The accumulation of foreign exchange by central banks: Fear of capital mobility?

The accumulation of foreign exchange by central banks: Fear of capital mobility?

Journal of Macroeconomics 38 (2013) 409–427 Contents lists available at ScienceDirect Journal of Macroeconomics journal homepage: www.elsevier.com/l...

752KB Sizes 0 Downloads 9 Views

Journal of Macroeconomics 38 (2013) 409–427

Contents lists available at ScienceDirect

Journal of Macroeconomics journal homepage: www.elsevier.com/locate/jmacro

The accumulation of foreign exchange by central banks: Fear of capital mobility? Andreas Steiner ⇑ University of Osnabrueck, Institute of Empirical Economic Research, 49069 Osnabrueck, Germany

a r t i c l e

i n f o

Article history: Received 4 March 2013 Accepted 2 August 2013 Available online 20 August 2013 JEL classification: E58 F31 F41 Keywords: International reserves Capital mobility Macroeconomic trilemma

a b s t r a c t Central banks’ foreign exchange holdings have increased significantly in the recent past. This article explains this development as a result of the liberalisation of international capital markets. First, central banks accumulate reserves in order to protect the economy from detrimental effects of sudden stops in capital flows and flow reversals. Second, central banks use the accumulation of reserves as a substitute for capital controls. Changes in reserves are a form to manage net capital inflows. They permit the central bank to preserve some leeway for an independent monetary and financial policy despite the classic policy trilemma. The empirical analysis of a large panel data set supports the hypothesis that the accumulation of reserves is the consequence of a ‘‘fear of capital mobility’’ suffered by central banks. Ó 2013 Elsevier Inc. All rights reserved.

1. Introduction Recent cyears have witnessed an enormous increase in central banks’ foreign exchange holdings. Whereas average holdings amounted to 6.2% of GDP in 1970, they reached an unprecedented level of 22.6% of GDP in 2010 in a sample of 180 countries. This increase is a puzzle for the literature on the demand for international reserves. Since at the same time exchange rates have become more flexible and countries more integrated in the international capital market, standard theory predicts a decline in the demand for foreign exchange. The existing literature usually explains the demand for foreign exchange as a buffer stock to defend the exchange rate. Whereas traditional approaches argue that reserves are needed to finance imbalances in the balance of payments under a fixed exchange rate system, the more recent literature, which emerged after the series of financial crises during the 1990s, focuses on the stock of reserves, which is seen as a lifejacket against financial crises.1 Both approaches coincide in the view that there exists an adequate level of reserve holdings, which is the outcome of an optimising behaviour of the central bank. This article extends the latter strand of the literature: It explains the accumulation of reserves as a side effect of the opening of national capital markets and, more particularly, of the integration of emerging and developing economies in the world capital market. According to this hypothesis, central banks suffer from a ‘‘fear of capital mobility’’. The accumulation of

⇑ Tel.: +49 541 9692556; fax: +49 541 9692757. E-mail address: [email protected] See Heller (1966) and Frenkel and Jovanovic (1981) for major contributions to the traditional approach. Aizenman and Lee (2007), Cheung and Qian (2009) and Jeanne and Rancière (2011) are examples of the more recent strand of the literature focusing on crises and competitive hoarding. 1

0164-0704/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jmacro.2013.08.002

410

A. Steiner / Journal of Macroeconomics 38 (2013) 409–427

foreign exchange is a response to capital inflows. It aims at reducing the interdependence of an open economy from developments in the rest of the world. This fear of capital mobility may manifest itself in two different forms: First, it is hypothesised that a central bank’s reserves increase in the degree of capital mobility. The motive for this behaviour might be the desire to protect the economy from potentially detrimental effects of sudden stops in capital flows and flow reversals. This idea deepens the evidence provided in the literature on precautionary reserve hoardings2: We explicitly distinguish between effects of de jure and de facto capital mobility. We show that (1) de facto capital mobility drives the positive relationship with reserves and (2) de jure capital mobility leads to an additional albeit less robust positive effect. The second form of fear of capital mobility is demonstrated by the management of capital inflows. A central bank might accumulate reserves in order to absorb net capital flows in the absence of capital controls. The management of capital inflows allows the central bank to preserve some leeway for the conduct of an independent monetary policy despite the classic policy trilemma. Furthermore, the central bank can limit the real effects of capital inflows, which might interfere with domestic policy objectives. In contrast to precautionary hoardings where the level of reserves matters the relevant variable for capital flow management is the change in reserves. Our empirical finding that changes in reserves are linked to contemporaneous capital inflows adds a complementary aspect to the literature on reserve accumulation: Reserve accumulation may be a by-product of the desire to manage capital flows in the absence of direct controls. In the case of precautionary reserve accumulation, the central bank supports the open capital account. According to the capital flow management hypothesis, the central bank intends to insulate domestic economic policy from the world capital market under a fixed or managed exchange rate. In either case the accumulation of reserves can be regarded as an instrument to manage capital flows.3 The fear of capital mobility – namely the tendency to accompany the removal of capital controls and the increasing volume of net capital flows by the accumulation of international reserves with the intention of combining an independent monetary policy with a stable exchange rate – is linked to other concerns of central banks that have been identified before: There is ‘‘fear of floating’’ (Calvo and Reinhart, 2002) implying that countries announce de jure floating exchange rates but de facto continue to stabilize them. Levy-Yeyati and Sturzenegger (2007) document a ‘‘fear of appreciation’’ in the sense that central banks intervene in the foreign exchange market to depreciate the exchange rate or postpone its appreciation. As a consequence of their desire to manage exchange rates, they fear capital mobility because it narrows their freedom of choice within the policy trilemma, and by implication, complicates exchange rate management. The accumulation of reserves may be used as an instrument to relax the restriction of the trilemma in the short run. The major contribution of this paper is twofold: First, we develop the concept of ‘‘fear of capital mobility’’ and describe its relation to the recent period of reserve accumulation. Second, we empirically confirm that international reserves are used to mitigate the effects of capital mobility and, in turn, appease central banks’ fear. Before proceeding, we would like to note one limitation: Our analysis is not suited to explain the Chinese reserve hoardings. China has accumulated an enormous amount of reserves although its capital account has remained relatively closed. A high national savings rate combined with an exchange rate policy of a fixed rate may explain the Chinese story. While China’s absolute amount of reserves is outstanding, the policy of reserve accumulation, however, is a global one: Between 2000 and 2010 each year 69% of the 180 countries of our sample have increased their reserves relative to GDP. This paper attempts to provide an explanation for this general phenomenon. The article is organised as follows. Section 2 describes the hypothesis that central banks suffer from a fear of capital mobility. Section 3 presents the data, discusses different measures of capital mobility and shows statistical evidence in support of the hypotheses. Section 4 presents and discusses the empirical results. The final section concludes. 2. The hypothesis: Fear of capital mobility The following section describes the hypothesis that central banks suffer from a fear of capital mobility. This fear of capital mobility arises in two different forms: First, central banks fear the openness of the capital account and, second, manage private capital inflows. 2.1. Capital mobility and the level of reserves: Self-insurance Financial liberalisation and economic globalisation both allow a country to benefit from international capital flows. However, they also make countries more vulnerable to sudden stops and capital flow reversals. Therefore, a central bank might back an open capital account by precautionary measures in the form of foreign exchange hoardings. On theoretical grounds, the effect of capital account liberalisation on the level of reserves is ambiguous. On the one hand, access to external credit sources reduces the importance of reserves in financing international transactions. Any balance of the current account can, at least theoretically, be counterbalanced by proportionate private capital flows. On the other hand, 2

See, among others, Aizenman and Lee (2007), Ghosh et al. (2012) and Obstfeld et al. (2010). In a similar vein, Ostry et al. (2012) argue that foreign exchange market intervention is – besides monetary policy – an instrument of central bank policy, which can be used to affect the exchange rate in an inflation-targeting framework, where interest rates are set to achieve the inflation target. The management of exchange rates is especially important in economies with a currency mismatch in foreign assets and liabilities. 3

A. Steiner / Journal of Macroeconomics 38 (2013) 409–427

411

open capital markets increase the exposure to external financial disturbances and speculative flows. Especially emerging and developing countries may be affected by sudden stops of capital flows and capital flight. The risk of capital flight originates from two sources: First, foreign investors might suddenly withdraw their capital invested in the domestic economy. Second, domestic agents might prefer to invest their wealth in foreign currency (see Obstfeld et al., 2010). This form of currency substitution may be restricted to currency in circulation or can comprise bank deposits. These savings, in turn, can be deposited at domestic or foreign commercial banks. Capital flight also increases the risk of speculative attacks and ensuing currency crises. Capital mobility has two dimensions: de jure and de facto capital mobility. De jure capital mobility corresponds to the extent and nature of regulations governing capital account transactions. It can be measured by the existence of legal restrictions of cross-border capital flows. It is a pre-condition for financial integration. De facto capital mobility refers to the actual mobility of capital measured by stocks and flows of cross-border assets and liabilities. Central banks might fear both forms of capital mobility: Countries that have liberalised their capital account but whose cross-border capital flows and stocks are limited, might fear that domestic investors prefer to invest abroad, leading to large capital outflows. Countries with substantial financial linkages to the rest of the world might fear a double outflow of capital, namely of domestic and foreign investors. This reasoning leads to the following hypotheses, which shall be tested empirically in Section 4: (1) The higher the degree of capital mobility, the more foreign exchange a central bank hoards. This hypothesis can be split into two different lines of argumentation: (1a) Central banks fear de jure capital mobility independently of a country’s actual degree of integration in the international capital market. (1b) Central banks fear de facto capital mobility, namely a country’s financial integration. They accumulate foreign exchange in order to protect the economy from potential sudden stops of capital flows and capital flight. 2.2. Capital mobility and changes in reserves: Management of capital inflows This section is devoted to the hypothesis that central banks actively manage capital flows.4 Both capital controls and changes in reserves allow a central bank to manage capital inflows. Changes in reserves are evidence for a ‘‘fear of capital mobility’’ if they are a counter movement to the removal of capital controls. 2.2.1. Foreign exchange accumulation as a substitute for capital controls Changes in foreign exchange reserves can be regarded as an imperfect substitute for capital controls because both have to a certain extent the same macroeconomic effects.5 In the following paragraphs different theoretical approaches are used to demonstrate these common effects. 2.2.1.1. Balance of payments implications. The balance of payments restriction implies that both capital controls and the accumulation of reserves (ceteris paribus) may lead to an increase in the current account balance. If private capital is perfectly mobile and domestic and foreign assets are perfect substitutes, the accumulation of reserves, which is a form of official capital export, will be offset by an additional import of private capital. Hence, the capital account balance including official reserve changes is unaffected while the gross sum of capital flows increases. In the case of imperfect capital mobility, the balance of payments restriction implies that the accumulation of foreign exchange increases the current account balance. To increase the current account balance, the exchange rate has to depreciate or – in the case of a fixed nominal exchange rate – domestic prices have to fall. In any case, the current account is distorted towards less domestic investment and consumption. Exports are driven up and imports are depressed. If, in turn, controls are used to reduce capital inflows, they distort the current account towards a surplus because controls restrict the supply of external credit by foreigners. So capital controls and the accumulation of foreign exchange are substitutes in the sense that both allow the government to increase the current account balance. Both lead to smaller net capital inflows and reduce the pressure towards an appreciation of the domestic currency. This might be in the interest of the central bank if a fixed nominal exchange rate regime is in place, if the economy pursues a development strategy of export-led growth or if a reallocation of resources towards the nontradable sector is not desired. 2.2.1.2. Implications for the domestic money market. The comparable effects of capital controls and reserve accumulation in the face of capital inflows also become evident on the domestic money market. Both cushion it from the effects of foreigners demanding domestic currency. Net capital inflows imply an increase in the supply of foreign currency and an 4 With respect to capital flows, the following definitions apply throughout this article. Net capital flows are the difference between capital inflows and capital outflows. If this difference is positive, it is also called net capital inflows. Capital flows are always measured as private flows excluding changes in official reserves. The empirical literature distinguishes two concepts of capital flows: In the terminology of the IMF (see World Economic Outlook) total net capital flows comprise direct investment, portfolio investment and other long- and short-term investment flows. In the standard balance of payments presentation total net capital flows are equal to the balance on the financial account minus the change in reserve assets. Other publications (see UNCTAD, 1999, p. 100) define net capital flows as the sum of the balance on capital and financial accounts, that is to say they additionally include capital account transactions like debt forgiveness, official grants and migrants’ transfers as well as the acquisition or disposal of nonproduced, nonfinancial assets (patents, trademarks, etc.). For the purpose of this article the first concept is more appropriate because it focuses on the financial account, which is the origin of the volatility of capital flows. Capital account transactions, in contrast, are fairly stable or even counter cyclical. 5 See Section 2.2.2 for a discussion of the features that make substitutability imperfect.

