A comprehensive classification of monetary policy frameworks in advanced and emerging economies

Abstract

The paper presents a new classification of monetary policy frameworks that it applies to advanced and emerging economies for the period since the end of the Bretton Woods international monetary system, with a focus on the monetary authorities’ objectives (domestic and external), and on pre-announced targets and actual performance. The classification (available at www.monetaryframeworks.org) also emphasizes the underlying monetary and financial infrastructure that conditions the instruments available to the authorities and, therefore, the coherence of different frameworks. It is constructed mainly from a close reading of IMF Article IV consultation reports. The two major changes revealed by the data are the trends over time towards a heavier focus on inflation, and towards more systematic and coherent monetary arrangements.

1. Introduction

This paper represents the first outcome of a wider research project on the evolution of countries’ monetary policy frameworks since the demise of the Bretton Woods international monetary system. The term ‘monetary policy framework’ is used here to refer to the objectives and the context that condition monetary policy decisions: primarily the objectives pursued by the monetary authorities,¹ but also the set of constraints and conventions within which their monetary policy decisions are taken. The data produced by the classification, together with the ‘Individual country details’ that explain the classification for each country in each period, are available on the www.monetaryframeworks.org website.

There is a substantial existing literature (see Tavlas et al., 2008, for a survey) on the classification of exchange rate regimes, which was stimulated by (among other contributions) Calvo and Reinhart’s (2002) finding that many countries that claimed to have floating exchange rates were, in fact, intervening directly in the Forex market and/or manipulating their interest rates in order to stabilize their exchange rates. This implied that there could be substantial differences between de jure (announced; i.e. declared by member countries to the IMF) and de facto (actual, realized) exchange rate regimes, and led to the construction of new, de facto, classifications of these regimes. The two most substantial and well-known such classifications are those by Levy-Yeyati and Sturzenegger (2005; see also their 2016), who used statistical data to classify exchange rate regimes by individual years, and Reinhart and Rogoff (2004; see also Ilzetzki et al., 2019), who used both announcements and realized data (particularly parallel market exchange rate data), typically for longer periods than one year. The IMF, which used to publish purely de jure classifications, began to publish an annual de facto classification in 1998, and this was made more consistent and comprehensive with the revisions introduced by Habermeier et al. (2009).²

For monetary regimes, on the other hand, the classification and de jure/de facto issues have attracted little attention.³ Researchers have mostly relied on monetary authorities’ announcements (e.g. of monetary or inflation targets)⁴ and there are no comprehensive de facto classifications of monetary regimes. Roger (2010), for example, presents graphs showing the evolution of monetary regimes, with only four categories: inflation targets, monetary targets, managed floats and multiple targets, and exchange rate targets. The IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions now (e.g. 2014) has a finely divided de facto classification of countries’ exchange rate anchors (not unlike that of Reinhart and Rogoff, 2004), and then identifies against that (on a de jure basis) whether countries have a monetary aggregate target, an inflation targeting framework, or some other framework.

Even in the absence of the de jure/de facto issue, however, there is a need to bring together the exchange rate and (other) monetary policy elements in a single comprehensive evaluation of monetary frameworks, because of the obvious relations between them. Inflation targeting typically (and, arguably, necessarily) involves floating exchange rates while hard exchange rate pegs largely preclude the active use of interest rate or money policies, though intermediate exchange rate flexibility can coexist with some monetary autonomy (however, see also Frankel et al., 2004, and Rey, 2015).

Reinhart and Rogoff (2004) implicitly acknowledge the importance of non-exchange rate monetary elements when they distinguish ‘freely floating’ exchange rate regimes with well-organized monetary policies, such as those of the USA, Australia, and Japan, from ‘freely falling’ regimes with poorly disciplined monetary policies resulting in high inflation (40% or more is their criterion) and inevitable depreciation, such as Argentina in the 1980s. Bailliu et al. (2003) address the issue by supplementing their exchange rate classification, for countries with intermediate or floating regimes, by information on other nominal anchors for monetary policy.

