The Effect of the ERM on Participating Economies

This section looks at the likely effects of changing from a floating exchange rate system to a (quasi) fixed exchange rate system such as the “hard” ERM.9 A simple, open economy macroeconomic model is presented, and the response of the economy to various shocks is computed for reasonable parameter values under both a fixed and a floating exchange rate regime. This approach allows the probable effects of the change in regime on the response of the economy to various shocks to be analyzed. In addition, the effect of the change in regime on the types of shocks faced by the economy is discussed in a rather less formal manner.

Abstract

This section looks at the likely effects of changing from a floating exchange rate system to a (quasi) fixed exchange rate system such as the “hard” ERM.9 A simple, open economy macroeconomic model is presented, and the response of the economy to various shocks is computed for reasonable parameter values under both a fixed and a floating exchange rate regime. This approach allows the probable effects of the change in regime on the response of the economy to various shocks to be analyzed. In addition, the effect of the change in regime on the types of shocks faced by the economy is discussed in a rather less formal manner.

II. A Model of the Effects of Different Exchange Rate Regimes

This section looks at the likely effects of changing from a floating exchange rate system to a (quasi) fixed exchange rate system such as the “hard” ERM.9 A simple, open economy macroeconomic model is presented, and the response of the economy to various shocks is computed for reasonable parameter values under both a fixed and a floating exchange rate regime. This approach allows the probable effects of the change in regime on the response of the economy to various shocks to be analyzed. In addition, the effect of the change in regime on the types of shocks faced by the economy is discussed in a rather less formal manner.

The effect of the change in regime is analyzed using a variation on the exchange rate overshooting model originally devised by Dornbusch (1976). This framework was chosen because it represents a simple, sticky price, open economy model based on the IS/LM framework outlined above, which has been used to analyze a variety of issues.

The model, which is made up of equations (6)—(9) below, represents a small open economy with sticky prices and forward-looking behavior in the exchange rate market. The equations represent the IS curve, money demand, a Phillips curve, and uncovered interest rate parity, respectively;10 where y represents the logarithm of actual output, Y is potential output, e is the exchange rate, ρ is the price level, Et is the expectations operator at time t, Δ is the first-difference operator, foreign variables are denoted by asterisks, and Greek letters represent coefficients.

yt=α(et+Pt*Pt)βit(6)
mt=Pt+ψytδit(7)
ΔPt=Φ(yty)(8)
Etet+1=et+itit*.(9)

The model can be solved for both a fixed and a floating exchange rate system. In the case of a floating exchange rate, it is assumed that a fixed money supply rule is followed. The model was solved using the techniques discussed in Taylor (1986). In the case of the fixed exchange rate regime, the model effectively collapses to equations (6) and (8), and the simulations are easy to compute. In order to make concrete comparisons of the responses, the model was solved for a specific set of parameter values. The following values were chosen: the elasticity of demand with respect to the exchange rate, α= 0.25; the interest rate semi-elasticity of demand, β= 0.5; the elasticity of money demand with respect to output, ψ = 1; the semi-elasticity of money demand with respect to the interest rate, δ = 0.5; and the coefficient on the Phillips curve, Φ=0.5. These coefficients were chosen to represent an approximate midrange of available empirical estimates for an annual model. When alternative parameter values were used it was found that the qualitative conclusions of the model were insensitive to the specific choice of values.

Two types of shocks are analyzed, namely a permanent rise in potential output, and a permanent shift in foreign interest rates.11 The former is a shock to aggregate supply. The latter represents one possible shock to aggregate demand. Clearly, other demand shocks could have been used; however, in most cases such alternative shocks have no impact in one or the other exchange rate regime. For example, shocks to money demand have no effect on output or prices under the fixed exchange rate system, while shocks to the IS curve are ineffective in the floating exchange rate system (at least in this particular model). The implications of this ineffectiveness of different shocks will be discussed further below.

Figure 2 shows the impulse response functions (the path traced out in response to a unit shock in potential output or foreign interest rates) for real output and prices under both a fixed and floating exchange rate regime. The most striking feature of the figure is that a fixed exchange rate regime produces a much more gradual adjustment of both output and domestic prices in response to shocks, reflecting the sticky price assumption. Under a floating exchange rate system the flexibility of the exchange rate allows the economy to respond relatively quickly to shocks; but when the exchange rate is fixed, the lack of exchange rate flexibility leads to a more sluggish response.

