Relative Prices and Economic Adjustment in the United States and the European Union: A Real Story About EMU
Author:
Mr. Tamim Bayoumi
Search for other papers by Mr. Tamim Bayoumi in
Current site
Google Scholar
Close
and
Mr. Alun H. Thomas
Search for other papers by Mr. Alun H. Thomas in
Current site
Google Scholar
Close

Structural vector autoregressions are used to analyze the relationship between real output and relative prices within the European Union and the United States. Relative price variability appears to be more important for adjustment within the European Union than in the United States, reflecting the lower integration of goods and factor markets. In the absence of higher market integration, the lower relative price variability implied by the introduction of a single currency in the European Union could well cause significant economic disruption.

Abstract

Structural vector autoregressions are used to analyze the relationship between real output and relative prices within the European Union and the United States. Relative price variability appears to be more important for adjustment within the European Union than in the United States, reflecting the lower integration of goods and factor markets. In the absence of higher market integration, the lower relative price variability implied by the introduction of a single currency in the European Union could well cause significant economic disruption.

The prospect of European Economic and Monetary Union (EMU) has created interest in a host of issues associated with the operation of currency unions.1 The most basic implication of adopting a common currency is that the participating countries are no longer able to vary their bilateral exchange rates. It is a widely held view that the loss of the exchange rate instrument will reduce the ability of economies to absorb disturbances. For example, the 1992 Economic Report of the President (of the United States) states: “A single currency would prevent exchange rate adjustments among European countries from absorbing external economic shocks or differences in domestic economic policies…” (p. 225).

Much of the academic analysis on EMU also assumes that the nominal exchange rate acts as a buffer in reducing disturbances; indeed, this is the basis of most of the theory of optimum currency areas. For example, work measuring the asymmetry of economic shocks in the European Union2 assumes that a currency union is less efficient at absorbing asymmetric disturbances across different countries than a regime involving individual currencies, as does much of the literature on the fiscal implications of EMU (surveyed in Bean (1992)). Krugman (1991) also assumes that the exchange rate adjusts to moderate shocks when he argues that there will be a greater need for national exchange rates in a future, more integrated Europe because of the increased likelihood of asymmetric output imbalances due to regional specialization. In a slightly different vein, macroeconomic simulations of the impact of EMU (surveyed in Masson and Symansky (1993)) compare a regime in which each individual country has a fixed monetary target with a situation in which there is a single European currency and hence a single monetary target. In practice, this amounts to assuming that the nominal exchange rate operates as a buffer under the flexible exchange rate regime.

This paper looks at the empirical relationship between fluctuations in relative prices and real output using structural vector autoregressions, focusing on behavior across European Union (EU) countries and across regions of the United States.3 Such a comparison of the behavior of EU countries, which have close economic ties but separate currencies, with regions within the United States, a currency union of roughly comparable economic magnitude, can be expected to shed light on how the existence of a currency union influences the response of the economy to underlying disturbances.

I. Theory

Movements in relative prices are one of the ways that economies buffer themselves against shocks to output. Consider the standard demand and supply model for a single industry shown in the top panel of Figure 1. A shock that moves the demand curve outward leads to a smaller rise in output than the overall shift in the curve because the relative price of output rises, reducing the full impact of the demand shock. In a similar manner, a positive supply shock leads to a fall in relative output prices, reducing the underlying output disturbance.

Figure 1.
Figure 1.

The Model

Citation: IMF Staff Papers 1995, 001; 10.5089/9781451957068.024.A005

The same diagram can be used to analyze regional behavior. However, there is an important difference between the analysis of an industry and the analysis of a region. For a single industry, it is reasonable to assume that changes in the incomes of workers and owners in the industry have negligible effects on the demand for the products from that industry, because they form such a small part of overall demand. In a region, on the other hand, regional incomes are an important determinant of the demand for regional goods (due, at least in part, to the existence of nontraded goods). This requires a more complex model than standard demand and supply analysis.

Appendix I contains a simple macroeconomic model of regional supply and demand. In the short run it is assumed that any changes in regional income accrue to owners of capital, whose consumption demand is highly diversified. Hence, the standard demand and supply analysis is adequate. Over time, however, much of the change in regional income is reflected in changes in labor income. This causes an endogenous change in the demand for home goods since workers are assumed to consume only these goods. Initial increases in regional output are therefore associated with a subsequent endogenous rise in demand for home goods, and falls in output, with a fall in demand for home goods.

Clearly, this is a very simple model. Not all short-run changes in income accrue to owners of capital, nor do workers spend all their money on local goods. At the same time, it does capture some stylized facts about the real world. Profits are indeed the most cyclically sensitive component of income, and their impact on the demand for local products is probably small.4 More generally, the high short-term elasticities on activity in most estimated import equations also imply that external sources are an important provider of goods in response to short-term changes in income.

The bottom two panels of Figure 1 show the impact of these considerations on the original demand and supply analysis. The left-hand panel shows the response to a rise in the demand for regional products, such as an increase in the demand for automobiles in the United States (which would raise the demand for products from the Great Lakes region), or a fiscal expansion in an EU country. The initial rise in demand from D to D′ causes an increase in both regional output and the relative price of regional goods. The rise in local wage income that follows this increase in regional output causes a further expansion in the demand for regional goods, which moves the economy toward its long-run equilibrium at E”. Hence, the induced increase in demand moves both regional real output and the relative price of regional goods further away from their initial values.