The aim of this research is to construct and apply a methodology to classify countries’ monetary policy frameworks over the period since the end of the Bretton Woods international monetary arrangements in the early 1970s (the 1950s and 1960s are less interesting because, in nearly all these countries, policy was dominated by fixed exchange rate parities, and there were no instances of anything that could be called monetary or inflation targets in the modern senses). The classification will take into account both pre-announced targets for exchange rates, monetary aggregates, and inflation, and the realized values of these and other indicators. In other words, the project will bring together exchange rate and (other) monetary elements in a single comprehensive classification of monetary policy frameworks that draws on de jure (pre-announced targets) as well as de facto (realized data) information.

The classification will facilitate a more precise account of the development of monetary policy strategies than is currently available. It should also be of significant value to researchers investigating the large number of questions where there is a need for clear identification of the monetary frameworks used by different countries at different times. Several examples may be given. First, current assessments of the effects on economic performance of inflation targeting, such as those of and surveyed by Ball (2010), rely on de jure identifications, and should be checked against a de facto classification. Second, attempts to assess the reasons why some countries experienced deeper and/or longer recessions as a result of the 2007–08 financial crisis (e.g. Ólafsson and Pétursson, 2011; Cecchetti et al., 2011) have included dummy variables for inflation targeting and for hard peg exchange rate regimes; such work would benefit from a more precise and de facto classification of monetary frameworks. A third example is the research by Frappa and Mésonnier (2011) that investigated whether asset price volatility was higher under inflation targeting. A fourth example is work on the impact of exchange rate regimes on international trade in gravity models, where it would be useful to investigate the effects on trade of monetary policy frameworks overall, rather than just currency unions (as in Rose, 2000), or even a wider menu of exchange rate regimes (as in Adam and Cobham, 2007). This could, for example, shed additional light on the Calvo and Mishkin (2003) argument for policymakers to focus on financial, fiscal, and monetary institutions, rather than just exchange rate regimes. A fifth example is work on the determinants of countries’ choices of exchange rate regimes (e.g. Juhn and Mauro, 2002; Levy-Yeyati et al., 2010), where the research question can be opened out into the choice of monetary policy framework, rather than just exchange rate regime, and a precise classification of frameworks is essential.

Section 2 discusses the key principles that should underlie a comprehensive classification, and then explains the precise criteria involved in distinguishing between the different frameworks identified. Section 3 discusses the implementation of these principles and criteria. Section 4 reports the results of the classification for 27 advanced countries/currency areas and 33 emerging economies, and provides the overall findings by subperiod on the basis of two different possible aggregations of the frameworks. Section 6 concludes.

2. Key principles and their application

Monetary policy frameworks can be thought of as combinations of the objectives of the monetary authorities (including their understanding of the trade-offs between those objectives) and the set of constraints and conventions—the former more binding, the latter more matters of established usage—within which specific (conjunctural) monetary policy decisions are made. The constraints and conventions that are relevant here include the rules or disciplines to which the authorities are subject (voluntarily or involuntarily), the nature of the financial and monetary markets and institutions in existence, the understandings (on the parts of the monetary authorities and of the society) of key macroeconomic relationships, and the political environment within which the monetary authorities operate. While some frameworks completely or almost completely dictate the actions of the monetary authorities (e.g. currency boards), others (such as inflation targeting) allow varying scope for (constrained) discretion, while yet others allow even wider discretion. In any case, any given framework can be operated more or less well or badly; that is, with more or less competence and commitment on the part of the monetary authorities. Here, the aim is to define a set of monetary policy frameworks that recognizes the crucial differences—in frameworks, not in specific decisions—across countries and time periods, but still allows comparisons to be made between broadly similar cases. Factors such as the degree of capital account openness, the degree of central bank independence, the use by the monetary authorities of credit targets within a reserve money programme, or a country’s participation in an IMF-monitored stabilization package are treated as important elements of the context, but the focus for the classification is the objectives of the monetary authorities and the extent to which they are attained.