Figure 2.
Figure 2.

Results from the Simulation Model

Citation: IMF Staff Papers 1992, 002; 10.5089/9781451947106.024.A005

Two further points are worth noting. While the policy response in a fixed exchange rate regime is uniquely defined, the floating regime covers a wider set of possible responses. The impulse response functions illustrated in Figure 2 assume that monetary policy involves adhering to fixed monetary targets. An alternative, and more general, class of policy functions could involve varying the money supply with deviations of prices from their expected level. Given this diversity in possible responses, it might be expected that the impulse response functions between different countries in a fixed exchange rate regime would be more similar than between countries with floating exchange rates.12 In addition, the fact that different shocks have different effects under the two policy regimes may mean that the types of shocks that affect the economy depend on the regime in place. For example, if aggregate demand shocks were dominated by shifts in money demand, the importance of such shocks might be lower in a fixed exchange rate system than in a floating one.

Although a model of a single economy is useful for looking at how the choice of exchange rate regime affects the response of a single economy, many of the issues involved in the choice of regime depend on the relationship between different economies. For example, it is generally acknowledged that a fixed exchange rate system is most effective when the shocks that hit the system are symmetric across different economies. Hence, countries that experience similar shocks might tend to join together in a fixed exchange rate regime. However, the causation may run the opposite way, with a fixed exchange rate regime actually causing idiosyncratic country shocks to have a larger effect on other members of the exchange rate system than under a floating rate, since the exchange rate “buffer” is no longer available.13 Finally, it has often been alleged that the core country in a fixed exchange rate regime (in this case, Germany) is far less constrained in terms of the conduct of economic policy than other participants.14 As will be seen below, no strong evidence was found that leadership made the effect of the ERM on macroeconomic performance different in Germany than in other members.

To summarize, the ERM could have several effects on the macroeconomic performance of members. By constraining the flexibility of exchange rates, it may elongate the response of members to shocks; by reducing monetary independence, it may make responses to shocks more alike. The ratio between different types of shocks may also depend on the exchange rate regime. The relationship between the ERM and the correlation of the underlying shocks is somewhat ambiguous: on the one hand, countries with similar shocks may be more likely to join a fixed exchange rate regime; on the other hand, the regime itself may cause shocks to become more contemporaneously correlated. Finally, the ERM may cause macroeconomic shocks across members to become more interrelated across time.

III. Results

Quarterly data on real and nominal GDP (or GNP) were collected for the seven major industrial countries, namely the United States, Japan, Canada, the United Kingdom, west Germany, France, and Italy.15 The first three countries are non-European economies with floating exchange rates; the last three are all members of the ERM. The United Kingdom represents an intermediate case, being within Europe, but not a member of the ERM over the available data period. Comparisons of the behavior of the ERM and non-ERM economies both before and after its inception allow the effects of the ERM on macroeconomic performance to be quantified.

The data were divided into two periods, corresponding to the period of floating exchange rates prior to the introduction of the ERM (1971:2–1979:2, hereafter the 1970s), and the period since the hardening of the ERM in the early 1980s (1982:1–1990:1, hereafter the 1980s). The ERM was introduced in mid–1979, hence the end point of the first period, whose starting point approximates to the breakup of the Bretton Woods fixed exchange rate regime. The second period corresponds to the time since the strengthening of the ERM, which is usually placed in 1982 or 1983 (an early date being chosen so as to maximize the length of the time series), and goes up to the end of the available data. The two periods are of equal length, which aids comparison.16

Vector autoregressions of the change in the logarithm of real output and the output deflator, plus a constant, were estimated for each country for both time periods.17 The results were transformed into demand and supply disturbances using the decomposition described above.18

The cumulative impulse response functions from this procedure are shown in Figures 3 and 4. These illustrate the effect of a unit shock in demand and supply on the level of output and prices. The left-hand side represents the results for the 1970s, the right-hand side, the results for the 1980s; they have been graphed on the same scale so that the relative importance of shocks in the two periods can be inferred. The results are shown for two sets of countries—the ERM members (Germany, France, and Italy), and the rest (the United States, Japan, Canada, and the United Kingdom). The first panel shows the effect of demand and supply shocks on output, the second, the effects on prices.

