c025b97a5b6a2388ef7b7ecdda1e286d

8/12/2019 c025b97a5b6a2388ef7b7ecdda1e286d

1/34

International Monetary Fund |April 2014 1

Summaryhe landscape of portfolio investment in emerging markets has evolved considerably over the past 15years. Teir nancial markets have deepened and have become more globalized. New asset class seg-ments have developed, including local currency sovereign debt, with increased direct participation ofglobal investors. Te mix of global investors has also changed. Te role of bond fundsespecially local

currency bond funds, open-end funds with easy redemption options, and funds investing only opportunistically inemerging marketshas risen.

Tis chapter aims to identify the effects of these changes on the stability of portfolio ows and asset prices inemerging markets with a range of methods using relatively unexploited data. We examine the sensitivity of owsfrom various types of global investors to assess whether the new mix of investors has made portfolio ows moreor less sensitive to global nancial shocks. We also investigate the role of investor herding and domestic macrofundamentals during crises. Moreover, we analyze how the strength of local nancial systems affects the sensitivityof local asset prices to global nancial shocks.

We nd that the structures of both the investor base and local nancial systems matter. Te new mix of globalportfolio investors is likely to make overall portfolio ows more sensitive to global nancial conditions. Te shareof more volatile bond ows has risen, and larger foreign participation in local markets can transmit new instability.Growing investment from institutional investors that are generally more stable during normal times is welcome,but these investors can pull back more strongly and persistently facing an extreme shock. While domestic macro-economic conditions matter, investor herding among global funds continues, and there are few signs of increasingdifferentiation along macroeconomic fundamentals during crises over the past 15 years. Nonetheless, the progressmade by emerging markets toward strengthening their nancial systems reduces their nancial asset prices sensitivity to global nancial shocks.

Our results suggest options to enable emerging markets to reap the benets of nancial globalization while mini-mizing its potential costs. Governments can promote larger local investor bases, deeper banking sectors and capitalmarkets, and better institutions. Initiatives to support local currency bond market development are benecial, butthe size of direct participation of foreign investors in local markets needs to be monitored and balanced with broadnancial system development policies. Knowing the investor base and its characteristics is critical for assessing therisks of capital ow reversals and designing macroprudential policies.

2 C H A P T E R

HOW DO CHANGES IN THE INVESTOR BASE AND FINANCIALDEEPENING AFFECT EMERGING MARKET ECONOMIES?


2/34

GLOBAL FINANCIAL STABILITY REPORT: MOVING FROM LIQUIDIT Y TO GROWTH DRIVEN MARKETS

2 International Monetary Fund |April 2014

IntroductionFinancial markets in emerging market economieshave deepened signicantly over the past 15 years and witnessed substantial changes in their global and localinvestor bases. Improved fundamentals in emergingmarket economies and the persistently low yields inadvanced economies have encouraged a broader rangeof investors to increase their investment in the nancialassets of emerging market economies. Tis has helpedfoster the development of local nancial markets andof new asset classes, such as local-currency-denom-inated sovereign debt. Global investors are directlyentering local currency bond markets, while the localinstitutional investor base has also been expanding. Atthe same time, the relative role of cross-border banklending has declined, and within portfolio ows, xed-income ows have gained in importance compared with equity ows. Te composition of internationalmutual funds investing in emerging markets has beenchanging, with a growing importance of globally oper-ating funds that do not focus on emerging markets. All these investors differ in their mandates, constraints,and incentives and behave differently during volatiletimes (Figure 2.1).

Despite potential benets, these changes may haveheightened the exposure of emerging markets to globalnancial conditions and to contagion and herd-ing. Around mid-2013 and again in January 2014,

for example, uncertainty over U.S. monetary policyroiled the markets for emerging market securities andgenerated substantial sell-offs, particularly amongretail investors. Tis raises the question of whether thestructural changes discussed above have contributed toenhance emerging markets resilience to external nan-cial shocks. For example, the increased foreign presencein local markets is likely to have been fundamentalfor the development of these markets, but may havemade local asset prices more exposed to global factors.Te ability of governments to issue their debt in localcurrency has reduced their currency mismatches, but

the transfer of exchange rate risk to investors may havemade portfolio ows more volatile. Similarly, the largerrole of global investors in emerging market economiesbond markets may have made these ows more depen-dent on the ups and downs in global risk appetite.

