This Selected Issues paper examines the long-term issues with pension expenditures in the Netherlands. The paper highlights that the public pension for a single person is equal to 70 percent of the (statutory) minimum wage. The minimum wage and public pensions thus move in lock-step; they are both by law indexed to contract wages in the private sector. This paper examines the structural policies of the Netherlands. Real wages and employment growth are also analyzed.

Abstract

This Selected Issues paper examines the long-term issues with pension expenditures in the Netherlands. The paper highlights that the public pension for a single person is equal to 70 percent of the (statutory) minimum wage. The minimum wage and public pensions thus move in lock-step; they are both by law indexed to contract wages in the private sector. This paper examines the structural policies of the Netherlands. Real wages and employment growth are also analyzed.

IV. Aspects of Monetary Policy Transmission 1/

1. Introduction

The mechanism through which the impact of monetary policy is transmitted to the economy has received considerable attention in recent years. While strong links between monetary or credit aggregates and nominal GDP have been well established at least since the work of Friedman and Schwartz (1963), significant questions remain concerning the precise impact of monetary policy on the economy—specifically, the pattern according to which adjustments of a central bank’s monetary policy instruments affect its intermediate targets and, ultimately, its nominal and/or real targets. The transmission mechanism for monetary policy, and appropriate methods to assess this, have recently become the subject of renewed theoretical debate—for example as regards often puzzling patterns in the response of different components of aggregate demand and the question whether monetary policy can play a counter-cyclical role. A better understanding of these issues is clearly important from a policymaker’s point of view.

This paper explores the patterns of monetary transmission in the case of the Netherlands. As in other countries, renewed attention has focused on this issue in the last few years, giving rise to a number of systematic empirical investigations of the transmission mechanism. In the past, the setting in which Dutch monetary policy has been implemented may have resulted in this being viewed as a second order issue for research—monetary policy has been clearly subordinated to supporting the exchange rate peg, and there has been no sense that it should be used as a countercyclical instrument. The question how monetary disturbances are transmitted through the economy has, however, recently become more policy-relevant.

Since early 1993, the guilder has appreciated slightly relative to its deutsche mark central rate and a (sometimes substantial) negative short-term interest rate differential has emerged vis-à-vis Germany. With the persistence of this trend, which could necessitate a further widening of the interest rate differential, the question arises how such developments may affect the economy. The prospect of EMU has also provided an impetus for researching the monetary transmission mechanism in the Netherlands, as in other EU countries, notwithstanding the fact that any conclusions in this area based on econometric analysis are likely to be exposed to the Lucas critique concerning inferences during periods of changing policy settings. 1/ These issues complement the more general reasons for investigating patterns and lags in the monetary transmission mechanism—including the light this can shed on an appraisal of monetary conditions and of the economy’s cyclical position.

On the methodological front, progress with vector autoregression (VAR) techniques, especially since the early 1980s, has provided an important impetus for research on the transmission mechanism. The attractiveness of this methodology reflects its simplicity and the need to impose relatively few restrictions to achieve parameter identification. At the same time, it is well-suited to address the impact of policy changes in ways that sidestep some of the criticisms levied against the large-scale reduced-form simultaneous equations models that were used extensively in the past.

This paper employs variants of the VAR methodology to gain insight into the main patterns of monetary transmission in the Netherlands. While it shares a number of features with other recent work on the subject, it also deviates in at least one major respect. Empirical work on the transmission mechanism in the Netherlands, as in other EU countries, has typically focused on a short-term domestic interest rate as the monetary policy instrument, very much in line with VAR models on the transmission mechanism for the U.S. economy. While such a choice would indeed appear natural for a relatively closed economy whose monetary authorities do not target the exchange rate, it appears less appropriate for a small open economy where monetary policy is subordinated to supporting a bilateral peg. This paper therefore examines separately the impact of changes in the short-term interest rate of the anchor currency and, respectively, the domestic interest rate premium. There are good theoretical reasons to postulate that the two components of the interest rate may affect the economy differently. Moreover, such a decomposition captures better the context in which Dutch monetary policy has been conducted—and may shed light on changes that will occur when the domestic interest rate premium disappears under EMU.

The estimation results presented in this paper suggest that this decomposition of the Dutch interest rate into a German rate and a domestic premium component is indeed fruitful. An important conclusion is that the impact of the interest rate premium on a number of key macroeconomic variables, notably prices, is rather strong. The paper proceeds to explore the implications of this conclusion for the conduct of monetary policy in the current context, and then examines its relevance as regards the likely impact of EMU. However, discussion of this latter issue is necessarily rather tentative, as most of the relevant inferences are subject to the reservation that existing stable relationships may break down with the occurrence of a regime change.

The plan of the paper is as follows. Section II describes the VAR methodology and provides a summary of recent research on the transmission mechanism in the Netherlands. Section III explores the impact of a change in the Dutch short-term interest rate, employing the standard VAR methodology. Section IV examines separately the impact of a change in the short-term interest rate of the anchor country and in the domestic interest rate premium on the basis of VAR models embodying different identification restrictions. Section V discusses the potential impact of EMU, in the light of the empirical results. Section VI concludes.

2. The VAR methodology and a survey of the literature

Empirical investigation of the monetary policy transmission mechanism typically attempts to describe how adjustment of a policy instrument affects intermediate monetary targets and, ultimately, key nominal and/or real variables that the central bank is seeking to influence. Since the application of the unrestricted VAR methodology by Sims (1980, 1981, 1982) to the study of monetary transmission in the United States, and its further refinement by Litterman and Weiss (1984), unrestricted VAR models have become a standard empirical tool in this area. In fact, this methodology has four notable advantages over large-scale reduced-form simultaneous equation models. 1/

First, in contrast to reduced-form simultaneous equation models, VAR models often do not impose a priori restrictions, such as exclusion of potential explanatory variables from the individual equations; significantly fewer such restrictions are required to achieve identification of the model’s parameters. Since economic theory is rarely so well defined as to suggest strong exogeneity assumptions (indeed competing models suggest very different exclusion restrictions), the rather “atheoretical” approach of VAR models has the advantage of allowing the historical data to “tell their own story”—without depending on the (often artificial) restrictions imposed by most large-scale reduced-form models.

Second, the richness of the lag structure in unrestricted VAR models provides a good safeguard against a host of econometric problems—notably spurious correlation and cointegration problems—that have been grounds for criticism of large-scale macro models.

Third, despite this richer lag structure, such models can be consistently estimated over much shorter periods: this is particularly helpful where the framework within which monetary policy has been conducted has undergone fundamental changes, or indeed the underlying economic structure has changed substantially over time.

