beyond liquidity: new uses for developing asia's foreign exchange reserves

8/22/2019 Beyond Liquidity: New Uses for Developing Asia's Foreign Exchange Reserves

1/50

Economics and REsEaRch dEpaRtmEnt

Bey Lquy: new Ue

fr develg a

Freg Exge Reerve

Donghyun Park

November 2007

RD WoRking PaPER SERiES no. 109


2/50


3/50

ERD Wrkin Paper N. 109

Beyond Liquidity:new usesfor deveLoping AsiAs

foreign exchAnge reserves

donghyun pArk

novemBer 2007

Donghyun Park is Senior Economist, Macroeconomics and Finance Research Division, Economics and Research

Department, Asian Development Bank.


4/50

Asian Development Bank6 ADB Avenue, Mandaluyong City1550 Metro Manila, Philippines

www.adb.org/economics

2007 by Asian Development BankNovember 2007

ISSN 1655-5252

The views expressed in this paper

are those o the author(s) and do notnecessarily refect the views or policies

o the Asian Development Bank.


5/50

FoREWoRD

The ERD Working Paper Series is a orum or ongoing and recently completedresearch and policy studies undertaken in the Asian Development Bank or onits behal. The Series is a quick-disseminating, inormal publication meant tostimulate discussion and elicit eedback. Papers published under this Series

could subsequently be revised or publication as articles in proessional journalsor chapters in books.


6/50


7/50

CoNtENts

Abstract vii

I. IntroductionI. Introduction 1

II. Asias oreign Reserves Buildupacts, uantitative Trends,II. Asias oreign Reserves Buildupacts, uantitative Trends,and Underlying Sources 3

III. Reserve Accumulation Costs and BenetsIII. Reserve Accumulation Costs and Benets 8

I. Spending Reserves on Domestic Projects Macroeconomic Implications 1I. Spending Reserves on Domestic Projects Macroeconomic Implications 17

A. iscal Reserves 2A. iscal Reserves 20 B. Central Bank Reserves 20

. Sovereign Investment Agencies Experiences and Challenges 2. Sovereign Investment Agencies Experiences and Challenges 23

A. Experiences o Existing SWs 2A. Experiences o Existing SWs 23 B. The Development Context Objective and unction 28 C. Design Principles or Good Practice 29

I. Summary and Conclusions 3I. Summary and Conclusions 35

Reerences 37


8/50


9/50

AbstRACt

Developing Asias oreign exchange reserves have grown explosively since2000. Evidence indicates that a substantial part o the regions reserves are nowsurplus to what is required or traditional liquidity purposes. There is consequentlya growing consensus or managing such surplus reserves more actively. The

notion that surplus reserves should be used to pursue higher returns is not onlypolitically popular but economically sound. Nevertheless, it is critical to notethat developing Asias reserves are not ree scal assets. Unlike the reserves

themselves, the income rom investing those reserves does represent a scaldividend or the government. This and other actors suggest that the rst-bestuse o developing Asias surplus reserves is to invest them abroad with the goalo maximizing risk-adjusted returns, subject to the governments broad guidance.

The resulting expansion o scal space will help the regions governments tacklethe huge long-term developmental challenges still acing the region.


10/50


11/50

I. INtRoDuCtIoN

Since the Asian nancial crisis o 19971998, developing Asia has run a sizable, sustained,

current account surplus with the rest o the world, transorming the region into a globally signicantnet exporter o capital. Central banks rather than the private sector have largely intermediatedthese surpluses. Indeed, the private sector has been the recipient o modest capital infows. As a

consequence, regional oreign exchange reserves have seen explosive growth since about 2000.

With no let-up in accumulation, there is a growing conviction that the regions central banksmay now have more than enough reserves to meet conceivable emergencies and contingencies. The

idea that excess reserves or reserves above those required or liquidity purposes should be more

actively invested is gaining appeal. Parking resources in liquid oreign currency assets that oerlittle additional value either in terms o sel-insurance or in terms o lowering the probability outure trouble has signicant opportunity costs. By raising the returns on these allow assets,

the argument runs, uture scal space will be expanded, thereby creating the potential to helpaccelerate development.

In practice, the size o the potential scal dividends rom more active management o oreign

exchange assets will depend on a wide range o actors. Indeed, estimating the size o the pool oreserves that are excess to liquidity requirements is not itsel straightorward. Also, the potentialstream o scal resources will depend on investment perormance, which is susceptible to many

infuences, or example, once the saety o high-grade, short-term investments in reserve currencyassets is let behind, there are no guarantees. But opportunities will also be infuenced by the

choices that developing Asian countries will make with regard to their scal objectives, and thedesign o and operation o their investment vehicles.

External actors will also be important. Commercial investments in oreign jurisdictions mustnecessarily adhere to local policies, regulations, and laws. Though these may curtail opportunities,their overall scope and impact is also likely to depend on the practices and reputation o the investor.

A basic principle is that investors and hosts are more likely to gain, i commercial considerationsare paramount. This is likely to have implications or the legal personality, orm, governance, aswell as investment strategy o any sovereign investment agency.

At a narrower level, the ability to generate scal dividends will depend on the internalcapabilities o the investment agency. actors such as skills and experience o sta, incentives,and the eectiveness o management and reporting and control systems will matter. or new unds

and those that are still on the drawing board, there is an interesting range o experiences, bothinside Asia and internationally, to draw on, but experience underlines the importance o learningby doing.

The central objective o this paper is to explain the context within which developing Asia coulduse its oreign exchange assets in new ways so that these can contribute more eectively to theregions growth, development, and welare. Although so called sovereign wealth unds (SWs) have


12/50

November 2007

beyoNd Liquidity: New usesfor deveLopiNg AsiAs foreigN exchANgereserves

doNghyuN pArk

generated considerable political discussion, the approach here ocuses on the economicopportunities

that they can create. The purpose here is to help the regions policymakers and interested readers tobetter understand the various conceptual and practical issues associated with spending or investingoreign currency assets, with a view to promoting national economic development.

Section II presents stylized acts about the trajectory o developing Asias reserves. DevelopingAsias appetite or oreign exchange reserves mirrors wider trends in the developing world, andis only marginally more accentuated. or most countries, the buildup o reserves refects current

account surpluses and oreign exchange market intervention. or a small number o countries,natural resource wealth and buoyant commodity prices have primed reserves. Though rom time totime capital infows have played their part in reserve accumulation, they have been the dominantsource o reserves only or India. This section explains important concepts that underpin reserves

and links reserves to notions o national wealth and scal resources.

Section III looks at the uses o reserves and possible motives or reserve accumulation.Observing that reserves have costs as well as benets, the section draws on the concept o optimal

reserves to ask whether the regions reserves now exceed those that are reasonably required or

sel-insurance, transactions, and other purposes. Data are presented using a variety o indicators oreserve adequacy, and calculations also use more sophisticated methods that capture the interplay

o multiple actors. These exercises point unambiguously to the presence o reserves that are surplusto those suggested by traditional benchmarks.

Section I observes that excess reserves are not usually a scal resource that can be reely

spent. Even where reserves have accrued through current rather than capital account surpluses,spending them in domestic currency could easily prove problematic rom a scal and a monetaryperspective, and could run counter to macroeconomic policy settings. Especially when reserveshave accrued through the capital account, care must be taken to avoid uses that risk currency

and maturity mismatches. Nevertheless, creative approaches to the use o surplus reserves thatencourage private sector participation in domestic inrastructure projects, say through their use to

underwrite guarantees, could be benecial, provided that complementarity steps are taken to removethe constraints that raise risks, thereby ensuring that the guarantees are not called.

The idea o SWs has attracted a lot o recent attention. In Section , it is pointed out thatthe dividends earned on sovereign investments would constitute a ree scal resource and so would

provide resources that could be directed at development priorities. Drawing on lessons rom existingunds, it is argued that new and nascent agencies or sovereign investment should be operationallyindependent and purely commercial in their orientation. These core principles should not only serveperormance objectives, but should also help allay concerns that have been raised about cross-

border dimensions o their operations. Key issues in the design o sovereign investment agenciesare examined, as are practical issues concerning building capabilities, establishing credibility, andsequencing o strategy.

The paper ends by summarizing key ndings and underlining the potential benets o sovereigninvestment agencies, both in terms o accelerating domestic development and contributing to theeciency and stability o global international capital markets. There are potential gains or the

host as well as investing countries and there are points where their interests intersect. Thoughpotential hosts have a legitimate interest in regulation, such regulation should be motivated byits economic merits and must not be used to camoufage nancial protectionism.


