V. Monetary Anchors
- Carlo Cottarelli, and Curzio Giannini
- Published Date:
- December 1997
Despite the gradual shift toward discretion illustrated in the previous sections, rule-based frameworks have been very common. This section focuses on these frameworks and on their role as credibility-enhancing devices. It will be argued that the relative diffusion of (multilateral or bilateral) exchange rate pegs—the most common and binding form of rule-based frameworks—may be explained by factors that go beyond the monetary sphere and role of these pegs as credibility-enhancing devices. Cases in which rules were introduced only for monetary purposes—as in examples of monetary targeting—were not only less common, but also involved much milder forms of rules. Thus, the use of strict rules as a device to enhance credibility was less marked than the data surveyed in Section III suggested. Finally, we discuss the role of exchange rate anchors in a world of increased capital mobility.
Economic literature has repeatedly focused on the alternative between monetary anchors and exchange rate anchors. The information surveyed in Section III suggests that exchange rate anchors have been much more common than monetary anchors, even in recent years. Remarkably, the gradual shift away from exchange rate pegs highlighted in Table 2 coincided only to a very limited extent with an increased use of monetary anchors.
Figure 7 provides further information on the diffusion of monetary targets. As discussed above, monetary announcements characterize both framework 5 and most of the cases included in framework 8. The figure highlights that throughout the past 25 years, monetary targeting has been a relatively uncommon phenomenon.30 The percentage of countries announcing monetary targets has steadily increased but at a very slow pace: in 1994, only 14 countries were announcing monetary targets. This slow trend increase reflects primarily the behavior of developing countries. In these countries, the discovery of monetary targets is a relatively recent phenomenon, still on the rise, but at a very slow pace and from a very low level. The limited use of monetary targets in developing countries is sometimes not recognized. For example, the literature on stabilization policies in the Southern Cone has frequently contrasted “exchange rate based” stabilizations with “money-based” stabilizations, with Chile in 1974–75, and Argentina in 1976 often quoted as examples of the latter (Kiguel and Liviatan (1994)). However, in neither of these cases were specific money targets or target ranges announced.
Figure 7.Rise and Fall of Monetary Targets1
Source: Appendix: Country Data.
1 Percentage of countries announcing monetary targets within each country group.
In industrial countries, monetary targeting became quite popular soon after the collapse of Bretton Woods and until the early 1980s: at that time, monetary targets were announced in almost one-half of the industrial countries. But since then, the percentage has steadily declined, dropping to less than one-fourth in the mid-1990s. The current diffusion of monetary targets is overestimated by this percentage, as we include in this group all countries announcing monetary aggregates until those aggregates are officially “downgraded” through some official statements. For example, it is well-known that monetary aggregates in the United States—while still announced by the Fed in order to fulfill the requirement of the 1978 Humphrey-Hawkins Act—became de facto less and less relevant throughout the 1980s (Melton and Roley (1990), Bernanke and Mishkin (1992), Crockett (1994), Madigan (1994)). However, we have included the United States in the group of the monetary targeters until February 1993, when Chairman Greenspan in a hearing at the U.S. Senate stated that “M2 has been downgraded as a reliable indicator of financial conditions in the economy, and no single variable has yet been identified to take its place” (Greenspan (1993)). It is also noticeable that broad money targets were, altogether, almost four times more common than narrow money targets. In the sample, there are 148 pairs of year-countries in which broad money was targeted, against only 39 cases in which narrow money was targeted.
The above evidence raises two questions. First, why are monetary targets less common than exchange rate targets? Second, why are monetary targets less common in developing countries? It is worth recalling that internal monetary targeting is quite a common procedure in many central banks of developing countries (Aghevli and others (1979)). What is unusual is the announcement of monetary targets.
One explanation can be found in the literature on optimal control and information variables. Exchange rates might bear a statistically tighter relation with the final targets of monetary policy (say inflation control) than monetary aggregates. Such a feature may be particularly relevant for small developing countries, whose price level is directly affected by exchange rate movements and money velocity may be more unstable, owing to the important role played, in basic financial environments, by monetary assets as a store of value.