Finally, unrestricted VAR models investigate the impact of economic policy by examining the effect of a shock only to the “innovation”—i.e., the estimated error term—of a policy variable, rather than its forecastable component. 1/ This contrasts with the approach in large-scale reduced-form models where policy variables are usually treated as exogenous: in such models, a policy shock is typically defined as a change of a particular magnitude and duration. To the extent this tends to impose deviations from the authorities’ policy reaction function, simulations using such models may be particularly vulnerable to the Lucas critique concerning inferences under circumstances where there is a shift in the policy setting. 2/

These advantages of unrestricted VAR models, however, come at a cost: since the testable implications of competing theories are not embedded into their formulation, they offer at best a rather loose account of how alternative theoretical models fit the observed patterns. 3/

A formal description of the standard unrestricted VAR model is provided in the technical annex. Here, we summarize its main features and the analytical tools it employs. A VAR model attempts to characterize the joint behavior of a set of economic variables; for the specific case of the monetary transmission mechanism, these would typically include policy instruments, intermediate transmission variables, and final target nominal and real variables (usually prices and real activity). For estimation purposes, no restrictions are placed on the lag structure of the model (except its length). On the other hand, for forecasting and policy analysis purposes, restrictions need to be imposed on the patterns of contemporaneous interaction between the variables under consideration. These restrictions crucially affect the so-called impulse-response functions. which summarize the expected impact of a shock to the innovation of a variable on all variables in the system, including itself, over a specified period of time.

In the standard unrestricted VAR model a strictly recursive contemporaneous structure is imposed: the variables are placed in a particular order, with the innovation of the first (typically a policy variable) assumed not to be affected contemporaneously by the innovation of any of the other variables, the innovation of the second assumed to be affected only by the innovation of the first, and so on. Therefore, results obviously depend on the choice of the ordering of the variables, and some sensitivity analysis involving altering this ordering is often pursued. Fortunately, for the purposes of the present paper which is mainly interested in tracing the impact of the policy instrument (which would naturally be placed first in the ordering), the results are not sensitive to the ordering chosen for the remaining variables. On the other hand, the assumed recursive structure itself could conceivably cause problems, which are, however, addressed by extensions of the standard VAR. 1/

Following the pioneering application of unrestricted VAR models to the study of monetary transmission in the United States, the methodology has more recently been applied to a number of European countries. 2/ In the case of the Netherlands, a substantial portion of the empirical work on the transmission mechanism, e.g. Boeschoten, van Els and Bikker (1994), Boeschoten and van Els (1995a), and Garretsen and Swank (1994), has relied heavily on the unrestricted VAR methodology.

The results of such work on the Dutch economy have so far been rather mixed. Boeschoten, van Els and Bikker (1994) and Boeschoten and van Els (1995a) estimate a VAR model over the period 1973-1993. 3/ The authors conclude that the impact on economic activity of a shock to the short-term interest rate is very weak. By contrast, the impact on real GDP of a shock to the long-term interest rate is both strong and persistent. On the other hand, the estimated impact of an innovation to interest rates on prices is rather puzzling: an increase in short- and long-term interest rates was estimated to result in higher prices, even after 16 quarters, leading the authors to conclude that this positive effect may reflect the expected inflation content of interest rates, thus indirectly casting doubt on the choice of the short-term rate as a proxy for the monetary policy instrument.

Garretsen and Swank (1994) employ a VAR system estimated over a shorter period, 1979-1993, coinciding with the operation of the ERM. 1/ The choice of variables is somewhat different from the papers summarized in the previous paragraph: for instance, credit is included along with money (M2)—both in real terms—a nominal effective exchange rate is introduced, and economic activity is proxied by industrial production rather than GDP. A more fundamental difference relates to the inclusion of real bond holdings of banks as part of the monetary transmission chain, suggesting that banks may respond to a monetary tightening by liquidating part of their bond portfolio. The authors’ conclusions are strikingly different from those of the literature summarized above: a positive shock to the short-term interest rate is estimated to have no discernible impact on prices but a persistent depressing impact on economic activity, with a negative impact on industrial production persisting even after four years, 2/ leading the authors to conclude that aggregate supply in the Netherlands is horizontal; this would obviously provide significant scope for countercyclical monetary policy. It should be emphasized, however, that such comparisons are hampered by the authors’ failure to indicate whether these estimated effects are statistically significantly different from zero.

In parallel to this recent analytical work using unrestricted VAR models, research on the transmission mechanism in the Netherlands has also continued in the framework of large-scale simultaneous equations models of the economy—notably Mallekoote and Moonen (1994) and Boeschoten and van Els (1995b). This work has focussed on the impact of a change in short-term interest rates, also regarded as a proxy for the Netherlands Bank’s policy instruments, often distinguishing between permanent and temporary changes. The main conclusions of this body of work imply that interest rate changes exert only a moderate impact on real GDP, 3/ and a somewhat larger and more persistent impact on prices. In terms of the transmission pattern of monetary policy, an exchange rate channel (impact of the short-term rate on the effective exchange rate) and a term structure effect (impact of the short-term rate on the long-term rate) are identified as particularly strong. By contrast, the results point to the money or credit transmission channels as being very weak.

3. The impact on the economy of the Dutch interest rate

In this section, a 7-variable VAR system is studied to explore some of the features of the monetary policy transmission mechanism in the Netherlands during the period 1984-1995, using monthly data. 1/ This sample period covers the period of ERM participation for the guilder during which its deutsche mark central parity remained unchanged. 2/ Since the hardening of the exchange rate peg at the beginning of this period could be viewed as a fundamental change in the setting for macroeconomic policy, empirical results based on the more recent sample may provide a more reliable basis for the study of the impact of monetary policy, relative to those based on a sample that covered the entire EMS (or a longer) period.

An important question concerns the choice of the variable to serve as a proxy for the monetary policy instrument. A natural choice would obviously be one of the Netherlands Bank’s official intervention rates, notably the rate on special loans which closely corresponds to the Bundesbank’s repo rate. However, given the changes over time in the central bank’s instruments and intermediate targets, 3/ it is difficult to identify an official interest rate that could accurately be described as the key policy instrument during the entire period under consideration. Accordingly, and in line with other research in this area—e.g. Boeschoten, van Els and Bikker (1994) and Garretsen and Swank (1994)—a short-term money market rate was chosen as the closest proxy of the policy instrument. 4/

Four types of intermediate or transmission variables were included in the VAR: the nominal effective exchange rate, which can be expected to be important in the transmission in view of the degree of openness of the Dutch economy; a long-term interest rate; money and credit aggregates. Prices and real GDP were included as the final target variables. 5/

Chart 1 presents the impulse response functions for each of the variables of the VAR, over a period of 24 months ahead, along with the 5 percent confidence intervals around the estimated impulse responses. 6/

CHART 1
CHART 1

NETHERLANDS: Impulse Responses: 7-Variable System 1/

Citation: IMF Staff Country Reports 1996, 080; 10.5089/9781451829303.002.A004

1/ For each column an asterisk (*) denotes the variable in which shock originates (for codes see Appendix II).
CHART 1a
CHART 1a

NETHERLANDS: Impulse Responses: 7-Variable System 1/

Citation: IMF Staff Country Reports 1996, 080; 10.5089/9781451829303.002.A004

1/ For each column an asterisk (*) denotes the variable in which shock originates (for codes see Appendix II).