412

A. Steiner / Journal of Macroeconomics 38 (2013) 409–427

Exchange rate stability

nc yb oa

nd

on

al

tre

at i ri z

ntr

al

lla

Ce

ob

Do rd /

Re

rre

se r C--a- ve a p--it-- cc a-l-- um c-o-- ul n-tr-- ati o-l-s- on

Cu

Monetary policy autonomy

Gl

ba

nk

Floating exchange rate

Freedom of capital movement

Fig. 1. Reconciliation of the classic policy trilemma in the short run.

increase in the demand for domestic currency. According to the monetary approach to the exchange rate, this relative increase in the demand for domestic currency leads to an appreciation of the exchange rate (see for example Frenkel (1976)). Capital controls, which decrease capital inflows, reduce directly the demand for domestic currency by foreigners and limit the supply of foreign currency. The accumulation of reserves satisfies the foreign demand for domestic currency: The central bank absorbs the increased supply of foreign currency and issues domestic money in turn. The domestic money market remains in equilibrium for a given price level. Both policies result in a more depreciated exchange rate than would be the equilibrium outcome without intervention.

2.2.1.3. Implications for macroeconomic policies: Reconciliation of the policy trilemma. The relationship between capital mobility and reserve accumulation can be analysed in the framework of the macroeconomic policy trilemma (see Fig. 1). The trilemma imposes a constraint on the choice of macroeconomic policies. It states that the objectives of exchange rate stability, monetary independence and capital mobility are mutually inconsistent. Only two out of these three possible objectives can be attained jointly. Recent empirical evidence in support of the trilemma is provided by Obstfeld et al. (2005), Popper et al. (2011) and Aizenman et al. (2010, 2011). The latter show empirically that a move towards one goal of the trilemma induces a shift away from at least one of the other two policy objectives. They note that the accumulation of reserves may be related to the changing configuration of the trilemma over time. The policy trilemma, however, constrains economic policy only in the long run. All three objectives are jointly attainable in the short run if they are supported by accompanying policies. Changes in reserves are such a policy to reconcile the trilemma. To illustrate the mechanism, consider the example of a country that opts for an independent monetary policy and capital mobility. According to the trilemma, the exchange rate cannot be stabilized. This, however, is not true in the short run. Assume that the given interest rate leads to an outflow of capital such that the exchange rate tends to depreciate. The central bank can stabilize the exchange rate if it counteracts the outflow of capital by the sale of foreign exchange reserves. This is the policy of exchange rate defence through an exchange market intervention.6 It allows to achieve all three objectives of the trilemma until reserves reach their lower bound if domestic and foreign bonds are imperfect substitutes. Whereas this policy configuration is extensively analysed in models of currency crises and speculative attacks, the opposite case of a foreign exchange intervention aimed at preventing an exchange rate appreciation has received less attention.7 According to the trilemma, an independent monetary policy and a fixed exchange rate system are incompatible in the presence of net capital inflows. However, if the central bank absorbs the capital inflow with the accumulation of reserves, it can reconcile an open capital account with an independent monetary policy and a fixed exchange rate.8 Basically, the central bank buys the excess supply of foreign assets and replaces private investors whose demand is short of supply at the given exchange rate. 6 It is assumed that the interest rate remains at its level consistent with domestic goals of monetary policy. Alternatively, the central bank might raise the interest rate to defend the exchange rate peg. Then, however, it would subordinate monetary policy to the objective of a stable exchange rate. 7 Levy-Yeyati and Sturzenegger (2007) and Korinek and Serven (2010) examine the relationship between reserve accumulation and the undervaluation of an exchange rate. Adler and Tovar (2011) show that sterilized exchange market interventions were effective to slow the pace of appreciation in the 2000s. 8 This mechanism can be illustrated in the framework of the portfolio balance model (refer to, among others, Branson (1977), Henderson and Rogoff (1982), Kouri (1983) and Blanchard et al. (2005)).

A. Steiner / Journal of Macroeconomics 38 (2013) 409–427

413

In general, if the central bank absorbs capital inflows and satisfies capital outflows through proportional changes of reserves, it can neutralise the effects of an open capital account. This is possible if assets are imperfect substitutes or if capital mobility is restricted. In comparison to the opposite policy of exchange rate defence in the face of capital outflows, this policy has the merit that there exists no upper bound of reserves above which this policy is no longer feasible. The recent experience of reserve accumulation has shown that this policy is sustainable over an extended time period. Since this article focuses on the explanation of the recent period of reserve accumulation, one has to raise the question whether the choices countries have made with respect to the trilemma in the recent past might have favoured systematic central bank interventions to prevent the exchange rate from appreciating. Reserve levels would be stable in the long run if central banks intervened symmetrically, preventing depreciations and appreciations of the exchange rate alike. It is, however, likely that interventions to prevent appreciation are dominant during our period of consideration: (1) In the aggregate, emerging and developing countries registered net private capital inflows in this period. (2) Countries might be reluctant to use their reserves during crises making reserve losses less probable. Aizenman and Sun (2012) confirm a ‘‘fear of losing reserves’’ for emerging markets during the financial crisis in 2008 and 2009. In summary, foreign exchange interventions provide a government with some leeway for domestic monetary policy in spite of exchange rate fixity. This leads to the second hypothesis: The accumulation of foreign exchange is a response of central banks to the removal of capital controls and the inflow of foreign capital. Central banks aim at managing capital inflows.

2.2.2. Explanations for the substitution of capital controls by foreign exchange interventions This hypothesis, however, raises new issues: Why have governments abolished controls on capital movements albeit they still have an interest in managing capital flows? Possibly they had to liberalise their capital account due to conditions associated with IMF lending or external consultants advised them to do so. Joyce and Noy (2008) find empirical evidence that the participation in an IMF programme during the 1990s is correlated with capital account liberalisation. Countries could might have followed the trend (bandwagon effects) and liberalised their capital account after neighbouring countries or members of their peer group had done so.9 When neighbours relax capital controls it becomes harder to justify them politically and economically given that financial liberalisation has been found to increase growth (see Honig, 2008). With respect to their appeal for foreign investors, countries that resist the tendency to relax capital controls fall behind countries that do remove controls. Finally, the relaxation of capital controls could be the result of some kind of bargaining between industrialised and emerging economies where the first loosened their trade barriers and, in return, the second had to open up their capital markets. The removal of capital controls can be in the proper interest of the central bank. The abandonment of capital controls means that the central bank loses one of its instruments of financial policy. However, there are several reasons why a central bank could be willing to give up capital controls anyway. First, capital controls are not an instrument that can be set independently by the central bank; on the contrary, the imposition of capital controls either has to be explicitly permitted by the central bank’s statute or be specified by other laws, which, in turn, have to be approved by government. In most cases, the central bank constitution defines a maximum percentage that can be required as mandatory reserves from capital importers. So, in principle, the central bank only enforces and administers the pre-defined capital controls. The accumulation of reserves, in contrast, is a policy that is independently set by the central bank. Capital controls are an inflexible instrument whereas the accumulation of reserves can be adjusted easily and without time lag to changing economic conditions and objectives of financial policy. Both capital controls and reserve accumulation are costly distortions but differ in the distribution of these costs: Whereas the cost of capital controls have to be born by borrower and lender, the costs of foreign exchange reserves fall on the society as a whole. Capital controls can lead to revenues for the central bank (for example if unremunerated reserves have to be held at the central bank), whereas reserves generally entail quasi-fiscal costs because of the interest differential between domestic bonds and bonds denominated in the reserve currency.10 In sum, central banks are financially worse off if capital controls are replaced by foreign exchange reserves. However, this might not bother central bankers. The economic theory of bureaucracy assumes that public officials are primarily interested in their power, prestige and independence (see Vaubel, 1997), but not in the profits of their institution. In fact, large reserve holdings may be preferred by a central bank because they increase its power and independence from government. The degree of substitutability between both forms of capital inflow management is imperfect because the accumulation of reserves cannot accomplish all effects, which are obtained (or at least intended) by capital controls. For instance, whereas capital controls, in the form of a fixed-term unremunerated reserve requirement, aim at changing the composition (from portfolio to direct investment flows) and maturity structure of capital inflows, the accumulation of foreign exchange cannot bias the nature of capital inflows. Moreover, whereas capital controls can be designed to target specific capital flows

9 Recent studies (e.g. Gassebner et al. (2011) and Simmons and Elkins (2004)) show empirically that countries are more likely to open their capital account when members of their peer group have done so. 10 This argumentation assumes that the effect of the accumulation of reserves on the domestic monetary base is sterilized by the issue of domestic bonds.

