هموارسازی نرخ بهره و سیاست پولی بهینه: مروری بر شواهد تجربی اخیر
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|24520||2000||24 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Journal of Economics and Business, Volume 52, Issues 1–2, January–April 2000, Pages 205–228
The Federal Reserve and other central banks tend to change short-term interest rates in sequences of small steps in the same direction and reverse the direction of interest rate movements only infrequently. These characteristics, often referred to as interest-rate smoothing, have led to criticism that policy responds too little and too late to macroeconomic developments. This paper, however, argues that interest-rate smoothing may in fact be optimal. We present empirical results from several recent papers that offer three explanations of interest-rate smoothing: forward-looking behavior by market participants, measurement error associated with key macroeconomic variables, and uncertainty regarding relevant structural parameters.
The conduct of monetary policy by central banks is of considerable interest to both academic researchers and to financial market participants. This interest has generated a large literature that attempts to describe and evaluate the interest rate policies of central banks. One characteristic that has been widely noted in this literature is that many central banks, including the Federal Reserve, adjust short-term interest rates in a smooth manner. In fact, some researchers have argued that observed interest rate movements are too sluggish, and that a less timid interest rate policy would be more effective at stabilizing output and inflation. Furthermore, the observed smoothness of interest rates has been considered evidence that central banks have a separate objective of minimizing interest rate volatility in addition to the goals of stabilizing output and inflation. This article, instead, argues that the observed smoothing of the interest rate may indeed be optimal, even if the central bank is not explicitly concerned with interest rate volatility. In particular, this article considers three characteristics of the policymaking environment that may render some degree of interest-rate smoothing optimal: forward-looking behavior by market participants, measurement error associated with key macroeconomic variables, and uncertainty regarding relevant structural parameters. The paper is organized as follows. This introductory section presents some evidence on the smoothness of the interest rate and clarifies the definition of interest-rate smoothing that is used throughout the paper. Sections II through IV describe each of the three arguments in detail and present supporting empirical results from recent research on the U.S. economy. Section V offers a brief conclusion. A definition of interest-rate smoothing The belief that the Federal Reserve, like most other central banks, deliberately chooses a smooth path for the short-term interest rate may stem from the dual observations that the federal funds rate tends to move in sequences of small steps in the same direction and that reversals in its direction are relatively infrequent. As apparent in Figure 1 , which displays the intended federal funds rate from 1984 through 1998, changes in the funds rate are frequently followed by further changes of the same sign. Indeed, in this sample 85 percent of funds rate changes represent “continuations” in the direction of policy. Furthermore, such continuations often occur in fairly rapid succession, with an average of 34 days separating changes when there is a continuation compared to an average of 97 days for a reversal, suggesting that these changes constitute steps within a single policy movement. The magnitude of these steps is modest, typically a quarter of a percentage point, as only 13 percent of funds rate changes in this sample have been by half a percentage point or larger.1 Similar patterns are observed in official interest rates in many other countries. Lowe and Ellis (1997) have presented evidence for Australia, the United States, the United Kingdom, Japan, and Germany. The implications of these results are discussed further by Lindsey (1997). Goodhart (1998) has, in addition, looked at interest rates in France, Italy, Canada, Spain, the Netherlands, Belgium, Sweden, and Austria.Of course, if monetary policy systematically reacts to macroeconomic variables such as inflation and output, these patterns in interest-rate behavior may simply reflect the persistence in the movements of those variables. This possibility becomes evident when central bank behavior is characterized by an interest rate reaction function—that is, a rule for setting the federal funds rate that responds systematically to variables such as output, inflation and past interest rates. A common specification of such a rule is equation(1) According to this rule, the current federal funds rate rt is determined by the lagged funds rate, the equilibrium real interest rate, rr∗, the inflation rate, πt, the inflation target, π∗, and the output gap, yt, which is the difference between real current and potential output. The coefficients α and β capture the response to output and inflation, whereas ρ measures the weight put on the past funds rate setting. Note that the well-known Taylor rule is a special case of equation (1) that sets ρ equal to zero [see Taylor (1993)]. Clearly, if output and inflation fluctuations are persistent, the systematic policy response will impart some smoothness to the federal funds rate. However, the degree of smoothness of the observed funds rate will be influenced importantly by two characteristics of interest rate reaction functions such as equation (1). First, a value of the coefficient on the lagged funds