13/50

sectioN ii

AsiAsforeigN reservesbuiLdup: fActs, quANtitAtivetreNds, ANd uNderLyiNgsources

erd workiNgpAper seriesNo. 109

II. AsIAs FoREIgN REsERvEs buIlDuP:FACts, QuANtItAtIvE tRENDs, AND uNDERlyINg souRCEs

In this section, the regions oreign exchange reserve accumulation is examined or the period19902007. The relative importance o current account surplus versus nancial account surplus as

the underlying source o the reserves is also explored. A basic understanding o the key acts aboutaccumulation is essential, and will help provide context or the later analysis o reserve adequacy,and what might be done with surplus reserves.

Throughout the paper, oreign exchange reserves reer solely to oreign currency assets recordedon central banks balance sheets, and exclude gold, special drawing rights, and reserve positions withthe International Monetary und (IM). Box 1 provides a more comprehensive discussion o oreignexchange reserves and the accumulation o reserves.igure 1 shows that between 1990 and 2006,developing Asias total oreign exchange reserves grew rom $203 billion to $2,295 billion in nominalterms ($267 billion to $1,960 billion in real terms). The overall regional picture is one o seculargrowth since 1990, punctuated by a noticeable acceleration since 2000. Interestingly, the impacto Asias nancial crisis on regional oreign exchange reserves is not readily detectable or the crisis

years o 1997 and 1998but its legacy, in subsequent reserve accumulation, is unmistakable.

FIGURE 1NOMINAL AND REAL FOREIGN EXCHANGE RESERVES OF DEVELOPING ASIA (BILLION US$)

1990 1992 1994 1996 1998 2000 2002 2004 2006

Nominal Real

2,500

2,000

1,500

1,000

500

0


14/50

November 2007


doNghyuN pArk

Box 1A primeron foreign exchAnge reservesAnd their AccumuLAtion

oreign exchange reserves have no universally accepted denition, and dierent countries dene

them dierently. or its part, IM (2001) provides the ollowing technical denition

Reserves consist o ocial public sector oreign assets that are readily available to andcontrolled by the monetary authorities. Reserve asset portolios usually have specialcharacteristics that distinguish them rom other oreign currency assets. irst and oremost,

ocial reserve assets normally consist o liquid or easily marketable oreign currency assetsthat are under the eective control o, and readily available to, the reserve management

entity. urthermore, to be liquid and reely useable or settlements o internationaltransactions, they need to be held in the orm o convertible oreign currency claims o the

authorities on nonresidents.

This technical denition thus excludes oreign currency assets that are relatively illiquid and

denominated in nonreserve currencies. It also excludes assets that are held by ocial institutions otherthan the central bank.

In addition to liquidity, saety is another dening characteristic o reserves, and so assets such asequities or bonds with low credit ratings are excluded rom reserves. Although in principle long-maturity

assets can be sae and liquid, in practice central banks preer assets with a short maturity, i.e., 03 years.Central banks portolios thereore typically consist o short-term government bonds or money market

instruments, largely denominated in either US dollars or euros.

The point o departure or understanding how a central bank accumulates oreign exchange reserves

is to look at a basic macroeconomic identity, namely the balance-o-payments identity

Crren Accn baance + Financia Accn baance1Cane in Frein Ecane Reere

The above identity is true by construction. By denition, the sum o the current account balance

and nancial account balance must be equal to the change in oreign exchange reserves. or example, i acountry has a current account surplus o $20 billion and a nancial account decit o $10 billion, then by

denition its reserves must increase by $10 billion. The demand and supply o oreign exchange is equal atthe prevailing exchange rate and i not, the gap is flled by adjustment in reserves. It is conceptually useulor our purposes to rearrange the above identity as ollows

Crren Accn baanceFinancia Accn baance + Cane in Frein Ecane Reere

In terms o the restatement, using the previous example, one hal o the current account surplus o

$20 billion is invested abroad by the private sector and the other hal by the public sector. An increase inoreign exchange reserves is, in eect, a nancial account decit since reserves are purchases o oreigncurrency assets. To an equivalent degree, we can think o the change in oreign exchange reserves as part

o a more broadly dened nancial account balance.

It might be tempting to think that, since the change in reserves is by denition equal to the sumo the current and nancial account balances, the level o reserves is determined mechanically by the

1 The nancial account balance used to be known as the capital account balance.


15/50

sectioN ii

AsiAsforeigN reservesbuiLdup: fActs, quANtitAtivetreNds, ANd uNderLyiNgsources


two balances and thus is beyond the control o a countrys central bank. In act, the contrary is the casethe change in reserves is determined by the central banks conscious and deliberate interventions in the

oreign exchange market. or example, suppose an exporter rom a country earns $1 million in the United

States. I the exporter keeps the dollars in a US bank account, then a $1 million nancial account decitexactly osets a $1 million current account surplus. Alternatively, i the central bank buys the $1 millionrom the exporter in the oreign exchange market, this will extinguish the new claims on nonresidents, and

the nancial account balance will be zero. The upshot is that nancial account balances are dependent onthe central banks oreign current market interventions and attendant changes in reserves.

The degree o central bank intervention in the oreign exchange market and hence the speed oreserve accumulation is largely infuenced by the exchange rate regime. Greater exchange rate fexibilityimplies less intervention and accumulation. Under perectly xed exchange rates, by denition, the

central bank must automatically purchase the countrys excess supply o oreign exchange to prevent thedomestic currency rom appreciating. In the previous example, the excess supply o $10 billion would

be automatically purchased by the central bank to prevent such an appreciation. In contrast, under acompletely fexible exchange rate regime, the central bank will not intervene at all and the excess supplys

eect will be entirely borne by a stronger domestic currency, which should eventually reduce the currentaccount surplus and restore external equilibrium.

The growth o the regions reserves mirrors, to some extent, the growth o regional outputover time. Thereore, to put the regions reserve growth in perspective, it is useul to scale regionalreserves by gross domestic product (GDP). igure 2 shows that the reserves-to-GDP ratio rose rom

12.3% in 1990 to 35.8% in 2006. Over the same period, developing Asias share o global reservesrose rom 23.8% to 44.0% (igure 3). The share o developing Asia in the developing worlds reservesrose rom 49.8% in 1990 to 58.7% in 2006 (igure 4).

FIGURE 2

RATIO OF FOREIGN EXCHANGE RESERVES TO GDP, DEVELOPING ASIA, 19902006

1990 1992 1994 1996 1998 2000 2002 2004 2006

0.45

0.35

0.25

0.15

0.05

Box 1. continued.


16/50

November 2007


doNghyuN pArk

FIGURE 3SHARE OF DEVELOPING ASIA IN WORLD FOREIGN EXCHANGE RESERVES,

19902006 (PERCENT)

1990 1992 1994 1996 1998 2000 2002 2004 2006

0.45

0.40

0.35

0.30

0.25

0.20

FIGURE 4SHARE OF DEVELOPING ASIA IN THE DEVELOPING WORLDS FOREIGN EXCHANGE RESERVES,

19902006 (PERCENT)

1990 1992 1994 1996 1998 2000 2002 2004 2006

65

60

55

50

45

Developing Asias emergence on the global reserve landscape is refected in the act that atthe end o 2006 no ewer than seven o the worlds top 10 reserve-holding economies came rom

the region. Developing Asias 12 largest reserve holders are, in descending order Peoples Republico China (PRC); Taipei,China; Republic o Korea (henceorth Korea); India; Singapore; Hong Kong,China; Malaysia; Thailand; Indonesia; Philippines; Kazakhstan; and iet Nam. Together they accountor almost all the regions reserves and highlight the pan-regional nature o developing Asias reserve

buildup (Table 1).


17/50

sectioN ii

AsiAs foreigN reservesbuiLdup: fActs, quANtitAtivetreNds, ANd uNderLyiNgsources


tABLe 1deveLoping AsiAs foreign exchAnge reserves:

stocksAsof 30 June 2007, And chAngein stocksduring firsthALfof 2007 And 2006

economy stockAsofJune 2007

($ BiLLion)

chAngein first hALfof yeAr

00 00

PRC 1,332.6 266.3 122.2

Taipei,China 266.1 0.1 7.1

Korea 250.2 11.8 15.3

India 206.1 35.9 25.0

Singapore 143.6 7.4 11.5

Hong Kong, China 136.3 3.1 2.3

Malaysia 90.8 9.1 8.6

Thailand 71.1 6.0 5.7

Indonesia 49.2 8.5 5.3Philippines 23.3 3.4 2.3

Kazakhstan 20.9 3.2 5.8

iet Nam 13.4 0.0 2.2

Rest o Developing Asia 30.0 5.5 3.8

Total Developing Asia 2,639.6 360.1 217.1

SourcesInternational Financial Statistics online database (International Monetary und 2007) and Bank o Korea website. Data orTaipei,China were downloaded rom www.cbc.gov.tw/enghome. All data downloaded 3 September 2007.