There is certainly some truth in this explanation. Figure 8 shows that the volatility of both narrow and broad monies is much higher in developing countries than in industrial countries (in the average of the whole period, by a factor of 1.9 and 2.5, respectively, for narrow and broad money).31 Moreover, in the average of the whole period, volatility in industrial countries is higher for narrow money than for broad money. All this is consistent with the evidence that monetary targets are more common in industrial than in developing countries, and that, in countries where monetary targeting is more common, broad money targets are more common than narrow money targets.
Figure 8.Volatility of Money Velocity
Source: Appendix: Country Data.
After pooling together all available year-framework pairs, we also computed the average volatility of broad money in countries targeting and “nontargeting” broad money: the latter exceeded the former by a factor of 2. Thus, empirical factors do seem to be important in the choice of the nominal anchor.
Yet, the above evidence does not say anything about the relative usefulness as anchor for monetary policy of exchange rates vis-à-vis monetary aggregates. It is somewhat surprising how little attention has been given to exploring systematically this issue. The adoption or rejection of monetary targets is typically premised on studies showing the stability or instability of the money demand equation.32 But the stability of the exchange rate to price relation has been subject to very limited scrutiny, despite its weak theoretical foundations. The relation is premised on the hypothesis that real exchange rates are constant or predictable, while there is ample evidence that this is not the case. Even assuming that long-run trends in the real exchange rate are relatively predictable, the short-term relation between exchange rates and prices may be unstable (as witnessed by the limited response of inflation to wide exchange rate adjustments in Europe in 1993–94). Moreover, the evidence in Figure 8 does not explain why monetary targets have gradually been abandoned in industrial countries, while they are becoming more common in developing countries. If anything, based on changes in the volatility of money velocity, the opposite could have been expected. All this suggests that other factors may be at play.
A second explanation refers to the “visibility” of exchange rate vis-à-vis monetary anchors: information on exchange rate movements is easily available every day and at virtually no cost, while monetary data may be available only with long lags and at low-frequency intervals (Aghevli, Khan, and Montiel (1991)). This explanation embodies some grain of truth, too, but its importance seems overestimated. Reliable high-frequency monetary statistics are now available in most countries, often with short lags (7–10 days).
We propose here a third explanation, namely, that in a number of cases, exchange rates are introduced and maintained because exchange rate stability is seen as a target in itself, either for economic reasons (minimizing the uncertainty related to foreign trade transactions, which are key to many developing economies) or extra-economic reasons. This hypothesis is supported by two types of evidence.
The first type of evidence refers to the historical origin of multilateral exchange agreements, such as the Bretton Woods system or the ERM. It is by now recognized that the fathers of the Bretton Woods system did not mean it to be a constraint on national monetary policies, but rather a mechanism to avoid “the chaos of the inter-war period,” including destabilizing currency speculation, beggar-my-neighbor devaluations, trade restrictions, and exchange controls (Bordo and Eichengreen (1993), Giovannini (1993), Dam (1982)). The ERM was to some extent motivated by the search for monetary stability and “convergence of economic trends” (van Ypersele and Koeune (1984)). But other factors were at play, including a direct dislike for exchange rate uncertainty and strong political goals. Giavazzi and Giovannini (1989) argue that the ERM was the product of the European dislike for exchange rate fluctuations, which they explain with (1) the openness of the European economies; (2) the belief that the floating rates of the 1920s and 1930s were responsible for the collapse of national economies and international trade; and (3) the dependence of postwar European institutions—particularly the common agricultural market—on exchange rate stability.33 In the same vein, it has been argued that the Monetary Union in Stage III of EMU has primarily a political objective, rather than an economic one.
The second type of evidence concerns the adoption of bilateral exchange rate pegs in developing countries. The empirical evidence in this area indicates that the choice between fixed and flexible exchange rates is influenced primarily by “political and external sector determinants” (Edwards (1995); see also Holden and Holden (1981)). Indeed, based on questionnaires compiled by central banks, Fry, Goodhart, and Almeida (1995) find that, in their sample of 34 developing countries, only 2 countries regarded the exchange rate peg as a nominal anchor.
The hypothesis that exchange rate pegs are adopted for reasons that go beyond the monetary sphere also explains why monetary targets have remained relatively uncommon. If the exchange rate is a target in itself, it can hardly be replaced by a monetary target. It is therefore interesting to focus on the form taken by actual rules when extramonetary factors were certainly not at play, specifically when monetary targets were used.