The right-hand panel shows the effect of a positive supply shock, such as a technological breakthrough in the computer industry or a change in labor market regulations. The supply curve shifts rightward from S to S′, which results in a rise in real output and a fall in the relative price of regional goods, as shown by point E′. This increase in real output again implies an expansion in local incomes (measured in terms of local goods) and a subsequent expansion in the demand for regional goods. The induced shift in the demand curve (from D to D′) causes a further expansion in regional output. However, unlike the earlier example, where the relative price of regional goods continued to diverge from their original values, in this case the induced increase in demand causes the relative price of regional goods to move back toward its original value, as shown in the new long-term equilibrium E”.

While these dynamics complicate the adjustment path, they do not detract from the role of relative prices in reducing output fluctuations, particularly in the short run. The importance of relative prices in this process depends upon the integration of regional goods markets and factor markets. If regional goods and factor markets are highly integrated, so that the demand and supply curves are relatively flat, relative price changes in response to disturbances will be relatively small. Hence, relative prices could be expected to be a less important part of adjustment in the United States, with its highly integrated markets, than in the European Union, with its less integrated goods and factor markets.

If regions have the same currency, then relative price movements have to occur through inflation differentials. If they have different currencies, however, these relative price movements can also come about through changes in the exchange rate. Since exchange rates appear to be more flexible than domestic prices, this implies that relative price changes may be more difficult to achieve in a currency union.5 Such slow adjustment can cause significant disruption to the local economy. Good examples would be the boom and bust in employment experienced by the Southwest region of the United States in the 1980s due to the rise and fall in oil prices; the difficulties experienced by (then) members of the Exchange Rate Mechanism (ERM) in 1992 and 1993 in the wake of German unification; and regional problems within European countries, such as the persistent differences in unemployment rates between the northern and southern regions of the United Kingdom and Italy. The remainder of this paper estimates the role of relative price movements (or, in other words, movements in real exchange rates) in the process of adjustment between members of the European Union, and compares it to the role of relative price movements in the United States, a smoothly functioning currency union.

II. Estimation Methodology

The model shown in Figure 1 is estimated using a structural vector autoregression (VAR) of the type proposed by Blanchard and Quah (1989), in which underlying disturbances are identified through their long-run responses to endogenous variables.6 Specifically, a bivariate vector autoregression involving real output and the relative output prices is estimated, and the underlying supply disturbances are identified by assuming that they have only a temporary effect on relative output prices, while demand disturbances are allowed to have a permanent effect. Hence, the estimation assumes that the long-term equilibrium for a supply disturbance (point E” in the bottom right-hand panel of Figure 1) involves the same relative price as the initial equilibrium. Details of the estimation procedure are given in Appendix II.

Clearly, identifying disturbances with only a temporary impact on the relative prices as supply disturbances and those with a permanent impact on relative prices as demand disturbances is a strong assumption. The underlying supply and demand framework only implies a tendency for relative prices to return to their initial level in response to a supply disturbance. However, as long as the long-term effect of a supply disturbance on relative prices is small in relation to the effect of a demand disturbance, the procedure will approximate the correct model.7 There are at least two reasons for believing that this is indeed the case. First, as discussed earlier, the induced changes in demand caused by movements in local incomes will tend to exacerbate long-run relative price movements caused by demand disturbances while reducing those due to supply disturbances. Second, it seems reasonable to assume that goods markets are more highly integrated than factor markets. This implies that the underlying supply curve will be steeper than the demand curve, which in turn implies larger relative price movements in response to demand disturbances than to supply disturbances.8

The advantage of using the structural vector autoregression approach is that it allows more structure to be put in the empirical analysis than would be possible from a more atheoretical examination of the data. In particular, as will be discussed in more detail below, it implies that most of the differences in macroeconomic adjustment between the United States and the European Union can be attributed to relatively intuitive differences in the slopes of the underlying curves and sizes of the underlying disturbances.

As a further check on the results, two “over-identifying” restrictions, which are implied by the underlying framework but are not imposed in the estimation, can also be used to test the plausibility of the results. The first is the sign of the relative price movements in response to each shock. The model implies that positive demand shocks should increase real output and raise the relative price of local output, while supply shocks that raise real output should lower the relative price of output. Second, as can be seen in Figure 1, the underlying supply curve can be identified both by the response to a demand disturbance and by the longer-term part of the response to a supply disturbance. These restrictions, which can be seen as additional tests of the reasonableness of the results, are broadly confirmed by the estimation.

III. Data

Annual data on real and nominal output in dollars were collected for 11 EU countries and 8 U.S. regions.9 For the European Union, the data came from the OECD Annual National Accounts. Real dollar GDP was calculated using 1985 prices and exchange rates, while nominal dollar GDP used current prices and exchange rates. The EU sample period is 1961-89.10 Regional real and nominal gross state product, the regional equivalent of GDP, were collected for the eight standard regions of the United States, as defined by the Bureau of Economic Analysis, from 1963-89.11

Relative prices were calculated by taking the implicit output deflators for each EU country or U.S. region and dividing them by the corresponding deflator for the remainder of the European Union or United States (that is, the EU or U.S. aggregate less the country or region). Hence, they measure relative output prices within the area, the proper concept for a demand and supply analysis. Output growth was also measured relative to behavior in the rest of the European Union or United States. Therefore, the data represent intra-area movements in relative prices and output across U.S. regions and EU countries.12 It is important to exclude the effects of the rest of the world from the analysis because, while U.S. regions have a fixed rate of exchange against each other, the United States as a whole has a variable exchange rate against the rest of the world. Similarly, EMU will fix only intra-EU rates of exchange, and not rates of exchange with the rest of the world.