The PRCs contribution to this buildup is notable. It accounted or only 14.1% o total regionalreserves in 1990, but its share rose sharply to 46.5% by 2006or 49.6% o total regional reserve

growth over the same period. Though it is clear that the PRCs contribution to the level and growtho regional reserves is substantial, others have in act contributed slightly more than hal the buildup,and broadly, contributions to reserves have not been too ar out o line with economic size.

As explained in Box 1, a country accumulates reserves i it has a balance o payment (BOP)

surplus, which can refect a current account surplus, nancial account surplus, or both. Table 2 showsthe average ratio o the current account surplus and nancial account surplus to the increase inreserves during 20002006. The numbers suggest that the regions BOP surplus is driven largely bycurrent account surplus. The current account surplus, in turn, largely refects exports o nonresource

goods and services although there are exceptions such as oil-rich Kazakhstan. Capital infows arealso signicant contributors to the BOP surplus in the PRC and Korea, and in the unique case oIndia, the BOP surplus is driven entirely by capital infows. inally, although the shares o the

current account balance and the nancial account balance should sum up to one, in theory, theyail to do so in practice due to errors and omissions in the data.


18/50

November 2007


doNghyuN pArk

tABLe 2AverAge current AccountsurpLus-to-chAngein reserves (cAs/r) rAtioAndAverAge finAnciAL AccountsurpLus-to-chAngein reserves (fAs/r) rAtioin

deveLoping AsiAs top 10 reserve hoLders, 20002006

economy CAs/R rAtio FAs/R rAtio

PRC 0.66 0.33

Taipei,China 0.88 -

Korea 0.53 0.36

India - 0.93

Singapore 2.41 -

Hong Kong, China 2.77 -

Malaysia 1.87 -

Thailand 0.70 0.15

Indonesia 2.93 -

Philippines 0.69 0.73

Note No value indicates a deicit. or example, India ran a current accountdecit during 20002006 as a whole.

III. REsERvE ACCumulAtIoN: Costs AND bENEFIts

oreign exchange reserves held by the central bank provide a range o possible benets. Most

immediately, liquid oreign currency assets are believed to provide an important source o sel-insurance against the consequences o uture nancial crisis. Indeed, it has been argued that a

suciently large stock o oreign exchange reserves might actually lower the probability o such acrisis occurring (see, or example, Hviding, Nowak, and Ricci 2004; or rankel and Wei 2004). Thissee, or example, Hviding, Nowak, and Ricci 2004; or rankel and Wei 2004). This, and Ricci 2004; or rankel and Wei 2004). Thisand Ricci 2004; or rankel and Wei 2004). This; or rankel and Wei 2004). Thisor rankel and Wei 2004). Thisrankel and Wei 2004). This

motive or developing Asia accumulating reserves has possibly been accentuated by their experiencein the Asian crisis. Having been hurt once, they are wary o being short o liquidity again. It ispossible, too, that an ample supply o oreign exchange reserves can improve a countrys sovereign

credit rating and, in this way, lower its overseas borrowing costs through the sovereign ceiling.Aizenman and Lee (2006 and 2005) provide extended discussions o the reasons or holding reservesand nd strong empirical support or sel-insurance motives.

The second main reason or a central bank to purchase oreign exchange reserves is to lowerthe price o the domestic currency, or at least slow its rate o appreciation. Indeed, some o theregions central banks seem to have been purchasing oreign currency to slow rates o domestic

appreciation, even though nominal exchange rates o East and Southeast Asian countries havebecome more fexible in the postcrisis period (see Box 2). Such currency market intervention isrequently justied by a variety o reasons, but government concerns about the impact o a rapidlyappreciating currency on macroeconomic stability and on the export sector oten loom large. The

analysis in theAsian Development Outlook 2007 Update (ADB 2007b) considers links between the realexchange rate and export perormance. It is empirically dicult to assess the relative importanceo the two main benets o holding reserves.


19/50

sectioN iii

reserveAccumuLAtioN: costsANd beNefits

erd workiNgpAper seriesNo. 109 9

Box 2exchAnge rAte fLexiBiLityin postcrisis eAstAsiA

The postcrisis exchange rate behavior o many East Asian countries seems to imply greater fexibility.Nominal exchange rates became noticeably more volatile in the postcrisis period relative to the precrisisperiod (see Box igure 2.1 and 2.2).

BOX FIGURE 2.1

STANDARD DEVIATION OF MONTHLY PERCENTAGE

CHANGE IN NOMINAL EXCHANGE RATE

BOX FIGURE 2.2

STANDARD DEVIATION OF MEAN ABSOLUTE MONTHLY PERCENTAGE

CHANGE IN NOMINAL EXCHANGE RATE

19 95 9 6 200 1 03 2 00 4 06

PRC INO MAL SIN TAPHKG KOR PHI JAPTHA GER UK

19 95 9 6 200 1 03 2 00 4 06

PRC INO MAL SIN TAPHKG KOR PHI JAPTHA GER UK

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0.0

3.0

2.5

2.0

1.5

1.0

0.5

0.0

Note Germanys currency was the mark beore the European Monetary Union and the euro thereater.

Exchange rate movementsper se may provide only a partial picture o exchange rate fexibility. or

example, the degree o volatility under a stable ree-foat and a heavily managed dirty foat may be similar.A methodology developed by Glick and Wihlborg (1997) and also used by Bayoumi and Eichengreen (1998)

takes into account movements o both nominal exchange rates and oreign exchange reserves to producea more inormative indicator o exchange rate fexibility, the fexibility index.

In Indonesia; Korea; Philippines; Taipei,China; and Thailand, the fexibility index declined noticeablyin the postcrisis period despite relatively high nominal exchange rate volatility (see Box igure 2.3).

In 20052006, the fexibility index o these economies fuctuated around 0.2, compared to more than0.3 during the crisis. In addition, the gap between the index o these currencies and the benchmark G3currencies (dollar, euro, and yen) widened over time, implying more central bank intervention in the

postcrisis period.continued next page.


20/50

10 November 2007


doNghyuN pArk

BOX FIGURE 2.3

GLICK AND WIHLBORGS INDEX OF NOMINAL EXCHANGE RATE FLEXIBILITY

Indonesia

Taipei,China

KoreaThailand

Philippines

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.01996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

Benchmark

0.6

0.5

0.4

0.3

0.2

0.1

0.0

Malaysia

Benchmark

China, Peoples Rep. of

Singapore

Hong Kong, China

The fexibility indices o the PRC and Malaysia (which have had dollar pegs until mid-2005) andHong Kong, China increased slightly in 20052006 but were still lower than those o the other regional

economies. In Singapore, the index remained quite stable over the past decade.

Trends in nominal exchange rate volatility imply that East Asian economies except PRC; Hong Kong,China; and Malaysia have moved toward greater exchange rate fexibility in the post-crisis period. However,trends in the fexibility index indicate that East Asian economies do not foat to the same extent as do

the G3 industrial countries.

Yet reserve accumulation also entails costs such as a combination o higher infation, expanded

scal liabilities, and a higher interest rate. The central banks issuance o domestic currency topurchase oreign exchange lits reserve money, which may percolate up to infation. So, in order tosterilize the potential infationary impact o reserve accumulation, central banks typically attempt

to withdraw domestic liquidity by selling debt (in the orm o bonds) to the nonbank public. But

such sterilization operations entail scal costs when, as is oten the case, the interest rate thecentral bank pays on its outstanding domestic bonds exceeds the yield on its oreign reserve assets.Higher interest rates may also ollowsince they may be required to persuade the nonbank public

to hold a larger stock o (sterilization) bonds. It is only an unusually avorable constellation oactors, such as the benign global infationary environment and ample global liquidity, which hasso ar limited the costs o the recent run-up o reserves in Asia, according to Mohanty and Turner(2006).

Box 2. continued.


21/50

sectioN iii



So what is the central banks optimal reserve level? Clearly, this is neither innite (since

reserves entail costs) nor zero (since they yield benets). Reserves are at their optimal level whenthe marginal benet o a dollar o extra reserves equals its marginal cost, provided that the reservesare in the range where benets exceed costs. Beyond a certain level, the marginal benet o reserves

is likely to diminish as they increase. Beyond some point, the level o reserves may do little toreduce the likelihood o crisis or the capacity to cope with one. Likewise, beyond a certain level,the marginal cost o reserves is likely to rise as they increase, as sterilization operations becomemore expensive, complicating monetary policy.

igure 5 presents a stylized illustration o the optimal reserve level R*, where the marginalbenet o reservesthe benet o an additional dollar o reservesequals their marginal costthecost o an additional dollar o reserves. Beyond R*, which is where many o the regions central

banks are thought to be, accumulating reserves reduces social welare since the cost o a dollarexceeds the benet. I, or example, the actual reserve level was R1, the amount o excess reservesis (R* - R1) and the total cost due to the excess is the triangle C. Reducing excess reserves would

thus theoretically increase social welare.

