Much of the conceptual framework discussed in this section represents refinements and extensions of an approach that has been prominent for several decades. As a complement to the methodology of basing exchange rate assessments on purchasing power parity (PPP) calculations or international competitiveness indicators, the macroeconomic balance approach, which features quantitative assessments of the exchange rates consistent with “appropriate” current account positions, has been used by the IMF staff since at least as far back as the summer of 1967, when views were being developed about the appropriate size of the prospective devaluation of sterling.1 Although it was not until 1973 that the staff first published in detail its Multilateral Exchange Rate Model (MERM),2 which provided consistent estimates of the trade effects of simultaneous changes in the exchange rates for the currencies of all industrial countries, an earlier version of MERM had become available in 1970 and provided a framework for analysis by the staff during the period preceding the Smithsonian conference in December 1971.3
The shift in the early 1970s to a system of floating exchange rates among many of the major currencies diminished the most obvious need for models that could be used to calculate appropriate par values. A second factor that acted to discourage further development of such models during the 1970s was a fairly common belief in the efficiency of foreign exchange markets. Such belief posed a challenge to the suggestion that econometric models could be superior to markets in providing assessments of appropriate exchange rates.4
By the mid-1980s, experience under the floating rate system had called into question the notion that freely functioning markets would always keep exchange rates closely aligned with their “equilibrium” values. As indicated in Figure 2.1, which shows nominal and real effective exchange rates for the major industrial countries since the inception of the system of generalized floating, currency values have exhibited wide fluctuations.5 Indeed, for the yen between early 1976 and early 1980, for the pound sterling between mid-1978 and the early 1980s, and for the U.S. dollar between the end of 1980 and early 1987, real effective exchange rates moved over ranges as wide as 40 to 60 percent. More recently, over the past five years, the yen has risen and fallen by more than 30 percent in real effective terms, while the pound has weakened and strengthened by about 30 percent. In contrast, the continental European countries have experienced more moderate fluctuations in their trade-weighted effective exchange rates, reflecting both the relatively large shares of intra-European trade in their total imports and exports and the relatively stable nominal exchange rates that have been maintained between European currencies. However, bilateral real and nominal exchange rates of the major continental European currencies against the U.S. dollar and the Japanese yen have sometimes fluctuated widely.
Major Industrial Countries: Nominal and Real Effective Exchange Rates
(1990 = 100; logarithmic scale)
Source: IMF, International Financial Statistics; and IMF staff estimates.1Real effective exchange rates are based on GDP deflators. For the last three quarters, data on GDP deflators represent. IMF staff estimates interpolated from the annual data and projections in the World Economic Outlook database.Economists have not been very successful in their attempts to explain the short-run behavior of exchange rates econometrically in terms of macroeconomic fundamentals.6 Partial explanations have been agreed upon for some of the wide swings in major currency exchange rates, such as the impact of the shift in the U.S. policy mix in pushing the dollar higher in 1980–82. However, for substantial parts of most of the wide swings of major currency exchange rates, convincing and generally accepted explanations in terms of movements in economic fundamentals do not exist.7
In this situation, it is not surprising that policymakers and other analysts have from time to time developed a broad consensus that certain exchange rates had become badly misaligned with fundamentals. Such views have typically been based on perceptions that exchange rates had moved to levels at which countries could not sustain their international competitiveness over the longer run, or at which countries would be projected to develop large macroeconomic imbalances. In this context, the framework described here may be seen as providing a firmer analytical foundation for such assessments. The premise of the analysis is that estimates of the equilibrium exchange rate levels that would be consistent with maintaining macroeconomic balance and international competitiveness in the medium run can be useful in helping to judge—along with other evidence and considerations—whether market exchange rates have become badly misaligned in the short run. As indicated in Section I, however, it is important to recognize that substantial deviations of exchange rates from their medium-run equilibrium levels can reflect cyclical factors or inappropriate policies and do not always indicate that exchange rates are misaligned.
As already mentioned, there are several approaches for assessing whether exchange rates seem broadly appropriate from a medium-run perspective. Perhaps the most widely used methodology for defining equilibrium exchange rates involves calculating PPP or international competitiveness ratios. Such calculations generally employ aggregate price or cost indices (such as indices of consumer prices, GDP deflators, export prices, or unit labor costs); their conceptual appeal comes primarily from the notion that the prices of (or the costs of producing) similar goods, when translated into a common currency unit, should be similar across countries—that is, should conform to the so-called “law of one price”— at least for tradable goods. As propositions about short-run behavior, both the law of one price for individual, narrowly defined categories of tradable goods and PPP for the aggregate outputs of countries are strongly rejected by the data.8 From a longer-term perspective, however, these propositions appear to have more empirical support, particularly when the PPP hypothesis is modified to allow for divergent trends in the prices of tradable and nontradable goods and services;9 and empirical testing of the long-run PPP hypothesis has undergone a remarkable rejuvenation in recent years.10
The approach of using indices of international price and cost competitiveness to gauge the equilibrium levels of exchange rates over the medium run has well-known limitations.11 First, because calculations of international competitiveness ratios are typically based on price or cost indices, rather than on data for the absolute levels of prices or costs, a country’s international competitiveness at prevailing exchange rates typically can only be assessed by comparison with its (average) international competitiveness ratio during some representative or normal base period; thus, such comparisons can be sensitive to the choice of base period. Second, assessments of international competitiveness can be sensitive to the type of price or cost indices on which they are based; for example, assessments based on GDP deflators can sometimes be quite different than those based on consumer prices, wholesale prices, export prices, or unit labor costs.12
Third, the theoretical assumptions that underlie the notion of a stable “equilibrium” level of international price or cost competitiveness can be challenged. In particular, the assumption that relative national price or cost levels (as measured by aggregate price or cost indices) should remain constant over time, at least for tradable goods, is called into question by several considerations: (1) the composition of tradable goods across countries can change over time; (2) changes over time in the relative prices of different tradables can contribute to deviations from PPP insofar as the weights of different categories of tradable goods in national price or cost indices differ across countries; and (3) the scope for arbitraging price or cost differentials across countries can be affected by the liberalization of trade and foreign exchange restrictions, reductions in transportation costs, or changes in other components of the costs of market penetration.13
These limitations notwithstanding, calculations of different measures of international price and cost competitiveness can often be helpful when judging whether exchange rates are reasonably close to medium-run equilibrium levels. The main focus here, however, is on the macroeconomic balance framework.
The macroeconomic balance approach, which addresses the requirements for achieving internal and external balance simultaneously, has been traced at least as far back as Nurkse (1945) and Metzler (1951), with pathbreaking contributions from Meade (1951) and Swan (1963). As refined by the IMF staff during the 1970s and 1980s,14 and also used by Williamson and others in their early work on “fundamental equilibrium exchange rates” (FEERs),15 the approach is rooted in the balance of payments identity, namely, the equality between the current account balance (CUR) and the net flow of private and official capital (CAP),
The current account balance is explicitly recognized to depend on the real exchange rate, which affects the volumes and values of imports and exports—with the complicating feature that the effects of changes in the exchange rate on CUR usually take some time to materialize fully. CUR also depends on levels of domestic and foreign incomes (or in some formulations on domestic and foreign output gaps), as well as on a variety of other factors that may shift the current account balance over time.
Early applications of the macroeconomic balance approach tended to treat CAP as the “normal” or “target” or “underlying” level of net capital flows. The equilibrium level of the exchange rate in this approach was the constant level of the exchange rate that would equate CUR to this normal level of CAP, with other factors affecting CUR usually assumed to be at their respective normal levels (e.g., domestic and foreign incomes were usually assumed to be at full employment).
While the two identities are closely related, the shift in emphasis has been in the direction of relying less on relatively ad hoc judgments about equilibrium capital flows and more on models of the equilibrium saving-investment balance,17 with particular attention to modeling the saving-investment balance in terms of its medium- or long-run determinants. This will be clarified later. Indeed, since S-I is a measure of the net outflow of saving, the model of S-I can be regarded as a model of the capital account balance—but to reiterate, from a medium- to-long-run perspective. In the framework developed below, short-term movements in international portfolio capital, which can be associated with strong pressures on market exchange rates in the short run, do not affect the medium-run equilibrium levels of real exchange rates unless they are also associated with factors that shift medium-run saving-investment balances.
A key distinction between the macroeconomic balance framework and other approaches to assessing exchange rates is that the former focuses explicitly on the current account—or more precisely (see below), on whether the outlook for current account positions, at prevailing real exchange rates, seems consistent with notions of normal or equilibrium saving-investment balances or capital flows. When policy authorities discuss the appropriateness of exchange rates, they often express opinions on whether current account imbalances seem inappropriately large; and in that context, the macroeconomic balance framework provides a relevant way of assessing the magnitudes of external positions. Approaches that rely on PPP measures and indicators of international competitiveness shed indirect light on the appropriateness of external positions but are usefully complemented by the direct focus that the macroeconomic balance approach provides.18
The Assessment Process
The methodology described here can be viewed as an attempt to strengthen the macroeconomic balance framework by building on equation (2.2), the national income accounting identity that links the current account (net exports) to the saving-investment balance. The assessment process has four steps, illustrated in Figure 2.2. The first step involves applying a trade-equation model to calculate the underlying current account positions that would emerge at prevailing market exchange rates if all countries were producing at their potential output levels; this focuses on the left-hand side of equation (2.2). As is clarified later, the relationship between a country’s underlying current account and its real exchange rate can be depicted by the line labeled UCUR in Figure 2.2; the negative slope of this line implies that a depreciation (or decline) in the real exchange rate improves the underlying current account. If the real exchange rate was R1, the first step in the assessment would identify the underlying current account position as UCUR1.
The second step uses a separate model to estimate an equilibrium or normal position for saving-investment balances based on the medium-run determinants of saving and investment, also assuming that countries are operating at potential output; this step focuses on the right-hand side of equation (2.2). In Figure 2.2, a country’s normal saving-investment balance is assumed to be independent of the level of its real exchange rate, as depicted by the vertical SI line.