Before considering the estimation results, it is useful to consider the characteristics of the raw data. Table 1 reports the standard deviation of the growth of real output and the change in the relative price of output (measured as the change in the logarithm of the underlying variables).13 To give an idea of the impact of adjusting by the regional aggregate, the standard deviations for the unadjusted data are shown in parentheses.

Table 1.

Growth of Output and Changes in the Relative Price of Output within the European Union (1973-89) and the United States (1966-89)

article image

For the European Union, the standard deviation of relative output growth varies considerably, from 0.010 to 0.027 (about 1.0 to 2.7 percent per annum since the data are in logarithms). The original members of the European Union (Belgium, France, Germany, Italy, and the Netherlands) all have lower variation in their output growth rates than those that joined later, with France having a particularly low value. The unadjusted data, shown in parentheses, generally show more variation than the adjusted data, indicating a common cycle. The results from the regional data for the United States are shown in the bottom part of Table 1. The variability of intra-U.S. output growth is generally smaller than the variability of intra-EU output, although the absolute values are similar. This suggests that fluctuations are more synchronized with aggregate behavior across the United States than across the European Union. However, this is not true for the Southwest and Rocky Mountain regions, whose output is heavily dependent on raw material production. In these regions, adjusting for aggregate U.S. activity has no impact on variability, indicating that growth is relatively uncorrelated with aggregate U.S. behavior.

The standard deviation of relative prices in the European Union varies from 0.032 in Denmark to 0.072 in the United Kingdom.14 There is a clear split between the original members of the European Union plus Ireland and Denmark, which have relatively low standard deviations, and the higher variation experienced by the United Kingdom, Greece, Spain, and Portugal. Table 2 shows the contribution of variability in national price levels and variability in nominal exchange rates to overall relative price variability for members of the European Union. The major difference between the original EU members, Ireland, and Denmark and the remaining countries is the lower level of nominal exchange rate variability in the first group (shown in the third column of figures), presumably reflecting the greater desire in these countries to stabilize intra-European exchange rates, as illustrated by their long-term membership in the ERM in the 1980s and the snake of the late 1970s.

Table 2.

Decomposition of Variance in Relative Prices

article image

Table 2 illustrates two other characteristics of intra-EU relative prices between 1963 and 1989. The first is the importance of nominal exchange rates in relative price movements. The standard deviation of the change in the nominal exchange rate is larger than that of relative national prices in every country except the Netherlands, and in most cases this difference is quite large. Another feature of the EU results is that the original EU members exhibit somewhat smaller variability of national prices than do newer entrants. Intra-U.S. relative price variability, reported in the bottom half of Table 1, is yet smaller.15

IV. Results

To identify supply and demand disturbances, structural vector autoregressions were estimated for each EU country and U.S. region by regressing the change in the logarithm of real output and of relative prices on the first and second lags of both series, and using the assumption that supply disturbances have no long-run impact on relative prices to identify demand and supply disturbances.16 The number of lags in the vector autoregressions was set at two since the Schwarz-Bayes information criterion indicated an optimal lag length of one or two in each case.17 The estimation period was 1963-89 for EU countries and 1965-89 for U.S. regions.

European Union and United States

Figure 2 shows impulse response functions for the 11 EU countries and 8 U.S. regions. Impulse response functions illustrate the impact on output and relative prices of a supply or demand disturbance equal to 1 standard deviation. Hence, the path shows the response of output and relative prices to a “normal” disturbance, and the size of the response is a measure of the importance of the particular disturbance. To aid comparison, the graphs for each response have the same scale.

Figure 2.
Figure 2.

Response to Supply Disturbance

Citation: IMF Staff Papers 1995, 001; 10.5089/9781451957068.024.A005

Note: The graphs show the response of output and relative prices to an average supply disturbance over time. Since the response are measured using logarithms, a change of 0.01 is approximately equal to 1 percent.

The response of relative prices to a supply disturbance, shown in Figure 2, illustrates the restriction imposed on the estimation procedure, namely, these disturbances have no long-run impact on relative prices. The responses of EU entrants are generally larger than those of the original EU members. Much more striking, however, are the much smaller relative price movements across U.S. regions. This is even true for the raw material producing regions (the Southwest and Rocky Mountains), which have the largest relative price responses of the U.S. regions.

Figure 2 also shows the response of output to supply shocks. The long-run response of output in the European Union is between 0.005 and 0.04 (since the data are in logarithms, this implies responses of between 0.5 and 4 percent), with the responses for the newer EU entrants being generally larger than those for the original members. The magnitude of the long-run responses of U.S. regions is similar to those of EU countries, although the speed of adjustment is slower. The two U.S. regions that are most specialized in raw material production, the Southwest and Rocky Mountains, again have relatively large responses.

