چگونگی سیاست های پولی انجام شده واقعی بانک باندس

Papers estimating the reaction function of the Bundesbank generally find that its monetary policy from the 1970s to 1998 can be captured by a standard Taylor rule according to which the central bank responds to the output gap and to deviations of inflation from target, but not to monetary growth. This result is at odds with the Bundesbank's claim that it followed a strategy of monetary targeting. This paper analyzes whether this apparent contradiction is due to (a) the use of ex-post data which do not necessarily match policy makers’ real-time information sets or (b) the omission of important explanatory variables. Accordingly, we compile a real-time data set for Germany including the Bundesbank's own estimates of potential output and use it to re-estimate the Bundesbank's reaction function. We find that the use of real-time data changes the results considerably. Moreover, when adding the change in the output gap as well as deviations of money growth from target to the set of explanatory variables, we find that both variables are highly significant. This suggests that the Bundesbank took its monetary targets seriously, but also responded to deviations of expected inflation and output growth from target.

The question of how the Bundesbank conducted monetary policy is of interest not only from a historical perspective. Given that the Bundesbank is usually seen as a comparatively successful central bank, it may be helpful to better understand its monetary policy in order to draw conclusions for current monetary policy. For example, there is an ongoing discussion about the interpretation of the U.S. Fed's monetary policy in the 1970s. While some commentators argue that the Fed was responsible for the “Great Inflation” of the 1970s, because its monetary policy was too expansionary compared with a Taylor rule, others stress that such an interpretation relies on the advantage of hindsight. Today, we know that the Fed's real-time assessment of the U.S. business cycle was too pessimistic. Taking this real-time problem into account, the Fed's monetary policy could be justified even from a Taylor-rule perspective. Or, to put it differently, the Fed's policy would not have been significantly different had it in fact followed a Taylor rule. The inflation record of the Bundesbank in the 1970s and early 1980s was better than that of the Fed and many other central banks and one could ask why this was the case. One obvious distinctive feature of the Bundesbank's monetary policy since 1975 was that it announced annual targets for monetary growth and – according to its own descriptions – based monetary policy decisions on deviations of actual money growth from these targets. However, recent empirical studies of the Bundesbank's monetary policy generally find that monetary aggregates did not play a significant role in the Bundesbank's interest rate decisions, but that its policy can well be described by a standard Taylor rule.1 There are several ways to explain this apparent contradiction. One is that the Bundesbank did not practice the strategy of monetary targeting that it preached. Alternatively, one can question whether the econometric estimations that led to these results are correctly specified. In order to test the second hypothesis, we concentrate on two potential sources of mis-specification: (a) the above-mentioned “real-time” problem and (b) the choice of explanatory variables and the way they actually enter the Bundesbank's reaction function. The first source of mis-specification relates to the fact that most empirical studies of the Bundesbank's monetary policy use the latest vintage of data available to the authors, i.e., they are based on ex-post revised data. This may not be adequate for the analysis of past monetary policy decisions, since some of the relevant data and estimates undergo major revisions in the course of time. By re-estimating policy reaction functions for the Fed, Orphanides has shown that the use of real-time information can considerably change the outcome of an analysis of past monetary policy decisions.2 To test whether this is the case for the Bundesbank's reaction function as well, we have compiled a real-time data set which includes real and nominal output, the Bundesbank's own estimates of potential output, the rate of change in the consumer price index and the growth rate of the official monetary-target variable. Apart from revised data, another source of mis-specification may be the choice of explanatory variables and the way they enter the reaction function. As a rule, the Bundesbank explained its monetary-policy measures with respect to its final goal, price stability, and its intermediate target, money growth. In this context, the “overall economic situation” was also taken into account. Interestingly, the Bundesbank's regular reports (as well as the internal briefing material available for the period in question) suggest that when assessing the overall economic situation, policy makers often focused on the growth rate of output relative to the growth rate of production potential rather than on the level of output relative to the level of potential. 3 In fact, for much of the period in question, it was more difficult to recover the Bundesbank's real-time estimates of the level of the overall production potential than estimates of the corresponding growth rates. 4 We, therefore, add the (real-time) change in the output gap (which is equivalent to growth in output relative to growth in potential) as well as the (real-time) growth rate of money relative to target to the set of explanatory variables in the Bundesbank's reaction function. The paper is organized as follows. In Section 2, we present the structure of our real-time data set and discuss the extent of the revisions. Section 3 contains the econometric approach. In the fourth part, we use our real-time data to re-examine the estimates of the Bundesbank's reaction function used in the empirical literature. We find that using real-time data instead of ex-post data considerably changes the results. Specifically, the coefficient of expected inflation (relative to target) becomes insignificant, which, in our view, strongly suggests that the underlying model is mis-specified. We, therefore, enlarge the set of explanatory variables to include the (real-time) growth rate of money relative to target as well as the (real-time) change in the output gap. We find that both additional variables are highly significant, whereas the level of the real-time output gap drops out. The final section interprets the results and concludes.

