International Currencies and Endogenous Enforcement
An Empirical Analysis
  • 1 https://isni.org/isni/0000000404811396, International Monetary Fund

This paper investigates the determinants of the international role of a currency. It argues that standard determinants such as monetary performance and financial openness are at best imperfect indicators of a currency’s stability prospects, because the issuer’s promise of stability is not exogenously enforceable. The paper advocates an enforcement approach to international currencies that make explicit the underlying incentive incompatibilities. Additional enforcement determinants of currency internationalization are identified. The model is estimated using time-series cross-sectional analysis for three data sets. Monetary performance-related standard determinants fail to exhibit explanatory power, whereas the enforcement determinants are strongly significant and robust.

Abstract

This paper investigates the determinants of the international role of a currency. It argues that standard determinants such as monetary performance and financial openness are at best imperfect indicators of a currency’s stability prospects, because the issuer’s promise of stability is not exogenously enforceable. The paper advocates an enforcement approach to international currencies that make explicit the underlying incentive incompatibilities. Additional enforcement determinants of currency internationalization are identified. The model is estimated using time-series cross-sectional analysis for three data sets. Monetary performance-related standard determinants fail to exhibit explanatory power, whereas the enforcement determinants are strongly significant and robust.

I. Introduction

Over the past three to four decades, the world economy has witnessed a key development: the trend toward emergence of greater symmetry in relative size and economic importance among the “big three”, that is, the United States, Germany, and Japan. These shifts have been accompanied by another trend that is no less notable: shifts in the relative importance of their currencies in the world economy. As Chart 1 shows, by several indicators, the deutsche mark (DM) and the yen have been emerging as major international currencies, challenging the long-held hegemony of the U.S. dollar.

Chart 1.
Chart 1.
Chart 1.

Selected Indicators of International Currency Use

Citation: IMF Working Papers 1997, 029; 10.5089/9781451844764.001.A001

Note:The currencies are abbreviated as follows: US$ for U.S. Dollar, DM for Deutsche Mark, Pound for British Pound, F Fr for French Franc, S Fr for Swiss Franc, N F1 for Netherlands Guilder, A$ for Australian Dollar, and C$ for Canadian Dollar.

The changing landscape of the world economy has inspired a great deal of speculation as to what the future world economic system, and the future international monetary system would look like. One vision of the future entails a “tripolar” world, with the Western Hemisphere, Europe, and the Asia and Pacific region as the three main global zones of economic activity, and the U.S. dollar, the duetsche mark (or a unified European currency), and the yen as the system’s three key currencies, each currency dominating its own individual sphere of influence. 2/ Skeptics, however, point to the potential obstacles that underlie European unification and the formation of a single European currency, or the limited extent of Japanese integration with Asia. This debate notwithstanding, observed shifts in the pattern of global influence have sparked renewed interest, both theoretical and empirical, in the topic of international currencies.

Key currencies, or perhaps more generally, international currencies, perform the important role of international money. 3/ By acting as a medium of exchange, unit of account, and/or store of value beyond the borders of the nation that issues it, an international money provides world liquidity and facilitates smooth functioning of the international payments mechanism. More importantly, an international currency entails, by virtue of that status, a certain degree of power and privilege for the issuer nation. When its domestic currency is held by foreigners, the foreigners in effect extend credit to the issuer nation. Moreover, this credit is “near free” in the sense that as long as these assets (foreign balances denominated in the home currency) are not withdrawn, the issuing country can use these funds almost free of charge (Grubel 1964). 4/ The issuer can obtain a kind of a free command over real resources and can use them to enlarge its investment at home or abroad, or consume them (Cohen 1971, Grubel 1964, Tavlas 1991). To put it differently, the privilege lies in the relaxation of the issuer’s balance of payments constraint. Additionally, a key currency country has the ability to wield global influence, because its currency is used as international money. Its policies can have profound implications for the world economy, from creation of a deflationary or inflationary impact (liquidity crises), to disruption of trade, finance, and systemic stability. It matters therefore, currencies of which countries are awarded international status by the international community, and why. 5/

The determinants of the international role of a currency are the subject of investigation in this paper. Specifically, what enables the national currency of a country to serve as international money in the world economy? The answer to this question is critical in shaping our understanding and interpretation of past, present, and future trends in the evolution of international currencies. The standard literature to date on the subject has identified a variety of factors, in the form of country and currency characteristics, that determine a national currency’s international comparative advantage. These determinants (hereafter, “standard” determinants) are inflation variability, exchange rate variability, inflation rate, exchange rate, financial openness, world export share, depth and breadth of the financial market, and net foreign asset position. The levels and variability of inflation and exchange rates relate to monetary performance of the issuing country, and along with financial openness, are broadly referred to in this paper as “currency stability criteria”.

The thrust of this paper’s critique of the standard approach to international currencies relates to the role of currency stability criteria as determinants of internationalization. Implicit in this approach is a view of agents who base their future expectations about the transaction costs associated with a currency, and thereby, decisions about holding and use of that currency, on the available evidence regarding currency stability (monetary performance and financial openness). 6/ In effect, they presume that future currency stability will mirror the past performance of the currency. Although at first glance it would seem that such a presumption is consistent with rational expectations formation, the paper argues that in fact this cannot be the case. The central argument in this paper is that there are certain incentive incompatibilities in the relationship between a key currency issuer and foreign holders of its currency, and when such incentive incompatibilities are recognized, the premise that a history of good performance (currency stability) is a reliable indicator of future performance is rendered naive. Indeed, once the issuer’s currency is held as international money in sizable amounts, the issuer has an incentive ex post to deviate from its previously established path of stability. That the issuer will not be tempted to do so cannot be guaranteed ex ante; actions of the issuer are not exogenously enforceable. It follows that in such a situation, the mere evidence of stability in the past cannot be taken as a reasonable indicator of future stability prospects.