The overidentifying restriction implied by the model is that the shortrun effect on relative prices should be the opposite of that on real output. This response, which is not imposed on the estimation, is satisfied for all of the EU countries. The results for the U.S. regions are less uniform. While most regions show a fall in relative prices, there are three exceptions: the Plains, the Far West, and the Great Lakes. However, in all three cases the responses are small compared with those of the other U.S. regions. Overall, combining the results for the European Union and United States, the relative price response is correct in 16 of 19 cases.18

Impulse response functions for demand disturbances are shown in Figure 3. The EU countries show relative price responses of between 0.02 and 0.07 (between 2 and 7 percent), with no clear difference between the behavior of the original EU members and newer entrants, although the original members do show some evidence of slower adjustment. As in the case of the supply disturbances, there is a big difference between the relative price responses in the European Union and in the United States, with the U.S. responses being much smaller than the EU ones. Within the U.S. regions, the largest relative price responses are again associated with the regions that specialize in raw material production.

Figure 3.
Figure 3.

Response to Demand Disturbance

Citation: IMF Staff Papers 1995, 001; 10.5089/9781451957068.024.A005

Note: The graphs show the response of output and relative prices to an average demand disturbance over time. Since the responses are measured using logarithms, a change of 0.01 is approximately equal to 1 percent.

In contrast to the relative price responses, the output responses to demand disturbances in the European Union are generally small, particularly for the original EU members, while the responses for the U.S. regions are significantly larger. The overidentifying restriction of the model is that the output response should have the same sign as the relative price response. Of the 11 EU country estimates, 7 have the anticipated response, while 4 responses are perverse. However, as with the U.S. regional results for supply disturbances discussed above, these perverse responses are all relatively small. The responses of the U.S. regions to demand shocks conform to the predictions of the model rather more closely than those for the EU countries, with only one region (the Far West) having a perverse response. Aggregating over both the European Union and the United States, the responses conform to the predictions of the underlying framework in 14 out of 19 cases.19

Overall, the results paint a fairly consistent picture. The European Union is characterized by large relative price responses to both types of disturbance, with particularly large responses to demand disturbances. The output responses to demand disturbances are relatively small, indicating that most of the adjustment to demand disturbances occurs through relative prices, and not through output adjustment. Within the United States, the relative price responses are both relatively small and relatively slow, presumably reflecting the highly integrated nature of the goods and factor markets and the limitations to relative price adjustment brought about by having a common currency.

Comparing Aggregate Responses

An alternative method of analyzing the impulse response functions is to look at the responses in relative price-output space. Recall from Figure 1 that a positive shift in the demand curve entails a movement up the supply curve. Similarly, a positive shift in the supply curve involves a short-run movement down the demand curve, followed by a movement back along the supply curve as the demand curve shifts outward over time. Hence, a scatter plot of the impulse response functions with respect to output against the impulse response functions with respect to relative prices should track the underlying demand and supply curves.

The results from such a scatter plot are shown in Figure 4. To compare the behavior of a typical EU country with that of a typical region in the United States, the graph uses the average responses across the EU countries and the average responses across the U.S. regions rather than showing the results for any specific country or region.20

Figure 4.
Figure 4.

Adjustment to Disturbances: EU Countries and U.S. Regions

Citation: IMF Staff Papers 1995, 001; 10.5089/9781451957068.024.A005

Note: The graph shows the average supply and demand curves for the European Union and United States derived from the estimated response of output and relative prices to underlying disturbances. Since the responses are measured using logarithms, a change of 0.01 is approximately equal to 1 percent.

The two curves in the upper right-hand quadrant of Figure 4 trace out the supply curve for the typical EU country and that for the typical region of the United States. In the case of the European Union, the supply curve is close to vertical, with almost all of the adjustment coming through movements in relative prices and virtually no output response. By contrast, the supply curve for the typical region in the United States has more of the adjustment coming through output than through prices. In addition, there is a distinct flattening of the estimated curve over time, which is consistent with labor migration in response to disturbances. Both results plausibly reflect the higher level of integration of factor markets within the United States.21

The length of the estimated curves indicates the average size of the underlying disturbances, while the check marks represent adjustment in each year. In the European Union, adjustment is relatively rapid, being virtually complete within two years. By contrast, in the United States less than half of the long-run adjustment occurs within two years. Hence, despite its significantly smaller size, relative price adjustment is considerably slower in the United States.

The initial segment of the lines in the bottom right-hand quadrant traces out the demand curves within the European Union and the United States. The demand curve for the typical EU country has a negative slope that is about 35 degrees below the horizontal, indicating that both output and relative prices respond significantly to supply disturbances. By contrast, the short-run demand curve for the typical region in the United States is very flat, with almost all of the adjustment coming through movements in real output and very little through movements in output prices. Regional goods are very close substitutes in the United States, while there is more differentiation of goods across the European Union. Both estimated demand curves are appreciably flatter than the estimated supply curves, indicating that goods markets are more integrated than factor markets.

The initial disturbance to output is somewhat larger in the European Union than in the United States, which may reflect a wider diversity of technologies in Europe. Correspondingly, the adjustment in the United States is significantly slower than in the European Union, which presumably indicates the greater importance of movements of factors of production (in particular labor) in the adjustment to supply disturbances within the United States.