As described in the previous section, the impulse response functions trace out how the impact of an innovation (i.e., shock) in each variable is propagated through the system, taking into account the interdependencies with all other variables implied by the model. For the purposes of this paper, and to facilitate comparison with other work in this area, a shock was defined as a 1 percent increase in the innovation of each variable. 1/

A first and main conclusion that emerges from the impulse response functions is that the impact of a change in short-term interest rates on real activity is remarkably weak. Specifically, a 1 percentage point shock to the short-term interest rate has no discernible impact on real GDP, even in the short-run. 2/ This conclusion is striking in that the experiments corresponds to an unanticipated tightening of monetary policy, which even under strong rational expectations is regarded as likely to have a short-run dampening effect on real activity. 3/ This result is largely in line with—albeit somewhat stronger than—the conclusions of Boeschoten, van Els and Bikker (1994), even though the sample here is much shorter. By contrast, the conclusions here lend no support to the finding of Garretsen and Swank (1994) that there is a substantial and persistent depressing effect of a rise in the short-term interest rate on economic activity.

A second conclusion is that there is a discernible relationship between an increase in the short-term interest rate and the price level. Thus, a 1 percentage point increase in the short-term interest rate is estimated to entail a depressing effect on the CPI, which returns to its baseline path only toward the end of the time horizon under consideration. Moreover, the estimated effect is significantly different from zero for at least 7 months after the shock. In conjunction with the above result on real activity, this would suggest rather flexible prices, permitting a relatively swift pass-through of a monetary policy shock to prices rather than quantities.

A third finding is that the impact of a shock to the policy instrument on the other intermediate transmission variables in the model is rather weak, in all cases being insignificantly different from zero. Thus, following a 1 percentage point rise in the short-term interest rate, the nominal effective exchange rate is estimated to appreciate only moderately, and there is no discernible impact on money (very much in line with the findings of Boeschoten, van Els and Bikker (1994)). More puzzlingly, perhaps, a 1 percentage point shock to the short-term interest rate, even if it persists for more than one year, has no discernible impact on the long term rate beyond the first two months. In that sense, these results fail to identify a term structure effect—and in this they differ from the recent studies based on large-scale simultaneous equation models of the Dutch economy, in which a term structure effect figures prominently. 1/

Finally, as regards the impact on the economy of shocks to other intermediate policy variables in the model, it is notable that the nominal effective exchange rate emerges as much more potent than the short-term interest rate. Indeed, a shock to the nominal effective exchange rate turns out to have a statistically significant impact on virtually all target and intermediate variables under consideration, with the exception of the money stock. Thus, following an unexpected 1 percent appreciation of the nominal effective rate the short-term interest rate is expected to fall, remaining some 0.5 percentage points below its baseline path even by the end of the time horizon under consideration; bank credit, the long-term interest rate, the price level, as well as GDP are also expected to fall. The strength of these effects clearly underlines the openness of the Dutch economy.

4. The impact on the economy of the German rate and the domestic premium

In the preceding section, the Dutch money market rate was treated as an adequate proxy for the monetary policy instrument in the Netherlands, in line with previous research. In an important way, this approach neglects the central feature of Dutch monetary policy: the fact that throughout the period under consideration it has been geared to supporting the guilder’s peg to the DM. Viewed in this way, the Dutch short-term rate is composed of two distinct components: the anchor currency interest rate and the differential vis-à-vis the anchor currency interest rate.

Theoretical as well as policy related considerations would suggest that separate investigation of the impact of each component may be warranted. In the first place, to the extent that the exchange rate peg is regarded as reasonably credible, Dutch long-term rates could be expected to be largely driven by German long-term rates. Accordingly, it can be conjectured that a change in the Dutch short-term that reflects a change in the premium may have less of an effect on Dutch long-term rates than a change originating from a shift in the short-term interest rate in Germany. In that sense, the impact of a change in the domestic interest premium on aggregate demand may be conjectured to be smaller than the impact of an interest rate change originating in Germany (at least to the extent that a positive term structure effect is in operation).

A second area in which the two components of Dutch short-term rate might have differing effects is the effect of the premium on the path of the effective exchange rate. On the one hand, an increase in the premium might be associated with appreciation of the guilder, since during a period of general ERM stress the Dutch authorities were much more successful than other ERM members in maintaining the DM peg. On the other hand, vis-à-vis non-ERM currencies, increases in the premium might be expected to have less effect than German rates. It is difficult to ascertain a priori which effect is likely to be stronger, although the preponderance of intra-EU trade might point to a predominance of the premium.

These considerations appear to have considerable policy relevance in the present context. First, since 1993, the guilder has exhibited substantial appreciating tendencies, leading the Netherlands Bank to maintain its intervention rates significantly below the German ones. In this setting, if the impact of a change in the premium on the economy turns out to be substantial, there could be inherent limits to such a strategy if the risk of some destabilizing effects on the economy is to be avoided. Second, monetary union implies elimination of the differentials in interest rates that currently exist between ERM members. Thus, a quantification of the role of the interest rate premium in the transmission mechanism could provide some indication of the extent to which EMU could change the patterns according to which monetary shocks get propagated throughout the economy.

To investigate separately the impact of the German short-term interest rate and the domestic premium on the Dutch economy, an 8-variable unrestricted VAR is employed, with the Dutch money market rate replaced by the German money market rate and the Dutch-German money market rate differential. The remaining variables are the same as in the previous section, as is the lag structure and estimation period. Chart 2 presents the estimated impulse response functions of the 8-variable system.

CHART 2
CHART 2

NETHERLANDS: Impulse Responses: 8-Variable System 1/

Citation: IMF Staff Country Reports 1996, 080; 10.5089/9781451829303.002.A004

1/ For each column an asterisk (*) denotes the variable in which shock originates (for codes see Appendix II).
CHART 2a
CHART 2a

NETHERLANDS: Impulse Responses: 8-Variable System 1/

Citation: IMF Staff Country Reports 1996, 080; 10.5089/9781451829303.002.A004

1/ For each column an asterisk (*) denotes the variable in which shock originates (for codes see Appendix II).

a. Results from the impulse responses

One conclusion that emerges from the estimated impulse response functions is that the impact of a shock to the Dutch interest rate premium is certainly not weaker, and in many cases actually stronger than the impact of a shock of the same size in the German rate, with regard to both target and intermediate transmission variables. Specifically, both components of the short-term rate turned out not to have a statistically significant impact on real activity. On the other hand, an increase in the premium appears to have more of an impact in reducing inflation than an increase in the German short-term rate. While the point estimates of the impulse response function suggest a broadly similar picture for the effect of both components of the short-term interest rate, examination of the calculated standard errors around these point estimates tells a different story. The impact of the German short-term rate on the price level turns out to be statistically insignificantly different from zero almost throughout the time horizon under consideration, but the impact of the premium remains continuously significant (and negative) during the last 15 months of the period under consideration. This result is all the more striking if one takes into account that a shock to the premium is estimated to die out, with the variable returning to its baseline path, much more quickly than is the case for the German rate.

