Committee on the Global Financial System

CGFS Papers
No 33
Capital flows and emerging market economies

Report submitted by a Working Group established by the Committee on the Global Financial System

This Working Group was chaired by Rakesh Mohan of the Reserve Bank of India

January 2009

JEL Classification: F21, F32, F36, G21, G23, G28

Copies of publications are available from:
Bank for International Settlements
Press & Communications
CH 4002 Basel, Switzerland

E mail: publications@bis.org
Fax: +41 61 280 9100 and +41 61 280 8100
This publication is available on the BIS website (www.bis.org).

© Bank for International Settlements 2009. All rights reserved. Brief excerpts may be reproduced or translated provided the source is cited.

A. Introduction

This Report reviews the growing integration of the major emerging market economies (EMEs) into global financial markets. Greater financial integration is evident from the sustained rise in both gross capital inflows (ie non-resident purchases of domestic assets) and outflows (ie resident purchases of foreign assets) to and from the EMEs. Although the structure of flows has become more stable, capital flows continue to be very volatile and this has major macroeconomic implications for recipient countries. The size and the structure of inflows are heavily conditioned by, and exert a major influence on, the state of development of local financial markets.

The benefits and costs of capital market integration have been a controversial topic of debate among academics and among policymakers. In principle, access to foreign savings helps a country to lift future income streams (by undertaking investments whose prospective returns exceed the cost of finance) and to better smooth consumption over time. In practice, however, capital inflows – in terms of sheer size, volatility and form – have very often put emerging market countries in major difficulty. During 2006 and 2007, the very rapid rise of gross private capital inflows into the EMEs (now over $1 trillion a year, compared with the previous peak of $300 billion in 1996) caused considerable strains in some countries. In contrast, the current year (2008) to date has witnessed large equity outflows by portfolio investors. Such large swings, over a very short period of time, complicate the conduct of monetary policy and liquidity management in the EMEs. And many feel that the financial stability risks have increased or at least become harder to monitor.

The next section briefly reviews the extensive academic debate on this topic. This debate, somewhat inconclusive on the surface, yields several valuable insights on closer inspection. The final section of this chapter outlines the plan of the Report.

The macroeconomic effects of capital account liberalisation

Economists have long debated the relationship between capital account liberalisation and economic performance. One theoretical approach, forcefully advocated in the mid-1990s, follows the first-best prescription of the neoclassical model (assuming, in particular, perfect capital markets): allowing the free flow of capital across borders would lead to a more efficient allocation of resources and be welfare-enhancing for both borrowers and lenders, in a fashion similar to the liberalisation of trade.

An alternative view took a second-best perspective: that removing one distortion – restrictions on capital movements – in the presence of other distortions that often exist in emerging markets may not necessarily enhance welfare (eg Newbery and Stiglitz (1984); and Stiglitz (2008)). This alternative view gained particular relevance after the onset of the Asian crisis. This crisis focused attention on how incomplete or malfunctioning domestic financial markets in recipient countries and poor risk management in capital-exporting countries could undermine the case for capital account liberalisation.¹

In the light of the debate on the macroeconomic and growth effects of capital account liberalisation, a large empirical literature has emerged during the past decade in an attempt to settle the issue. The Report therefore begins with a brief survey of this empirical evidence.

¹ The first-best prescription is, of course, that policies can address those market failures directly – and indeed did so after the Asian crisis.

Cross-country studies

The cross-country empirical literature on capital account liberalisation has been extensively surveyed by several authors – Eichengreen (2001); Edison, Klein, Ricci and Sløk (2004); Prasad, Rogoff, Wei and Kose (2003); Henry (2007); and Reinhart and Reinhart (2008). This section draws out some key themes of selected studies.

A majority of cross-country studies on the growth effects of capital account liberalisation follow a similar methodological approach, although some of them examine developed and developing countries together, while others focus only on developing countries. Typically, these studies use a proxy to measure capital account openness (eg the number of years a country has had an open capital account), and regress a measure of economic performance (eg average economic growth) on this proxy. Very often, the proxy measure for capital account liberalisation (but certainly not the only one) is constructed as a binary indicator of an “open” or “closed” capital account using information from yearly issues of the IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER).

Despite the numerous cross-country attempts to analyse the effects of capital account liberalisation, there appears to be only limited evidence that supports the notion that liberalisation enhances growth. This failure to find robust evidence has been interpreted by the critics of capital account liberalisation to mean that liberalisation does not promote growth, as the proponents of the alternative view would argue. For example, one of the earliest (and most widely cited) papers about the macroeconomic effects of capital account liberalisation is Rodrik (1998). He uses data for about 100 countries (both developed and developing) between 1975 and 1995 and regresses growth of income per capita on the IMF indicator-based binary variable mentioned above. He finds no correlation between liberalisation and per capita growth.

However, some argue that this result may be due partly to the crudeness of the binary variable that proxies for liberalisation. Quinn (1997) constructs a proxy that measures not only the presence but also the intensity of a liberalised capital account and finds that, for a sample of 66 countries between 1960 and 1989, there is a positive correlation between the change in his indicator and growth; however, Quinn uses relatively fewer low-income countries than Rodrik, which may be one source of the different results. Edwards (2001) uses Quinn’s measure and finds that liberalisation enhances growth in high-income countries but decreases it in low-income ones. This result suggests that perhaps the growth effects of capital account liberalisation are contingent on a country’s level of development.

Contingent effects and sequencing

Given the general inability to find unconditional positive growth effects of liberalisation, many studies have attempted to determine whether such effects are dependent on other conditions and policies that accompany liberalisation (“contingent effects”). For example, Kraay (1998) tests whether the growth effects of liberalisation are contingent on the quality of policy and institutions but finds no evidence, regardless of the use of the IMF-based binary proxy of Quinn’s indicator. In contrast, Klein (2005) finds evidence of a non-monotonic interaction between institutional quality and the effect of capital account openness on growth. Using panel data for 71 countries, a measure of capital account openness similar to the approach in Rodrik (1998), and the average of five variables to proxy for institutional quality (bureaucratic quality, control of corruption in government, risks of expropriation, repudiation of government contracts, and rule of law), he finds that the effect of capital account openness on economic growth is greatest for countries with better, but not the best, institutional quality.

Arteta, Eichengreen and Wyplosz (2003) ask whether the positive effect (if any) of capital account liberalisation on growth is limited to countries in a more advanced stage of financial and institutional development (where distortions that may result in a perverse effect of liberalisation are presumably low). They also examine whether it is limited to countries that have been deemed to have followed a proper sequencing of reforms (that is, where macroeconomic imbalances have been first eliminated and a high degree of trade openness has been achieved). They find only weak evidence that the effects of capital account liberalisation vary with financial and institutional development. On the other hand, they do find evidence that the positive effects of capital account openness on growth are contingent on the absence of macroeconomic imbalances, but not on openness to trade. In the presence of macroeconomic imbalances, capital account liberalisation is as likely to hurt as to help. This suggests that the sequencing of reforms shapes the effects of capital account liberalisation, which underlines the need for caution in approaching liberalisation in practice.

However, such cross-country studies that document conditional or unconditional positive effects of capital account liberalisation on aggregate economic growth cannot be considered representative of the considerable literature on this issue. In a survey of 10 studies on the subject, Edison et al (2004) find that only three uncover an unambiguous positive effect of liberalisation on growth. Similarly, Prasad et al (2003) survey 14 studies and find that only three of those studies identify a statistically significant positive relationship between capital account liberalisation and economic growth. This is consistent with the observation by Eichengreen (2001) that the literature finds, at best, ambiguous evidence that liberalisation has any impact on growth.

Reconciling the evidence

If theory predicts a positive effect of capital account liberalisation on growth for an emerging market economy, why have empirical studies been unable to unequivocally establish this link? Henry (2007) offers two compelling explanations. First, the IMF’s AREAER-based measure of capital account liberalisation used in several studies is fraught with imperfections. Second, the common econometric specification and data used in previous studies test for permanent effects of capital account liberalisation on growth, while theory only suggests a temporary growth effect and a permanent level effect.

A line of research that possibly circumvents imperfections in the IMF’s AREAER-based measure of capital account liberalisation does so by focusing on what happens before and after episodes of capital account liberalisation. In this literature, a capital account liberalisation event is assumed to occur when a country changes regulation to allow foreigners to purchase shares on the domestic stock market or when there is a significant increase in the S&P/International Finance Corporation’s Investability Index for the country. Studies based on this approach have been more successful in documenting a positive, but temporary, effect of capital account on investment and growth. In particular, countries appear to derive substantial benefits from opening their equity markets to foreigners (see eg Henry (2007) for a survey).² Henry (2003) documents the channels – consistent with prediction by theory – through which the effect of capital account liberalisation operates. In the years following capital account liberalisation, the cost of capital declines. The lower cost of capital in turn boosts investment and, hence, economic growth.³

The relative success of event studies in documenting a positive effect could be attributed to the fact that these studies generally focus on a shorter time window around the date of the capital account liberalisation. As pointed out by Henry, theory suggests that the growth effect of capital account liberalisation should be temporary, and that level effects should be permanent. Estimating the effect with a sample that covers a long time period or a cross section of countries, as is the case in many of the studies, would fail to find a significant positive growth effect even when there is one because it is implicitly testing for a permanent growth effect, which is not a prediction of the neoclassical growth models. If this hypothesis is indeed correct, it would suggest that the long-run effect of capital account liberalisation on the emerging market economies should be recast in terms of its effect on levels of aggregate economic variables and on welfare.

² Coulibaly (2009) uses an event study based on imposition and removal of an economic embargo on South Africa. He documents a negative effect of the embargo on economic growth and a positive effect of the removal of the embargo on growth.

³ Additional references on the effect of capital account liberalisation on the cost of capital include Kim and Singal (2000) and Martell and Stulz (2003).

Collateral benefits

Despite the progress made over the last decade in understanding the effect of capital account liberalisation on economic activity in emerging market countries, some unresolved issues remain. In studies – mostly event studies – that have found positive effects of capital account liberalisation on economic growth, the magnitude of the output growth effect implied by the observed boost in investment and the capital stock falls short of the actual growth rates observed in the periods following capital account liberalisation (see Henry (2003) for a detailed discussion). In other words, following capital account liberalisation, output grows at a rate faster than can be justified by the increase in the capital stock given the share of capital in production. If the observed growth in output following capital account liberalisation cannot be fully accounted for by the liberalisation, what other forces are at play when the capital account opens?

A tentative answer to this puzzle lies in the growing literature on the “collateral benefits” of capital account liberalisation. According to this literature, the benefits of capital account liberalisation do not just operate through the cost of capital and investment. Opening capital accounts serves as an important catalyst for a number of indirect benefits. These indirect benefits include development of the domestic financial markets, improvements to local institutions, and better macroeconomic policies (Prasad et al (2006)). It is also conceivable that the presence of knowledgeable foreign investors increases competition and forces local market participants to become more efficient. The better governance, competition, and the enhanced efficiency that ensues could possibly explain the additional growth observed in the years after a country opens its capital account. Indeed, studies have documented an important increase in total factor productivity following the liberalisation of capital accounts, which could account for the additional increase in output growth. In a more recent study by Prasad et al (2008), the authors find that de jure capital account openness has a robust positive effect on total factor productivity growth, but the effect was less clear for the de facto financial integration.

However, more research is needed to establish a causal link between capital account liberalisation and total factor productivity growth. Empirical testing of the collateral benefits hypothesis is complicated by the wrinkle that proponents of this view also argue that, for these collateral benefits to kick in, a minimal degree of financial development needs to be in place already, which they call the required “thresholds”. This introduces considerable non-linearities in the relationships, where the relationships between capital account liberalisation, growth, and the variables that could be considered the outcomes of collateral benefits change at levels that are uncertain. Thus, while the hypothesis of collateral benefits is intriguing, it is in its infancy, and more research in this area is needed to evaluate its merits. However, it does reiterate that appropriate sequencing of policy changes may be very important; and the ideas of collateral benefits threshold effects are also related to the earlier literature on contingent effects. In sequencing policies of capital account liberalisation, the possible increase in vulnerability arising from volatility in cross-border flows has to be weighed against the potential benefits.⁴

An outline of the Report

The aim of this Report is to shed further light on these issues by examining what has happened in the major emerging market economies over the past decade. The CGFS asked this Working Group to pay particular attention to the implications for the financial system (see the mandate prepared in May 2007 in Annex 1). In preparing this, the Working Group has had considerable assistance from central banks, from academics and from representatives of the financial industry (see Annex 3).

Chapter B summarises the main trends of aggregate capital flows in the 2000s. It analyses the macroeconomic factors that have determined the volume and the composition of capital flows. Weak or unstable macroeconomic conditions in capital-importing countries can lead to destabilising forms of capital flow. Macroeconomic conditions in capital-exporting countries can also exert an influence.

Chapter C reviews how the composition of capital flows – mainly foreign direct investment, portfolio investments (equity and debt securities) and flows intermediated through banks – has changed over time. The composition of flows does have a significant bearing on monetary policy dilemma. It also has major implications for the sustainability of flows, for the nature of risk-sharing and for financial stability more generally.

An unusual feature of the most recent period of heavy capital inflows to the large EMEs is that they have not been “needed” to finance current account deficits. In fact, the EMEs as a group have had a large and growing current account surplus. Several countries have resisted currency appreciation. Only part of the foreign currency inflow (that is, from the current account surplus plus private capital inflows) has been recycled by institutional and other private sector investors from the emerging markets (private capital outflows). The monetary authorities have in effect done the bulk of the recycling (that is, via increased foreign exchange reserves). This has had major monetary and financial implications that are analysed in Chapter D.

Chapter E explores the various linkages between capital flows and the development of local financial markets. It also explores how the correlations between local and international financial markets have changed in recent years. There is clear evidence that domestic financial markets in most EMEs have become both broader and deeper than a decade ago.

The increased size of the local operations of international banks in EMEs is examined in Chapter F. Foreign banks have often spurned a shift in bank lending from the commercial sector to households. The macroeconomic and financial implications of such lending are quite different from those of the direct cross-border lending in foreign currencies that characterised earlier periods. In some cases, the growth and structure of international bank lending has given rise to some financial stability returns.

Chapter G documents the large increase in capital outflows from EMEs over the past decade. The marked home bias of local investors is weakening. Households in the EMEs have begun to increase the share of foreign assets in their portfolios, directly and indirectly via institutional investors.

