Dominant Currency Debt

UAA

BIS Working Papers

No 783

 

Dominant Currency Debt

by Egemen Eren and Semyon Malamud

 

Monetary and Economic Department

May 2019

 

JEL classification: D22, D84, E31

Keywords: inflation expectations, firms' survey, new information.

This publication is available on the BIS website (www.bis.org).

 

© Bank for International Settlements 2019. All rights reserved. Brief excerpts may be reproduced or translated provided the source is stated.

ISSN 1020-0959 (print)

ISSN 1682-7678 (online)

 

Dominant Currency Debt

by Egemen Eren and Semyon Malamud

 

Abstract

We propose a “debt view” to explain the dominant international role of the dollar. We develop an international general equilibrium model in which firms optimally choose the currency composition of their nominal debt. Expansionary monetary policy in downturns prevents Fisherian debt deflation through its effects on inflation and exchange rates, and alleviates financial distress. Theoretically, the dominant currency is the one that depreciates in global downturns over horizons of corporate debt maturity. Empirically, the dollar fits this description, despite being a short-run safe-haven currency. We provide broad empirical support for the debt view. We also study the globally optimal monetary policy.

JEL classification: D22, D84, E31

Keywords: inflation expectations, firms' survey, new information.

The dollar is the most common currency of choice for debt contracts worldwide. According to the Bank for International Settlements, the dollar-denominated credit to non-banks outside the United States amounts to around $11.5 trillion. While the dominance of the dollar had declined prior to 2008, the dollar has strengthened its international role since the Global Financial Crisis (Figure 1).

Image irrp

In this paper, we study how a single currency can become the most common currency of choice for denominating debt contracts in general equilibrium, i.e. the dominant currency, why that choice is the dollar, and why the dominance of the dollar may have declined and recovered in the last two decades.

We develop an international general equilibrium model with multiple countries where firms optimally choose the currency composition of their debt. All firms are exporters, prices are flexible, and the firms have fully diversified cash flows. Firms issue equity and nominal, defaultable debt to optimize the trade-off between tax benefits of debt and the risk of default. Debt can potentially be issued in any currency. Since debt is nominal, exchange rate movements affect their real debt burdens. When debt servicing costs are high relative to profits, firms face debt overhang. They cut production and reduce demand for their intermediate inputs imported from other countries. This global debt overhang channel spreads debt overhang costs along the global value chain and serves as a key mechanism for international spillovers.

Expansionary monetary policy in downturns prevents Fisherian debt deflation through its effect on inflation and exchange rates. We model a central bank policy as a countercyclical monetary policy rule that attempts to ease financing conditions for firms in times when output gap is high due to inefficient production with debt overhang. Central banks differ from each other in their willingness and ability to pursue inflation stabilization policies that protect borrowers and real debt burdens. In the model, relative inflation between two countries determines the exchange rates via a relative purchasing power parity (PPP) condition where higher inflation leads to exchange rate depreciation.

Our first main result is the existence of a dominant currency debt equilibrium defined as an equilibrium in which a single common currency is chosen as the currency for denominating all outstanding debt contracts, even though there may be other currencies with almost identical characteristics. Theoretically, the dominant currency is the one that depreciates in global downturns over the horizons of corporate debt maturity, which is typically around seven years. When an adverse shock hits, the risk of default is minimal if debt is issued in currencies that co-move positively with firms' discounted profits. When profits drop, such currencies tend to depreciate as well, thereby reducing these firms’ debt service costs and alleviating financial distress.

Empirically, the dollar fits the theoretical description of the dominant currency. We show that the dollar robustly depreciates during stock market downturns on horizons that accord with typical duration of corporate liabilities. Since typical corporate debt maturity is around seven years, the risk profile of the dollar over the medium run determines its use for the denomination of corporate debt. As a result, even though the dollar does co-move negatively with the stock markets for shorter horizons up to a year (i.e. it is a safe haven currency), this is less relevant for the choice of currency denomination of corporate debt.

This pattern of the correlations between the dollar and the stock market has direct implications for the maturity choices of firms for dollar-denominated debt contracts. As the dollar co-movement with the stock market increases over longer horizons, our model predicts that the propensity to issue dollar-denominated debt increases with debt maturity. We use granular bond issuance data to formally test this prediction, and find strong support.

In our model, expectations of firms about the ability of a central bank to pursue inflation stabilization policies in global downturns impact the currency denomination choice of debt contracts to the extent that they affect exchange rates. While expectations about such counter-cyclical inflation policies are not directly observable, they can be backed out from the inflation risk premium (IRP). In fact, our model predicts that in the dominant currency debt equilibrium, the dominant currency is always the currency with the highest IRP.

We formally test our theoretical prediction about the link between the share of dollar- and euro-denominated debt and inflation expectations, as captured by the IRP. We find that the dynamics of IRP in the United States and the Eurozone explains a large part of the variation in the share of dollar debt with the signs of the regression coefficients consistent with our theory, even at a quarterly frequency. We interpret this fact as strong evidence of a distinctive prediction of our theory: Changes to the currency composition of debt can occur in high frequency.

We provide additional broad empirical evidence in favor of the debt view. We show that the risk properties of currencies discussed in this paper shed light on patterns of debt issuance in other major currencies, such as the pound and the yen, as well as the dynamics of the shares of dollar- and pound-denominated debt during the interwar years as documented by Chifu, Eichengreen and Mehl (2014). We further show that an extension of our basic model can also be used to explain the distribution of local currency and dominant currency mix of corporate debt for a cross-section of emerging market economies.

Finally, our general equilibrium framework allows us to discuss the macroeconomic and monetary policy implications of a dominant currency debt equilibrium. We run the following thought experiment: In the dominant currency debt equilibrium, with firms around the entire world issuing dominant currency debt, how should the dominant currency central bank assign weights to the output gaps of each country to maximize the global welfare? We derive the optimal weights analytically and show that they are lower for countries with volatile TFP shocks, high debt restructuring costs and for those countries that are more important in world trade. Limiting the insurance given to those countries reduces the leverage of firms in these countries, thereby improving global (and domestic) welfare.

Related literature. The international role of the dollar has received a lot of attention in the recent literature. The dollar is omnipresent in all parts of the global financial system, in­cluding international trade invoicing (see Goldberg and Tille (2008), Gopinath (2015), Casas, Diez, Gopinath and Gourinchas (2017)); global banking (Shin (2012), Ivashina, Scharfstein and Stein (2015), Aldasoro, Ehlers and Eren (2018)); corporate borrowing (Bruno, Kim and Shin (2018), Bruno and Shin (2017)); central bank reserve holdings (Bocola and Lorenzoni (2018)); and global portfolios (Maggiori, Neiman and Schreger (2018)). Our paper adds to the growing literature that studies the dominant role of the dollar in a general equilibrium framework.

Our main contribution is the introduction of the “debt view” in explaining the interna­tional role of the dollar. Current explanations can be broadly classified into three categories. First is “the trade view,” wherein trade invoicing in dollars is the reason for the dollar’s role in the global economy (see, for example, Gopinath and Stein (2018)). Second is “the safe asset view,” in which the dollar is dominant because of its safe haven properties (see, for example, He, Krishnamurthy and Milbradt (2019), Farhi and Maggiori (2018)). Third is the “vehicle currency view,” wherein the dominance of the dollar arises from its international use as a vehicle currency (see for example Goldberg and Tille (2008)).