A second conclusion is that the impact of the two components of the short-term Dutch interest rate on a number of intermediate transmission variables also turned out to be significantly different. Indeed, the impact of changes in the premium turn out in several cases to be both stronger and in an intuitively more intelligible direction. Thus, the impact of the premium on bank credit is not only statistically significant, but is also opposite in sign relative to the impact of the German rate: a 1 percentage point shock is expected to lead to a reduction of bank credit by some 7 percent below its baseline path six months after the shock, a pattern that persists for twelve months. By contrast, the estimated impact on credit of a 1 percentage point increase in the German rate is positive (although statistically insignificantly different from zero during the last 18 months of the time horizon under consideration). A similarly asymmetric pattern can be observed for the impact of the two components of the short-term interest rate on money, with an increase in the premium estimated to have a negative impact and an increase in the German rate a positive one.

With regard to the impact of each component of the short-term interest rate on the long-term rate, the results remain rather puzzling. Specifically, a 1 percentage point shock to the premium was estimated to be associated with an increase on the long term rate during the first year after the shock, peaking at some 0.4 percentage points above its baseline path; however, the change was significantly different from zero only during the first three months after the shock. By contrast, a 1 percentage point increase in the German rate was estimated to be associated with a negative impact on the long-term rate, with the impact turning positive only 15 months after the shock; moreover, this impact turned out to be insignificantly different from zero throughout the time horizon under consideration. As in the previous section, this approach again fails to identify a strong term structure effect.

The exchange rate once again turns out to be the most potent transmission variable with respect to most variables of the system under consideration, with the exception of the money and credit variables. Thus, a one percent appreciation of the nominal effective exchange rate was estimated to result in: an expected reduction in the long-term interest rate by some 0.25 percentage points below its baseline path, with the effect being statistically significant for more than one year after the shock; a depressing effect on prices, with the effect being statistically significant for just under a year; and a rather strong depressing effect on real economic activity, on average pushing real GDP down by 0.5 percentage points below its baseline path during the time horizon under consideration, with the effect being statistically significant for some 16 months after the shock.

The money and credit variables were once again shown to have a statistically insignificant impact on both prices and real GDP throughout the time horizon under consideration. This result underlines the endogeneity of these variables in a small open economy operating under a pegged exchange rate and near-perfect capital mobility, and is in line with the conclusions of other recent VAR-based work on the monetary transmission in the Netherlands. 1/

On the other hand, the estimated impact of the long-term interest rate weakens somewhat under the system that allows for different treatment of the German and domestic premium components of the Dutch short-term interest rate relative to the 7-variable system of the previous section: in particular, while a 1 percentage point increase in the long-term interest rate is associated with a strong and significant fall in the expected stock of money and credit, its effect on prices and real GDP turns out to be statistically insignificantly different from zero throughout the time horizon under consideration. This latter conclusion would appear to be in contrast to recent results based on large-scale simultaneous equation models of the Dutch economy—e.g., Mallekoote and Moonen (1994), and, to a lesser extent, on VAR-based approaches—e.g., Boeschoten, van Els and Bikker. 2/

b. Other econometric tests

The relative importance of the German component and the domestic premium component of the Dutch short-term interest rate in affecting the target nominal and real variables, as well as the intermediate transmission variables, can be also assessed by recourse to two other types of econometric tests. In the first place, block exogeneity tests were performed, which provide an indication of statistical significance, rather than magnitude of the effect. These tests are based on F-statistics, and look at the extent to which each variable (including all its lags) can be regarded as “exogenous” to (or, more loosely, not significantly affecting) the path of all other variables, including itself. On the basis of the calculated F-statistics, the significance level at which such exogeneity is rejected can then be computed.

On the specific question of the relative impact of the two components of the Dutch short-term rate, the results of the block exogeneity tests suggest that, at the 10 percent significance level, exogeneity of the premium can be rejected for five of the eight variables of the system: the German rate, 1/ the premium itself, bank credit, money, and the price level. By contrast, exogeneity of the German rate can be rejected for only two variables: the German rate itself and the price level. Moreover, the degree of certainty with which exogeneity with regard to the price level can be rejected is distinctly superior for the case of the premium relative to the German rate: the F-statistics suggest non-rejection of exogeneity at a significance levels of 1.4 percent versus 9.8 percent, respectively. The block exogeneity test thus essentially corroborates the conclusions regarding the significance of the estimated impact of the premium versus the German rate based on the confidence intervals around the impulse response functions of Chart 2.

To gain additional insight into the relative magnitude of the impact the German rate component and the domestic premium component of the Dutch short-term interest rate, variance decomposition tests were used. These tests attempt to capture the proportion of the variance of each variable that can be explained by other, contemporaneous and lagged, variables of the VAR system (including the variable itself). These variable decompositions are presented in Chart 3, where for expositional convenience the contribution of the own lags of each variable under consideration is not included (hence, the bars in the chart do not add up to 100 percent). 2/ The variance decomposition tests of Chart 3 lend additional support to the conclusions that the impact of the interest rate premium on almost all the variables under consideration is stronger than that of the German short-term interest rate—the only exception being real GDP, the contribution to the variance of which of each variable was, however, estimated to be less than 10 percent. With regard to the impact on the consumer price index, the variance decomposition test suggests that the effect of the interest rate premium is particularly strong: by the end of the time horizon under consideration, the premium accounts for around 40 percent of variance of the CPI (by contrast, the contribution of the German rate is less than 10 percent).

Chart 3
Chart 3

NETHERLANDS: Variance Decomposition

(Percentage Contributions)

Citation: IMF Staff Country Reports 1996, 080; 10.5089/9781451829303.002.A004

Chart 3a
Chart 3a

NETHERLANDS: Variance Decomposition

(Percentage Contributions)

Citation: IMF Staff Country Reports 1996, 080; 10.5089/9781451829303.002.A004

c. Sensitivity of results to alternative identifying restrictions

In this subsection we attempt to assess the sensitivity of the results obtained so far to the particular contemporaneous relationships between variables. As discussed above, the standard VAR imposes a strictly recursive structure on the contemporaneous correlations between the parameters of the VAR system. In particular, it imposes that policy variables do not respond to the contemporaneous realizations (as opposed to the lags) of any other variable of the system. For the specific case of monetary transmission in the Netherlands this could appear rather arbitrary from a theoretical perspective, as well as rather unrealistic given the context in which Dutch monetary policy has been conducted over the period under consideration. Given that monetary policy in the Netherlands has been geared to supporting the guilder’s deutsche mark peg, it could be surmised that the Netherlands Bank’s policy reaction functions could well involve within-period adjustment of its policy instrument in response to an exchange rate shock, as well as to a shock to the long-term interest rate to the extent that this reflects perceived policy credibility. To the extent that such contemporaneous feedbacks into the policy instrument are indeed important, their exclusion by standard methodology used in the previous sections could have affected our estimated impact of the policy instrument on certain key variables. Thus, it may account for our failure to identify a statistically significant term structure effect, which is an important element of large-scale simultaneous equations models of the Dutch economy.