⁴ According to Reinhart and Reinhart (2008), capital inflow “bonanzas” tend to be associated with economic crisis (debt defaults, and banking, inflation and currency crashes). Similarly, Calvo (2008) concluded that the probability of a “sudden stop” of capital flows initially increases in the early stages of financial integration but then gradually decreases, and is virtually nil at high levels of integration. Emerging markets largely stand in a grey area between developed and other developing countries, where the probability of a sudden stop is the highest, suggesting that financial integration can be risky when not accompanied by the development of institutions that will support the use of more sophisticated and credible financial instruments.

Chapter H provides a preliminary assessment of the impact on EME capital flows of the financial crisis that started in the main centres in August 2007. Although the immediate impact was limited, a major adverse impact has developed since August 2008. A prolonged period of deleveraging in the financial system in major countries, a loss of confidence in large financial firms and an extreme lack of liquidity across financial assets have had a dramatic effect on exchange rates, equity prices and bond yields across the emerging market world. This has confronted policy makers with many difficult dilemmas.

Many of the trends analysed in this Report are too new to permit definitive conclusions. The links between openness to international capital flows and economic welfare are in any case very complex. Nevertheless, one theme recurs in the chapters that follow: a larger number of EMEs now satisfy the macroeconomic and financial system preconditions needed to fully realise the benefits of international capital mobility than was the case even a decade ago. Nevertheless, capital flows cause changes in financial exposures that need to be monitored, even for countries with a comparatively well developed financial system. Many members of the Working Group viewed capital account liberalisation as a process to be managed, and the challenges this poses for policymakers are discussed in the final chapter.

B.The macroeconomic context of capital flows

Introduction

The very sharp rise in capital flows into the EMEs in the period 2002–07 took place in macroeconomic and financial circumstances quite different than in the past. Importantly, such flows were not “needed” to supplement inadequate savings. Indeed, since 1999, increases in saving rates have outpaced investment while the current account balance of EMEs as a whole has not only been in surplus but has also expanded rapidly (Graph B1).⁵ Private capital outflows from emerging market (EM) residents have also risen sharply (as local pension funds have diversified into foreign assets (see Chapter G) and as local corporations have expanded their operations overseas), in effect partly recycling these surpluses. But they have not expanded enough to offset the growing current account surpluses and capital inflows, resulting in an accumulation of foreign exchange reserves at an unprecedented level.

¹ Includes 142 emerging and developed countries as defined by the IMF World Economic Outlook Database; as a percentage of GDP. Source: IMF, World Economic Outlook Database, October 2008.

This chapter looks at the main macroeconomic or financial factors behind the major trends of capital flows over the period 2000–07. The first section summarises how capital flows have evolved over the decades. The policy issues raised by capital inflows into EMEs are not new. This brief historical review therefore illustrates the central importance of the macroeconomic context, underlines the significance of sustainable exchange rate regimes, shows how the form that capital movements take matters a great deal, and outlines the policy dilemmas that have arisen. The second section examines net capital flows within the context of current account developments, drawing comparisons with the previous episode in the 1990s.⁶ The final section discusses the factors that drive these flows.

⁵ While the aggregate current account surplus might be contributed by a few large countries, the number of surplus countries rose steadily from 22 in 1998 to around 50 in the mid-2000s. Among those surplus countries, the median current account surplus increased from $0.2 billion to over $3 billion over that period.

⁶ The volume and composition of gross capital flows, which have an important bearing on financial stability issues, are discussed in Chapter C.

Capital flows in historical perspective⁷

Standard neoclassical theory suggests that international capital movements should respond to differences in expected rates of return on capital across countries. Accordingly, capital “should” flow from high-income countries to developing countries (where capital/labour ratios are lower and the productivity of capital higher), boosting growth for some years and allowing developing countries to run current account deficits. In a world with perfect capital markets, capital flows can be used to smooth consumption or finance profitable investment opportunities.

One qualification to this perspective is that the expected variance of returns also matters: potential investors can be deterred from investing in developing countries because of greater risks.⁸ Nevertheless, allocating capital to where risk-adjusted returns are higher should raise global welfare.

A second qualification is that capital flows are also known to be volatile. Expected returns can change sharply. A shock in one emerging country can lead foreign investors with limited local knowledge to indiscriminately withdraw from several countries (“contagion”). In addition, financial or monetary shocks in investor countries can greatly destabilise capital movements in emerging markets. There have been many instances of sharp and very disruptive reversals in capital flows hitting small open economies. Many of the policy debates about capital flows to the EMEs depend on how much emphasis is placed on these disruption costs compared with the gains from a more efficient use of global capital.⁹

The neoclassical perspective broadly fits the pattern seen from the late 19th century up to 1914. Capital flowed to developing areas where the expected return on capital was high. The associated current account imbalances were larger, measured in relation to GDP, than in subsequent periods. Four features of this classical period are worth noting:

• First, flows were almost entirely denominated in gold standard currencies, mainly sterling – nominal exchange rates were in fact fixed. This arrangement also served to stabilise inflation expectations.

• Second, the main investment vehicle was bonds, and nominal long-term interest rates were comparatively stable.

• A third, partly related, feature was that the range of financial assets was extremely limited. Denomination of contracts in gold standard currencies and the stability of long-term interest rates eliminated much of the need for financial diversification and hedging. In any case, the high costs of communication and of computation impeded the development of such activities. Hence the forms that capital flows took were much more uniform than has been the case in recent decades. Nor were there the huge two-way flows of capital that prevail today.

• The final feature was that much of the movement of capital was long-term in nature, going to finance investment in capital-intensive infrastructure and other real investment. The scope for profitable foreign investment was considerable at that time because real output was expanding twice as fast in capital-importing countries outside Europe than in capital-exporting countries in Europe and because of confidence that bonds would be honoured. Increased real investment led to a deterioration in the current account of recipient countries so that the transfer of capital could be “requited” without a change in the real exchange rate.

⁷ This section draws on BIS (1995), Calvo, Goldstein and Hochreiter (1996), Lamfalussy (2000), Kindleberger (1973) and Turner (1991).

⁸ This may explain the “Lucas paradox” of capital movement, ie in recent decades capital has moved in the opposite direction to that implied by capital/labour ratios (Lucas (1990)). Inadequate protection of investor rights may be another explanation (Alfaro, Kalemli-Ozcan and Volosovych (2008)).

Because of these features, many of the problems associated with capital flows in more recent decades did not arise.¹⁰

The “golden” age of international capital mobility was not re-established after 1919. Attempts made to stabilise currencies in the mid–1920s did not produce an exchange rate regime which lasted long, nor could the large US current account surplus of the 1920s find a durable counterpart in long-term US investment abroad. Borrowing in Europe and Latin America thus led to defaults on a scale not seen in the 19th century. By the late 1920s, attempts to keep exchange rates fixed among the major currencies increasingly strained monetary policies. The system of open capital accounts with stable currencies had, by the early 1930s, collapsed.

For the following 40 years, the role played by capital flows was greatly limited by government restriction. Under the Bretton Woods system established in 1944, comprehensive capital account restrictions were allowed. Such controls were regarded by many as essential for prudent economic policymaking domestically and for permitting the gradual restoration of liberal trading arrangements internationally. Capital account restrictions also allowed the government to control the level of interest rates and gave local banks (often subject to government direction) the effective control of domestic finance.

The 1950s and 1960s were decades of substantial trade liberalisation and strong global growth. Although most countries maintained a tight control on capital movements (despite some easing), their effectiveness became progressively weaker. But the strong pressures that built up on major exchange rates from the mid-1960s were not mainly due to “autonomous” capital flows but to divergent current account positions. Governments resisted these pressures for several years, responding with a series of exchange rate realignments (Brittan (1970)).

The advent of generalised floating among the major currencies in March 1973 occurred against a background of strongly stimulative fiscal and monetary policies in the largest industrial countries. Domestic money growth in the industrial world reached rates unprecedented in the postwar period.¹¹ Nevertheless, the adoption of floating exchange rates did enable current account surplus countries to regain control of domestic monetary conditions in the face of very strong global inflation pressures.

The 1973–74 oil shock accentuated inflation pressures and created a particularly unstable structure of capital flows. The oil-producing countries’ surpluses were placed in short-term deposits with international banks. With recession and large current account deficits curbing fixed investment in the industrial world from 1975, the international banks looked for borrowers in the developing world.¹² These forces helped foster a borrowers’ market in international bank lending, leading to a major underpricing of risk. The central banks supervising the major banks were well aware of these risks but were unable to curb the growth of bank lending.<sup>13

10</sup> But flows in the classical period did share one important feature with recent flows: capital flows were still responsive to cyclical developments in lending countries. UK capital outflows, for instance, tended to increase when depressions at home lowered local interest rates (Imlah (1958)). Deane and Cole (1966) demonstrate that, from 1875 to 1914, foreign and domestic investment moved in opposite directions. Bordo (2008), who provides a useful summary of more recent literature, argues that “sudden stops” were also important in emerging market crises pre-1914: for instance, a rise in the Bank rate of the Bank of England would cut off capital flows to the emerging markets and produce a contraction of domestic demand that would often lead to banking crises.

¹¹ On this, see Black (1977).

Capital flows to the developing world rose sharply for several years, taking the form of short-term (or variable rate) bank lending in dollars (or other international currencies). Banks provided finance at rates much lower than long-term bonds (even World Bank finance) in the same currency (Lessard and Williamson (1985)). Capital inflows were in effect used to finance fiscal deficits or sustain private consumption – and not necessarily lift domestic capital formation (Turner (1995)).¹⁴ In many instances, capital inflows led to large currency appreciations which, by making imported goods cheaper, encouraged consumption (Ffrench-Davis and Reisen (1998), Rodrik and Subramanian (2009)).

It was this unstable form of capital flow that created the currency mismatches and short-duration debt structures that played a key role in almost all financial crises affecting the EMEs in the 1980s and the 1990s. Heavy reliance on such forms of capital flow, and the substantial subsequent adjustment costs, made policymakers very wary of reliance on external finance more generally. From the onset of the debt crisis in 1982 until the end of the decade, investors in industrial countries avoided those countries which had borrowed heavily. There were substantial write-offs of bank claims on developing countries and increased risk aversion discouraged new investments: measured outflows from crisis-hit countries increased markedly.

In the late 1980s and early 1990s, however, there was a revival of capital flows to the EMEs as growth in the industrial world picked up. After a short-lived tightening in 1994, US policy rates were reduced in 1995 and the decline in European rates continued: this easing of monetary conditions in major countries increased the supply of low-cost finance through banks and the international capital markets.¹⁵ International bank lending moved from Latin America to the rapidly growing Asian developing countries (“Tigers”). The issuance of debt securities, mostly denominated in dollars, in international capital markets increased. East Asia and Russia benefited from declining risk premia on sovereign (and even bank) debt. Substantial current account deficits were financed in ways that created large risk exposures (see below). This period of expansion in capital flows, punctuated by the Mexican crisis at the end of 1994, really came to an end only with the Asian and Russian crises in 1997–98.

These crises demonstrated that capital flows into countries with weak banking systems and underdeveloped capital markets create huge risks:

¹² This summarises the flows from industrial to developing countries in this period. Flows between industrial countries also became less stable. Unlike the situation prevailing pre-1914, private flows between major industrial countries tended to compound, rather than offset, current account imbalances – exchange rate expectations, rather than returns on real investments, came to dominate capital flows. Strong currency appreciation put a heavy burden on the traded sector of current account surplus countries, often leading to an easing of monetary policy that was unwarranted on domestic grounds.

¹³ Lamfalussy (2000, pp 9–13) describes in some detail the options considered by the Governors of the G10 central banks.

¹⁴ This experience prompted at that time much analysis about the sequencing of reforms – the optimum order of economic liberalisation. Particularly notable was the work of McKinnon (1993) who concluded, “Only when domestic borrowing and lending take place freely at equilibrium (unrestricted) rates of interest and the domestic rate of inflation is curbed so that ongoing depreciation in the exchange rate is unnecessary, are the arbitrage conditions right for allowing free international capital mobility”.

¹⁶ The importance of liquidity risks in national balance sheets was examined in detail in BIS (2000).

¹⁷ A similar debate took place after the 1982 debt crisis: Lessard and Williamson (1985) argued that the sources of finance needed to be broadened and financial instruments had to achieve a better distribution of risks and rewards. See also Santiso (2003) for a perspective on the political economy of financial markets.

¹⁸ One book that brings together research on contagion from many different angles is Claessens and Forbes (2001).

¹⁹ See Financial Stability Forum (2000).

²⁰ Korinek (2008) presents a model of external financing decisions in countries prone to collateral-dependent financing constraints. Decentralised agents do not internalise how their repayments could accentuate selling pressure on their country’s exchange rate in a crisis. Hence they take on too much systemic risk with their borrowing and so impose an externality on the rest of the economy.

Developments in the dollar and nominal effective exchange rates are summarised in Table B1.
¹ 1994–2008 = 100 (ie the average over the past 15 years).
Source: BIS.

After the Asian crisis, net private capital inflows to EMEs fell abruptly to around one third of the pre-crisis levels. Most of this reflected declines in the more volatile or short-term types of flow: cross-border lending by foreign banks shrank. Foreign direct investment held up comparatively well. The currencies of crisis-hit countries gradually recovered, although real effective exchange rates remained below their pre-crisis levels.

A renewed upswing set in from 2002, as global measures of financial market size (equity market capitalisation, outstanding debt securities, etc) rose strongly. Inflows increased sharply and generally assumed forms that were more sustainable, much less likely to provoke external financing crises than in the past (see Chapter C). Reliance on short-term foreign currency denominated debt flows was greatly reduced: currency mismatches were generally avoided. Perhaps most importantly of all (discussed more fully in Chapter E), the domestic financial system became more resilient: long-term local currency debt markets developed; equity markets deepened; derivatives markets developed; and domestic financial firms became stronger. The intermediation channels through which capital flowed have thus become much more diversified. Aggregate outflows have grown significantly (partly intermediated by local institutional investors) and a large current account surplus has emerged (Table B2).

The greater attractiveness of EME assets was also the result of improved macroeconomic performance. Table B3 summarises three key differences between 1993–96 and 2004–07 (the two periods when there were large capital inflows to the EMEs):

• The EM growth advantage has widened. Changes in Brazil, China and India mean that the potential growth rate of the developing world has risen.