Classification is about collecting some cases in one category and other cases in other categories, so what is important is the distinctions between the various categories. In this connection, there are six major distinctions that are appropriate for countries with relatively modern, or developed, financial and monetary systems:

• Do the monetary authorities (central bank and/or ministry of finance) publish targets for some (intermediate or final) objective, or do they exercise short-term discretion over what objectives they pursue and how?

• Where such targets exist, are they for monetary aggregates, exchange rates, inflation or, indeed, some other variable?

• Where such targets exist, are they precise and narrow, or wide and broad-brush?

• Are such targets static or stationary (the same each year), or are they converging (e.g. involving an exchange rate crawl or a declining trend from high to low monetary growth or inflation)?

• Are these targets fulfilled?

• If policy is not focused on one (or more) specified and quantified objectives, is the policy framework well-defined and clearly structured (i.e. do the authorities have both a clear view of what they want and the means to achieve their various objectives)?

In addition, for countries with less-developed financial and monetary systems, particularly in earlier periods, two additional distinctions are required. First, it is useful to identify an exchange rate fix, where the exchange rate is fixed entirely by central bank action in interactions that the central bank dominates,⁵ as opposed to an exchange rate target, where there is an autonomous foreign exchange market not dominated by the central bank, where intervention in that market is only one of the tools used, and where a more or less active monetary policy is focused on controlling the exchange rate. Within the fix category, a ‘pure’ exchange rate fix, where no monetary policy instruments are deployed, can be distinguished from an ‘augmented’ exchange rate fix, where some basic instruments are used (but typically directed towards objectives other than exchange rate stabilization, in a context of limited capital mobility).

It is necessary to identify separately the case where multiple direct controls are employed, including multiple exchange rates, direct controls on bank lending, and/or administered interest rates, with no sense of monetary policy objectives and with the financial system essentially geared to the provision of finance for investments determined in a state plan. This is the monetary policy framework associated with command economies.

The approach developed here therefore starts (like the exchange rate regime classification of Reinhart and Rogoff, 2004) by asking if there is a pre-announced target for monetary policy. Here, targets that do not drive monetary policy are excluded—either because they represent government aspirations that are not internalized by the monetary authorities, or because they are merely internal projections on the part of the authorities with no element of pre-commitment or pre-announcement. Next, the approach asks what variable is being targeted: the exchange rate (where it differentiates between fixes and targets), or some monetary aggregate, or inflation;⁶ how precise that target is; and whether it is stationary or converging. The next stage is to examine whether the target is fulfilled: a framework that includes an identified target must be observed in practice as well as announced. In addition, the approach allows for combinations of (specific) objectives, distinguishing between situations where one of the two objectives is dominant and where they are equally weighted.

Where no target is announced (so that the extent to which the—presumably multiple but unquantified—objectives are met cannot be assessed), or where the announced target is clearly not attained, the analysis considers (as far as it can) the clarity of the objectives of the monetary authorities, including their perceptions of the trade-offs between them. The analysis then turns to the effectiveness of the instruments available: do the monetary authorities in some particular case have the ability to pursue serious targets? The classification therefore identifies cases of multiple direct controls as above and then distinguishes between three types of discretion: ‘unstructured’, ‘loosely structured’, and ‘well-structured’, where the triage between these three focuses on the monetary policy objectives of the monetary authorities and their instruments. ‘Well-structured discretion’ indicates that the monetary authorities have a coherent set of objectives (in the sense that they have a clear view of their preferences, on the one hand, and of the trade-offs between them, on the other), and a precise and effective set of instruments. ‘Unstructured discretion’ indicates that the priorities of the monetary authorities and the trade-offs between objectives as perceived by them are not clear, and that the instruments available to them are largely ineffective (i.e. not capable of delivering the desired outcome). Between these two categories, there is what can be called ‘loosely structured discretion’, covering cases where the instruments are effective but the objectives and trade-offs are unclear; or where the objectives and trade-offs are clear but the instruments are ineffective; or (more often) where both criteria are partly, but only partly, fulfilled. This last category covers a range of different monetary arrangements and, partly for this reason, is very common; however, there seems to be no clear principle that would make it possible to disaggregate it further.