Financial integration, especially if not managed

well, can make asset prices and portfolio ows moresensitive to global push factors and pose challengesto nancial stability in emerging markets.1 Tesemarkets have strengthened buffers, including larger

1Capital ows are driven by so-called push factors reectingcommon global conditions (such as monetary and scal policies inadvanced economies, global liquidity, and global risk aversion) andcountry-specic pull factors (such as local macroeconomic funda-mentals and institutional quality).

Te authors of this chapter are Hiroko Oura (team leader), Nico-ls Arregui, Luis Brandao-Marques, Johannes Ehrentraud, HibikiIchiue, and Prachi Mishra with contributions and research assistancefrom Soya Avramova.

Globalretail,340

Localinstitutional,

1,300

Globalinstitutional,

260

40

20

0

20

40

60

80

100

120

2004 05 06 07 08 09 10 11 12 13

U.S./European retail Japanese retai lGlobal institutional

About 80 percent of bonds of emerging markets are owned byinstitutional investors

but retail investor ows remain important and have been more

volatile.

1. Ownership of Emerging Market Bonds(Billions of U.S. dollars; as of 2013)

2. Bond Flows to Emerging Markets(Billions of U.S. dollars)

Figure 2.1. Investor Base for Bonds in Emerging Markets

Source: J.P. Morgan.Note: Global retail investors consist of European and U.S. mutual funds andJapanese investment trusts. Global institutional investors include investors withlong-term strategic mandates such as pension funds, insurance companies, andofcial funds. Local institutional investors encompass emerging market insurancecompanies and pension funds. Some market participants consider the guresunderestimate the assets and ows from global institutional investors.


3/34

CHAPTER 2HOW DO CHANGES IN THE INVESTOR BASE AND FINANCIAL DEEPENING AFFECT E MERGI


international reserves, exible exchange rates, andreduced exchange rate mismatches at the sovereignlevel. However, sudden large capital outows can stillinduce nancial distress with their effects on exchangerates and the balance sheets of banks, rms, and

households. Capital inows driven by global nancialconditions can help generate credit booms that sow theseeds of crises (Rey, 2013). Similarly, greater local assetprice exposure to global conditions makes fundingconditions for households, rms, and sovereigns moredependent on external nancial conditions.

Against this backdrop, this chapter aims to identifythe effects of changes in the global investor base andnancial deepening in the recipient emerging marketson their exposure to global nancial conditions. Inparticular, it assesses how these developments haveaffected the sensitivity of bond and equity ows as

well as asset prices (foreign and local currency bonds,equities, and currencies) to global nancial shocks.Te chapter complements research that has focused onmacroeconomic aspects of capital ows and macropru-dential and capital ow management policies.2

Specically, this chapter aims to answer the follow-ing questions: What do the changes in the global and local inves-

tor base of emerging markets over the past 15years imply for the sensitivity of portfolio flows toglobal financial conditions? Have global investorsbecome more discerning about local fundamentals?

Has herding declined as the new asset classes havematured?

What forms of financial deepening can reduce thesensitivity of emerging markets asset prices to globalfinancial shocks? Have developing local currencybond markets contributed to financial stability orhave they made local yields more exposed to globalfactors when combined with increased foreignparticipation?

Our ndings indicate that the sensitivity of portfolioows to global nancial conditions is likely to increaseand that herding among funds is on the rise. Weinvestigate global ows involving institutional inves-torsdened as large pension and insurance funds,international reserve funds, and sovereign wealthfundsby using a unique custodian database fromBank of New York Mellon (BNY). Flows from pre-

2For instance, see IMF (2012, 2013a, 2013b, 2013c) and Ghoshand others (2009).

dominantly retail-oriented mutual funds are examined with the Emerging Portfolio Fund Research (EPFRGlobal) database.3 Fixed-income flows are substantially more sensitive

to global push factors than are equity flows, and

their importance in overall capital flows is growing. Mutual fund investor flows (Box 2.1) are generallymore sensitive to global factors than those of insti-tutional investors, and they are expanding exposuresto emerging markets, making flows more sensitive topush factors (as witnessed during the recent bout of withdrawals from emerging market economies).

hese developments have been somewhat moderatedby large institutional investors, which contribute tothe stability of flows during normal times. How-ever, they pulled back more strongly and persis-tently when faced with extreme shocks. hey have

also been increasing their allocations to emergingmarkets, but not to the extent that they becomerelatively larger players in these markets (Figure 2.1).

