اطلاعات در زمان واقعی و سیاست پولی
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|25804||2005||6 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : The North American Journal of Economics and Finance, Volume 16, Issue 3, December 2005, Pages 271–276
This paper provides an introduction to the problems and opportunities provided by the availability of real-time data. We stress the importance of analyzing policy issues relying on real-time data. A summary of papers presented at a Conference hosted by the Bundesbank in 2004 is also provided.
Successful monetary policymaking is replete with real-world challenges. A particularly vexing one is that monetary policy decisions are, by necessity, always based on imperfect knowledge of the evolution and state of the economy. Decisions are made in real time. Preliminary estimates of aggregate measures of production, employment, prices, money and credit, and other variables, are frequently subject to substantial errors and are prone to subsequent revisions. Furthermore, key concepts whose availability could greatly improve forecasts of the economy and simplify policy decisions, such as the level of the economy's potential supply, its natural rate of growth, the natural rate of unemployment, or the equilibrium values of interest rates and exchange rates, are always significantly easier to assess with the benefit of hindsight, long after they have ceased to be of value for real-time policy decisions. While waiting to assess the economic environment and re-evaluate the context in which a policy decision may have been made with the benefit of hindsight is typically the strategy adopted in monetary policy analyses, policymakers plainly do not always have this option. Despite its pervasive influence on real-world policymaking, this aspect of decision-making is often neglected in theoretical treatments of the monetary policy problem. It is common practice to estimate models and simulate economic decisions with only the latest rendition of historical data. Indeed, it is often convenient to use the latest available revised data, even if these did not exist or were substantially revised long after they could possibly have been employed by policymakers and other agents in the economy, who supposedly relied on such data for their decisions. In some settings, the complexity of modeling the underlying information problem proves too great. In other cases, the data needed for more careful real-time analysis may not be readily available. To be sure, theoretical simplicity can be a virtue when it does not interfere with the purpose of the analysis. For many questions of interest in monetary analysis, however, neglect of the real-time nature of the policy problem can render the analysis unreliable, if not outright misleading. Characterizations of historical monetary policy can be particularly sensitive to the treatment of information. Consider, for instance, econometric attempts to characterize historical policy by estimating policy reaction functions. To the extent such estimated reaction functions correctly identify how policymakers responded to available information over time, comparisons across periods with better or worse macroeconomic performance can help identify better from worse policy practices. But estimated policy reaction functions could potentially lead to very different conclusions, depending on whether estimation is performed with data actually available to policymakers in real time or with ex-post revised proxies. Evidence from the U.S. experience, where the availability of data permits such comparisons, suggests that in some periods the resulting estimation issues are indeed severe (Orphanides, 2001). Interpretation of estimates based on the latest revised data appears untrustworthy. The foregoing considerations also suggest that a retrospective analysis of the behavior of inflation over the past few decades needs some rethinking. For example, why was inflation so much higher and more volatile during periods of the 1960s, 1970s and 1980s in so many countries? A variety of reasons have been put forward and the search continues for a convincing understanding of the success of central banks to tame inflation to low and stable levels over the past decade in particular. Some explanations focus on institutional aspects of the relationship between the central bank and the government (and public), to whom it is ultimately accountable for the longer term effects of central bank successes or failures in the realm of monetary policy (inter alia, see Siklos, 2002). Other explanations simply point to policymaker competence or lack thereof. However, shorter-run explanations of inflation performance must, in large part, rely on the persistent misinterpretations of the state of the economy. Testing such a view is impossible, unless the investigator can reconstruct the information set available to policymakers when policy decisions that subsequently proved incorrect were made. For such analysis, the availability of real-time data is, of course, essential. Further, consider the question of how forward-looking monetary policy should be. Under idealized conditions, forecasts of the economy may be quite accurate and forecast-based policies may anticipate and counteract, at least to some extent, macroeconomic disturbances. Real-time forecasting exercises, however, might point to limitations in the quality of preliminary data and to uncertainty about model specification as limiting the reliability of forecasts, thereby complicating evaluations of the efficacy of forecast-based policy approaches. Over the past few years, a burgeoning literature has begun investigating the implications of real-time data analysis for monetary policy. Much of this literature has concentrated on the U.S. experience, where data and historical records for monetary policy have generally been more readily available than for other countries. This early work has served to confirm many of the suspected pitfalls of neglecting the real-time nature of the policy problem in monetary policy, suggesting that additional investigations are in order. With these considerations in mind, the North American Journal of Economics and Finance (NAJEF), and the Deutsche Bundesbank, called on researchers from academia and central banking institutions around the world to participate in a conference that would deal with questions concerning “Real-Time Data and Monetary Policy.” Selected papers were considered for publication in a Special Issue of the NAJEF. Thanks to the generous hospitality of the Deutsche Bundesbank, a conference was held on May 28–29, 2004 in Eltville, Germany. 1 The conference had two principal objectives, namely, to stimulate the creation and dissemination of data and empirical evidence dealing with real-time data for countries other than the U.S., and to explore some of the broader implications for monetary policy when decisions are taken in real time. Readers will find a mix of approaches being used to investigate the role and impact of real-time data. Some of the studies published here rely on the Taylor rule to illustrate the various difficulties and pitfalls in the estimation of output gaps; others resort to the specification of structural models to investigate much the same questions. Taken together, the papers are a sobering reminder of the difficulties monetary authorities face in having to act on the interest-rate front and the challenges of communicating information that is noisy and highly variable over time. Nevertheless, the papers lead to the unmistakable conclusion, that it is vital to maintain a historical record of what policymakers knew at the time decisions were made and the models that were used as inputs into such decisions. The editors also wish to acknowledge Hermann Remsperger, Vitor Gaspar, and Bundesbank President Axel Weber for delivering interesting and illuminating commentary about the role of real-time data from a central banker's perspective. The Bundesbank has undertaken to serve as a home for some of the real-time data sets developed for this conference. No doubt this will stimulate the development of more data sets in the future.