The paper thus introduces an element of novelty into the standard literature on international currencies, by advocating a theoretical framework that makes explicit these incentive incompatibilities and the potential for strategic acts of opportunism on the part of the issuer. The analysis in this paper is an attempt to situate the issue of determinants of international currency status within a growing body of literature on exchange relations called the enforcement literature. This literature has shown that when contracts, implicit or explicit are exogenously (third-party) unenforceable, endogenous enforcement must regulate the exchange -- parties to the exchange themselves must adopt or institute certain enforcement activities such as surveillance and sanctioning mechanisms that alleviate the incentive incompatibility and elicit desired levels of the attribute in question. The paper explores these endogenous enforcement strategies in the context of the international money issuer. It identifies four observable characteristics of the issuer nation as “enforcement” determinants of the currency’s international role: stock of foreign direct investment assets abroad, multinational rent earnings from abroad, central bank independence, and military strength. These characteristics embody information about the extent to which the issuer is committed against imprudence. They reflect the credibility of the issuer’s promise of stability and signal what future prospects for currency stability might be expected.

If agents are rational and aware of the underlying incentive incompatibilities, then, when forming future expectations about the transaction costs associated with the currency and assessing its desirability, they should invoke, in addition to the observed historical record of monetary performance and financial openness, consideration of these enforcement determinants. The enforcement approach to international currencies is thus consistent with a broader concept of rational expectations, whereby expectations about future currency stability are informed by the record of past stability, as well as by considerations of the likelihood that the issuer might engage in strategic opportunism in the future.

The paper estimates empirically a model of the determinants of international currency status, whereby standard determinants along with enforcement determinants constitute the complete set of factors that influence the international role of a currency. This analysis is a significant contribution to the literature in that it represents the first empirical effort to test the theory of determination of currency internationalization. The model is estimated using time-series cross-sectional analysis for three data sets as samples, and two different measures of the international role of a currency. The empirical results indicate that the enforcement determinants are strongly significant and robust in explaining international currency share. None of the standard determinants related to monetary performance has any explanatory power. The performance of the rest of the standard determinants can be characterized as mixed. The empirical findings demonstrate the validity of the endogenous enforcement approach as a theoretical framework for analyzing the issue of determinants of international currency status. In particular, they support the proposition that the enforcement determinants are better indicators of a currency’s international role than the standard criteria commonly associated with currency stability.

The remainder of this paper is organized as follows. Section II presents the theoretical rationale underlying the standard determinants, and critiques the role of currency stability criteria in this approach. Section III introduces the endogenous enforcement perspective and identifies the so-called enforcement determinants. Section IV estimates the model empirically, and discusses results. Section V draws implications and concludes the paper.,

II. Theoretical Considerations: The Standard Approach to International Currencies

1. Standard determinants

Determinants of the international role of a currency have been well documented in the literature, so their discussion will be kept brief. 7/ All together, these constitute a set of country and currency characteristics that relate to one of three general categories: (1) monetary performance, (2) trade patterns, and (3) financial market comparative advantage and international financial intermediation. Inflation and exchange rate levels, and inflation and exchange rate variability are the determinants relating to monetary performance of the issuing country. Share of world exports is a determinant relating to trade patterns of the issuing country. Financial market depth and breadth, financial openness, and net foreign asset position are the determinants relating to financial market comparative advantage and international financial market intermediation. The theoretical rationale underlying each of these factors is discussed next.

The basic framework of analysis is one where agents are assumed to use and hold those particular international currencies that offer relatively low costs of information (transaction) associated with their use. Currencies that are relatively stable in external and internal value, as well as offer relative certainty with respect to financial openness, that is, a low risk of exchange or capital controls, imply a low degree of uncertainty associated with their use, and thus, low transaction costs. The price of an international currency plays an important role in disseminating information, and instability of value distorts the ability to embody and provide sufficient information to transactors, making it necessary for them to undertake costly investigation. Likewise, excessive regulations on financial transactions raise transaction costs by impairing quick movements in and out of foreign currency assets, thereby increasing the risk factor and uncertainty associated with international use of a currency. They also increase transaction costs by restricting foreign access to domestic financial markets and reducing foreign opportunities to hold claims denominated in the currency, and by restricting domestic financial institutions from being competitive offshore. High inflation rates distort movements in relative prices through which market information is transmitted, and create uncertainty. A weak exchange rate implies an eroded or diminished purchasing power over other currencies and increases the costs associated with a currency.

Besides the above-mentioned currency stability criteria, another important indicator of the transaction costs associated with a currency is the extent of the currency’s use in world trade and payments. Currencies of countries that are predominant in world trade and payments, that is, countries with large shares of world exports, offer large markets in those currencies. The existence of a large market in a currency implies greater familiarity with the currency, and lower uncertainty and search costs.

The depth and breadth of the financial market is another factor relating to transaction costs. The breadth of a financial market refers to a large assortment of financial instruments traded, while the depth refers to the existence of well-developed securities markets. Financial markets that have breadth and depth offer the most capital certainty for their currencies. In other words, from the point of view of a risk-averse investor, the risk of capital loss on the sale of an asset is smaller in a broad and deep market than in a thinner market. There are two main reasons: first, an economic agent acting alone is less likely to have an influence on asset prices in a market that is deep and broad, and second, an exogenous disturbance is likely to induce greater price variations in a market that lacks depth and breadth. Thus, broad and deep markets tend to be resilient, and by offering the advantage of capital certainty for assets denominated in that currency, they lower overall transaction costs of that currency. They also lower costs by offering international holders a stable and diverse supply of assets.