The debt view of the dollar’s dominance assigns an important role to the choice of debt currency denomination of firms, monetary policy and exchange rates. The debt view focuses on the medium run to account for typical corporate debt maturity, and in that complements other theories which focus on the short run frictions such as price stickiness, or the short-run appreciation of the dollar in bad times as an insurance to investors. In contrast to other theories, we show that a dominant currency equilibrium can arise without relying on network effects, price stickiness, pricing complementarities and safety demand.

The most closely related paper to ours is by Gopinath and Stein (2018), which demon­strates how the dollar can emerge as a key international currency starting from its role in trade invoicing and in turn affecting global banking, which in turn affects currency denomination of bank deposits and firm borrowing endogenously. In our setup, dollar’s dominance arises from debt currency choice of firms. In their paper, the dollar’s role for denominating debt depends on its role in trade invoicing. In our paper, it depends on the risk properties of the dollar compared to other currencies.

A growing literature studies reasons for persistent uncovered interest rate parity (UIP) violations and its implications for the foreign currency borrowing of emerging market firms (see, for example, Hassan (2013), Baskaya, di Giovanni, Kalemli-Ozcan and Ulu (2017) and Salomao and Varela (2018)). Our primary interest lies in understanding why it is the dollar that is the “dominant foreign currency” as opposed to similar currencies with similarly deep and liquid markets, such as the euro. Thus, we abstract from many features of currencies that lead to UIP deviations in reality to clearly highlight our main mechanism, even though it would be possible to capture these in reduced form, through differential issuance costs. That said, in section 6, we provide theoretical and empirical results for the dominant and local currency mix in the debt of a cross-section of emerging market firms which depends on the properties of local inflation and its relation to the US inflation that could inform this literature.

Numerous papers in international macroeconomics study how the exchange rate pass­through into prices of real goods depends on price stickiness and the invoicing currency choice. We highlight a novel pass-through mechanism that operates through the financial channel. With dollar debt, a dollar appreciation shock puts leveraged firms in distress and increases their effective operational costs. Firms respond to this circumstance by raising prices, consistent with the mechanism highlighted in Gilchrist, Schoenle, Sim and Zakrajsek (2017) and Malamud and Zucchi (2018). Importantly, this pass-through channel operates even when prices are fully flexible.

  1. Model: Main mechanisms and general equilibrium

    1. Households and exchange rates

Time is discrete, indexed by t = 0,1. There are N countries, indexed by i =1, N. Households work and consume. They maximize

Image qa7a

is the constant elasticity of substitution (CES) consumption aggregator, with the elasticity of substitution n. Here, Ci>t(j,u) denotes the consumption of type-u goods imported from country j into country i, with u E [0,1].

The price at which country j firms sell type-u goods in country i is denoted by Pti(j, u). This price is always in the domestic, country-i currency. The country-i price index is defined by the standard formula

Image 0bg9

For simplicity, we assume that prices are fully flexible and the aggregate nominal price level follows an exogenous, country-specific stochastic process Vi.t.

Households have access to a complete, frictionless financial market with a domestic, nominal pricing kernel Mi,t,T in the domestic currency, for any t < t. The following lemma characterizes their optimal consumption choices.

Image 4mlr

We denote by £i.j.t the value of a unit of currency i in the units of currency j. That is, when Ei.j.t goes up, currency i appreciates relative to currency j. We select one reference currency (the dollar), denoted by $, and use Ei.t = Ei.$.t to denote the nominal exchange rate against the dollar. Due to assumed market completeness, consumers in different countries attain perfect risk sharing and the pricing kernels Mi.t.t+1 and Mj.t.t+1 of any two countries i,j are linked through the no-arbitrage identity:= EHy . For the sake of convenience, everywhere in the sequel, we normalize the initial price levels Vо so that Ei,0 = 1 for all

Image iydc

Each country’s production sector is populated by a continuum of ex-ante identical firms, as indexed by ш E [0,1], with firm ш producing type-ш goods. We use (i,u) to denote firm ш in country i. All firms use labour as well as goods produced by other firms (domestic and foreign) as inputs in a standard Cobb-Douglas production technology: Absent debt overhang (as defined below), the production function of an (^ш) firm is given by

where, for each i =1, N,

  • Ziyt(u) > 0 is the firm (i,ш) idiosyncratic production shock that is drawn from the distribution with the density ф(г) = £ze-1 on [0,1] with a country specific parameter £ > 0 and the cumulative distribution function

Image 5a5q

We assume that Zi,t are i.i.d. over time and across firms within a given country;

  • ai,t is the country-i productivity shock;
  • Lt(i,u) is labour hired by the (i,u) firm at time t;
  • Xt(i,u) is the CES aggregator of goods used by the (i,u) firm as inputs:

Image ib7v

Here, Xt>(i>u)(j, s) is the demand of a (i, u) firm for goods of a (j, s)-firm in country j.

We assume that firms are taxed on profits at the tax rate t. As we show in the Appendix (see Lemma A.1), total nominal after tax profits of an (i, u)-firm measured in country-i currency are given by

Image hlx4

for some explicit, country-specific random variables Qi)t, i = 1, N.

1.2.2 Debt

In order to highlight the mechanisms through which debt affects real outcomes in our model, we first introduce debt exogenously. We assume that all firms in each country i have nominal debt. When firms receive a low draw of either the idiosyncratic shock Zi)t or the country-specific TFP shock ai)t, they become distressed and produce less efficiently. This form of debt overhang is the key mechanism by which debt is linked to real outcomes in our model.

Assumption 1 (Debt Overhang) Firms born at time t — 1 live for one period, [t — 1, t] and are endowed with nominal debt. The random face value, Bi,t, of this debt in domestic currency is realized at time t.

• Having observed an idiosyncratic shock realization, Zi,t, and before starting production, the firm computes its optimal potential profits (4) :

  • If the after-tax profits (4) are sufficient to cover the debt servicing cost, the firm then produces according to (2).
  • If the after-tax profits are insufficient to cover the debt servicing cost, then the firm enters a financial distress state, and it is only able to produce at a fraction Z(n-1) 1 g (0,1) of its capacity. That is, its production function is given by

Image er8

and, similarly, by direct calculation, total nominal profits are given by nit = Щу(ш)(1 + (Z — 1)1distress) .

The simple nature of debt overhang in Assumption 1 implies that profits in distress are identical to those in (4), but with Zi,t being replaced by ZZi,t. Furthermore, Assumption 1 also implies that the firm enters financial distress when Zi t falls below the distress threshold

Image jr9y

In the sequel, we also frequently refer to (5) as leverage: Indeed, Ti,t is the ratio of the face value of debt, Bi,t, to the measure of country i nominal value of profits as given by Qi,t .

Assumption 1 allows us to capture two key features of the behaviour of financially constrained firms. When the debt service costs are high relative to profits, financially constrained firms often have problems paying their suppliers and employees, and they are thus often forced to fire employees and cut production. As a result, in such a distress state effective marginal costs surge, forcing these firms to raise prices and cut production. Both features are important for our results: In equilibrium, high prices hit global demand; firms respond by cutting their production and raising prices even further, thereby spreading the debt overhang costs worldwide.