Overall, the most striking differences between the two areas are the very different underlying supply curves and the slower rate of adjustment in the United States. These results can be interpreted as meaning that the lower integration of goods and (particularly) factor markets in the European Union implies the need for a significant degree of relative price flexibility in Europe in order to adjust to underlying disturbances. As discussed earlier, much of this flexibility is currently provided by adjustable nominal exchange rates. Such relative price flexibility is less important in the United States since more of the adjustment occurs through real output via integrated goods and factor markets. This allows the United States to operate a common currency without significant economic disruption, although the importance of movements of labor and capital does mean that adjustment is relatively slow.

The implication is that, without higher factor mobility, introduction of a common currency in the European Union could cause significant economic disruption by limiting the flexibility of relative prices. There is, however, an alternative interpretation of Figure 4 that is rather more supportive of a single European currency. Instead of being an equilibrating response to underlying disturbances, the large movements in relative prices observed in the demand response of the European Union could reflect excess volatility in nominal exchange rate markets. According to this view, the introduction of a common currency would simply lower the observed volatility in relative prices. While there may well be some truth to this view, there are at least two reasons for thinking that it does not present the whole story. While smaller than those across EU countries, estimated long-run relative price movements across U.S. regions are not inconsiderable.22 Also, as will be discussed below, the long-run relative price response of EU countries that were members of the snake and the ERM (arrangements designed to limit nominal exchange rate fluctuations) are similar to those of countries that were not involved in these arrangements, but their relative prices adjust at a slower rate. Hence, these mechanisms appear to have slowed the speed of response of relative prices without having reduced the size of the underlying adjustment.

It is also of interest to look at the size of the underlying disturbances over time. Economic integration in the European Union could have led to a significant fall in the size of the underlying disturbances over time, implying smaller required movements in relative prices and output across countries. To investigate this possibility, Table 3 reports the average size of the underlying disturbances for the European Union and the United States for three periods: 1963-72 (1966-72 for the United States), 1973-80, and 1981-89. The aggregate values are measured by calculating the variation in the disturbances for each country or region separately and averaging them.23 Since each shock is normalized to have a standard deviation of 1 over the estimation period, any deviation from unity indicates disturbances that were larger or smaller compared with the entire period. The results indicate a fall in the variability of both demand and supply disturbances in the European Union in the 1980s compared with the 1972-80 period, but very little change compared with the 1960s. This is very similar to the path evident in the U.S. data, indicating that the higher variability of the disturbances in the 1970s probably reflects the economic turbulence caused by the two oil shocks. There does not appear to have been an independent downward trend in demand and supply shocks within the European Union over time.24

Table 3.

Variability of the Underlying Disturbance over Time

article image

Behavior within the Regions

Thus far the comparisons have been between the typical behavior of an EU country and that of a region in the United States. However, as noted earlier, countries that were original members of the European Union appear to behave somewhat differently from those that entered the European Union at a later date. Similarly, within the United States the behavior of regions that are heavily concentrated in producing raw materials appears somewhat different from the remainder.

Figure 5 shows the responses for these groups separately. The results for the European Union show both interesting differences and some striking similarities between the two sets of countries. Consider first the supply curves traced out in the upper right-hand quadrant of the graph. Both curves are very steep and have almost identical lengths. However, while the later EU entrants reach the new equilibrium almost immediately (relative prices move by almost 4 percent in the first year), the speed of adjustment for the original EU members is significantly slower. This slower rate of adjustment plausibly reflects the greater commitment to inter-European exchange rate stability of the original EU members, through mechanisms such the ERM in the 1980s and the snake in the 1970s, as reflected in their lower observed relative price volatility (see Table 1).25

Figure 5.
Figure 5.

Adjustment within the European Union and the United States

Citation: IMF Staff Papers 1995, 001; 10.5089/9781451957068.024.A005

Note: The graph shows average supply and demand curves for different types of countries within the European Union and regions within the United States. The curves are derived from the estimated responses of output and relative prices to underlying disturbances. Since the responses are measured using logarithms, a change of 0.01 is approximately equal to 1 percent.

Turning to the lower right-hand quadrant, the results indicate that the slope of the short-run demand curve is relatively similar across the two groups of countries, implying a similar level of goods market integration. However, there does appear to be a significant difference in the size of the underlying disturbances. While an average disturbance changes real output in the original EU countries by 1 percent, it increases output for newer entrants by over 2 percent. The smaller underlying supply disturbances in the original EU members may reflect their greater homogeneity in technology and industrial structure.26 In both cases, the adjustment occurs very quickly, essentially within a year.

The top right-hand quadrant of the graph for the U.S. regions (Figure 5) indicates that the slopes of the supply curves for the raw material producers and for the more industrial regions are very similar. However, the underlying demand disturbances, which are traced out by these lines, are much larger for the raw material producing regions than for the industrial regions, although the speed of adjustment is not dissimilar. The basic pattern of similarly sloped curves but larger disturbances for the raw material producing regions also holds in the lower right-hand panel of the graph. Both sets of regions have relatively flat demand curves. However, the size of the supply disturbances for raw material producing regions is double that for industrial regions.

These results indicate that the slopes of the estimated demand and supply curves are relatively stable both in the European Union and in the United States. Hence, while differences in adjustment between the European Union and the United States largely reflect the lower level of market integration in the European Union compared with the United States, differences in behavior within the United States and the European Union reflect variations in the size of the underlying disturbances.