Finally, the exercise identifies the case where a country uses another sovereign’s currency (dollarization or euroization); the case where a country has chosen to join a currency union and therefore has no national monetary policy framework of its own (so the empirical classification focuses instead on the framework of the union-level central bank); and two separate categories for currency boards. In both of the latter, the domestic currency must typically be backed 100% by foreign exchange held by the domestic currency issuer,⁷ but it is useful to differentiate between the ‘pure’ case (where a currency board operates within a very limited financial system) and the ‘augmented’ case (where there is a more developed financial and monetary system, and some monetary instruments can be deployed).⁸

Table 1 presents the full menu of monetary policy frameworks. It lists, first, the most basic frameworks: multiple direct controls, (pure and augmented) exchange rate fix, and (pure and augmented) currency board. These are followed by the four types of each of exchange rate targeting, monetary targeting, and inflation targeting, with a progression within each four from looser to full. The list is completed with the 10 possible variants of mixed targets, the three types of discretion (with a progression from looser to full) and, finally, the two cases where there is no national currency. In implementing the classification, however, some further criteria and/or definitions are needed, and those that will be used here are as follows. First, ‘full’ targeting requires that ‘narrow’ stationary announced targets are ‘typically attained’. To be narrow, targets need to be point targets or, for exchange rates, parities with margins of no more than 2.25% on either side (those are the margins set at the Smithsonian agreement at the end of the Bretton Woods period, and the narrow margins originally used within the European Monetary System). For monetary aggregates, there should be ranges of no more than 3% (which includes, for example, most German and French monetary targets but not all UK or US ones). For inflation, ranges should be no greater than 2% (the most common range for inflation targeters; see Hammond, 2012). To be typically attained, the outcome for a monetary aggregate or inflation should be within the target range over the period specified, or no more than 1% below or above the range, or, where there is a point target only, within 2% on either side of that target; slightly wider outcomes could be accepted for much higher targets (e.g. monetary targets above 10%). For exchange rate targets, the actual rate should remain within the margins specified. In addition, within a run of years in which targets are mostly attained, a single year deviation is ignored, or a longer deviation is ignored where it is clear that expectations remain anchored.⁹ The point of these relatively generous criteria is to identify the monetary policy frameworks as they are understood and as they influence both policy decisions and expectations: small occasional deviations do not compromise the perceived existence of frameworks, while large and repeated deviations do.

Loose targeting, on the other hand, requires either that narrow targets are missed by no more than 1% beyond the criteria defined above, or that wider targets (e.g. target ranges for monetary aggregates of 4%) are hit according to those criteria. Wider targets have wider ranges than narrow targets or are less clearly specified, where less clearly specified targets include, for example, definitions of price stability goals rather than inflation targets, or even cases where no precise targets are specified. In the latter case, where the monetary authorities are consistently pursuing some unannounced and unquantified target, that pursuit is likely to be identified in the sources used and its attainment can be checked, but the lack of announcement rules out full targeting.

Converging targets are those that change (in most cases, decline) regularly over time, as opposed to, for example, stationary inflation targets, which are constant over time. Finally, in the mixed target categories ‘dominant’ is decided on the basis of which of two targets of different kinds is more fully met, and ‘primacy unclear’ refers to cases where both are attained to an equal extent. The (rare) combinations of three objectives—exchange rate, monetary aggregate, and inflation—are also considered together in a single category, whichever is dominant.

Table 2 collects these criteria and definitions together for convenience, and these should be sufficient to distinguish between all the various cases of exchange rate, monetary, and inflation targeting (categories 6–27 in Table 1). However, there are a number of pairs of categories for which it is useful to indicate more clearly the basis on which they will be distinguished from each other in the implementation of the classification:

• Pure or augmented exchange rate fix versus (any form of) exchange rate targeting: The key difference is that, in the fixes, more or less all transactions involve the central bank as one of the counterparties, or are transacted at rates that the central bank sets. There is no separate or autonomous foreign exchange market where banks or other agents freely transact with each other. In exchange rate targeting, however, there is an autonomous foreign exchange market in which other agents operate and the central bank intervenes more or less frequently. The margins in fix cases are typically much narrower (e.g. 0.5% or less) than those in targeted markets.