The third step is to calculate how much exchange rates would have to change, other things equal, to be consistent with medium-run fundamentals—that is, to equilibrate the underlying current account positions with the medium-run saving-investment norms for each country simultaneously. Although this calculation is made in a multilateral framework, it is broadly similar to estimating the difference between R1 and R*, where R* corresponds to the medium-run equilibrium exchange rate at which the UCUR and SI lines intersect.
The final step involves judgmental assessments of whether the calculations in step three suggest that any currencies are badly misaligned. As already emphasized in Section I, large deviations between prevailing real exchange rates and estimates of the equilibrium levels consistent with medium-run fundamentals do not necessarily imply large misalignments.
Before describing the successive steps in the assessment process further, it is important to emphasize several points. First, the primary motivation for the analysis is to look for cases of badly misaligned exchange rates (“wrong rates”), not to prescribe exchange rate targets (“right rates”). Second, by focusing on the current account and saving-investment positions that would emerge if countries were producing at their potential output levels (i.e., were in positions of internal balance), the approach provides a framework for assessing whether current accounts and exchange rates are appropriately related to other fundamentals from a medium-run perspective. Third, the approach also has the attractive features of assessing external positions and exchange rates within a multilateral framework and potentially in a manner that is globally consistent. Fourth, the direct focus of the analytic framework is on real multilateral exchange rates—that is, on trade-weighted averages of nominal exchange rates adjusted for relative national price levels; when judging the appropriateness of current exchange rates, the distinction between real and nominal rates is typically irrelevant, but when focusing on exchange rates at some point in the future, the prospect of international inflation differentials can make the distinction quite important. Fifth, as mentioned in Section I, the analytic framework is intended to generate inputs to use as a starting point when assessing the appropriateness of prevailing exchange rates in the context of a broader range of considerations, including the cyclical positions of national economies and market participants’ expectations of exchange rate movements over the medium term. And sixth, at this point, applications of the quantitative methodology have been largely confined to industrial countries, partly in light of data limitations, but also because the particular model that is used to assess equilibrium saving-investment balances implicitly assumes a high degree of access to international capital markets.19
Underlying Current Account Positions
As already noted, the first step in the assessment process is to estimate each country’s underlying current account position, defined as the external balance that would emerge at prevailing market exchange rates if all countries were operating at potential output. For this, the IMF staff has focused on two alternative sets of estimates. One comes from the current account projections generated in connection with the semiannual World Economic Outlook (WEO) exercise20 and has the advantage of incorporating the country-specific knowledge and judgments of the IMF’s country experts. The other is generated from a standard trade model that has a relatively simple structure and employs common equation specifications and parameter values across countries. Here, the two alternative sets of estimates are referred to as the WEO projections and the model-based estimates;21 the trade model from which the latter estimates are generated is described in Section V. While relative simplicity and lack of country-specific detail are disadvantages, the model-based estimates have the positive attributes of global consistency and transparency. The trade model is also important in the third step of the assessment process (see below) for calculating the changes in exchange rates that would be needed to make current account balances consistent with medium-run equilibrium levels of saving-investment balances.
Various factors enter the calculations of underlying current account positions based on the trade model. The model has a standard structure: export volumes depend on the current and lagged values of the real effective exchange rate and on the weighted-average level of foreign activity (or aggregate demand); import volumes depend on the current and lagged values of the real exchange rate and the level of domestic activity. Other fundamental factors may also influence the current account, but rather than modeling them explicitly, the analytic framework incorporates them implicitly by allowing the intercept (or baseline) for the current account to shift over time in accord with actual experience. Thus, fundamentals that are not explicitly modeled are diagnosed to have changed whenever the observed (or estimated) base-period level of the current account differs from its previous value by an amount that is not entirely explained by movements in exchange rates and activity levels.
The model-based estimate of the underlying current account is calculated by adjusting the most recent WEO estimate of the current account in the present year (the baseline) for the effects of closing the foreign and domestic output gaps (i.e., of setting the levels of domestic and foreign activity at potential output), as well as for whatever effects of past exchange rate changes are estimated to be still in the pipeline. Consistent with the stylized facts reported in surveys of standard trade equations,22 the trade model assumes that it takes three years for trade volumes to respond fully to changes in exchange rates. It is also assumed that import prices respond fully and with no lags to exchange rate changes, and that export prices (measured in the exporter’s currency) are not directly influenced by exchange rates. In addition, the model assumes that the percentage responses (elasticities) of export and import volumes to a given percentage change in the real exchange rate are identical across countries; thus, in terms of absolute magnitudes, an economy with relatively high ratios of exports and imports to GDP will experience relatively large changes in trade volumes in absolute terms (and as shares of GDPs) in response to a given percentage change in its real exchange rate.
Table 2.1 provides several hypothetical examples. Reflecting the considerations just discussed, the calculations start from current account positions during a base year (column 1) and depend on ratios of trade to GDP (column 2) along with the base-year values of domestic and foreign output gaps (columns 3 and 4, constructed as actual output minus potential output) and the amounts that real effective exchange rates have changed during the current and previous two years (columns 5, 6, and 7). The three hypothetical country cases, which are assumed not to constitute the entire world, have been distinguished in several ways. Country 1 has a lower ratio of trade to GDP than countries 2 and 3. Country 1 is operating somewhat above potential, while countries 2 and 3 are experiencing considerable cyclical slack. Trade-weighted-average foreign output gaps are somewhat lower for countries 2 and 3 than for country 1. All three have experienced exchange rate changes of equal magnitudes during the present and past two years, but countries 1 and 2 have experienced these changes more recently than country 3.
Underlying Current Account Calculations
Positive number indicates appreciation.
Underlying Current Account Calculations
| Base-Year Data | Projected Change in Current Account Due to | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Current account | Ratio of trade to GDP | Domestic output gap | Foreign output gap | Percent change in real effective exchange rate1 | Closing domestic output gap | Closing foreign output gap | Effects of exchange rate changes | Underlying Current Account | |||
| (In percent of GDP) | Current year | Previous year | Two years previous | (In percent of GDP) | (In percent of GDP) | ||||||
| Country | (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) | (10) | (11) |
| 1 | -2.0 | 10 | 1 | -3 | 10 | 0 | 0 | 0.15 | 0.45 | -0.65 | -2.0 |
| 2 | 2.0 | 25 | -3 | -2 | -10 | 0 | 0 | -1.15 | 0.75 | 1.60 | 3.2 |
| 3 | 0.0 | 25 | -3 | -2 | 0 | 0 | -10 | -1.15 | 0.75 | 0.60 | 0.2 |
Positive number indicates appreciation.
Underlying Current Account Calculations
| Base-Year Data | Projected Change in Current Account Due to | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Current account | Ratio of trade to GDP | Domestic output gap | Foreign output gap | Percent change in real effective exchange rate1 | Closing domestic output gap | Closing foreign output gap | Effects of exchange rate changes | Underlying Current Account | |||
| (In percent of GDP) | Current year | Previous year | Two years previous | (In percent of GDP) | (In percent of GDP) | ||||||
| Country | (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) | (10) | (11) |
| 1 | -2.0 | 10 | 1 | -3 | 10 | 0 | 0 | 0.15 | 0.45 | -0.65 | -2.0 |
| 2 | 2.0 | 25 | -3 | -2 | -10 | 0 | 0 | -1.15 | 0.75 | 1.60 | 3.2 |
| 3 | 0.0 | 25 | -3 | -2 | 0 | 0 | -10 | -1.15 | 0.75 | 0.60 | 0.2 |
Positive number indicates appreciation.
The last four columns of the table show the implications of these contrasting cases for underlying current account estimates, based on approximately the same elasticity parameters as those used in the trade model. The assumed closing of domestic output gaps over the medium run has a small positive effect on the underlying current account for country 1 and large negative effects on the current account positions of countries 2 and 3 (column 8); this reflects both the relative sizes of the domestic output gaps and the fact that countries 2 and 3 have considerably higher ratios of trade to GDP than country 1. The closing of foreign output gaps has larger effects in countries 2 and 3 than in country 1 (column 9), despite the smaller size of the gaps in the former cases; this results from the smaller openness (i.e., trade-to-GDP) ratio in country 1. The relative impacts of exchange rate changes in the three cases (column 10) reflect, first, that country 2 is more open than country 1, and hence that a given percentage change in the exchange rate generates a larger absolute change in the ratio of the current account to GDP for country 2, and second, that base-year current account positions already include most of the effects on country 3, which experienced an earlier depreciation than country 2.
It should be apparent that the illustrative calculations are based on a streamlined trade model and have not taken account of any projected changes in net factor income payments or transfers over the medium run.23 As mentioned earlier, the WEO projections of current account positions at the end of the five-year WEO horizon (under the assumption of constant real exchange rates) can be viewed as an alternative set of measures of underlying current accounts. The availability of two alternative estimates provides a useful check to identify anomalies in either estimating procedure. One important conceptual difference between the two estimates is that the WEO projections apply to the terminal year in the five-year WEO horizon, whereas the underlying current account estimates based on the trade model are measures of what current account positions would be in the present year if countries were producing at potential and the effects of past exchange rate changes had been fully realized.24
Without even proceeding beyond step one, the estimates of underlying current account imbalances can be useful in forming preliminary judgments about possible exchange rate misalignments. Specifically, if the underlying current account balance is significantly outside the range of current account balances that a country has normally experienced, this can be an indicator of potential misalignment. The focus on the underlying current account, rather than the present current account balance, can be important. If a country has a substantial present current account imbalance but the imbalance may reasonably be expected to shrink (to a smaller underlying position) because of the effects of past exchange rate changes and future cyclical developments, then the present configuration of exchange rates may not appear problematic.25 On the other hand, a somewhat smaller prevailing imbalance that is expected to grow substantially at prevailing real exchange rates may be more problematic.
Saving-Investment Norms
The first step in the assessment process, as described above, focuses on the determinants of the current account balance, CUR, on the left-hand side of equation (2.2). The second step seeks to establish a standard for the “equilibrium” level of the current account by assessing the “normal” level of the saving-investment balance, S-I, on the right-hand side of the equation. There are several ways in which estimates of a “normal” saving-investment balance might be derived.