The output responses reflect the restrictions imposed in the estimation. Demand shocks result in a temporary rise in output, which then comes back to zero (with some cycling in the case of Japan and Italy), while supply shocks generally result in a gradual rise in output. The price responses to demand and supply shocks are in almost total accord with the predictions of the model, with positive demand shocks causing prices to rise, and supply shocks resulting in reductions in the price level. The only deviation from the expected results is that in two cases in the 1970s (Germany and Italy), the long-run response to a supply shock is incorrectly signed or very small. Given that no constraints were imposed on the price responses, these results represent some confirmation that the decomposition being used is actually differentiating between aggregate supply and demand shocks. Furthermore, since the results hold for both the 1970s and for the period starting in 1982, they cannot simply be ascribed to the effect of the oil shocks in the 1970s.

The impulse response functions are almost universally higher in the 1970s than the 1980s, indicating that shocks played a larger role in determining the levels of output and prices in the earlier period. 19 In addition, some of the predictions from the theoretical section appear to be confirmed. Comparing the impulse response functions of the ERM in the 1980s with the other response functions, they do appear to be more elongated and more correlated. These issues are explored in a more formal manner below.

Figure 3.
Figure 3.

Impulse Response Functions for Output

Citation: IMF Staff Papers 1992, 002; 10.5089/9781451947106.024.A005

Figure 4.
Figure 4.

Impulse Response Functions for Prices

Citation: IMF Staff Papers 1992, 002; 10.5089/9781451947106.024.A005

Table 1a.

Correlation of Growth and Inflation, 1970s

(Correlation coefficients)

article image
Note: In this and succeeding tables, US = United States, CA = Canada, JP = Japan, UK = United Kingdom, DE = Germany, FR = France, and IT = Italy.
Table 1b.

Correlation of Growth and Inflation, 1980s

(Correlation coefficients)

article image
Note: Outlined areas represent interaction of ERM members.

IV. Effects of ERM Membership on the Macroeconomy

The ERM may cause the performance of participating economies to change in a number of ways. In this section, the possible effects discussed in Section II are investigated: elongation of the responses to shocks; convergence of the impulse response functions; the relationship between the ERM and the underlying shocks; the effect of the ERM on the interconnection of shocks across countries; and the effect on the mix of the different types of shocks.20 However, first, it is useful to consider the raw data.

Behavior of inflation and growth. Tables 1a and 1b show the correlations of the two variables, growth and inflation, for the full set of countries over the 1970s and 1980s.21 The most striking feature of the data is the high correlation of inflation performance among members of the ERM in the 1980s. The correlations are stronger than those among the same countries in the 1970s, and higher than the correlations among non-ERM members in the 1980s. When the significance of these differences is formally tested,22 all the correlations among ERM members are significantly higher in the 1980s than in the 1970s at conventional significance levels, while the median value of the three ERM correlations is significantly higher than all but one of the other cross correlations reported in the table. There appears to be strong evidence that the ERM has caused inflation performance to become more correlated.23

As far as real growth is concerned, the situation is less clear cut. Real growth is more highly correlated among ERM countries than non-ERM countries; however this is as much a feature of the pre-ERM period (the 1970s) as of the 1980s; the 1980s do not show any strengthening of this bond associated with the introduction of the ERM, although there are some weak indications that growth rates for other economies have become less correlated over time.24

Elongation of the response functions. Table 2 shows the ratio of the impulse response four quarters after the shock to the long-run level, which is used as a simple measure of the speed with which the economy responds to shocks for the three responses where the long-run value is not zero (recall that the long-run response of output to a demand shock is zero by construction).25

Table 2.

Elongation of Response

article image
Note: The figures represent the ratio of the response four quarters after the shock to the long-run response. Outlined values represent ERM countries in the I980s.

Comparing the ratios for the 1970s with those for the 1980s, seven of the nine ratios associated with ERM members fell between the 1970s and the 1980s. This contrasts with the results for the non-ERM countries where six of the twelve ratios rose and six fell. These results provide some evidence that ERM member responses have become more elongated. A formal nonparametric test can be constructed by assuming that in the absence of a change in behavior, the ratio is equally likely to rise as to fall, as indicated by the non-ERM data. Using the binomial distribution, the null hypothesis of no fall in the ratio can be rejected at the 10 percent significance level, but not at the 5 percent level, for the ERM members.