Although country-level macroeconomic conditionsin emerging markets matter for resilience, their roleduring crises does not seem to have grown over timesince the late 1990s, and herding among globalmutual funds has been increasing.

Nonetheless, the progress made by emerging marketstoward nancial deepening and better institutionsmitigates some of the unpleasant side effects of nan-

cial globalization. Having a larger domestic investorbase, deeper banking sectors and capital markets, moreliquid markets, and better institutions all bring quan-titatively large benets. In particular, relying more onlocal currency debt makes bond prices more resilientto the ups and downs of international capital markets.

Yet, while foreign participation has often played akey role in developing local markets, a large share offoreign holdings of domestic debt comes with a height-ened bond price sensitivity to global nancial shocks.Tis further underscores the importance of developinga local investor base.

3Mutual funds are generally sold to retail investors, although arising number of institutional investors purchase mutual fund shares.In most of the past research, the EPFR Global data were analyzedonly at the aggregate level without differentiating across types offunds. Further details are given in Annex 2.1.


4/34


5/34



Evolving Emerging Market Assets and TheirInvestor BasesRising Importance of Portfolio Flows in Total CapitalFlows

Gross capital ows to emerging markets have quin-tupled since the early 2000s, and the most volatilecomponentportfolio owshas become a moreimportant part of the mix.4 Since the global nancialcrisis, portfolio ows to these economiesespeciallybond owshave risen sharply (Figure 2.2). Temarked swings of these ows around the time ofannouncements about the tapering of U.S. unconven-tional monetary policy have raised nancial stabilityconcerns. In contrast, deleveraging at European bankshas accelerated shrinkage of cross-border bankingows. While foreign direct investment is still the larg-est component of capital ows to emerging markets,it has been relatively stable through a number ofcrises.

Growing Importance in Global Portfolios and DeepeningFinancial Systems

Te nature of portfolio investment in emerging mar-kets has evolved as these markets have deepened andbecome more globally integrated. Over the past two

4See IMF (2013d and 2013e) for details on macrolevel trends ofcapital ows.

decades, capital markets in emerging market economiesoften developed in tandem with nancial integra-tion and liberalization (Box 2.2). Foreign participa-tion in emerging market equity markets took off inthe 1990s.5 In the 2000s, changes were concentratedin xed-income markets: many emerging marketsovereigns managed to shift from issuing hard cur-rency external debt to local currency domestic debt.

In doing so, they partially overcame original sin,a key historical source of vulnerability of emergingmarkets (Burger, Warnock, and Warnock, 2011).6,7 International investors now purchase local currencydebt in domestic markets and play a dominant role

5See Bekaert and Harvey (2000), Henry (2000), and World Bank(2013).

6Some of these trends mirror conscious policy efforts. See Felmanand others (2011) and Goswami and Sharma (2011) on the AsianBond Markets Initiative. Te G20 have also emphasized the develop-

ment of local currency bond markets, which led to a joint paper bythe IMF, the World Bank, the European Bank for Reconstructionand Development, and the Organization for Economic Coopera-tion and Development, titled Local Currency Bond MarketsADiagnostic Framework, in July 2013.

7Original sin refers to the inability of emerging market bor-rowers to issue debt to foreigners in local currency (Eichengreen,Hausmann, and Panizza, 2005), which leads to currency mismatchesunless accompanied by natural hedging and makes these economiesmore vulnerable to sudden stops of capital ows. Because of dataconstraints, the extent of original sin is difficult to assess for cross-border banking ows.

Table 2.1.2. Shares of Types of Mutual Funds(Percent of total assets under management)

Structure

Strategy Domicile Geography Currency

ETF

Non-ETF

TotalOpen-end

Closed-end Active Passive U.S. Offshore Global

EMRegional

HardCurrency

LocalCurrency

Bond Funds2003 0 100 89 11 100 0 54 46 0 19 66 32010 7 93 89 4 93 7 58 37 56 9 27 152013 9 91 89 2 91 9 58 38 47 7 23 28Equity Funds1996 0 100 92 8 100 0 65 20 0 38 2003 0 100 97 3 98 2 64 26 37 16 2010 19 81 79 1 77 23 50 32 35 25 2013 27 73 73 1 56 44 64 20 51 14

Sources: EPFR Global; and IMF staff calculations.Note: Global funds correspond to those that are categorized as Global or Global ex-U.S. by EPFR Global. The offshore markets are denedaccording to IMF (2008). Numbers may not add up due to rounding. The numbers reect the subsamples of EPFR Global data used in our analyses.EM = emerging market; ETF = exchange-traded fund.