• EM saving ratios have risen while the average in the industrial world has fallen (mainly because of the decline in the United States). In earlier episodes (particularly in Latin America), capital inflows were often used to supplement declining domestic saving rates rather than to increase fixed investment. Using capital flows to boost consumption reduces future domestic disposable income. While potential output is unchanged, foreigners’ claims on income have increased.

• EMs’ fiscal positions are stronger.

Have policies of monetary accommodation in advanced economies also shifted demand towards EME assets? Measured by both real short-term interest rates and broad money, the policy stance in advanced countries was more accommodative in 2004–07 than in 1993–96 (Table B4). Given that potential growth rates in the EMEs are higher than in industrial countries, the monetary policy stance appears to have been even more accommodating – on these two simple metrics – in emerging Asia and other EMEs outside Latin America. This has been particularly true in the past two years. However, policies of monetary accommodation in the EMEs have themselves been influenced by external factors (Hannoun (2008), see below for a discussion of the role of the “global liquidity” factor).

At the same time, capital outflows have risen as EM investors (including central banks) have invested heavily in industrial country assets and (to a lesser extent) in assets of other EMEs. By the end of 2006, total holdings of foreign assets by major EMEs excluding China had reached $6.7 trillion, compared with $3.2 trillion in 2001.²¹ Including China, the total rose to $8.4 trillion.²² A large part of the increase in foreign assets was due to reserve accumulation; but excluding the reserve assets, EMEs still held $5.3 trillion foreign assets in 2006 (Table B5).

²¹ Henceforth, unless otherwise stated, major EMEs include Argentina, Brazil, Chile, China, Colombia, the Czech Republic, Hungary, Hong Kong SAR, India, Indonesia, Korea, Malaysia, Mexico, Peru, the Philippines, Poland, Russia, Singapore, South Africa, Taiwan (China), Thailand, Turkey and Venezuela.

²² China first reported international investment position data in 2004.

This rise in both foreign assets and foreign liabilities of all major EMEs as a whole shows that these countries are playing an increasingly important role in the process of financial globalisation. Nevertheless, the scale of this shift has not matched that of the EME share of world trade. Data computed by Lane and Milesi-Ferretti (2008) show that, while the advanced countries’ share of world trade has fallen from around 70% in 1990 to under 60% by 2006, their share of cross-border financial positions has continued to rise.²³ According to these authors, the main explanatory factor for cross-country differences in integration with global financial markets is the depth of the domestic financial system: as domestic markets deepen and grow in sophistication, they argue, the private sector’s capability to both acquire foreign assets and sustain issuance of foreign liabilities improves. (This subject is discussed further in Chapter E).

The accumulation of current account surpluses has lifted the net external asset position of most EMEs in Asia and Latin America. Between 2001 and 2006, net foreign assets held in Asia (including China, Hong Kong SAR and Singapore) and Latin America rose from 8% and –37% of regional GDP to 21% and –29% respectively. Excluding China and the two banking centres, net foreign assets held in Asia improved from –11% to –2% of regional GDP over the same period. In contrast, the current account deficits in central Europe, Turkey and South Africa have resulted in deterioration in external balance sheet positions.

Capital flows in the 2000s

According to the IMF World Economic Outlook, net private capital flows to 143 emerging markets and five small open economies rose from $90 billion in 2002 to around $600 billion in 2007. The Institute of International Finance (IIF) estimates a larger figure of $780 billion for 2007 for a smaller number of countries.²⁴ For consistency, unless otherwise stated, capital flow statistics from the IMF will be the baseline data for analysis in this Report, supplemented with other sources when necessary. Furthermore, in order to focus on the key issues, this report will analyse developments in the major EMEs and refer to other countries only for specific developments. As shown in Graph B3 (left-hand and centre panels), the group of major EMEs captures well the broad trends in capital flow developments – both in US dollar terms and as a percentage of combined GDP.

The recent episode of net private capital inflows to EMEs draws comparison to the experience in the early 1990s when foreign capital inflows also expanded at a rapid pace. During the earlier period, a large volume of foreign private capital, mostly in short-term and foreign currency, has helped finance a current account deficit – an excess of domestic investment over savings – in many emerging economies. As a result, these EMEs were highly exposed to the risks of currency and maturity mismatches. In the event of a loss of confidence, holders of short-term paper were able to demand quick repayment, forcing local issuers to meet their foreign debts by prematurely liquidating longer-dated assets or by sharply reducing spending. These mismatches aggravated the financial crises in 1990s of many EMEs. The sudden capital reversals often took place through banking withdrawals, which are recorded as “other investment” in the balance of payments accounts (Graph B3, right-hand panel).

²³ They show that a typical EME has a much smaller cross-border asset and liability position (a median of 70– 80% of GDP) than an advanced economy (a median well over 200% of GDP).

²⁴ Annex 4 compares the differences of three commonly cited capital flow data sources: the IIF, IMF World Economic Outlook and World Bank Global Development Finance databases.

¹ Comprises 142 emerging and developing countries and five small open economies as defined in the IMF WEO 2008 October database. ² Argentina, Brazil, Chile, China, Colombia, the Czech Republic, Hungary, Hong Kong SAR, India, Indonesia, Korea, Malaysia, Mexico, Peru, the Philippines, Poland, Russia, Singapore, South Africa, Taiwan (China), Thailand, Turkey and Venezuela. ³ In billions of US dollars. DI = direct investment; PI = portfolio investment; OI = other investment. Sources: IMF, Balance of Payments Statistics and World Economic Outlook; Central Bank of China (Taiwan).

The current episode differs from previous cycles in most regions, with the exception of central Europe, countries have been recording strong current account surpluses or manageable deficits in recent years (Graph B4).²⁵ Under the balance of payments identity, ignoring the over- or understatement of the recorded components (net errors and omissions), the sum of private capital inflows and current account balance should equal to the change in reserve holdings by the monetary authorities. Thus large private capital inflows in combination with observed current account surplus have meant that foreign exchange reserves have increased markedly in most EMEs. Between 2000 and 2007, major economies accumulated more than $2 trillion of foreign exchange reserves. Foreign exchange reserve accumulation on this scale has important macroeconomic and financial implications. This will be examined in detail in Chapter D.

The current account deficits and heavy reliance on foreign financing in central Europe, South Africa and Turkey have become more vulnerable to a reversal in capital inflows (Graph B3, right-hand panel). The Baltic states and several southeastern European countries are in a similarly vulnerable position, with current account deficits around 15% of GDP and high levels of external debt. Box C3 in the next chapter compares the vulnerability of emerging Europe now with East Asia prior to the 1997–98 crisis, and a main source of vulnerability in these countries – cross-border banking flows – will be discussed in Chapter F.

²⁵ Among the three countries grouped under “others”, South Africa and Turkey have been recording current account deficits in recent years; however, they were offset by the large surpluses in Russia.

¹ China, Hong Kong SAR, India, Indonesia, Korea, Malaysia, the Philippines, Singapore, Thailand and Taiwan (China). ² The Czech Republic, Hungary and Poland. ³ Argentina, Brazil, Chile, Colombia, Mexico, Peru and Venezuela. ⁴ Russia, South Africa and Turkey.
Sources: IMF, Balance of Payments Statistics and World Economic Outlook; Central Bank of China (Taiwan).

The macroeconomic drivers of capital flows

Economists have long debated the factors that drive international capital flows. One popular distinction is between those that reflect conditions in investor countries (external factors) and those that are specific to the recipient economies (domestic factors).²⁶ The proponents of the external factor view argue that capital flows into EMEs rise when financing conditions in investor countries ease. Such flows may also move inversely with the business cycles in industrial countries. Others argue that the demand for foreign capital depends largely on prospective returns on domestic investment. In practice, the balance between external and domestic factors is likely to vary according to circumstances.

External factors

Ample “global liquidity” is often cited as an important external factor that has “pushed” capital into EMEs. The term global liquidity, however, is rarely precisely defined. It is generally associated with a wide range of price and quantity measures of the monetary stance in industrial economies. These include low world interest rates and strong money growth in developed countries (see IMF (2007b), pp 34–7, Box 1.4).

Several members of the Working Group noted the dominant role of the US dollar in the global economy. The widespread denomination of international contracts in dollars and the use of the dollar as the reference currency for many developing countries with fixed or quasi-fixed exchange rates meant that the impact of movements in US dollar rates on global economic conditions was much greater than a simple calculation of the US weight in the global economy would suggest.²⁷ In some EMEs, dollar interest rates have a large impact on the financing decisions of households and firms.²⁸ In addition, lending by international banks tends to respond to monetary policy or liquidity shocks in their home country.²⁹

²⁶ See Fernández-Arias (1996), Eichengreen and Mody (1998); and Ferrucci et al (2004) for a summary of the external and domestic factors.

²⁷ On this subject, see Greenwald and Stiglitz (2008).

In any event, changes in interest rates in investor countries are likely to influence capital flows to the EMEs. There are several channels. First, lower interest rates in the main centres encourage international investors, particularly those with shorter investment horizons, to search for higher yields elsewhere. Second, for the net debtor countries, low international interest rates reduce their servicing costs and indirectly improve their creditworthiness. Third, low international rates encourage borrowers in emerging markets to borrow in international currencies rather than in their own domestic currency.

The left-hand panel of Graph B5 shows that declining real interest rates in the G3 economies in the early 1990s and 2000s indeed coincided with rising capital flows to EMEs. In addition, the turning of interest rate cycles, especially in the United States, also matched closely the starting dates of the Latin America debt crisis in the 1980s and the Mexican crisis in 1994. In recent years, however, this link has been harder to detect – perhaps because US policy rates have changed in a more gradual way and have remained comparatively low. For instance, capital flows to EMEs continued to expand between mid-2004 and late 2007 when real interest rates in the United States and the euro area were rising.

One quantitative complement to the interest rate-based measures of global liquidity could be one based on money growth. The right-hand panel of Graph B5 shows the GDP-weighted changes of base money of the G3 economies. This indicator reveals a strong monetary expansion between 2002 and 2004, which coincided with the beginning of the current episode of strong capital flows to EMEs. But again, the sharp contraction in money growth from 2005 to 2007 did not trigger sharp movements in capital in the opposite direction as in earlier episodes.

¹ In per cent.² GDP-weighted change over three years for the G3 economies.Shaded areas are capital flow cycles.
 Source: IMF, World Economic Outlook.

In addition to the accommodative monetary stance in the industrial countries, some emerging market economies themselves have contributed to the significant increase in global liquidity by channelling their excess of domestic saving over investment abroad (Bernanke (2005)). An excess of domestic saving over investment in many EMEs and resistance to currency

²⁸ In central Europe, interest rates in the euro and the Swiss franc exert a similar effect because households borrow from local banks in these currencies.

²⁹ Cetorelli and Goldberg (2008) analyse a very large set of bank-specific data on intragroup flows to show that the lending of foreign offices of US banks is affected by US monetary policy.

¹ In per cent. ² Savings as a percentage of GDP; GDP-weighted. ³ Up to 2002, 10-year nominal Treasury yield minus 10-year inflation expectations from survey data; from 2002, 10-year inflation-indexed Treasury yield.
Sources: Federal Reserve Bank of Philadelphia; Federal Reserve Board; IMF, International Financial Statistics.

A related external factor is the business cycle in industrial countries. International business cycles can have both positive and negative impacts on capital flows to EMEs. For example, the economic downturns in most industrial countries during the early 1990s made investment opportunities in EMEs more profitable and encouraged flows into them (Calvo, Leiderman and Reinhart (1996)). When economic conditions in the industrial economies started to improve in the mid-1990s, this factor became less important in driving capital flows to EMEs. However, it is also noted that stronger growth in industrial countries may boost profitability of local firms, which will in turn increase their gains abroad (Ferrucci et al (2004)).

While the external macroeconomic factors discussed above could contribute to the cyclical behaviour of international capital movements, a microeconomic factor – global investors’ portfolio diversification – could lead to a more stable trend growth in capital flows to EMEs (see CGFS (2007a)). In particular, the aggregate financial assets of pension funds and insurance companies in the euro area, Japan, the United Kingdom and the United States reached $34 trillion at end-2007. A small percentage increase in their portfolio allocations to EM assets could potentially generate substantial capital flows to the EMEs. Data on institutional investors’ asset allocation to emerging market assets are not readily available, but the IMF Coordinated Portfolio Investment Surveys show that the holdings of emerging market equities and debt securities by developed countries and emerging market economies have increased steadily since 2001 (Graph B7)

³⁰ Hannoun (2008) argues that the global credit excesses that led to the current financial market turmoil had their origins in: first, accommodative monetary policies; second, large global imbalances and forex reserve accumulation; and third, rapid financial innovation. He analyses the multidimensional linkages between these three powerful forces.

³¹ Following Jones and Obstfeld (2001), the saving ratio is calculated implicitly, via the current account identity, as the sum of investment rate and the ratio of current account to GDP. The investment rate is calculated as the ratio of gross domestic capital formation (gross fixed investment plus changes in stocks/inventories) as a percentage of GDP. Note that the saving ratio defined as such incorporates all measurement error from both investment and the current account.

¹ In billions of US dollars. ² Australia, Canada, Sweden, Switzerland, the United Kingdom and the United States.
Sources: IMF Coordinated Portfolio Investment Survey.

Domestic factors

A major element that “pulls” international capital to the EMEs is higher expected risk-adjusted returns. Better growth performance and a more stable macroeconomic outlook are thus important external factors. Graph B8 shows that the pace of net private capital flows into EMEs tends to pick up whenever the growth in the emerging markets surpasses that in the advanced economies.

Equally, the volatility of growth and inflation has come down sharply from the 1990s, when so many countries went through very disruptive crises, and is now close to that prevailing in industrial countries (Table B6). Reforms of macroeconomic policies played a key part. For example, the successful disinflation programmes implemented in many Latin American economies during the late 1980s and early 1990s, which were accompanied by fiscal adjustment, helped reduce perceived risk on real domestic investment and stimulate capital inflows.

As discussed more fully in Chapter G, institutional reforms and policies that ease the access of foreign investors to domestic financial markets, such as the removal of capital controls and the liberalisation on foreign direct investment, have also facilitated international capital movements.