Comparing the ratios for ERM and non-ERM members in the 1970s and 1980s yields further evidence of a change in behavior. The data for the 1980s indicate that the values for this ratio for the three ERM members (Germany, France, and Italy) were relatively low, as would be expected if their responses were more elongated. Indeed, in two cases the ERM members had the lowest three ratios among the seven reported, while in the third case they had three of the four lowest ratios; the data for the 1970s. however, showed no such pattern for these economies. A simple test of the significance of the results for the 1980s is to calculate the probability of the observed ranking occurring randomly. The probability of the ERM countries being the three lowest ranked countries on a random basis is less than 5 percent, while the probability of all three being represented in the lowest four rankings is less than 15 percent. This is further evidence that membership in the ERM has indeed elongated the responses of participating economies to shocks.26

Correlation of the impulse response functions. Another possible effect of the ERM is that, by constraining monetary policy, it could make the response of the participating economies to shocks more similar.27 The visual impression from Figure 3, that the responses are indeed similar for the ERM countries in the 1980s, is confirmed by formal tests. Tables 3a and 3b show the correlation matrices for the impulse response functions of output and prices with respect to both demand and supply shocks for the 1970s and the 1980s. The most striking feature is the very high level of correlation for the ERM countries in the 1980s, compared with both their own past history and nonmembers. The minimum correlation for ERM members in the 1980s across both response functions is 0.89, which is higher than all of the six correlations between the same countries in the 1970s; formal tests indicate that every intra-ERM correlation rose significantly between the two periods. Turning to the comparison of ERM with non-ERM countries in the 1980s, only one of the twelve correlations between the non-ERM members is larger than the minimum correlation for the impulse between the ERM members. In the vast majority of cases, the ERM correlation coefficients are significantly different from those for non-ERM members. The ERM does indeed appear to have produced a convergence in responses.

Table 3a.

Impulse Response Function Correlations, 1970s

(Correlation coefficients)

article image
Table 3b.

Impulse Response Function Correlations, 1980s

(Correlation coefficients)

article image
Note: Outlined areas represent interaction of ERM members.

Correlation of the shocks. The ERM could have two effects on the shocks hitting members. Most authors agree that a fixed exchange rate regime is most effective if the shocks are symmetric, rather than idiosyncratic. This implies, on the one hand, that countries that join a mechanism such as the ERM will be those subject to symmetric shocks.28 On the other hand, by increasing the interrelationship between members, the ERM could increase the contemporaneous correlation across countries.

Variance covariance matrices for the demand and supply shocks derived from the model are shown in Tables 4a and 4b. The continental European countries appear to have somewhat higher correlations among themselves than other countries; however, this tendency is as strong in the (pre-ERM) 1970s as in the 1980s. Indeed, the most obvious example of significantly higher correlations for this subset of countries concerns supply shocks in the 1970s. The correlation between the United Kingdom and the continental economies is also consistently high in this case. The picture for the 1980s is less clear-cut. However, all the continental European economies have significant positive correlations between each other, while the coefficients on the other cross correlations are positive and negative in approximately equal quantities. Demand shocks show a less obvious pattern. Of the six correlations between the continental European economies (three in the 1970s and three in the 1980s), five are significantly positive, and one is insignificantly different from zero. Of the other 36 coefficients in the table, 16 are significantly positive, 17 are insignificant, and 3 are significantly negative.

Table 4a.

Correlation of Demand and Supply Shocks, 1970s

(Correlation coefficients)

article image
Table 4b.

Correlation of Demand and Supply Shocks, 1980s

(Correlation coefficients)

article image
Note: Outlined areas represent interaction of ERM members.

An alternative way of summarizing the data is to use principal component analysis. Table 5 shows the percentage of the variance of demand and supply shocks explained by the first principal component of the data—that is, the orthogonal component most correlated with the underlying series. This is a measure of the degree to which the shocks are symmetric across groups of countries. The results are shown for each shock over both the 1970s and the 1980s. The rows indicate the four groups of countries being analyzed: the ERM members (Germany, France, and Italy); the United States, Japan, and Canada; the ERM plus the United Kingdom; and the four non-ERM members (the United States, Japan, the United Kingdom, and Canada). In analyzing the results, it is important to remember that the groups with three members will tend to show higher correlations than those with four members; hence, the results from the first two rows are not directly comparable with those in rows three and four.