414

A. Steiner / Journal of Macroeconomics 38 (2013) 409–427

(prohibitions, requirement of special permission for pre-defined types of flows), the accumulation of reserves cannot prohibit certain kinds of capital movements. In this sense, the accumulation of reserves is a simple instrument that cannot be designed to achieve specific objectives concerning the nature of capital flows. 3. Data and descriptive statistics 3.1. Description of the data The empirical analysis is carried out on the basis of a pooled data set of cross-country and time-series observations. It contains annual data from 1970 to 2010. Since data for several explanatory variables are missing for some countries, the number of countries used in the econometric analysis depends on the particular specification and is indicated in the respective tables. It ranges from 87 to 180 countries. With a few exceptions data are taken from the International Financial Statistics of the IMF and the World Development Indicators of the World Bank. A detailed description of the sample, variables and their data sources can be found in Appendices A and B.

3.2. Traditional control variables Concerning the set of control variables, we follow the existing literature of the demand for reserves (Aizenman and Lee, 2007; Lane and Burke, 2001; Obstfeld et al., 2010). To control for the stage of development, output per capita is included. Since less developed countries are more affected by sudden changes in the flows of goods and capital, they are expected to hold more reserves. However, they might be unable to afford the desired level of reserves. Therefore, the expected impact of the stage of development on reserves is ambiguous. Trade openness is included to control for the effects of real linkages with other economies. The more open the economy, the more vulnerable it is to external shocks and is expected to hold more reserves. In line with this precautionary motive, higher volatility measured as the standard deviation of the previous years of the growth rate of exports is also expected to be associated with a higher level of reserves. External debt is another source of vulnerability. Empirical studies show that both a high level of external debt and a low level of reserves increase the probability of a financial crisis. Reserves might offset this vulnerability. Therefore, it is expected that countries with a high level of external debt hold more reserves for precautionary reasons. Additionally, short-term external debt is included. Finally, different types of exchange rate regimes are accounted for by the inclusion of dummy variables. The more flexible the exchange rate, the less reserves are needed for its management.

3.3. Measures of capital mobility The empirical analysis uses different measures of capital mobility, which allow to distinguish between the effects of de jure and de facto capital mobility. Most indices of de jure capital account openness are based on the information provided in the IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER). It offers a binary variable that informs about the existence of restrictions in different categories of international financial transactions. Chinn and Ito (2002, 2006) provide an index of capital mobility, which embodies four binary dummy variables on restrictions taken from the AREAER. In the empirical analysis I refer to this index as de jure capital mobility (Chinn-Ito). An alternative index of de jure capital mobility is provided by Edwards (2007). He combines the information of the indices of Quinn (2003) and Mody and Murshid (2005), which are based on data from the IMF. Country-specific information is used to revise and refine the index. Since the index provides data only until the year 2000, regressions including the index cover a reduced period ending in 2000. This variable is called de jure capital mobility (Edwards) in the empirical analysis. A measure of de facto capital mobility is constructed from data on external capital stocks. As proposed by Lane and MilesiFerretti (2007) an index is given by the sum of total external assets and total external liabilities as a proportion of GDP. This index measures financial integration of an economy with the rest of the world. Finally, we use an index of economic globalisation. It is a sub-index of the KOF index of globalisation proposed by Dreher (2006). The index of economic globalisation has two main components, which are weighted equally: actual flows of goods and capital and restrictions to these flows. Hence, this index combines information of de jure and de facto capital mobility with information of trade openness. Fig. 2 illustrates the evolution of capital mobility over time. For all indices higher values indicate that countries are more open to cross-border financial transactions. All four measures capture a trend of increasing capital mobility, which started in the mid 1980s. This trend was temporarily halted by the Mexican Tequila crisis in 1994 and the East Asian financial crisis. Countries have removed capital controls and have faced increasing international capital flows. This trend is observable in industrial, emerging and developing countries alike (see right-hand panel of Fig. 2). Whereas on average industrial countries are in every single year more open than the other two country groups, capital mobility in emerging markets does not differ

415

65

4

60

2

55

0

50

-2

45

-4

40

2

Index number

6

Index number

Index number

A. Steiner / Journal of Macroeconomics 38 (2013) 409–427

1.5 1 .5 0 -.5 -1

1970 1975 1980 1985 1990 1995 2000 2005 2010

1970 1975 1980 1985 1990 1995 2000 2005 2010

Year

Year

De jure capital mobility (Chinn-Ito) De jure capital mobility (Edwards)

Note:

Financial integration Economic globalisation

Industrial countries Emerging markets

Developing countries

The left-hand panel visualizes the evolution of different measures of capital mobility over time. The scale on the left-hand side axis corresponds to the Chinn-Ito index (multiplied by ten) and the measure for financial integration. For the index of Edwards and economic globalisation the right-hand axis applies. The right-hand panel shows the evolution of the Chinn-Ito index for different country groups over time. Fig. 2. Measures of capital mobility.

Billions of US$

1500

1000

500

0 1970

1980

1990

2000

2010

Year Current account balance Change in reserves

Net private capital flows

Data sources: IMF (2012) and World Bank (2011a) Notes: A positive sign in the change of reserves indicates an increase in foreign exchange holdings. Net private capital flow is equal to the balance of the financial account.

Fig. 3. Balance of payments of emerging and developing countries.

much from that in developing countries. However, emerging markets are characterised by a higher volatility of capital mobility over time.

3.4. Statistical evidence Since the balance of the current account equals the change in a country’s net foreign assets, a deficit (surplus) in the current account has to be offset by net capital inflows (outflows). If these flows are private in nature, they figure in the financial account balance. Alternatively, if the central bank sustains the current account balance, it accumulates or sells international reserves, which are official capital flows. In the case that a central bank does not intervene in the foreign exchange market or intervenes only temporarily to lean against the wind, the current account and the capital account excluding reserve changes are the main components of the balance of payments. Changes in official reserves are only a residual entry that brings about the overall balance. This picture has changed dramatically since the Asian financial crisis (see Fig. 3). The current account deficits of developing countries as a group have been replaced by surpluses since 1999. At the same time, these countries still have registered net private capital inflows (excluding changes in reserves). This joint incidence of net private capital inflows and a current account surplus is somewhat unusual. Capital inflows cannot be explained as the counterpart of a current account deficit, that is to say, they do not finance the current account deficit. The puzzle can be solved when changes in official reserves are added. In fact, when total capital flows including changes in reserves are considered, the group of developing countries has registered net financial outflows since 1999. Hence, they have increasingly become net exporters of capital. Central bank intervention reverses the direction of net capital flows.

416

A. Steiner / Journal of Macroeconomics 38 (2013) 409–427

Table 1 Capital inflows and change in reserves. Capital inflow (in billions of US$)

Change in reserves (in billions of US$)

Change in reserves as percentage of capital inflows

1970–79

World Emerging-market countries Developing countries

1295 122.6 223.3

296 44 130

22.9 35.6 58.4

1980–89

World Emerging-market countries Developing countries

5207 214.6 374.9

339 42 36

6.5 19.6 9.7

1990–99

World Emerging-market countries Developing countries

15114 1611 1946

996 495 709

6.6 30.7 36.5

2000–2010

World Emerging-market countries Developing countries

54162 5720 8879

7465 4584 6136

13.8 80.1 69.1

Index number

1.0

0.5

0.0

-0.5

1970

1975

1980

1985

1990

1995

2000

2005

2010

Year Mean index over all countries Mean index over ten largest reserve accumulaters

Fig. 4. Index of de jure capital mobility (Chinn-Ito) and reserve accumulation.

Table 1 shows the absolute values of capital inflows11 and changes in reserves over different time periods and for different country groups. The variable of primary interest is the change in reserves expressed as a percentage of capital inflows. In the 1970s, changes in reserves accounted for a relatively large part of capital inflows. This is primarily due to the low level of cross-border capital flows during this period. Since the second wave of capital account liberalisation, which took place in the late 1980s and early 1990s, this has changed fundamentally: In the 1980s, capital flows increased such that the accumulation of reserves only offset a minor part of capital inflows (between 6.5% in the world and 19.6% in emerging market countries). Since then, these figures have increased steadily. Between 2000 and 2010, the majority of capital inflows to emerging and developing countries was reversed via the accumulation of reserves, namely 80.1% and 69.1%, respectively. This means that in developing countries only 30.9% of capital inflows (emerging markets 19.9%) could be used for domestic investment. Central banks in emerging and developing countries increasingly offset net capital flows. Both country groups show the same pattern, which, however, is more pronounced in emerging markets. This is also first evidence that central banks replaced capital controls by a policy of reserve accumulation, thereby still pursuing the objective of regulating capital flows. In comparison with developing countries, emerging markets as a group are characterised by both less capital account restrictions (see Fig. 2) and a larger extent of capital inflow management. Fig. 4 shows the Chinn-Ito index of capital mobility. It compares the average value of the index over all countries with its mean value calculated for the ten countries with the largest absolute increase in reserves over the period 1990–2010.12 It is striking that these ten countries have been characterised by a significantly higher degree of de jure capital mobility than the average country. This is first evidence that countries with loose capital controls tend to hoard reserves. Only around the Mexican Tequila crisis, capital mobility of the reserve accumulators was similar to that of an average country. Finally, Fig. 5 presents some country examples that illustrate the effects of a removal of capital controls. The graphs show the time-series of capital inflows, reserve changes and de jure capital mobility (Chinn-Ito index) for Argentina, Brazil, the

11 In this table, capital inflows are defined as the change in external liabilities of the recipient economy. External liabilities comprise FDI in the reporting economy, portfolio equity liabilities and other investment liabilities. 12 These ten countries – beginning with the largest accumulator – are China, Japan, Saudi Arabia, Brazil, Korea, India, Hong Kong, Singapore, Switzerland and Algeria. This list shows that the phenomenon of reserve accumulation is present across continents and across different levels of development.