In sum, the accommodative monetary stance and low interest rates in major industrial countries have played some part in encouraging both equity and debt flows to EMEs, particular in the early stage. At the same time, the strong economic outlook and sound domestic policies might appear to be important factors in attracting foreign capital into these countries. The fact that capital continues to flow into EMEs after monetary policies were tightened in the past few years up to August 2007 might suggest that improved fundamentals could outweigh other factors in international investors’ investment decisions (this is discussed further in Chapter G). Furthermore, emerging economies which have surplus domestic savings over investment export their surpluses officially via foreign exchange reserve accumulation or through private channels. The domestic financial implications of these official outflows will be discussed in the next chapter while the intermediation of private capital outflows will be examined in Chapter G.

¹In per cent.² Comprises 142 emerging markets and five small open industrial economies; in billions of US dollars.
Source: IMF, World Economic Outlook Database, October 2008.

C. Composition of capital flows and financial stability

Composition and risk exposures

The composition of capital flows matters for monetary policy, for the management of liquidity and for financial stability. It matters for monetary policy because some forms of capital flow are more sensitive to the central bank’s policy rates than others. At one end of the spectrum, short-term capital flows are typically very sensitive to domestic short-term rates (given exchange rate expectations). Because domestic short-term rates are regarded as a key policy device, restrictions on cross-border investments in short-term debt instruments have often been the last to be removed.³² It matters for the management of liquidity because the maturity or duration structure will influence the choice of instruments for sterilisation.

Capital flow composition matters for financial stability because it determines how risks are shared between provider and recipient (and so defines risk exposures), impacts domestic fixed capital formation, affects the cyclical sensitivity of aggregate flows and influences the future pattern of international adjustment.³³

From the perspective of financial stability, four dimensions are of general importance:

a. Equity versus debt. Equity forms of investment serve to transfer risk to the supplier of funds and away from the user of funds.Debt has to be serviced irrespective of the returns earned on the investment financed by borrowing. The servicing of equity liabilities, on the other hand, depends on the returns actually earned.

b.Short-term versus long-term. Borrowers reliant on long-term debt to finance long- term projects are less vulnerable to interest rate and refinancing risks. A decline in the market value of long-term debt paper – eg because of changed market assessment of risk or higher interest rates – will be borne by the lender.

c. Investment versus consumption. Capital inflows to finance current public sector spending allow governments to delay measures of fiscal consolidation. Equally, inflows that in effect finance increased private consumption (eg thanks to exchange rate overvaluation or the sale of domestic assets) can be less welcome than inflows associated with increased real fixed capital formation.

d.Foreign versus domestic currency and tradables versus non-tradables. National balance sheet considerations could suggest that foreign currency inflows should be used to invest in foreign currency earning assets – that is, an expanding productive capacity in tradables, rather than in non-tradables. Otherwise, the country faces currency mismatches.

In addition, borrowing in foreign currency to finance investment in tradables increases productive capacity in the tradable sector and so, at constant terms of trade, tends to increase future trade surpluses to finance foreign currency debt obligations.

³² Attempts by industrial countries to maintain such controls in the 1960s and 1970s were, however, partly frustrated by the growth of the eurodollar market (ie bank deposits denominated in currencies other than that of the country in which the bank is domiciled). An excellent account of this issue is Chapter 24 of Caves and Jones (1973). The recent growth of NDF markets offshore for many EME currencies has a similar effect of undermining the effectiveness of capital controls (see Chapter E).

³³ Williamson (2005) provides a careful analysis of stabilising properties of the different types of capital flows. He argues that countries need to be as concerned about the form in which they borrow as with the aggregate level of borrowing.

According to this – admittedly simple – perspective, foreign direct investment (FDI) is a “good” capital flow from the point of view of the potential vulnerability of the capital-importing country. It represents equity rather than debt, is long-term rather than short-term, and is associated with increased domestic capital formation rather than increased consumption (public or private).

Portfolio investment into equities also provides useful risk diversification. As noted in Chapter A, there is evidence that increased foreign investment in equities can lift the rate of domestic fixed capital formation for several years. At the other end of the spectrum, reliance on short-term foreign currency debt is the most risky strategy.

In practice, however, the hedging strategies of investors and borrowers can blur these sharp distinctions: an FDI investor, for instance, may simultaneously take on short-term hedges that can be liquidated very rapidly.

Macroeconomic policies (fiscal policies, exchange rate regime, etc) have often had a major influence on the composition of capital flows. Countries with poor fiscal discipline tend to become too dependent on short-term or volatile foreign finance (Rajan and Tokatlidis (2005)). Another example is the exchange rate regime. The combination of pegged exchange rates (or some equivalent) and high interest rates in inflation-prone countries creates attractive arbitrage opportunities and so induces large volumes of foreign investment funds into short-term debt instruments.

Mexico faced this problem in the early 1990s – but the adoption of a flexible exchange rate in 1995 led to a significant reduction in debt flows (Sidaoui (2005)). In short, an undesirable structure of foreign capital inflows often reflects poor macroeconomic policies.³⁴ The sequencing of capital amount liberalisation measures (and the periodic re-imposition of controls) can also influence the composition of capital flows. There is evidence (eg De Gregorio et al (2000)) that capital controls on short-term debt instruments, but not on bonds, can lengthen the maturity of debt inflows.

Volatility

Another approach to gauge the stabilising (or destabilising) properties of particular forms of capital movement is to examine historical volatilities. Most studies concur that FDI is regularly the most stable form of capital flow: see, for instance, the recent study by Felices et al (2008).³⁵ Portfolio investment tends to be more volatile, and there is some evidence that such flows are procyclical. Within the portfolio category, Felices et al report evidence that flows into equities in EMEs over the period of 1980 to 2007 have been more stable than flows into debt securities. Many have found that the most unstable source of finance is cross-border bank lending.

The volatility of flows is often a symptom of more fundamental causes. Box C1 summarises a recent study by Broto et al (2008) of the determinants of volatility. The authors examine the impact of a set of domestic macroeconomic and domestic financial factors and global factors on the volatility of gross inflows (ie, net purchases by non-residents of domestic assets) of FDI, portfolio investment and other investment (mostly bank credit). Particularly notable is their finding that the development of the domestic financial system tends to reduce the volatility of portfolio and banking flows.

³⁴ And, it must be added, imprudent bank lenders – see Lamfalussy (2000).

³⁵ Dobson and Hufbauer (2001) reach this conclusion in their survey of the literature.

Box C1

Factors affecting the volatility of capital flows: some empirical evidence

Although the volatility of capital flows has long been a concern for policymakers in the EMEs, there is only limited empirical evidence on the determinants of such volatility. Broto et al (2008) examine the impact of a set of domestic macroeconomic and domestic financial factors and global factors on the volatility of gross FDI, portfolio and other investment (mostly bank credit) flows. They use a panel of 48 emerging economies, covering the period from 1980 to 2006. The key results are highlighted below (Table C1.1).

First, global factors have become more important in determining the volatility of capital flows, particularly in the case of FDI flows. Specifically, since 2000, movements in international equity prices, as measured by the S&P 500, are correlated with the volatility of all three types of capital flows. But global activity (measured in terms of world GDP growth) and global liquidity are both negatively correlated with FDI volatility. This confirms the sensitivity of FDI to the global cycle found by other studies.

Second, FDI is the flow whose volatility is more associated with macroeconomic soundness. Higher GDP per capita, the ratio of reserves to imports (a measure of self-insurance) and lower inflation in these countries all reduce the volatility of FDI.

Third, development of the domestic financial system is relevant for all types of flows but is particularly important for portfolio and banking flows. This could help explain the volatility of all three types of flows, especially in more recent periods. The size and competitiveness of the domestic banking system (as measured by assets/GDP and interest rate spreads in the banking sector) are negatively related to the volatility of FDI and banking flows. But these factors are related to higher volatility of portfolio flows only in the recent years. The depth of the domestic stock markets (as measured by the stock market capitalisation) is negatively associated with the volatility of portfolio and other flows.

Despite the increasing importance of global factors in explaining the volatility of capital flows, domestic policies can help reduce their volatility, particularly those aimed at reinforcing the depth and soundness of the domestic financial system and, for the specific case of FDI flows, developing policies to maintain macroeconomic soundness.

Finally, some studies have examined what might be termed the “crisis-resistance” of different forms of capital flows. Levchenko and Mauro (2007), for instance, examine episodes of “sudden stops”, which they define as a year-to-year decline in total net financial flows of more than 5 percentage points. They find that equity-like forms of financial flows (and FDI in particular) have proved to be very stable during such episodes. But bank-intermediated flows fall sharply and often remain depressed for several years. Portfolio debt flows also drop but tend to recover relatively quickly.

Changing composition

The composition of capital flows (in both directions) has changed significantly over the past 25 years. Graph C1 shows the evolution of gross private debt (portfolio debt and other investment) and equity (direct and portfolio equity investment) flows since 1980 (see Box C2 for definitions of gross inflows and outflows). The volume of capital flows in the 1980s was relatively small and comprised more of debt flows. Entering the 1990s, equity inflows started to catch up with debt inflows. The onset of the Asian crisis in 1997 had led to a sharp reversal in debt flows but a slight expansion in equity flows in the subsequent years. The increase in equity flows partly reflected the sales of distressed EME assets to foreign investors. Debt flows started to recover in the early 2000s, contributing to a more balanced pattern of capital flows.

¹ In billions of US dollars. Includes Argentina, Brazil, Chile, China, Colombia, the Czech Republic, Hungary, Hong Kong SAR, India, Indonesia, Korea, Malaysia, Mexico, Peru, the Philippines, Poland, Russia, Singapore, South Africa, Taiwan (China), Thailand, Turkey and Venezuela. ² Includes portfolio debt investment and other investment. ³ Includes direct investment and portfolio equity investment.
Sources: IMF, Balance of Payments Statistics; Central Bank of China (Taiwan).

Box C2

Definition of gross private capital flows¹

Terms used for the different measures of international capital transactions are potentially ambiguous. The conventions used for flow measures in this report are as follows:

Gross inflows refer to purchases by non-residents of domestic assets less their sales of such assets. Similarly, gross outflow measures the purchases by residents of foreign assets less their sales of such assets. Therefore, the gross flow measures are effectively consolidated magnitudes which capture the balance of all increases and decreases of foreigners’ and residents’ holdings of domestic and foreign assets respectively. Accordingly, there can be negative reported gross inflows or outflows.

Net flows are defined as gross inflows less gross outflows.

Conceptually, private capital flows should equal total flows less all official components. But when one side of transactions is an official entity, the deal may be recorded as official flows in one country and private in another. For consistency, private capitalflows are defined with respect to the reporting country by taking out the portfolio and other investment asset components that are attributed to monetary authorities and general government.

¹ The data are derived from the IMF Balance of Payments Statistics. There are two main groups of accounts: the current account, and the capital and financial account (previously known as capital account). Data on private capital flows used in this report are derived from three standard components of the financial account: direct investment, portfolio investment and other investment. Another component – financial derivatives – is not included in our compilation of flow measures as reported figures are largely negligible for most EMEs. Each of the three components separates the transactions between residents and non-residents. For direct investment, there are directional distinctions (abroad or in the reporting country). The portfolio investment and other investment components are divided into assets and liabilities. On this basis, gross inflows include items recorded as direct investment in that country, and its portfolio and other investment liabilities, while the outflow measures comprise direct investment abroad, portfolio and other investment assets.

¹ In billions of US dollars. Includes Argentina, Brazil, Chile, China, Colombia, the Czech Republic, Hungary, Hong Kong SAR, India, Indonesia, Korea, Malaysia, Mexico, Peru, the Philippines, Poland, Russia, Singapore, South Africa, Taiwan (China), Thailand, Turkey and Venezuela.
Sources: IMF, Balance of Payments Statistics; Central Bank of China (Taiwan).

¹ In billions of US dollars. Includes Argentina, Brazil, Chile, China, Colombia, the Czech Republic, Hungary, Hong Kong SAR, India, Indonesia, Korea, Malaysia, Mexico, Peru, the Philippines, Poland, Russia, Singapore, South Africa, Taiwan (China), Thailand, Turkey and Venezuela.
Sources: IMF, Balance of Payments Statistics; Central Bank of China (Taiwan).

Foreign direct investment

Inflows

Foreign direct investment has been the major source of capital flows into EMEs since the 1980s. Throughout that decade, total FDI to the EMEs was stable at less than $50 billion. But entering the 1990s, FDI inflows accelerated, and by the end of the decade, total FDI flows reached $250 billion, or one third of the world total (Graph C4). The spate of financial crises in the late 1990s had slowed FDI flows into the EMEs, but they rebounded strongly in recent years to reach a record of $350 billion in 2006.

From the international investors’ perspective, the main criteria for investing in emerging market firms include competitive labour costs, growth prospects and the size of the host country. In this context, perhaps it was not surprising to see the rapid increase in FDI to EMEs in the early 1990s concentrated mostly in Asia, and China in particular. The 1997–98 financial crisis in Asia might have affected investor confidence in the region, but post-crisis market liberalisation such as granting foreign investors easier access to local firms, and the acquisition of distressed assets, have led to a strong expansion in merger and acquisition (M&A) activity in the 2000s. Furthermore, China, boosted by accession to the World Trade Organisation in 2001, continues to be the largest recipient of FDI among the EMEs, accounting for around 20% of the total inflows in 2006. Part of this, however, could be attributed to the “round-tripping” of Chinese savings for tax purposes (see World Bank (2002)). India, another fast-growing country in the region in recent years, has recorded FDI inflows which are modest when compared to those of China, but which have been observed to increase substantially in 2007 and 2008.

Sources: IMF, Balance of Payments Statistics; Central Bank of China (Taiwan).

Privatisation has played an important role in the large FDI flows in Latin America and central Europe during the past two decades. In Latin America, an important part of the structural reforms which started in the 1980s was to privatise government-owned firms. These privatisation programmes gained force during the late 1980s. From 1980 to 1992, a total of more than 1,600 publicly owned firms were privatised in Argentina, Brazil, Chile, Mexico and Venezuela. But behind the regional trend were some country-specific characteristics. While the increase in FDI was largely driven by privatisation of state-owned assets in the service sectors (particularly telecommunications and banking) in Argentina, Brazil and Chile, investments in Colombia, Peru and Venezuela concentrated in the commodity-related sector.³⁶ The signing of the North American Free Trade Agreement in 1992 boosted investments in Mexico. In the 2000s, the events in Argentina and the slowdown in the privatisation process in other countries contributed to a reduction in FDI flows to Latin America in recent years.

¹ In billions of US dollars.
Sources: IMF, Balance of Payments Statistics; Central Bank of China (Taiwan).

