III. The Post-Bretton Woods Era

Carlo Cottarelli, and Curzio Giannini
Published Date:
December 1997
  • ShareShare
Show Summary Details

Against the background of the preceding discussion, we now move to reviewing the actual evolution of monetary frameworks during 1970–94.

Country Sample

Information was gathered on the evolution of monetary frameworks in 100 countries (or country unions). This country group includes almost all independent nations in existence for the 25 years covered in this paper, with the exception of countries whose monetary system was seriously disrupted by external or internal wars and countries under central planning for most of the observed period.8 Thus, our country group does not include any of the so-called transition economies, a painful exclusion as those economies have recently become the locus of several monetary experiments (Begg (1996)). However, this drawback is offset by the advantage of working for the entire sample period with a panel of relatively homogeneous countries. Needless to say, it would be dangerous to carry over to countries introducing new currencies (as many transition economies) the experience of other countries (Selgin (1994) discusses the problems involved in introducing new fiat money).

This country group includes 22 industrial and 78 developing countries or country unions (currency unions).9 A currency union is here considered as an individual entity distinct from its members: the monetary framework of the latter is implicit in the choice of joining a currency union, while the monetary framework of the union itself depends on the rules regulating the actions of the union’s central bank. The three existing currency unions (West African Monetary Union, Central African Monetary Union, and Eastern Caribbean Currency Area) are thus included as separate entities. The countries belonging to the first two unions are also included individually. The members of the East Caribbean Currency Area were not included, as they became formally independent only after 1970.

Taxonomy of Monetary Frameworks

Monetary frameworks can, in principle, be differentiated according to three aspects: (1) whether there exist announced rules or formal institutions affecting the behavior of the monetary authorities; (2) the costs associated with repudiating announced rules; and (3) the costs of monitoring the possible deviation of the monetary authorities from announced rules. The latter are relevant because high monitoring costs favor noncompliance with the announced rules even in the absence of formal repudiation.

Our classification of monetary frameworks focuses primarily on the first aspect. In principle, this choice may bias some of the conclusions reached in this study regarding the relation among discretion, credibility and monetary frameworks, as the factors mentioned under (2) and (3) do affect the credibility of (and the degree of discretion implicit in) a certain monetary framework (Flood and Isard (1989), Lohmann (1992)). However, focusing on the first aspect has three practical advantages. First, it allows us to avoid highly subjective decisions on repudiation and monitoring costs. For example, multilateral pegs are often regarded as more costly to repudiate than unilateral pegs, as parity revisions require consensus among all countries participating in the peg (Persson and Tabellini (1994)). But Cukierman, Rodriguez, and Webb (1995) find that, in practice, unilateral pegs may have imposed equally binding constraints on monetary policies than multilateral constraints.

Second, it reduced significantly the number of potential frameworks (some of which would have included only one or two countries), thus simplifying the analysis of the main trends. For example, incorporating the difference between multilateral and unilateral pegs would have required adding three additional frameworks (corresponding to frameworks 4, 5, and 6 of Table 1).

Table 1.A Taxonomy of Monetary Frameworks
Monetary FrameworkCan the Monetary Authorities
Set short-term interest rates independently from monetary conditions abroad?Adjust their inflation target in response to country specific shocks?Set short-term interest rates independently from monetary conditions abroad?Adjust their inflation target in response to country specific shocks?
(In the short run)(In the long run)
Foreign currencyNNNN
Currency unionNNNN
Currency boardNNNN
Exchange rate peg without capital controlsNNNN
Exchange rate peg with capital controls and a short-term intermediate targetYNNN
Exchange rate peg with capital controlsYYNN
Inflation targetingYYNN
Short-term intermediate targetYNYY
Source: Appendix: Country Data.
Source: Appendix: Country Data.