A final factor relating to transaction costs is the net foreign asset position of the currency issuer. 8/ It has been argued that a key currency nation is, in effect, a world banker or an international financial intermediary. 9/ When foreigners hold short-term claims denominated in its currency, for investment or transaction purposes, they supply short-term liquid capital (on net) in the currency of that country. The ability of the issuer to attract short-term liquid capital or liquid deposits denominated in its own currency enables it to play the role of a world banker, supplying long-term capital in the form of loans and investments denominated in its own currency to the rest of the world. Net debtor or net creditor status signals the financial health of the intermediary. A net debtor position, for example, jeopardizes the functioning of the world banker, and by signaling solvency risk, it creates uncertainty and raises transaction costs associated with the currency.

In sum, the various factors highlighted above are indicators of the transaction costs associated with a currency’s international use. Potential international holders, assumed to be seeking to minimize their transaction costs (both present and future expected) of holding and using international currencies, thus consider these factors in assessing the desirability of a currency.

2. The role of currency stability criteria

As the previous section indicated, the importance of currency stability has been duly recognized in the existing literature. It is argued (correctly) that national currencies possess international comparative advantage in the form of low costs of information if they have a protracted record of stability in the sense of a low degree of inflation and exchange rate variability, and a consistent record of a high degree of financial openness. Furthermore, it is implied that costs -- not just present but future expected as well -- to international money holders and users will rise if the value of a currency, internal or external, fluctuates considerably, or if financial restrictions such as exchange controls are imposed. It follows that if a currency in international use begins to display unsatisfactory performance in the sense of poor stability, its international attractiveness and therefore its international use will be undermined.

A careful evaluation of this argument shows that it appears to rely on a key premise -- that the actual observed evidence with respect to currency stability is a reasonable indicator of the currency’s future stability. Implicit is a view of agents who base their future expectations about the transaction costs associated with a currency, and thereby decisions about holding and use of that currency, on the available evidence regarding monetary performance and financial openness. It would seem that reliance on a history of good performance (currency stability) in the past as a reasonable indicator for the future conforms to a rational expectations formation process. The paper takes issue with this point. It notes that there exist potential incentive incompatibilities (agency problems) in the relationship between the issuer of international money and foreign holders of its currency, which the standard approach to international currencies fails to account for. Once these incentive incompatibilities are explicitly recognized, it is no longer rational to rely solely on the observed record of past stability.

Briefly, the underlying incentive incompatibility is this. The foreigner holds some amount of the issuer’s currency, expecting the issuer to maintain stability, that is, to uphold the real value of his holdings through good monetary performance, as well as to maintain financial openness. However, the issuer can at best offer only an exogenously unenforceable promise to do so. Indeed, once its currency is held as international money in sizable amounts, the issuer has an incentive ex post to renege on the implied contract, in other words, to deviate from its past record of stability. A strategy of unanticipated inflation, for example, enables a country with an international currency to inflate away an arbitrarily large portion of the real purchasing power represented by its nominal debt (Tavlas 1991), resulting in large-scale wealth transfers from holders to the issuer. Foreigners are not protected against such opportunism or imprudence on the part of the issuer, because they can only hedge against the anticipated component of inflation. Likewise, policy surprises such as financial regulations and exchange and capital controls that restrict financial openness can also mean substantial opportunistic gains for the issuer in the form of increased flexibility in pursuing domestic macroeconomic objectives, but they impose considerable costs on the foreign holders by limiting their freedom and flexibility in moving their funds. 10/ If such policies are surprises and therefore not foreseeable, there is little foreign holders can do, ex ante, or ex post, to hedge against these losses. 11/

If the underlying model is assumed to be one where all agents are rational and understand the incentive incompatibility, then they also understand that in such a situation, a promise is only a promise. That the issuer might renege cannot be precluded ex ante; so the observed evidence on monetary performance and financial openness is at best an imperfect indicator of future performance. Thus, enforcement concerns are paramount. In the absence of exogenous enforcement, how then is prudent (non-opportunistic) performance on the part of the currency issuer ensured?

In recent years, an entire body of literature 12/ has emerged that makes explicit the distinction between the class of economic exchanges that involve comprehensive exogenous enforcement of contracts (implicit and explicit), and the class of economic transactions that does not. 13/ This literature (enforcement literature) holds that in situations not amenable to exogenous enforcement, endogenous enforcement is operative. Endogenous enforcement is said to occur when parties to the exchange themselves adopt or institute certain enforcement activities such as surveillance and sanctioning mechanisms that regulate the exchange and elicit desired levels of the attributes in question. These endogenous enforcement mechanisms may constitute a combination of penalties and incentives that reduce the incentive incompatibilities and thereby endogenously enforce the exchange relation.