Taken together, our results throw serious doubts on the widespread view that the monetary policy of the Bundesbank can well be described by a generalized Taylor rule. Instead, they suggest that policy makers at the Bundesbank followed a strategy of “flexible” monetary targeting with a significant response to the money-growth gap as well as to the (expected) inflation gap and the output-growth gap. For several reasons, we believe that this result is very much in line with the Bundesbank's own descriptions of its strategy: • In our view, the significance of the money-growth gap reflects the Bundesbank's commitment to its monetary targets. As mentioned above, the derivation of the targets was based on the Bundesbank's conviction, that in the long run inflation is pinned down by the economy's steady-state growth rate of money. The monetary targets were thus intended to anchor long-term inflation expectations as well as the trend rate of inflation. Another important aspect was to give guidance to other policy makers, especially to fiscal and wage policy. In terms of the more recent literature on the time inconsistency of optimal monetary policy, policy makers at the Bundesbank used the monetary targets as a commitment device to avoid the “classic” inflation bias of discretionary policy.27 • But why should the Bundesbank react to deviations of money growth from target independently of its concern about (expected) inflation? Several (non-exclusive) explanations are possible. First, in order to make the commitment to the target credible, the Bundesbank had to show some response to deviations of money growth from target even at times when inflation and output growth were in line with the respective targets.28 Second, in view of the long transmission lags, the money-growth gap might contain information about future price developments over a longer horizon than the four to six quarters of the inflation-gap variable we have considered. To put it in the words of Karl Klasen, President of the Bundesbank from 1970 to 1977: “in the absence of the monetary target, we would not have responded so early or so often”. 29 Third, monetary aggregates may contain useful information for specific (often unobserved) economic developments, e.g., the build-up of financial-market turbulences and asset-price bubbles. Finally, and closely related to the topic of our paper, as data on money growth were subject to very few revisions in Germany during the period in question, 30 they may have played a significant role in providing timely and steady information about the state of the economy. 31 On the other hand, the formulation of the target as a corridor and the willingness to tolerate temporary deviations from this corridor, left some room for other considerations or “side targets”, as they were called. It was this flexibility which led Otmar Issing (chief economist from 1990 to 1998) to describe the Bundesbank's variant of monetary targeting as “pragmatic monetarism” or “disciplined discretion”.32 Helmut Schlesinger, one of the “fathers” of the Bundesbank's monetary-targeting strategy, once stated that the monetary policy of the Bundesbank could be characterized as a mixture of two “pure” strategies: a medium-term monetarist and an anticyclical orientation.33 He stressed the importance of the anticyclical component, because typically an economy is not in a general equilibrium situation, and therefore one of the preconditions of the pure medium-term monetarist strategy – which aims at maintaining an equilibrium situation – is not met. Therefore, “the Bundesbank was moderately anticyclical, but at heart took a long-term view,” 34 giving justification to the inclusion of short-term objectives like the stabilization of inflation and output – or output growth – around their steady state values in the Bundesbank's reaction function. In order to assess an economy's current position in the business cycle and its likely development over the horizon relevant to monetary policy, both the level of the output gap as well as the change in the gap are certainly important. However, the real-time measurement problem seems to be much more severe in case of the level compared to the change of the output gap. Given this high degree of uncertainty, one would not expect the Bundesbank to put a strong weight on the real-time level of the output gap which is confirmed by our estimation results. In this respect, our results are in line with those of Orphanides (2003a), who concludes that successful central banks – like the Bundesbank and the Fed after the appointment of Volcker – have placed much less emphasis on (real-time estimates of) the output gap than suggested by simple activist policy rules such as the Taylor rule. The simultaneous presence of the inflation gap and the output-growth gap in the Bundesbank's reaction function (and the closeness of the point estimates of their coefficients), suggests a certain kinship of the Bundesbank's policy with the concept of nominal-income-growth targeting, which has been advocated, among others, by Tobin (1980), McCallum (1985) and McCallum and Nelson (1999). Recently, academic interest in policy rules, which focus on growth rates (differences) as opposed to levels, has revived. Orphanides (2003a), Walsh (2003) and others have presented evidence that “difference rules” or “speed limit policies” may perform well in the presence of imperfect information about the level of potential output.35 Whether these results are robust to other sources of uncertainty has yet to be proved. However, our finding that the Bundesbank during its most successful period (1979–1998) responded to deviations of money growth, inflation and output growth from their respective targets lends support to the view that difference rules deserve serious consideration.