Third, focusing also on aspects (2) and (3) would not have significantly altered the conclusions on the main trends in the evolution of monetary frameworks highlighted in the following sections. This is because there have been limited shifts of countries between frameworks differing for how easily announcements can be repudiated. In this respect, the most relevant change was the shift from Bretton Woods multilateral pegs toward bilateral pegs. But, assuming that multilateral pegs can be regarded as more costly to repudiate, this shift, if anything, strengthened the trends toward more discretionary frameworks highlighted in the following sections. As to monitoring costs, exchange rate-based frameworks are usually considered easier to monitor than monetary rules. But, as we will see, in our classification, exchange rate frameworks (with the exception of crawling pegs) are, albeit for reasons unrelated to monitoring costs, attributed a lower weight in terms of discretion than frameworks based on monetary targeting.

Thus, based on announced policy rules, nine basic monetary frameworks can be identified, which are listed in Table 1 by increasing degree of discretion. In assessing the degree of discretion, two aspects are here considered, namely whether the monetary authority (that is, the agency—the government or the central bank—in charge of monetary policy) has the possibility of (1) setting short-term interest rates independently from monetary conditions abroad and (2) surprising the private sector through unanticipated inflation without repudiating the announcements. These two aspects are considered both in the short run and in the long run, and the table reports whether there is (Y) or there is not (N) scope for discretion under each reported framework. The relative degree of discretion of the frameworks has been assessed based on the number of yeses in the tables, and on the principle that long-term discretion is to be weighted more heavily than short-term discretion.

The first four frameworks (use of foreign currency as the only legal tender, membership in a currency union, replacement of a central bank with a currency board, and pegging the exchange rate in the absence of capital controls) leave no room for an independent monetary policy, either in the short run or in the long run, as domestic interest rates are continuously pegged, through arbitrage operations, to foreign interest rates.10 In defining currency boards, some ambiguity is unavoidable. In principle, a currency board is characterized by the fact that 100 percent of base money creation must be matched by the acquisition of foreign exchange reserves by the central bank at a preannounced exchange rate. But, in practice, the percentage is almost never 100 percent, as some limited flexibility is always admitted, at least in the very short term. This survey considers as currency boards those arrangements in which at least 90 percent of base money must be covered by foreign exchange.

The fifth framework (pegged exchange rate with capital controls and a short term intermediate monetary target) allows the monetary authority to set the stance of monetary policy in the short run independently from abroad. However, after the intermediate target is announced, monetary policy is in principle geared to the attainment of the intermediate target and cannot respond to shocks affecting the economy. Moreover, through the price-specie flow mechanism, in the long run, domestic inflation must be brought in line with foreign inflation, thus constraining the discretion of monetary policy.

The sixth framework differs from the fifth in the absence of a domestic short-term target, which increases short-run discretion.

The seventh framework (inflation targeting) has attracted much attention in recent years, particularly in industrial countries. “Inflation targeting” is not purely the announcement of some short-run inflation target by the government—something that to a different degree occurs in most countries—but the announcement of a targeted inflation path extending to a few years ahead, coupled with the setting up of procedures for public monitoring of how the monetary authorities pursue their objective.11 Owing to medium- to long-term announced inflation targets, monetary policy has no medium- to long-term discretion. However, in the presence of flexible exchange rates to which inflation targeting is typically associated, there is room for keeping domestic monetary conditions independent from foreign monetary conditions.

The eighth framework is characterized by the announcement of a short-term intermediate target, either in the form of a monetary aggregate or of a (typically crawling) peg. This framework constrains the monetary authorities in the short run, but not in the long run, as the targets are revised periodically (at least annually).

Finally, the last framework is characterized by the lack of any announcement binding the monetary authorities (full discretion).