FIGURE 5

OPTIMAL LEVEL OF FOREIGN EXCHANGE RESERVES

$

Marginal

Cost

C

Marginal

Benefit

R* R1

Reserves


22/50

1 November 2007


doNghyuN pArk

Evidently, i a central bank has accumulated more reserves than is theoretically optimal, one

must ask why, and what policy adjustments are needed to improve welare? The most obvioussolution to the problem o welare-reducing excess reserves is to avoid accumulating them tobegin with, an issue explored in Box 3. But such a response will necessarily entail structural

adjustments and rebalancing o the economy, and as such has greater relevance over the longerrun. More fexible exchange rate policies may also help, but they oer no guarantees o reversingreserve accumulation. or example, Koreas currency has appreciated sharply since early 2002, yetits reserves have continued to grow.

Box 3first-BestsoLutiontothe proBLemof excess reserves

By denition, excess reserves entail social costs (see Box igure 3.1). In a undamental sense, theissue o what to do with them once they have already been accumulated is a study o the second best.The obvious rst-best solution (to the problem o too many reserves) is not to accumulate them in the

rst place!

But slowing or reversing reserve accumulation may not be easy. Even sharp appreciations o some

regional currencies, such as the Korean won, have not slowed the pace o accumulation, as Box igure 3.1shows. There are questions about the extent and the speed with which nominal exchange rate changes

pass through to the real exchange rate; in addition, the impact o real exchange rate changes on short-runexport and import responses has been changing over time (see the chapterExport dynamics in East Asia

inAsian Development Outlook 2007 Update [ADB 2007]). Nevertheless, attempts to resist relative pricechanges over a protracted period will exact costs that may surace in excess reserves or in other ways.

Ultimately, the root cause o excess reserve accumulation lies in the combination o excess netsaving (the dierence between domestic saving and domestic investment) and the inability o capital

markets to intermediate surpluses eciently. Tackling the underlying structural problems rom whichexcess reserves emerge is dicult, and likely to take time. A closer alignment between actual and sociallyoptimal saving would require governments to consider aordable ways o widening access to merit goods,

such as social insurance and pensions. Expanding the availability o credit, particularly or households andsmall enterprises, would also help the private sector better arrange its consumption and saving over time.

On the investment side o the equation, there is a wide variety o possible impediments, and over the yearstheAsian Development Outlookpublished by ADB has repeatedly emphasized the importance o improving

the climate or private sector investment.

In addition to these steps, governments could consider liberalizing opportunities or capital outfows.

Some countries have already moved in this direction and this should help shit intermediation unctionsrom the public to the private sector (IM 2007). In April 2005 or example, Malaysia raised the maximum

amount that institutional investors can invest abroad. In April 2006, the PRC launched the ualiedDomestic Institutional Investor (DII) Program, which enables domestic individuals and companies tohold overseas portolio assets up to a government-determined aggregate quota. The DII Program also

allows commercial banks to sell yuan-denominated nancial products and use the proceeds to buy oreign

currency-denominated xed-income products abroad within the predetermined quota. Subsequently, inMay 2007, the scope o the DII Program was expanded to allow approved DII banks to invest up to 50%o their overseas investment in equities and equity-linked structured products, with a single stock capped

at 5% o a products asset value. Since the securities regulator o Hong Kong, China, is the only regulatorthat has currently entered into a memorandum o understanding with the banking regulator o the PRC,banks are currently only permitted to invest in stocks listed in Hong Kong, China.


23/50

sectioN iii



Box 3. continued.

Korea, too, has taken steps to ease capital outfows. In January 2007, the government announcedtax breaks and other incentives to acilitate overseas portolio investments by domestic institutional

investors and banks, and raised the ceiling on speculative overseas real estate investment rom $1 millionto $3 million. The authorities have also used a variety o promotional measures to acilitate outward DI,

such as new insurance schemes to help hedge DI-related risks and the expansion o the Export-ImportBank o Koreas overseas investment support capacities.

BOX FIGURE 3.1

TRENDS IN KOREAS EXCHANGE RATE

Billion

US$

300

260

220

180

140

100

1,350

1,260

1,170

1,080

990

900

Won/US$

Foreign reserves, left scale Exchange rate, r ight scale

Jan-02 Jan-03 Jan-04 Jan-05 Jan-06 Jan-07

Though there is broad theoretical agreement that reserves entail costs as well as benets, a

wide range o views exist about the magnitude o actual costs and benets. No consensus exists onhow to calculate the optimal reserve level. Moreover, there is no obvious reason why the benets

and costs o reserves should be similar across countries or remain constant over time. Countries thatpeg their exchange rates are obviously going to need more reserves than those whose currency isallowed to move more reely. The need or reserves is also likely to be infuenced by the degree ocapital account openness, structure o external liabilities, health o the domestic nancial system,susceptibility to contagion, and other actors. Given a large element o uncertainty, including the

possibility o unknowable shocks with unknowable consequences, erring on the upside o optimalreserve calculations is understandable.

The case or heightened caution is especially strong or countries whose reserves are mainly

built on capital infows, especially i those infows are primarily short-term portolio investmentsand debt rather than long-term DI. The only large regional country that alls into this category isIndia. Other things equal, Indias optimal reserve level is likely to be higher because its reserves

largely refect a nancial account surplus rather than a current account surplus.

The experience o Thailand prior to the Asian nancial crisis oers potentially very valuablelessons. Thailand had the worlds 10th largest reserves in 1995 and the 11th largest the ollowing

year. Yet even these apparently robust levels ailed to prevent the economy rom alling prey tocrisis in 1997. The lesson is that short-term capital infows are vulnerable to disruptive reversals,increasing the potential benets o precautionary reserves. Comprehensive measures o gross external


24/50

1 November 2007


doNghyuN pArk

liability positions are perhaps most relevant. A net-creditor country may still be vulnerable to a

nancial shock i, or example, nonresidents hold large amounts o shares in the domestic stockmarket.

The idea that reserve assets should be invested in higher-yielding assets conuses the true

need or reserves with the belie that some reserves are surplus to this need. As the primary purposeo reserves is to provide international liquidityduring a crisis, they can only perorm this unctioni they are invested in sae and liquidassets. The typically low rate o return on such assets is

unortunate, but beside the point. Thereore, conceptually, it may be helpul to view surplus reservesnot as reserves per se but as an entirely dierent kind o public sector oreign currency asset thatcan be used or purposes other than sel-insurance. Since these assets are publicly owned, oneobjective might be to maximize the scal dividend that they yield. Yet i the government ails to

make a clear-cut distinction between reserves (properly dened) and surplus oreign currency assetsthat can be put to alternative uses, reserve management may be subject to extraneous (perhapspolitical) infuences. The result could be a ailure o reserve management to deliver liquidity when

needed and to use surplus oreign currency assets in the national interest.

To gauge the magnitude o developing Asias surplus oreign currency assets, we now turnto some well-known measures o reserve adequacy. Comprehensive discussions o these measures

include Edison (2003) and ECB (2006). Although these measures are based on general economicintuition rather than rigorously derived theoretical concepts, empirical studies nd them to behelpul guides or policymakers. In particular, many such studies nd one such rule o thumbtheratio o reserves to short-term external debtto be a signicant determinant o an economys

vulnerability to nancial crisis. More precisely, according to the well-known Greenspan-Guidottirule, the critical value o this ratio is one. The idea here is that a country with reserves equal to ormore than all external debt alling within one year should be able to service its immediate oreignexchange obligations even during a crisis. igure 6 reveals that developing Asia comortably passes

the Greenspan-Guidotti test o adequacy.

FIGURE 6RATIO OF FOREIGN EXCHANGE RESERVES TO SHORT-TERM EXTERNAL DEBT IN

DEVELOPING ASIAS TOP 10 RESERVE HOLDERS, 19902006

1990 1992 1994 1996 1998 2000 2002 2004 2006

20

16

12

8

4

0


Indonesia

Philippines

Thailand

Singapore

Hong Kong, China

Korea, Rep. of

India

Malaysia

Taipei,China


25/50

sectioN iii



Two other well-known reserve adequacy measures are the reserves-to-M2 ratio and the monthso imports that reserves can pay or. The basic intuition is that the higher the M2 ratio, the greaterthe degree to which the risks o crisis-provoking domestic capital fight are covered. The suggested

critical values range rom 5% to 20%. igure 7 shows that the reserves/M2 ratio are either above or

close to the upper limit o the 5%20% range or the major reserve holders o developing Asia. Thebasic idea behind the import cover measure is that a large stock o reserves will reduce vulnerabilityto adverse current account shocks. The suggested critical value is 3 to 4 months. igure 8 shows

that the number o months that imports can cover is well above our in developing Asia.