Box 2.1(continued)


6/34


7/34



0

5

10

15

20

25

30

35

Dec.2002

Dec. 04 Dec. 04Sep.2003

Jun. 07Mar. 06Dec. 06 Sep. 08Dec. 08 Dec. 10 Dec. 12 Dec. 09 Mar. 11 Jun. 12 Sep. 13

AllFinancialNonnancial corporationsGovernment

10

0

10

20

3040

50

60

70

FinancialNonnancial corporations

Government

0

200

400

600

800

1,000

1,200

1,400

Dec.1997

Dec.2000

Dec. 03 Dec. 06 Dec. 09 Dec. 12

Local bondsInternational governmentInternational corporate

140 120 100 80 60 40 20 0MexicoBrazil

Average AETurkeyChina3South AfricaPoland3MalaysiaHungaryColombiaChileThailandIndonesiaIndiaRussia3PeruEgypt3Philippines

Argentina

2000:Q42013:Q4

0.6

0.7

0.8

0.9

1.0

March

1995

March

98

March

2001

March

04

March

07

March

10

March

13

Emerging markets totalBrazil, China, India, Russia, South Africa

Relying less on hard currencies(overcoming"original sin")

Relying more on hard currencies(severe "original sin")

0

10

20

30

40

50

60

70

2009 2013 latest

Figure 2.3. Transformation of Investment Options in Emerging Markets

Emerging markets are shifting issuance from international todomestic debt, except for rms.

1. Share of Emerging Market Debt Issued in International Marketsover Total Emerging Market Debt (Percent)

2. Net E merging Market Debt Issued in International Markets(Four-quarter average; billions of U.S. dollars)

3. Trading Volume of Emerging Market Bonds(Four-quarter average; billions of U.S. dollars)

4. Equity Market One-Way Trading Costs(Quarterly average fees and comissions; basis points)

5. Reliance on Hard Currencies for Emerging Market Governmentand Non-Government Bonds Issued in International CapitalMarkets (Simple average across countries)

6. Foreign Investor Participation in Local Government Bond Markets(Share of local government bond held by foreigners; percent oftotal outstanding)

Firms now issue more international debt than governments.

Trading volume has also shifted away from international governmentbonds and toward local bonds.

Fees and commissions for trading have declined, and someemerging markets offer cheaper trading costs than some advancedeconomies.

Emerging markets can now sell local currency debt to foreigners,partially overcoming original sin

and foreign investors have entered domestic governmentdebt markets, increasing external debt in disguise.

Sources: Asian Development Bank, AsianBondsOnline; Bank for International Settlements; Elkins-McSherry; Emerging Market Trading Association; J.P. Morgan;national authorities; and IMF staff calculations.Note: AE = advanced economy.Debt issued by former and current emerging markets based on the nationality of issuers (including debt issued by foreign subsidiaries of issuers headquarteredin emerging market economies). Sample includes Argentina, Brazil, China, Taiwan Province of China, Chile, Colombia, Croatia, Czech Republic, Hungary, India,Indonesia, Malaysia, Mexico, Pakistan, Peru, Philippines, Poland, Russia, South Africa, South Korea, Thailand, and Turkey.The gure shows the "original sin" measure following Eichengreen, Hausman, and Panizza (2005) and is calculated as max (1-[debt issued in the currency ofcountry i]/[all debt issued by country i], 0). Debt refers to international debt securities based on nationality issued by all sectors. Debt denominated in localcurrencies is assumed to be zero if data are not available.3Data for 2000:Q4 are not available.

P e r u

M a l a y s i a

M e x i c o

H u n g a r y

P o l a n d

I n d o n e s i a

R o m a n i a

T u r k e y

T h a i l a n d

B r a z i l

C z e c h

R e p u

b l i c

I n d i a


8/34



information on the allocation to emerging marketassets for some subgroups of investors.

Assets managed by mutual funds and institutionalinvestors have grown both in nominal amounts andrelative to GDP. In advanced economies, mutualfunds have gained in relative importance over thepast two decades, despite a decline since the globalfinancial crisis.12

Among U.S. investors, allocations to emerg-ing market assets have increased for equities andbonds. Both institutional and retail investors have

12Tis trend in advanced economies is also pointed out in Chapter2 of the September 2011 GFSR.

allocated more to emerging market assets (see Box2.3). Among mutual funds, global funds with moreglobally diversified portfolios have strengthenedtheir engagement in emerging markets over thepast decade despite some retrenchment since 2011.Still, portfolio flows to emerging markets continueto be very small compared with those to advancedeconomies.