1.2. General equilibrium

In our model, all the goods are used for both consumption and for production. Further, by assumption, households and firms in all countries are completely symmetric: They all have identical preferences and identical production functions, and hence, they only differ in productivity shocks agt. Under this assumption, marginal rates of inter-temporal substitution are perfectly aligned across countries, so that

Image ui64

The nominal exchange rates are fully driven by inflation dynamics:

Image rfpp

plays an important role in the subsequent analysis. It captures the total drop in productivity of country—i firms. Indeed, by (4), the total profits for country i firms are given by

As explained above (see (5)), distressed firms are those firms with idiosyncratic shock realizations Zi,t below the threshold Ti,t. Thus, by (3), the fraction of distressed firms is provided by the measure^ : Zi,t(u) < Ti,t] = min(Ti t, 1}^. We restrict our attention to normal equilibria in which a strictly positive fraction of firms in each country is not in distress, so that Ti,t < 1 for all i = 1, ••• , N, and Ct are monotonically increasing in aj,t, j = 1,N. The following result shows explicitly how global debt overhang influences equilibrium consumption.

Image giwd

Image p7gh

Image icky

Image 4z56

We complete this section with two results that are crucial for understanding the real effects of debt overhang. First, since financial distress lowers production, debt overhang leads to an output gap and unemployment. Second, due to the input-output linkages, rising debt burdens in one country always transmit to other countries. Here we show that, under natural conditions, a rising debt burden in one country always leads to higher debt overhang costs in all other countries. The following two corollaries formalize this intuition.

Image bpyc

Corollary 1.3 Denote by Lt(i) and Ot(i) the country i equilibrium labour demand (employ­ment) and output, respectively. Let L((i) and Ot*(i) also denote the corresponding frictionless benchmarks absent the debt overhang. 

Corollary 1.4 (Default transmission) Suppose that f) > 0.20 Then, a shock to debt bur­den Bj,t of a country j always leads to an increase in the fraction of distressed firms in all other countries i = j.

2. Dominant Currency Debt: Theory

In this section, we study firms’ choice of leverage and the composition of the currency denomination of their debt in general equilibrium. We assume that firms finance themselves by issuing both equity and defaultable short-term nominal bonds in any of the N currencies. Each bond has a nominal face value of one currency unit, and the firm is required to pay a coupon of c currency units per unit of outstanding debt. We denote by Bj,t(i) the stock of outstanding nominal debt at time t of country i firms, denominated in the currency of country j. We also denote by Bt = (Bj-,t(i))N=1 the N x N matrix of debt stocks in different currencies. As in Gomes, Jermann and Schmid (2016), we assume that coupon payments are shielded from taxes, so that

Image bp0

is the total debt servicing cost in country—i currency, net of tax shields. The choice of firm leverage, therefore, depends on the trade-off between tax advantages and the distress costs. Then, absent any default, nominal distribution to shareholders at time t + 1 is given by

Image 2676

If the cash flows (10) are non-positive, shareholders will default on a firm’s debt. Upon default, debt-holders take over the firm, and shareholders get zero.

1.1. Optimal leverage and the dominant currency debt

Image 8xoh

As we explain above, shareholders default whenever cash flows (10) are non-positive, that is, when nt+i(i,w) < Bi,t+1(Bt). Thus, default occurs whenever Zi,t+1 falls below the default threshold

Image 9z2d

We assume that, upon default, debt-holders only recover a fraction pi of their promised value, 1 + c. Thus, by direct calculation, the nominal price in country i currency of one unit of debt denominated in currency j is given by

Image 3o0g

We assume that country i firms face a proportional cost qi(j) of issuing in country j currency for i, j = 1, N and it maximizes equity value plus the proceeds from the debt issuance net of issuance costs:

Image lc06

Everywhere in the sequel, we use E$ and Cov$ to denote conditional expectation and covari­ance under the dollar risk neutral measure with the conditional density Et[M$,t,t+1]-1M$,t,t+1. Furthermore, for each stochastic process Xt, we consistently use the notation

Image x7vk

We need the following assumption to ensure that the leverage choice problem has a non-trivial solution.

Image opum

The first condition ensures that the cost qi(j) of issuing debt is less than the gains, as measured by the value of tax shields, so there is positive debt issuance. The second condition ensures that the recovery rate pi is sufficiently small: Otherwise, funding becomes so cheap for the firm that the firm may want to issue infinite amounts of debt. The following is true.

Image w3cx

The condition (11) shows that the incentives for issuing in dollars are determined by two forces: The effective cost of issuance, qi($), and the risk profile of the dollar. Dollar capital and derivative markets are deep and liquid (see Moore, Sushko and Schrimpf (2016)); thus, the low effective cost of issuance, qi($), is an obvious factor favoring the dollar as the dominant currency of choice for debt contracts. However, our main result does not rely on the dollar debt having low issuance costs: Theorem 2.1 implies that the dollar can rise as the dominant debt-denomination currency due purely to its risk profile. To understand the underlying mechanism, we note that, absent heterogeneity in effective issuance costs (that is, when qi(j) is independent of j), (11) takes the form of

Image j2qi

Here, Ei,t+iQi,t+i is the factor that determines the value of profits of country i firms in dollars (see (6)). Intuitively, at time t, firms when deciding on the currency composition of their debt choose to issue in dollars if they anticipate the dollar to depreciate at those times when their time t + 1 profits are low; the condition (13) provides a precise formalization of this intuition. Since (Ei,t+iHi,t+i)-i attains its largest value when dollar profits Ei,t+iQi,t+i are close to zero, covariance (13) overweighs the distress states: When I is sufficiently high, (13)  essentially requires the dollar to depreciate against all its key competitors during times of major economic downturns.

It is also important to note that condition (13) corresponds to the problem a firm faces when choosing between dollar debt and debt denominated in other key currencies, such as e.g., the euro, the yen, the Swiss franc and the pound. For an emerging markets’ firm that is choosing between local currency debt and dollar debt, heterogeneity in issuance costs may be as (if not more so) important as the currency risk profile. However, even for the choice between dollar- and euro-denominated debt, ignoring differences in issuance costs puts the dollar at a disadvantage: Existing evidence (see e.g., Velandia and Cabral (2017)) suggests that issuing debt in dollars is significantly cheaper than issuing in euros.

1.2. Dominant currency debt in general equilibrium

In this section, we combine the equilibrium characterization in Theorem 1.2 with the dominant currency debt condition of Theorem 2.1 to answer the following question: When does dominant currency debt arise in general equilibrium? We then make the following simplifying assumption:

Assumption 3 We assume that

  •  issuing costs are independent of currency denomination: ^(S) = qi(j) for all i, j =1
  • TFP shocks satisfy aj,t = at + ea,t for some common shock at and idiosyncratic TFP shocks ej,t with a small variance that are independent across countries and also inde­pendent of at.

As we explain above, in our model, consumption is perfectly aligned across countries, real exchange rates equal 1, and nominal exchange ratechanges are determined purely by

Image jdyz

the relative inflation rates: £i,t,t+1 = P—t1t+1P$,t,t+1. Thus, substituting profits Qj,t from (7) and using Assumption 3, we determine from (13) that issuing all debt denominated only in dollars is optimal if and only if for all.

In our model, a central bank policy that leads to a positive shock to Pgt has two main functions: First, it eases the real debt burden of firms’ borrowing in country’s i currency; and secondly, it leads to a depreciation of country i’s currency. Therefore, an increase (respectively, decrease) in Pgt can be interpreted as monetary easing (respectively, tightening). We further assume a counter-cyclical monetary policy rule whereby the central bank eases (respectively, tightens) when employment or output falls (respectively, rises) relative to the frictionless benchmark (see (9)):

Assumption 4 The inflation rate in country i is determined by

Image 0ao8

Here, фг > 0, i = 1, ■ ■ ■ , N is a country specific parameter that measures the responsiveness of domestic monetary policy to economic conditions, and £г.t+1 is a country-specific monetary policy shock with bounded support and variance.