• Pure versus augmented exchange rate fix: In both cases, the central bank fixes the exchange rate via its actions within a market that it dominates, but in PERF it deploys no other monetary policy instruments, whereas in AERF it uses from time to time some basic instruments, such as reserve requirements, typically aimed at other objectives.

• Pure or augmented exchange rate fix versus pure or augmented currency board: The key difference is that, in the currency board, all domestic currency is backed by the central bank’s holdings of foreign currency; this is therefore a more tightly structured arrangement.

• Augmented exchange rate fix versus unstructured discretion: Under the former, some basic monetary instruments are deployed, typically aimed at objectives other than the exchange rate itself, but nevertheless the exchange rate is the central concern of policy. Under the latter, the exchange rate may still be subject to a (probably varying) fix but the authorities have more concern with other objectives as well (typically economic activity or growth and inflation) and use a wider range of instruments.

• Pure versus augmented currency board: The difference lies in the extent to which other monetary policy instruments are deployed. None are in use in the first case, but a few—typically including reserve requirements, but also standing facilities and some limited possibility of lender of last resort operations—are in use in the second.¹⁰

• Multiple direct controls versus unstructured discretion: The former corresponds to a command economy, in which the financial system is merely the counterpart of the planning process. In unstructured discretion, there is some kind of an autonomous banking system that is at least partly independent of any state planning mechanism.

• Unstructured discretion versus loosely structured discretion: In the unstructured case, the monetary policy instruments are weak, and actual as well as potential fiscal dominance is common, while the monetary authorities do not have a clear idea of the priorities they should be following, or of the trade-offs between their various objectives. In the loosely structured case, the instruments are weak but the objectives are clear and coherent, or the instruments are effective but the objectives are unclear and incoherent, or the instruments are partly effective and the objectives partly clear and coherent. In loosely structured discretion, but not in unstructured discretion, there would typically be a money market and some sort of government securities market so that monetary policy is operated, at least in part, through indirect instruments. In the unstructured discretion case, fiscal dominance is the norm whereas, in the loosely structured discretion case, it is typically possible but not always realized.

• Loosely structured discretion versus mixed targets (monetary, exchange rate and/or inflation): In the former case, the objectives are unclear, or at least unquantified, but in the latter they are quantified (and met).

• Loosely structured discretion versus well-structured discretion: Under well-structured discretion, the authorities have a complete set of effective indirect policy instruments (which requires a full set of money and securities markets, and the absence of fiscal dominance), a clear ranking of their own objectives, and a full understanding of the trade-offs between those objectives.

These distinctions are summarized in Table 3. In total, there are 32 different categories in this classification. It would, of course, be possible to construct an even finer grid of frameworks, but the danger is that the classification ends up identifying every country episode separately, and no useful comparisons can be made. On the other hand, 32 categories is obviously too large for many purposes, notably for econometric work. However, the classification has been constructed, in part, with an eye to aggregation, and there are at least two useful aggregations that can be made. First, the categories can be aggregated by target variable: exchange rate (but keeping the distinction between fix and target), money, inflation, mixed targets, and different types of discretion.¹¹ Second, they can be aggregated by the degree of control of overall monetary conditions provided by the framework. In this case, the aggregation consists of ‘rudimentary’ (i.e. multiple direct controls, or pure exchange rate fix); ‘intermediate’ (e.g. augmented exchange rate fix, or unstructured discretion); ‘substantial’, where there is significant but not comprehensive monetary control (e.g. loose targeting and loosely structured discretion); and ‘intensive’, which includes full targeting and well-structured discretion. These aggregations are made precise in Table 4.