rate ρ that is positive but below one induces partial adjustment of interest rates to changes in the other macroeconomic variables. In particular, it implies that a deviation of output from potential or inflation from target will trigger a sequence of policy moves in the same direction until output and inflation have been restored to their desired paths. Second, the overall responsiveness of the interest rate rule is determined by the response coefficients α and β. For a given value of ρ, smaller coefficients on output and inflation imply more timidity in the responsiveness of the interest rate, which results in lower interest rate volatility. We use these two characteristics of interest rate reaction functions to define interest-rate smoothing. In the remainder of the paper, interest-rate smoothing refers to a high degree of partial adjustment and limited overall responsiveness of the interest rate. As to the observed degree of partial adjustment and responsiveness, there exists a large literature on estimating interest rate rules that has produced estimation results for many different countries and time periods [see Clarida et al 1998, Clarida et al 1999 and Judd and Rudebusch 1998, and Orphanides (1998a), among others]. Rather than providing a detailed survey of the many different reaction functions that can be found in the literature on policy rules, we present an example taken from Orphanides and Wieland (1998) estimated by means of instrumental variables techniques using quarterly U.S. data from 1980:Q1 to 1996:Q4: equation(2) The variables are defined as in equation (1), and unexplained residuals are denoted by u. 2 The results indicate that federal funds rate behavior during the 1980s and 1990s can be largely described by a systematic response to the lagged funds rate, the inflation rate, and the current and lagged output gap. Furthermore, while the lagged output gap is important in the early part of the 1980s, it could be dropped without any loss in fit if the rule were estimated from 1987 onward. 3 Two findings that are common to much of the literature are of particular interest. First, the size and significance of the partial adjustment coefficient, which is about 0.8, provides direct evidence that the observed degree of persistence in federal funds rates is greater than can be attributed to systematic policy responses to persistent output and inflation fluctuations.4 Second, the federal funds rate responds significantly to output and inflation. In particular, a favorable property of equation (2) is that the federal funds rate responds sufficiently aggressively to an increase in inflation to induce an increase in real rates and an effective tightening of monetary policy. Thus, the criticism leveled against monetary policy in the 1970s, namely that it did not respond sufficiently to keep inflation under control, does not apply to policy estimated over the 1980s and 1990s.5 Nevertheless, interest rate rules estimated over the latter period have been subjected to a different criticism. Criticism of federal reserve policy In the context of interest rate reaction functions such as and , asking whether the observed degree of interest rate smoothing is optimal is equivalent to asking whether the observed degree of partial adjustment and the overall responsiveness of interest rates are optimal. Some analysts who have evaluated the performance of policy rules in terms of stabilizing output and inflation fluctuations have argued that an optimal policy rule would not exhibit as high a degree of partial adjustment and that estimated interest-rate reaction functions such as equation (2) are too timid. Of course, forming such a judgement requires a specific macroeconomic model that describes the determination of output and inflation when interest rates are set according to a specific rule. Quantitative studies that have come to the conclusion that policy rules that fit interest rate behavior in the 1980s and 1990s are too inertial and timid have typically relied on models that assume adaptive expectations and ignore important sources of uncertainty [see, for example, Ball 1999, Rudebusch 1998 and Rudebusch and Svensson 1999 and Williams (1999)]. Because these models assume that the short-run relationship between interest rates and output and inflation is known with certainty and invariant to policy, output and inflation can be stabilized more effectively by making interest rates more responsive to macroeconomic variables than observed historically. That is, the policymaker can achieve greater macroeconomic stability in these models by accepting higher volatility of the interest rate. As a result, the Federal Reserve has been criticized for responding too little and too late to macroeconomic developments. Furthermore, it has been suggested that the Federal Reserve’s practice of smoothing interest rates reflects an objective of lowering interest rate volatility, in addition to the policy goals mandated in the Federal Reserve Act.6 In fact, many empirical studies of optimal monetary policy assume that policymakers have a threefold objective of minimizing deviations of output from its potential level, deviations of inflation from its target, and the variability of the level or the change in the short-term interest rate. Some researchers, using backward-looking models, have found that a large weight on interest-rate variability in the policymakers’ loss function is required to obtain an optimal rule that matches historical estimates. In contrast, this article argues that the observed degree of interest-rate smoothing reflects to a large extent optimal behavior on the part of central banks that are concerned only with output and inflation stability. Three arguments help to explain why the observed degree of interest rate smoothing may be optimal: the forward-looking behavior of market participants, data uncertainty or measurement error regarding key macroeconomic variables, and uncertainty regarding relevant parameters of the economic structure. Three motives for interest-rate smoothing 1. In models with forward-looking expectations, estimated policy rules such as equation (2) are more effective in stabilizing output and inflation for a given level of interest rate volatility than rules without partial adjustment. If policy exhibits a high degree of partial adjustment, forward-looking market participants will expect a small initial policy move to be followed by additional moves in the same direction, which increases the impact of policy on current output and inflation without requiring large interest rate changes. 