Before proceeding, two caveats are in order. First, the above nine frameworks do not include all monetary frameworks suggested by the economic literature. There are two remarkable omissions: the announcement of medium- to long-term intermediate monetary targets (or of a money growth rule a la Friedman); and the announcement of nominal income targets. The reason for these omissions is the absence of practical examples of these frameworks in our country group. As discussed in Section V, monetary targeting has (almost) always been characterized by short-term announcements, rather than by the announcement of money rules. Nominal income targeting is also not a common practice, although in some cases, short-term monetary targets are explicitly derived from consistent nominal income targets. For example, it has been argued that the Bundesbank follows an income targeting procedure (Fischer (1987)). However, while in Germany monetary and income targets are defined consistently at the beginning of the planning period, during the period German monetary policy is supposed to be geared to the attainment of the monetary objective. One likely reason why nominal income targeting is not a common practice, despite its theoretical appeal (see, for example, Hall (1983), Taylor (1985), and Frankel and Chinn (1995)) is the lag with which reliable nominal income data become available even in industrial countries (Fischer (1995b)).

Second, the above classification is centered on announced targets or rules affecting monetary policy. The definition of unannounced targets—a practice followed by many central banks with respect to either exchange rates or monetary aggregates—was not considered to be relevant in this context for two reasons. First, unannounced targets cannot be monitored by the public and therefore are less binding. Second, there are theoretical reasons to believe that unannounced targets are less binding. Indeed, monetary theory has focused on the expectational effect of announcements: “tying one’s hands” without announcing it is not an appropriate policy as it forgoes the advantages of flexible response to shocks, without acquiring the advantages related to the possibility of affecting expectations (see, for example, Griffiths and Wood (1981 b)).12 Thus, unannounced targets are unlikely to be as binding as announced targets. Of course, the practice of defining unpublished monetary projections is far from useless. Many central banks use those projections to assess possible deviations of economic developments (including nonmonetary variables) from initially expected outcomes, thus deriving indications to be used to revise the setting of monetary instruments. This practice, however, corresponds to using monetary aggregates as information variables (in the sense defined by B. Friedman (1975) and (1990)), rather than as policy targets.

Evolution of Monetary Frameworks

Based on the detailed information reported in the Appendix, countries were classified by monetary framework for each year between 1970 and 1994. The framework assigned to each country in each year is the one prevailing for most of the year. The monetary frameworks listed in Table 1 are, in principle, mutually exclusive. In practice, however, there were a few cases of “mixed” frameworks, such as France during 1993, and the United Kingdom during 1991–92. In both cases, a (quasi) fixed exchange rate in the absence of capital controls was coupled with the announcements of money targets. In the presence of such inconsistent announcements (which in the case of France have been frequently discussed; see Icard (1994), Bryant (1994)), it has been assumed that the external target was the overriding one (de Larosière (1994)).

Table 2 summarizes the distribution of countries by monetary framework during that period. As the total sample includes 100 countries, the number of countries reported in each year/framework pair provides also the percentage of countries for the corresponding pair. In the tables referring separately to industrial and developing countries, percentages, rounded to unity, are reported in parentheses.

Table 2.Industrial and Developing Countries: Distribution by Monetary Framework
YearMonetary Framework
Source: Appendix: Country Data.
Source: Appendix: Country Data.

The following features stand out. At the beginning of the period, the “Bretton Woods order” is reflected in the overwhelming share of the fixed-exchange rate cum capital controls framework (framework 6) covering over one-half of the countries considered. Most other countries applied more binding frameworks, either by forsaking central bank activity altogether (regime 1–3) or by coupling the exchange rate constraint with the absence of capital controls. Only 10 countries followed a purely discretionary approach to monetary policy. It is noticeable that currency boards—a monetary framework that has attracted much attention in recent years (Liviatan (1993))—have been used in the last 25 years very rarely, although they were more common in the 1950s and 1960s, at the time when many colonies turned into independent states. The drop from 6 to 2 in the number of countries using foreign currency (first column of Table 2) also reflects the gradual introduction of own currencies by newly independent countries.

The breakdown of the Bretton Woods regime is mirrored by the decline of countries in framework 6. However, the shift, after a sharp drop of “peggers” between 1972 and 1973, is quite gradual and continues throughout the period: initially many countries replaced the Bretton Woods multilateral peg system with bilateral pegging, or, within Europe, by alternative multilateral arrangements (first the monetary “snake,” then the ERM).