Across regions and countries, portfolio flows frominstitutional investors and mutual funds have gener-ally grown in tandem. However, institutional bondinvestors appear to differentiate more across regions.

Hedge fund investment in emerging markets hasstagnated since the global financial crisis.

1. Share in Global GDP(Percent of global GDP)

3. Share in Global Bond Market Capitalization and Index(Percent)

2. Share in Global Equity Market Capitalization and Index(Percent)

4. Sovereign Ratings Distribution(Cumulative distribution; percent)

Emerging markets share in world GDP has grown...

Emerging markets share in bond market capitalization has grown,and their share in global high-yield index has caught up recently

and their share in global indices has risen, though less than theirshare in market capitalization would suggest.

helped, in part, by a narrower rating gap with advancedeconomies.

Source: IMF, World Economic Outlook database; J.P. Morgan; Morgan Stanley Capital International (MSCI); Standard and Poor's; and IMF staff calculations.Note: AE = advanced economy; EM = emerging market; WEO EMs =emerging market economies classied as such in the World Economic Outlook database. SeeTable 2.4 for sample economies.

Figure 2.4. Emerging Markets: Shares in Economic Activity and Financial Markets

05

1015202530354045

1 9 9 4

9 5

9 6

9 7

9 8

9 9

2 0 0 0

0 1

0 2

0 3

0 4

0 5

0 6

0 7

0 8

0 9

1 0

1 1

1 2

1 3

WEO EMsSample EMsSample EMs excluding China

0

20

40

60

80

100

120

AA A AA A+ A BBB BB+ BB B CCC+ CCC C

2013 EMs

2000 EMs

2013 AEs

2000 AEs

Improvement

0

2

4

6

8

10

12

14

16

1 9 9 4

9 5

9 6

9 7

9 8

9 9

2 0 0 0

0 1

0 2

0 3

0 4

0 5

0 6

0 7

0 8

0 9

1 0

1 1

1 2

1 3

EM bond in global market capitalizationEM share in J.P. Morganglobal high-yield bond index

0

5

10

15

20

25

30

1 9 9 4

9 5

9 6

9 7

9 8

9 9

2 0 0 0

0 1

0 2

0 3

0 4

0 5

0 6

0 7

0 8

0 9

1 0

1 1

1 2

1 3

EM equity in global market capitalizationEM share in MSCI global equity index


9/34



Recent developments Financial depth can be dened by the size of nancialmarkets relative to economic activity and by the variousfunctions those markets perform. Functions includeintermediation, price discovery, and hedging. Tis widerange of functions can be measured by bank-based indi-cators (such as credit to GDP), market-based indicators(such as the market sizes for government and corporatebonds, foreign exchange, and derivatives), and indica-tors of nancial access.

Financial markets in emerging market economieshave generally deepened over the past decade butunevenly over time and across different dimensions(Figure 2.2.1). Te period since the early 2000s has witnessed broad-based nancial deepening in most

segments of the nancial systems. At the same time,growth in international government debt has beenlimited, partly because of emerging markets efforts toreduce external vulnerabilities. Some market activi-ties, such as measured by stock market capitalization,mutual fund assets, and interest derivatives, haveshrunk since 2007.

Emerging markets nancial systems, however,remain thinner than those of advanced economies(AEs), with substantial variations among emergingmarkets (Figure 2.2.1, panels 2 and 3). In particular,emerging markets insurance companies, mutual funds,international corporate debt, and interest rate deriva-

tives markets are generally small compared with thoseof AEs. While this is not surprisingnancial systemstend to deepen as countries developsome segmentsof nancial systems, such as the size of domesticmoney markets and spot foreign exchange markets, arelarger in some emerging markets than in AEs.