Based on the considerations of the previous paragraph, we therefore specify the interest rate premium to depend contemporaneously both on the exchange rate and on the long-term interest rate. Further deviating from strict recursiveness, we also allowed innovations in the long-term interest rate to be contemporaneously reflected in changes in the exchange rate and the long-term interest rates to be contemporaneously affected by innovations in prices. 1/

The impulse-response functions estimated on the basis of the restrictions described above are presented in Chart 4 and suggest that the overall picture of the transmission mechanism in the Netherlands is not materially different from the one derived on the basis of the standard VAR model. In particular, the effects of innovations in the instruments on other variables remain almost unchanged; the downward impact of the premium on prices is still strong, while the expected effect on the long-term interest rate of changes in German rates and the premium are well within the confidence intervals shown in Chart 2.

CHART 4
CHART 4

NETHERLANDS: Impulse Responses: Structural VAR 1/

Citation: IMF Staff Country Reports 1996, 080; 10.5089/9781451829303.002.A004

1/ For each column an asterisk (*) denotes the variable in which shock originates (for codes see Appendix II).
CHART 4a
CHART 4a

NETHERLANDS: Impulse Responses: Structural VAR 1/

Citation: IMF Staff Country Reports 1996, 080; 10.5089/9781451829303.002.A004

1/ For each column an asterisk (*) denotes the variable in which shock originates (for codes see Appendix II).

On the other hand, as regards the impact of an unexpected change in intermediate transmission variables, a few small differences can be detected. In particular, a positive innovation to the long-term interest rates is now associated with a rise of the premium and a depreciation of the exchange rate in the short run (the magnitude of the differences is, nevertheless, of slightly less than 2 standard deviations).

d. Policy implications

In sum, the estimated impulse response functions, block exogeneity tests, and variance decompositions, taken together, suggest that the domestic interest rate premium has played at least as important a role in the monetary transmission mechanism as the German component of the Dutch short-term interest rate. As regards prices, in particular, it appears that the impact of the premium is both strong and statistically significant.

These results suggest that, to the extent independent interest rate adjustments are made by the central bank to support the bilateral exchange rate peg, these adjustments should not be viewed as neutral with regard to the course of prices—with independent cuts carrying, beyond a certain point, some potential inflationary risk (and conversely). However, this should not be interpreted as implying an inherent inflationary bias associated with the choice of the deutsche mark anchor: 1/ the point is that, ceteris paribus, some degree of greater inflation variability could be an inherent by-product of such a monetary policy strategy.

On a more general note, these considerations illustrate the dilemmas and choices that policymakers can face in the run-up to monetary union. A potential policy tradeoff can be identified involving, on the one hand, greater emphasis on exchange rate fixity, as implied by EMU-related considerations, and on the other the of limiting undesirable interest rate movements by allowing part of the adjustment to occur through the nominal exchange rate. 2/ It should be emphasized, however, that the choices involved are in general not straightforward, and that the cyclical position of the economy may in fact be quite relevant. The estimation results presented here would suggest that, for the case of the Dutch economy in the current context, emphasis on exchange rate fixity would probably imply some inflationary impact, albeit of a modest amount in absolute terms, but that the alternative option, while potentially holding inflation more tightly in check, would probably entail some depressing impact on real activity.

5. The impact of EMU

The question arises whether the empirical results discussed above have any bearing on an assessment of the likely impact of EMU on the Dutch economy. It should be emphasized at the outset that any conclusions in this area would be of a highly tentative nature, as the Lucas critique concerning inferences during periods of changing policy settings is likely to be particularly relevant. After all, monetary union would constitute a fundamental policy regime shift, entailing major changes in the monetary authorities’ policy instruments, as well as intermediate and ultimate targets. Under these conditions, parameter estimates derived on the basis of a previous monetary regime should be treated with a great deal of caution when utilized as the basis of trying to assess what the main features of the new regime are likely to be. While the VAR methodology employed in this paper can avoid some of the Lucas critique-type problems that affect large-scale simultaneous equation econometric models (provided that the estimation period and policy shocks are appropriately specified), the VAR methodology remains after all a reduced-form equation model, and parameter estimates derived along these lines are unlikely to remain invariant to a regime shift of this magnitude. 1/

With these caveats in mind, the results reported above should provide at least some qualitative indications of the likely impact of EMU on the monetary transmission mechanism. In this regard, the separate analysis of the impact of the German rate and the domestic premium component of the Dutch short-term interest rate proves rather useful. In fact, one could investigate the likely impact of monetary union on the transmission mechanism, at least to some extent, in terms of the impact of the elimination of this interest rate premium.

The empirical results of the previous sections suggested a rather strong, persistent, and statistically significant effect of the premium on prices. At first glance, it could be argued that, given the impact of the premium on the price level, elimination of the scope for a premium would be rather costly, in effect depriving the Dutch monetary authorities of an important policy instrument. In that sense, it could be argued that, with EMU, the authorities would no longer be in a position to offset the impact of an asymmetric real or (especially) monetary shock relative to the anchor currency country. To that extent, EMU might render the Dutch economy more vulnerable to the cycle. 2/ The costs associated with this loss of a policy instrument would be rather difficult to calculate, but would in principle depend crucially on the incidence of such asymmetric shocks, the degree of labor and product market flexibility in the Netherlands, and a host of related factors.

The main problem with applying this type of argument to the Netherlands relates to the fact that it essentially ignores the mode in which Dutch monetary policy has been pursued. In fact, the Dutch monetary authorities have probably very rarely used their interest rate policy as a countercyclical (or, more narrowly, as a short-run anti-inflationary) tool, at least over the period under consideration; instead, the primary function of Dutch interest rate policy has been to support the bilateral exchange rate peg to the deutsche mark, and to anchor the price level indirectly in this way. Viewed thus, the question of whether the domestic premium could have been used effectively as an anti-inflationary instrument becomes rather academic. At all events, EMU should not be characterized as involving a cost deriving from the loss of a policy instrument that had almost never been used with an eye to domestic policy objectives.

Moreover, given this primary function of interest rate policy, the domestic interest premium could be viewed as a (generally unintended) implication of the monetary authorities’ attempt to support the exchange rate peg. 1/ In that sense, and given the estimated strength of the premium’s impact on prices, it could be argued that the interest rate premium has constituted an additional source of inflation variability. To the extent that the variability of inflation (in addition to its level) entails a cost, what was previously viewed as a net loss associated with EMU under the “loss of a policy instrument” argument could instead be featured as a net gain. In effect, EMU, by eliminating the need for an interest rate premium to support the guilder’s “parity” vis-à-vis other members of the monetary union, it could be expected to reduce inflation variability.