Table 5.

Percentage of Variance Explained by First Principal Component

article image
Note: The ERM represents Germany, France, and Italy; the non-ERM group is represented by the United States, Canada, Japan, and the United Kingdom. Outlined areas show the results for ERM members in the 1980s.

As with the correlation coefficients, the principal components data indicate that supply shocks are more correlated within the ERM members, both before and after ERM’s inception. while there is relatively little difference in the correlation of demand shocks. In the case of supply shocks, the principal component for the ERM members explains about 10 percent more of the variance than for the United States, Japan, and Canada in the 1980s, with even larger differences in the 1970s. When the United Kingdom is added to both groups, the results indicate that it is an intermediate case; supply shocks in the United Kingdom are more correlated with the rest of the ERM than other countries, but not as correlated as the core members of the mechanism.

This evidence indicates that the ERM has produced no increase in the correlation of the shocks hitting members. Rather, it has attracted countries whose supply shocks are relatively similar.

Interconnection between shocks. While the ERM may not have made the shocks hitting its members more contemporaneously correlated, it could, by binding participants more closely together, make shocks more interrelated over time. In particular, the German leadership hypothesis would imply that shocks to the German economy would tend to affect other members of the ERM. To investigate this possibility, Granger causality tests, which measure the importance of lagged values of one variable for the outcome of another, were carried out using both the demand and supply shocks derived from the estimation.29 Table 6 shows the results using demand shocks; the results for supply shocks, which were similar, are not reported for the sake of brevity. For entries above the diagonal, the direction of causation runs from the country defined by the column to the country defined by the row; for those below the diagonal the direction of causation is reversed. An asterisk indicates that the F-test for Granger causality cannot be rejected at the 5 percent significance level, while no asterisk indicates no evidence for causality. As can be seen from the table, the data indicate very little interaction between shocks. In particular, there is no discernible difference between ERM and non-ERM countries or within ERM countries across time, and hence, no evidence that the ERM has had any effect on the temporal correlation between shocks.30

Table 6.

Results of Granger Causality Tests on Demand Shocks

article image
Note: An asterisk implies Granger causality cannot be rejected at the 5 percent significance level. Outlined area represents the ERM in the 1980s.
Table 7.

Percentage of Unconditional Variance Explained by Demand Shocks in the Long Run

article image
Note: Outlined area represents ERM members in the 1980s.

Relative importance of different types of shocks. Table 7 shows the percentage of the unconditional variance of growth and inflation attributable to demand and supply shocks in the short run. No particular pattern emerges between the performance of ERM members in the 1980s and the other results, indicating that there is no evidence that the switch in regime has led to a change in the types of macroeconomic shocks being experienced across countries.31

V. Conclusions

This paper has looked at the effect of the ERM on the macroeconomic performance of its members. In order to do this, a procedure for identifying aggregate demand and supply shocks was proposed and executed. Data on the seven largest industrial countries were used, both for the period between the breakup of the Bretton Woods fixed exchange rate system and the formation of the ERM in 1979, and for the period since 1982, when the ERM is generally felt to have been most effective, in particular in reducing the inflation rates of other members down to German levels.

The results indicate that while the ERM has had little effect on the nature of the shocks hitting the economies, it has had a significant effect on the response of member countries to these shocks, making them both more elongated and more similar. Longtime members of the ERM, and to some extent the United Kingdom, have relatively correlated supply shocks compared to the other economies studied; however, this is as true of the pre-ERM period as it is of the 1980s. Similarly, the interrelation between shocks and the ratio between demand and supply shocks does not appear to be affected by ERM membership. The evidence from this paper indicates that members of the ERM have had, and continue to have, relatively symmetric supply shocks; this similarity of shocks may be one reason for the desire to move to more fixed exchange rates across members; however, the ERM itself has not affected the nature of the underlying shocks.