Rising portfolio capital inflows have reflected portfolio diversification by international institutional investors such as pension funds and insurance companies in industrial countries (see CGFS (2007a)). The fact that most EMEs liberalised their equity markets more than a decade ago made this possible (Bekaert et al (2005)).³⁹ It is difficult to obtain information on institutional investors’ international asset allocation, but IMF data suggest that total portfolio investment assets held by major industrial countries rose from $574 billion in 2001 to almost $2 trillion in 2006; the share of equities increased from 60% to 74% over the same period.

The pattern of equity flows appears to have reflected relative economic performance between source and recipient countries. Funds tended to flow into countries where growth was accelerating. Accordingly, interest in emerging market equities strengthened considerably in recent years. Between 2002 and 2007, emerging stock markets have outperformed most mature markets by a great margin (Graph C6, left-hand panel).

³⁹ Note that portfolio debt flows in some EMEs such as China and India are still not fully liberalised

¹ 1990–2005 = 100. ² Includes both Chinese-owned (H shares) and affiliated companies (red chips), in billions of US dollars. Sources: Bloomberg; Hong Kong Stock Exchanges and Clearing Limited.

The large volume of equity flows was also driven by a significant increase in international placements by emerging market companies. In 2006, firms in China, Mexico, Russia and Thailand all recorded cross-border initial public offering (IPO) deals worth more than $1 billion. Chinese companies in particular have raised a cumulative $169 billion in the Hong Kong stock exchanges since 2000, of which $104 billion were through IPOs (Graph C6, right-hand panel). Apart from fund-raising, the listing of companies in major financial centres could help improve corporate governance because of the more established legal and accounting standards in these centres.

Outflows

Despite the strong gross portfolio equity inflows, major EMEs on average recorded net private portfolio flows abroad between 2000 and 2006, suggesting even stronger gross flows in the opposite direction. The main component in driving these gross outflows was the acquisition of foreign debt securities by local residents, particularly in Asia. In 2006, the gross outflows almost doubled from the previous year to $264 billion, of which $109 billion was from China. Given that the capital account in China is not fully liberalised, a large part of these transactions might be sourced from other quasi-government entities such as state-owned commercial banks. This highlights a deficiency in measuring the true private sector flows using balance of payments data as they may contain flows related to other non­government and non-central bank official entities such as sovereign wealth funds.⁴⁰

Bank flows

Banks act as agents in virtually all capital account transactions. Balance of payments statistics classify these transactions under other headings wherever possible (eg direct investment, portfolio investment or changes in reserves). Bank flows that are not captured in these categories are classified as other investment (which also includes other external transactions such as trade credits and accounts receivable and payable). Thus, the other investment category is essentially a residual item that covers capital transactions through banks, but the bank flows in this category are not included elsewhere in the balance of payments statistics.

To some extent, banks’ role is to “take up” the counterpart foreign exchange exposure desired by non-banks; their exposures are then passed on to others (eg central banks). The right-hand panel of Graph C2 shows that bank inflows into EMEs under the other investment category are large and volatile. A striking point was that gross inflows turned “negative” in the few years after the Asian financial crisis, implying that foreign banks sold more emerging market assets than they purchased during those years. Similarly, the negative outflows from EMEs over the same period meant that domestic banks’ sales of foreign assets have exceeded their purchases (Graph C3, right-hand panel). In sum, in the aftermath of the financial crisis, foreign banks withdrew large amounts of capital from Asia while Asian banks reduced their holdings of foreign assets.

It should be noted that substantial amounts of these banking flows were directed through the two banking centres in Asia – Hong Kong SAR and Singapore – during the early post-crisis years. As McCauley and Mo (1999, 2000) noted, a large part of these outflows reflected the reversal of the round-tripping of funds from Japanese banks in Japan to Hong Kong SAR and back to non-banks in Japan. Excluding the flows from the banking centres, the aggregate gross inflows and outflows of EMEs were less volatile (the red lines in Graph C7).⁴¹

⁴⁰ Under the current balance of payments framework, private portfolio flows are calculated by taking out those related to monetary authorities and general governments from the total figures.

⁴¹ It should also be noted that Hong Kong SAR first reported balance of payments statistics in 1998, so the banking centre total was just the capital flowing in and out of Singapore.

¹Other investment flows that are related to banks; in billions of US dollars.² Banking centres: Hong Kong SAR and Singapore.
Sources:IMF,Balance of Payments Statistics;Central Bank of China (Taiwan (China)).

In recent years, non-resident bank flows have returned to the EMEs. Emerging Asia, Hong Kong SAR and Singapore in particular accounted for more than half of the transactions; there was also strong growth in gross bank inflows into Russia and Turkey.

National balance of payments data, however, often fail to fully capture the transactions between residents and non-resident banks. The BIS banking statistics provide useful supplementary information. In particular, the BIS consolidated banking statistics, which are based on the nationality of the reporting bank and include the positions of their foreign subsidiaries, provide valuable insight into international banks’ changing strategies.

Until the mid-1990s, international bank lending to the EMEs primarily took the form of cross-border lending in dollars (or other foreign currency). Since then, however, international banks have greatly increased their local presence. The result has been a substantial increase in local currency lending through their local affiliates (Table C3).

Despite the recent rebound in US dollar terms, the BIS consolidated banking data reveal that the size of international banking claims relative to GDP in most EMEs has been lower than that in the 1990s (Table C4). Notable exceptions are Turkey and central Europe, where the size of international claims is much higher than the average in the early 1990s. While the current account in central Europe can be compared to the pre-crisis Asian experience, some commentators started to draw parallels between the recent expansion of international banking claims in central Europe and the Latin American experience in the 1980s, when some regional economies held substantial amounts of bank-intermediated external debt (see Box C3). However, unlike Latin America in the 1980s, the banking systems in central Europe are now largely foreign-owned with more advanced risk management. Furthermore, instead of using the external funds to finance the local import substitution industries, the increase in emerging market external borrowing is taking place against the background of rapid economic and financial integration with the euro area.

Cross-border bank borrowing has played an important role in the emerging market financial crises in the 1990s. First, heavy short-term borrowing of domestic banks in the international interbank market creates a risk for large-scale sudden reversals of bank loans. Second, domestic banks which borrow in foreign currency and then onlend to finance domestic projects face a currency mismatch (or pass this mismatch on to their domestic customers). The BIS data show that the share of short-term international claims in total claims fell sharply in Asia after 1997, and like in other regions, the ratio has stabilised in recent years (Graph C8, left-hand panel). The ratio of international claims on banks in total also declined steadily between 1997 and 2002 and has since stabilised in all regions (Graph C8, right-hand panel). These developments add support to the perception that the nature of emerging market bank flows is less risky now than in the mid-1990s.

¹ In per cent. ² With remaining maturity up to and including one year.
Source: BIS.

Box C3

There are clear similarities between the current economic and financial conditions in emerging Europe and East Asian economies prior to the 1997–98 crisis. The crisis was preceded by rapid growth in credit to the private sector, with a significant share of loans denominated in foreign currency. East Asian economies also recorded large current account deficits, mainly induced by the private sector. These deficits were financed by strong debt inflows, which reversed sharply following the crisis (Chart C3.1). A further similarity lies in exchange rate policies. Prior to the crisis, East Asian economies had fixed nominal exchange rates (in their case against the US dollar). Moreover, the economies relied heavily on a single foreign creditor – Japanese banks. Emerging European countries currently show a similar level of dependence on a few European banking system creditors. For example, claims by Austrian-owned banks are equivalent to 20% of annual GDP in the Czech Republic, Hungary and Slovakia, while claims of Swedish-owned banks on the Baltic states are equivalent to 90% of their combined GDP. An adverse shock to one or more of these foreign banks could result in them withdrawing funds from emerging European countries.

¹The Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Slovakia and Turkey.² Indonesia, Korea, Malaysia, the Philippines and Thailand.² Also includes holdings of currency and deposits and trade credits. Source: IMF, International Financial Statistics.

But there are also clear differences. First, the strong growth in external and domestic credit in emerging Europe may anticipate a strengthening in GDP growth and policy frameworks due to closer EU integration. Second, the growth in domestic bank credit has been concentrated mainly in the household sector, partly through mortgage lending, rather than in the corporate sector as was the case in East Asia. Third, a large part of most emerging European banking systems is foreign-owned. These banking groups appear to be financially strong currently, as reflected in standard – albeit backward-looking – measures of financial strength such as capital adequacy ratios and profitability. The foreign subsidiaries should have better risk management techniques in place, more geographically dispersed assets and, in principle, good supervision (from the home country on the consolidated entity). In contrast, Japanese banks, the largest foreign creditor to east Asian economies prior to the crisis, were facing big losses on their domestic exposures at the time. Moreover, the increase in information on borrowers is higher than was the case in 1997 reducing the likelihood of an ill-informed liquidity run. These issues may now merit closer examination in the light of the current ongoing financial crisis.

Arguing in the other direction, the main supervisor of the consolidated entity for many emerging European countries is likely to be the foreign (home) one. They might be less concerned about any systemic risk that is felt in the host rather than home country (ie the risk to a European EME of borrowing from a foreign creditor is likely to be bigger than the risk to the creditor from lending to the EME). The banks – and possibly their supervisors – might also be encouraged by the strong profit margins they are making currently on business in the region without fully factoring in the probability of future losses should risks materialise.

D. Intervention, sterilisation and domestic financial intermediation

Introduction

Substantial inflows of foreign capital to the EMEs in the 2000s have not financed current account deficits. Instead, in contrast to earlier episodes, they have in almost all cases taken place against the background of widening current account surpluses. In effect, foreign currency inflows beginning in 2000 have resulted in a substantial increase in foreign exchange reserves (Table D1).

The unusual context for capital flows in this decade has implications for absorption in the macroeconomic sense of the term. That is, a rise in capital inflows achieves a real transfer of resources only to the extent that the current account deficit (present or future) increases.⁴² In that sense, the absorption of capital inflows into the domestic economy has, in aggregate and over this time period, been limited.

A second implication is that current account surpluses have probably driven up market expectations of long-run equilibrium exchange rates – whatever policies were being pursued in capital-importing countries. Unlike capital inflows – particularly portfolio inflows – which may be prone to sudden reversals, current account surpluses can endure and thereby can have persistent effects on the exchange rate. Therefore, investors and currency traders have had a tendency to associate persistent current account surpluses with an appreciation of the long-run equilibrium exchange rate.

The payments pattern in this decade has arisen in part because of policy choices in the EMEs. As noted in Chapter B, these policy choices have been conditioned by the accommodative monetary policies in the advanced economies. A series of crises in the 1990s made many governments in the EME world more wary of reliance on capital inflows, particularly when such inflows were attracted by the combination of excess demand and high nominal interest rates that often resulted from large budget deficits. Most governments became convinced that a much larger buffer of foreign exchange reserves was needed if their economies were to be able to withstand possible future difficulties in external financing.

⁴² This condition is analogous to the famous “transfer problem” that Keynes and others analysed in the 1920s. The effective transfer of resources by Germany to pay for reparations imposed after World War I required not only fiscal surpluses to raise local funds but also current account surpluses to effect the real transfer.

That desire for a buffer was the main motive for reserves accumulation in most EMEs during the early 2000s. Several EME central banks echoed the argument of Obstfeld et al (2008) that greater financial openness and the increased size of domestic financial liabilities that could be converted into foreign currency had lifted desired precautionary reserves holdings. As Reddy (2002) has stressed, the cost of holding reserves can be viewed as an insurance premium to maintain confidence.⁴³Sharp turbulence in global financial markets during September and October 2008 – associated with difficult dollar funding conditions for banks worldwide – put substantial pressures on markets in EMEs. As discussed more fully in Chapter H, substantial foreign exchange reserves helped several EMEs cushion the effect of these disturbances on their financial systems.

Some EME central banks represented in the Working Group noted that, even when an objective of building reserves was in place, central bank purchases of foreign exchange were designed to manage undue volatility in the exchange rate so that undershooting or overshooting could be prevented. By the later 2000s, however, most of the larger EMEs had met or exceeded conventionally accepted thresholds for precautionary reserve holdings (Table D2), and thus limiting the pressures of currency appreciation associated with foreign currency inflows has gradually become the dominant objective. Such behaviour has also been influenced by the perception of volatility in capital flows arising from monetary and financial developments in the advanced economies (Chapter B, Graph B5).

After experimenting with a number of fixed and managed exchange rate regimes over many years, Australia eventually responded by floating the exchange rate and removing longstanding capital controls in 1983. Although the country has continued to face high and sometimes volatile capital flows, they are no longer the focus of, nor a constraint on, monetary policy.

Between 1931 and 1983, when the country went through four different fixed exchange rate regimes, volatile capital flows threatened on several occasions to destabilise domestic financial conditions so much that abrupt currency realignments were forced upon the authorities.

After pegging successively to sterling and the US dollar from 1931 until September 1974, the Australian dollar was pegged to a basket of trade-related currencies. The aim was to reduce the Australian dollar’s exposure to fluctuations in the US dollar. In an attempt to regain control of monetary conditions, Australia devalued its currency 17.5% against the basket in November 1976. With the aim of avoiding the need for such large adjustments in the future, the regime was then changed so that the exchange rate could be adjusted each day. The adjustments under the crawling peg became progressively larger and more frequent as it became obvious that they were insufficient to give the Reserve Bank control over monetary conditions in the face of sometimes volatile capital flows.

These crawling peg arrangements, however, failed to prevent heavy capital outflows in the run-up to the election in the first half of 1983. The new government devalued the currency 10% within days of taking office. The move left a perception in the market that speculators could precipitate a large exchange rate adjustment despite a regime designed to discourage exactly that outcome.

During the second half of 1983, the situation was reversed as exports grew strongly, the current account recovered, and Australia's high interest rates encouraged capital inflows, which again threatened to destabilise domestic monetary conditions. Some technical changes were made to the crawling peg system, but they had little effect.

In December 1983 the Australian dollar was finally floated, and controls on foreign exchange wereremoved. The Reserve Bank was no longer obliged to clear the foreign exchange market at a predetermined rate. Participants took positions against one another rather than the central bank

By the time the decision was made to float, the change had gained widespread political support. Another important factor was the fact that Australian banks had gained market experience by participating in the non-deliverable foreign currency hedge market (which effectively circumvented exchange controls but which the authorities had not sought to suppress) and in trading in third currencies.