417

A. Steiner / Journal of Macroeconomics 38 (2013) 409–427

Argentina

Brazil

10 0.0 0 -1.0

0.0 100

-0.5 -1.0

50

-1.5

-10

0 -2.0

-2.0

1970 1975 1980 1985 1990 1995 2000 2005 2010

1970 1975 1980 1985 1990 1995 2000 2005 2010

Year

Year

Philippines

10

0.5

5

0.0

0

-0.5

-5

-1.0 1970 1975 1980 1985 1990 1995 2000 2005 2010

Year Capital inflows De jure capital mobility (Edwards)

200

0.0

150 -0.5 100 -1.0

50

Index

1.0

Billions of dollars

Russia

15

Index

Billions of dollars

Index

1.0

Billions of dollars

20

0.5

150

2.0

Index

Billions of dollars

30

-1.5

0 -50

-2.0 1970 1975 1980 1985 1990 1995 2000 2005 2010

Year Change in reserves

Capital inflows De jure capital mobility (Edwards)

Change in reserves

Fig. 5. Capital account openness, capital flows and reserve accumulation: Country examples.

Philippines and Russia. Despite some downward outliers in crisis years, capital inflows increased after the liberalisation of the capital account. More importantly, changes in reserves increased simultaneously. Especially since the East Asian financial crisis, reserve changes seem to mimic capital inflows. Countries learnt their lessons: In the run-up to crises, the Philippines (until 1996) and Argentina (until 2002) faced large capital inflows but did not accumulate reserves at the same pace. Since then, central banks pace of reserve accumulation seems to be a function of capital inflows. These country cases may be regarded as first evidence that the accumulation of foreign exchange and capital controls are substitutes. 4. Regression analysis The remaining task consists in testing empirically whether the degree of capital mobility has an effect on the level of reserves (Hypothesis 1) and whether capital inflows are managed through changes in reserves (Hypothesis 2). The dependent variable international reserves is measured net of gold holdings and scaled by GDP. Scaling is motivated by the fact that a comparison of reserve levels across countries has to take differences in countries’ economic size into account. As a by-product, scaling by GDP secures that the dependent variable is stationary.13 4.1. Capital mobility and the level of reserves: Self-insurance We first test the hypothesis that foreign exchange holdings are the larger, the higher the degree of capital mobility is (Hypothesis 1). To this end, we use both static and dynamic regression specifications. 4.1.1. Static regressions We first use the static fixed effects estimator to examine correlations in our cross-country time series data.14 In particular, we estimate the following relationship:



 IR ¼ a  X it þ b  CapOpenit þ ci þ eit GDP it

ð1Þ

where IR represents international reserves, which are scaled by GDP and expressed in decimal form. X is a vector of control variables and CapOpen is one of the four measures of capital account openness described in Section 3.3. c represents the 13 A Fisher-type panel unit root test, developed by Maddala and Wu (1999), was applied. The individual time-series are examined by the Augmented Dickey Fuller (ADF) and the Phillips-Perron test. Independently of the number of lags included and of the specification (trend), the hypothesis of nonstationary series can be rejected. 14 The Hausman test recommends the use of a fixed effects specification instead of the random effects model.

418

Table 2 Reserves and capital mobility: static models. (1)

Number of countries Number of observations R2 (overall)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(11)

(12)

(13)

(14)

(15)

(16)

(17)

(18)

0.0034⁄⁄⁄ (4.97) 0.1279⁄⁄⁄ (6.22) 0.0998⁄⁄⁄ (2.88) 0.0081 (0.98)

0.0036⁄⁄ (2.10) 0.0642⁄ (1.80) 0.0845⁄⁄ (2.20) 0.0065 (1.34)

0.0019⁄⁄ (2.10) 0.0710⁄⁄⁄ (2.77) 0.0407 (1.38) 0.1062⁄⁄⁄ (3.32)

0.0022⁄⁄ (2.57) 0.0754⁄⁄⁄ (3.39) 0.0520 (1.48) 0.0117 (1.29)

0.0018⁄⁄ (2.04) 0.0691⁄⁄⁄ (2.66) 0.0430 (1.42) 0.1013⁄⁄⁄ (3.11)

0.0020 (1.24) 0.0492 (1.28) 0.0457 (1.48) 0.0907⁄ (1.93)

0.0015⁄⁄ (2.05) 0.0634⁄⁄ (2.49) 0.0433 (1.58) 0.1203⁄⁄⁄ (4.84)

0.0019 (1.21) 0.0486 (1.30) 0.0451 (1.52) 0.1051⁄⁄ (2.31)

0.0037⁄⁄⁄ (5.09) 0.1217⁄⁄⁄ (6.36) 0.0908⁄⁄⁄ (2.71) 0.0057 (0.48)

0.0033⁄⁄⁄ (4.60) 0.1231⁄⁄⁄ (6.44) 0.0949⁄⁄⁄ (2.82) 0.0038 (0.32)

0.0040⁄ (1.84) 0.0474 (1.13) 0.0782⁄⁄ (2.08) 0.0255⁄⁄ (2.09)

0.0023⁄⁄ (2.41) 0.0609 ⁄⁄ (2.24) 0.0354 (1.26) 0.0995⁄⁄⁄ (2.75)

0.0022⁄⁄⁄ (2.63) 0.0747⁄⁄⁄ (3.39) 0.0489 (1.44) 0.0075 (0.64)

0.0021⁄⁄ (2.27) 0.0620⁄⁄ (2.24) 0.0386 (1.35) 0.0964⁄⁄ (2.58)

0.0027 (1.37) 0.0320 (0.72) 0.0431 (1.39) 0.0760 (1.56)

0.0018⁄⁄ (2.37) 0.0539⁄⁄ (2.06) 0.0368 (1.45) 0.1137⁄⁄⁄ (4.22)

0.0026 (1.36) 0.0298 (0.68) 0.0421 (1.41) 0.0923⁄⁄ (1.99)

0.0402 (0.95)

0.0402 (0.96)

0.1165⁄⁄ 0.0464 (2.36) (1.15)

0.0225 (0.39)

0.0412 (0.98)

0.0842⁄⁄ 0.0693⁄ (2.47) (1.84)

0.0142 (1–17) 0.0088 (1.15)

0.0114 (0.97) 0.0088 (1.19)

0.0164⁄ (1.71) 0.0024 (0.23)

0.0083 (0.72) 0.0104 (1.45)

0.0042 (0.48) 0.0120 (1.53)

0.0064 (0.66) 0.0067 (0.71)

0.0083⁄⁄ (2.25)

0.0178⁄⁄ (2.34)

0.0052 (1.59) 0.0004⁄⁄ (2.13)

0.0004 (1.46) 0.0933⁄⁄⁄ (3.19)

0.0901⁄⁄⁄ (3.04)

0.0883⁄ (2.01)

0.0065⁄ (1.79)

0.0821⁄⁄ (2.05)

0.0042 (0.49) 0.01225 (1.60)

0.0003 (1.27) 0.0951⁄⁄⁄ (3.07)

⁄⁄⁄

0.0925⁄⁄⁄ (2.93)

0.0082 (0.87) 0.0052 (0.55)

0.0180⁄⁄ (2.47)

0.0040 (1.20) 0.0004⁄ (1.78)

0.0001 (0.21) 0.0974⁄⁄⁄ (4.27)

0.0032 (0.33) 0.0123 (1.52)

0.0920⁄⁄⁄ (2.64)

0.0886⁄ (1.89)

0.0002 (0.68) 0.1014⁄⁄⁄ (4.37)

0.0813⁄ (1.96)

0.0124⁄⁄ (2.03)

0.0003⁄⁄⁄ (2.98)

⁄⁄⁄

0.0028 (5.30)

0.0027 (5.20)

0.0110 (1.77)



0.0003 (2.47)

⁄⁄

107

106

96

107

89

107

96

106

96

106

105

95

105

89

105

95

105

95

2812

2743

1774

2652

2447

2812

1764

2597

1764

2746

2687

1715

2586

2416

2541

1705

2541

1705

0.11

0.12

0.05

0.21

0.19

0.11

0.15

0.25

0.16

0.11

0.12

0.04

0.21

0.19

0.21

0.13

0.25

0.14

Dependent variable: Reserves/GDP. Estimation method: Fixed effects estimator. Notes: t-statistics (in brackets) computed with heteroskedasticity-consistent standard errors. ⁄,

⁄⁄

, and

⁄⁄⁄

denote significance at the 10%, 5%, and 1% levels, respectively.

A. Steiner / Journal of Macroeconomics 38 (2013) 409–427

0.0038⁄⁄⁄ (5.63) 0.1309⁄⁄⁄ (6.46) Volatility 0.0975⁄⁄⁄ (2.85) Total external 0.0102 debt (per cent (1.29) of GDP) Short-term external debt (per cent of GDP) Fixed exchange rate,dummy Intermediate exchange rate, dummy De jure capital mobility (Chinn-Ito) De jure capital mobility (Edwards) Financial integration Economic globalisation De jure and de facto openness interacted Real GDP per capita Trade openness