We can now characterize the conditions when a dominant currency debt equilibrium emerges.

Theorem 2.2 Suppose that monetary policy uncertainties ог. *, i =1 N are sufficiently small and the indices фг are all pairwise different, and 1 — ( is not too large. Then, there exists a Dominant Currency Debt equilibrium. The dominant debt currency is always the currency of the country with the highest index фг.

In addition to inflation stabilization indices фг, countries may also differ in the volatility of inflation shocks, аг,е. Naturally, firms view this uncertainty as an additional and undesirable form of risk. The following is true:

Proposition 2.3 Absent heterogeneity in the indices фг, i = 1, N, firms always issue debt in the currency of the country that has the lowest volatility of inflation shocks, аг,е.

While аг,£ does represent idiosyncratic inflation volatility in our model, Proposition 2.3 holds true for any shocks to exchange rates that are unrelated to economic fundamentals, for example, monetary policy shocks or temporary demand pressures and liquidity shocks in currency markets. Proposition 2.3 suggests that, in addition to insufficient market liquidity (modelled by high issuance costs), the significant idiosyncratic volatility of emerging market currencies may serve as an additional important mechanism that explains why firms do not want to issue in these currencies, despite the fact that such currencies do tend to significantly depreciate during crises (see also Du, Pflueger and Schreger (2016)). As an illustration, consider a typical emerging market currency, the Argentinian Peso (ARS). During the period of November 1995-September 2018, the standard deviation of the monthly returns on the dollar index was 1.9%, while the standard deviation of monthly returns on the ARS/USD exchange rate was 7.1%. Further, this volatility was almost entirely due to idiosyncratic shocks, as indeed, the R2 of a regression of the monthly ARS/USD returns on the returns on the dollar index was only 0.0033.

3. Dominant Currency Debt: Empirical Evidence

As we show above (see condition (13)), firms prefer to issue in dollars if the dollar exchange rate co-moves positively with the present value of their profits, which we proxy by their stock market value. For the stock market, we use S&P 500 and MSCI AC World indices measured in dollars as we also measure profits in dollars in our theoretical results. As a proxy for the global dollar exchange rate, we use the trade-weighted dollar index against major currencies, including those in the Eurozone, Canada, Japan, United Kingdom, Switzerland, Australia, and Sweden, as obtained from the FRED database. We also provide results using the bilateral exchange rates between the dollar and the euro, the yen, the pound and the Swiss franc.

1.1. Why is the dollar the dominant currency? Results with the dollar index

Given that the dollar is the most common currency of denomination in international debt markets, the first prediction of our model is that the returns on the dollar index positively correlate with the returns on the stock market indices at horizons that correspond to the weighted average corporate debt maturity, that is around 6-7 years (see section 5, Choi, Hack- barth and Zechner (2018), Cortina, Didier and Schmukler (2018)). To test this prediction, we first run the following regressions for the horizons of h G {3,12, 24, 36, 48, 60, 72,84, 96,108,120} months:

Image 8mho

Here, Return-USD^ and Return SP500t^ denote the rolling returns on the dollar index and the SP500 index over h months, respectively. The left-hand panel of Figure 2 reports the results for the regression coefficient f3h for different horizons, together with the 95% confidence intervals for the period between January 1994 and

Image 2h72

The results show a pattern of negative correlations at short horizons and positive and mostly increasing correlations at longer horizons. These findings, together with condition (13), suggest that US firms are better off if they borrow in dollars rather than in other major international currencies if their debt maturity exceeds roughly two years, which is the case here.

Next, we turn to the rest of the world and compute the correlation of the returns on the dollar index with the returns on the MSCI AC World Index. We follow the same procedure as before and run the following regressions, for the same horizons h as shown above:

Image nwy

The right-hand panel of Figure 2 reports the results for the regression coefficient fth for different horizons, together with the 95% confidence intervals for the period between January 1994 and September 2018. The results from these regressions are aligned with the results for the S&P 500 stock index. While the dollar is negatively correlated with the MSCI AC World Index at short horizons, this correlation turns positive at horizons that are longer than three years. Thus, condition (13) suggests that international firms with average debt maturities longer than three years (which is indeed the case for most firms; see section 5, and also Choi, Hackbarth and Zechner (2018), Cortina, Didier and Schmukler (2018)) are better off issuing debt that is denominated in dollars.

1.2. Why is the dollar the dominant currency? Results with bilat­eral exchange rates

In this section, we provide the results for the same regressions as in Section 3.1, but using bilateral exchange rates for the dollar against four other major currencies. As Figure 3 shows, the dominant currency condition (13) holds empirically with i=dollar and currency j being the euro (EUR), the yen (JPY), or the Swiss franc (CHF). The only exception is British pound (GBP), for which our empirical proxy estimates in Figure 3 for the covariance in (13) have a negative sign. However, these covariance estimates are statistically insignificantly different from zero at the horizons of average corporate debt maturity. Thus, even absent differences in issuance costs, firms would strictly prefer issuing debt denominated in dollars, even if they could issue in EUR, JPY, or CHF. And even a slight difference in issuance costs favouring dollar to GBP would also make dollar immediately dominate over GBP.

Image npg9

1.3. Yen vs. pound

As we show in Section 3.2, the risk properties of the dollar alone can explain why the dollar dominates the euro, the yen and the Swiss franc in the sense of Theorem 2.1. One notable case is the pound: By Figure 3, the pound has favorable risk properties for debt issuers compared to most of the other major currencies. In reality, it is not reasonable to expect the pound to compete with the dollar on a large scale because of the mere size of the US economy. However, it is reasonable to compare the dynamics of corporate debt issuance in GBP to that in JPY, since Japan and the Great Britain have similar size in the world economy. In this case, Figure 3 shows that (13) holds empirically with i=GBP and j=JPY, and hence corporates should strictly prefer issuing in GBP to issuing in JPY.

Figure 4 confirms this prediction of our model. Indeed, surprisingly, despite the slightly larger share of Japan in the world economy and lower nominal interest rates in Japan, the share of pound-denominated debt is higher than the share of yen-denominated debt.

Image k20g

1.4. The fall and the rise of the dollar

In our model, relative inflation dynamics in the two countries determine exchange rates. This is a simplifying assumption that we use to compute general equilibrium. An explicit relationship between inflation and exchange rates is not crucial for our previous empirical results. However, in this section we do take the model seriously and explore whether there is actually an empirical link between inflation and the choice of the dominant currency.

Image nhr8

The explicit link between relative inflation dynamics and exchange rates in our model is the key element behind Theorem 2.2. The latter states that firms should be issuing dollar debt only if they expect the United States to have the most counter-cyclical inflation over the horizon of their debt maturity. These expectations about inflation cyclicality can be backed out from the inflation risk premium, given by the difference between inflation expectations under the risk-neutral and the physical measures:

The covariance term, Covt(Mi,t,t+1,Pi,t,t+1), reflects the basic intuition, namely, that the inflation risk premium is determined by market expectations regarding inflation cyclicality. The following is true.