Since 1983, Australia’s exchange rate has floated freely, and the Reserve Bank has been able to determine monetary conditions without the distraction of having to sterilise volatile capital flows. The Reserve Bank has intervened on occasion since the adoption of the floating currency, but at no time has it sought to effectively fix the exchange rate through its intervention. Pressures in the foreign exchange market that previously bore on domestic liquidity now bear on the exchange rate, and at times the exchange rate has changed by large amounts.

Swings in the exchange rate, whatever their source, have proved to be less destabilising than the volatility in domestic financial conditions that Australia experienced under earlier fixed rate regimes. The economy has coped relatively well with large movements in the exchange rate; indeed, those movements have been an important part of Australia’s adjustment to substantial changes in the terms of trade, both upwards and downwards. In addition, the pass-through from large currency depreciation into domestic prices has not been as great as was initially expected, partly because of more competitive product markets both in Australia and abroad and an increasingly more flexible industrial relations system in Australia.

These changed arrangements do not mean that capital flows have become unimportant to Australia – its current account deficit still averages around 4½% of GDP. But they do mean that even large shocks to capital flows do not pose the large problems for monetary policy that they did under capital controls, a pegged exchange rate and regulated financial prices. The 1997 Asian crisis is a case in point.

At the onset of the Asian crisis in mid-1997, monetary policy was guided by an inflation target. The Australian economy was growing at around trend rates, with domestic demand beginning to accelerate and underlying inflation running below 2%. Given the inflation performance, monetary policy had been eased over the previous year, as required by the 2–3% inflation target. Thus, the shock hit the Australian economy at a time when it was in reasonably good shape and its monetary policy stance already relatively expansionary. The brunt of the negative shock was mostly borne by the floating exchange rate, and the Australian dollar depreciated about 20%.

On some previous occasions, such a large currency depreciation of the currency had led to both a pickup in inflation and a rise in inflation expectations because of higher import prices. The rise in inflation expectations required an increase in interest rates to contain, and eventually reverse, the inflation impulse. In contemplating whether that policy response was appropriate on this occasion, the Reserve Bank came to the view that, even though inflation was forecast to rise above 3% in the short term as the depreciation was passed through to consumer prices, performance was likely to be consistent with the target thereafter. The forecasted rise in inflation was not expected to be persistent, partly because the contractionary impulse from the decline in export demand would dampen growth. But, in addition, the credibility of the inflation target served to keep inflation expectations in check. The flexibility of the monetary policy framework allowed the validity of this judgment to be reassessed as time passed. In the event, inflation rose less than was forecast. By the end of 1998, not only had the Reserve Bank not lifted interest rates, it had actually reduced them slightly.

So the flexible inflation target served as a useful framework within which to manage the effects of the Asian crisis and the policy response to capital flows. Allowing the exchange rate to move provided a part of the mechanism that helped the economy adapt to the crisis and the changes in capital flows that it had brought about. This mechanism limited the disruption to the domestic economy. Capital flows and exchange rate changes did not compromise the conduct of monetary policy, which remained relatively expansionary, given with the needs of the economy. In the world of fixed exchange rates, the Reserve Bank would not have been able to set policy in this way.

¹ This box is a summary of Veale (2008), on the Working Group’s website. See also Stevens (2006).

With the exception of Hong Kong and Saudi Arabia (whose currencies are pegged to the dollar), most EMEs in principle maintain managed floating exchange rate regimes. In practice, however, the degree of exchange rate management varies appreciably across countries. How far EME currencies have been allowed to move can be measured in several ways. One is to examine historical volatilities. The volatility of daily exchange rates has a significant impact on short-term capital flows because interest rate arbitrage positions are safer the more stable are spot exchange rates. The volume of currency carry trades, for instance, tends to fall when volatility rises. The greater the degree of volatility, the greater the degree of monetary independence that is being maintained. Calculations of annualised volatilities of daily exchange rates are shown in Table D3.

Compared with the yen and the euro, most emerging Asian currencies have been significantly less volatile; whereas the Brazilian real, the Colombian peso, the Turkish lira and the South African rand have been much more volatile. But the latter have not been more volatile than the Australian dollar (which was affected, as discussed in Box D1, by commodity prices). It is also clear from Table D3 that some currencies that were unusually stable only a few years ago (eg India and Malaysia in 2003–04) have become more flexible. China’s exchange rate, however, has remained very stable.

A second approach is to examine the size of medium-term changes. The Brazilian real and Russian rouble have changed most, both against the dollar and in nominal effective terms. During the period 2004–07 Brazil and Russia both allowed their currencies to appreciate substantially in real terms and increased their reserves via large-scale intervention in foreign exchange markets.⁴⁵ Thus the potential inflationary impact of intervention in both countries has been offset by substantially higher real exchange rates. The Indian rupee has risen but by much less,⁴⁶ and the Chinese renminbi has been the most stable. The Indian context has differed from that of other Asian countries by exhibiting a consistent current account deficit and even larger and rising trade deficits. Moreover, the Indian rupee has made significant two-way movements in the medium term. Since 2000 the currencies of other Asian countries have risen about 20% against the dollar but have been broadly stable in nominal effective terms.

Members of the Working Group were also very aware that issuing local currency liabilities to finance foreign currency assets exposes the public sector to a currency mismatch. At the same time, several members stressed that ex ante policy choices given the information available to policymakers at the time looked rather different than this ex post characterisation might suggest. As potential growth rates in EMEs rise and capital inflows increase, one can expect that the equilibrium real exchange rate needs to rise as capital inflows increase, or growth prospects improve or the terms of trade rise. But policymakers will not know how far and how quickly the equilibrium real exchange rate has to rise in practice. In addition, the desirable policy response in the short term depends on three additional elements:

• One is whether the positive shocks are temporary or permanent. If temporary, the authorities may want to intervene or resist changes to the exchange rate. Many view surges in capital inflows (especially in portfolio inflows) as temporary and unpredictable and therefore worth counteracting with short-run interventions. If intervention continues much longer than expected, however, central banks or governments will have to re-examine their initial assessment of the permanent versus temporary nature of flows. They will also need to design their intervention instruments in such a manner that ex post developments can be managed.

• A second element concerns the formation of exchange rate expectations in the short term. Allowing the exchange rate to move towards its long-run equilibrium level should help to choke off speculative inflows. Short-run volatility in the daily spot rate may have a similar deterrent effect. The counterview is that the market may have only a very imprecise idea of the long-run equilibrium exchange rate – particularly for rapidly developing economies. Accordingly, some believe that an appreciation may merely increase expectations of further appreciation, attracting yet further capital inflows. Extrapolative expectations can create a major dilemma for policymakers.

⁴⁵ The real effective value of the Brazilian real doubled between January 2003 and June 2008.

⁴⁶ Between January 2006 and January 2008 the Indian rupee has appreciated 12.4% against the dollar and 0.8% in nominal effective terms. BIS real effective exchange rates (Chapter B, Graph B2), which use consumer prices as the deflator, may not be very accurate measures of competitiveness of countries where volatile food prices dominate the movement in consumer prices.

• A third element of relevance for policymakers is that the adjustment costs for the domestic economy may be much lower when the exchange rate moves only gradually to a new equilibrium level. As forex reserves have risen and inflation pressures have emerged, several EME central banks have become more willing to accept currency appreciation. For almost all EME central banks represented in the Working Group, a market-determined exchange rate was their medium-term goal; and several said that prolonged intervention had to be accompanied by a willingness to accept currency appreciation. But few believed that a fully flexible exchange rate alone would be enough to “take care” of capital inflows at all times. Views diverged on the precise balance to strike between exchange rate flexibility and intervention in responding to strong currency inflows.⁴⁷

A key policy dilemma

A major dilemma for policymakers arises when foreign exchange intervention to hold down the exchange rate leads to inflation. While generalisation is difficult (given the very diverse circumstances in the EMEs), the nature of the recent economic cycle had eased this dilemma and thus played a key role in allowing high levels of intervention to be sustained for several years. Until 2005 or so, many Asian countries accumulating reserves had faced few inflation pressures, given substantial excess capacity in their economies and growing manufacturing capacity in other countries, which was holding down world prices of tradables. Thus, easing policy rates in the face of upward pressure on the currency entailed only minimal inflation risks. As noted in Chapter B, policy rates in some EMEs were reduced more than in the major countries. In those circumstances, reserve accumulation did not always create the dilemma policymakers had faced in earlier episodes in which high inflation forced them to choose between their inflation targets and their exchange rate objectives. The steady rise in inflation rates since 2006, however, has brought this dilemma back.⁴⁸

Capital controls for monetary policy purposes?

Large inflows of foreign capital can further complicate monetary policy. Several countries have used controls on capital inflows to give them somewhat greater independence in setting monetary policy.⁴⁹

As discussed in Chapter C, there are good grounds – in a phased liberalisation of capital controls – for freeing foreign direct investment and portfolio equity flows before allowing non­residents to conduct debt-related transactions such as portfolio flows into domestic debt securities and international debt issuance by local firms. In addition, there are grounds for limiting short-term flows; hence restrictions on short-term instruments are frequently maintained longer than other restrictions. Such restrictions are often supported by the imposition on non-residents of minimum holding periods for domestic securities. In the 1990s several countries resorted to new market-based controls. They were generally able to drive a wedge between domestic and foreign interest rates. Nevertheless, most also had to intervene in foreign exchange markets. Box D2 outlines Chile’s experience. The Working Group did not conduct an analysis of the efficacy of capital controls. However, it was agreed that in certain circumstances controls on capital inflows could in principle, at least in the short run, help monetary policy by moderating the size or the volatility of inflows and by modifying their composition in favour of more stable flows. Temporary measures of control could be particularly appropriate when volatile capital movements reflect abrupt changes in financial market conditions in advanced economies and are unrelated to conditions in recipient countries.<sup>50

47</sup> Another issue of relevance to EMEs is that many of their labour-intensive exports are much more sensitive to exchange rate movements. Hence abrupt movements in the exchange rate can have a significant impact on domestic industrial growth and employment, which is then difficult to reverse.

⁴⁸ As Vargas and Varela (2008) put it, drawing on Colombian experience, “a negative output gap made it possible for the central bank to engage in forex intervention at the same time that it was easing monetary policy ... but as the output gap closed and inflation pressures began to emerge, sterilised intervention lost effectiveness”.

⁴⁹ For a useful taxonomy of policy choices that incorporate capital controls as well as fiscal and monetary policies, see Ghosh et al (2008).

Controls on capital inflows for monetary policy purposes: the example of Chile

Heavy capital inflows in 1991 presented the Chilean authorities with a choice between lowering interest rates (in the face of strong demand) or allowing the currency to appreciate in real terms. In order to ease this dilemma, the authorities put in place market-based regulations designed to reduce the incentives for capital inflows while avoiding outright prohibition.

• The cost of short-term external financing was raised by the imposition of an unremunerated reserve requirement (URR). This requirement evolved over the years to a dollar denominated one-year mandatory deposit on nearly all foreign capital inflows associated with foreign debt or with foreign portfolio investment. The effectiveness of the measure, which did not cover all flows, gradually declined over time as traders exploited loopholes in the requirement. In September 1998, after macroeconomic conditions had changed (capital inflows weakened in the wake of the Asian crisis), the URR rate was lowered to 0%.

• A minimum holding period for direct and portfolio investment from abroad was aimed at limiting “in and out” financial operations performed by large institutional investors.

• The international public offerings of Chilean securities were subjected to tighter regulation. Over time, these regulations have been gradually eased.

Summarising the many studies of Chile’s experiences, Desormeaux et al (2008) report that, on the empirical evidence, the URR had only limited effects on the home-foreign interest rate differential, on the appreciating trend and on total capital inflows. Several studies also suggest that the microeconomic costs were significant: the domestic financial services industry was held back, and the policy discriminated against small and medium-sized firms that could not access long-term financing to reduce the burden of the tax.

The use of sovereign wealth funds may also serve as a mechanism for keeping capital inflows out of the domestic banking system. A summary is provided in Annex 5.

⁵⁰ On the wider issue of capital controls, which involves the question of financial stability as well as that of monetary policy, which is the focus of this chapter, the Working Group shared the conclusion of the FSF report in 2000: “The use of controls on capital inflows may be justified for a transitional period in the face of very strong inflows or as countries strengthen the institutional and regulatory environment in their domestic financial systems, especially if the process of liberalisation had not been carried out in a well-sequenced manner. In other words, some measures to discourage capital inflows may be used to reinforce or complement prudential requirements on financial institutions and other resident borrowers. But it is vital that controls should not be seen as providing a way of allowing countries to pursue unsound macroeconomic policies or to delay actions to strengthen the financial system”, see FSF (2000), page 35.

Intervention and sterilisation: the main elements⁵¹

The accumulation of foreign exchange reserves has significant implications for a central bank’s balance sheet. Table D4 presents a stylised version of a monetary authority’s balance sheet. Its assets consist of foreign currency and domestic assets; its liabilities comprise currency, bank reserves (taken together as monetary liabilities), its own securities, other liabilities (taken together as non-monetary liabilities) and equity capital. In most cases (but not all – see Box D3), foreign exchange reserve assets are financed by the liabilities within the domestic financial system. Of these, currency is largely determined by the public’s demand for cash balances.

An injection of equity capital by the government to finance reserve accumulation would not result in monetary expansion, provided that the injection of capital did not come from deposits or an overdraft facility at the central bank. Without increased equity capital, and assuming other things (ie the demand for cash) equal, the accumulation of reserves requires financing in some form. One simple measure of such a financing need is the excess of foreign currency reserves over currency in circulation. Up until the late 1990s, such a financing gap was either small or negative in many developing countries. In other words, foreign exchange reserve assets and currency in circulation were of a similar order of magnitude.⁵²

In recent years, however, the gap has widened appreciably. In many emerging market economies, it has become very large in relation to the size of the domestic financial system (Graph D1). The large gap means that domestic financing operations of the central bank, which are the counterpart of the accumulation of foreign exchange reserves, are likely to have a significant impact on the operation of the local financial system.

⁵¹ This and the following sections draw on Mohanty and Turner (2006).

⁵² As the economy grows and the central bank is buying forex reserves, the rising demand for cash means that interest-bearing paper issued needs to increase by less than the increase in forex reserves. Conversely, when the central bank is selling reserves the volume of interest-bearing paper retired will have to exceed the decline in reserves by the amount that the demand for cash rises.