Collateralization is a widely known endogenous enforcement mechanism in this literature. It is said to ensue when both parties somehow come to share a stake in satisfactory performance by the potentially belligerent party. 14/ When agents are ensured that their exchange partners have too great a stake in the satisfactory outcome of a transaction to fail to live up to its explicit or implied conditions, the relationship is endogenously enforced. Collateralization, as an endogenous enforcement mechanism in problems of agency, is in fact a derivative of a broader concept called pre-commitment. 15/ In many social and economic situations, threats and/or promises are invoked by one party or the other. But a threat or promise must be credible in order to have its intended effect. Pre-commitment is a technique for lending credibility to both threats and promises. Schelling (1956, 1960), who pioneered the work on pre-commitment, notes that the essence of such a class of tactics as pre-commitment lies in some voluntary but irreversible sacrifice of freedom of choice. There are several ways to pre-commit, that is, to tie one’s hands towards or against a particular action. Pledging one’s reputation, hostage-offering, undertaking specific costly social investments, and volunteering of strategic information are but a few of the many specific acts of commitment. 16/ More generally, these are acts that generate a stake in prudent performance by making the belligerent party vulnerable either in terms of punitive sanctions initiated by the disgruntled party, or in terms of some self-imposed institutional constraints that render imprudence costly. The next section explores various pre-commitment strategies in the context of the relationship between the international currency issuer and foreign holders of its currency.

III. Endogenous Enforcement in the Case of Key Currencies

Applying the logic laid out above, an international currency issuer may be understood as appropriately collateralized if it demonstrates a stake in prudent performance. This may be achieved in two ways: via relative exposure to foreign retaliation in the form of punitive sanctions, and/or via relative exposure to internal rigidities and resistance to imprudent behavior. The former may be interpreted somewhat loosely, as the “external tying of hands”, and the latter as the “internal tying of hands”. It is important to acknowledge that this is only a crude distinction, because strictly speaking, even the external forces (foreign sanctions) that may tie the issuer’s hands against opportunism might ultimately bind the issuer via an internal channel, that is, by impacting adversely on some key interest group within the key currency nation. Nevertheless, the distinction is useful for two reasons. First, it helps to point out that the international community can, by initiating external sanctions, play a role in enforcing prudent performance on the part of the issuer. Second, it helps to underscore the point that even if foreign punitive retaliation might not be forthcoming (for example, if foreigners lack the ability or the willingness to impose sanctions), internal rigidities within the issuer nation could, on their own, enforce prudent behavior. Howsoever its hands may be tied, ultimately the important point from an enforcement perspective is for the issuer to demonstrate a stake in maintaining prudence. Such an international currency issuer is collateralized, committed, and credible.

Drawing on the accumulated literature on enforcement and pre-commitment, several factors in the form of issuer nation characteristics can be identified as potential commitment devices. These are discussed next.

1. Enforcement determinants

One characteristic of the issuer nation that helps to generate collateralization is the physical presence abroad of a large stock of foreign direct investment assets. Taking a cue from the institution of hostages (see Section II), we conjecture that this stock serves the important function of a hostage. The hostage, by exposing the issuer nation to potential costs in the form of foreign retaliation (hostage/asset appropriation), ties the issuer’s hands against imprudent actions. 17/ An international currency issuer with a larger stock of such exposed assets is, in effect, more heavily collateralized (compared with one with a smaller stock) since it has a larger stake in maintaining satisfactory performance.

Another characteristic of the issuer nation that reduces the incentive incompatibility between the parties is the accrual of sizable economic rents to the key currency nation’s multinationals from their trading and financial activities abroad. This represents another dimension of exposure -- in the event that the issuer nation’s performance is deemed unsatisfactory, foreigners could take aim at this exposed target by imposing trade and financial sanctions that would restrict the multinationals’ access to valuable foreign markets and associated opportunities. Since their economic futures are tightly linked with an open international economy, such rent-earning multinationals could serve as an effective hedge against imprudent policy actions on the part of their government. 18/

Still another feature can be identified by recognizing the importance of an exposed, that is, a pledged reputation, as a commitment strategy. Besides the seizure of foreign direct investment assets as hostages, and the imposition of trade and financial sanctions on multinationals, foreign retaliation can take another potentially potent form -- censure of the issuer nation. 19/ But loss of prestige is a deterrent only if the issuer is appropriately collateralized, that is, demonstrates a strong interest in preserving its reputation. It follows, therefore, that a heavily collateralized issuer must be a nation that has significantly “invested” in its reputation, for example, by assuming an international leadership position in issues of world peace, conflict, and security. In general, it would seem that military strength is a prerequisite for effective international leadership. The exercise of military power -- in the form of military threats as well as military favors and defense umbrellas -- is indeed part and parcel of successful international political diplomacy. Military strength can thus impart to a key currency nation, a vital resource necessary for it to assert international dominance and earn international stature. At the same time, it ties the issuer’s hands against imprudence by exposing it to potential retaliation in the form of adverse reputation effects. 20/

The broad characteristics discussed above are features of the key currency nation that reflect the extent of its exposure to external sanctions, that is, sanctions potentially initiated by foreigners. Additional characteristics may be identified by focusing internally, that is, by exploring within the key currency nation potential domestic rigidities that serve to constrain its actions. These rigidities would deter opportunism even if foreign sanctions were absent. In effect, they act as internal sanctions.

The commitment literature has shown that in some situations, the potentially belligerent party can earn credibility by investing in certain social institutions that restrain its behavior. Independent central banks are precisely such institutions. An independent central bank represents a structural impediment to unanticipated inflation. Typically, it is the fiscal arm of the government that is faced with the temptation of engaging in short-term wealth transfers through unanticipated inflation. Therefore, an institutional structure that severs federal government control over monetary policy and concentrates it within an independent body with an explicitly stated mandate for price stability can be an extremely effective commitment device. 21/

It might seem, on the surface at least, that the deterrent effect of independent central banks emanates from the mere mechanical separation of monetary and fiscal authority in an economy. However, the role played by central bank independence in generating collateralization can only be understood well if one delves into the sources of independent central bank preferences. Several researchers in the political-economy tradition have suggested that the independent central bank preferences for price stability and financial openness are in fact a reflection of the preferences of the central bank’s primary political constituency: the domestic (typically also multinational) financial sector. 22/ The financial community values domestic price stability in the interest of protecting the real value of its assets denominated in the domestic currency. Likewise, this sector maximizes its gains in a steady policy environment of unregulated financial activities at home and abroad. From the point of view of international money holders, therefore, the existence of an independent central bank signals a low propensity or predisposition to imprudence (unanticipated inflation and/or arbitrary financial regulation). Additionally, if an independent central bank commands a strong international reputation (as does the Bundesbank), fear of adverse reputation effects from abroad may serve as a deterrent to deviation from the banks’ publicized preferences and mandates.