A. Steiner / Journal of Macroeconomics 38 (2013) 409–427

419

country specific effect and e is the error term. i denotes a specific country and t represents the time period. The slope parameters, represented by the vectors a and b, are assumed to be constant across countries and time. t-statistics are computed with heteroskedasticity-consistent standard errors. Table 2 presents the results. Column 1 contains the benchmark regression, which includes the set of control variables that accounts for precautionary motives of reserve holdings. The results confirm previous findings: Economic development and trade openness are positively associated with reserves. More developed and more open countries hold higher levels of reserves. In contrast to our expectations, a higher volatility of exports reduces reserves. This finding may be explained by the fact that volatility is measured over the preceding 5 years. Low reserves indicate that reserves have been used to cushion the economy from volatile exports. Total external debt does not significantly affect reserves. Columns 2 to 5 add different measures for capital mobility to test the hypothesis that countries increase their reserve holdings in the presence of increasing capital mobility. Columns 2 and 3 examine the effect of de jure capital mobility (Hypothesis 1a) and columns 4 and 5 are devoted to de facto mobility (Hypothesis 1b). Column 2 analyses the effect of de jure capital mobility proxied by the Chinn-Ito index. The coefficient is significant and positive implying that countries with fewer restrictions on capital flows hold a larger amount of reserves. This effect is not only statistically significant but also economically important. To evaluate the size of the effect we rank countries according to their de jure capital mobility in the year 2000 and calculate the difference in the level of reserves (relative to GDP) between a country at the 75th percentile of capital mobility and a country at the 25th percentile. The level of reserves as a ratio to GDP in the relatively open country is 2.4 percentage points higher than in the rather closed economy. Given that the average ratio of reserves to GDP was 14.3% in 2000, this effect is substantial. Sign and significance of the control variables are unchanged with respect to column 1, which does not control for capital mobility. Column 3 examines the robustness of these results. It uses an alternative measure for de jure capital mobility, namely the index of Edwards (2007). The effect of capital mobility is again positive and significant. It shows that the results of column 2 are insensitive with respect to the definition of de jure capital mobility. The smaller magnitude of the effect comes from the fact that both indices use different scales with the Edwards index usually being larger than the Chinn-Ito index. We repeat our analysis of above and calculate the predicted effect for countries characterised by different levels of capital mobility in the year 2000. The level of reserves (relative to GDP) in a country at the 75th percentile of capital mobility according to the Edwards index is 1.5 percentage points larger than in a country at the 25th percentile. We now turn to the question whether de facto capital mobility has an effect on reserve holdings (Hypothesis 1b). To this end, column 4 adds a measure of de facto capital mobility, namely the Lane and Milesi-Ferretti measure for financial integration. Its effect is positive and significant. Countries with a large stock of cross-border assets relative to their economic size hold more international reserves. Reserves as a share of GDP are 1.1 percentage points higher in financially integrated economies (75th percentile of financial integration in the year 2000) compared to closed economies (25th percentile). The inclusion of this measure for de facto capital mobility affects the results with respect to the standard control variables. While the volatility measure becomes insignificant, total external debt is significant, but with an unexpected negative sign. This might be due to the fact that total external debt and the measure for de facto capital mobility are correlated because the construction of the latter contains total external debt. Hence, the assumption of exogenous regressors is violated and the results might be biased. Therefore, we re-estimate the effect of de facto capital mobility after dropping total external debt. Financial integration is still positive and significant (result not reported in table). A country’s reserves increase with the degree of its de facto openness to the world capital market. Economic globalisation, which is added in column 5, confirms the previous results that capital mobility – both de jure and de facto – is associated with larger reserve holdings. As long as capital controls are effective, de jure capital mobility affects de facto mobility and both measures are correlated. Since our regressions include either a measure of de jure or de facto mobility, we cannot determine whether both independently affect central banks’ reserve policies. To gain further insight in the nature of the fear of capital mobility, columns 6 and 7 include a measure of de jure capital mobility along with our de facto measure of financial integration.15 Results suggest that reserve policies respond to de facto financial integration. The de jure measures turn out to be insignificant. In addition, we interact financial integration with a measure of de jure mobility to examine whether the response in reserves to a given level of financial integration depends on the level of de jure capital mobility. When the Chinn-Ito index is used (column 8), de jure capital mobility becomes significant. De jure and de facto capital mobility both increase reserves individually. The interaction term is negative implying that for a given level of financial integration the effect of de jure mobility on reserves is smaller when fewer restrictions are imposed. When de jure capital mobility is proxied by the Edwards index (column 9), the index itself is insignificant (as in column 7), but the interaction is positive and significant. The positive effect of financial integration on reserves is the larger, the less capital controls are in place. This result confirms our hypothesis of a fear of capital mobility. Column 10 adds short-term external debt (relative to GDP) and dummy variables for de jure fixed and intermediate exchange rate regimes as additional regressors. These three determinants turn out to be insignificant. Columns 11 to 14 add the four measures of capital mobility one after another to this extended set of control variables. The finding that reserve holdings increase in the mobility of international capital can be confirmed. Columns 15 to 18 include de jure and de facto measures jointly and allow for interactions. Previous results are confirmed. 15 We do not include the index for economic globalization in these regressions because it combines information on de jure restrictions with de facto capital flows.

420

A. Steiner / Journal of Macroeconomics 38 (2013) 409–427

4.1.2. Dynamic regressions Since the time-series of reserves are characterised by a high degree of persistence, the determination of the level of reserves is a natural candidate for a dynamic specification that includes the lagged level of reserves as one of its determinants. This specification can be motivated by a partial adjustment or habit-persistence model. Table 3 therefore replicates the regressions of Table 2 using a dynamic specification. As a consequence, the fixed effects estimator is asymptotically biased. Therefore, the difference GMM estimator, also known as the Arellano-Bond estimator, is used.16 The tests of the validity of instruments, which are reported at the bottom of the table, support these specifications. Previous findings are confirmed: With the exception of the Chinn-Ito index, all measures of de facto and de jure capital mobility positively affect the level of reserves. They remain significant when de jure and de facto measures are included jointly. The effects of the control variables have the expected sign. 4.1.3. Robustness analysis Reverse causality: Since we regress the level of reserves on contemporaneous measures of capital mobility, the results rather show the correlation than the causality between capital mobility and reserves. Since both the level of reserves and the extent of de jure capital account restrictions are the outcome of policy decisions, their timing rather reveals preferences than causality. Countries may accumulate reserves after they have abolished capital controls. Alternatively, the liberalisation of the capital account itself may be a function of the level of reserves: Countries only dismantle capital controls when they consider their reserve buffer large enough. The pre-liberalisation period is characterised by increasing levels of reserves. For our hypothesis, however, the timing – and hence causality – is not crucial. The finding of a correlation is sufficient to conclude that central banks fear capital mobility. In contrast to de jure mobility, de facto financial integration is determined by the decisions of individual investors. Central banks might counteract the increased potential of capital flight in financially integrated economies by an increase in reserves. However, there exists also an argument for reversed causality: If investors prefer countries with large reserve holdings, reserves affect de facto financial integration and not vice versa. To address potential endogeneity of our measures of capital mobility, we lag them by one time period. Results are presented in columns 1 to 6 of Table 4. Previous results carry through: When included individually, all lagged measures of capital mobility raise the level of reserves significantly. The regressions including de jure and de facto measures concurrently suggest that de facto integration is the driving force. cDomestic component of capital flight: The previously used measures of de facto capital mobility implicitly assume that the fear of capital mobility stems from the foreign component of capital flight. A central bank has less instruments to steer foreign liabilities and capital flows when the capital account is liberalised. Columns 7 to 9 of Table 4 turn to the domestic component of capital flight: Domestic capital flees when domestic agents attempt to convert their bank deposits in foreign exchange.17 Since this domestic component of capital flight is proportional to the liabilities of the domestic banking system, broad money supply (M2) is used as an indicator of potential capital flight. Column 7 adds M2 scaled by GDP as an additional regressor to the benchmark regression (Table 2, column 1). Whereas the findings with respect to the other control variables are qualitatively unchanged, broad money increases the level of reserves significantly. Central banks’ reserves are higher, the larger the potential for domestic capital flight is. As columns 8 and 9 show, this effect comes in addition to the effect of de facto capital mobility, which preserves its positive impact. Moreover, de jure capital mobility increases reserves either individually or via the interaction term. Mercantilist reserve accumulation: To control for the fact that reserve accumulation might be intended to resist an exchange rate appreciation in the face of net capital inflows, we include the rate of change (in decimal) of the nominal exchange rate measured over the preceding 5 years as an additional regressor labelled Depreciation. An increase in this variable is defined as a depreciation. Results are presented in columns 10 to 12 of Table 4. The evidence suggests that a nominal depreciation raises the level of reserves. Central banks increase reserves after currency crises (large depreciations) or the depreciation itself is the result of reserve increases. With respect to our measures of capital mobility results are not qualitatively affected by the inclusion of this additional control variable. 4.2. Changes in reserves as management of capital inflows So far it has been shown that there exists a fear of capital mobility in the sense that central banks increase their holdings of foreign exchange when capital controls are dismantled and when the integration in the international financial market deepens (Hypothesis 1). We now analyse the related but different question whether central banks’ accumulation of foreign exchange is a direct response to capital flows. The accumulation of reserves is a form of capital inflow management and allows a central bank to influence the amount of foreign capital channelled to domestic uses in the absence of capital controls. Hence, the accumulation of foreign exchange might be a substitute for capital controls. We therefore proceed by testing whether capital flows cause changes in reserves. According to the hypothesis, higher net capital inflows imply that central banks absorb part of these inflows via the accumulation of reserves. The dependent variable is nominal changes in reserves net of gold (DIR), denominated in millions of US$. Net capital flows (CapFlows) are mea16 When the system GMM estimator (see Arellano and Bover (1995) and Blundell and Bond (1998)) is used, results are qualitatively unaffected. Estimation results may be obtained upon request. 17 Rothenberg and Warnock (2011) show that a substantial fraction of sudden stop episodes is caused by domestic capital flight.

Table 3 Reserves and capital mobility: dynamic models. (1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(11)

(12)

(13)

(14)

(15)

0.8065⁄⁄⁄ (19.55) 0.0006 (0.08) 0.0463⁄⁄⁄ (4.11) 0.0419⁄⁄ (2.50) 0.0020 (0.70)

0.8034⁄⁄⁄ (17.64) 0.0008 (1.31) 0.0320⁄⁄ (2.02) 0.0545⁄⁄ (2.11) 0.0086 (1.63)

0.7252⁄⁄⁄ (18.08) 0.103⁄⁄ (2.28) 0.0317⁄⁄ (2.31) 0.0266 (1.53) 0.0492⁄⁄⁄ (3.54)

0.8291⁄⁄⁄ (21.72) 0.16⁄⁄⁄ (4.32) 0.0302⁄⁄⁄ (2.68) 0.0200 (1.40) 0.0043 (1.42)

0.7328⁄⁄⁄ (16.84) 0.0010⁄⁄ (2.14) 0.0329⁄⁄ (2.40) 0.0329 (1.55) 0.0460⁄⁄⁄ (3.40)

0.7564⁄⁄⁄ (16.67) 0.0014⁄ (1.85) 0.0302⁄⁄ (2.00) 0.0320 (1.34) 0.0351⁄ (1.79)

0.7230⁄⁄⁄ (17.56) 0.0012⁄⁄⁄ (3.14) 0.0308⁄⁄ (2.33) 0.0262 (1.64) 0.0561⁄⁄⁄ (5.47)

0.7602⁄⁄⁄ (16.90) 0.0014⁄ (1.84) 0.0303⁄⁄ (1.99) 0.0326 (1.36) 0.0335⁄ (1.83)

0.8298⁄⁄⁄ (22.54) 0.0364 (0.77) 0.0367⁄⁄⁄ (3.32) 0.0369⁄⁄ (2.30) 0.0044 (0.63)

0.8222⁄⁄⁄ (21.54) 0.0432 (0.85) 0.0387⁄⁄⁄ (3.65) 0.0363⁄⁄ (2.25) 0.0059 (0.73)

0.8251⁄⁄⁄ (24.02) 0.0901 (1.11) 0.0211 (1.39) 0.0485⁄⁄ (2.00) 0.0204⁄⁄ (2.46)

0.7450⁄⁄⁄ (17.95) 0.11⁄⁄ (2.43) 0.0291⁄ (1.92) 0.0239 (1.30) 0.0453⁄⁄⁄ (3.00)