V. Conclusions and Implications for EMU

This paper has used structural vector autoregressions of the type proposed by Blanchard and Quah (1989) to look at the response of output and relative prices to supply and demand shocks within the European Union and within the United States. The results indicate that, at least between the 1960s and 1980s, the United States had significantly more integrated goods and factor markets than the European Union. As a result, relative prices were more important for adjustment within the European Union than within the United States, even though the estimated underlying disturbances were of a similar size. Adjustment occurred more quickly within the European Union, presumably reflecting the flexibility in relative prices implied by adjustable nominal rates of exchange.

What are the implications for EMU? By adopting a single currency, the European Union is likely to reduce the short-run flexibility of relative prices, making it more difficult and costly to adjust to underlying disturbances. Given the very steep estimated supply curve, this will be particularly important in response to demand disturbances. Indeed, the exchange rate turmoil in 1992 and 1993 can be seen as an example of this, with the ERM structure making it difficult for relative prices in the European Union to respond sufficiently quickly to the rise in demand for west German products caused by German unification.

In the longer term, increasing integration of EU goods and factor markets should reduce the need for large movements in relative prices. Institutional changes, such as the recent completion of the single market in the European Union, are important in promoting this integration. In addition, EMU itself will probably promote greater flexibility than has been seen in the past. With the exchange rate instrument no longer available, the incentive for greater flexibility in domestic prices and factor markets is clearly greater, while a single monetary policy may allow larger inflation differentials than those deemed acceptable by national central banks.

Having said this, it does not appear likely that the European Union will achieve anything like the levels of integration of U.S. regions in the immediate future. In the shorter run, disruptive relative price adjustments can probably be best avoided by reducing the size of underlying disturbances in demand for regional products. Coordination of domestic aggregate demand policies across EU countries, such as the fiscal restraints incorporated in the Maastricht Treaty, can be seen as one method of easing the growing pains likely to be associated with EMU.

APPENDIX I

A Simple Macroeconomic Model

This appendix presents a simple macroeconomic model of regional supply and demand. Consider a world made up of a large number of regions with identical economic structures. Output in region i is produced using a Cobb-Douglas production function:

y i = φ i + α l i + ( 1 α ) k i , ( A 1 )

where φi is the logarithm of a productivity disturbance, and yi, li, and ki are the logarithms of output, labor, and capital, respectively. Assume that labor and capital respond to changes in relative prices in the following way: li = γwi/(1 + γ) and ki = Kri/(1 + K), where wi, and ri are the logarithms of wages and the rental rate of capital and γ and K measure the responsiveness of factor demands to charges in factor prices. Then, for small changes in the underlying variables, the supply curve is

y i = ( 1 + γ α ( 1 α ) ( 1 + γ ) + k ( 1 α ) α ( 1 + k ) ) φ i + ( γ α ( 1 α ) ( 1 + γ ) + k ( 1 α ) α ( 1 + k ) ) p i . ( A 2 )

The slope of the curve depends upon the degree to which factors move in response to differences in rates of return. If factors do not move at all between regions, then the curve is vertical; by contrast, if both labor and capital are fully mobile, the coefficient on pi is (α/(1 - α) + (1 - α)/α).

Demand for local goods comes from workers, who receive wages, and capitalists, who earn income from the ownership of capital. It is assumed that workers consume only their locally produced good, yi, while capitalists have consumption that is fully diversified across all regions. All capitalists have the same preferences, given by

U k = Σ j ( 1 σ ) ( c j j ) , ( A 3 )

where UK is the utility function of capitalists, and cj and ∈j are the logarithms of consumption of good j and of a preference disturbance, respectively. For small changes in variables, the logarithm of the demand curve for local goods is

c i = α ( w i + l i ) + ( 1 α ) ( i p i σ ) , ( A 4 )

where wi is the real wage of workers in terms of local goods. The first term shows the demand for local goods from workers; since they only consume local products it depends only on their real income. The second term, which represents demand from capitalists, depends only on local prices and not on incomes. (Since the demand of capitalists for local goods makes up only a small fraction of the total incomes of capitalists, changes in the rate of return on regional capital have no impact on demand.) The coefficients a and (1 - α) reflect the weight of the two types of demand in total consumption.

It is assumed that both real wages and labor input are fixed in the short run, so that all short-run changes in income accrue to capitalists. In this case, the short-run demand and supply curves for local products are

s i = y i = ( 1 + λ ) φ i + λ p i    a n d d i = y i = ( 1 α ) ( i p i σ ) , ( A 5 )

where λ = K(1 - α)/α(l + K). Demand and supply disturbances act exactly as in the usual demand and supply diagram. As the labor market adjusts, real wages and labor input both change. The resulting redistribution of income from capitalists to workers causes an endogenous change in demand for products. Assuming that labor remains immobile,27 the long-run demand and supply curves are

s i = y i = ( 1 + λ ) φ i + λ p i   

and

d i = y i = α y i + ( 1 α ) ( i p i σ ) .    ( A 6 )

A positive demand or supply disturbance that raises output induces a subsequent expansion in demand equal to proportion a of the original change in output. This process will continue, causing demand to steadily expand over time.

APPENDIX II

The Estimation Methodology

This appendix presents the estimation methodology. Consider a system where the model can be represented by an infinite moving average representation of a (vector) of variables, Xt, and an equal number of shocks, ∈t. Formally, using the lag operator L,

X t = A 0 t + A 1 t 1 + A 3 t 2 + A 4 t 3 .... = Σ t = 0 L i A i t ( A 7 )

where the matrices Ai represent the impulse response functions of the shocks to the elements of X.