The migration away from framework 6 is matched by an increase in the percentage of countries following more discretionary policies. The migration, however, follows three different phases. During the 1970s, the migration is toward both framework 8 (short-run intermediate target) and framework 9 (full discretion). Within framework 8, the announcement of monetary targets is overwhelming. Short-run exchange rate targets, in the form of an announced depreciation (or crawling peg) rate over the coming quarters, have been used in Argentina (1979–80), Brazil (1980), Chile (1978), Jamaica (1978), Portugal (1978–90), Uruguay (1979–82) and (1992–94), Mexico (1988–90), Israel (1992–94), and Colombia (1994). Unannounced crawling pegs were more common (see, for example, Williamson (1981)). During the 1980s, the number of countries using short-run intermediate targets stops increasing, while the number of countries following full discretion keeps increasing. Finally, in the 1990s, inflation targeting is virtually the only framework on the rise.

Important differences in the above trends exist between industrial and developing countries (Tables 3 and 4).13 At the beginning of the 1970s, framework 6 was the most common one for both industrial and developing countries. However, full discretion was relatively common for industrial countries even in the early 1970s (Table 3). With the breakdown of the Bretton Woods regime, industrial countries experienced a sharp but short-lived shift toward full discretion, which was, however, rapidly replaced by exchange rate targets (primarily in Europe) and by short-term monetary targets. With the beginning of the 1980s, short-term monetary targets (framework 8, whose diffusion had reached 41 percent of the sample in 1980–82) became less and less popular, with their relative share dropping to 23 percent by the early 1990s. At the same time, exchange rate targets (this time coupled with free capital mobility) peaked. Their drop in 1993–94 is related to the crisis of the ERM.14

Table 3.Industrial Countries: Distribution by Monetary Framework(Number of countries and percentage composition)
Source: Appendix: Country Data.
Source: Appendix: Country Data.
Table 4.Developing Countries: Distribution by Monetary Framework(Number of countries and percentage composition)
Source: Appendix: Country Data.
Source: Appendix: Country Data.

Developments are somewhat different for developing countries (Table 4). In the early 1970s, only a handful of developing countries (Argentina, Brazil, Korea, Lebanon, Maldives, and the Philippines, representing less than 10 percent of the total) followed a discretionary approach to monetary policy. Moreover, the shift away from framework 6 (and, more generally, from fixed exchange rate frameworks) was much more gradual than for industrial countries, as it continued at an approximately steady pace throughout the 1970s and 1980s. The announcement of short-run intermediate targets is a relatively uncommon and, if anything, new phenomenon. Inflation targeting is unknown.

Summary Indicator of the Degree of Discretion

A summary indicator of the above trends can be built by assigning to each of the nine frameworks a “weight” reflecting its degree of discretion, and by building weighted-average indexes of discretion for different country groups.

The discretion indexes in Figure 1 are built after assigning to each framework one point for each “Yes” reported in the first two columns of Table 1 and one and a half points for each “Yes” reported in the last two columns of the same table. The choice of the weight is, admittedly, largely arbitrary. However, it reflects two considerations. First, that long-term discretion involves more uncertainty on price developments than short-term discretion. Second, that the difference cannot be too large as, in many cases, the announcement of short-term targets is accompanied by an implicit commitment to monetary moderation also in the long run. In this way, long-term discretion received a higher weight than short-term discretion.

Figure 1.Monetary Discretion Index

Source: Appendix: Country Data.

The figure reports the weighted average degree of discretion by country groups using as weights the number of countries adopting each framework. In doing so, each country is treated as equally important, regardless of its population or income size. This approach was followed because we are mainly interested in studying trends in the behavior of policy authorities, rather than in deriving a “world index of monetary discretion.” However, the indexes in this figure were also recomputed by weighing each country based on relative GDP (at PPP in 1989). The income-weighted indexes move in line with those of the figure for both industrial and developing countries. Quite obviously, the income-weighted index for the whole sample follows closely the movement of the index for industrial countries (the correlation coefficient is 0.86), albeit at a higher level of discretion, reflecting the relatively higher level of discretion of the three major economies.