Proposition 4.1 Under the hypotheses of Theorem 2.2 , the inflation risk premium, IRPi>t, has the largest value for the dominant currency country.

The key prediction of Proposition 4.1 is a direct link between the IRP and the dominance of a currency. To test this prediction empirically, one can consider an extension of our benchmark model featuring time variation in the coefficients фе ,t and ф$у, the respective abilities of the ECB, and the Federal Reserve to pursue counter-cyclical stabilization policies (see Assumption 4). In such an extended model, Theorem 2.2 implies that a change in the sign of ф$у — фе ,t from positive to negative will immediately trigger a regime change, thereby making the dollar lose its dominant currency status to the euro. Proposition 4.1 further predicts that this status change should be accompanied with a simultaneous change in the sign of IRP$,t — IRPe ,t from positive to negative. Importantly, what matters is not the actual value of these coefficients, but rather the firms’ and market participants’ expectations about them: The former determine debt issuance policies; the latter determine the IRP.

We compare the pre- and post-crisis trends in the shares of euro- and dollar-denominated debt (Figure 1) with the pre- and post-crisis dynamics of inflation risk premia (18) in the United States and the Eurozone. Figure 1 and Figure 5 show a behaviour that is consistent with our key prediction: In the pre-crisis period, the inequality IRPe ,t > IRP$,t held true most of the time, and the fraction of euro-denominated debt rose; post crisis, the relationship reverted to IRPEURt < IRP$,t, and the fraction of euro-denominated debt started to fall.

Image nhr8

Consistent with Figure 1, Maggiori, Neiman and Schreger (2018) show that the share of dollar-denominated debt in cross-border corporate holdings has drastically increased during the post-crisis period compared to that for the euro. We argue that this pattern is to a large extent driven by the bond-supply channel of Figure 1; bond investors hold what the firms issue to clear the markets in general equilibrium. That said, while bond investors might generally dislike holding nominal bonds with a high inflation risk premium, there is an opposite force in our model that increases the attractiveness of dollar-denominated bonds for lenders: The default probability of these bonds is lower because it is easier for firms to repay dollar debt due to lower real debt burdens in bad times. However, when in equilibrium, the former dominates, and hence the dominant debt currency has a higher inflation risk premium.

We further investigate the dynamic link between IRP and debt issuance in different currencies. Clearly, the presence of any type of inertia in debt currency denomination (e.g., due to some implicit or explicit adjustment costs or smooth variations in beliefs) will make the transition between the euro- and dollar-dominance regimes smooth, with the share of dollar debtincreasing in IRP$t and decreasing in IRPe ,t. Stated formally, we test the following hypothesis.

Hypothesis IRP-1: The share of dollar-denominated debt, USDShr, positively relates to IRP$,t — IRPe ,t. More specifically, it positively relates to IRP$,t, while it negatively relates to IRP€ ,t.

Table 1 presents the results of a linear regression. First, in column (1), we find a positive, statistically significant relationship between the share of dollar-denominated debt

Image k6m1

Notes: Robust standard errors are shown in parentheses. Denote significance at the 10, 5, and 1% levels respectively. USDfhr refers to the share of dollar debt, including both bank loans and debt securities. IRPand IRPJY refer to the 5-year inflation risk premia for the dollar and the euro respectively, as measured by Hordahl and Tristani (2014). Q refers to the quarterly frequency since 2000. Д refers to quarterly first difference.

on the IRP differential with an R2 of 52.8%. In column (3), even after including year dummies, the positive sign remains positive and statistically significant. The estimates are also economically large: A 1 percentage point increase in the IRP differential between the dollar and the euro is associated with an increase in the share of dollar-denominated debt of about 10 percentage points.

An alternative specification includes both inflation risk premia for the dollar and the euro separately as independent variables in the regression. In columns (2) and (4), we report the results without and with year dummies, respectively. These results are again consistent with our hypothesis: The share of dollar-denominated debt is positively related to /RP$5Y, while it is negatively related to IRP^^.. Furthermore, the R2 of the specification in column (2) is much higher than that in column (1), and the coefficient for the ZRPjpt is economically very large. A 1 percentage point increase in the euro IRP is associated with a drop in the share of dollar-denominated debt of about 22 percentage points. The negative coefficient remains economically significant even after the inclusion of year dummies in the quarterly regressions in column (4). In column (5), we run the regression with quarterly first differences which yields qualitatively similar results as in column (4).

5. Debt currency and maturity choice

Our results in section 3 have direct implications on the link between debt maturity and the incentives to issue dollar-denominated debt. Namely, as the dollar’s co-movement with the stock market increases over longer horizons, we expect that firms with a longer maturity of debt have a preference for issuing this debt in dollars.

We use data at the bond issuance level in order to formally test the hypothesis that the propensity to issue dollar-denominated debt increases with debt maturity. We restrict our attention to non-financial corporations that issued bonds between 2000 and 20 1 9.

1.1. Data

We use data from Dealogic where observations are at the ISIN level of bond issuance. In order to keep the timing of our analysis similar to the previous sections, we restrict the sample to bonds issued between January 2000 and February 2019. Our dataset includes a total of 102,159 bonds, issued by 23,992 firms that are headquartered in 110 different countries.

The dataset includes information on the identity of the firm, the country where it is headquartered, the industry as well as information on the bonds, such as the currency denomination, date of issuance, maturity date, issued amount denominated in the local currency of the firm’s headquarters, and whether the bond is investment-grade or is not.

In the full sample, the mean of the winsorized maturity is 2,950 days, with a standard deviation of 2,646 days; the minimum value is 375 days and the maximum value is 10,958 days. We present the summary statistics regarding debt maturity by different currencies in Figure 9 in a box plot in the Appendix. As Figure 9 shows, the maturities of bonds issued in pounds have the largest median, followed by bonds issued in dollars.

1.2. Results

Following our results in section 3 and the patterns in Figure 1, we test the following hypotheses using micro-level data on bond issuance.

Hypothesis BI-1: A longer debt maturity is associated with a higher propensity to issue

Then, the independent variables of interest in our regressions become: Maturityw, which is the winsorized and standardized value of maturity at the 5% level; 1 (Maturity > 1y), which is a dummy variable that takes the value 1 if the non-winsorized maturity is greater than one year. According to our hypotheses, we expect a positive coefficient for these variables.

Other control variables are the size of the issuance and a dummy variable that is equal to 1 if the bond is investment-grade. Moreover, depending on the specification, we include Industry, Country * Month and Firm * Month fixed effects. We cluster the standard errors at the Country * Year level.

We run different linear regressions, varying the fixed effects used and making different cuts of the sample in order to test the predictions of our theory. Table 2 presents the results.

The first three columns control for bond characteristics as well as industry and Country * Month fixed effects. In column (1), we run the regression using the full sample. The coef­ficient on Maturityw suggests that a one standard deviation increase in maturity increases the likelihood of the currency denomination of the bond to be dollars by 2 percentage points. In column (2), in order to address worries that this result might be driven by the ease of issuing longer maturity bonds in dollars, we still use the full set of firms, but we restrict the sample to bonds with maturity of 10 years or less. For this specification, we use the non-winsorized sample and test whether 1 (Maturity > 1y) predicts a higher likelihood of dollar debt issuance. Indeed, dollar bond issuance is more likely for maturities of 1y-10y, in line with our findings in section 3. In column (3), we restrict the sample to bonds that are issued by non-US, non-Eurozone, non-JP, non-GB, non-CHF firms and ones that are only issued in USD, EUR, JPY, GBP and CHF. We obtain the same result that longer maturities correspond to a higher likelihood of issuance in dollars.