Specifically, let Xt be made up of the change in relative prices and the change in output, and let ∈t be supply and demand shocks. Then the model becomes

[ Δ r t Δ y t ] = Σ i = 0 L i [ a 11 i a 21 i a 12 i a 22 i ] [ s t d t ] , ( A 8 )

where rt and yt represent the logarithms of relative prices and output respectively;∈st and ∈dt are independent supply and demand shocks; and a11i represents element a11 in matrix Ai

Demand and supply shocks are identified by assuming that, while demand shocks have permanent effects on relative prices, supply shocks have only temporary effects. Since relative prices are measured as first differences, this implies that the cumulative effect of supply shocks on the change in relative prices (Δrt) must be zero:

Σ i = 0 a 11 i = 0. ( A 9 )

The model defined by equations (A8) and (A9) can be estimated using a vector autoregression (VAR). Each element of Xt is regressed on lagged values of all the elements of X. Using B to represent these estimated coefficients, the estimating equation becomes

X t = B 1 X t 1 + B 2 X t 2 + + B n X t n + e t = ( I B ( L ) ) 1 e t = ( I + B ( L ) + B ( L ) 2 + ) e t = e t + D 1 e t 1 + D 2 e t 2 + D 3 e t 3 + , A 10

where et represents the residuals from the equations in the VAR. In the case being considered, et is comprised of the residuals of a regression of lagged values of Δrt and Δyt on current values of each in turn, labeled ert and eyt, respectively.

To convert equation (A10) into the model defined by equations (A8) and (A9), the residuals from the VAR, et, must be transformed into the demand and supply shocks, ∈t. Writing et = C∈t, it is clear that, in the 2x2 case considered, four restrictions are required to define the four elements of the matrix C. Two of these restrictions are simple normalizations, which define the variance of the shocks est and ∈dt. A third restriction comes from assuming that demand and supply shocks are orthogonal.

The final restriction, which uniquely defines the matrix C is that supply shocks have only temporary effects on relative prices. This implies

Σ i = 0 [ d 11 i d 21 i d 12 i d 22 i ] [ c 11 c 21 c 12 c 22 ] = [ 0 . . . ] . ( A 11 )

This restriction allows the matrix C to be uniquely defined and hence the demand and supply shocks are identified.

REFERENCES

  • Bayoumi, Tamim, and Barry Eichengreen, “Shocking Aspects of European Monetary Union,” in Adjustment and Growth in the European MonetaryUnion, ed. by Francisco Torres and Francesco Giavazzi (Cambridge: Cambridge University Press, 1993), pp. 193229.

    • Search Google Scholar
    • Export Citation
  • Bean, Charles R., “Economic and Monetary Union in Europe,” Journal ofEconomic Perspectives, Vol. 6 (Fall 1992), pp. 3152.

  • Blanchard, Olivier Jean, and Lawrence F. Katz, “Regional Evolutions,” Brookings Papers on Economic Activity: 1 (1991), The Brookings Institution, pp. 1— 75.

    • 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
  • Clarida, Richard, and Jordi Gali, “Sources of Real Exchange Rate Fluctuations: How Important Are Nominal Shocks?” CEPR Working Paper No. 951 (June 1994).

    • Search Google Scholar
    • Export Citation
  • Cohen, Daniel, and Charles Wyplosz, “The European Monetary Union: An Agnostic Evaluation,” in Macroeconomic Policies in an InterdependentWorld, ed. by Ralph C. Bryant and others (Washington: The Brookings Institution; International Monetary Fund; London: Centre for Economic Policy Research, 1989) pp. 31137.

    • Search Google Scholar
    • Export Citation
  • Eichengreen, Barry, “Should the Maastricht Treaty be Saved?” Studies in International Finance, No. 74, Princeton University (Princeton, New Jersey: Princeton University, Department of Economics, International Finance Section, December 1992).

    • Search Google Scholar
    • Export Citation
  • Eichengreen, Barry, “One Money for Europe? Lessons from the U.S. Currency Union,” Economic Policy, Vol. 10 (April 1990), pp. 11887.

    • Search Google Scholar
    • Export Citation
  • Krugman, Paul, “Increasing Returns and Economic Geography,” Journal ofPolitical Economy, Vol. 99 (June 1991), pp. 48399.

  • Masson, Paul, and Steven Symansky, “Evaluating the EMS and EMU Using Stochastic Simulations: Some Issues,” IMF Working Paper 93/28 (Washington: International Monetary Fund, March 1993).

    • Search Google Scholar
    • Export Citation
  • Minford, Patrick, “Do Floating Exchange Rates Insulate?” in Exchange Ratesand Open Economy Macroeconomics, ed. by Ronald MacDonald and Mark P. Taylor (Oxford: Basil Blackwell, 1989), pp. 27594.

    • Search Google Scholar
    • Export Citation
  • Mussa, Michael, “Exchange Rates in Theory and in Reality,” Essays in International Finance, No. 179, Princeton University (Princeton, New Jersey: Princeton University, Department of Economics, International Finance Section, December 1990).

    • Search Google Scholar
    • Export Citation
  • Organization for Economic Cooperation and Development, OECD NationalAccounts (Paris: OECD, various issues).