Besides an independent central bank, other social institutions internal to a key currency nation are also significant in that they have the capacity to restrain its behavior. For example, a well-established stable democratic political system and a well-developed and efficient legal infrastructure ensure sufficient domestic restraint and minimize the possibility of abrupt radical changes in policies.

In sum, several broad characteristics of the issuer nation have been identified as potential commitment devices: large stock of foreign direct investment assets situated abroad, accrual of sizable economic rents to multinationals from their trading and financial activities abroad, high international profile, independence of the central bank, a well-established stable democratic political system, a well-developed legal infrastructure, and military strength. For purposes of empirical analysis, these features are narrowed down to four specific quantifiable indicators, called“enforcement”determinants of the international role of a currency: stock of foreign direct investment assets abroad, multinational rent earnings from abroad, central bank independence, and military strength.

The enforcement determinants reflect the extent to which the issuer is collateralized against imprudence, or in other words, the extent to which the incentive incompatibilities are alleviated. Rational agents, who are assumed to be aware of the incentive incompatibilities, therefore invoke -- in addition to the observed historical record of monetary performance and financial openness (currency stability) -- consideration of enforcement determinants, when forming future expectations about the transaction costs associated with the currency and assessing its desirability. Moreover, the enforcement determinants are expected to embody greater and more meaningful information content about future currency stability prospects than what is contained in the mere statistical record of past performance. This so, because they are derived from an in-depth consideration of the issuer’s incentives and constraints in reneging on the implied contract.

IV. Empirical Analysis

1. An econometric model of the determinants of currency internationalization

This section estimates empirically a model of the determinants of the international role of a currency. Standard determinants (Section II), in conjunction with enforcement determinants (Section III), constitute all the factors that influence international holders’ decisions to hold and use various international currencies. Estimation of the model reveals the magnitude and direction of the separate effects of various determinants on currency internationalization. In particular, it enables an explicit comparison between the standard determinants and the enforcement determinants in terms of relative explanatory power.

The specification of the international role of a currency, as well as of its determinants, in the form of econometric variables was necessarily informed by several considerations, both theoretical and practical. Owing to limitations of space, discussion of data issues and variable selection is confined to Appendix I. Table 1 summarizes the definitions (measures) of all the independent variables, their expected signs, and data sources. Two different definitions (measures) of the international role of a currency were used -- the currency’s share in international bond issues, and its share in world official foreign exchange reserves. Note that one definition -- currency composition of foreign exchange reserves -- relates to the currency’s official role and is a stock-based measure, while the other -- currency composition of international bond issues -- is a flow measure and reflects the private aspect of a currency’s international role. Admittedly, each series captures a different (disaggregated) aspect of a currency’s international role, and perhaps ideally different (sub)sets of determinants should explain these different aspects of a currency’s international role; however, it is important to keep in mind that for the purposes of this econometric study, the sense in which we use these two alternative definitions is ultimately only to serve as potential indicators (proxies) of the overall international role of the currency.

Table 1.

Independent Variables: Measures and Sources

article image

As Table 2 shows, the availability of data for the two series varied over time and across countries; the sample size varied accordingly in each case. The data were sorted into three samples.

Table 2.

Dependent Variable: Measures and Sources

article image

The model was specified as a pooled time-series cross-sectional equation. 23/ The equation was estimated for each of the three samples, using Generalized Least Squares (GLS). 24/

CURSHi,t=β0+β1INFLi,t+β2INFLVi,t+β3EXRi,t+β4EXRVi,t+β5EXPSHi,t+β6KKIi,t+β7FINSHi,t+β8NFARi,t+β9CBIIi,t+β10FDISHi,t+β11NFISHi,t+β12MILSHi,t+ui,t(1)

t= 1,…, T

i= 1,…, N

where

CURSHi, t = Share of currency i at time t, relative to world

INFLi, t = Inflation rate of country i at time t, relative to the average for N countries

INFLVi, t = Inflation variability of country i at time t, relative to the average for N countries

EXRi, t = Exchange rate of country i at time t

EXRVi, t = Exchange rate variability of country i at time t, relative to the average for N countries

EXPSHi, t = Export share of country i at time t, relative to world

KKIi, t = Capital controls index for country i at time t

FINSHi, t = Financial market share of country i at time t, relative to world

NFARi, t = Net foreign assets as a share of GNP for country i at time t, relative to the average for N countries

CBIIi, t = Central bank independence index for country i at time t

FDISHi, t = Foreign direct investment assets stock share of country i at time t, relative to world

NFISHi, t = Net factor income share of country i at time, relative to N countries

MILSHi, t = Military expenditure share of country i at time t, relative to world

β’s are parameters to be estimated, and u is the stochastic error term. Note that T and N vary by data set.