For identification, we rely on firms that issue multiple bonds in at least two different currencies in a given month. This choice allows us to tightly identify that the same firm

Image u7n

Notes: Standard errors clustered by Country * Year in parantheses. *, **, *** denote significance at the 10, 5 ,and 1% levels, respectively. 1 (USD) is a dummy variable that takes the value 1 if the currency of the issued bond is the dollar. Maturityw is the standardized value of maturity winsorized at 5% and 95% levels. t (Maturity > 1y) is a dummy variable that is 1 if maturity is greater than 1 year. Controls include the local currency amount of the size of the issuance and a dummy variable for the status of investment-grade status of the bond. The full sample includes all observations. Partial & FC refers to observations where the nationality of the company is not the United States, a country in the Eurozone, Japan, Great Britain or Switzerland, but the currency is either USD, EUR, JPY, GBP or CHF. t means that the sample is further restricted only to those firms that issued debt in multiple currencies in a given month.

that has access to multiple markets chooses to issue the longer maturity bond in dollars as opposed to issuing in other currencies. In column (4), we repeat the results for a modification of the regression in column (1), but use Firm * Month fixed effects instead in a sample that is restricted to those firms that issued debt in multiple currencies in a given month. In column (5), we repeat column (3) in the new sample with Firm * Month fixed effects. In both cases, the results corroborate Hypothesis BI-1.

6. Local currency and dominant currency debt mix

In this section, we test the predictions of our model using a cross-section of the emerging market economies for which data on corporate debt in different currencies are available. To this end, we prove the following extension of Theorem 2.1 for the case wherein firms issue a mixture of local currency (LC) and dollar-denominated debt (see Theorem G.1 in the Appendix for the proof. Note that, while Proposition 6.1 is a partial equilibrium result, it still holds true in general equilibrium when debt overhang costs are sufficiently small).

Proposition 6.1 Suppose that (1) qi(i) = qi($) (that is, issuing in LC costs the same as issuing in dollars); (2) the variance of all shocks is sufficiently small; (3) inflation in all countries is determined by (15); and (4) issuing debt in both LC and dollars is optimal.

Then,

  1.  the fraction B.B$)|g) ,t is monotone increasing in the covariance Covt(ei,t+i,e$,t+1) if and only if BM(i) > Вм($)£$,м;
  2. the fraction B.B$) E). t is always monotone decreasing in aiy£.

Items (a)-(b) of Proposition 6.1 directly translate into the following two empirical hy­potheses:

Hypothesis CS-1: The local currency share of corporate debt is higher for countries in which domestic inflation correlates more with US inflation when controlling for relevant factors.

Indeed, local currency debt partly replicates insurance properties of the dominant cur­rency in downturns, while it is a better hedge against domestic productivity shocks. In order to test this hypothesis, we proceed as follows. For each country i in our sample, we estimated the following time series regression:

П = 7o + Yi ■ ReturnM SCI ACWorldt + Г ■ Return^DomesticStockIndex] + nrtes’1, (19)

where n't is the domestic monthly inflation rate in country i and ReturnM SC IACWorldt is the monthly return on the MSCI AC World Index. Return-DomesticStockIndext is the monthly return on the domestic stock market index. щее’г are the residuals from this regression. We also run the following regression for the US:

Image sqto

We then run the following regression to compute a proxy for the covariance Covt(ei)t+1, e$)t+1) between the residual domestic inflation and residual US inflation (see item (a) of Proposition 6.1), where nrtes,г is the residual domestic monthly inflation rate in country i from (19) and п is the residual monthly inflation rate in the US from (20). We denote the estimated slope

We then run the following cross-sectional regression:

Image 25fr

Here, LCD. is the average ratio of debt denominated in local currency to debt denominated in dollars for corporates in the countries of the dataset; X. denotes other control variables. Item (a) of Proposition 6.1 predicts that the coefficient в1 in the regression (21) should be positive. The first three columns of Table 3 show that this is indeed the case. In column (1), we show the results of a univariate regression. In column (2), we add an additional control variable kaopen. : a financial openness index obtained from Chinn and Ito (2006). In column (3), we take the predictions of the model literally as they appear in item (a) of the Proposition 6.1: в1 > 0 for countries where LCU. > 1 and we exclude the two countries where

Image its6

Hypothesis CS-2: Firms in countries with more volatile domestic inflation tend to have less debt denominated in local currency.

Image xil

To test this hypothesis, we calculate the standard deviation of nrtes’1 as a proxy for a£>i in Proposition 6.1, and then run the following cross-sectional regression:

Image 5648

Proposition 6.1, item (b) predicts that в2 < 0. Column (4) of Table 3 shows the results of regression (22). Although the result is lacking statistical significance, the sign of the coefficient is indeed consistent with our theoretical prediction.

Image wwi3

 Notes: Robust standard errors in parentheses. *, **, *** denote significance at the 10, 5, and 1% levels respectively. LClJi is the mean share of local currency debt obtained from the IIF for each of the 17 emerging market economies between 2005 Q1 and 2018 of the Chinn-Ito financial openness index for each country. а)1"* is the standard deviation of the residuals of the monthly domestic inflation rate obtained from (19). In column (3), Hong Kong and Mexico are excluded.

Q2. (3it ’ * is the estimated regression coefficient for a linear regression of residuals of monthly domestic inflation rate from (19) on the residuals of the US inflation rate from (20). kaopeni is the mean

7. Optimal monetary policy

Our general equilibrium framework allows us to discuss the macroeconomic implications of a dominant currency debt equilibrium and the role of the central bank of that dominant currency country as the world’s central bank.

An active stabilization policy of the dominant currency country lowers ex-post the real debt burdens of firms through higher inflation and exchange rate depreciation. Therefore, this policy reduces the effective cost of issuing dominant currency debt, thereby prompting firms to take on a higher leverage ex-ante. However, higher leverage also means higher distress costs in the face of more severe shocks. Even though active monetary policy in economic downturns is optimal ex-post when a crisis state is realized, that policy is never optimal ex-ante. Namely, expected welfare gains from reducing the distress costs of firms are more than offset by the welfare costs of higher leverage. Central banks would prefer not to provide this insurance to firms ex-ante, but they cannot credibly do so.

We thus undertake the following thought experiment. Assume that the global central bank optimally assigns weights on the output gaps of different countries in order to maximize global welfare, taking into account all spillovers that arise from the interconnectedness of different countries due to global value chains. We then make the following assumption:

Image lble

The following is thus true.

Proposition 7.1 The welfare maximizing policy is to react only to the output gap in coun­tries with:

  • low TFP variance of ai>t
  • low importance in global trade,
  • low restructuring cost, 1 — Q

The global central bank chooses weights to precisely limit the leverage of firms in countries where the adverse effects of leverage are the highest as shown in Proposition 7.1. That in turn reduces leverage ex-ante and improves global welfare. Our results also have implications for the recent literature on the Global Financial Cycle. First, it might be optimal for the dominant currency country to respond to global economic conditions; and second, the dynamics of global expectations for the monetary policy of the dominant currency central bank might be as important as the monetary policy itself because it is these expectations that determine the ex-ante leverage of firms.

