آموزش بانک مرکزی، شوکهای قوانین و مقررات تجارت و ریسک ارز: آیا برای سیاستهای پولی فقط نرخ تورم اهمیت دارد؟

This paper examines the role of interest rate policy in a small open economy, subject to terms of trade shocks and time-varying currency risks. The private sector makes optimal decisions in an intertemporal, non-linear setting with rational, forward-looking expectations. In contrast, the monetary authority chooses an optimal interest rate reaction function, given a loss function that is conditional on the state of the economy and given its “least squares learning” about the evolution of inflation and exchange-rate depreciation. The simulation results of the effects of different policy scenarios on welfare show that, on balance, the preferred stance should be strict inflation targeting.

This paper examines the role of interest rate policy in a small open economy subject to terms of trade shocks and time-varying currency risks. A central bank committed to low inflation controls neither the terms of trade nor the evolution of currency risk, both of which condition the response of inflation to its policy instruments. In this context, the best the central bank can do is to learn the effects indirectly by frequently updating estimates of inflation dynamics and re-adjusting its policy rules accordingly. However, when an economy is subjected to large adverse external shocks and the exchange rate depreciates rapidly, it should not be surprising if a central bank also comes under strong pressure to incorporate exchange-rate volatility targets in its policy objectives. Should exchange-rate changes then be included as one of the monetary policy targets along with inflation targets? Much of the discussion of monetary policy is framed by the well-known Taylor, 1993 and Taylor, 1999 rule, whereby interest rates respond to their own lag, as well as to deviations of inflation and output from respective targets. Taylor (1993) points out that this rule need not be a mechanical formula, but something which can be operated informally, with recognition of the “general instrument responses which underlie the policy rule.” Not surprisingly, the specification of this rule, which reflects the underlying objectives of monetary policy, has been the subject of considerable controversy.1 In a closed economy setting, Christiano and Gust (1999), for example, argue that only the inflation variable should appear as a target. Rotemberg and Woodford (1998) concur, but they argue that a higher average rate of inflation is required for monetary policy to be effective over the medium to long term. They base their argument on the zero lower bound for the nominal interest rate, since at very low inflation rates there is little room for this instrument to manoeuvre.2 In an open economy setting, McCallum (2000) takes issue with the Rotemberg and Woodford policy ineffectiveness argument under low inflation and zero lower bounds for nominal interest rates. McCallum argues that the central bank always has at its disposal a second tool, the exchange rate, so if the economy is stuck at a very low interest rate, there is the option of currency intervention. Christiano (2000) disagrees: McCallum's argument rests on the assumption that currency depreciation is effective. Furthermore, the central bank must be willing to undermine public confidence that it stands ready to cut interest rates in the event of major adverse shocks. For small emerging market economies, Taylor (2000) contends that policy rules that focus on a smoothed inflation measure and real output, and which do not “try to react too much” to the exchange rate might work well. However, he leaves open the question of a role for the exchange rate. Ball (1999) argues that inflation targeting “can be dangerous” in an open economy setting because exchange-rate changes have a direct effect on inflation via changes in import prices. Hence, adoption of a strict inflation targeting stance can result in large output variations. More recently, Gali and Monacelli (2002) found, in a small open economy setup with sticky-price setting behavior, that domestic inflation targeting dominates, from a welfare point of view, both 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). However, practically all of these studies are based on linear stochastic and dynamic general equilibrium representations, or linearized approximations of non-linear models. The Taylor-type feedback rules are either imposed or derived by linear quadratic optimization. While these approaches may be valid if the shocks impinging on the economy are indeed small and symmetric deviations from a steady state, they may be inappropriate if the shocks are large, persistent, and asymmetric, as they are in many highly open economies. Furthermore, few if any of these studies incorporate “learning” on the part of the monetary authority itself. Bullard and Mitra (2002) incorporate private sector learning of the specific Taylor rules used by the central bank in the Rotemberg–Woodford closed economy framework. They argue for Taylor rules based on expectations of current inflation from target levels, rather than rules based on lagged values or forecasts further into the future. In contrast to Bullard and Mitra (2002), we assume that the private sector uses the true, stochastic dynamic, non-linear model for formulating its own laws of motion for consumption, investment, and trade, with forward-looking rational expectations. In this analysis, the monetary policy authority learns the laws of motion of inflation dynamics from past data, through continuously updated least squares regression. From the results of these regressions, the monetary authority obtains 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. 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. Our approach also departs from typical new Keynesian models, with the friction coming from sticky pricing of domestically produced goods and/or sticky pricing of imported goods and/or stickiness in wage setting. In this paper, we have only one very simple nominal rigidity: that of varying degrees of exchange-rate pass-through in conjunction with the stickiness in learning on the part of the policy authority. Unlike most new Keynesian models, which assume that the aggregate capital stock is fixed, we allow for adjustment costs with capital accumulation and production in the model for commodity exports and for import-substituting manufacture goods. This is in keeping with recent research on optimal monetary policy which allows for different types of adjustment costs, such as the assumption in Altig et al. (2004), that the capital used by firms is completely firm specific. See also Schmitt-Grohé and Uribe (2004) for analysis with multiple distortions. The focus of our paper is to examine a different type of environment from the one typically used in the new Keynesian open economy literature. We are interested in a small semi-industrialized or emerging market open economy for which terms of trade shocks are the key determinants of business cycles and in which foreign investment and production are center stage in the macro adjustment process. We also model a central bank setting in which the monetary policy has to learn the laws of motion of inflation and formulate policy in this evolving, limited information context. This context is different from the setup in which the Calvo (1983) pricing mechanism is typically used, which is to assess short run response of the economy to monetary policy in industrialized countries. In this model, the monetary authority is interested in how to implement monetary rules when investment and the terms of trade are the driving forces for economic growth. Our results show that when a central banker is all-knowing and acts to optimize the intertemporal welfare of the consumer, there is not much difference in terms of welfare outcomes between fixed and flexible inflation targeting. In contrast, if the central bank decides to incorporate, in addition to inflation, exchange-rate dynamics in its learning and policy objectives, it does so at some welfare costs. In a learning environment, there is always the risk that the perceived laws of motion lag behind the actual laws of motion. Hence, expanding the range of policy objectives increases uncertainty. For this reason, strict inflation targeting dominates monetary policy based on multiple targets in an environment with central bank learning. Section 2 describes the theoretical structure of the model for the private sector and the nature of the monetary authority learning. Section 3 discusses the calibration, as well as the solution method, while Section 4 analyzes the simulation results of the model. Section 5 concludes.

This paper has compared two alternative policy scenarios—strict and flexible inflation targeting—for an economy facing terms of trade shocks and time-varying currency risk. The central bank is also assumed not to have full information about all behavioral aspects of the economy; instead, it has to learn the laws of motion for its key target variables in order to set the interest rate according to a feedback rule. The results show that including exchange-rate changes in its learning and policy targeting framework improves welfare only in the case of very high exchange-rate pass-through. In this case, the probability of low welfare outcomes is reduced. This result has intuitive appeal. With information stickiness and a high degree of “pass-through”, it is harder for the central bank to learn effectively the laws of motion of inflation, since inflation is more volatile under high pass-through. Adding exchange-rate changes into the learning and targeting framework makes sense, since there are high information spillovers from exchange-rate changes to inflation. However, with low pass-through, it is easier for the central bank to learn the laws of motion of inflation. Finally, if information becomes more transparent, with no information stickiness, there is little or no case for including exchange-rate depreciation targets. The policy implication is that central banks which are already targeting inflation should resist pressures to adopt exchange-rate targets, except in very special cases of high pass-through and information stickiness.