2. Estimation results

Results for SAMPLE1 are discussed first. Table 3 presents GLS estimates for equation (1). As the table shows, the variables Milsh, Fdish, Cbii, and Expsh exhibited strong t-ratios. None of the monetary performance variables Infl, Inflv, Exr, and Exrv performed well as explanatory variables. Finsh, Nfar, and Kki were not statistically significant, while Nfish was significant in only one instance. These results are indicative of the strong performance of the enforcement variables, individually, as explanatory variables. To test their joint significance, the enforcement variables were next restricted to zero, and equation (1) was re-estimated. At the 5 percent level of significance, an F-test comparing the restricted equation with any of the versions of equation (1) presented in Table 3 rejected the joint hypothesis that the true coefficients for the enforcement variables were all zero. 25/

Table 3.

GLS Estimatesa: SAMPLE1 (1964–90, Six Countries, Dependent Variable: Currency Share in International Bond Issues), Equation (1)

article image
Note:t-statistics in parentheses, *significant at the 10 percent level, **significant at the 5 percent level or less aCorrections for both autocorrelation and heteroscedasticity.I Cbii measured as Annual Turnover Rate of Central Bank Governor.II Cbii measured as Central Bank Legal Independence Index.III Cbii measured as Epstein-Schor Central Bank Independence Index.

Given that, in the first round of estimation above, neither the inflation- nor the exchange rate-based variables proved statistically significant, in subsequent stages of the empirical analysis, estimation was restricted to two alternative versions of equation (1). In these versions, the inflation-based and the exchange rate-based variables were omitted in turn so as to reduce the number of explanatory variables, increase the degrees of freedom, and make the results somewhat more tractable. 26/

CURSHi,t=α0+α1INFLi,t+α2INFLVi,t+α3EXPSHi,t+α4KKIi,t+α5FINSHi,t+α6NFARi,t+α7CBIIi,t+α8FDISHi,t+α9NFISHi,t+α10MILSHi,t+ui,t(2)
CURSHi,t=λ0+λ1EXRi,t+λ2EXRVi,t+λ3EXPSHi,t+λ4KKIi,t+λ5FINSHi,t+λ6NFARi,t+λ7CBIIi,t+λ8FDISHi,t+λ9NFISHi,t+λ10MILSHi,t+ui,t(3)

t = 1,…, T

i = 1,…, N

α’s and λ’s are parameters to be estimated, and u is the stochastic error term.

Tables 5 and 6 present for SAMPLE1 the results of GLS estimation for equations (2) and (3), respectively. As the tables indicate, regression results for (2) and (3) were very similar to those of (1) estimated earlier. Moreover, there did not appear to be much difference in the outcomes of the inflation and the exchange rate versions. By and large, Milsh, Fdish, Cbii, and Expsh were significant in both equations. Similarly, Kki and Nfar were not statistically significant in either equation. In both equations, the performance of Finsh and Nfish was poor. Infl and Inflv did not perform well in equation (2): Exr and Exrv performed poorly in equation (3).

Table 4.

GLS Estimatesa: SAMPLEI (1964–90, Six Countries, Dependent Variable: Currency Share in International Bond Issues), Equation (1) with Exchange Rate Regime Dummy

article image
Note:t-statistics in parentheses, *significant at the 10 percent level, **significant at the 5 percent level or less.

Corrections for both autocorrelation and heteroscedasticity.

Exchange Rate Regime Dummy: 1 if ≥ 1972, 0 otherwise.

I Cbii measured as Annual Turnover Rate of Central Bank Governor.II Cbii measured as Central Bank Legal Independence Index.III Cbii measured as Epstein-Schor Central Bank Independence Index.
Table 5.

GLS Estimatesa: SAMPLE1 (1964–90, Six Countries, Dependent Variable: Currency Share in International Bond Issues), Equation (2)

article image
Note:t-statistics in parentheses, *significant at the 10 percent level, **significant at the 5 percent level or less.

Corrections for both autocorrelation and heteroscedasticity.

I Cbii measured as Annual Turnover Rate of Central Bank Governor.II Cbii measured as Central Bank Legal Independence Index.III Cbii measured as Epstein-Schor Central Bank Independence Index.
Table 6.

GLS Estimatesa: SAMBLE1 (1964–90, Six Countries, Dependent Variable: Currency Share in International Bond Issues), Equation (3)

article image
Note:t-statistics in parentheses, *significant at the 10 percent level, **significant at the 5 percent level or less

Corrections for both autocorrelation and heteroscedasticity.

I Cbii measured as Annual Turnover Rate of Central Bank Governor.II Cbii measured as Central Bank Legal Independence Index.III Cbii measured as Epstein-Schor Central Bank Independence Index.

Equations (2) and (3) were also estimated for two of the remaining samples in turn. The results are summarized in Tables 7 and 8 (SAMPLE2), and Tables 9 and 10 (SAMPLE3).

Table 7.

GLS Estimatesa: SAMPLE2 (1973–90, Nine Countries, Dependent Variable: Currency Share in International Bond Issues), Equation (2)

article image
Note:t-statistics in parentheses, *significant at the 10 percent level, **significant at the 5 percent level or less.

Corrections for both autocorrelation and heteroscedasticity.

I Cbii measured as Annual Turnover Rate of Central Bank Governor.II Cbii measured as Central Bank Legal Independence Index.III Cbii measured as Epstein-Schor Central Bank Independence Index.
Table 8.

GLS Estimatesa: SAMPLE2 (1973–90, Nine Countries, Dependent Variable: Currency Share in International Bond Issues), Equation (3)

article image
Note:t-statistics in parentheses, *significant at the 10 percent level, **significant at the 5 percent level or less.