Box D3

Circumventing the domestic banking system

In some cases central banks acquire foreign assets by making transactions outside the banking system, sometimes circumventing the foreign exchange market (Moreno (2005)). Such transactions might serve to limit the impact of the accumulation on the banking system. In Chile, for instance, the foreign assets accumulated by the state-owned copper company appear to be held in accounts abroad that are managed by the central bank and that do not enter the domestic banking system. If government deposits arising from copper sales are matched by central bank foreign assets, accumulation of the latter would have no impact on domestic financial intermediation.

In other cases, however, such transactions still feed through to the local banking system. For example, in Mexico, foreign exchange transactions executed by the state-owned oil company (and also by the federal government) are intermediated directly by the central bank outside the foreign exchange market (Sidaoui (2005)). These transactions have been a primary source of foreign reserve accumulation by the central bank. However, as the oil company acquires pesos in exchange for the foreign assets that it deposits in the central bank, the result is an increase in the commercial banks’ balance sheets that closely resembles unsterilised intervention. Indeed, the costs of sterilising the monetary implications of reserve accumulation has been an issue in Mexico and in many other countries.

¹ M1, also called narrow money, comprises transferable deposits and currency outside deposit money banks. ² Weighted average of the economies listed, based on 2005 GDP and PPP exchange rates. ³ Hong Kong SAR, Indonesia, Korea, Malaysia, Philippines, Singapore, Taiwan (China) and Thailand. ⁴ Argentina, Brazil, Chile, Colombia, Mexico, Peru and Venezuela. ⁵ Czech Republic, Hungary and Poland. ⁶ M2 is a broad measure of money which comprises, in addition to M1, time, savings, and foreign currency deposits of resident sectors other than central government.
Sources: IMF; Datastream; national data; BIS.

As intervention takes place, the central bank can cover the financing gap by issuing domestic monetary liabilities (typically commercial bank monetary reserves). Should these increased monetary reserves be allowed to put downward pressure on the short-term interest rate (unsterilised intervention), bank credit would tend to expand, and inflationary pressures would eventually mount. If central banks have a target for the short-term rate (usually the policy rate), they would attempt to offset increases in bank reserves through changes in other balance sheet items (usually selling domestic assets or issuing their own securities) over which they have control (sterilised intervention).

What have been the monetary implications of intervention in practice? Most analyses based on changes in the balance sheet quantities in Table D3 suggest that changes in net domestic credit and non-monetary liabilities (ie not monetary expansion) have generally offset the greater part of changes in net foreign assets.⁵³ In that narrow sense (focusing on base money), the bulk of intervention thus appears to have been sterilised in most countries. In some countries, however, the monetary base has expanded as central banks sought to ease monetary policy in the environment of low inflation and large excess capacity that prevailed before 2006. But an easy monetary stance has continued in several countries even as inflation has risen. As Graph D2 shows, real short-term interest rates have fallen during the past few years and are now very low (particularly compared with the potential growth rate).

But even if base money is kept constant, broad money and credit aggregates are still likely to rise because commercial bank deposits tend to rise with capital inflows.⁵⁴ To counter an unwarranted rise in liquid commercial bank assets, central banks would have to withdraw excess liquidity. How they choose to do so is a matter of some importance.

Withdrawing excess liquidity: non-market versus market-based approaches

The growing size of foreign exchange reserves tends to increase the scale of central bank liquidity operations. Partly because the heavier use of a limited range of existing methods would become increasingly distortionary, many central banks have introduced new techniques to manage liquidity. Each method has its advantages and drawbacks: as Mohan (2008) notes, a key operational issue for central banks is “how much” each approach needs to be used. The choice of method depends not only on the intrinsic merits of particular techniques but also on the conjuncture of financial and economic factors (capital flows, market conditions and monetary developments). After briefly discussing the choice of non-market versus market-based approaches, this section outlines the non-market methods that have been used and some considerations regarding the selection of a specific market instrument.<sup>55

53</sup> See Mohanty and Turner (2006) and Aizenman and Glick (2008). There are, however, some grounds for doubting Aizenmann and Glick’s conclusion that sterilisation has been near perfect. In their regression analysis, they control for nominal GDP growth to account for money demand. While that approach is standard, the control they use would tend to obscure the fact that intervention and reserve money growth have boosted aggregate demand and inflation.

⁵⁴ Banks’ holdings of bonds and bank deposits (M2) both rise: see footnote 20 below.

⁵⁵ For a country by country summary, see Table A4 of Mohanty and Turner (2005), from which the following paragraphs draw.

¹ Overnight interbank rate for Brazil and Chile, one–month interbank rate otherwise; deflated by consumer prices; in per cent.
² Korea, Malaysia, Taiwan (China) and Thailand; weighted average of the econmies listed, based on 2005 GDP and PPP exchange rates.
Sources: IMF; Datastream; national data; BIS


In principle the use of instruments that have a high degree of marketability is preferable to reliance on non-market techniques because market instruments have fewer distortionary effects and do not undermine the capacity of local financial institutions to intermediate credit.

In practice, however, an uncomfortably large price response on market instruments in illiquid or thin markets can make recourse to non-market methods attractive. Hence, the state of development of financial markets can be an important consideration: reliance on market instruments tends to increase as local securities markets become more developed and as local borrowers get greater access to foreign sources of finance. But it is also true that the greater official use of market instruments can itself serve to stimulate the development of debt markets.⁵⁶

Non-market approaches

Three forms of non-market measures have been used: direct controls on bank lending; reserve requirements on commercial banks; and the shifting of deposits to the central bank.


⁵⁶ BIS (2006) discusses difficulties of sterilisation in several African countries (Mozambique, Uganda and Tanzania) where financial markets are not very well developed.

a) Direct controls on bank lending

Since the mid-1980s, most direct controls on bank lending have been dismantled because they undermined the efficiency of financial intermediation. Nevertheless, several countries have used credit ceilings, eg Indonesia (see Goeltom (2008)), and China has used window guidance to curtail lending.⁵⁷

b) Reserve requirements on commercial banks

Reserve requirements are less distortionary than direct controls on bank lending. The attractiveness of relying on increased reserve requirements – particularly great when nominal market rates are high – is their low cost to the authorities.⁵⁸ Indeed, to the extent that compulsory deposits are remunerated at below market rates (or not remunerated at all), their imposition acts as a tax on commercial banks; thus it is the fragility of the domestic banking system that in practice is often the main impediment to a tightening of reserve requirements. Where the banks have a sufficient profit to absorb this tax, such a measure can serve as an effective macroprudential tool. Increasing reserve requirements in the expansion phase of a cycle (eg when capital inflows are boosting the liquidity of the banking system) can provide the authorities with a liquidity “cushion” that can be released when banks face greater funding difficulties (eg because of capital outflows). This proved to be unexpectedly useful during the period of extreme stress in October 2008 (see Chapter H).

One disadvantage of reserve requirements should be noted: their effectiveness tends to decline over time because they encourage the use of alternative mechanisms that circumvent banks. In some cases, firms can even be induced by higher bank intermediation spreads to borrow abroad, thus defeating the very purpose of sterilised intervention. For all these reasons, many central banks have, in normal times, gradually reduced the size of their reserve requirements.

In the episodes of capital inflows in the early 1990s, however, several countries in Latin America did increase reserve requirements, including Brazil, Chile and Colombia. ⁵⁹ Indonesia, Malaysia and the Czech Republic sharply raised reserve requirements in the 1990s. In the recent episode, it is China that has placed most reliance on increased reserve requirements (Graph D3). As noted in Box D4, banks receive an interest rate of 1.89% on reserve deposits, compared with 4.06% on one-year central bank bonds. India, Korea and Taiwan (China) have also increased reserves requirements.

In the Working Group discussions, some central banks noted that because deposits made under reserve requirements do not specify a particular maturity, they represent an effective instrument to achieve a one-time withdrawal of liquidity that is not expected to be reversed in the very near term.

⁵⁷ Peng (2008) argues that such measures, which have served to keep M2 largely in line with the growth of nominal GDP despite increased bank liquidity, have led to credit rationing by banks, which tends to favour the larger and state-owned enterprises at the expense of smaller enterprises.

⁵⁸ In some cases (eg Mexico), however, reserve requirements are remunerated at market-related interest rates.

⁵⁹ See Turner (1995) and references therein. This outlines the various ways that reserve requirements have been tailored as desired: they can be based on average or marginal requirements (ie based on a change in loans); they can differ according to foreign or local currency denomination; and they can apply differentially to non-resident deposits.

c) Shifting deposits to the central bank

A third non-market technique for absorbing liquidity is to shift the bank deposits of state or regulated institutions from commercial banks to the central bank.⁶⁰ During the early 1990s episode of capital inflows, many East Asian economies transferred large amounts of provident fund deposits from the banking system to the central bank at below market interest rates.

Selecting specific instruments for market-based measures

Market-based measures fall into three broad categories: direct borrowing from banks at market rates, the issuance of short-term securities and long-term bonds and repurchase transactions. The choice of instrument has several implications. Sterilisation will be complete only if the paper issued in sterilised operations is long term and is ultimately held by households and non-bank firms. ⁶¹ In addition, the different instruments will impact the balance sheet of various sectors in a different way and will entail different costs to the government (Mohan (2008)).

⁶⁰ See Reisen (1993) and Griffith-Jones and Bhattacharya (2001) for a discussion of the experiences in the 1990s.

⁶¹ This is because as non-banks purchase government bonds they reduce their deposit holding with banks, thus lowering broad money. Ultimately capital inflows change the composition of non-bank balance sheets – higher foreign holdings of domestic assets (purchased from the non-banks) by non-residents are offset by increased domestic bond holding by non-banks. Such sterilisation bonds have to be sufficiently attractive for this adjustment to occur. In other words, a liquid and long-term bond market is essential for complete market- based sterilisation.

Box D4

Sterilisation operations in China

In the face of large and continuing accumulation of foreign exchange reserves in recent years, the People’s Bank of China (PBC) has used a mixture of market-based and non-market based tools, described below, to actively implement sterilisation and flexibly adjust the liquidity in the banking system.

•Issuance of central bank bills. The PBC started to issue three-month, six-month, and one-year central bank bills in 2003. In 2007, the PBC re-initiated the issuance of three-year central bank bills with a view to deeply freezing liquidity in the banking system. At the same time, market-based issuance of central bank bills was combined with targeted issuance. Targeted central bank bills were issued to commercial banks that had rapid credit growth and relatively abundant liquidity. By end-March 2008, the outstanding balance of central bank bills stood at RMB 4 trillion.

•Open market operations of special government bonds. Starting in August 2007, the Ministry of Finance sold a cumulative total of RMB 1.55 trillion of special government bonds in three issues. The proceeds were used to purchase $200 billion of foreign exchange reserves, which was then used as capital for the China Investment Company. The PBC purchased RMB 1.35 trillion of special government bonds through market operations and used them to gradually strengthen repo operations.

•Increases in the required reserve ratio. Between January 2006 and July 2008, the PBC increased the required reserve ratio for financial institutions a total of 10 percentage points, to 17.5%. During the same period, the PBC implemented a higher required reserve ratio for banks with relatively lower capital adequacy ratios and higher non-performing loan ratios. As of mid-2008 the central bank paid 1.89% on reserve deposits, compared with 4.06% on one-year PBC bonds.

•Currency swaps with commercial banks. In November 2005, in an effort to absorb liquidity to a reasonable extent and keep money market interest rates stable, the PBC conducted currency swaps with some banks, exchanging a certain amount of US dollars for renminbi and getting back the original amount of US dollars at the same exchange rate. Accordingly, the PBC charged the interest rate differential between the US dollar and the renminbi.

•Strengthening of window guidance on bank lending. To cope with the rapid expansion of bank loans resulting from excess liquidity, the PBC uses moral suasion to alert banks to the risks arising from rapid increases in lending. The PBC guides banks to make appropriate lending, optimise their credit structures, and stick to the principle of different treatments for different borrowers.

The PBC estimated that these operations helped sterilise more than 80% of the increase in liquidity between 2003 and 2007. The growth rates both of broad money and of RMB loans fell during that period.

However, the effectiveness of sterilisation is facing more constraints with the continuing and rapid increases in foreign exchange reserves. The ratio of the money supply associated with reserve accumulation to total base money in each period keeps increasing. More and more central bank bills are issued to roll over previous bills. Sterilisation operations are also affected by the limitation of sterilisation instruments and rising sterilisation costs. Moreover, the effect of sterilisation operations on the soundness of the banking system, including increases in the required reserve ratio and holdings of central bank bills, also need to be closely monitored.

a) Direct borrowing from commercial banks

An instrument used extensively by many countries in this cycle is direct borrowing by the central bank from commercial banks through an overnight or similar deposit facility.⁶² In Malaysia, direct borrowing by the central bank remains the single most important instrument for sterilising excess liquidity, although the Bank of Malaysia is gradually shifting to the use of repo operations and central bank securities (Ooi (2008)). Another example of extensive use of such an instrument is Hungary.⁶³ The speculative attack on the forint on the strong edge of the exchange rate band in January 2004 resulted in very large capital inflows over a span of few days (about €5 billion in two days). While intervening in the foreign exchange market, the central bank absorbed most of the excess liquidity through the overnight and two-week deposit facility offered to banks. At the same time, the central bank sharply cut the interest rate on overnight deposits to reduce the profitability of speculation. The central bank in the Philippines uses its tiering deposit system (interest rates varying with the amount of the deposit) as an instrument for sterilising intervention. Similarly, the central bank of Mexico offers a special deposit facility to banks at market interest rates to withdraw long-term liquidity from the banking system.

b) Sterilisation securities: maturity

Sterilisation securities have traditionally tended to be very short term – perhaps naturally so when capital inflows themselves are very short term and comparatively modest in size and when inflows in one month are reversed in the next. Although central banks might run up intervention-related debts during an inflow burst, they can run them down during periods of heavy outflow and of depreciation pressures on the exchange rate.

But it is much less natural for sterilisation securities to be very short term when inflows endure so that intervention continues in the same direction for many years. In some cases the issuance of short-term debt securities by the central bank or government – which are particularly attractive to foreign investors who wish to take a short-term position on the currency – may actually encourage additional capital inflows, partly undermining the main aim of intervention.

Several EME central banks noted in the workshops that the issuance of central bank sterilisation notes at somewhat longer maturities allowed them to develop a yield curve for their (nascent) money markets.⁶⁴ In the case of Mexico, sterilisation nowadays involves the selling of securities with maturities of between three and five years. In China, the central bank began to issue three-year securities beginning in 2005 with a view to lengthening the maturity of sterilisation bonds. To the extent that such bonds replace the shorter-term central bank bills, they could have a longer-term impact on the excess liquidity in the banking system, thereby enhancing monetary control.