In this paper, the starting point for specifying S-I norms is to focus on the fitted values of a set of equations—here referred to as the S-I model—that relate the saving-investment balance to various medium-run determinants.26 Insofar as the approach involves the estimation of a set of equations with consistent specifications and parameter values across countries, the methodology can be viewed as a step forward—in terms of global consistency and transparency—from the relatively ad hoc approaches that others have taken in specifying equilibrium levels for current account positions.27
In the specification of the S-I model, the variables that are assumed to be the main direct determinants of saving and investment in the medium run are different than the variables that enter the current account model. In particular, each country’s saving-investment balance is assumed to depend on five variables: its stage of development, as represented by its per capita income position; its demographic structure, as summarized by a dependency ratio;28 its fiscal position; the gap between its actual and potential output levels; and the level of world interest rates.
As will become apparent below, the analytic framework essentially focuses on whether ex ante projections of medium-run S-I balances seem badly out of line with ex ante estimates of underlying current account positions, given prevailing exchange rates and policies. Such focus aims to identify situations in which exchange rates, policies, or some other economic factors will have to adjust, given that the saving-investment balance and the current account must be identical ex post.
The fact that aggregate saving must equal aggregate investment for the world as a whole provides a condition that relates the level of world interest rates to the other variables in the S-I model. This implies that the world interest rate can be substituted out of the reduced-form equations for the saving-investment balances of individual countries, even though it is an important determinant of both saving and investment separately (see Section IV). It also leads to reduced-form equations in which each country’s saving-investment balance depends on its per capita income level relative to a GDP-weighted average of per capita incomes in the world as a whole, as well as on the relative levels (compared with world GDP-weighted averages) of its dependency ratio, fiscal position, and output gap.29
As already noted, the purpose of estimating the S-I equations is to come up with a basis for quantifying levels of S-I balances that can be regarded as “historically normal,” given the levels of the principal explanatory variables for saving and investment over the medium run. To fit historical data, the output gap was included as a cyclical variable in the estimating equations, along with the medium-run determinants of saving and investment. Other variables that might help capture cyclical influences on S-I, such as country-specific interest differentials, were not included in the reduced form.
The S-I equations have been estimated with country-specific constant terms to capture the effects of “omitted variables” that may influence the relative saving and investment rates of different countries (see Section VI). These constant terms capture, for example, the fact that a relatively low saving-investment balance has been historically normal for the United States, other things equal, consistent with a relatively low observed saving rate for the United States, and perhaps also reflecting a relatively strong desire of nonresidents to accumulate claims on the United States. It may also be noted that the reduced-form equations do not include the exchange rate among the main direct determinants of saving or investment.30
Estimation of the S-I model produces the following results when other things are held constant. An increase in the domestic output gap (excess of actual over potential output) has a negative effect on the saving-investment balance. Countries with higher relative per capita incomes tend to have relatively high saving-investment balances. Higher dependency ratios imply lower saving-investment balances. Changes in the fiscal position tend to have “non-Ricardian” effects; in particular, an increase in the fiscal surplus is not fully offset by a decline in private saving and therefore has a positive effect on the saving-investment balance.31
To calculate medium-run norms for the saving-investment balance, the estimated equations are evaluated with output gaps set to zero, with per capita incomes corresponding to the levels that would prevail if output was at potential, and with fiscal balances at (smoothed values of) the structural budget positions reported in the World Economic Outlook database.32 Note that the structural budget positions projected in the World Economic Outlook do not necessarily correspond to “desirable” fiscal balances. Correspondingly, the S-I norms can be interpreted as normal or equilibrium saving-investment balances, given the projected fiscal positions, but not necessarily as desirable S-I balances.
The model-based estimates of the normal levels of S-I balances for the major industrial countries—corresponding to actual and projected structural fiscal positions, dependency ratios, and per capita incomes—are illustrated in figure 2.3 for the period from 1982 through 2002, the terminal year in the five-year WEO projection horizon.33 For the United States, the S-I norm is always in deficit during this period; that is, the normal level of domestic saving is less than the normal level of domestic investment, partly reflecting the relatively low propensity to save in the United States, other things equal. The modest upturn in the U.S. S-I norm beginning in 1995 mainly reflects the actual and projected improvement in the public sector’s structural budget position. For Japan, the S-I norm is persistently in surplus, in part reflecting a relatively high national saving propensity associated with a relatively low dependency ratio. The time profile of Japan’s normal S-I position partly reflects (smoothed) movements in the structural budget deficit of the public sector. In addition, projected population aging in Japan has direct effects on the saving-investment balance (at a given fiscal position), as well as indirect effects via its influence (through the social security system) on the projected fiscal balance. For Germany, the big shift in the S-I norm from significant surplus to deficit reflects the direct impact of German unification and the associated shift in German fiscal policy; the estimates for Germany include (in addition to the variables discussed earlier) a set of dummy variables to capture both the impact and the gradual erosion over time of the unification effect, and counterpart adjustments are made to the norms for other countries in proportion to their shares in Germany’s imports.34 The uptrend of the German S-I norm after 1991 also reflects the gradual correction of the structural fiscal position. For France and the United Kingdom, the S-I norms are close to zero, but increase moderately in 1991 as counterparts to the unification effect for Germany. For the United Kingdom, the modest trend over time mainly reflects (smoothed) movements in the structural fiscal position of the public sector. Similarly, for Canada the substantial rise of the S-I norm over the two-decade period covered in figure 2.3 mainly reflects the cumulative shift in the structural fiscal position. For Italy, about half of the rise in the S-I norm reflects a relatively large decline in Italy’s dependency ratio over the period from 1982 through 1995.35
Model-Based Estimates of Normal Saving-Investment Balances for the Major Industrial Countries
(in percent of GDP
Like the underlying current account estimates derived from the trade model, the norms derived from the S-I model have the characteristics of (approximate) global consistency and transparancy. From this perspective, they represent an attractive starting point for the exchange rate assessments. But the present S-I model also has limitations—for example, in failing to capture how the relationship between S-I balances and structural fiscal positions changes when fiscal deficits become large.
Discussions of current account positions sometimes focus on the issue of “sustainability.” In applying the framework described here to the industrial countries, it is presumed that moderately wide ranges of fiscal and S-I positions are probably sustainable over the medium run for these economies, but no attempt has been made to develop precise views about the ranges that are sustainable.36 More attention to sustainability issues would be important in extensions of the analysis to countries facing limited or highly variable access to international capital markets or with relatively high ratios of net foreign liabilities (NFL) to GDP. For such cases, the composition of capital inflows and the projected trends in ratios of NFL to GDP would appear to be quite relevant in defining sustainable and appropriate S-I balances.
Exchange Rates Consistent with Medium-Run Fundamentals
When the underlying current account balances determined in step one of the assessment process differ from the norms for the saving-investment balances determined in step two, the implication is that either exchange rates or other variables (including policies) will need to change from present levels over the medium term to be consistent with (projected) medium-run fundamentals. Step three assumes that other things remain equal and determines the direction and magnitude of the implied exchange rate changes.
To illustrate the logic of what is done in step three, applied to the real effective exchange rate of a single country viewed in isolation, it is helpful to focus again on Figure 2.2. Recall that the negatively sloped UCUR line plots the underlying current account position of the country in question as a function of that country’s real effective exchange rate. The UCUR line is constructed under the assumption that the exchange rate is held constant at the level shown on the vertical axis, and that all lagged effects of past exchange rate changes have been accounted for. It is also based on the assumption that output is equal to potential in the country in question and in all of its trading partners. It is negatively sloped to reflect the presumption that a lower real effective exchange rate (which strengthens a country’s international competitiveness) is associated with an improvement in the current account over the medium run, other things equal. Changes in economic fundamentals that directly affect the underlying current account at a given level of the real exchange rate—such as changes in relative productivity that alter competitiveness—are reflected in shifts of the position of the UCUR line.
The vertical SI line in Figure 2.2 shows the normal level of the saving-investment balance at given relative levels of potential outputs, dependency ratios, and structural fiscal positions, as determined in step two of the assessment process. The line is vertical because the normal level of the S-I balance is assumed not to depend on the exchange rate.37 The point at which the SI line intersects the UCUR line determines the equilibrium level of the real effective exchange rate (R*) that is assessed to be consistent with medium-run fundamentals.
From Figure 2.2 it is clear that changes in economic fundamentals that shift either the position of the vertical SI line or the negatively sloped UCUR line will alter the real effective exchange rate that is consistent with medium-run fundamentals. In particular, from the analysis in step two it follows that, other things constant, a higher relative level of real per capita income, or a reduction in the (relative) dependency ratio, or a rise in the (relative) structural fiscal surplus will shift the SI line in Figure 2.2 to the right and will imply a lower medium-run equilibrium level of the real effective exchange rate.38 The size of this implied change in the equilibrium exchange rate will depend on the extent of the shift in the vertical SI line and on the slope of the UCUR line. Shifts in the position of the UCUR line, due to changes in medium-run fundamentals that affect the current account through channels other than real exchange rates, also change the equilibrium level of the real effective exchange rate that is assessed to be consistent with medium-run fundamentals, but do not change the medium-run levels of either the saving-investment balance or (correspondingly) the current account balance.
More generally, changes in economic fundamentals may affect both the medium-run saving-investment norm and the position of the UCUR line. For example, a relative improvement in a country’s productivity in its tradable goods industries will improve international competitiveness (provided that relative real wages and other factor costs rise less than proportionately in the tradables sector); this will be reflected in an upward (and rightward) shift of the UCUR line. Such a relative productivity improvement will also tend to increase the S-I norm by raising relative per capita income and perhaps also by improving the government’s fiscal position. Provided that the rightward shift of the UCUR line exceeds that of the SI line, the medium-run equilibrium level of the real exchange rate will appreciate and the current account surplus will rise as a consequence of such a change in economic fundamentals.