3. Implementation

While recent inflation targets are easily located from central banks’ websites, it is often more difficult to find the details of monetary and even exchange rate targets for earlier decades. However, there is a source that can be tapped for this: the IMF Article IV consultation reports, including both Staff Reports (SR) and Recent Economic Developments (RED) papers (and their successors in later years¹²), which are now available in the IMF’s electronic archives. This material is essentially real time—such consultations are held every one to two years—so that the evolution of policy frameworks can be traced as it happened and as it was seen at the time. The reports typically provide information on the aims and objectives of the monetary authorities, so that any serious and consistent pursuit of an informal (unannounced) target can be identified.¹³ They also provide information on the outcome for target variables, which is particularly important for monetary aggregates and some inflation targets where the targets are set for national definitions not covered in standard statistical sources such as the IMF’s International Financial Statistics (IFS).¹⁴ The classification process involves the examination of (parts of) 50 or so documents per country for the full 44 years.

In principle, there is a question about the independence and objectivity of these IMF reports. It could be, for example, that the IMF staff typically push the same standard policies from the same standard perspective on all countries; thus, the reports may provide a distorted view of the issues and developments concerned.¹⁵ However, the internal evidence is that the IMF has supported different monetary policies in different countries at different times, and the reports—which are negotiated and agreed with the authorities in the relevant countries—often set out the points on which IMF staff and national authorities agree and disagree.¹⁶ There is also a contrast between the monetary policy and the fiscal policy content of these reports. Regarding the latter, the IMF staff do, indeed, seem to recommend the same medicine on nearly every occasion—cuts in public spending and in the budget deficit—in a way that validates the old joke (that IMF stands for It’s Mainly Fiscal). The explanation for this difference may be that on monetary questions the IMF staff are discussing largely technical issues with central bankers who have at least some technical expertise, whereas on fiscal matters they are discussing unavoidably politicized budgetary issues with politicians who (IMF staff believe) have political axes to grind. Moreover, the IMF has a legitimate concern about fiscal dominance and the effect of deficits on monetary growth.

This material can also be supplemented with, and checked against, overviews of the development of monetary frameworks from central bank and other sources. Later perspectives sometimes provide clearer views about long-run developments, as well as revised data on targeting outcomes.¹⁷

The definitions of advanced and, even more, of emerging economies are not clear-cut, with the latter often depending on the investment opportunities identified by investment banks. Here, for want of a better principle, the groupings of advanced and emerging countries used in Laurens et al. (2009) are used throughout. These are different, for example, from the classification in the IMF’s International Financial Statistics, which appears to treat all members of the Euro area as advanced economies (but not, for example, Poland). On this basis, the paper covers 26 advanced countries plus the Euro currency area, and 33 emerging economies.¹⁸

Before reporting the overall results, an individual country illustration will be useful. Table 5 is for Turkey, which has had a number of different frameworks over the years. Table 5 and others corresponding to each of the other 60 countries are available on the www.monetaryframeworks.org website. Each table has, at the bottom, a selection of the most relevant IMF references and, in some cases, additional sources used; thus, in principle, the reader can find the rationale for the classifications.

Turkey, as can be seen from Table 5, had a long period of incoherent policy, with some improvements in its monetary instruments in the 1990s. The country then moved to inflation targets from 2002, but it struggled to meet those targets on a consistent basis and, in some years, its framework has to be reclassified as loosely structured discretion.

4. Findings

We can now consider the overall results of the classification. Table 6 shows the number and percentage of countries using each framework by subperiods, where 1974–84 can be thought of as the pre-Great Moderation subperiod. The Great Moderation itself is divided into pre-EMU (1985–98) and EMU (1999–2007) subperiods. The final subperiod is the global financial crisis (GFC) and its aftermath (2008–17). A number of comments should be made.

First, there are 6 frameworks for which no cases are recorded here (though there may be some cases when the classification is extended to developing countries): pure exchange rate fixing, pure currency board, full converging exchange rate targeting, full converging inflation targeting, monetary with inflation targeting, and inflation with monetary targeting. Second, the frameworks with the highest frequency overall are loosely structured discretion, followed by currency union membership (here, EMU), full inflation targeting, full exchange rate targeting, unstructured discretion, loose inflation targeting, and loose exchange rate targeting. Third, if we focus on the changes over time, the rise from early to later subperiods of full and also loose inflation targeting is clear, as is the varying importance of full and loose exchange rate targeting, and the relatively low frequency and then demise of (any type of) monetary targeting.¹⁹ Of the discretion frameworks, unstructured declines rapidly in the 1980s and 1990s, loosely structured rises and falls but remains important, and well-structured is found only in the third and fourth subperiods. The sharp rise in membership of currency unions in 1999 (the start of the EMU) and the frequency before 1999 of combinations of exchange rate with monetary targeting (mostly in countries that were moving towards the EMU), can also be seen.