It would appear worthwhile to obtain at least a broad sense of the magnitude of the impact just described. Since the postulated benefit is related to the variability of inflation, resorting to variance decomposition test, like those presented in Chart 3, would appear a natural choice of framework for addressing this question. It should be recalled that a variance decomposition test provides the contribution to each variable’s variance by all variables of the VAR system, including the (lagged values of the) variable itself. For the specific case of the price variable, we can calculate that, over the two-year time horizon considered in this paper, the domestic interest rate premium was responsible for more than 25 percent of total price variability. While such quantitative estimates should be treated with considerable caution in view of the likely prevalence of the Lucas critique, as discussed at the beginning of this section, this result suggests that the likely impact of EMU in reducing inflation variability could be significant.

6. Summary and concluding remarks

This paper has explored aspects of the monetary transmission mechanism in the Netherlands, within the overall framework set by the exchange rate anchor, in an attempt to identify the main channels through which monetary impulses affect the performance of the economy. The main objective was to identify the impact of domestic monetary policy instruments on final policy target variables—essentially prices and real activity—as well as on intermediate transmission variables such as the exchange rate, money and credit aggregates, and long-term interest rates. The discussion in the paper relies on some variants of the VAR methodology, which has proved a particularly attractive technique for studying such effects.

The main innovation of the paper relates to the specification of the monetary policy instrument. Most of the literature in this area focuses on a domestic short-term interest rate as the policy instrument. However, for a small open economy under a fixed exchange rate regime, there is a case for examining separately the impact of the anchor country interest rate component and the domestic premium component of the short-term interest rate. The two components could well be expected on theoretical grounds not to have an identical impact on the variables of interest. Furthermore, such a decomposition may better capture the framework within which Dutch monetary policy is conducted, to the extent that the domestic premium may better represent the policy instrument at the monetary authorities’ disposal.

The results presented in this paper suggest that neither component of the short-term interest rate has a significant impact on real activity, even in the very short-run. By contrast, a significant impact on prices was identified; indeed, the domestic premium component, in particular, turned out to have a strong and rather persistent impact on the price level in the conventional direction. The results also pointed to a discernible impact of the domestic premium on several key intermediate transmission variables.

The paper then explored the policy relevance of these conclusions. On the one hand, it was argued that the strong estimated impact of the domestic premium on prices pointed to some rather delicate tradeoffs involved in a pegged exchange rate regime. This issue appeared particularly relevant to the current situation in the Netherlands, where the guilder’s underlying strength has led the monetary authorities to cut interest rates significantly, opening up a substantial negative interest rate differential vis-à-vis Germany at the short end. The results of this paper suggest that there are inherent limits to such an approach if potential risks of an adverse effect on inflation are to be contained; on the other hand, there are costs in terms of economic growth to an unchecked appreciation.

The results presented in the paper also suggest some possible implications for the impact of EMU on the Dutch economy—albeit subject to the caveat that applies to any extension of inferences to the period that follows a regime change. On the one hand, it was argued that the longstanding subordination of Dutch monetary policy to support of the exchange rate peg rendered the relevance of the welfare losses associated with the loss of a policy instrument rather limited. On the other hand, the disappearance of the domestic interest rate premium under EMU could be seen as eliminating a source inflation variability for the Dutch economy.

APPENDIX I: The VAR Methodology

1. Identification Issues

Unrestricted VAR models of the monetary transmission mechanism attempt to explain a set of variables in terms of the lags of all the variables under consideration. Thus, suppose that Yt is a k × 1 vector of economic variables (policy instruments, intermediate targets or other transmission variables, final targets), observed at time t, whose joint behavior is of interest. It is postulated that the behavior of Yt is governed by the following model:

Yt=Σ0nBiYti+Aut(1)

where Bi is an unrestricted k × k matrix of coefficients, ut is a vector of serially uncorrelated disturbances, and E(utut’) is a diagonal matrix. It is the unrestricted nature of matrix Bi which is the main source of the methodology’s attractiveness. In typical VAR models, A is usually constrained to be an identity matrix.

Estimation of the system described in (1) is relatively straightforward. The system of equations summarized by equation (1) can then be solved, expressing each contemporaneous variable as a function only of the lags of the variables under consideration. Thus, a reduced form of (1) can be written as follows:

Yt=Σ1nCiYti+yt(2)

where Ci = (1-B0)-1 Bi and yt is a serially uncorrelated vector of residuals. The vector yt, the error term of the reduced-form equations, in turn satisfies:

yt=B0yt+Aut(3)

As a first step, the reduced form system (2) can be efficiently estimated by unrestricted ordinary least squares, obtaining the estimated coefficients of matrix C. These parameter estimates, however, are of no special interest in themselves, and are typically not even presented. Instead, these parameter estimates are utilized to construct estimates of yt, the vector of residuals or innovations.

A VAR model wants to study the impact of changes in the elements of the vector of innovations yt on all the variables of the system. However, to achieve this, one typically utilizes a vector of orthogonal components of the elements of yt, i-.e. a vector whose elements are uncorrelated to each other, rather than the vector yt itself. There are a number of advantages in using the orthogonal components of yt over the (in general non-orthogonal) estimated vector. From an econometric viewpoint, because orthogonalized innovations are by definition uncorrelated, it is very simple to compute the variances of linear combinations of them. More fundamentally, to the extent that a variable has historically tended to move together with other variables, it would be rather misleading to talk about a shock to this variable in isolation; orthogonalization takes such co-movement into account. Once this orthogonalization is achieved, the so-called impulse response functions, which are viewed as capturing the essence of the transmission of shocks across the economy, can then be computed.

The particular way to achieve this orthogonal decomposition of vector yt is a crucial non-trivial aspect of the methodology; indeed, the particular type of orthogonalization employed is the main feature distinguishing the standard VAR described from alternative VAR methodologies. The method to achieve the orthogonalization of yt under the standard VAR is based on the Choleski decomposition, and in essence amounts to assuming that the matrix Bq is lower triangular, thus imposing a strictly recursive contemporaneous structure to the system. 1/

This unrestricted VAR approach has come under some criticism in recent years. While the strength of unrestricted VARs in providing a good characterization of the historical patterns of the data under consideration remains uncontroversial, their ability to aid in discriminating among structural hypotheses about the economy has been put into question. 2/ In particular, it has been emphasized that even if a general specification like our equation (1) of Section II in fact describes the true structure of the economy, the imposition of a strictly recursive contemporaneous structure is problematic, as it is usually not motivated in a satisfactory way by the relevant economic theory. It should be emphasized that the usual approach of ensuring that the results are not overly sensitive to the particular ordering of the variables under consideration that has been chosen does not remove the essential arbitrariness of the Choleski decomposition, which fundamentally emanates from restricting attention to strictly recursive models.

