual inflation rate. H is the forecast-feedback horizon, which is distinguished from the policy target horizon, i.e., the expected time before inflation has returned to its target level (see also Batini and Nelson, 2001). 2 The rule allows for interest rate smoothing, i.e., the partial adjustment of the interest rate, which may be important for the central bank to have a beneficial influence on private sector behavior by affecting private agents expectations about future policy (see Woodford, 2003b). Although research on forecast feedback rules, which is discussed below, has almost exclusively been based upon the assumption of rule-consistent forecasts,3 i.e., equation(2) View the MathML sourceπ¯^t+H=Et[π¯t+H], Turn MathJax on central banks have typically used an unchanged-interest-rate assumption (see, e.g., the italicized text in the quotations) in deriving the inflation forecasts, that is, deriving expected inflation conditional on the interest rate not being changed throughout the forecast-feedback horizon, i.e., equation(3) View the MathML sourceπ¯^t+H=Et[π¯t+H|r¯t-1]. Turn MathJax on We use the abbreviation PCF for denoting the policy-consistent forecast-feedback rule (Eqs. (1) and (2)), and UIF for denoting the unchanged-interest-rate forecast-feedback rule (Eqs. (1) and (3)). The forecast assumption is important when the forecast-feedback horizon is longer than the policy control lag and the inflation forecast does not only depend on the present policy stance, but also on the future policy stance. Although the assumption of a policy-consistent interest rate throughout the forecast horizon ensures consistency, it may be somewhat unrealistic from a practical point of view. Forecasts based on assumptions about specific future interest rate changes may be of little guidance to the interest rate decision body that may have a hard time just deciding about the present interest-rate stance. Svensson (1999a) argues that the forecast should be based on an unchanged interest rate, which allows the decision body to focus on the current interest rate setting, and not having to form expectations about future interest rate decisions. Moreover, Svensson argues that it may be easier to incorporate outside-of-the-model information under such a procedure, since such information may take the form of the policymakers’ judgment conditional on the policy stance remaining unchanged. So far, however, there has been no procedure for handling the unchanged-interest-rate assumption in models where agents display forward-looking behavior. This paper presents a method for doing exactly that. Using this method, we evaluate the difference between the two forecast assumptions using an empirical optimizing New Keynesian model of the UK economy. An important result is that we find the difference between the PCF and UIF rules to be small, given that the forecast-feedback horizon is short. It becomes more important at horizons exceeding six quarters. Although both assumptions make the rule prone to rational expectations (RE) indeterminacy at long forecast horizons, the assumption of an unchanged-interest-rate forecast increases the region of RE indeterminacy. We find that both the UIF and PCF rules are successful at stabilizing inflation but less so at stabilizing the output gap, thus supporting and extending the result in Rudebusch and Svensson (1999) and Levin et al. (2003) to the small open economy. The optimal forecast-feedback horizon is found to be around a year and a half, irrespective of how strict the central bank is on inflation stabilization. This length is close to the forecast-feedback horizon used by many inflation-targeting central banks. PCF rules have been extensively discussed in the literature. Batini and Haldane (1999) argue that the rule is “lag encompassing”, i.e, takes account of the fact that monetary policy works with a lag on inflation by focussing on the inflation forecast. By responding to the forecast of inflation at a sufficiently long length, it ensures that policy preemptively responds to those inflationary shocks monetary policy may indeed counteract. The policy rule includes the inflation forecast as an indicator and therefore, embodies all relevant knowledge about future inflation. The rule is therefore “information encompassing”. Finally, they show that within a small forward-looking macroeconomic model, the rule is successful at stabilizing both inflation and output (i.e., rule is “output encompassing”) without causing too strong movements in the interest rate. Batini and Nelson (2001) evaluate the rule in both a vector autoregressive (VAR) model and a small forward-looking macroeconomic model and find that the optimized rule performs close to the optimal commitment policy. The optimal forecast-feedback horizon, however, depends greatly on the particular model, being two quarters for the forward-looking model and as long as 15 quarters for the VAR model. Rudebusch and Svensson (1999) compare the performance of several rules for inflation targeting in a backward-looking model of the US economy and find that the forecast-feedback rules perform close to the optimal rule, as long as the central bank does not put a relatively large weight on stabilizing output. Their study suggests the performance to be relatively independent of the choice of forecast feedback horizon, as long as it is beyond 2 years. Levin et al. (2003) study forecast feedback rules in five models of the US economy and also find, as noted above, that the rule is successful at stabilizing inflation but lacks some of the output encompassing features found by Batini and Haldane (1999). However, by extending the rule to include the output gap as an indicator and responding to the 1-year-ahead inflation forecast, an appropriately calibrated rule does not only more efficiently stabilize output, but also becomes more robust to model uncertainty, i.e., it works well in all five models. They find that rules with a longer forecast feedback horizon are prone to RE indeterminacy, and also less robust to model uncertainty. The remainder of the paper is organized as follows: Section 2 discusses the use of the unchanged-interest-rate assumption in constructing inflation forecasts. Section 3 presents a New Keynesian model of a small open economy due to Monacelli (2003) with both domestic goods producers and firms importing goods from abroad experiencing rigidities in price setting. Section 4 presents the stabilization and determinacy properties of the two types of policy rules and discusses the best choice of the forecast-feedback horizon. Finally, Section 5 provides the main conclusions.
نتیجه گیری انگلیسی
This paper has evaluated inflation forecast-feedback rules in an estimated, micro-founded model of the UK economy. We find that these rules bring inflation and output close to the optimal policy inflation-output variance frontier, and that they are close to the optimal rule as long as the central bank puts sufficiently large weight on stabilizing inflation. This confirms results obtained in models of relatively closed economies. The optimal forecast-feedback horizon is surprisingly stable at six or seven quarters, and independent of the weight the central bank attaches to inflation versus output stabilization. A potential problem with both forecast-feedback rules is that neither rules do not necessarily ensure determinacy of the RE equilibrium. This problem is especially acute at long forecast-feedback horizons where the set of rule parameters that creates determinacy is quite small. We find that the forecast-feedback rule using an unchanged-interest-rate forecast of inflation in general decreases the parameter determinacy space, and does not improve on the performance of the rule. The implied dynamics of the rules do, however, show an important difference when the central bank applies a long forecast-feedback horizon.