It should be noted that changes in economic fundamentals that affect the position of the UCUR line appear to be quite important in practice. Specifically, in terms of the analytical framework summarized in Figure 2.2, a good deal of the movements observed over the medium run in real exchange rates and current account balances cannot be explained by shifts of a vertical SI line against a relatively fixed UCUR line. For example, the upward trend through the early 1990s in the S-I norm for Japan (recall Figure 2.3), which corresponds to a rightward shift in the SI line, cannot be reconciled with the trend real appreciation of the yen unless account is taken of a contemporaneous rightward shift in the UCUR line in association with trends in trade flows. Unfortunately, while it is possible to conclude after the fact that underlying fundamentals must have been changing in ways that shifted significantly the position of the UCUR line (as well as the SI line), it is often not possible to diagnose such changes in fundamentals as they are happening, let alone to predict them in advance.
To move from the analysis of the equilibrium level of the real effective exchange rate of a single country summarized in Figure 2.2 to conclusions about the equilibrium configuration of exchange rates in a multicountry world, some difficulties must be con-fronted. Because in an n-country world there are only n-1 independent exchange rates, it is not feasible to apply Figure 2.2 independently to all countries (or country groups) without imposing a mathematical requirement for global consistency. It is also important to assure that the sum of the current account positions implied by the configuration of equilibrium exchange rates is globally consistent, or in practical applications, that a simultaneous movement in all exchange rates from their prevailing to their equilibrium levels would not significantly affect the global current account discrepancy. Section VII describes the procedures used to derive an internally consistent set of equilibrium exchange rates and to assure global consistency (to a close approximation) between the saving-investment norms and the underlying current accounts associated with equilibrium exchange rates. It may be noted, however, that given the global consistency properties of the trade and S-I models, the results of calculating equilibrium exchange rates in a multilateral framework (when the underlying current account positions and S-I norms are generated from the models) generally do not deviate materially from those derived by focusing on countries individually (as depicted in Figure 2.2).
Judgmental Assessments
The final step in the assessment process is to reflect on whether the estimates of the exchange rates consistent with medium-run fundamentals suggest that any currencies are badly misaligned. In general, if differences between present market exchange rates and estimated medium-run equilibrium exchange rates are small—less than 5 percent or so—there is a very strong presumption against any conclusion of serious misalignment. After all, the framework used to determine estimates of medium-run equilibrium exchange rates clearly has its limitations, and considerable deference should be accorded to the market before suggesting any conclusion of serious exchange rate misalignment. For the same reasons, modest differences between market exchange rates and estimated medium-run equilibrium rates—differences of 10 percent or so—would not trigger any automatic conclusion of misalignment, although they might suggest further investigation in some circumstances where other judgmental considerations (discussed below) raised concerns. Significant deviations from estimated medium-run equilibrium exchange rates—deviations of about 15 percent or larger—raise the warning flag of possible exchange rate misalignment and warrant serious consideration of other judgmental factors before concluding either that exchange rates are misaligned or that the situation is apparently benign.
More is said below about the basis for judging 10 to 15 percent to be an appropriate threshold for triggering serious consideration of the possibility that exchange rates may be misaligned. It may be noted here that the “significance” of deviations from estimated equilibrium exchange rates does depend to some degree on the specific countries involved. Among the currencies of closely linked economies, such as those of the continental European countries, or of Canada and the United States, the range of bilateral real exchange rate movements that can plausibly be regarded (or tolerated) as consistent with an unchanged set of medium-run fundamentals—given the relatively high responsiveness of bilateral trade flows, in such cases, to movements in real bilateral exchange rates—is somewhat narrower than that among the currencies of Germany, Japan, and the United States.
Before reaching any conclusions about whether exchange rates are misaligned, it is important to adjust judgmentally for several factors that are not adequately addressed in the formal analytic framework. Given the limitations of the present trade and S-I models, the following types of judgmental considerations are relevant.
First, it is important to take account of cyclical and related monetary and financial conditions in various countries. If cyclical conditions are strong and monetary conditions are also firm in one country, while in another cyclical conditions are relatively weak and monetary conditions easy, then it is reasonable to expect that the exchange rate of the first country’s currency against that of the second will be strong relative to the medium-run equilibrium value suggested by the analytical framework. In this relative cyclical situation, an overvaluation of country one’s currency (vis-à-vis currency two) by 10 or 15 percent relative to estimates of its medium-run equilibrium position might be interpreted as a normal and desirable reflection of cyclical conditions, rather than as an indication of exchange rate misalignment. On the other hand, in this same relative cyclical situation, an undervaluation of currency one by 10 or 15 percent would raise serious concerns about a possible exchange rate misalignment.
Second, to get a quantitative notion of how the shorter-run behavior of exchange rates may be influenced by relative cyclical positions and related factors, including current and anticipated monetary policies, it is often useful to examine short- to medium-term interest rate differentials adjusted for differences in expected inflation rates.39 In the example discussed in the preceding paragraph, a real interest rate in country one that was 2 percent a year higher than that in country two over a five-year horizon would suggest that financial markets are expecting a 10 percent real depreciation of country one’s currency relative to country two’s currency over the five-year horizon.40 Of course, in this example if the real interest rate differential pointed in the other direction—toward further real appreciation of an exchange rate that already appeared potentially over-valued—then there would be increased cause for concern.
Third, the exchange rate implications of the fiscal situation require careful attention when public sector imbalances are large. Recall that the estimates of medium-run equilibrium exchange rates reflect norms for saving-investment balances that incorpo-rate the effects of structural fiscal positions. If a country has a large structural fiscal deficit, it tends (other things equal) to have a relatively low S-I balance and, according to the framework summarized in Figure 2.2, a relatively strong equilibrium real exchange rate. There are numerous examples of this phenomenon, including the strong U.S. dollar in the early 1980s and the strong deutsche mark after German unification. However, there are also situations where a country’s structural fiscal deficit may be viewed as unsustainable in the medium or longer run, and the financial market reaction to such a situation may, quite understandably, tend to produce a relatively weak currency rather than a relatively strong one. In such situations, it is not reasonable to base an assessment of a country’s medium-run equilibrium exchange rate on an estimate that assumes an apparently unsustainable fiscal position. Rather it is appropriate to come to a judgment about how much the estimate of the medium-run equilibrium exchange rate should be modified downward to be consistent with a significantly smaller and more sustainable fiscal deficit. Moreover, if the market exchange rate is observed to be significantly below the revised estimate of medium-run equilibrium, a second judgment must be made as to whether the market is overreacting to an adverse situation for which corrective fiscal actions are already being implemented, or whether the apparent undervaluation is more appropriately ascribed to the authorities failing to take sufficiently convincing action to assure the correction of their fiscal imbalance.
Fourth, it is clear that underlying structural conditions that affect medium-run equilibrium real exchange rates may change over time. In particular, there has been a long-term trend toward real effective appreciation of the Japanese yen (especially if real exchange rates are measured using consumer price indices), and there was a substantial apparent downward shift in the real exchange rate of the U.S. dollar against European currencies and the Japanese yen sometime in the early 1970s. Similarly, it appears that German unification has also induced a persistent shift in equilibrium real exchange rates (after allowing separately for the effects of the change in Germany’s fiscal position). And the list does not end with these examples. The difficulty is in knowing when such structural changes are occurring and are likely to persist in the future. There is no simple resolution of this difficulty, but the detailed economic analysis of experts on individual countries is a valuable input in any attempt at resolution.
In summary, the four-step assessment process for diagnosing possible exchange rate misalignments is a combination of formal analysis using an explicit multicountry framework, detailed understanding of the economic situations in individual countries, and bottom-line judgments factoring in other relevant considerations. The formal analysis imposes an important degree of rigor and consistency. The detailed knowledge provides essential input to the formal analysis and helps to inform the bottom-line judgment. The final judgment takes account of realities beyond the formal framework and recognizes the limitations of the whole exercise.
Limitations and Evaluation
Like other applications of the macroeconomic balance approach, the framework described here can be criticized for not using a more fully specified and dynamic multicountry econometric model. Such dynamic models generate complete future timepaths for equilibrium exchange rates; unfortunately, they generally do a very poor job of replicating the historically observed empirical behavior of exchange rates. By contrast, the macroeconomic balance approach simply generates a point estimate of the medium-run “equilibrium” exchange rate, rather than a timepath of the equilibrium exchange rate stretching from the short run to the long run. Thus, even though the framework described here might be viewed as an advance over other macroeconomic balance frameworks41—in the sense that it moves toward a global and multilaterally consistent set of trade equations and also embodies a set of S-I norms that are generated in a systematic and globally consistent way—it needs to be recognized that its numerical assessments are derived from a simplified analytic framework. As emphasized above, in the absence of a complete dynamic framework, judgments about cyclical factors and other considerations need to be superimposed on the model-based estimates of medium-run equilibrium exchange rates.
In this connection, it has been suggested that an attempt could be made to incorporate the Fund’s multicountry econometric model, MULTIMOD, into the exchange rate assessment framework.42 Such an effort would pose a number of difficulties. This is be-cause MULTIMOD has been designed to analyze the effects of various shocks on a World Economic Outlook baseline scenario, not to generate a baseline forecast itself. Thus, while MULTIMOD can be used to explore the interactions between policy adjustment and exchange rate adjustment in responding to shocks and to analyze the behavior of exchange rates over macroeconomic cycles generated by different types of shocks, it is not as well suited for analyzing whether the baseline forecast itself—namely, the judgmental forecast associated with the WEO—is likely to give rise to exchange market tensions.43 That being said, Section VIII of this study presents a dynamic application of the macroeconomic framework that uses the Japan block of MULTIMOD to generate a plausible baseline forecast for Japan and to simulate a dynamic path of the equilibrium exchange rate for the yen.44
A second limitation of the analytic framework is the uncertainty surrounding the estimates of medium-run “equilibrium” exchange rates. Uncertainty about the appropriate specifications for trade equations and the estimated values of elasticity parameters implies imprecision both in the underlying current account estimates and in the calculated magnitudes of the changes in exchange rates required to reconcile given estimates of underlying current account positions with given saving-investment norms. Additional uncertainty surrounds the equation specifications and estimated parameters that underlie the saving-investment norms.