The figure suggests that, primarily as a consequence of the shift away from exchange rate anchors, monetary discretion has, on the average of all countries, increased in the last 25 years. The trend is, however, particularly strong in the first 17 years, while in the last period, the index seems to have stabilized. The figure also shows that, owing to the composition of the sample, the total average follows very much the movements of the developing countries’ index. The discretion index for industrial countries jumps in 1973–74, but then declines slightly through the end of the 1980s. After dropping dramatically in 1991–92 (reflecting the capital movement liberalization in European countries in the context of pegged exchange rates), it increases again in 1993–94 (ERM crisis).

Another feature highlighted by Figure 1 is that, with the exception of 1991–92, the average degree of discretion is higher in industrial than in developing countries. However, the difference seems to have narrowed significantly during the 1970s and early 1980s.

Frameworks During Disinflations

It has been argued that the choice of monetary framework is influenced by inflationary conditions. For example, Bernanke and Mishkin (1992) argue that “central bankers are more likely to adopt targets for monetary growth, or to increase their emphasis on meeting existing targets, when inflation is perceived as the number one problem.” If this were the case, it could be argued that the trend toward greater discretion may be related to the decline of inflation observed since the early 1980s, at least in the indusdiscretion may be related to the decline of inflation observed since the early 1980s, at least in the industrial world.15 To explore this possibility, we have focused on the monetary framework of countries that undertook successful stabilizations. There is a second reason to focus on these countries, namely that in these countries the formal monetary framework is likely to be taken more seriously than in countries with high inflation. For example, in countries with high inflation, the constraint imposed by exchange rate pegs may be watered down by frequent parity revisions or by increasingly tight exchange rate controls.16

In defining stabilizations, we distinguish between high inflations (hyperinflations) and moderate inflations. Successful stabilizations are respectively defined as achieving a reduction of the CPI inflation rate from above 100 percent to below 20 percent in 4 years and from above 15 percent to below 5 percent in 4 years.17

There are only a few cases of stabilizations from inflation levels over 100 percent (Table 5, on previous page), and they all concentrate in a relatively short time span, which prevents any analysis of the evolution of monetary frameworks. In four of the seven cases considered here, a discretionary frame-

Table 5.Breaking the Inflation Momentum(High level and hyperinflations)
CountryStabilization PeriodInitial InflationFinal InflationFramework
Source: Appendix: Country Data.

Framework 6 in 1992.

Source: Appendix: Country Data.

Framework 6 in 1992.

The number of stabilizations from moderate inflation is higher (Table 6) and allows the computation of a discretion index based on the same methodology followed above (Figure 2). The index is based on all-year framework pairs reported in Table 6. Each country was included in the computation only during the disinflation years. As this index is computed on a more limited sample, it has a strong erratic component (and was not computable for some periods). Nevertheless, there seems to be a clear trend toward discretion, if anything more pronounced than the one observed for the total country sample.

Figure 2.Monetary Discretion Index: Countries Under Disinflation

Source: Appendix: Country Data.

Table 6.Breaking the Inflation Momentum(Moderate inflation)
CountryStabilization PeriodInitial InflationFinal InflationFramework
Saudi Arabia1975–7823.6–1.64
Côte d’Ivoire11979–8316.35.92
Dominican Republic1980–8316.84.86
United Kingdom1980–8318.04.68
United States1980–8313.53.18
New Zealand1987–9115.72.69
Dominican Republic1990–9259.44.09
Source: Appendix: Country Data.

“Borderline case” for which the definition of “successful disinflation” repeated in the text does not strictly apply because inflation did not fall below 5 percent.

Framework 9 in 1993.

Framework 9 in 1983.

Framework 7 in 1990.

Source: Appendix: Country Data.

“Borderline case” for which the definition of “successful disinflation” repeated in the text does not strictly apply because inflation did not fall below 5 percent.

Framework 9 in 1993.

Framework 9 in 1983.

Framework 7 in 1990.

    Other Resources Citing This Publication