2. Model-based evaluations of simple policy rules typically assume that policymakers observe key macroeconomic variables without error. In practice, however, some measures of output and inflation tend to be revised several times following the initial data release, while estimates of potential output or the natural unemployment rate may even be revised years later. Recent research has shown that under simple policy rules such as equation (1) the central bank should moderate the responsiveness of the interest rate to initial data releases when that data is measured with error, because an aggressive policy response would induce unnecessary fluctuations in the interest rate resulting in unintended movements in output and inflation. 3. Policymakers are not only uncertain about the exact state of the economy, but also about key parameters of the economic structure that affect the transmission of monetary policy. When parameters are uncertain, aggressive policy moves that might otherwise be expected to offset inflation and output deviations fairly quickly are more likely to have unpredicted consequences on output and inflation. It may instead be optimal to implement a gradual response of the interest rate that would return output and inflation more slowly to the targeted values. Other explanations of interest-rate smoothing have been offered in the literature, including concern about the soundness of the financial sector, maintaining the reputation of the central bank, issues regarding consensus decision-making, and adverse reactions of financial markets to frequent changes in the direction of interest rates. Although these arguments may have merit, empirical evidence is yet to be provided, and several of the arguments still have to be modeled rigorously. As a result, this paper focuses on the three arguments outlined above. In the following sections we discuss empirical evidence for each of these arguments. While for each argument we refer to several supporting recent studies, we limit the more detailed review of empirical methodology and results to one study per argument. In particular, this paper reviews the results of Levin et al. (1999) regarding efficient simple interest-rate rules under forward-looking expectations, Orphanides (1998b) regarding the impact of data uncertainty on optimal Taylor rules, and Sack (2000) regarding optimal policy under parameter uncertainty. The three studies have in common that they apply to the U.S. economy and were conducted at the Federal Reserve Board. While we have not extended our review to cover the experience of other countries and central banks, we again note that choosing a smooth path for short-term interest rates is a strategy common to many central banks.
نتیجه گیری انگلیسی
This paper has offered several explanations for why interest-rate smoothing by central banks may be an optimal approach for stabilizing inflation and output, without assuming an interest-rate smoothing objective for the central bank. In particular, some degree of interest-rate smoothing is optimal when the policymaker is faced with forward-looking behavior by market participants, measurement error associated with key macroeconomic variables, and uncertainty regarding relevant structural parameters. Supporting empirical evidence for this contention appears in three recent studies: Levin et al 1999 and Orphanides 1998b, and Sack (2000). Based on this research, we find plausible reasons for the inertial response of U.S. monetary policy in the 1980s and 1990s to fluctuations in key macroeconomic variables such as output and inflation. The three factors considered—forward-looking expectations, parameter uncertainty, and data uncertainty—would each rationalize a substantial degree of interest-rate smoothing in U.S. monetary policy. Of course, the results presented also indicate that engaging in too much interest-rate smoothing can have a detrimental impact on macroeconomic performance. For example, it has been argued that in the 1970s, policy did not respond sufficiently aggressively to increases in inflation to induce a tightening. In the 1980s and 1990s, however, interest rates clearly responded sufficiently aggressively to inflation, and the degree of interest-rate smoothing observed has not been far from the optimal level found under the three factors considered in this paper. Although none of the three factors, as far as existing empirical results are concerned, is able to account completely for the observed degree of interest-rate smoothing, taken together they may be more than sufficient. Furthermore, in each case we suggest extensions that might further enhance the empirical support for interest-rate smoothing as optimal policy. A useful avenue for future research would seem to be an empirical evaluation of all three factors in a single model that would make it possible to test the relative importance of the factors and the interaction between them. Goodhart (1997).