The specific forms taken by monetary targeting strongly suggests that this framework often represented a way of giving a rule flavor to policies that remained to a large extent discretionary, although within the constraint of achieving (more or less ambitious) inflation targets.34 In this respect, two aspects are relevant. A first aspect refers to the extent to which the attainment of monetary targets was considered binding. In most cases, it was explicitly recognized that monetary targets had to be considered not as binding constraints, but as indicative figures.35 Even the Bundesbank applied monetary targeting with a feedback mechanism, never refraining “from using new information, for fear of being caught in a credibility dilemma” (Issing (1995)). Indeed, the record of monetary targeting has been quite poor in a number of countries although the monetary targets have frequently been set in terms of, sometimes wide, target ranges (see Griffiths and Wood (1981a), Isard and Rojas-Suarez (1986), Argy, Brennan, and Stevens (1990)).
Second, in almost all cases, monetary (or credit) targets have been announced only for the short run (typically one year ahead) and not over a mediumto long-term horizon. There are two exceptions. First, in the United Kingdom in 1980, as part of the so-called Medium-Term Financial Strategy, monetary targets were announced for four years ahead (Townend (1991)). However, already in 1981, the figure announced for the years beyond the first were downgraded from targets to “illustrative ranges.” The second, and more relevant, exception is Switzerland where, as of end-1990, monetary targets have been stated in terms of average annual base growth rate over a 3–5 year period. In this case, however, no specific target was announced for the short run, thus implying that, in principle, the short-term (annual) growth rate of money remained highly uncertain. One can also note that while extreme forms of exchange rate anchors (such as freezing the exchange rate parity for an indefinite period of time) are quite common, correspondingly extreme forms of monetary anchors (such as freezing the quantity of money or its growth rate a la Friedman) are entirely unknown in practice. In many countries, more or less binding constraints have been imposed on base money growth. An example is given by legislation limiting the growth of central bank credit to the government, which in a few cases has been expressed in terms of maximum contribution to the growth rate of base money (Cottarelli (1993)). But we are not aware of countries in which the total growth rate of base money has ever been legislatively set.
The fact that money targets have been mostly announced for the short term is significant because the recent literature on monetary anchors has stressed that the main purpose of monetary announcements is to affect expectations. As most private sector contracts (including labor contracts) are forward looking, expectations are quite important in shaping price dynamics. But a large share of private sector contracts extends much beyond one year. Moreover, as monetary announcements are typically made once a year, at the moment when the announcement comes, the share of private sector contracts that has already been set may be quite large. Consequently, announcing short-term money targets does not prevent the central bank from surprising the private sector with an unanticipated inflation. Thus, the role of monetary aggregates as expectation anchor was per se limited. Therefore, we submit that the role of monetary targets was mainly to reinforce somewhat the effects of other announcements that monetary policy was turning restrictive and serious about inflation.
Pegs and Capital Mobility
Even though exchange rate pegs have often been introduced for reasons unrelated to monetary control, they have obviously acted as a constraint on monetary policy. In light of this, can they be expected to remain popular in the future as well?
As a matter of fact, Figure 9 shows that the percentage of countries announcing exchange rate pegs (including crawling pegs) has declined over the last 25 years, although the overall trend is more erratic for industrial than for developing countries. Before proceeding, it is useful to recall that the issue here discussed concerns the announcement of exchange rate rules, and not the stability of exchange rates. The latter does not necessarily require the former.
Figure 9.Percentage of Countries with Pegged Exchange Rates
Source: Appendix: Country Data.
The fall in the popularity of exchange rate pegs is correlated with the gradual removal of capital controls across the world. As shown in Figure 10, the percentage of countries with capital controls dropped from 78 percent in 1978 to 59 percent in 1994. This decline reflects primarily the removal of capital controls in industrial countries,36 while developing countries have experienced cyclical movements around what appears to be a substantially constant mean. The downward swings in these cyclical movements coincide with favorable macroeconomic conditions. Bartolini and Drazen (1995) have developed a theoretical model of capital controls explaining this feature.
Figure 10.Percentage of Countries with Capital Controls
Source: Appendix: Country Data.