In view of the problem described above, recent research has focused on implementing econometric approaches which entail an orthogonalization based on a contemporaneous structure explicitly embodying testable hypotheses that can be readily derived from economic theory, while at the same time retaining the advantages of an unrestricted lag structure. For the specific case of monetary transmission in the Netherlands we imposed the interest rate premium and the exchange rate to depend contemporaneously on the long-term interest rate; and the latter to depend on prices. In order for the system to remain just-identified, and to ensure that the system of equations described by matrix Bq has a solution or that this matrix is invertible, we placed three additional exclusion restrictions on the contemporaneous structure. Specifically, we imposed no contemporaneous feedback from the effective exchange rate to domestic premium, 1/ and no contemporaneous feedback from bank credit to long-term interest rate, nor from the long-term interest rate to prices. 2/

The relevant methodology to compute the factorization implied by the above restrictions is rather complicated and it is not presented here. 3/ Suffice it to say that it involves minimizing the log-likelihood of a function resulting from multiplying a transformation of the Bq matrix by the sample variance-covariance matrix of the errors from the VAR estimation, under the assumption that the matrix A is diagonal.

2. Confidence intervals

Although the a VAR is efficiently estimated using ordinary least squares, confidence intervals for the impulse-response functions are quite difficult to estimate. Not only impulse responses are non-linear functions of the autoregressive coefficients obtained from the VAR, but the covariance matrix of the autoregressive coefficients is not the same as that for OLS (because of the cross-equations covariances). In fact, the variance of the posterior distributions of the coefficients depends on the posterior distribution of the (orthogonalized) VAR errors, which is described by a Normal-inverse Wishart function (Zellner, 1971). Owing to the non-linearity of the impulse-response functions, their standard errors are seldom computed using analytical methods, but instead are estimated using Monte Carlo or bootstrapping methods (Hamilton, 1994). In the case of the Choleski decomposition we used the Monte Carlo method suggested in Doan (1992); but we were unable to compute confidence intervals for the decomposition allowing a non-recursive error structure. Nevertheless, the magnitude of the errors in both cases should be similar because the factorization matrix were not very different and both models are just-identified (imposing more restrictions tends to reduce the standard errors).

APPENDIX II: Definitions of Variables and Data Sources

Thus, the VAR to be estimated consists of the following variables, in the order presented (the variables and data sources are described in more detail in the data annex); a constant was also included. The variables are in logs, except for the interest rates which are in percent:

SHISH : Dutch 3-month money market rate (IFS).

SHIG : German 3-month money market rate (IFS).

DIFSH : Dutch 3-month domestic premium, i.e., the interest rate differential.

XRATE : Nominal effective exchange rate (MERM).

BKLOAN : Stock of banking loans (IFS databank).

LONGI : Dutch 10-year benchmark bond yield (IFS).

Ml : Stock of Ml, seasonally adjusted (IFS).

CPI : Consumer price index (IFS).

GDPSAL : Real GDP, interpolated from the raw quarterly data on the basis of the monthly retail sales volume (GDP from IFS, retail from OECD databank).

References

  • Barran, F., V. Coudert and B. Mojon, The Transmission of Monetary Policy in the European Countries,” CEPII Document de Travail No. 96-03, February 1996.

    • Search Google Scholar
    • Export Citation
  • Bank for International Settlements, “National Differences in the Interest Rate TransmissionBasle, March 1994.

  • Barro, R.J., Unanticipated Money Growth and Unemployment in the United States,” American Economic Review, Vol. 67 No. 2, March 1977.

    • Search Google Scholar
    • Export Citation
  • Barro, R. J., Unanticipated Money, Output, and the Price Level in the United States, Journal of Political Economy, Vol. 86, No. 4, August 1998.

    • Search Google Scholar
    • Export Citation
  • Bemanke, B. S., Alternative Explanations of the Money-Income Correlation,” Carnegie-Rochester Conference Series on Public Policy, Vol. 25, 1986.

    • Search Google Scholar
    • Export Citation
  • Bernanke, B. S. and A. S. Blinder, The Federal Funds Rate and the Channels of Monetary Transmission,” American Economic Review, Vol. 82, No. 4, September 1992.

    • Search Google Scholar
    • Export Citation
  • Bernanke, B.S. and M. Gertler, Inside the Black Box: The Credit Channel of Monetary Policy Transmission,” Journal of Economic Perspectives, Vol. 9, No. 4, Fall 1995.

    • Search Google Scholar
    • Export Citation
  • Blanchard., O. J. and M. W. Watson, Are Business Cycles All Alike?,” NBER Working Paper No. 1392, June 1984.

  • Boeschoten, W. C., P. J. A. van Els and J. A. Bikker, Monetary Transmission in a Small Open Economy: The Case of the Netherlands,” The Netherlands Bank. Research Memorandum WO&E No. 406/9413, August 1994.

    • Search Google Scholar
    • Export Citation
  • Boeschoten, W. C. and P. J. A. van Els, The Importance of Monetary Transmission, The Netherlands Bank. Quarterly Bulletin, March 1995a.

    • Search Google Scholar
    • Export Citation
  • Boeschoten, W. C. and P. J. A. van Els, Interest Rate Transmission in the Netherlands: Results for the Nederlandsche Bank’s Model MORKMON II,” in Bank for International Settlements. “Financial Structure and the Monetary Transmission Mechanism.” Basle, March 1995b.

    • Search Google Scholar
    • Export Citation
  • Cooley, T. F. and S. F. Leroy, Atheoretical Macroeconomics: A Critique,” Journal of Monetary Economics, 16, 1985.

  • Dale, S. and A. G. Haldane, A Simple Model of Money, Credit, and Aggregate Demand,” Bank of England Working Paper No. 7, 1993.

  • De Grauwe, P., The Economics of Monetary Integration,” Oxford: Oxford University Press, 1994.

  • Doan, T., Rats User’s Manual - Version 4, Evaston, Estima, 1992.

  • Escrivá, J. L. and A. G. Haldane, The Interest Rate Transmission Mechanism: Sectoral Estimates for Spain,” Bank of Spain. Working Paper No. 9414, 1994.

    • Search Google Scholar
    • Export Citation
  • Friedman, M. and A. Schwartz, A Monetary History of the United States,” Princeton: Princeton University Press, 1963.

  • Garretsen, H. and J. Swank, The Transmission of Interest Rate Changes and the Role of Bank Balance Sheets: A VAR Analysis for the Netherlands,” De Nederlandsche Bank. MEB Series No. 4, 1994.

    • Search Google Scholar
    • Export Citation
  • Keating, J., Identifying VAR Models under Rational Expectations,” Journal of Monetary Economics. 25, 1990.

  • Levy, J., Exchange Markets and Interest Rates in the Netherlands: Recent Experience,” in IMF, SM/94/100.

  • Litterman, R. and L. Weiss, Money, Real Interest Rates, and Output: A Reinterpretation of Postwar U.S. Data,” Econometrica. 53, 1985.

    • Search Google Scholar
    • Export Citation
  • Lucas, R. E., Jr., Econometric Policy Evaluation: A Critique,” in K. Brunner and A. H. Meltzer, The Phillips Curve and Labor Markets,” New York: North-Holland, 1976.

    • Search Google Scholar
    • Export Citation
  • Mallekoote, P. M. and R. Th. L. Moonen, The Effects of Changes in Interest Rates on the Dutch Economy,” in Bank for International Settlements, “National Differences in Interest Rate Transmission,” Basle, March 1994.