The operational choices made by the authorities on the maturity of their sterilisation note issuance will depend on several factors. One is their assessment of debt servicing costs – for instance, issuing long-term instruments when market conditions permit. Another is the impact on the interest rate structure and the risk of attracting additional inflows.

⁶² The treatment of such deposits, however, differs; some central banks have classified them as non-market instruments and some have not.

⁶³ See Érsek (2005).

⁶⁴ Ooi (2008) notes, “given the magnitude of surplus liquidity to be sterilised relative to the size of the economy, the advantages of issuing longer-term [central bank] notes is that it facilitates a more efficient longer-term sterilisation and reduces the need for larger turnover of shorter-term sterilisation transactions”.

c) Government versus central bank securities

From the perspective of the consolidated budget of the public sector, the distinction between government and central bank securities would seem unimportant, as both are liabilities of the official sector. And approaches to the use of government versus central bank paper for sterilisation have differed considerably by country. A first approach is for the central bank to issue sterilisation securities. A majority of central banks issue their own securities rather than using government-issued paper.

A second approach has been to use both government and central bank paper. In Mexico, the market for both types of paper has grown simultaneously in the past few years. In Malaysia, the central bank has sparingly used its own securities for sterilisation operations – reserving them mostly for withdrawing longer-term liquidity – while the government issues its own bonds for financing deficits. Similar models are seen across a number of emerging market economies (eg the Czech Republic, Hungary, Turkey and South Africa).

Third, some central banks (eg the Philippines) used to depend exclusively on government issuance for sterilisation operations. The Reserve Bank of India offers a recent striking example of reliance on government issuance. Under a new market stabilisation scheme (MSS, outlined in Box D5) introduced in April 2004, the central government began to issue securities, over and above its own borrowing requirement, exclusively for sterilisation operations. On the grounds of good governance it would seem appropriate that interventions directed by the government should be borne in a transparent way by the government. Showing the costs of intervention – ultimately a claim on the taxpayer – in the budget would promote such transparency and facilitate parliamentary scrutiny.⁶⁵ A further argument for issuance of government rather than central bank securities has a financial market angle. Given that local bond markets in EMEs are usually relatively small and illiquid, issuing large amounts of central bank securities in such EMEs could fragment the market and further reduce liquidity.

One suggestion in that regard has been that the central bank and the government might issue securities of different maturities, reducing the problem of multiple yield curves. Korea has adopted such a strategy. The Bank of Korea’s monetary stabilisation bond commonly stretches out to three years, whereas the government has issued primarily longer-term securities.

d) Repo transactions

In recent years, several central banks in countries with comparatively well-developed markets have introduced repurchase operations on government paper. In many cases (eg Korea and India), such operations are used primarily for the “fine tuning” of liquidity shortages or surpluses rather than for offsetting large or prolonged liquidity inbalances. The Central Bank of Brazil, however, has relied heavily over the past few years on repo operations. Such operations have been conducted in a wide range of maturities – up to seven months. The vast bulk of their repo operations now have a maturity exceeding one month; in 2003–05, by contrast, repo operations were primarily very short term. Singapore and South Africa have also developed longer-term repo transactions.

⁶⁵ The Fiscal Responsibility Law of Brazil led the Central Bank of Brazil to cease issuing its own bonds in 2002.

Box D5

Sterilisation operations in India

In India, a number of instruments have been used to sterilise the impact of the large capital inflows that have been witnessed since the mid-1990s. The liquidity impact of large capital inflows was initially (mid-1990s to 20003) managed mainly through outright open market operations (OMO), ie outright sales of government securities through repo and reverse repo auctions. The OMOs were supplemented by operations under the daily Liquidity Adjustment Facility (LAF). In addition to LAF and OMO, excess liquidity from the financial system was also absorbed through the building up of surplus balances of the government with the Reserve Bank, particularly by raising the notified amount of 91–day Treasury bill auctions, and through foreign exchange swaps. The Reserve Bank of India (RBI) also uses a cash reserve ratio (CRR) as an instrument of sterilisation. In the context of an increase in the CRR, the cost is borne by the banking sector, as CRR balances are no longer remunerated. Greater flexibility in exchange rate movements, further liberalization of capital outflows, encouragement of pre-payment of external debt obligations and modulation of interest rate ceilings on non-resident deposits have also been used to manage the capital account inflows.

In view of the sustained large volume of net capital inflows, the finite stock of government securities at the central bank and the absence of the option of issuing central bank securities under the RBI Act, an internal Working Group on Instruments of Sterilisation (2004) recommended that the government issue a special variety of bills and bonds for sterilisation purposes, called Market Stabilisation Bills/Bonds (MSBs). The MSBs could be used for mopping up enduring surplus liquidity from the system over and above the amount that could be absorbed under the day to day repo operations of the LAF.

On the basis of the 2004 working group recommendations, a new instrument, named the Market Stabilisation Scheme (MSS), was made operational in April 2004. Under MSS, which is meant exclusively for sterilisation purposes, the Reserve Bank has been empowered to issue government Treasury bills and medium-term dated securities for the purpose of liquidity absorption. The scheme works by impounding the proceeds of auctions of Treasury bills and government securities in a separate identifiable MSS cash account maintained and operated by the RBI. The amounts credited to the MSS cash account are appropriated only for the purpose of redemption or buyback of the Treasury bills or dated securities issued under the MSS. The MSS securities are indistinguishable from normal Treasury bills and government dated securities in the hands of the lender. The payments for interest and discount on MSS securities are not made from the MSS account but are instead shown transparently in the Union budget and other related documents as distinct components under separate subheads. The introduction of MSS has succeeded broadly in restoring the LAF to its intended function of daily liquidity management.

Since its introduction in April 2004, the MSS has served as a useful instrument for medium term monetary and liquidity management. It has been unwound at times of low capital flows and built up when excess capital flows could lead to excess domestic liquidity (Table D5.1; see also Mohan (2006a)).

The impact on commercial banks

The discussion in the previous section makes clear that the links between intervention and money or credit expansion are complex: many other determinants are also changing; there are lags between changes in central bank balance sheets and changes in commercial bank balance sheets; offsetting policies are being pursued; and central banks have a wide range of feasible instruments. The different structures of local financial systems make generalisations about intervention and credit expansion hazardous.

It is nevertheless clear that the sterilisation operations of central banks during this episode of heavy capital inflows have relied much more heavily on the issuance of market paper than on previous occasions. But it is also clear that the banks, not private non-banks, have been the primary counterparty in the expansion of central bank and government securities. Most sterilisation bonds were of short-term maturity (treasury bills, etc) and so highly liquid.

During a period when nominal long-term interest rates are declining (such as during 2000–06 in many EMEs), bonds can be particularly attractive investments. According to a survey from an earlier CGFS working group, commercial banks hold almost one-half of local-currency government paper outstanding in the emerging markets.⁶⁶ Such paper can serve to restrain commercial lending because, in the short run, banks may find the risk-adjusted returns on government bonds more attractive at the margin than expanding credit to the private sector.

How commercial banks respond will, in addition, depend on the capital, reserves, liquidity or other balance sheet constraints they face. Currently such constraints do not seem to bind in most countries. Capital ratios in most EMEs are currently well above regulatory minima.⁶⁷ Reserve requirements do not appear to have been binding in general, as reserve money has grown persistently. Indeed, this lack of constraint has led some central banks to increase reserve requirements, but the effect has often proved to be short-lived, with monetary conditions remaining very easy and intervention continuing at a heavy pace. As for liquidity, very large holdings of government bonds make the asset side of bank balance sheets highly liquid. To the extent that banks with liquid balance sheets and no capital constraints feel better placed to expand less liquid bank loans, the apparent restraining influence of sterilised intervention on monetary growth could prove temporary.⁶⁸

These considerations suggest that the sizable expansion in central bank balance sheets as a result of forex intervention over many years has put upward pressure on the growth of bank credit to the private sector. The pace of such expansion has quickened in the past couple of years (Table D5).

⁶⁶ See CGFS (2007b).

⁶⁷ See, eg, IMF (2008).

⁶⁸ See Kumhof (2004), who provides a formal model for examining the effectiveness of sterilisation using short- term bonds. He shows that increases in short-term bonds with monetary characteristics do lead to increased demand.

The precise quantification of the impact of forex intervention on the growth of bank credit is impossible. Nevertheless, simple cross-country comparisons suggest that these links can be very important. The eventual effects of intervention on domestic credit expansion could be large.

Graph D4 illustrates, for a representative cross-section of countries, the correlation in this decade between growth in net foreign assets and growth in (a) reserve money, (b) M2 (approximately corresponding to growth in bank demand and other deposits), and (c) credit to the private sector. All correlations are positive and highly significant. Over the sample period, a 1% increase in central bank net foreign assets was associated with a 0.5% increase in commercial bank reserve money assets, a 0.4% increase in M2, and a nearly 0.5% increase in credit to the private sector. In contrast, there is no correlation between the increases in net foreign assets and equity prices (see Moreno (2008) for further analysis).

¹ Average annual changes 2001–07, in per cent; horizontal axis: monetary authorities’ net foreign assets. Vertical axis: reserve money, M2, credit to the private sector and equity prices respectively. The dots represent Argentina, Bahrain, Brazil, Chile, China, Colombia, the Czech Republic, Hong Kong SAR, Hungary, India, Indonesia, Iran, Korea, Kuwait, Libya, Malaysia, Mexico, Oman, Peru, the Philippines, Poland, Qatar, Russia, Saudi Arabia, Singapore, South Africa, Taiwan (China), Thailand, Turkey, United Arab Emirates and Venezuela.
Sources: IMF; Datastream; national data.

Impact on the public finances and balance sheets

The increased size of central bank balance sheets associated with intervention has fiscal costs and creates exposures to exchange rate risk, interest rate risk and other risks for the public sector. First, the running costs of holding foreign exchange reserves while maintaining most liabilities in the local currency depend on the differential between the yield on foreign exchange assets and that on local currency liabilities issued to finance forex purchases.⁶⁹ Before 2000, such running costs were large because local interest rates were well above international levels. In the past it was often these rising costs which weakened fiscal positions and led to policy reversals. During much of the 2000s, however, domestic rates fell more rapidly than international rates, and that served to curb running costs. Since mid-2007, sharp declines in US short-term rates have raised the running costs of a significant portion of foreign exchange reserves.

Second, duration mismatches arise (because central bank assets tend to have a longer duration than their liabilities) and create an exposure to a general increase in interest rates. The mismatch has been profitable in recent years: the steady decline in long-term yields in international markets from 2001 to 2007 generated capital gains for central banks with long-term (foreign currency) assets but domestic short-term liabilities.

Third, a currency mismatch arises: an appreciation of the local currency will reduce the local currency value of the foreign exchange reserves, while the value of local currency debt does not change. In addition, movements in cross-rates between major currencies have an impact on the local currency value of foreign exchange reserves, even if the exchange rate between the local currency and the US dollar remains unchanged. This valuation effect may focus public attention on the currency composition of reserves. On the other hand, a rapidly growing EME undergoing financial deepening needs monetary expansion in excess of nominal GDP growth. The monetary base also has to expand correspondingly so that only a part of the foreign exchange reserve accumulation needs to be financed by interest-bearing securities. This can in practice limit sterilisation costs.

Some central banks publicly report their calculations of these sterilisation costs and the accounting gains or losses: Box D6 illustrates the estimates of the Bank of Mexico and explains how these costs and accounting adjustments can result in fluctuation of the Bank of Mexico’s accounting capital.

Most central bank participants in the workshops agreed that the direct economic consequences of sterilisation costs and valuation effects on public sector balance sheets had so far been generally limited. Valuation losses do not reduce the central bank’s ability to sell local currency to limit further appreciation, nor do they reduce the purchasing power of its foreign currency reserves in terms of foreign goods. Nevertheless, in some circumstances large valuation losses may lead to a negative capital position for the central bank, which in turn may compromise its financial independence from the government and weaken its policy credibility. Another aspect is that weakness in (or uncertainty about) the balance sheet of the central bank may undermine the central bank’s credibility in markets, particularly in conditions of stress. Finally, how “losses” are reported may well influence the public debate on the pros and cons of foreign reserve accumulation. Against the various pecuniary and other costs incurred through sterilisation must be weighed the benefits of accumulating foreign exchange reserves. These benefits can include reduced volatility in financial markets and in the exchange rate as well as an increase in overall financial stability.

⁶⁹ This is a simplification. As noted above (and illustrated in Graph D1), the issuance of currency also finances forex reserves.

Box D6

Accounting for the costs of sterilisation

The Bank of Mexico produced the following calculation to illustrate accounting for the costs of sterilisation. First, the wide spread between domestic and international interest rates leads to significant running costs of sterilisation; but as that interest rate gap has narrowed over recent years, running costs have fallen. The table below calculates the cost of sterilisation assuming that the entire stock of sterilisation instruments is used to fund the stock of international reserves (which is, in fact, the case in Mexico). Under such circumstances the annual sterilisation cost is approximately equal to the interest rate spread (domestic-external) multiplied by the stock of sterilisation instruments.

Second, as the stock of international reserves represents a long position in foreign currency (surpassing by about one third of the stock of the monetary base) it involves a substantial exchange rate risk exposure. In the case of an appreciation (in 2005 the Mexican peso appreciated 4.7%), the international reserves stock is worth less in terms of the domestic currency, a decline reflected in the central bank’s accounting as a capital loss.

The impact of these two adjustments on the central bank’s accounting capital is shown in the last column of the table below: eg in 2005 a capital loss of MXP 35 billion came on top of a running cost of MXP 32 billion. These two factors can help explain a large part of the annual change in Bank of Mexico’s accounting capital.

One potential challenge for the public sector as a whole is that a larger and larger stock of sterilisation securities and other forms of interest bearing non-monetary liabilities increases the interest payment liabilities of the public sector. Higher interest rates could then require a significant further issuance of securities just to finance such interest payments. One of the major factors that have facilitated the smooth absorption of sterilisation securities in recent years has been the strong demand for risk-free assets by banks. If that demand were to abate or reverse, sustaining sterilised intervention through large-scale debt issuance could become more difficult and more costly. Several observers (eg Bank of Korea (2005)) have expressed concerns about rising interest costs implied by the growing reliance on monetary stabilisation bonds.