  1. Additional results and further evidence

    1. Dollar debt and international trade

Our model also makes predictions about the relationship between international trade and dollar-denominated debt. As we show in the Appendix (see the proof for Proposition 7.1), in the dominant currency debt equilibrium of Theorem 2.2, an increase in the coefficient ф of monetary policy effectiveness of the dominant currency country's central bank is always associated with (i) more issuance of debt denominated in the dominant currency; and (ii) a drop in the conditional expectations for the amount of international trade. This result is intuitive: An aggressive monetary policy provides incentives for firms to choose higher leverage, which ex post leads to more debt overhang and a drop in international demand. Thus, in the extended version of the model that is discussed in the previous section, shocks to ф$^ should move trade and the amount of debt denominated in the dominant currency in opposite directions. Figure 6 shows the joint dynamics of dollar-denominated debt and international trade over the last two decades. Consistent with our theory, the pre- and post-crisis trends in Figure 1 move-to-one with opposite trends in international trade.

Image la8i

1.2. Pound vs. dollar in the interwar years

One can use the expectations channel to shed light on the history of multiple, repeated

switches between the pound and the dollar and their roles in the main reserve currencies

Image wegn

during the inter-war period (see Chitu, Eichengreen and Mehl (2014)). Consider the two currencies (the pound and the dollar) with sufficiently similar indices (Figure 7). At the same time, the dollar faced greater deflation during the Great Depression, (1929-1939) with a subsequent partial regaining of dominance by the pound, based on the evidence provided by Chitu, Eichengreen and Mehl (2014).

9. Conclusion

We propose a “debt view” to explain the dominant international role of the dollar using an international general equilibrium model in which firms optimally choose the currency composition of their nominal debt. Theoretically, the dominant currency is the one that depreciates in global downturns at the horizons of corporate debt maturity (around six years). Empirically, the dollar fits this description, despite being a safe haven currency over horizons up to a year. Monetary policy is a key determinant in the choice of the dominant currency. Expansionary monetary policy in global downturns lowers real debt burdens of firms through its impact on inflation and exchange rates. Indeed, differences in inflation risk premia between the United States and the Eurozone can explain the fall and the rise of the dominance of the dollar.

What do our results imply for the future of the dollar? Many explanations of the dominant role of the dollar in the international monetary system feature arguments like inertia, size, network externalities, and market liquidity. All these arguments suggest that changes in the dominance status of a currency occur very slowly. By contrast, our results suggest that the dollar can lose its dominance if the expectations that the Federal Reserve is able to stimulate the economy and reduce real debt burdens of firms during global crises actually decline. As this view relies on the beliefs of market participants, this change might occur abruptly.

Our model can be extended in multiple directions. First, addressing the interactions between the role of the dollar in trade and finance may shed important light on endogenous inflation dynamics and the role of the dollar in trade invoicing. Second, modelling the demand for safe assets would help in understanding the role of the dollar for financial intermediation and household balance sheets. We leave these important questions for future research.


References

Aldasoro, Inaki and Torsten Ehlers, “Global liquidity: changing instrument and currency patterns,” Technical Report, BIS 2018.

Aldasoro, Inaki, Torsten Ehlers, and Egemen Eren, “Business Models and Dollar Funding of Global Banks,” BIS working papers, 2018, (708).

Baskaya, Yusuf Soner, Julian di Giovanni, Sebnem Kalemli-Ozcan, and Mehmet Fatih Ulu, “International Spillovers and Local Credit Cycles,” Working Paper 23149, NBER 2017.

BIS, “Statistical release: BIS global liquidity indicators at end-March 2018,” Technical Report, BIS 2018.

Bocola, Luigi and Guido Lorenzoni, “Financial Crises and Lending of Last Resort in Open Economies,” Technical Report, Northwestern University 2017.

__  and _____  , “Financial Crises, Dollarization, and Lending of Last Resort in Open Economies,” 2018.

Bodnar, K., “Survey evidence on the exchange rate exposure of Hungarian SMEs,” MNB Bulletin, 2006, 1 (1), 6-12.

Bruno, Valentina and Hyun Song Shin, “Global dollar credit and carry trades: a firm-level analysis,” Review of Financial Studies, 2017, 30 (3), 703-749.

__ , Se-Jik Kim, and Hyun Song Shin, “Exchange rates and the working capital channel

of trade fluctuations,” Working Paper, BIS 2018.

Caballero, Ricardo J. and Arvind Krishnamurthy, “Global Imbalances and Financial Fragility,” American Economic Review, 2009, 99 (2), 584-588.

Caballero, Ricardo J, Emmanuel Farhi, and Pierre Olivier Gourinchas, “An equilibrium model of global imbalances and low interest rates,” American Economic Review, 2008, 98 (1), 358-393.

Casas, Camila, Federico Diez, Gita Gopinath, and Pierre-Olivier Gourinchas,

“Dominant currency paradigm,” Technical Report, BIS 2017.

Cerutti, Eugenio, Stijn Claessens, and Andrew K Rose, “How important is the Global Financial Cycle? Evidence from capital flows,” Technical Report, BIS 2017.

Chahrour, Ryan and Rosen Valchev, “International Medium of Exchange: Privilege and Duty,” Technical Report, Boston College 2017.

Chinn, Menzie D. and Hiro Ito, “What Matters for Financial Development? Capital Controls, Institutions, and Interactions,” Journal of Development Economics, October 2006, 81 (1), 163-192.

Chitu, Livia, Barry Eichengreen, and Arnaud Mehl, “When did the dollar overtake sterling as the leading international currency? Evidence from the bond markets,” Journal of Development Economics, 2014, 111(C), 225-245.

Choi, Jaewon, Dirk Hackbarth, and Josef Zechner, “Corporate debt maturity profiles,” Journal of Financial Economics, 2018.

Chowdhry, Bhagwan, Richard Roll, and Yihong Xia, “Extracting Inflation from Stock Returns to Test Purchasing Power Parity,” American Economic Review, 2005, 95 (1), 255-276.

Craure, Benoit, “The euro’s global role in a changing world: a monetary policy perspec­tive,” speech at the Council on Foreign Relations, NYC, 2019.

Corsetti, Giancarlo and Paolo Pesenti, “Endogenous Exchange-Rate Pass-Through and Self-Validating Exchange Rate Regimes,” in Ricardo J. Caballero and Klaus Schmidt- Hebbel, eds., Economic Policies in Emerging-Market Economies Feschrift in Honor of Vittorio Corbo, Vol. 21 of Central Banking, Analysis and Economic Policies Book Series, Central Bank of Chile, 2015, chapter 11, pp. 229-261.

Cortina, Juan J, Tatiana Didier, and Sergio Schmukler, “Corporate debt maturity in developing countries: Sources of long and short-termism,” World Economy, 2018, Special Issue, 1-29.

Devereux, Michael B. and Shouyong Shi, “Vehicle Currency,” International Economic Review, 2013, 54 (1), 97-133.

Drenik, Andres, Rishabh Kirpalani, and Diego Perez, “Currency Choice in Con­tracts,” Technical Report, NYU 2018.

Du, Wenxin, Carolin E. Pflueger, and Jesse Schreger, “Sovereign Bond Portfolios, Bond Risks and Credibility of Monetary Policy,,” Working Paper 22592, NBER 2016.

ECB, “The international role of the euro,” Technical Report, ECB July 2017.

Engel, Charles, “Equivalence results for optimal pass-through, optimal indexing to ex­change rates, and optimal choice of currency for export pricing,” Journal of European Economic Association, 2006, 4 (6), 1249-1260.