0.8274⁄⁄⁄ (21.07) 0.14⁄⁄⁄ (4.14) 0.0348⁄⁄⁄ (2.96) 0.0191 (1.29) 0.0112⁄ (1.86)

0.7489⁄⁄⁄ 0.7776⁄⁄⁄ (17.59) (20.93) 0.0010⁄⁄ 0.0011 (2.27) (1.14) ⁄⁄ 0.0304 (2.02)0.0235 (1.35) 0.0265 0.0339 (1.39) (1.39) 0.0439⁄⁄⁄ 0.0180 (2.82) (0.98)

0.0114 (0.37)

0.0173 (0.59)

0.0746⁄⁄ (2.38)

0.0026 (0.01)

0.0411 (1.59)

0.0086 (0.37)

0.0119⁄⁄⁄ (2.67) 0.0011 (0.57)

0.0092⁄⁄ (2.08) 0.0008 (0.70)

0.0141⁄⁄⁄ (3.24) 0.0037 (0.63)

0.0097⁄⁄ (2.12) 0.0019 (0.71)

0.0003 (0.36) 0.0044 (0.76)

0.0092⁄ (1.92) 0.0024 (0.90)

0.0019 (0.89)

0.0014 (0.38) 0.0004⁄⁄⁄ (3.91)

0.0004⁄⁄⁄ (3.07) 0.0479⁄⁄⁄ (3.77)

Economic globalisation De jure and de facto openness interacted Number of countries Number of observations Sargan test (p-level) Arellano-Bond-test (p-level)

0.0065 (1.39)

0.0440⁄⁄⁄ (3.62) 0.0009 (3.44)

0.0393⁄⁄ (2.31)

0.0009 (0.28) 0.0004⁄⁄ (2.53)

0.0489⁄⁄⁄ (4.70)

(16)

(17)

(18)

0.7415⁄⁄⁄ (18.50) 0.0012⁄⁄⁄ (2.92) 0.0028⁄ (1.94) 0.0251 (1.47) 0.0489⁄⁄⁄ (3.92)

0.7780⁄⁄⁄ (21.33) 0.0010 (1.14) 0.0213 (1.35) 0.0345 (1.38) 0.0184 (1.07)

0.0586⁄ (1.93)

0.0177 (0.82)

0.0612⁄ (1.76)

0.0116⁄⁄ (2.52) 0.0004 (0.29)

0.0084⁄ (1.86) 0.0012 (0.62)

0.0125⁄⁄ (2.42) 0.0014 (0.30)

0.0048 (0.83)

0.0004 (0.30) 0.0003⁄⁄ (2.54)

0.0401⁄⁄ (2.15)

0.0002⁄ (1.77) 0.0436⁄⁄⁄ (3.50)

⁄⁄⁄

0.0436⁄⁄⁄ (3.38)

0.0334⁄⁄ (2.11)

0.0002 (1.02) 0.0465⁄⁄⁄ (4.29)

0.0336⁄ (1.90)

0.0041 (1.32)

0.0000 (0.07)

⁄⁄

0.0007 (2.16) 0.0042 (1.55)

0.0000 (0.39)

107 2654

106 2589

92 1637

106 2496

89 2310

105 2444

92 1628

105 2444

92 1628

106 2591

105 2535

87 1581

104 2434

88 2281

104 2390

87 1572

104 2390

87 1572

1.0 0.84

1.0 0.97

1.0 0.59

1.0 0.59

1.0 0.75

1.0 0.67

1.0 0.45

1.0 0.67

1.0 0.45

1.0 0.99

1.0 0.94

1.0 0.63

1.0 0.67

1.0 0.72

1.0 0.73

1.0 0.51

1.0 0.74

1.0 0.51

A. Steiner / Journal of Macroeconomics 38 (2013) 409–427

Lagged endogenous 0.8151⁄⁄⁄ (21.38) variable Real GDP per capita 0.0003 (0.01) Trade openness 0.0449⁄⁄⁄ (3.87) Volatility 0.0429⁄⁄⁄ (2.59) Total external 0.0020 debt (per cent of (1.08) GDP) Short-term external debt, (per cent of GDP) Fixed exchange rate, dummy Intermediate exchange rate, dummy De jure capital mobility (ChinnIto) De jure capital mobility (Edwards) Financial integration

Dependent variable: Reserves/GDP. Estimation method: Difference GMM estimator (Arellano-Bond). Notes: t-statistics (in brackets) computed with heteroskedasticity-consistent standard errors. The interaction term consist of those variables of capital mobility that are included individually in the respective specification. ⁄, ⁄⁄, and ⁄⁄⁄ denote significance at the 10%, 5%, and 1% levels, respectively.

421

422

Table 4 Reserves and capital mobility: robustness analysis.

Real GDP per capita

Volatility Total external debt (per cent of GDP) M2 (per cent of GDP)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(11)

(12)

0.0034⁄⁄⁄ (4.87) 0.1282⁄⁄⁄ (6.13) 0.1011⁄⁄⁄ (2.92) 0.0079 (0.95)

0.0038⁄⁄ (2.19) 0.0661⁄ (1.86) 0.0873⁄⁄ (2.26) 0.0067 (1.39)

0.0025⁄⁄⁄ (3.50) 0.1009⁄⁄⁄ (4.68) 0.0667⁄⁄ (2.44) 0.0552⁄⁄⁄ (2.79)

0.0023⁄⁄⁄ (2.72) 0.0834⁄⁄⁄ (4.18) 0.0650⁄ (1.92) 0.0119 (1.23)

0.0021⁄⁄⁄ (3.14) 0.0989⁄⁄⁄ (4.39) 0.0700⁄⁄⁄ (2.65) 0.0612⁄⁄⁄ (2.77)

0.0031⁄ (1.83) 0.0648⁄ (1.82) 0.0691⁄⁄ (2.16) 0.0327⁄ (1.72)

0.0028⁄⁄⁄ (3.06) 0.1010⁄⁄⁄ (4.99) 0.0726⁄⁄ (2.13) 0.0111 (1.35) 0.0015⁄⁄⁄ (4.13)

0.0000 (0.83) 0.0482⁄⁄ (2.05) 0.0298 (1.20) 0.1177⁄⁄⁄ (4.56) 0.0009⁄⁄ (2.53)

0.0018 (1.10) 0.0436 (1.23) 0.0429 (1.45) 0.1050⁄⁄ (2.27) 0.0002 (0.45)

0.0039⁄⁄⁄ (5.09) 0.1292⁄⁄⁄ (5.72) 0.1152⁄⁄⁄ (3.17) 0.0137 (1.43)

0.0013 (1.53) 0.0603⁄⁄ (2.19) 0.0558⁄⁄ (2.06) 0.1234⁄⁄⁄ (4.83)

0.0018 (0.98) 0.0454 (1.12) 0.0616⁄ (1.97) 0.1067⁄⁄ (2.15)

0.0000⁄ (1.90)

0.0000⁄ (1.85)

0.0000 (1.02)

Depreciation De jure capital mobility (Chinn-Ito) De jure capital mobility (Edwards) Financial integration

0.0077 (2.14)

⁄⁄

0.0140 (2.24) ⁄⁄

0.0170 (2.39)

⁄⁄

0.0003 (1.05)

0.0004 (2.08)

0.0504⁄⁄⁄ (2.97)

0.0200 (2.35)

⁄⁄

0.0509⁄⁄⁄ (3.19)

0.0399⁄ (1.82)

0.0943⁄⁄⁄ (4.06)

0.0001 (0.28) 0.0810⁄⁄ (1.99)

0.0990⁄⁄⁄ (4.21)

0.0002 (0.55) 0.0803⁄ (1.94)

0.0075 (1.34)

0.0000 (0.01)

0.0110⁄ (1.87)

0.0003⁄⁄ (2.55)

0.0116⁄ (1.74)

0.0003⁄⁄ (2.57)

106 2612 0.20

98 1796 0.10

106 2588 0.29

96 1749 0.16

106 2352 0.26

95 1553 0.16

0.0028⁄⁄⁄ (5.42)

Economic globalisation De jure and de facto openness interacted Number of countries Number of observations R2

⁄⁄

106 2685 0.12

98 1810 0.05

107 2678 0.17

89 2471 0.19

107 2792 0.17

107 2544 0.10

Dependent variable: Reserves/GDP. Estimation method: Fixed effects estimator. Notes: t-statistics (in brackets) computed with heteroskedasticity-consistent standard errors. In columns 1 to 6 all variables related to capital mobility (de jure, de facto and interaction terms) are lagged by one period. The interaction term consist of those variables of capital mobility that are included individually in the respective specification. ⁄, ⁄⁄, and ⁄⁄⁄ denote significance at the 10%, 5%, and 1% levels, respectively.

A. Steiner / Journal of Macroeconomics 38 (2013) 409–427

Trade openness

(1)

423

A. Steiner / Journal of Macroeconomics 38 (2013) 409–427

sured by the balance of the financial account of the balance of payments, i.e., excluding the central bank. It encompasses the categories direct investment, portfolio investment, financial derivatives and other investment. We estimate the following regression equation by OLS using heteroskedasticity-consistent standard errors:

DIRit ¼ a  X it þ b  CapFlowsit þ ci þ eit

ð2Þ

where as in Eq. (1) c is a fixed country effect and e the error term. i denotes a specific country and t represents the time period. The vector of control variables X includes changes in money supply and a dummy variable for currency crises. Given a central bank’s balance sheet, an increase in money supply induces either a decrease in domestic bonds in private hands or an increase in international reserves (unsterilized intervention). The dummy for currency crises controls for the effect that reserves generally fall during a crisis. It takes the value one in years where a speculative attack – unsuccessful or successfully leading to a crisis – is identified by an exchange market pressure index. Additional control variables, derived from the regressions in Table 2, are changes in real GDP per capita, of trade openness and of total external debt. However, they all turn out to be insignificant and are not included in the presented results. Net private capital flows are positively associated with changes in reserves in all specifications (see Table 5). Central banks accumulate reserves when capital flows in and when private agents increase their external indebtedness. This holds for all country samples, in particular the group of developing countries and emerging markets. Whereas changes in money supply are uncorrelated with reserve changes, reserves are significantly lower in years with a currency crisis. To further unveil central bank behaviour, we test whether they respond asymmetrically to capital flows. The action of a central bank in the face of capital inflows (= positive net capital flows) might differ from its response to capital outflows (= negative net capital flows). We hypothesise that central banks manage capital inflows via the accumulation of reserves since capital inflows cause an appreciation of the exchange rate and imply increasing external indebtedness of the country. If there are capital outflows, the central bank does not intervene in the foreign exchange market as long as these outflows do not lead to a systemic crisis. Table 5 Management of capital inflows. Full sample