Turning to the responses of ERM members to shocks, members of the ERM in the 1980s appear to have had both more elongated and more correlated responses to shocks, compared either with their own past behavior or the responses of non-ERM countries in the 1980s. It appears that the ERM, by taking away the flexibility afforded by floating exchange rates, has both lowered the speed at which members respond to shocks while at the same time making these responses more coordinated across members.

These results, that ERM members have relatively symmetric underlying macroeconomic shocks and that membership has produced more elongated and correlated responses to these shocks, conform to the thesis that the ERM represents a move by countries with relatively similar structures to coordinate macroeconomic policy by limiting monetary independence, at the cost of lower flexibility in the face of shocks. They also indicate that the core ERM members have at least some of the characteristics desirable for a common currency area.

REFERENCES

  • Artis, Michael J., and S. Gazioglu, “Modelling Asymmetric Exchange Rate Unions: A Stylized Model of the EMS,” Greek Economic Review, Vol. 11 (June 1989), pp. 177202.

    • Search Google Scholar
    • Export Citation
  • Artis, Michael J., and Dilip Nachane, “Wages and Prices in Europe: A Test of the German Leadership Thesis,” CEPR Discussion Paper No. 296 (London: Centre for Economic Policy Research, March 1989).

    • Search Google Scholar
    • Export Citation
  • Artis, Michael J., and Mark P. Taylor, “The Achievements of the European Monetary System,” Economic and Social Review, Vol. 20 (January 1989), pp. 12145.

    • Search Google Scholar
    • Export Citation
  • Barrell, Ray, “European Currency Union and the EMS,” National Institute Economic Review, No. 132 (May 1990), pp. 5966.

  • Bayoumi, Tamim, and Barry Eichengreen, “Shocking Aspects of European Monetary Unification,” in The Transition to Economic and Monetary Union in Europe, ed. by F. Giavazzi and F. Torres (Cambridge: Cambridge University Press, forthcoming 1992).

    • Search Google Scholar
    • Export Citation
  • Blanchard, Olivier Jean, and Danny Quah, “The Dynamic Effects of Aggregate Demand and Supply Disturbances,” American Economic Review, Vol. 79 (September 1989), pp. 65573.

    • Search Google Scholar
    • Export Citation
  • Canzoneri, Matthew B., “Exchange Intervention Policy in a Multiple Country World,” Journal of International Economics, Vol. 13 (November 1982), pp. 26789.

    • Search Google Scholar
    • Export Citation
  • Cohen, Daniel, and Charles Wyplosz, “The European Monetary Union: An Agnostic Evaluation,” in Macroeconomic Policies in an Interdependent World, ed. by Ralph C. Bryant, David A. Currie, Jacob A. Frenkel, Paul R. Masson, and Richard Portes (Washington: The Brookings Institution, Centre for Economic Policy Research, and International Monetary Fund, 1989).

    • Search Google Scholar
    • Export Citation
  • Cohen, Daniel, and Charles Wyplosz, “France and Germany in the EMS: the Exchange Rate Constraint,” in The External Constraint on Macroeconomic Policy: The European Experience, ed. by George Alogoskoufis, Lucas Papademos, and Richard Portes (Cambridge: New York: Cambridge University Press, 1991).

    • Search Google Scholar
    • Export Citation
  • de Grauwe, Paul, “The Cost of Disinflation and the European Monetary System,” Open Economies Review, Vol. 1, No. 2 (1990), pp. 14773.

    • Search Google Scholar
    • Export Citation
  • Dornbusch, Rudiger, “Expectations and Exchange Rate Dynamics,” Journal of Political Economy, Vol. 84 (December 1976), pp. 116176.

    • Search Google Scholar
    • Export Citation
  • Dornbusch, Rudiger, and Stanley Fischer, Macroeconomics, 3rd ed. (New York: McGraw Hill, 1986).

  • Fratianni, Michele, and Jürgen von Hagen, “The European Monetary System Ten Years After,” Carnegie–Rochester Conference Series on Public Policy, Vol. 32 (Spring 1990), pp. 173242.

    • Search Google Scholar
    • Export Citation
  • Giavazzi, Francesco, “The Exchange Rate Question in Europe,” in Macroeconomic Policies in an Interdependent World, ed. by Ralph C. Bryant, David A. Currie, Jacob A. Frenkel, Paul R. Masson, and Richard Portes (Washington: The Brookings Institution, Centre for Economic Policy Research, and International Monetary Fund, 1989).