  • Poloz, Stephen, “Real Exchange Rate Adjustment between Regions in a Common Currency Area” (unpublished; Ottawa: Bank of Canada, 1990).

    • Search Google Scholar
    • Export Citation
  • Sims, Christopher, “Macroeconomics and Reality,” Econometrica, Vol. 48 (January 1980), pp. 148.

  • U.S. President and Council of Economic Advisors, Economic Report of thePresident (Washington: U.S. Government Printing Office, February 1992).

    • Search Google Scholar
    • Export Citation
  • Weber, Axel, “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
*

This paper has benefited from the comments of participants at a seminar at Georgetown University, particularly those of Matt Canzoneri and Susan Collins, as well those of Peter Clark, Chris Towe, and Mark Griffiths at the IMF. Tamim Bayoumi is an Economist in the Research Department. He is a graduate of Cambridge and Stanford Universities. Alun Thomas is an Economist in the Western Hemisphere Department. He is a graduate of MIT.

1

There is a rapidly expanding literature on almost every aspect of EMU. For a survey see Eichengreen (1992).

3

We are unaware of any earlier empirical work of this type. Minford (1989) looks at this issue, but not from an empirical point of view.

4

Profits can be retained, spent on investment goods (which have a large import component), or distributed to investors, who need not live in the region. None of these uses is likely to have a particularly large short-term impact on the relative demand for local products.

5

Mussa (1990) discusses the role of flexible nominal exchange rates in real exchange rate variability.

6

A VAR is a system of two or more variables in which each variable is related to lagged values of all of the variables in the system (Sims (1980)).

7

This is proved in the technical appendix of Blanchard and Quah (1989).

8

Some empirical evidence in support of this assumption can be found in Clarida and Galí (1994), who analyze a three-variable VAR for Germany, Japan, Britain, and Canada, all relative to the United States. Using a slightly different approach from ours, they isolate aggregate demand, aggregate supply, and monetary shocks. They find that supply shocks explain only a small amount of the conditional variance of real exchange rate in these countries, and that the long-run real exchange rate responses to supply shocks are small.

9

The 11 countries include every EU country except Luxembourg, which apart from being very small was also in a currency union with Belgium over the sample period.

10

This period includes the collapse of the Bretton Woods exchange rate system in the early 1970s. Since Chow tests indicated no significant change in behavior in the early 1970s, the full data set was used to conserve degrees of freedom in the estimation.

11

New England, the Mideast, Great Lakes, Southeast, the Plains, Southwest, Rocky Mountains, and Far West.

12

As a result, external disturbances, such as the oil shocks of the 1970s, will only have an influence in so far as the individual regions or countries reacted in a different way from the region as a whole.

13

Results for the period after the breakup of the Bretton Woods system are very similar to those for the whole period.

14

In every case, the values in parentheses, which show relative price variability against the United States, indicate much higher variation, presumably reflecting movements in the nominal exchange rate of the dollar over the period.

15

Earlier work on real exchange rate variability within currency unions includes Poloz (1990) and Eichengreen (1990).

16

For both data sets, Dickey-Fuller tests indicated that the logarithms of the level of both real output and relative prices were nonstationary, but that the first differences were generally stationary. Accordingly, all variables were transformed into first differences.

17

A uniform lag of two was chosen in order to preserve symmetry across the estimation.

18

The probability of getting 16 or more correct responses if the responses were random is 1160/219, or about 0.2 percent. Hence, there is considerable evidence that the overidentifying restriction is satisfied.

19

The probability of this occurring randomly is 16664/219, or 3.2 percent.

20

There is a problem in aggregating countries or regions with responses that are perverse, since it is not clear which “incorrect” sign should be used. Fortunately, when the normalizations shown in Figures 2 and 3 are used, the perverse responses are generally very small, and hence the incorrect values make very little difference to the results. The one exception is the output response to a demand shock in the Far West region of the United States. Hence, for the demand responses the Far West region was excluded from the U.S. data.

21

Blanchard and Katz (1991) discuss the operation of regional labor markets in the United States in detail.

22

As will be discussed below, the long-run real exchange rate adjustment in the raw material producing regions of the United States is about half that estimated for EU countries, despite the existence of a common currency and large associated movements in regional output.

23

One reason for averaging them was that the individual results indicated the existence of a considerable amount of noise.

24

The disturbances were also divided between long-term members of the European Union and more recent entrants. Except for the relatively low level of demand disturbances for long-term EU members in the 1970s, there were no large differences in behavior between the groups.

25

Table 1 indicates that Denmark and Ireland, which are later EU entrants, also had relatively low real exchange rate volatility. When Denmark and Ireland are included with the original EU members and excluded from the later EU entrants, the results are similar. In a similar vein, results for a smaller group of core EU countries, consisting of Germany, the Netherlands, and Belgium, show very similar results to those for all long-term members of the European Union.

26

Bayoumi and Eichengreen (1993) find a similar result, namely that aggregate supply disturbances are more highly correlated for Germany and its immediate neighbors than for the rest of the European Union, the implication being that relative disturbances (which are measured in this paper) will be smaller for core members than for the rest of the European Union.

27

If labor is regionally mobile this will cause a flattening of the supply curve over time (see equation (A2)).

  • Collapse
  • Expand
IMF Staff papers: Volume 42 No. 1
Author:
International Monetary Fund. Research Dept.