FIGURE 7

RATIO OF FOREIGN EXCHANGE RESERVES TO M2 IN

DEVELOPING ASIAS TOP 10 RESERVE HOLDERS, 19902006

1990 1992 1994 1996 1998 2000 2002 2004 2006

1.0

0.8

0.6

0.4

0.2

0

Thailand


Indonesia

Philippines

Taipei,China

Malaysia

India

Korea, Rep. of

Hong Kong,China

Singapore

FIGURE 8IMPORTS COVERED BY FOREIGN EXCHANGE RESERVES (NUMBER OF MONTHS)

IN DEVELOPING ASIAS TOP 10 RESERVE HOLDERS, 19902006

1990 1992 1994 1996 1998 2000 2002 2004 2006

20

16

12

8

4

0

Thailand

Philippines

Hong Kong,China

Singapore

India

Indonesia

Korea, Rep. of

Malaysia

Taipei,China



26/50

1 November 2007


doNghyuN pArk

Let us now examine two additional measures o reserve adequacy that are less widely used

than the three measures discussed above reserves-to-GDP ratio and reserves-to-total external debtratio. igure 9 shows rising reserves/GDP ratios throughout the region. igure 10 shows that thereserves-to-total debt ratio is currently substantially above 1 or close to 1, which implies that

reserves are sucient to cover not only short-term external debt but all external debt in PRC;Korea; Taipei,China; India; Singapore; Hong Kong, China; Malaysia; and Thailand. Thereore, bothreserves/GDP and reserves/total external debt ratios provide urther evidence o reserves that aresurplus to needs.

FIGURE 9

RATIO OF FOREIGN EXCHANGE RESERVES TO GDP


1990 1992 1994 1996 1998 2000 2002 2004 2006

1.2

0.9

0.6

0.3

0


IndonesiaPhilippines

Thailand

Hong Kong, China

Korea, Rep. ofSingapore

India

Malaysia

Taipei,China

FIGURE 10

RATIO OF FOREIGN EXCHANGE RESERVES TO TOTAL EXTERNAL DEBT


1990 1992 1994 1996 1998 2000 2002 2004 2006

6

5

4

3

2

1

0


Indonesia PhilippinesThailand

Hong Kong, China

Korea, Rep. of

Singapore

India

Malaysia

Taipei,China


27/50

sectioN iv

speNdiNg reservesoN domesticprojects: mAcroecoNomic impLicAtioNs


Policymakers oten rely on simple rules o thumb, which give reasonably good policy guidance,

to assess reserve adequacy. Nevertheless, it is worthwhile to take a look at the limited but growingtheoretical and empirical literature consisting o more rigorous and systematic studies o reserveadequacy. Pertinent studies include Edison (2003), Wyplosz (2007), Jeanne and Ranciere (2006),

Gosselin and Parent (2005), Mendoza (2004), Aizenman, Lee and Rhee (2004), Aizenman andMarion (2004 and 2002), and Dooley et al. (2004). The overall balance o evidence in this emergingliterature conrms the story told by the rules o thumbthat the regions current reserve build-up has overshot what might be deemed optimal, especially since 2000although the studies

dier considerably about the extent o the overshooting. Another signicant general nding is anapparent structural increase in optimal reserves in the post-1997 period, which is consistent witha stronger precautionary demand or reserves ater the Asian crisis.

Iv. sPENDINg REsERvEs oN DomEstIC PRojECts: mACRoECoNomIC ImPlICAtIoNs

At the broadest level, ollowing Hildenbrand (2007), the regions oreign exchange reserve

accumulation can be split into two types (i) accumulation based on government budget surpluses,

prots o state-owned companies, or other government net income; and (ii) accumulation basedon oreign exchange market interventions by central banks within the context o current account

surplus and/or capital infows. The classical example o the rst type is associated with exportrevenues rom natural resources such as oil. Governments typically either own the natural resourcesor heavily tax their private sector owners. or example, the post-2002 commodity boom has provideda scal bonanza or the governments o resource-rich countries around the world. Oten export

revenues are kept in separate unds and thus excluded rom ocial reserve statistics. The secondtype refects the central banks purchases o oreign exchange. These reserves, which originate romcurrent account surpluses or capital infows, become part o the central banks stock o oreignexchange reserves.

or conceptual clarity and simplication, the rst type o reserves can be considered scal

reserves and the second type central bank reserves. rom the viewpoint o the balance sheeto the sovereign (i.e., the consolidated public sector = government plus central bank), there isa critical conceptual dierence between scal reserves and central bank reserves. iscal reservesprovide additional scal resources or the government and can be spent without incurring debt.Central bank reserves, on the other hand, which are nanced by issuing bonds or currency, do not

constitute ree scal assets as they have counterpart liabilities, i.e., currency or debt (bonds).It ollows that i the government wishes to spend such reserves, it must borrow to cover its newliabilities. or the most part, developing Asias reserve build-up refects central bank reserves; thesedo not represent ree scal resources at governments disposal.

Regardless o whether the reserve build-up refects accrual o scal assets or the centralbanks oreign exchange market intervention or both, it usually takes place in the context o a

balance o payments or external surplus. In this connection, it is conceptually useul to distinguishamong the three main types o external surplus (i) resource-based current account surplus basedon natural resource export revenues; (ii) nonresource current account surplus based on exportso manuactured goods and services; and (iii) nancial account surplus, i.e., capital infows rom

abroad. or example, the sharp increase in oil price since 2002 has produced a sizable externalsurplus or oil exporters such as Kazakhstan (Type 1). On the other hand, Koreas external surplusis driven largely by current account surplus (Type 2) while Indias external surplus is driven by


28/50

1 November 2007


doNghyuN pArk

capital infows (Type 3). or the region as a whole, the external surplus is predominantly Type 2,

in some cases augmented by Type 3, rather than Type 1. In general, scal reserves originate romresource-based current account surplus whereas central bank reserves originate rom nonresourcecurrent account surplus, nancial account surplus, or both. Singapore is a particularly interesting

case since the government has accumulated large amounts o oreign assets even though it doesnot have any natural resources.

Consider the implications o developing Asias reserve build-up or national wealth as

opposed to the sovereign balance sheet. The surge in oil prices since 2002 has clearly improvedthe terms o trade or oil-exporting countries and added to their wealth. It is equally true is thata nonresource current account surplus also adds to a countrys net wealth. Box 4 elaborates thispoint. This type o surplus mirrors an excess o savings over domestic investment and is refected

in the accumulation o nancial claims on nonresidents. Essentially, the stock o such claims cannance uture consumption. inally, the impact o nancial account surplus on national wealth islikely to be more limited. To the extent that oreign debt and equity infows enhance a countrys

productivity capacity, they allow or higher output and consumption in the uture. However, suchinfows constitute oreign residents uture claims on a share o domestic output, which will reduceuture consumption by domestic residents.

Box 4Are deveLoping AsiAs foreign exchAnge reserves free fiscAL resources?

or the most part, developing Asias oreign exchange reserves are the consequences o oreign

exchange purchases by central banks, not scal surplus, prots o state-owned companies, or othergovernment income. A hypothetical example will clariy that such reserves do not constitute a ree scal

resource. Suppose that Samsung, a large private-sector Korean conglomerate, exports $50 billion andimports $30 billion. The rm has earned more than it spent, so it is in eect saving and thus adding $20

billion to its net wealth. Koreas national net wealth has unambiguously increased. In terms o Samsungsbalance sheet, the $20 billion is a oreign currency asset, as ollows

Samsung

Assets Liabilities

US dollars 20

Now suppose that instead o investing the $20 billion abroad in assets such as a US dollar deposit

account, Samsung decides to bring its US dollars home and exchange them or Korean won in the Koreanoreign exchange market. Let us suppose urther that the Bank o Korea, or whatever reason, decides to

purchase the $20 billion dollars with Korean won it issues. The increase in Samsungs net worth is still

worth $20 billion, except that the net wealth is now held in won. As or the Bank o Korea, the increase inthe value o its assetsthe $20 billion dollarsis exactly oset by the increase in its liabilities, or $20billion dollars worth o liabilities, as ollows


29/50

sectioN iv

speNdiNg reservesoN domesticprojects: mAcroecoNomic impLicAtioNs


Bank o Korea

Assets Liabilities

US dollars 20 Korean won 20

or Korea as a whole, there is an increase in the net worth o Samsung but not in the net wealtho the Bank o Korea. Since the $20 billion worth o Korean won increases the monetary base, a centralbank concerned about infation might attempt try tosterilize this additional liquidity by selling bonds to

the nonbank public. or the sake o simplicity, let us assume that Samsung buys these sterilization bonds.Then Bank o Koreas and Samsungs balance sheets change as below. It still remains the case that private

sector and national net wealth has increased, although there are no additional scal resources or thesovereign.