In applications of the framework in the absence of any formally calculated confidence bands, the IMF staff has tended to regard 10 or even 15 percent deviations from estimates of equilibrium exchange rates as within a range that is not necessarily significant. There is reason to conjecture that a 50 percent statistical confidence band around the “equilibrium” exchange rates calculated from the combined trade and S-I models could be considerably wider than plus or minus 15 percent.45 However, drawing on the retrospective assessments provided later in this section, it may also be noted that setting the threshold value as high as 20 percent would come close to ignoring the “signals” provided in some of the cases for which there is now a fairly strong expost consensus that exchange rates were substantially misaligned.
It may be noted, as well, that the range of 10 to 15 percent is not used as a basis for drawing statistical inferences. Instead it serves as a threshold for indicating the possibility of serious misalignment and triggering a more judgmental assessment of the appropriateness of prevailing exchange rates in the context of a broad range of considerations. Such considerations include the cyclical positions of national economies and the stances of monetary and fiscal policies. Moreover, such broader assessments by the IMF staff take into account both the World Economic Outlook projections and the model-based estimates of underlying current accounts, as well as alternative views on saving-investment norms. They do so on the grounds that the degree of confidence inspired by the estimates of equilibrium exchange rates generated from a single pair of trade and S-I models may depend much less on the width of the statistical confidence bands for that pair than on the extent to which other models (or projection frameworks) lead to similar impressions about equilibrium positions.
A third limitation of the assessment methodology is that it has not yet been adapted to the developing and transition economies. In principle, the steps of estimating underlying current account positions and “equilibrium” saving-investment balances can be taken for any country, implying that applications of the framework need not be restricted to the industrial countries. However, the S-I model has not been estimated for the developing and transition economies, partly because of data limitations, but also because the model does not seem appropriate for countries that have limited access to international capital markets.46
How well does the assessment framework perform? There is not a long and continuous history of applying the framework, and the methodology used within the IMF to generate S-I norms has been evolving. Nonetheless, it is possible to construct approximate estimates of what the present methodology would have suggested on various occasions for which there is now a fairly strong ex post consensus that prevailing exchange rates were substantially misaligned.
For this purpose, the remainder of this section provides retrospective assessments of the exchange rates of the United States, Japan, and Germany in February 1985 and April 1995, as well as the exchange rates of the major European countries in June–July 1992. Such retrospective assessments abstract from two important considerations. First, the historical data that are used to calculate underlying current account positions and S-I norms may sometimes represent significant revisions from the estimates or projections that would have been used during the particular months for which exchange rates are being retrospectively assessed. Second, the assessments are based entirely on the trade and S-I models, without the benefit of whatever relevant adjustments might have been suggested at the time by the IMF’s country experts.
With these caveats in mind, Table 2.2 provides a retrospective assessment of what the methodology would have suggested about prevailing market exchange rates for the United States, Japan, and Germany in February 1985, the month in which the effective exchange rate of the U.S. dollar reached its peak. The calculations use annual data for 1985 as the base-year magnitudes of current account positions, output gaps, and so forth. The assessments in the table indicate that on a real multilateral basis, the dollar was substantially overvalued (46 percent) and the yen substantially undervalued (35 percent), while the deutsche mark was slightly undervalued (6 percent).47
Assessment of Exchange Rates Prevailing in February 1985
Positive number indicates that prevailing exchange rates were undervalued relative to their estimated medium-run equilibrium levels.
Assessment of Exchange Rates Prevailing in February 1985
| Current Accounts | Deviation of Multilateral Real Exchange Rate from Medium-Run Equilibrium Level1 | |||
|---|---|---|---|---|
| 1985 Actual | Underlying | S-I Norm | ||
| (In percent of GDP) | (In percent) | |||
| United States | -3.0 | -3.4 | -1.3 | -46 |
| Japan | 3.7 | 4.5 | 1.4 | 35 |
| Germany | 2.5 | 3.3 | 2.5 | 6 |
Positive number indicates that prevailing exchange rates were undervalued relative to their estimated medium-run equilibrium levels.
Assessment of Exchange Rates Prevailing in February 1985
| Current Accounts | Deviation of Multilateral Real Exchange Rate from Medium-Run Equilibrium Level1 | |||
|---|---|---|---|---|
| 1985 Actual | Underlying | S-I Norm | ||
| (In percent of GDP) | (In percent) | |||
| United States | -3.0 | -3.4 | -1.3 | -46 |
| Japan | 3.7 | 4.5 | 1.4 | 35 |
| Germany | 2.5 | 3.3 | 2.5 | 6 |
Positive number indicates that prevailing exchange rates were undervalued relative to their estimated medium-run equilibrium levels.
Estimates of real equilibrium bilateral rates against the U.S. dollar indicate a significant overvaluation of the dollar versus the deutsche mark (also 46 percent), but the multilateral rate for the mark, with its heavy weight on other European currencies, provides only a pale reflection of this situation. Notably, the dollar and yen respectively depreciated and appreciated substantially over the following two-year period (recall Figure 2.1), while the deutsche mark appreciated slightly on a real multilateral basis and quite substantially against the dollar.
The next retrospective assessment provides a snapshot of what the methodology would have suggested about the alignments of the deutsche mark, the French franc, the lira, and the pound sterling just prior to the summer-1992 crisis in European exchange markets. It may be recalled that the tremendous pressures that were unleashed in exchange markets that summer led to the withdrawal (in mid-September) of sterling and the lira from participation in the Exchange Rate Mechanism (ERM) of the European Monetary System, and subsequently to sharp depreciations of both currencies. The franc also came under strong downward pressure, which persisted off and on through August 1993, when the fluctuation bands of the ERM were widened considerably.
Table 2.3 presents a hypothetical assessment of the average exchange rates that prevailed in June–July 1992, using annual data for 1992 as the base-period magnitudes of the other variables that enter the calculations. Analysis at the time, based on the present methodology, would have focused first on the size of underlying current account imbalances (column 2), noting that Italy and the United Kingdom had underlying deficits in the neighborhood of 2½ percent and 3¾ percent of GDP, respectively, while Germany and France had relatively small underlying imbalances. Thus, Italy and the United Kingdom would have stood out, independently of specific estimates of S-I norms, as the subset of the four countries whose currencies seemed most vulnerable at the time to strong market pressures, other things equal.
Assessment of Exchange Rates Prevailing in June–July 1992
Positive number indicates that prevailing exchange rates were undervalued relative to their estimated medium-run equilibrium levels.
Assessment of Exchange Rates Prevailing in June–July 1992
| Current Accounts | Deviation of Multilateral Real Exchange Rate from Medium-Run Equilibrium Level1 | |||
|---|---|---|---|---|
| 1992 Actual | Underlying | S-l Norm | ||
| (In percent of GDP) | (In percent) | |||
| Germany | -1.0 | -0.4 | -1.8 | 10 |
| France | 0.3 | 0.3 | 0.4 | 1 |
| Italy | -2.4 | -2.5 | 1.3 | -29 |
| United Kingdom | -1.7 | -3.7 | -0.3 | -21 |
Positive number indicates that prevailing exchange rates were undervalued relative to their estimated medium-run equilibrium levels.
Assessment of Exchange Rates Prevailing in June–July 1992
| Current Accounts | Deviation of Multilateral Real Exchange Rate from Medium-Run Equilibrium Level1 | |||
|---|---|---|---|---|
| 1992 Actual | Underlying | S-l Norm | ||
| (In percent of GDP) | (In percent) | |||
| Germany | -1.0 | -0.4 | -1.8 | 10 |
| France | 0.3 | 0.3 | 0.4 | 1 |
| Italy | -2.4 | -2.5 | 1.3 | -29 |
| United Kingdom | -1.7 | -3.7 | -0.3 | -21 |
Positive number indicates that prevailing exchange rates were undervalued relative to their estimated medium-run equilibrium levels.
Had the assessment framework been in operation during June–July 1992, a range of judgments on appropriate S-I norms would probably have been collected and then compared with alternative estimates of underlying current account imbalances. For present purposes, however, Table 2.3 restricts attention to norms calculated from the S-I model (column 3). Based on these norms, the analysis would have suggested that the lira and pound sterling were overvalued by about 30 and 20 percent, respectively, while the franc was appropriately valued on a multilateral basis and the deutsche mark about 10 percent undervalued.48
The third retrospective exercise applies the present methodology to the exchange rates of the United States, Japan, and Germany during April 1995. As summarized in Table 2.4, the retrospective assessment, using realized data for 1995 as base-year numbers, suggests that the yen was about 30 percent overvalued on a multilateral basis, with the dollar nearly 20 percent undervalued. On a real multilateral basis the deutsche mark appears to have been only modestly overvalued (by 8 percent), but comparison of actual bilateral rates with estimates of equilibrium bilateral rates suggests a more substantial overvaluation vis-à-vis the U.S. dollar (20 percent).
Assessment of Exchange Rates Prevailing in April 1995
Positive number indicates that prevailing exchange rates were undervalued relative to their estimated medium-run equilibrium levels.
Assessment of Exchange Rates Prevailing in April 1995
| Current Accounts | Deviation of Multilateral Real Exchange Rate from Medium-Run Equilibrium Level1 | |||
|---|---|---|---|---|
| 1995 Actual | Underlying | S-l Norm | ||
| (In percent of GDP) | (In percent) | |||
| United States | -1.8 | -0.7 | -2.1 | 19 |
| Japan | 2.2 | 0.9 | 2.6 | -31 |
| Germany | -0.9 | -1.9 | -1.0 | -8 |
Positive number indicates that prevailing exchange rates were undervalued relative to their estimated medium-run equilibrium levels.
Assessment of Exchange Rates Prevailing in April 1995
| Current Accounts | Deviation of Multilateral Real Exchange Rate from Medium-Run Equilibrium Level1 | |||
|---|---|---|---|---|
| 1995 Actual | Underlying | S-l Norm | ||
| (In percent of GDP) | (In percent) | |||
| United States | -1.8 | -0.7 | -2.1 | 19 |
| Japan | 2.2 | 0.9 | 2.6 | -31 |
| Germany | -0.9 | -1.9 | -1.0 | -8 |
Positive number indicates that prevailing exchange rates were undervalued relative to their estimated medium-run equilibrium levels.