To highlight better the relationship between the exchange rate regime and financial openness, Figure 11 shows the frequency of countries without capital controls within the two groups of “peggers” and “floaters.” The figure shows that free capital mobility is clearly more frequent among floaters than among peggers: on the average of the whole period, the percentage of countries with free capital mobility was about 36 percent among floaters, while it was only 23 percent among peggers. The t statistics for testing whether the two series reported in Figure 9 have the same average is 9.0, which allows us to reject the null hypothesis at substantially less than the 1 percent level. This means that free capital mobility is more likely to be associated with flexible exchange rates than with pegged exchange rates. The difference between floaters and peggers is much more marked for industrial countries (62 percent against 35 percent) than for developing countries (26 percent against 20 percent), possibly reflecting the lower de facto capital mobility in the latter, even in the absence of capital controls.37
Figure 11.Percentage of Countries Without Capital Controls
Source: Appendix: Country Data.
The figure also shows that the difference between peggers and floaters in their attitude toward capital controls (the vertical distance between the two lines) has not changed substantially throughout the last 25 years: thus, there is no evidence that it is becoming relatively easier for countries with free capital mobility to maintain exchange rate pegs. There is, however, evidence that for both groups of countries, free capital mobility is becoming more common: since the end of the 1980s, both lines in Figure 11 are upward sloping. But this is only a recent phenomenon. During the 1970s and most of the 1980s, the increase in the number of countries without capital controls had been associated with a shift of countries from the group of peggers to the group of floaters, rather than to any significant change of behavior within the two groups.
Further insight can be gained from Table 8, which reports the joint distribution of countries by exchange regime and capital mobility regulations. As the table shows, the increase in the number of countries without capital controls is entirely due to the subgroup of countries adopting flexible exchange rates. There is no trend increase in the number of peggers with free capital mobility (last row)—the number of developing countries with free capital mobility and fixed exchange rates has even declined from 13 countries in 1970 to 9 countries in 1994.
|Countries with capital controls|
|and flexible exchange rates||5||17||17||28||31||32||29||31||32||31||32||29||25|
|and pegged exchange rates||74||59||56||47||44||43||44||41||39||37||35||31||34|
|Countries without capital controls|
|and flexible exchange rates||5||8||12||14||14||15||15||16||17||17||16||25||27|
|and pegged exchange rate||61||16||15||11||11||10||12||12||12||51||17||15||14|
Number of countries in each group.
Number of countries in each group.
More specifically, in 1994, there were only 14 countries with pegged exchange rates and free capital mobility. Five of them adopted very rigid monetary frameworks to bolster the credibility of the peg. These countries include Panama and Kiribati, which do not use domestic currency; and Swaziland, Lesotho, and Argentina, which have established currency boards. The nine remaining countries include (1) five ERM countries that have maintained relatively stable exchange rates even after the widening of the ERM band (Austria, Belgium, France, Ireland, and the Netherlands; Germany is classified as a floater, because it is assumed that it acts as leader of the ERM group); (2) three oil-exporting countries (Bahrain, Oman, and Saudi Arabia); and (3) Uruguay. Thus, even including the wider-bands ERM, maintaining exchange rate pegs, with free capital mobility and in the absence of strong institutional constraints on domestic base money creation, is to be regarded as a very unusual condition.
There is an obvious reason why fixed exchange rates and free capital mobility form an unlikely couple. A necessary condition for the coexistence of free capital mobility and exchange rate pegs is that the follower country is willing to accept the automatism in monetary policy management implicit in such a rigid framework. For some countries, this automatism may be politically unacceptable (Gros (1988)); it is also suboptimal unless economic shocks are highly correlated across countries. However, the composition of the above sample suggests that the de facto acceptance of automatism may not be sufficient; rather it seems necessary to accept some form of ironclad institutional constraint, such as the adoption of currency boards or monetary unions, going beyond the mere announcement of the exchange rate parity (the relatively good holding of the ERM after the 1992–93 turbulence may indeed be explained by the prospect of the participation in the European Economic and Monetary Union, and the rules related to EMU admission). The number of countries willing to adopt such constraints seems to be limited. The above evidence may be taken as a belated vindication of the Bretton Woods consensus, which pointed to the basic incompatibility of pegged exchange rates with free capital mobility (Marston (1987), Feldstein (1993)).
The above evidence also points at one important effect of increased financial integration: when faced with the choice between maintaining monetary flexibility and pegging the exchange rate, monetary authorities around the world seem to have more and more opted for monetary flexibility. Thus, if the trend toward increased capital mobility continues, exchange rate pegs will possibly lose their role as monetary anchor that they have played in a very large share of countries even after the demise of Bretton Woods.