Farhi, Emmanuel and Matteo Maggiori, “A Model of the International Monetary System,” Quarterly Journal of Economics, 2018, 133 (1), 295-355.

Favara, Giovanni, Erwan Morellec, Enrique Schroth, and Philip Valta, “Debt enforcement, investment, and risk-taking across countries,” Journal of Financial Eco­nomics, 2017, 123 (1), 22-41.

Ferrari, Massimo, Jonathan Kearns, and Andreas Schrimpf, “Monetary Policy’s Rising FX Impact in the Era of Ultra-Low Rates?,” Working Paper, BIS 2017.

Froot, Kenneth A., David S Scharfstein, and Jeremy C Stein, “Risk Management: Coordinating Corporate Investment and Financing Policies,” The Journal of Finance, 1993, 48 (5), 1629-1658.

Gabaix, Xavier and Matteo Maggiori, “International liquidity and exchange rate dynamics,” The Quarterly Journal of Economics, 2015, 130 (3), 1369-1420.

Gilchrist, Simon, Raphael Schoenle, Jae Sim, and Egon Zakrajsek, “Inflation Dynamics During the Financial Crisis,” American Economic Review, 2017, 107 (3), 785-823.

Goldberg, Linda and Cedric Kiel Tille, “A Bargaining Theory of Trade Invoicing and Pricing,” Working Paper 1839, Kiel Institute for the World Economy 2013.

__  and Cedric Tille, “Vehicle currency use in international trade,” Journal of Interna­tional Economics, 2008, 76(2) (2), 177-192.

Gomes, Joao, Urban Jermann, and Lukas Schmid, “Sticky Leverage,” American Economic Review, 2016, 106 (12), 3800-3828.

Gopinath, Gita, “International Price System,” paper prepared for the Jackson Hole Symposium, 2015.

__  and Jeremy C. Stein, “Banking, Trade, and the Making of a Dominant Currency,”

Working Paper, Harvard University 2018.

__ , Oleg Itskhoki, and Roberto Rigobon, “Currency Choice And Exchange Rate

Pass-Through,” The American economic review, 2010, 100, 304-336.

Gourinchas, Pierre-Olivier and Helene Rey, “From World Banker To World Venture Capitalist: US External Adjustment And The Exorbitant Privilege,” in “G7 Current Account Imbalances: Sustainability and Adjustment,” University of Chicago Press, 2007,

pp. 11-66.

__ , Nicolas Govillot, and Helene Rey, “Exorbitant Privilege And Exorbitant Duty,”

2010.

Hassan, Tarek A, “Country Size, Currency Unions, And International Asset Returns,” The Journal of Finance, 2013, 68, 2269-2308.

He, Zhiguo, Arvind Krishnamurthy, and Konstantin Milbradt, “A Model of Safe Asset Determination,” American Economic Review, 2019, 109 (4), 1230-1262.

Hordahl, Peter and Oreste Tristani, “Inflation Risk Premia in the Euro Area and the United States,” International Journal of Central Banking, 2014, 10, 1-47.

Hugonnier, Julien, Semyon Malamud, and Erwan Morellec, “Capital Supply Uncertainty, Cash Holdings, and Investment,,” Review of Financial Studies, 2015, 28 (2), 391-445.

Imbs, Jean, Haroon Mumtaz, Morten Ravn, and Helene Rey, “PPP Strikes Back: Aggregation and the Real Exchange Rate,” Quarterly Journal of Economics, 2005, 120 (1), 1-43.

Itskhoki, Oleg and Dmitry Mukhin, “Exchange Rate Disconnect in General Equilib­rium,” Working Paper 2017.

Ivashina, Victoria, David S Scharfstein, and Jeremy C Stein, “Dollar Funding And The Lending Behavior Of Global Banks,” The Quarterly Journal of Economics, 2015, 130 (3), 1241-1281.

Jiang, Zhengyang, Arvind Krishnamurthy, and Hanno Lustig, “Dollar Safety and the Global Financial Cycle,” Technical Report, Stanford GSB 2018.

Kang, Johnny and Carolin E. Pflueger, “Inflation Risk in Corporate Bonds,” Journal of Finance, 2015, pp. 115-162.

Maggiori, Matteo, “Financial Intermediation, International Risk Sharing, and Reserve Currencies,” American Economic Review, 2017, 107 (10), 3038-3071.

__ , Brent Neiman, and Jesse Schreger, “International currencies and capital alloca­tion,” Technical Report, Harvard University 2018.

__ , ____ , and _____  , “The Rise of the Dollar and Fall of the Euro as International

Currencies,” AEA Papers and Proceedings, 2019, 109.

Malamud, Semyon and Francesca Zucchi, “Liquidity, Innovation, and Endogenous Growth,” Journal of Financial Economics (forthcoming), 2018.

Matsuyama, Kiminori, Nobuhiro Kiyotaki, and Akihiko Matsui, “Toward a theory of international currency,” Review of Economic Studies, 1993, 60 (2), 283-307.

Miranda-Agrippino, Silvia and Helene Rey, “U.S. Monetary Policy and The Global Financial Cycle,” Technical Report, London Business School 2018.

Moore, Michael, Vladyslav Sushko, and Andreas Schrimpf, “Downsizing FX mar­kets: interpreting causes and implications based on the 2016 triennial survey,” BIS Quarterly Review, December 2016, 2016.

Mukhin, Dmitry, “An Eqilibrium Model of the International Price System,” Technical Report, Princeton University 2017.

Niepmann, Friederike and Tim Schmidt-Eisenlohr, “Foreign currency loans and credit risk: Evidence from U.S. banks,” Technical Report, Federal Reserve Board 2017.

Praet, Peter, “The ECB’s fight against low inflation: reasons and consequences,” 2016.

Rampini, Adriano, Amir Sufi, and S. Viswanathan, “Dynamic risk management,” Journal of Financial Economics, 2014, 111, 271-296.

Rey, Helene, “International trade and currency exchange,” Review of Economic Studies, 2001, 68 (2), 443-464.

Rey, Helene, “Dilemma Not Trilemma: The Global Cycle And Monetary Policy Indepen­dence,” in “Proceedings-Economic Policy Symposium-Jackson Hole” Federal Reserve Bank of Kansas City 2013.

Salomao, Juliana and Liliana Varela, “Exchange Rate Exposure and Firm Dynamics,” Working paper, University of Minnesota 2018.

Shin, Hyun Song, “Global Banking Glut And Loan Risk Premium,” IMF Economic Review, 2012, 60, 155-192.

Velandia, Antonio and Rodrigo Cabral, “Why are more sovereigns issuing in Euros? Choosing between USD and EUR-denominated bonds,” World Bank Group, 2017.

Image xe32

Image b00a

Image ch11

Image peo

Image ifl0

Image g76v

Image tpga

Image ozx9

Image 2inb

Image 2inb

Image q138

Image nhoj

Image nhoj

Image 27be

Image mvxj

Image ba0q

Image lopa

Image lopa

Image n300

Image yrcj

Image x7kq

Image dbry

Image wffk

Image ttb7

Image pouz

Image 2znc

Image k4ri

Image 8u2g

Image vze

Image 494t

Image r5va

Image y4xz

Image u9q4

Image k5ro

Image zadk

Image e31j

Image 7fuy

Image b7si

Image ar3l

Image e0v4

Image gy7h

Image 86tn

Image 62gl