    • Search Google Scholar
    • Export Citation
  • Sims, C. A., Macroeconomics and Reality,” Econometrica. 48, 1980.

  • Sims, C. A., Policy Analysis with Econometric Models,” Brookings Papers on Economic Activity, 1982.

  • Sims, C. A.,Are Forecasting Models Useful for Policy Analysis?,” Federal Reserve Bank of Minneapolis Quarterly Review, Winter 1986.

    • Search Google Scholar
    • Export Citation
  • Sims, C. A., Interpreting the Macroeconomic Time Series Facts: The Effects of Monetary Policy,” European Economic Review. 36, 1992.

    • Search Google Scholar
    • Export Citation
  • Wellink, A. H. E. M., Experience Gained with Monetary Policy Instruments in the Netherlands,” in Bankhistorisches Archiv, “Monetary Policy Instruments: National Experiences and European Perspectives,” Frankfurt: 1994.

    • Search Google Scholar
    • Export Citation
  • Zellner, A., An Introduction to Bayesian Inference in Econometrics, New York, Wiley, 1971.

1/

Prepared by Ioannis Halikias and Joaquim Levy.

1/

Monetary union will clearly constitute a policy regime shift to some degree, entailing changes in the monetary authorities’ policy instruments, as well as intermediate and ultimate targets. For instance, the Dutch monetary authorities intend to introduce a set of policy instruments that would be similar to the instruments likely to be employed by the European Central Bank. Thus, the Netherlands Bank has already announced the introduction of a Lombard facility; it is planning to increase the frequency of its open market operations; and it intends to introduce a system of reserve requirements in the near future. Most other EU countries have embarked in a similar exercise of policy instrument harmonization.

1/

In the case of the Netherlands, models of this latter type include the Central Planning Bureau’s FKSEC and the Netherlands Bank’s MORKMONII models.

1/

Indeed, a shock to the overall policy variable would be rather meaningless in the context of a VAR model, as the policy variable itself is postulated to be affected by its own lags and those of all other variables of the system.

2/

Of course, if the results of the VAR were to be applied to assess the impact of a fundamental change in the policy framework, a “regime shift” in Lucas’ terminology, it would also be vulnerable to the Lucas critique, as a VAR model is itself a reduced-form, rather than a structural model. This point is further discussed in connection with the likely impact of EMU.

3/

These difficulties are acknowledged in Sims (1992), where the unrestricted VAR methodology is utilized to distinguish between an IS/LM and a real business cycle explanation of the transmission mechanism in five industrial countries.

1/

See Bernanke (1986), Blanchard and Watson (1984), and Sims (1986).

2/

See, for instance, Dale and Haldane (1993) on the U.K., Escrivá and Haldane (1994) on Spain, and, for a comparative analysis of a number of European countries, Sims (1992), BIS (1995) and Barran, Coudert and Mojon (1996).

3/

This model consisted of money (Ml), a short-term interest rate, a long-term interest rate, real GDP, the GDP deflator, real exports, and the export price deflator, with the short-term interest rate treated as a proxy for the monetary policy instrument. The model employed quarterly data, and imposed a 4-lag structure.

1/

On the other hand, the paper also presents estimates based on a longer time period, 1963-1993.

2/

It should be noted, however, that this latter conclusion is reversed on the basis of the estimation results of the longer 1963-1993 sample.

3/

In fact, work on the FKSEC model has identified a somewhat stronger, and more persistent, impact of short-term interest rate changes on real GDP, relative to that based on the MORKMONII model.

1/

The use of monthly data provides sufficient degrees of freedom to employ a richer lag structure, consisting of 12 lags, while at the same time utilizing a relatively more recent sample, namely 1984-1995.

2/

The guilder’s only central parity change against the deutsche mark was a 2 percent devaluation in early 1983.

3/

On the evolving nature of the Netherlands Bank’s instruments and operating targets, see We11ink (1994).

4/

There is strong evidence that the Netherlands Bank’s official rates, notably the rate on special loans, essentially drives short-term money market rates is provided by Mallekoote and Moonen (1994).

5/

A more detailed description of the variables considered, as well as of the data sources, is provided in the data annex.

6/

The method used to compute the standard errors is explained in the Appendix.

1/

Although the primary focus will be on the impact of a change in the policy instrument, the main patterns associated with changes in other variables will also be briefly discussed.

2/

On the basis of the calculated standard errors the impact on real GDP is insignificantly different from zero throughout the time horizon under consideration. To the extent any pattern is suggested, it is that real GDP may rise in the immediate wake of the interest rate increase, peaking at almost 1 percent above its baseline path 7 months after the shock; thereafter, real GDP essentially returns to its baseline path. The result reported in this paper was quite robust to a number of alternative specifications of the real activity variable.

3/

See, for example, Barro (1977, 1978).

2/

The difference with this latter body of research could only be apparent, however, as-comparisons are hampered by the authors’ non-inclusion of confidence intervals around their estimated impulse response functions. In fact, based on our “point” estimates of the impulse response functions alone, one would have concluded that an increase in long-term interest rates exerts a strong depressing effect on real GDP during the second half of the time horizon under consideration.

1/

Non-exogeneity of the premium with regard to explaining the German rate is in line with the conclusions of Levy (1994). On the other hand, it casts some doubt on the treatment of the German rate as an exogenous variable by Garretsen and Swank (1994).

2/

For most of the variables under consideration, the contribution of the variables’ own lags emerged as very large in the early part of the time period under consideration, but their impact tended to diminish over time.

1/

The criteria on which these choices were made, and other identification restrictions introduced, broadly followed Blanchard and Watson (1984).

1/

In fact, during the largest part of the period under consideration, the Dutch short-term interest rate differential vis-à-vis Germany had been positive.

2/

In the case of the Netherlands, this would presumably imply making greater use of the guilder’s narrow bilateral fluctuation band against the deutsche mark.

1/

See, however, Keating (1990) for examples of rational expectation VAR models which are immune to this problem.

2/

For a particularly clear presentation of this type of “loss of policy instrument” argument resulting from monetary integration and its implications, see, for example, de Grauwe (1994).

1/

This is all the more so since the Dutch monetary authorities have typically refrained from resorting to exchange market intervention to support the peg.

1/

See, for instance, Hamilton (1994) for an explanation of the relationship between the B0 matrix and the Choleski decomposition.

2/

See, for example, Blanchard and Watson (1984), Cooley and LeRoy (1985), Bernanke (1986), and Sims (1986).

1/

The impact of an innovation to the bilateral guilder-DM rate cam be expected to be better captured by the innovation to the long-term interest rate.

2/

These exclusion restrictions are less strong than one might think. It should be recalled that the coefficients of matrix B0 refer to the correlations between the innovations of the variables under consideration, rather than the variables themselves. Thus we impose no restrictions on the contemporaneous relation between the forecastable part of any set of variables.

3/

See Hamilton (1994) and Doan (1992).

Kingdom of the Netherlands - Netherlands: Selected Issues
Author: International Monetary Fund