    • Search Google Scholar
    • Export Citation
  • Giavazzi, Francesco, and Luigi Spaventa, “The ‘New’ EMS,” CEPR Discussion Paper No. 369 (London: Centre for Economic Policy Research, January 1990).

    • Search Google Scholar
    • Export Citation
  • Giovannini, Alberto, “How Do Fixed Exchange Rate Regimes Work: The Evidence from the Gold Standard, Bretton Woods and the EMS,” CEPR Discussion Paper No. 282 (London: Centre for Economic Policy Research, October 1988).

    • Search Google Scholar
    • Export Citation
  • Haldane, A. G., “The Exchange Rate Mechanism of the European Monetary System: A Review of the Literature,” Bank of England, Quarterly Bulletin, Vol. 31 (February 1991), pp. 7382.

    • Search Google Scholar
    • Export Citation
  • Hall, Robert E., and John B. Taylor, Macroeconomics: Theory, Performance, and Policy (New York: Norton, 1988).

  • Italianer, A., and Jean Pisani-Ferry, “Exchange Rate Regimes in the EC,” European Economy, No. 44 (1990), pp. 30334.

  • Kendal, M., and A. Stuart, The Advanced Theory of Statistics, Vol. 2 (New York: Hafner Publishing Company, 1967).

  • Mundell, Robert A., “A Theory of Optimum Currency Areas,” American Economic Review, Vol. 51 (September 1961), pp. 65765.

  • Portes, Richard, “Macroeconomic Policy Coordination and the European Monetary System,” CEPR Discussion Paper No. 342 (London: Centre for Economic Policy Research, September 1989).

    • Search Google Scholar
    • Export Citation
  • Roubini, Nouriel, “Leadership and Cooperation in the European Monetary System: A Simulation Approach,” NBER Working Paper No. 3044 (Cambridge, Massachusetts: National Bureau of Economic Research, July 1989).

    • Search Google Scholar
    • Export Citation
  • Sims, C., “Macroeconomics and Reality,” Econometrica, Vol. 48 (1980), pp. 149.

  • Sterne, Gabriel, and Tamim Bayoumi, “Temporary Cycles or Flexible Trends: Evidence from 21 OECD Countries” (unpublished; London: Bank of England, 1991).

    • Search Google Scholar
    • Export Citation
  • Taylor, John B., “New Econometric Approaches to Stabilization Policy in Stochastic Models of Macroeconomic Fluctuations,” in Handbook of Econometrics, Vol. 3, ed. by Zvi Griliches and Michael D. Intriligator (Amsterdam: North Holland, 1986).

    • Search Google Scholar
    • Export Citation
  • Ungerer, Horst, Owen Evans, and Peter Nyberg, The European Monetary System: The Experience 1979–1982, IMF Occasional Paper No. 19 (Washington: International Monetary Fund, 1983).

    • Search Google Scholar
    • Export Citation
  • Ungerer, Horst, Owen Evans, Thomas Mayer, and Phillip Young, The European Monetary System: Recent Developments, IMF Occasional Paper No. 48 (Washington: International Monetary Fund, 1986).

    • Search Google Scholar
    • Export Citation
  • van der Ploeg, Frederick (1989a), “Fiscal Aspects of Monetary Integration in Europe,” CEPR Discussion Paper No. 340 (London: Centre for Economic Policy Research, August).

    • Search Google Scholar
    • Export Citation
  • van der Ploeg, Frederick (1989b), “Monetary Interdependence Under Alternative Exchange Rate Regimes: A European Perspective,” CEPR Discussion Paper No. 358 (London: Centre for Economic Policy Research, November).

    • Search Google Scholar
    • Export Citation
  • Weber, Axel A., “EMU and Asymmetries and Adjustment Problems in the EMS: Some Empirical Evidence,” CEPR Discussion Paper No. 448 (London: Centre for Economic Policy Research, August 1990).

    • Search Google Scholar
    • Export Citation
9

The results in this section pertain to a completely fixed exchange rate system, rather than the bands that have operated in the ERM. It seems probable, however, that any characteristics of a completely fixed exchange rate system will also tend to he true for members of the ERM, at least since the hardening of the system in the early 1980s. Earlier, the ERM had the characteristics of a “crawling peg” exchange rate regime.