The conclusion that the major currencies were misaligned would presumably have been reinforced by judgmental assessment of the implications of real interest rate differentials. Looking forward over a five-year horizon from April 1995, U.S. real interest rates, by any reasonable estimate, exceeded Japanese real interest rates by at least 2 percent a year, suggesting that financial markets were implicitly expecting a further real appreciation of the yen against the dollar of at least 10 percent over the coming five years from its already very strong level.
Thus, based on the retrospective applications summarized in Tables 2.2 through 2.4, it seems valid to conclude that the assessment framework would have delivered “correct signals” in the cases that are widely regarded as the most extreme misalignments of major currencies during the 1980s and 1990s. This is only a weak test of the framework and should not be taken to suggest that the methodology would never have delivered incorrect signals about major misalignments. One important safeguard against incorrect signals is the practice of focusing on at least two alternative sets of underlying current account estimates, and of analyzing the basis for any substantial differences between the model-based estimates and the World Economic Outlook projections. There can be cases in which the World Economic Outlook projections of underlying current account positions incorporate important country-specific factors that are overlooked by the simplified trade-model framework, and there can be situations in which the systematic and globally consistent properties of the model-based calculations may raise questions about the projections for certain countries.
Conclusions
This section has provided a general description of a macroeconomic balance framework that the IMF staff has extended in recent years and found useful for assessing current account positions and exchange rates. In concluding, the following points deserve emphasis.
-
Exchange rate assessments are an integral part of the IMF’s surveillance responsibilities; and although such assessments necessarily rely on information and judgments that are not easily integrated into any formal analytical framework, a systematic and transparent framework can impose an important degree of rigor and multilateral consistency.
-
While the macroeconomic balance framework described here has been refined and extended in several significant ways during recent years, it basically reflects an approach that the IMF has been relying upon for the past three decades.
-
In applying the assessment framework, the IMF’s primary motivation is to look for cases of badly misaligned exchange rates (”wrong rates”), not to prescribe exchange rate targets or target zones. Estimates of “equilibrium” exchange rates cannot be placed within narrow confidence bands.
-
As a framework for helping to identify cases of badly misaligned exchange rates, the methodology is oriented toward assessing the extent to which prevailing exchange rates are consistent with medium-run fundamentals. It is important to recognize that substantial deviations of exchange rates from their medium-run equilibrium levels may reflect cyclical considerations or in-appropriate policies and do not always indicate that “markets are wrong.”
-
In applying the macroeconomic balance framework at the IMF, it has been useful to divide the process into several steps. The first step focuses both on the World Economic Outlook projections and on calculations based on the trade-equation model as two separate estimates of what underlying current account positions would be if all countries moved to potential output and if real exchange rates remained constant at prevailing market levels. The second step looks at the determinants of saving and investment, using a different type of model, and derives estimates of “equilibrium” or “normal” current account positions consistent with the underlying behavior of saving and investment in the medium run. The third step calculates the extent to which prevailing real exchange rates differ from the medium-run equilibrium levels that would bring the underlying current account positions into line with the saving-investment norms. The fourth step, which draws inter alia on the detailed knowledge of the IMF’s country experts, involves judgmental assessments of whether the estimates of exchange rates consis-tent with medium-run fundamentals suggest that any currencies are badly misaligned.
-
Whether anything should be done when it is concluded that prevailing exchange rates differ substantially from their medium-run equilibrium levels is appropriately addressed case by case and in the context of considering whether monetary and fiscal policies are appropriate from a broader perspective. In general, the IMF bases its advice primarily on assessments of the extent to which policies seem consistent with ultimate targets, rather than on the extent to which exchange rates deviate from estimates of their equilibrium levels.
Artus and Rhomberg (1973) present the original published version of MERM; Artus and McGuirk (1981) describe an extended version.
Polak (1995). The earlier version was based on the work of Paul Armington and Rudolf Rhomberg.
Among policymakers, significant concerns that floating exchange rates could be driven to inappropriate levels remained, and in the context of such concerns, discussions of the IMF’s surveillance procedures emphasized that countries with pegged and floating exchange rates should be treated evenhandedly. Consistently, the Second Amendment to the Fund’s Articles of Agreement (adopted by the Board of Governors in April 1976 and effective two years later) stressed that members should “avoid manipulating exchange rates” and, in the view of Polak (1995, p. 747), “conveyed, in much stronger terms than the par value regime had ever done, the notion that members should avoid serious deviations of their exchange rates from an equilibrium rate….”
Each country’s real exchange rate is constructed by calculating a trade-weighted average of foreign GDP deflators (price levels) converted at nominal exchange rates into the domestic currency unit, and then dividing by the country’s own GDP deflator. The calculations are based on trade weights from the Fund’s Information Notice System, which take into account competition between imports and locally produced import-substituting goods, competition between own exports and locally produced foreign goods, and competition between own exports and exports of other countries in third markets; see Zanello and Desruelle (1997) and McGuirk (1987).
For perspectives on empirical attempts to explain the short-run behavior of exchange rates, see Meese and Rogoff (1983a, 1988), MacDonald and Taylor (1992), Frankel and Rose (1995).
Surveys of institutional investors (e.g., Taylor and Allen (1992) and roup of Ten Deputies (1993)) have found that market participants, in driving exchange rates up or down, base their currency positions to a significant extent on charts or other forms of “technical analysis” of very recent trends or other patterns in the observed behavior of exchange rates, rather than basing their trading entirely on analysis of macroeconomic fundamentals. Similar characteristics are found in markets for equities and other financial assets; see Shleifer and Summers (1990). Whether reliance on technical analysis represents irrational behavior is a matter of debate. A literature is emerging that tries to rationalize such behavior in an environment of limited information about macroeconomic fundamentals; see Gennotte and Leland (1990), De Long and others (1990a, b), Banerjee (1992), Romer (1993). Related to this, it is evident from survey data that market participants have heterogeneous exchange rate expectations; see Ito (1990), Bryant (1995). It has also been found in simulation experiments that various types of trading strategies based on technical analysis generate statistically significant profits; see Dooley and Shafer (1983), Sweeney (1986), Cumby and Modest (1987).
See, for example, Isard (1977), Engle and Rogers (1996), Rogoff (1996).
Quantitative comparisons of living standards in different countries have observed that prices of nontradable goods and services, relative to prices of tradables, tend to be higher in high-income countries than in low-income countries; see, for example, Gilbert and Kravis (1954), Kravis, Heston, and Summers (1982), Summers and Heston (1991). Balassa (1964) and Samuelson (1964) hypothesized that this empirical regularity reflected a tendency for productivity in the tradable goods sector to rise relative to productivity in the nontradables sector as real incomes expanded. Isard and Symansky (1996) provide a summary of econometric tests of the Balassa-Samuelson hypothesis, which have been conducted mainly with data for Organization for Economic Cooperation and Development countries, along with some empirical perspectives for the Asia-Pacific Economic Cooperation Council region.
Recent reviews of the literature are provided by Breuer (1994), Isard (1995), MacDonald (1995), Froot and Rogoff (1995), Rogoff (1996).
See Artus and Knight (1984) and Clark and others (1994). In some cases, reliance on PPP for policy prescription has led to major difficulties, as in Great Britain’s return to the gold standard in April 1925, which was based on calculations showing that a return to the prewar gold parity would approximately restore sterling’s PPP against the U.S. dollar, and which turned out to have disastrous consequences for the British economy; see Moggridge (1972) and Kindleberger (1984).
See Lipschitz and McDonald (1992), Turner and Van ‘t dack (1993), and Marsh and Tokarick (1994) for perspectives on different competitiveness indicators. As an illustration, data for Japan show markedly divergent trends in different real exchange rate measures over recent decades.
Specifically, saving (S) minus investment (I) equals output (Y) minus absorption (where absorption is the sum of private consumption (C), government consumption (G), and investment), which also equals exports (X) minus imports (M), corresponding to the national income concept of the current account (CUR). When applying the macroeconomic balance approach, definitional distinctions between different concepts of the current account must be taken into account, especially with respect to the treatment of payments and receipts for factor services and transfers.
See, for example, Knight and Masson (1988), and Williamson (1994).
Single-equation reduced-form models, with exchange rates as dependent variables, are also used to assess the extent to which exchange rates deviate from estimates of their medium- or long-run equilibrium levels. In some of these models, the derivations of the reduced-form equations draw on both the macroeconomic balance framework and a modified PPP framework that allows, inter alia, for trends in relative productivity levels (Balassa-Samuelson effects) and terms-of-trade variables. But unlike the macroeconomic balance framework, this approach does not directly shed light on whether the outlook for current account positions seems appropriate. See Mussa (1984) and Frenkel and Mussa (1985) on the conceptual foundations for these reduced-form models. See Faruqee (1995), MacDonald (1997), and Goldman Sachs (1996, 1997) for empirical implementation.
See Hinkle and Montiel (1997) for a selection of recent work on estimating equilibrium exchange rates in developing countries.
The WEO projections are conditional on unchanged real exchange rates and assume that economies operate at potential output during the final year of the five-year projection period.
It may be noted that for many countries the WEO projections are also generated from formal models, but the structures of these models differ considerably across countries.
Work is in progress within the IMF to refine the standard trade model in certain directions, and to move from a calibrated to an estimated model, partly for modifying the exchange rate assessment framework in the context of European Economic and Monetary Union and partly to introduce country-specific estimates of various parameters. Given resource constraints and the priority attached to global coverage and multilateral consistency, the structure of the refined trade model will also be relatively streamlined.
The trade model is not well suited for projecting current account positions five years ahead, or for estimating underlying current account positions as of that year, since it has a simplified structure that precludes explicit allowance for projected changes over time in explanatory variables other than output and real exchange rates.
Although the discussion here abstracts from the fact that current account developments sometimes reflect significant structural changes, it is important to take account of relevant structural shifts in the assessment process, preferably in arriving at modified estimates of underlying current account positions, and otherwise in the final judgmental step of the process.