Corrections for both autocorrelation and heteroscedasticity.

Market Exchange Rate-based measures.

Effective Exchange Rate-based measures.

I Cbii measured as Annual Turnover Rate of Central Bank Governor.II Cbii measured as Central Bank Legal Independence Index.III Cbii measured as Epstein-Schor Central Bank Independence Index.
Table 9.

GLS Estimatesa: SAMPLE3 (1973–90, Nine Countries, Dependent Variable: Currency Share in Official Foreign Exchange Reserves), Equation (2)

article image
Note:t-statistics in parentheses, *significant at the 10 percent level, **significant at the 5 percent level or less.

Corrections for both autocorrelation and heteroscedasticity.

I Cbii measured as Annual Turnover Rate of Central Bank Governor.II Cbii measured as Central Bank Legal Independence Index.III Cbii measured as Epstein-Schor Central Bank Independence Index.
Table 10.

GLS Estimatesa: SAMPLE3 (1973–90, Nine Countries, Dependent Variable: Currency Share in Official Foreign Exchange Reserves), Equation (3)

article image
Note:t-statistics in parentheses, *significant at the 10 percent level, **significant at the 5 percent level or less.

Corrections for both autocorrelation and heteroscedasticity.

Market Exchange Rate-based measures.

Effective Exchange Rate-based measures.

I Cbii measured as Annual Turnover Rate of Central Bank Governor.II Cbii measured as Central Bank Legal Independence Index.III Cbii measured as Epstein-Schor Central Bank Independence Index.

3. Summary of main findings and final regressions

The main findings from the empirical analysis may be summarized as follows. Overall, the results were fairly stable across the three samples estimated in this study, thus indicating that the two different measures that served as alternative specifications of the international role of a currency are fairly close substitutes and could be treated as such in future empirical research as well.

The most striking finding of the econometric analysis was that all but one of the enforcement variables corresponding to the enforcement determinants proved to be strongly significant and robust in explaining international currency share, with signs as expected. Three enforcement variables -- Milsh, Fdish, and Cbii -- repeated their strong performance as explanatory variables consistently across all three samples. The fourth enforcement variable, Nfish, achieved significance in all SAMPLE3 regressions, and occasionally in the SAMPLE1 regressions. Of the three measures of central bank independence Cbii, the legal aggregate index, and the Epstein-Schor index worked fairly well, while the turnover rate measure occasionally proved problematic (sign contrary to expectations) in SAMPLE3 regressions.

The strong and consistent performance of the enforcement determinants as explanatory variables was, however, not echoed by the set of variables corresponding to the standard determinants. In particular, monetary performance-related variables -- Infl, Inflv, Exr, and Exrv -- generally did not achieve explanatory power. Occasionally, Exr attained significance with the appropriate sign, but only when the effective exchange rate index-based measure was used. To test whether multicollinearity might have been responsible for the low t-ratios, we ran auxiliary regressions with Infl, Inflv, Exr, and Exrv each as the dependent variable regressed on the rest of the independent variables in the model. The regressions revealed little evidence that multicollinearity was a contributing factor in the poor performance of these variables. Additionally, in light of the fact that central bank independence, Cbii, proved to be strongly significant while none of the standard monetary performance variables did, we checked pairwise correlation coefficients between Cbii and each of variables Infl, Inflv, Exr, and Exrv. These coefficients were weak in every sample, confirming that pairwise collinearity was not responsible for the low t-ratios. 27/

As to the other explanatory variables corresponding to the standard determinants -- Expsh, Finsh, Kki, and Nfar -- their performance can be characterized as mixed at best. Expsh was consistently significant in all the 1964–90 regressions, but its performance weakened somewhat in the 1973–90 regressions. Finsh attained significance in the 1973–90 regressions (only SAMPLE2), but was not statistically significant otherwise. Kki was significant in SAMPLE2 regressions, and occasionally so in SAMPLE3. Nfar was not significant in the 1964–90 regressions, but when it did attain significance (for example, in a number of instances in the 1973–90 regressions), its sign was contrary to expectations.

These mixed results are not easily comprehended. It is not clear, for example, whether the significance of a particular variable in some time periods as opposed to others was simply a spurious outcome, or alternatively, indicative of a structural time-shift in the model itself, affecting the relevance of that variable as a regressor. Moreover, explanatory variables such as Nfish, Finsh, and Kki at best only imperfectly gauge and capture the three underlying determinants of multinational rent earnings from abroad, financial market depth and breadth, and financial openness, respectively (see Appendix I). Their performance as explanatory variables should therefore be interpreted with caution. It is possible that perhaps better designed variables might have resulted in better performance. Finally, the unexpected outcome registered by Nfar raises interesting questions. Contrary to the standard logic that a higher net debtor status should diminish international confidence in a currency and lower its international status, the empirical results indicated that, ceteris paribus, a net debtor position is associated with a larger currency share. This result warrants a re-examination of the traditional theoretical relationship between net foreign asset position and international currency share. While a full investigation of this issue is beyond the scope of this paper, one preliminary thought relates to the possibility of reverse causation, that is, the international role of a currency might in fact be a cause of deterioration in the net foreign asset position.

We conclude this section by presenting some final regressions. For each of the three samples, one final “best” equation was estimated whereby all the independent variables that were not significant in previous rounds for that sample were eliminated. Table 11 shows the estimated coefficients, along with elasticities for the independent variables, calculated at mean values. The elasticity figures indicate the relative strength (magnitude) of the impact of the various explanatory variables on the international role of a currency.

Table 11.