10

The nominal, as opposed to real, interest rate in the IS curve is used for analytic convenience. A real interest rate specification produces the same qualitative results.

11

Permanent shocks are analyzed in order to make the results comparable to the impulse response functions calculated in the empirical work.

12

Indeed, one of the main arguments in favor of the ERM is that, by reducing monetary independence, it enforces a more coordinated policy response across members (see Fratianni and von Hagen (1990)).

13

See Canzoneri (1982), van der Ploeg (1989a, 1989b), and Cohen and Wyplosz (1991) for theoretical discussions of these points.

14

For example, Giovannini (1988). There is also a large empirical literature on the German leadership hypothesis. For an empirical investigation of this issue, plus more general data on the effects of the ERM, see Weber (1990).

15

The data come from Organization for Economic Cooperation and Development (OECD) quarterly national accounts. The data are seasonally adjusted except for Germany and Japan, which were seasonally adjusted by the author.

16

Limited experimentation with different time periods indicates that the exclusion of 1982 has little effect on the results. Without the 1971 data, however, the 1970s period becomes dominated by the 1974 oil shock, and the responses to demand shocks become small and unstable.

17

Likelihood ratios tests were constructed to test for the optimal lag length. All of the values came to between 3 and 6; in the interests of standardizing and to conserve degrees of freedom, all lags were set to 4.

18

The decomposition matrix provides a useful method for assigning current data into demand and supply shocks; simply work Out the implied residuals from the VAR and transform the residuals using the matrix C. This is explained in Sterne and Bayoumi (1991).

19

A comparison of the variance of real growth and inflation over the two periods also shows this effect.

20

Other empirical studies of the effects of the ERM on behavior include Weber (1990) and Cohen and Wyplosz (1989), who looked at effects of the ERM using the raw data without distinguishing between shocks and responses and concluded that the ERM is relatively closely economically integrated. Rouhini (1989), Artis and Gazioglu (1989), Italianer and Pisani-Ferry (1990), and Barrell (1990) used large econometric models to look at the impact of various types of shocks under different exchange rate regimes. The overall conclusion from these studies is that the ERM makes countries more economically interdependent.

21

Throughout this paper real growth is used as the measure of the variability of real activity, in preference to some measure of the deviation of the level of output from trend, because the theory implies that many of the macroeconomic shocks on the economy involve permanent changes in the level of output, making the potential level of output difficult to measure.

22

The value ½ In((1+r)/(1r)) is distributed approximately as normal with expected value ½ In((1+ρ)/(1ρ)) and variance 1 /(T – 3) (Kendall and Stuart (1967, pp. 292–93)).

23

Similar results are shown in Fratianni and von Hagen (1990). There is some controversy as to whether membership in the ERM actually helped the European disinflation in the 1980s, or whether it was part of a wider experience of all industrial countries. See Artis and Nachane (1989), Fratianni and von Hagen (1990), de Grauwe (1990), and Portes (1989).

24

Interestingly, Granger causality tests on growth and inflation do not indicate any particularly close interaction between the current and past performance of different ERM members in either period.

25

While the choice of four quarters as the intermediate point in the measurement is somewhat arbitrary, alternative choices yield the same results.

26

de Grauwe (1990) also found evidence of prolongation of responses.

27

See van der Ploeg (1989a, 1989b) and Cohen and Wyplosz (1990) for analyses of the effect of the ERM on policy behavior. One of the arguments for the ERM is that it enforces policy coordination, which might otherwise be difficult to achieve; see Portes (1989).

28

The general issue of the effect of fixed exchange rate regimes on shocks is discussed in Canzeroni (1982) and van tier Ploeg (1989a, 1989b). The specific issue of symmetric versus asymmetric shocks is discussed in Giavazzi (1989), Cohen and Wyplosz (1989), and Bayoumi and Eichengreen (1992).

29

These used a VAR comprising four lags of each pair of variables. Results using two lags produced similar results.

30

In particular, whatever the role of Germany in terms of policymaking, there is no evidence for German leadership in macroeconomic shocks.

31

Data for the long-run decomposition of variance between the shocks show a similar lack of pattern.