Net capital inflows

Developing countries

(1)

(2)

(3)

(4)

(5)

(6)

(7)

0.7657⁄⁄ (1.99)

1.2017⁄⁄⁄ (2.86) 0.0010 (0.31)

1.3344⁄⁄⁄ (3.11) 0.0301 (0.41) 2180⁄⁄ (2.51)

1.2666⁄⁄⁄ (2.69)

1.4777⁄⁄⁄ (3.56) 0.0125 (0.42) 1190⁄⁄ (2.24)

1.4267⁄⁄⁄ (2.91)

1.5497⁄⁄⁄ (3.47) 0.0186 (0.62) 3980⁄ (2.06)

180 5031 0.23

107 1608 0.42

154 4107 0.43

77 690 0.63

19 636 0.50

11 142 0.67

DM1 Currency crisis, dummy Number of countries Number of observations R2 (overall)

Emerging markets

Dependent variable: Changes in reserves. Estimation method: Fixed effects. Notes: t-statistics (in brackets) computed with heteroskedasticity-consistent standard errors. ⁄, ⁄⁄, and ⁄⁄⁄ denote significance at the 10%, 5%, and 1% levels, respectively. Table 6 Management of capital inflows. Full sample

Capital inflows Capital outflows

Developing countries

Emerging markets

(1)

(2)

(3)

(4)

(5)

(6)

(7)

1.0169⁄⁄ (2.00) 0.1003 (0.65)

1.4313⁄⁄⁄ (3.59) 0.1885 (1.09) 0.0159 (0.52)

1.5117⁄⁄⁄ (3.85) 0.3436 (1.65) 0.0343 (0.49) 1900⁄⁄ (2.51)

1.6500⁄⁄⁄ (3.50) 0.0822 (0.32)

1.6124⁄⁄⁄ (4.32) 0.4777 (1.34) 0.0040 (0.14) 1270⁄⁄⁄ (3.05)

1.6343⁄⁄⁄ (3.23) 0.3823 (1.55)

1.6564⁄⁄⁄ (4.37) 1.7632⁄⁄ (2.72) 0.0020 (0.06) 3530⁄⁄ (2.34)

180 5031 0.31

107 1608 0.30

89 881 0.57

154 4107 0.55

77 690 0.68

19 636 0.56

11 142 0.67

DM1 Currency crisis, dummy Number of countries Number of observations R2 (overall)

Dependent variable: Changes in reserves. Estimation method: Fixed effects. Notes: t-statistics (in brackets) computed with heteroskedasticity-consistent standard errors. ⁄, ⁄⁄, and ⁄⁄⁄ denote significance at the 10%, 5%, and 1% levels, respectively.

424

A. Steiner / Journal of Macroeconomics 38 (2013) 409–427

Table 7 Management of capital inflows. Full sample

DNet capital inflows

Developing countries

Emerging markets

(1)

(2)

(3)

(4)

(5)

(6)

0.5792⁄⁄⁄ (3.24)

0.7983⁄⁄⁄ (4.59) 0.1019 (1.01) 1880⁄⁄⁄ (2.68)

0.7289⁄⁄⁄ (3.09)

0.8726⁄⁄⁄ (5.40) 0.0687 (1.02) 734 (1.57)

0.7564⁄⁄ (2.67)

0.9595⁄⁄⁄ (6.25) 0.0694 (1.00) 28.6 (0.02)

180 4884 0.09

88 874 0.25

154 3978 0.12

76 684 0.17

19 621 0.12

11 141 0.17

DM1 Currency crisis, dummy Number of countries Number of observations R2 (overall)

Dependent variable: Changes in reserves. Estimation method: Fixed effects. Notes: t-statistics (in brackets) computed with heteroskedasticity-consistent standard errors. ⁄, ⁄⁄, and ⁄⁄⁄ denote significance at the 10%, 5%, and 1% levels, respectively.

Table 6 shows the results of estimating Eq. (2) where CapFlows is split into capital inflows and capital outflows. The variable capital inflows equals the amount of net capital flows if they are larger than zero and is set to zero otherwise. The variable capital outflows, respectively, equals capital outflows and zero otherwise. The results fully support the hypothesis. In all specifications and across different country samples capital inflows lead to a significant increase in reserves. Central banks offset net capital inflows via the accumulation of foreign exchange and thereby export capital. The coefficient, which is larger than one, implies that the accumulation of reserves by central banks reverses the direction of private capital inflows: When private capital flows in, central banks react by exporting capital such that a net capital outflow results. This mirrors the Lucas paradox according to which capital flows from relatively poor to rather rich countries. Since reserve accumulation has taken place predominantly in emerging and developing countries, our findings show that the paradox disappears when solely private capital flows are considered. Central bank intervention is responsible for net capital flowing to rich countries.18 In the face of capital outflows, central banks do not adjust their reserves. Reserves are neither affected by changes in money supply. Currency crises are associated with losses of reserves, which are especially pronounced in emerging markets. Finally, Table 7 investigates whether central banks dislike large swings in the financial account. It analyses whether changes in capital flows – i.e. from a moderate level of capital inflows to a much larger amount – are offset by changes in reserves. Changes in net capital inflows are defined as changes of the financial account on a year-to-year basis. Hence, we again estimate Eq. (2) where CapFlows is replaced by the variable DNet capital inflows, which is defined as (CapFlowst  CapFlowst1). Independently of the specification, this variable has a positive and significant effect on reserve changes. A positive change in capital flows with respect to the previous year is counteracted by an increase in reserves. The effects of the control variables are in line with previous findings. 5. Conclusion The empirical analysis has shown that the accumulation of foreign exchange may be regarded as an indication of a ‘‘fear of capital mobility’’ suffered by central banks. First, central banks fear that capital inflows are volatile and subject to sudden reversals. Therefore, they demand reserves as a buffer stock against potential capital flight. Second, central banks accumulate reserves in order to manage net capital flows in the absence of capital controls. They fear the real effects that capital flows might have on the real exchange rate and, consequently, on the domestic economy. The finding that central banks manage capital flows differs in an important way from the standard analysis concerning the accumulation of reserves. If the accumulation of foreign exchange is explained as a buffer stock, which will be used to defend the exchange rate in a period of crisis, the level of reserves matters. The accumulation of reserves itself has no function and, more precisely, the timing of the accumulation is irrelevant. However, if the objective of the foreign exchange accumulation consists in managing capital flows, the accumulation itself – and its effects – is the target of central bank policy. The level of reserves does not matter. Only changes in reserves have macroeconomic effects. These results also affect the literature on the costs and benefits of capital account liberalisation. These studies have to take the costs of increased foreign exchange holdings into account when capital account liberalisations are evaluated. There exists widespread evidence that international reserve holdings reduce domestic volatility: output volatility (Aizenman et al., 2011), terms of trade volatility (Aizenman et al., 2012) and the volatility of the real exchange rate (Aizenman and Riera-Crichton, 2008) are lower in economies characterised by large reserve holdings. Further research may examine whether these reserve benefits in terms of lower volatilities are positively correlated with the degree of capital mobility. If this were the case, our finding that reserves increase with capital mobility can be rationalised by a cost–benefit analysis: 18

Alfaro et al. (2011) come to a similar conclusion.

A. Steiner / Journal of Macroeconomics 38 (2013) 409–427

425

The optimal amount of reserves increases in capital mobility. Moreover, one could analyse whether the positive effects of a reserve lifejacket are temporary, namely restricted to the process of liberalisation, or long-lasting, thus arising also in countries that have been characterised by financial openness throughout history. In sum, the accumulation of foreign exchange has to be analysed in a broad context and is the result of a variety of factors. Central banks might deliberately distort the balance of payments. Foreign reserves are not only used to defend the exchange rate in periods of crisis but also to manage capital flows in periods without major economic disturbances. The liberalisation of capital markets is to a certain extent compensated by the accumulation of official reserves. A microeconomic policy distortion – capital controls – is replaced by a macroeconomic one – the accumulation of foreign exchange. To put it differently: Methods have changed, but the objective of regulating net capital flows persists. Acknowledgements The author would like to thank Roland Vaubel, Frank Westermann, anonymous referees and participants of seminars at the University of Mannheim, University of Osnabrück, the Dutch National Bank and the EEA-ESEM conference for helpful comments and suggestions. Part of this research was undertaken while the author was affiliated with the University of Mannheim. Appendix A. Country list

Afghanistan Albania Algeria Angola Antigua and Barbuda Argentina Armenia Aruba Australia Austria Azerbaijan Bahamas, The Bahrain Bangladesh Barbados Belarus Belgium Belize Benin Bhutan Bolivia Bosnia and Herzegovina Botswana Brazil Brunei Darussalam Bulgaria Burkina Faso Burundi Cambodia Cameroon Canada Cape Verde Central African Republic Chad Chile China Colombia Comoros Congo, Rep.

Djibouti Dominica Dominican Republic Ecuador Egypt, Arab Rep. El Salvador Equatorial Guinea Eritrea Estonia Ethiopia Fiji Finland France Gabon Gambia, The Georgia Germany Ghana Greece Grenada Guatemala Guinea Guinea-Bissau Guyana Haiti Honduras Hong Kong, China Hungary Iceland India Indonesia Iran, Islamic Rep. Iraq Ireland Israel Italy Jamaica Japan Jordan

Lebanon Lesotho Liberia Libya Lithuania Luxembourg Macao, China Macedonia, FYR Madagascar Malawi Malaysia Maldives Mali Malta Mauritania Mauritius Mexico Moldova Mongolia Montenegro Morocco Mozambique Myanmar Namibia Nepal Netherlands New Zealand Nicaragua Niger Nigeria Norway Oman Pakistan Panama Papua New Guinea Paraguay Peru Philippines Poland

Senegal Serbia Seychelles Sierra Leone Singapore Slovak Republic Slovenia Solomon Islands Somalia South Africa Spain Sri Lanka St. Kitts and Nevis St. Lucia St. Vincent and the Grenadines Sudan Suriname Swaziland Sweden Switzerland Syrian Arab Republic Tajikistan Tanzania Thailand Timor Leste Togo Tonga Trinidad and Tobago Tunisia Turkey Turkmenistan Uganda Ukraine United Kingdom United States Uruguay Vanuatu Venezuela, RB Vietnam (continued on next page)

426

Costa Rica Cote d’Ivoire Croatia Cyprus Czech Republic Denmark

A. Steiner / Journal of Macroeconomics 38 (2013) 409–427

Kazakhstan Kenya Korea, Rep. Kosovo Kuwait Kyrgyz Republic Lao PDR Latvia

Portugal Romania Russian Federation Rwanda Samoa Sao Tome and Principe Saudi Arabia

Yemen, Rep. Zambia Zimbabwe

Note: Countries marked with an asterisk belong to the group of emerging-market countries.