Bank o Korea

Assets Liabilities

US dollars 20 Sterilization bonds 20

Samsung

Assets Liabilities

Sterilization bonds 20

A central conclusion is that developing Asias reserves do not, in general, create ree scal resources at

the disposal o government. The exception is where government itsel earns oreign exchange. In particular,or oil-rich countries such as Azerbaijan, Kazakhstan, and Timor-Leste, the increase in reserves clearly addsto scal resources and national wealth. Singapore is somewhat o a special case, in which a commercially

active government earns signicant amounts o oreign exchange in a nonresource economy.

Asias mountain o oreign currency reserves has sparked an important debate about how theymight be put to better use. Why cant excess oreign currency assets, which do not serve liquidityneeds, be spent on vital domestic projects? Presumably, the social returns on many such projectsare much larger than the rewards o investing in an internationally diversied nancial portolio,

even when nancial markets are booming.

In evaluating the macroeconomic implications o spending reserves, clarity is needed on threebasic issues (i) Are the reserves a ree scal resource? (ii) I they are not, how is spending to

be unded? (iii) Does spending entail oreign currency intervention?

The source o reserves determines whether they are a ree resource, or not. In addition toscal reserves and central bank reservesthe two main sources o reserve accumulation already

discussedreserves can accrue through public sector oreign borrowing and grant ocial developmentassistance (ODA). In general, scal reserves and grant ODA can be considered as scal resources

Box 4. continued.


30/50

0 November 2007


doNghyuN pArk

since the government can spend them without incurring debt. In this sense they are ree. By

contrast, central bank reserves and oreign borrowing are not ree since they have counterpartliabilities.

A. Fica ReereConsider rst circumstances where the oreign currency reserves o the central bank are scal

assets. In this case, spending or transers will entail debiting the government account at thecentral bank, with the central bank providing the needed liquidity. oreign currency spending onimports or domestic projects would have no monetary or scal implications. The drawdown o thegovernments account would be oset by a reduction in the oreign exchange assets on the central

banks balance sheet. But i government receives and spends domestic currency, the central bankmust rst sell oreign exchange or domestic currency. This leads to a withdrawal o reserve money,and an appreciation o the domestic currency. But as the government spends and receipts aredeposited by the private sector in the commercial banking system, reserve money is replenished.

Complications occur i the central bank wants to avoid an appreciation. It would thenrepurchase oreign currency in the market, expanding reserve money. Let unsterilized, the monetaryexpansion would raise the domestic price level and real exchange rate appreciation would occur

through infation. I oreign exchange market intervention is accompanied by sterilizationthesale o domestic currency securities to the private sectordomestic interest rates rise. iscal costsoccur not because o government spendingper se, but because the monetary authorities choose to

maintain the pre-existing exchange rate parity. In intermediate cases, drawing on oreign currencyreserves to und domestic spending could entail some combination o a nominal appreciation,rising domestic prices, and higher domestic interest rates. On the real side o the economy, netexports and/or interest sensitive components o domestic expenditure would adjust depending on

real exchange rate and interest rate impacts.

b. Cenra bank Reere

Now imagine that government spending requires borrowing reserves rom the central bank.Assume too that the exchange rate is pegged and there is ull sterilization o oreign currency

market interventions. urther assume that government spending takes the orm o oreign currencyspending on imports or domestic projects.

In the rst leg o the transaction, the reduction in oreign exchange reserves is oset by an

expansion o government securities held by the central bank. There is no impact on the balancesheet o the commercial banking system, as government purchases do not create private sectordeposits. Moreover, as this operation entails no oreign exchange market purchases or sales, theexchange rate parity is unaltered. Eventually, however, the government would have to service its

debt with the central bank.


31/50

sectioN iv

speNdiNg reservesoN domestic projects: mAcroecoNomic impLicAtioNs


Alternatively, government could sell debt directly to the nonbank public, using the proceeds

to purchase oreign currency rom the central bank. Now reserve money is withdrawn and interestrates rise, but there are no oreign exchange market impacts. Through its defationary impact, thecontraction o reserve money induces a depreciation o the real exchange rate and capital infows

are encouraged by rising domestic interest rates. Any induced balance o payments surplus will raisethe demand or domestic currency requiring oreign currency purchases to maintain the peg, henceexpanding reserve money. This raises the price level, counteracting the earlier defation. Governmentborrowing and spending o oreign currency has increased domestic currency debt and has required

that the oreign currency be repurchased to check a nominal appreciation. Changes in interest ratesand in the real exchange are elt on domestic expenditure and net exports.

It would make little sense or the government to nance domestic currency spending, rather

than oreign currency purchase o imports, by borrowing reserves rom the central bank. or example,inrastructure projects typically have high local content. In this case, the government must purchasedomestic currency using the oreign currency reserves it borrowed. Though this operation has no

direct impact on base money since the government spends the purchased domestic currency, it exertspressure or an appreciation o the nominal exchange rate, with the attendant need or interventionand sterilization. I a government wants to spend on domestic inrastructure, it would be muchsimpler and more transparent to do so through orthodox domestic-debt-unded scal operations.

In the nal analysis, government spending o scal reserves is comparatively straightorward.However, central bank reserves are not a ree asset and government attempts to spend them mayimply policy reversals and responses that cause reserves to fow back onto the balance sheet o

the central bank.

Box 5 looks at the specic case o spending oreign exchange reserves or domestic purposes,i.e., whether India should use its soaring reserves to nance its large and growing need or

inrastructure. Given the complications o using oreign currency assets to und domestic currencyspending, the Indian government has rightully exercised a great deal o caution in exploring this

issue. Box 5 describes in detail the various potential pitalls associated with using reserves toimprove Indias inrastructure.

Nevertheless, the Indian experience also suggests that there is some scope or innovation andcreativity in using reserves productively or high-priority scal needs. In particular, a more creative

use o Indian reserves or inrastructure might be to deploy them to mitigate risks on private sectorinvestments. But or this to work, the obstacles that block potential private sector participation indomestic inrastructure would ultimately have to be removed. I this could be achieved, contingentliabilities in the orm o guarantees would not result in debt and could boost national productivity.

urthermore, such guarantees can help to bring in oreign inrastructure companies with superiortechnology or know-how, resulting in, or example, better roads at lower costs. This will not onlydirectly benet Indian inrastructure but also contribute to learning-by-doing on the part o Indian

inrastructure companies.


32/50

November 2007


doNghyuN pArk

Box 5using indiAs excess reservesfor infrAstructure?

The Government o India has proposed a mechanism or using oreign exchange reserves or

inrastructure investment. The proposal was announced in the ederal Budget Speech in ebruary 2007,and the Reserve Bank o India (RBI) and the Ministry o inance are currently working on implementingthe proposal. One o the recommendations o a high-level government committee on inrastructure

nancing is to use a small part o the oreign exchange reserves without the risk o monetary expansion.Thus, it is proposed in the budget to establish two wholly owned subsidiaries o the India Inrastructure

inance Corporation (IIC) with the ollowing objectives (i) to borrow unds rom the RBI and lend toIndian companies implementing inrastructure projects in India, or to co-nance their external commercial

borrowings (ECBs) or such projects, solely or capital expenditures outside India; and (2) to borrow undsrom the RBI and invest such unds in highly rated collateral securities, so as to enhance the credit ratingso Indian companies that raise unds in international markets or inrastructure projects in India. The loans

by RBI to these two subsidiary companies will be guaranteed by the Government o India, and the RBI willbe guaranteed a rate o return higher than the average rate o return on the reserves.

The specic proposals outlined is part o a broader line o thinking within the Indian government,which is seriously considering inrastructure nancing as a major avenue or using RBIs soaring reserves

more actively. The government has orecast that the country needs $320 billion in inrastructure investmentduring the next ve years as the countrys rapid economic growth tests the limits o its long-neglected

ports, roads, airports, and power utilities. The combination o woeully inadequate inrastructure andstrong growth prospects means that the social rate o return on inrastructure investments is likely to be

high. This is all the more so since inadequate inrastructure is widely seen as one o the biggest obstaclesto Indias sustained economic growth. Thereore, in principle, it seems that India would do better to investpart o its reserves on high-return domestic highways than on low-return US government bonds. This is

especially true i the inrastructure investments are limited to spending on imports, since this entails lessinfationary risk than domestic spending.