Section IV provides a derivation of a general globally consistent, reduced-form specification for the S-I balances of individual countries. Section VI describes several specific estimated models, along with details of the reduced-form equations that have been adopted as a starting point for quantifying normal S-I balances.
Compare, for example, the approach of Williamson and Mahar (1998). Polak (1995) points to the lack of agreed analytic procedures for modeling equilibrium current account positions as a major weakness in the IMF’s applications of the macroeconomic balance approach during the 1970s and 1980s.
The measure used here is the ratio of the population aged 19 and younger or 65 and older (combined) to the population aged 20 to 64, as calculated from data in the United Nations’ World Population Prospects, 1992. Masson, Bayoumi, and Samiei (1995) found that separating the overall dependency ratio into dependency ratios for the young and the old did not make a significant difference in explaining saving behavior. In estimating equations for the S-I balance, Debelle and Faruqee (1996 and Section VI of this study) found that the projected change in the dependency ratio over the five-year WEO horizon did not have significant explanatory power in a model that included both the level of the dependency ratio and the fiscal position.
The S-I equations have been estimated using panel data for the industrial countries. Because of data limitations, the estimation has not been extended to the developing and transition economies.
This is a simplifying assumption, consistent with most other empirically estimated models of saving and investment. A more complete macroeconomic framework could recognize that real exchange rates may influence saving and investment through their effects on income distribution, the terms of trade, and the profitability of the tradable goods sector.
There is a long-standing debate on the economic implications of public deficits; see Barro (1989) and Bernheim (1989) for reviews of the Ricardian and neoclassical perspectives. Whether the effects of an increase in the fiscal surplus on aggregate saving are positive, negligible, or even negative is related to the particular economic circumstances, including debt sustainability and capital market imperfections. Various empirical studies (see the discussion in the May 1996 World Economic Outlook, pp. 56–62) tend to suggest that for a country with a large public deficit, fiscal consolidation can have an expansionary effect on economic activity that operates through reductions in interest rate premiums; and this can presumably be associated with an increase in investment, a reduction in precautionary saving, and a decline in the country’s S-I balance—the opposite of the pattern that has been estimated to be historically normal in the present S-I model. Further work investigating cross-country experiences with respect to the effects of fiscal policy may prove useful in better understanding these issues and perhaps refining the model on which the S-I norms are based.
The data on countries’ structural fiscal positions reveal more year-to-year variation than seems desirable in S-I norms; hence, to calculate these norms, the structural fiscal positions are smoothed using a Hodrick-Prescott filter.
The extension of these S-I norms through 2002 is based on the WEO projections, as of August 1997, of the variables that enter the saving and investment equations.
The German unification adjustments are described in more detail in Section VI.
The norm for Italy is based on a different relationship between the S-I balance and the structural fiscal position than the norms for the other major industrial countries; see Section VI for details.
It is recognized that the analysis requires adjustment (see below) for cases in which prevailing fiscal positions are not sustainable (e.g., when Italy had a structural fiscal deficit of 10 percent of GDP in 1992), since it does not make sense to view the associated S-I norms as equilibrium balances.
As noted earlier, this simplifying assumption could be relaxed in principle, implying a non-vertical SI line, but is consistent with many other empirical models of saving and investment behavior.
In a more refined model, fiscal consolidation might generate a leftward shift in the SI line, and hence an appreciation of the medium-run equilibrium exchange rate, whenever structural public deficits were initially large.
As an agenda for future work, there is scope for linking judgments about the cyclical behavior of exchange rates to simulations of a fully specified dynamic macroeconometric model such as MULTIMOD, recognizing that the behavior of exchange rates generally depends on the nature of the shocks that generate the cycle. Section VIII takes a first step in this direction.
A substantial body of empirical research shows that interest rate differentials are not good predictors of changes in exchange rates over short horizons; the differentials often predict the exchange rate change with the wrong sign. By contrast, the presumption here is that real interest rate differentials are more reliable in assessing market expectations of medium-term developments in real exchange rates. This presumption is subject to question in particular circumstances.
For a recent alternative application of the macroeconomic balance framework to the major industrial countries, see Wren-Lewis and Driver (1998), who use “S-I norms” (current account targets) generated by Williamson and Mahar (1998).
Descriptions of MULTIMOD are provided in Masson, Symansky, and Meredith (1990), and Laxton and others (1998).
MULTIMOD is not normally used as a forecasting model, but one can think of the baseline path for a country’s real exchange rate as a model-consistent “equilibrium” path for which the real exchange rate in the long-run steady state satisfies the same type of condition that the macroeconomic balance framework of this paper uses to determine the medium-run equilibrium exchange rate. However, the baseline path in MULTIMOD reflects fairly arbitrary assumptions about the speed of convergence of the growth rates of different countries, and the steady-state levels of real exchange rates are moderately sensitive to these assumptions.
The analysis in Section VIII involves some modifications of the general specification of MULTIMOD as well as re-estimation of the trade equations for Japan. The conclusions about equilibrium exchange rates for the yen are broadly similar to the conclusions reached using the framework described in this section.
The different sources of uncertainty in the trade and S-I models interact in a way that makes it virtually impossible to calculate (or simulate) statistical confidence bands unless one resorts to un-realistic simplifying assumptions. Under the extreme assumption that the estimates of underlying current account positions were precise, the standard error for the saving-investment model would imply that the width of an interval of one standard deviation—or approximately 50 percent confidence—around the equilibrium exchange rate would be roughly plus-or-minus 8.5 percent for the European major industrial economies and Canada, and even larger for the more closed economies of the United States and Japan. Allowing for uncertainties about the specification forms and parameters of the model used to estimate underlying current account positions would presumably make the confidence bands considerably wider; see Kramer (1996).
By contrast, as described in detail in Section V, the standard trade model is used to generate estimates of underlying current account positions for six groups of developing and transition economies. See Reinhart (1995) and Senhadji (1997) for evidence that the standard trade model adequately captures the relationships between trade volumes and real exchange rates for these countries.
These crude retrospective assessments do not make adjustments for any “expected” exchange rate changes that may have been reflected in medium-term interest rate differentials.
A more traditional approach that focused on purchasing power parities and other competitiveness indicators and noted also the substantial current account deficits of Italy and the United Kingdom would also have suggested that these currencies were somewhat overvalued in the summer of 1992. Moreover, the cyclical situation in the United Kingdom suggested that it was not appropriate from a domestic policy perspective to maintain a high interest rate policy to defend sterling’s peg in the ERM in the face of the Bundesbank’s necessary efforts to combat German inflation. Accordingly, such a policy was somewhat lacking in financial market credibility.
References
Artus Jacques R 1978, “Methods of Assessing the Long-Run Equilibrium Value of an Exchange Rate,” Journal of International Economics, Vol. 8, pp. 277–99.
Artus Jacques R, and Malcolm D. Knight 1984, Issues in the Assessment of Exchange Rates of the Industrial Countries, IMF Occasional Paper No. 29 (Washington: International Monetary Fund).
Artus, Jacques R., Anne Kenny McGuirk 1981, “A Revised Version of the Multilateral Exchange Rate Model,” Staff Papers, International Monetary Fund, Vol. 28, pp. 275–309.
Artus, Jacques R., Rudolf R. Rhomberg, 1973, “A Multilateral Exchange Rate Model,” Staff Papers, International Monetary Fund, Vol. 20, pp. 591–611.
Balassa, Bela, 1964, “The Purchasing-Power-Parity Doctrine: A Reappraisal,” Journal of Political Economy, Vol. 72, pp. 584–96.
Banerjee, Abhijit V., 1992, “A Simple Model of Herd Behavior,” Quarterly Journal of Economics, Vol. 107, pp. 797–817.
Barro, Robert J., 1989, “The Ricardian Approach to Budget Deficits,” Journal of Economic Perspectives Vol. 3, pp. 37–54.
Bernheim, B. Douglas, 1989, “Neoclassical Perspectives on Budget Deficits,” Journal of Economic Perspectives, Vol. 3, pp. 55–72.
Breuer, Janice Boucher, 1994, “An Assessment of the Evidence on Purchasing Power Parity,” in Estimating Equilibrium Exchange Rates, ed. by John Williamson (Washington: Institute for International Economics), pp. 245–77.
Bryant, Ralph C., 1995, “The Exchange Risk Premium, Uncovered Interest Parity, and the Treatment of Exchange Rates in Multicountry Macroeconomic Models” (unpublished; Washington: The Brookings Institution).
Clark, Peter B., Leonardo Bartolini, Tamim Bayoumi, Steven Symansky, 1994, Exchange Rates and Economic Fundamentals: A Framework for Analysis, IMF Occasional Paper No. 115 (Washington: International Monetary Fund).
Cumby, Robert E., David M. Modest, 1987, “Testing for Market Timing Ability: A Framework for Forecast Evaluation,” Journal of Financial Economics, Vol. 19, pp. 169–89.
Debelle, Guy, Hamid Faruqee 1996, “What Determines the Current Account? A Cross-Sectional and Panel Approach,” IMF Working Paper 96/58 (Washington: International Monetary Fund).
De Long, J. Bradford, Andrei Shleifer, Lawrence H. Summers, Robert J. Waldmann, 1990a, “Positive Feedback Investment Strategies and Destabilizing Rational Speculation,” Journal of Finance, Vol. 84, pp. 379–95.
De Long, J. Bradford, Andrei Shleifer, Lawrence H. Summers, Robert J. Waldmann, 1990b, “Noise Trader Risk in Financial Markets,” Journal of Political Economy, Vol. 98, pp. 703–38.
Dooley, Michael P., and Jeff Shafer 1983, “Analysis of Short-Run Exchange Rate Behavior: March 1973 to November 1981,” in Exchange Rate and Trade Instability: Causes, Consequences and Remedies, ed. by David Bigman Teizo Taya (Cambridge: Ballinger), pp. 43–69.
Engle, Charles, and John H. Rogers 1996, “How Wide Is the Border?” American Economic Review, Vol. 86, pp. 1112–25.
Faruqee Hamid 1995, “Long-Run Determinants of the Real Exchange Rate: A Stock-Flow Perspective,” Staff Papers, International Monetary Fund, Vol. 42, no. l, pp. 80–107.