GLS Estimatesa: All SAMPLES, Final Regressions

article image
Note:t-statistics in parentheses, *significant at the 10 percent level, **significant at the 5 percent level or less.

Corrections for both autocorrelation and heteroscedasticity.

Elasticity at mean values of dependent and independent variables.

Cbii measured as Epstein-Schor Central Bank Independence Index in all regressions.SAMPLEI: 1964–90, 6 Countries, Dependent Variable is Currency Share in International Bond Issues.SAMPLE2: 1973–90, 9 Countries, Dependent Variable is Currency Share in International Bond Issues.SAMPLE3: 1973–90, 9 Countries, Dependent Variable is Currency Share in Official Foreign Exchange Reserves.

V. Conclusions and Implications

The empirical investigation of the determinants of currency internationalization indicated that with one exception (multinational rent earnings from abroad), all the enforcement determinants -- stock of foreign direct investment assets abroad, central bank independence, and military strength -- were strongly significant and robust in explaining international currency shares. All the standard determinants related to monetary performance -- inflation, inflation variability, exchange rate, and exchange rate variability -- failed to attain explanatory power, while the performance of the rest of the standard determinants -- financial openness, world export share, financial market depth and breadth, and net foreign asset position -- was mixed at best. With the usual word of caution about the limitations of any econometric endeavor, the empirical results from this research permit the following interpretation.

The strong empirical performance of the enforcement determinants demonstrates the validity of the endogenous enforcement approach as a theoretical framework for analyzing key currencies. Indeed, the empirical results lend support to the theoretical proposition (introduced in Section III) that the enforcement determinants are better indicators of a currency’s international role than the standard currency stability criteria of monetary performance and financial openness, the reasoning being that they embody greater and more meaningful information content about the propensity of a country to exhibit currency stability than what is contained in the mere statistical record relating to stability. With the exception of financial openness, none of the other standard determinants commonly understood as indicators of a currency’s stability prospects had any explanatory power whatsoever.

Endogenous enforcement as a theoretical framework for analyzing the issue of determinants of international currency status, as well as the empirical findings from the estimation of the model of determinants, carries important implications and raises several issues for future research. One implication relates to the possibility of substitution between the standard determinants and the enforcement determinants. Using a familiar tool from microeconomics, the production function, one can visualize a highly stylized scenario: key currency status as the “output”, and the two sets of determinants as two hypothetical inputs, a “standard” input, and an “enforcement” input (Chart 2). Points on the isoquant, or more appropriately the “iso-key currency share curve”, indicate various combinations of the two inputs necessary to achieve a fixed level of international currency share, and movements along the curve indicate potential trade-offs between the two inputs necessary to maintain the currency share constant. Our empirical estimates of the coefficients in fact allow for the calculation of elasticity of substitution at mean values of the inputs, taking two specific inputs at a time. For example, the ratio of the estimated coefficients for Fdish, an enforcement input (0.418), and Expsh, a standard input (1.464), yields an elasticity of substitution at 0.286 (Table 11). This implies that, ceteris paribus, at mean values of Fdish and Expsh, a 1 percent reduction in Expsh must be matched by a 0.286 percent increase in Fdish, for the international share of the currency to remain constant.

Chart 2.
Chart 2.

The Key Currency Model as a Production Function

Citation: IMF Working Papers 1997, 029; 10.5089/9781451844764.001.A001

Any potential for substitution between the standard determinants and the enforcement determinants would have a far-reaching implication -- that key currency nations have more autonomy than is traditionally believed to be the case. Conventional wisdom is that key currency nations are bound by their inflation performance; the restraint on monetary policy constitutes a constraint on national autonomy. However, if the relative strength of one or more enforcement determinants could compensate for the weakness in monetary performance, a key currency nation would no longer face the monetary-policy constraint to the extent envisaged.

Another implication of this paper relates to our understanding of trends in the evolution of international currencies in the world economy. The post-Bretton Woods period has witnessed several interesting shifts in the relative importance of international currencies, ranging from virtual dominance of the dollar in the 1970s, to the emergence of the duetsche mark and the yen as major challengers to the dollar’s hegemony in the 1990s. Typical attempts to analyze these trends have concentrated entirely on movements in the standard determinants. The paper contributes to such analyses by suggesting that these trends cannot be explained by standard determinants in isolation; shifts in the distribution of enforcement capabilities among potential key currency-issuing nations play a fundamental and critical role in determining these trends.

With respect to the future and the evolution of the international roles of the U.S. dollar, the duetsche mark, and the yen, popular predictions often paint a picture of an evolving tripolar world where “no dominant trend toward any [currency] will develop” (Bryant 1990, p.33). Purportedly, these forecasts of relative currency shares are based on expected levels of the usual fundamentals defined by standard theory. We would beg to differ from such tripolar predictions (no hegemony) in terms of a crucial qualification. Only if the trend toward convergence of enforcement capabilities among the big three continues in the future, might the tripolar system evolve in the form envisioned. Huge disparities among the United States, Germany, and Japan in their enforcement determinants, for example, could drive a wedge between their international currency shares and potentially steer the future course of evolution away from the tri-currency system. Analysis of trends in the enforcement environment in the world economy and their impact on currency internationalization thus promises to be an important avenue for future research.

Finally, this paper is an exercise in situating the topic of international currencies within new, emerging insights in the discipline of economics, as represented by the enforcement approach to exchange relations. Mainstream economics has long taken for granted the problem of agency and enforcement as a solved political problem. This paper is an attempt to address this oversight in the international money literature.

International Currencies and Endogenous Enforcement: An Empirical Analysis
Author: Roohi Prem