Though attractive in principle, a number o considerations suggest that the Indian government should

approach the issue o using reserves to nance inrastructure with a great deal o caution. ortunately,as the limited scope and specic targeting o the two specic proposals shows, the government is doingprecisely that so ar. irst, Indias reserves are a result o the central banks oreign exchange market

interventions and, as such, are not sovereign wealth, and hence are not scal resources. Second, underlyingmuch o Indias reserve build-up are potentially volatile short-term portolio capital infows, which callsor setting aside more reserves. Third, inrastructure projects in India yield low or negative returns due to

political and economic risks associated with adjusting the tari structure, introducing labor reorms, andupgrading technology. A more direct and eective policy approach to nancing inrastructure is to create

a more avorable business environment or private sector investment. ourth, using reserves to nanceinrastructure will soten the governments budget constraint and weaken scal discipline, a major risk

in light o the governments unhealthy overall scal position. ith, much o inrastructure investment islikely to revolve around domestic spending, so it makes more sense or the government to issue rupee-denominated inrastructure bonds directly. Not only is the alternative o issuing bonds to purchase reserves

to buy rupees roundabout, it increases the monetary base and thus entails risk o infation.


33/50

sectioN v

sovereigN iNvestmeNt AgeNcies: experieNcesANd chALLeNges


v. sovEREIgN INvEstmENt AgENCIEs: ExPERIENCEs AND ChAllENgEs

As was seen in the previous section, spending on domestic projects represents one possibleuse or developing Asias large and growing surplus reserves. Another policy option is to use thesurplus reserves to make overseas investments with the aim o maximizing risk-adjusted returns. In

this context, sovereign investment agencies, better known as sovereign wealth unds, are attractingthe interest o policymakers throughout the region.

In this section, the experiences o existing SWs, which can provide valuable lessons or the

new Asian SWs, are rst surveyed. The development context that denes the broader objectivesand unctions o the Asian unds is then explored. inally, based on the experiences o the existingSWs and the Asian development context, a number o specic design principles that may help

unds maximize scal dividends are set orth.

A. Eperience Eiin sWF

The past experiences o existing SWs can lead to a better understanding o thecan lead to a better understanding o thethe

challenges that lie ahead or the new Asian SWs. Although the term SW was coined onlythat lie ahead or the new Asian SWs. Although the term SW was coined onlyahead or the new Asian SWs. Although the term SW was coined onlyor the new Asian SWs. Although the term SW was coined onlyrecently (Rozanov 2005a and 2005b), SWs have been around or quite some time. In act,the oldest und, the Kuwait Investment Authority, goes all the way back to 1953.While SWsare hardly new, they have come into the spotlight recently due to their sheer size. There isno single authoritative denition o an SW. However, regardless o their dierences, SWsshare two dening characteristics namely (i) ownership and control by the government,and (ii) pursuit o risk-adjusted returns as the central objective.

According to Jen (2007), SWs now have as much as $2.5 trillion in assets, a numberthat could possibly double by 2012. To put the numbers into better perspective, as omid-2007, Jen (2007) estimates that the worlds ocial oreign exchange reserves stand

at around $5.1 trillion; hedge unds are considered to control between $1.5 trillion and$2 trillion in assets; and total global nancial assets are roughly around $140 trillion.I current trends persist, the assets at the disposal o SWs may expand rapidly, both inabsolute terms and relative to the size o global oreign exchange reserves.

Table 3 lists the worlds biggest and most well-established SWs. In terms o theirunderlying resource base, SWs can be broadly classied into two types, commodity andnoncommodity. Commodity unds have been established rom export revenues o key naturalresources such as oil and gas. As such, SWs are sometimes called oil unds or naturalresource unds because a majority o them were created rom revenues rom oil, gas, copper,phosphates, and other minerals. In this case, countries are converting real assets buried

in the ground (e.g., oil) into nancial assets. Majority o the worlds most well-establishedsovereign unds are commodity unds, including the two biggest, Norways GovernmentPension und (see Box 6) and the Abu Dhabi Investment Authority.


34/50

November 2007


doNghyuN pArk

tABLe 3sovereign weALth fundsinthe worLd

economy nAmeof fund Assets

(us$ BiLLion)

yeArof

inception

type

United Arab Emirates Abu Dhabi Investment Authority 875 1976 Commodity Oil

Singapore Government o Singapore

Investment Corporation

330 1981 Noncommodity

Norway Government Pension und 300 1990 Commodity Oil

Saudi Arabia arious types 300 n/a Commodity Oil

PRC China Investment Corporation 300 2007 Noncommodity

Kuwait Kuwait Investment Authority 160205 1953 Commodity Oil

Singapore Temasek Holdings 100 1974 Noncommodity

Hong Kong, China Investment Portolio (HongKong Monetary Authority)

100 1998 Noncommodity

Australia uture und 50 2004 Noncommodity

atar atar Investment Authority 40 n/a Commodity Oil Alaska Permanent Reserve und 35 1976 Commodity Oil

Russia Oil Stabilization und 32 2003 Commodity Oil

Note Due to lack o ocial inormation rom the unds themselves, asset sizes are largely estimates rom unocial sources such asMorgan Stanley (Jen 2007).

Sources Jen (2007), Rozanov (2005), Setser and Ziemba (2007), Temasek Holdings (2007), Rietveld and Pringle (2007), United StatesTreasury (2007), and Government o Singapore Investment Corporation website (http//www.gic.com.sg).

Box 6norwAys government pension fund

In striking contrast to other SWs, Norways oil-based SW, the Government Pension und (GP)Global, is characterized by a high degree o transparency and disclosure o inormation. The Ministry oinance, which owns the unds, reports to the parliament on all important issues related to the und, such

as changes in investment strategy, perormance, risks, and costs. The central bank, which manages theund, publishes quarterly reports on the unds management, along with an annual report that lists all o

the unds investments. Those reports are comparable in terms o inormation disclosure to the reportso publicly listed private sector companies. Due to its high level o transparency and good investmentperormance, GP und has become a model or new resource-based SWs in developing Asia and beyond.

or example, Kazakhstans National Oil und has explicitly modeled itsel ater the GP.

Norway began producing oil in the North Sea in 1970 and is currently the worlds third largest

exporter o oil. Such large-scale exports o oil have continuously generated substantial scal revenues orthe Norwegian government. In 1996, the Norwegian government created the GP, which has since been

renamed the Government Pension und Global in 2006. The und had assets o around $285 billion at theend o 2006. The renaming refects the unds role in mobilizing government savings to meet the rapid

uture growth o public pension expenditures. However, the und is not an earmarked pension und. Rather,the unds basic economic unction is essentially that o an endowment und, and its central mandate is

to transorm Norways oil wealth into a diversied portolio encompassing oreign equities and bonds. Thetwo main expected benets rom such transormation is the reduction o total risk through diversicationas well as higher returns.


35/50

sectioN v

sovereigNiNvestmeNt AgeNcies: experieNcesANd chALLeNges


The und is included in the government budget and money is not transerred to the und until thebudget is balanced. This means that the government cannot build debt at the same time as it accumulatesassets in the oil und. The use o petroleum is separated rom the ongoing accumulation o income and

serves as a buer or government nances. The buer unction is related to the act that petroleum incomeaccounts or a substantial portion o government revenues. Those revenues are uncertain due to the wide

variation in oil prices.

At the end o 2006, xed income instruments and equities accounted or 60% and 40% o the unds

strategic benchmark portolio, respectively. The xed instruments subportolio consists o instrumentsrom Europe (60%), America/Arica (35%), and Asia/Oceania (5%), and its benchmark portolio is based

on Lehman Brothers global indices or government and corporate bonds. The equity subportolio consistso instruments rom Europe (50%), America/Arica (35%), and Asia/Oceania (15%), and its benchmark

portolio is based on the TSE All-World indices. The unds investment perormance relative to the strategicbenchmark portolio o equities and bonds rom three broad geographic regions has been satisactory, asevidenced by an average annual excess return o 0.48 percentage point relative to the benchmark during

19982006. The accumulated excess return is NOK 28.9 billion or $4.7 billion. At a broader level, GP has

eectively saeguarded national wealth or uture generations and shielded government nances as well asthe economy at large rom the volatility o oil prices. Another signicant contribution o the GP has beento prevent Norway rom being inected with Dutch disease, or the phenomenon o resource-rich countries

suering loss o competitiveness due to currency appreciation, as evident in the act that the country hasa airly well-diversied economy.

uite clearly, the Norwegian SW experience can provide valuable lessons or resource-rich countriesin developing Asia. However, some dening characteristics o the Norwegian model need to be discussedbeore the models applicability to other countries can be meaningully assessed. irst, all o the unds

capital is invested outside Norway and in oreign currencies. The underlying motivation here is to preventinfationary pressures and currency appreciation that would result i a large proportion o the oil revenues

were spent domestically. Second, a new business unit was created within the central bank, the NorgesBank Investment Management (NBIM), which was independent rom the rest o the central bank. or

example, the compensation structure is dierent and much more perormance-linked at NBIM. The nanceministry acts as the owner but only provides broad strategic guidance to NBIM, which actively managesthe assets. Third, the und is explicitly and transparently included in the government scal balance.