Appendix B. List of variables and data sources Variable

Source

Description

Reserves

World Bank (2011a)

Real GDP per capita

World Bank (2011a)

Trade openness Total external debt (divided by GDP)

World Bank (2011a) World Bank (2011a)

Short-term external debt (divided by GDP) M1

World Bank (2011b) IMF (2012)

M2

IMF (2012)

Interest rate (money market) De jure capital mobility (Chinn-Ito)

IMF (2012)

De jure capital mobility (Edwards)

Edwards (2007)

Financial openness

Lane and MilesiFerretti (2007) Dreher (2006)

Net international reserves comprise special drawing rights, reserves of IMF members held by the IMF, and holdings of foreign exchange under the control of monetary authorities. Gold holdings are excluded. Data are in current U.S. dollars GDP is measured as gross domestic product in constant international dollars with the year 2000 as base. An international dollar has the same purchasing power over GDP as the U.S. dollar has in the United States. This measure of GDP is divided by the population which counts all residents regardles of legal status or citizenship Openness is defined as the sum of exports and imports divided by GDP. Data are expressed in per cent Total external debt is the sum of public, publicly guaranteed, and private nonguaranteed long-term debt, use of IMF credit, and shortterm debt. Data are in current U.S. dollars divided by GDP Short-term external debt includes all debt that has an original maturity of 1 year or less. Data are in current U.S. dollars divided by GDP Money (line 34 IFS) is the sum of currency outside banks and demand deposits (excluding those of the central government). Data are in millions of current national currency M2 is the sum of M1 and quasi money. Quasi money (line 35 IFS) is defined as the sum of time, savings and foreign currency deposits of residents (excluding the central government). Data are in millions of current national currency Money market rate (line 60b IFS): interest rate on short-term lending between financial institutions, measured in per cent Measure of the de jure openness of the capital account. Calculation is based on the binary dummy variables of the IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions (AREAR) Index that combines the measures from Quinn (2003) and Mody and Murshid (2005) [both are based on the information provided by the AREAR] and information from country-specific sources Sum of total external assets and liabilities divided by GDP

Economic globalisation Depreciation

Chinn and Ito (2002, 2006)

Net capital inflows

World Bank (2011a) IMF (2012)

Currency crisis, dummy

Own calculations

Index based on actual flows of goods and capital and restrictions concerning these flows Rate of change of the nominal exchange rate (defined as domestic currency per dollar), calculated over the preceding 5 years Financial account (IFS line 78bjd) is defined as the difference of net capital inflows (investment from domestic residents abroad) and net capital outflows (investment from foreigners in the domestic economy). It comprises direct investment, portfolio investment, financial derivatives and other investment The identification of a currency crisis is based on an exchange market pressure index. The calculation follows the procedure as described in Eichengreen et al. (1996)

A. Steiner / Journal of Macroeconomics 38 (2013) 409–427

427

References Adler, G., Tovar, C.E., 2011. Foreign Exchange Intervention: A Shield Against Appreciation Winds? IMF Working Paper 11/165. Aizenman, J., Chinn, M.D., Ito, H., 2011. Surfing the waves of globalization: Asia and financial globalization in the context of the trilemma. Journal of the Japanese and International Economies 25 (3), 290–320. Aizenman, J., Chinn, M.D., Ito, H., 2010. The emerging global financial architecture: tracing and evaluating the new patterns of the trilemma’s configurations. Journal of International Money and Finance 29 (4), 615–641. Aizenman, J., Edwards, S., Riera-Crichton, D., 2012. Adjustment patterns to commodity terms of trade shocks: the role of exchange rate and international reserves policies. Journal of International Money and Finance 31 (8), 1990–2016. Aizenman, J., Lee, J., 2007. International reserves: precautionary versus mercantilist views, theory and evidence. Open Economies Review 18 (2), 191–214. Aizenman, J., Riera-Crichton, D., 2008. Real exchange rate and international reserves in an era of growing financial and trade integration. The Review of Economics and Statistics 90 (4), 812–815. Aizenman, J., Sun, Y., 2012. The financial crisis and sizable international reserves depletion: from ‘fear of floating’ to the ‘fear of losing international reserves’? International Review of Economics and Finance 24, 250–269. Alfaro, L., Kalemli-Ozcan, S., Volosovych, V., 2011. Sovereigns, Upstream Capital Flows and Global Imbalances. NBER Working Paper No. 17396. Arellano, M., Bover, O., 1995. Another look at the instrumental variable estimation of error-component models. Journal of Econometrics 68 (1), 29–51. Blanchard, O., Giavazzi, F., Sa, F., 2005. International investors, the U.S. current account, and the dollar. Brookings Papers on Economic Activity 1, 1–49. Blundell, R., Bond, S., 1998. Initial conditions and moment restrictions in dynamic panel data models. Journal of Econometrics 87 (1), 115–143. Branson, W.H., 1977. Asset markets and relative prices in exchange rate determination. Sozialwissenschaftliche Annalen, Band 1, 69–89. Calvo, G.A., Reinhart, C.A., 2002. Fear of floating. The Quarterly Journal of Economics 117 (2), 379–408. Cheung, Y.-W., Qian, X., 2009. Hoarding of international reserves: Mrs. Machlup’s wardrobe and the Joneses. Review of International Economics 17 (4), 824–843. Chinn, M.D., Ito, H., 2002. Capital Account Liberalization, Institutions and Financial Developments: Cross Country Evidence. NBER Working Paper 8967. Chinn, M.D., Ito, H., 2006. What matters for financial development? Capital controls, institutions, and interactions. Journal of Development Economics 81 (1), 163–192. Dreher, A., 2006. Does globalization affect growth? Evidence from a new index of globalization. Applied Economics 38 (10), 1091–1110. Edwards, S., 2007. Capital controls, sudden stops, and current account reversals’’. In: Edwards, S. (Ed.), Capital Controls and Capital Flows in Emerging Economies: Policies, Practice, and Consequences. University of Chicago Press, Chicago. Eichengreen, B., Rose, A.K., Wyplosz, C., 1996. Contagious currency crises. Scandinavian Journal of Economics 98 (4), 463–484. Frenkel, J.A., 1976. A monetary approach to the exchange rate: doctrinal aspects and empirical evidence. The Scandinavian Journal of Economics 78, 200–224. Frenkel, J.A., Jovanovic, B., 1981. Optimal international reserves: a stochastic framework. The Economic Journal 91, 507–514. Gassebner, M., Gaston, N., Lamla, M.J., 2011. The inverse domino effect: are economic reforms contagious? International Economic Review 52, 183–200. Ghosh, A.R., Ostry, J.D., Tsangarides, C.G., 2012. Shifting Motives: Explaining the Buildup in Official Reserves in Emerging Markets since the 1980s. IMF Working Paper 12/34. Heller, H.R., 1966. Optimal international reserves. Economic Journal 76, 296–311. Henderson, D.W., Rogoff, K., 1982. Negative net foreign asset positions and stability in a world portfolio balance model. Journal of International Economics 13, 85–104. Honig, A., 2008. Addressing causality in the effect of capital account liberalization on growth. Journal of Macroeconomics 30 (4), 1602–1616. International Monetary Fund, 2012. International Financial Statistics, Online Database (Various Months), Washington, DC. Jeanne, O., Rancière, J., 2011. The optimal level of international reserves for emerging market countries: a new formula and some applications. Economic Journal 121, 905–930. Joyce, J.P., Noy, I., 2008. The IMF and the liberalization of capital flows. Review of International Economics 16 (3), 413–430. Korinek, A., Serven, L., 2010. Undervaluation through Foreign Reserve Accumulation: Static Losses, Dynamic Gains. Policy Research Working Paper 5250. The World Bank. Kouri, P.J.K., 1983. The balance of payments and the foreign exchange market: a dynamic partial equilibrium model. In: Bhandari, J.S., Putnam, B.H. (Eds.), Economic Interdependence and Flexible Exchange Rates. MIT Press, Cambridge, MA. Lane, P., Burke, D., 2001. The empirics of foreign reserves. Open Economies Review 12 (4), 423–434. Lane, P., Milesi-Ferretti, G.M., 2007. The external wealth of nations mark II: revised and extended estimates of foreign assets and liabilities, 1970–2004. Journal of International Economics 73 (2), 223–250. Levy-Yeyati, E., Sturzenegger, F., 2007. Fear of Appreciation. The World Bank, Policy Research Working Paper WPS 4387. Maddala,G.S.,Wu,S.,1999. Acomparative study ofunitroottests withpaneldataandanewsimpletest. Oxford BulletinofEconomics andStatistics 61,631–652. Mody, A., Murshid, A.P., 2005. Growing up with capital flows. Journal of International Economics 65 (1), 249–266. Obstfeld, M., Shambaugh, J.C., Taylor, A.M., 2010. Financial stability, the trilemma, and international reserves. American Economic Journal: Macroeconomics 2 (2), 57–94. Obstfeld, M., Shambaugh, J.C., Taylor, A.M., 2005. The trilemma in history: tradeoffs among exchange rates, monetary policies, and capital mobility. Review of Economics and Statistics 87 (3), 423–438. Ostry, J.D., Ghosh, A.R., Chamon, M., 2012. Two Targets, Two Instruments: Monetary and Exchange Rate Policies in Emerging Market Economies. IMF Staff Discussion Note No. 12/01. Popper, H., Mandilaras, A., Bird, G., 2011. Trilemma Stability and International Macroeconomic Archetypes in Developing Economies. School of Economics Discussion Papers 0311, School of Economics, University of Surrey. Quinn, D.P., 2003. Capital account liberalization and economic globalization, 1890–1999: a synoptic view. International Journal of Finance and Economics 8 (3), 189–204. Rothenberg, A.D., Warnock, F.E., 2011. Sudden flight and true sudden stops. Review of International Economics 19 (3), 509–524. Simmons, B.A., Elkins, Z., 2004. The globalization of liberalization: policy diffusion in the international political economy. American Political Science Review 98 (1), 171–189. UNCTAD, 1999. Trade and Development Report, New York and Geneva. Vaubel, R., 1997. The bureaucratic and partisan behavior of independent central banks: German and international evidence. European Journal of Political Economy 13, 201–224. World Bank, 2011a. World Development Indicators, Washington, DC. World Bank, 2011b. Global Development Finance, Washington, DC.