The number of frameworks (32, or 26 if we exclude the no-shows) is too many for a useful and readable figure, but the interested reader can see a range of figures from the Visualizations page on the www.monetaryframeworks.org website. Figures 1–3, however, show the percentages of countries using each of the aggregated sets of frameworks set out in the first part of Table 4, aggregation by target variable, while Tables 7.1–3 provide the data, separating out the advanced from the emerging economies.²⁰ It should be noted that these figures exclude not only countries that do not (yet) exist (the Xs in the tables), but also countries with no national framework (UASC and CU). There is an intrinsic awkwardness here in that, from 1999, 11 countries (a number that subsequently increased) disappear from the tables and figures to be replaced by a single entry for the Euro area. One obvious way of dealing with this issue is to weight the data by GDP (or population), and this will be explored in a companion paper due to be completed shortly.²¹ The effect in the European context is greatly to increase the share of loose inflation targeting (the classification of the European Central Bank: ECB).

Table 7.1 and Fig. 1 make clear the very high degree to which (the broad category of) inflation targeting comes to dominate countries’ monetary policy frameworks by the end of the period, to the detriment of both loosely structured discretion and exchange rate targeting. Table 7.2 and Fig. 2 show that this trend is even stronger for the advanced economies. Table 7.3 and Fig. 3 show that it is also present among the emerging economies. There is a later and less strong rise in inflation targeting at the expense of loosely structured discretion and exchange rate targeting, while direct controls and exchange rate fixes disappear in the second half of the period, and there are no instances of monetary targeting.

Tables 8.1–3 and Figs 4–6 provide comparable results for the aggregation by degree of monetary control. Overall, the rudimentary category disappears (by 1991), frameworks that offer substantial monetary control rise in the second and third subperiods, and intensive control frameworks rise throughout. For the advanced economies, there are never any rudimentary control frameworks and intermediate control frameworks disappear by the early 1990s. By the fourth subperiod, intensive control frameworks account for 61% of the total, and substantial for 39%. For the emerging economies, rudimentary control frameworks (multiple direct controls) continue to exist until the early 1990s, and intermediate control frameworks remain until the early 2000s (but have reappeared in Venezuela in the last few years), while substantial and intensive control frameworks both become more important. By the fourth subperiod, intensive control frameworks account for 39% of the total and substantial control frameworks for 58%.

The main historical trends highlighted by this classification are the evolution of monetary policy frameworks over time, first, towards inflation targeting of one sort or another, and, second, towards better structured and disciplined frameworks that use more precise indirect instruments, rather than less precise direct instruments. While more work needs to be done on the determinants of countries’ choices, in broad terms these trends should probably be seen as the result of:

• increasing clarity on the part of monetary authorities (and also governments and societies) with respect to the feasible objectives of monetary policy, reflecting both experience and the changes in understanding of basic macroeconomic relationships; and

• the development of financial infrastructure (especially government securities and interbank money markets, but also central bank autonomy), which was sometimes the result of deliberate government policy designed to enable better monetary control, but sometimes more incidental.

In addition, the movement towards monetary union in Europe (in pursuit of both political and economic objectives) has had a major impact on monetary policy frameworks. This can be seen both in the efforts made by countries to qualify for the EMU (in particular, the use of hard exchange rate targets by smaller countries, and that of mixed targets by larger ones) and in the resulting replacement of individual national frameworks by the (loose inflation targeting) framework of the Euro area. Of the other frameworks discussed, successful monetary targeting of different kinds has been much less frequent in advanced economies and absent in emerging economies, while exchange rate targeting of different kinds remains important for emerging economies (and even more so for the developing economies as yet unclassified). Loosely structured discretion has been common for emerging economies, and continues to be so: this has involved different mixes of objective clarity and instrument effectiveness, with considerable variations over time in individual countries. Well-structured discretion, on the other hand, is rare, perhaps because, when countries’ monetary authorities are really clear about their objectives and in possession of fully effective instruments, they tend to opt for specified inflation targets.