Although in theory the und serves as a buer or government nances, in practice annual transers romthe und to the government budget has been limited to a maximum o 4% o the unds market value.

ourth, and this is related to the third point, there is a strong underlying philosophy that oil revenuesshould benet all generations o Norwegians. The government has made lots o eorts to explain the

endowment nature o the oil wealth and rmly resisted populist calls or spending more rom the und.ith, as already noted, GP-G has been unique among SWs in terms o its exceptionally high degree o

transparency and inormation disclosure. The nance ministry and central bank jointly provide virtually allrelevant und-related inormation to the general public.

However, not all SWs were created out o natural resource wealth. or noncommodity SWs,unding has usually come rom one o the ollowing sources (i) transer o oreign exchangereserves, (ii) government budget surplus and other government net income, and (iii) oreign aid inthe orm o trust unds. Among noncommodity economies, Singapore is unique in its long history

and abundant experience with SWs. The countrys two SWs, Temasek Holdings and Governmento Singapore Investment Corporation (GIC), jointly manage assets o about $430 billion, and were

Box 6. continued.


36/50

November 2007


doNghyuN pArk

created as vehicles to deploy scal surpluses (broadly dened) rather than natural resource income

(see Box 7). Asias newer SWs (see Table 4) are also largely noncommodity unds but nancedthrough the central banks oreign exchange reserves.

Box 7temAsek hoLdingsAnd governmentof singApore investment corporAtion

Temasek is the older o the two Singaporean SWs and is essentially the governments investment

holding company. Temasek, which currently has assets o between $80100 billion, began its operationsin 1974. As such, it is one o the oldest nonresource SWs in the world. The Ministry o inance hadbegun acquiring ownership interests in strategic companies throughout the economy, rom banking to

telecommunications to transportation, since Singapores early nation-building years, and those investmentsreached substantial levels by the early 1970s. Temasek thus has its origins in the governments strategic

decision to run its investments on a sound commercial basis and prevent any confict with the governmentspublic interest role o policy making and market regulations. At a deeper level, Temasek refects Singapores

unique brand o state-led capitalism in which the government is extensively involved in the production o

goods and services. The major reason Temasek has attracted so much attention rom the regions emergingSWs is its exceptional investment perormance. According to Temasek (2006), its market value grew on

average by a remarkable 18% per year on a compounded basis between 1974 and 2006. During the sameperiod, the market value o Temaseks portolio rose rom $170 million to $64 billion. Although there is

some debate about the accuracy o the numbers, there is almost universal consensus about Temasekscredentials as a highly successul investor.

Temasek is characterized by a number o noteworthy structural characteristics. Above all, Temasekseems to enjoy a very high degree o operational autonomy and reedom rom political intererence in

its day-to-day operations. In terms o its corporate governance, Temasek has an independent board,which consists largely o private sector business leaders and only one representative rom the Ministry

o inance. As o March 2006, around 40% o senior managers were non-Singaporeans. urthermore,more than a quarter o all proessional sta were non-Singaporeans. Operational independence has givenTemasek a great deal o discretion in terms o its investment decision-making. In addition to its bread

and butter o large equity stakes in domestic and oreign companies, Temasek has ventured into areassuch as private equity, real estate, and venture capital. Temaseks appetite or high risk and high return

is also evident in its medium-term portolio target o one-third exposure to the emerging markets o Asia.The remaining two-thirds are to be equally divided between Singapore and OECD economies. Thereore,

although all o Temaseks investment holdings were initially in Singaporean companies, the share oSingaporean companies in Temaseks portolio is now only one-third. Temaseks prots are paid back tothe government as dividends.

Temasek is very much an active investor and seeks to control or infuence the management o

the companies in which it buys equity stakes. In recent years, around 6070% o Temaseks portolioconsisted o investments where the companys share o total equity exceeded 20%. Indeed Temaseksoverall investment philosophy can be summarized as that o an active shareholder that seeks to maximize

sustainable shareholder value. Although Temasek has the luxury o adopting either long or short investment

horizons, in practice its guiding principle has been to manage or long-term value. While it might betempting or outsiders to dismiss the commercial orientation o a government-owned institution, it is inact accurate to say that Temasek has been run on a purely commercial basis, more or less like Goldman

Sachs. The one area where Temasek lags ar behind the likes o Goldman Sachs is transparency, although ithas recently made eorts to improve its transparency, such as publishing audited nancial statements.


37/50

sectioN v

sovereigN iNvestmeNtAgeNcies: experieNcesANd chALLeNges


GIC was initially established in 1981 ollowing the governments strategic decision to invest some

o the countrys oreign exchange reserves in higher-yielding oreign assets. It has become substantially

bigger than its older sister Temasek, and currently has estimated assets o at least $100 billion, with someestimates as high as $330 billion. Even though GIC started out as an investment management company

entrusted with investing part o reserves, its unds now come primarily rom the Singaporean governmentrather than the central bank. In contrast to Temasek, GIC does not own the assets it manages but manages

them on behal o its two clients, the government and the central bank. GIC is, in eect, an external undmanager or the two clients. GIC is external rather than internal because GIC, like Temasek, is run on a

purely commercial basis and enjoys a high degree o operational independence. Although the governmentsets GICs broad investment objectives (such as asset class mix, expected return, and risk tolerance)and monitors GICs perormance, it does not interere with the day-to-day operations. GIC nances its

operating expenditure by receiving a ee rom its two clients.

The biggest common actor binding the two SWs is that they are owned and unded by theSingaporean government. More specically, the Ministry o inance acting on behal o the governmentis the sole shareholder o both unds even though Temasek, unlike GIC, owns the assets it manages. As

such, Temasek and GIC pay all their dividends back to the nance ministry, and Temaseks average dividendyield to its shareholder during 19742006 has been an impressive 7%. What is less than ully transparent

is the mechanism or determining how much prot the two unds pay back to the government and howthe government spends the money it gets rom the unds. It is widely believed that GIC also serves as the

asset manager o the Central Provident und (CP). However, there is no ormal systematic relationshipbetween the two, and in particular, the rate o return on CP savings is unrelated to the rate o return onGICs investments.

It is conceptually helpul to think o both as managers o the wealth o a commercially active

and successul government. None o Temaseks unds and only a minor part o GICs unds comes romSingapores reserves. As noted beore, the two unds dier rom developing Asias new SWs in that theyare based on scal reserves, much like the oil unds o the Gul, rather than central bank reserves. That

is, the Singaporean unds are based on scal surplus, the prots o state-owned companies, and other

government net income, whereas the new Asian unds are based on oreign exchange market interventionby the central bank. Both GIC and Temasek use external und managers as an integral part o theirinvestment strategy. In the case o GIC, the share o assets under management by EM may be as high as

25% while Temasek uses external und managers when there are potentially high returns or as a means toestablish market presence.

A more undamental similarity between Temasek and GIC is that they have ully used their operationalautonomy to invest in a wide range o risky asset classes in pursuit o high returns. They have also madegood use o their long-term investment horizons to earn premiums or oregoing liquidity. or GIC, the

range o asset classes includes government and corporate bonds, equity, oreign exchange, commodities,real estate, private equity, venture capital, and inrastructure. Just like Temasek, GIC has established a

solid track record o consistently good investment perormance. According to company sources, between1981 and 2006, the average annual return has been 9.5% in US dollar terms and 8.2% in Singapore dollar

terms. The average annual return above global infation has been 5.3%. Moreover, GIC has added value inboth equities and bonds against the relevant industry indices. The two SWs have thus not only used their

operational independence vigorously but eectively as well. The overall picture that emerges is that otwo well-run and successul private-sector companies, not altogether dierent rom Goldman Sachs. Theperormance o GIC and Temasek so ar also indicates a high level o risk management capacity that has

eectively managed the overall risk o their portolios.

Box 7. continued.


38/50

November 2007


doNghyuN pArk

tABLe 4sovereign weALth fundsin deveLoping AsiA

economy nAmeof fund Assets

(us$ BiLLion)

yeArof

inception

type

Singapore Government o SingaporeInvestment Corporation

330 1981 No

beyond liquidity: new uses for developing asia's foreign exchange reserves

Documents