Frankel, Jeffrey A., Andrew K. Rose, 1995, “A Survey of Empirical Research on Nominal Exchange Rates,” in Handbook of International Economics, ed. by Vol. 3 Gene Grossman and Kenneth Rogoff (Amsterdam: North-Holland).
Frenkel, Jacob A., Michael L. Mussa, 1985, “Asset Markets, Exchange Rates and the Balance of Payments,” in Handbook of International Economics, ed. by Vol. 2 Ronald W. Jones Peter B. Kenen, (Amsterdam: North-Holland), pp. 619–141.
Froot, Kenneth A., and Kenneth Rogoff 1995, “Perspectives on PPP and Long-Run Real Exchange Rates,” in Handbook of International Economics, ed. by, Vol. 3 Gene Grossman and Kenneth Rogoff (Amsterdam: North-Holland).
Gennotte, Gerard, and Hayne Leland, 1990, “Market Liquidity, Hedging, and Crashes,” American Economic Review, Vol. 80, pp. 999–1021.
Gilbert, Milton, and Irving B. Kravis, 1954, An International Comparison of National Products and the Pur-chasing Power of Currencies: A Study of the United States, the United Kingdom, France, Germany, and Italy (Paris: Organization for European Economic Co-operation).
Goldman Sachs, 1996, The Foreign Exchange Market (July).
Goldman Sachs, 1997, The Foreign Exchange Market (September).
Goldstein, Morris, and Mohsin S. Khan, 1985, “Income and Price Effects in Foreign Trade,” Chap. 20 in Handbook of International Economics, ed. by, Vol. 2 Ronald W. Jones and Peter B. Kenen (New York: North Holland Press), pp. 1041–105.
Group of Ten Deputies,1993, International Capital Movements and Foreign Exchange Markets (Rome: Bank of Italy).
Hinkle, Lawrence E., and Peter J. Montiel, 1997, Estimating Equilibrium Exchange Rates in Developing Countries (preliminary draft; Washington: The World Bank).
International Monetary Fund 1987, Report of the Working Group on the Global Current Account Discrepancy (Washington: International Monetary Fund).
Isard, Peter, 1977, “How Far Can We Push the ‘Law of One Price’,” American Economic Review, Vol. 67, pp. 942–48.
Isard, Peter, 1995, Exchange Rate Economics (Cambridge: Cambridge University Press).
Isard, Peter, , and Steven Symansky 1996, “Long-Run Movements in Real Exchange Rates,” in Exchange Rate Movements and Their Impact on Trade and Investment in the APEC Region, IMF Occasional Paper No. 145, ed. by Takatoshi Ito and others (Washington: International Monetary Fund).
Ito, Takatoshi, 1990, “Foreign Exchange Rate Expectations: Micro Survey Data,” American Economic Review, Vol. 80, pp. 591–611.
Kindleberger, Charles P., 1984, A Financial History of Western Europe (London: George Allen & Unwin).
Knight, Malcolm D., and Paul R. Masson, 1988, “Fiscal Policies, Net Saving, and Real Exchange Rates: The United States, the Federal Republic of Germany, and Japan,” in International Aspects of Fiscal Policies, ed. by Frenkel Jacob (Chicago: University of Chicago Press), pp. 21–59.
Kramer, Charles, 1996, “FEERs and Uncertainty: Confidence Intervals and the Fundamental Equilibrium Exchange Rate of the Canadian Dollar,” IMF Working Paper 96/68 (Washington: International Monetary Fund).
Kravis, Irving B., Alan Heston, Robert Summers, 1982, World Product and Income: International Comparisons of Real Gross Product (Baltimore: Johns Hopkins University Press).
Laxton, Douglas, Peter Isard, Hamid Faruqee, Eswar Prasad, Bart Turtleboom, 1998, MULTIMOD Mark III: The Core Dynamic and Steady-State Models, IMF Occasional Paper No. 164 (Washington: International Monetary Fund).
Lipschitz, Leslie, and Donogh McDonald, 1992, “Real Exchange Rates and Competitiveness: A Clarification of Concepts and Some Measurement for Europe,” Empirica—Austrian Economic Papers, Vol. 19, pp. 37–69.
MacDonald, Ronald, 1995, “Long-Run Exchange Rate Modeling,” Staff Papers, International Monetary Fund, Vol. 42, pp. 437–89.
MacDonald, Ronald, 1997, “What Determines Real Exchange Rates? The Long and Short of It,” IMF Working Paper 97/21 (Washington: International Monetary Fund).
MacDonald, Ronald, and Mark P. Taylor 1992, “Exchange Rate Economics: A Survey,” Staff Papers, International Monetary Fund, Vol. 39, pp. 1–57.
Marsh, Ian W., and Stephen P. Tokarick, 1994, “Competitiveness Indicators: A Theoretical and Empirical Assessment,” IMF Working Paper 94/29 (Washington: International Monetary Fund).
Masson, Paul, Tamim Bayoumi, Hossein Samiei, 1995, “Saving Behavior in Industrial and Developing Countries,” Staff Studies for the World Economic Out-look (Washington: International Monetary Fund), pp. 1–27.
Masson, Paul, Steven Symansky, Guy Meredith, 1990, MULTIMOD Mark II: A Revised and Extended Model, IMF Occasional Paper No. 71 (Washington: International Monetary Fund).
McGuirk, Anne K., 1987, “Measuring Price Competitiveness for Industrial Country Trade in Manufactures,” IMF Working Paper 87/34 (Washington: International Monetary Fund).
Meade, James E., 1951, The Theory of International Economic Policy Volume One: The Balance of Payments (London: Oxford University Press).
Meese, Richard A., and Kenneth Rogoff, 1983a, “Empirical Exchange Rate Models of the Seventies: Do They Fit Out of Sample?” Journal of International Economics, Vol. 14, pp. 3–24.
Meese, Richard A., and Kenneth Rogoff, 1983b, “The Out-of-Sample Failure of Empirical Exchange Rate Models: Sampling Error or Misspecification?” in Exchange Rates and International Macroeconomics, ed. by Jacob Frenkel (Chicago: Chicago University Press), pp. 67–112.
Meese, Richard A., and Kenneth Rogoff, 1988, “Was It Real? The Exchange Rate-Interest Differential Relationship over the Modern Floating-Rate Period,” Journal of Finance, Vol. 43, pp. 933–48.
Metzler, Loyd A., 1951, “Wealth, Saving, and the Rate of Interest,” Journal of Political Economy, Vol. 59, pp. 93–116.
Moggridge, Donald E., 1972, British Monetary Policy 1924–1931: The Norman Conquest of $4.86 (Cambridge: Cambridge University Press).
Mussa, Michael, 1984, “The Theory of Exchange Rate Determination,” in Exchange Rate Theory and Practice, ed. by John F.O. Bilson Richard C. Marston (Chicago: University of Chicago Press), pp. 13–78.
Nurkse, Ragnar, 1945, “Conditions of International Monetary Equilibrium,” reprinted in The International Monetary System: Highlights from Fifty Years of Princeton’s Essays in International Finance, ed. by Kenen Peter B. (Boulder, Colorado: Westview Press), pp. 1–24.
Polak, Jacques J., 1995, “Fifty Years of Exchange Rate Re-search and Policy at the International Monetary Fund,” Staff Papers, International Monetary Fund, Vol. 42, pp. 734–61.
Reinhart, Carmen M., 1995, “Devaluation, Relative Prices, and International Trade,” Staff Papers, International Monetary Fund, Vol. 42, pp. 290–312.
Rogoff, Kenneth, 1996, “The Purchasing Power Parity Puzzle,” Journal of Economic Literature, Vol. 34, pp. 647–68.
Romer, David, 1993, “Rational Asset-Price Movements Without News,” American Economic Review, Vol. 80, pp. 1112–30.
Samuelson, Paul A., 1964, “Theoretical Notes on Trade Problems,” Review of Economics and Statistics, Vol. 46, pp. 145–54.
Senhadji, Abdelhak, 1997, “Time-Series Estimation of Structural Import Demand Equations: A Cross-Country Analysis,” IMF Working Paper 97/132 (Washington: International Monetary Fund).
Shleifer, Andrei, Lawrence H. Summers, 1990, “The Noise Trader Approach to Finance,” Journal of Economic Perspectives, Vol. 4, pp. 19–33.
Summers, Robert, and Alan,Heston, 1991, “The Penn World Table (Mark 5): An Expanded Set of International Comparisons, 1950–1988,” Quarterly Journal of Economics, Vol. 106, pp. 327–68.
Swan, Trevor W., 1963, “Longer-Run Problems of the Balance of Payments,” in The Australian Economy: A Volume of Readings, ed. by Heinz W. Arndt and W. Max Corden (Melbourne: Cheshire), pp. 384–95.
Sweeney, Richard J., 1986, “Beating the Foreign Exchange Market,” Journal of Finance, Vol. 41, pp. 163–82.
Taylor Mark P., and Helen Allen, 1992, “The Use of Technical Analysis in the Foreign Exchange Market,” Journal of International Money and Finance, Vol. 11, pp. 304–14.
Turner, Philip, Jozef Van ‘t dack, 1993, “Measuring International Price and Cost Competitiveness,” BIS Economic Papers, Vol. 39 (Basel: Bank for International Settlements).
Williamson, John, 1985, The Exchange Rate System (Washington: Institute for International Economics, 2d ed.).
Williamson, John, 1994, Estimating Equilibrium Exchange Rates (Washington: Institute for International Economics).
Williamson, John, and Mahar, Molly 1998, “Current Account Targets,” Appendix A in Real Exchange Rates in the Year 2000, by S. Wren-Lewis and R. Driver (Washington: Institute for International Economics).
Wren-Lewis, Simon, and Rebecca Driver, 1998, Real exchange Rates for the Year 2000 (Washington: Institute for International Economics).
Zanello, Alessandro, and Dominique Desruelle, 1997, “A Primer on the IMF’s Information Notice System,” IMF Working Paper 97/71 (Washington: International Monetary Fund).