It would be useful here to be able to compare these results with those of other classifications, but this is not straightforward. For the Roger (2010) graphs, there is no published database (the source is simply given as IMF), so no individual country comparisons are possible. Moreover, the graphs cover a shorter period, 1989–2006; have a coarser set of categories (only four, with everything other than monetary, exchange rate, or inflation targets put into ‘managed floating and multiple targets’); and are essentially de jure rather than de facto. They show a rise in inflation targeting in the industrial countries to the end of the 1990s, while the non-industrial countries continue to have a high proportion of exchange rate targets. But the main change in the graphs over time is the creation of the EMU in 1999, when the proportion of exchange rate targets in industrial countries drops sharply, to be offset by a large rise in managed floats and multiple targets.²²

The IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions has, as already indicated, a finer distinction between exchange rate regimes than the current classification, but it has a coarser and de jure classification of other monetary policy elements. A rough comparison can be made in terms of target variables (but not in terms of the degree of monetary control). In its 2014 edition, for example, of the 60 countries covered in the present classification, the IMF has 10 operating some kind of exchange rate stabilization; 1, China, operating monetary targets (in a crawl-like arrangement); 23 operating inflation targeting; and 25 others. The present classification has 6 operating exchange rate targeting (or fixing); 0 operating monetary targeting; 13 operating inflation targeting (some loose and some full); 10 operating discretion of different kinds; and 18 with no national monetary framework (in the EMU). Crucial differences at country level are the following: Singapore is classified by the IMF as having an exchange rate-stabilizing arrangement, whereas here it is characterized as loose inflation targeting (operated via adjustments of the exchange rate); Egypt, Croatia and Venezuela for the IMF are exchange-rate-stabilizing, but here they are classed as loosely structured discretion; China is classified here as loosely structured discretion; Turkey is an inflation targeter for the IMF, but here it is classed as loosely structured discretion because it overshot its inflation target by a wide margin (and its inflation expectations were no longer well-anchored); and the EMU members are ‘other’ for the IMF but are classed as having no national monetary framework here (while the Euro area operates loose inflation targeting). Overall, what this amounts to is that the present classification provides a more integrated and nuanced assessment of countries’ (internal and external) monetary policy frameworks, with de facto assessments of internal as well as external elements, with a finer grid of internal elements, and over a period (so far) of 44 years. The IMF classification, on the other hand, provides a finer description of exchange rate arrangements supplemented by a coarser and largely de jure description of domestic monetary policy objectives, for a single year (in fact, a single day; in this case, 30 April 2014). Some researchers will find that the IMF classification (or one of the de facto exchange rate regime classifications mentioned above) fits their requirements, but researchers who want to examine monetary policy processes and operations more widely will find the present classification more useful.

5. Conclusions

This paper represents the first outcome of a wider research project on monetary policy frameworks that will eventually encompass developing economies as well. It is innovative in its sources and in its categories, which permit a perspective that is not just finer, but also multi-dimensional, on the range of frameworks that countries have used in different periods. The classification will allow researchers to address more carefully a number of questions about the comparative performance of different frameworks, and about the choice of frameworks made by different countries in different periods.

At this stage, the work has shown how monetary policy frameworks in both advanced and emerging economies have shifted over time towards heavier emphasis on inflation targeting (with less focus on exchange rates and the demise of monetary targets), and more precise and systematic (full rather than loose) monetary strategies. Future work will try to tease out the relative importance of these trends for macroeconomic performance, and to investigate the determinants of countries’ choices of monetary policy framework.²³

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Footnotes

Acknowledgments

The author is grateful to Christopher Adam, Charles Goodhart, and Aerdt Houben, for comments on the working paper version, and to two anonymous referees for their constructive comments.

References