Financial deepening and economic and nancialstability in emerging markets Financial deepening does not guarantee nancialstability. Te benets of nancial deepening in reduc-ing economic and nancial volatility emanate mainlyfrom the nancial sectors role in allocating savings toproductive use, smoothing consumption, and provid-ing price discovery mechanisms and hedging oppor-tunities. Empirical evidence suggests that in emergingmarkets, deepening nancial marketsin particularstock and money marketsand making markets moreliquid (as measured by reducing bid-ask spreads in

Box 2.2. Financial Deepening in Emerging Markets

Bank private creditNonbank private credit

Stock marketcapitalization

Dom. governmentdebt

Int.government debt

Dom. nongov. debt

Int. nongov. debt

Dom. money marketSpot FX turnover

FX derivative turnover

OTC interestderivative turnover

OTC interestderivative turnover

Insurance companyassets

Mutual fund assets

Pension fund assets

Insurance companyassets

Pension fund assets

2001 2007 Latest

Bank private creditNonbank private credit

Stock marketcapitalization

Dom. governmentdebt

Int. governmentdebt

Dom. nongov. debt

Int. nongov. debtDom. money market

Spot FX turnover

FX derivative turnover

Mutual fund assets

AE EM median EM 75thpercentile

EM 25hpercentile

020406080100120140160180

Bank privatecredit

Nonbank privatecredit

Stock market

size

Dom.gov.debt

Int.gov.debt

Dom.nongov.

debt

Int.nongov.

debt

EM 2001 EM 2007

EM latest AE latest

Figure 2.2.1. Financial Deepening inEmerging Markets

1. The Evolution of Financial Deepening inEmerging Markets(Percent of 2001 emerging market median)

2. Financial Deepening in Emerging MarketsRelative to Advanced Economies(Percent of latest advanced economy median)

3. Financial System Size by Segments(Group median; percent of GDP)

Sources: Bank for International Settlements; European

Fund and Asset Management Association; Organization forEconomic Cooperation and Development; national mutualfund associations; World Bank, Global Financial Develop-ment database; and IMF staff calculations.Note: AE = advanced economy; dom. = domestic; EM =emerging market; FX = foreign exchange; int. =international; OTC = over the counter. Latest data are for2013, except for bank credit, money market instruments,and mutual fund assets (2012); and for insurance andpension fund assets (2011).

050

100150200250300350

0255075

100125150175

Te authors of this box are Johannes Ehrentraud, PrachiMishra, Kenji Moriyama, Papa NDiaye, and Hiroko Oura.


10/34



Identifying the Financial Stability Effects ofChanges in the Investor Base and in LocalFinancial SystemsApproach

Tis section examines how changes in the investorbase and in local nancial systems of emerging marketeconomies have affected portfolio ows and asset pricesin these economies. he evolving role of global and local factors over

time: We document the evolution of correlationsin emerging market asset returns with global assetreturns. We then consider whether macro funda-mentals have become more important over timein explaining cross-country differences of emerg-ing market asset returns during crises (Box 2.4).his section also explores trends in investor herdbehavior.

Investor characteristics and portfolio flows: We measurethe sensitivity of portfolio flows to global risk factorsby estimating a panel model with global and domes-tic factors. he focus is on differences in sensitivitiesto global risk factors across types of investors. Sinceinvestors may behave differently when faced withextreme shocks, flows of mutual funds and institu-tional investors are also examined specifically aroundcrises.

Local financial systems and emerging market asset prices: We explore the impact of financial deepeningon the sensitivity of emerging market asset returnsto global risk factors using a technique similar tothat for portfolio flows.13 he panel model includes

various global push and domestic pull factors. heanalysis encompasses the role of local investors, who

13 Annex 2.1 shows additional results on the relationship betweenlocal macroeconomic indicators and asset prices and ows.

foreign exchange and bond markets) can enhance mac-roeconomic resilience. In contrast, additional benetsfrom deeper banking systems are likely to be more

limited. Deepening debt markets on the other handmay increase economic volatility.1

Hence, the overall effect of nancial deepening onan economys exposure to global nancial conditionsexplored in this chapter is ambiguous a priori. More-over, the empirical analysis is complicated by the factthat nancial deepening has often occurred alongsidenancial integration, and separating the effects of thesetwo interrelated but distinct dimensions is difficult.Te following summarizes recent literature on therelationship among nancial deepening, asset prices,and capital ows. Financial deepening and asset price sensitivity to global financial conditions: Increased market transparencyand liquidity, coupled with a broader local investorbase, should allow local markets to absorb externalshocks more easily. For example, a broader domesticinvestor base can prevent prices from overshootingor undershooting in response to sales or purchasesby foreigners that are driven by external factors.2

1Te relationship between nancial deepening and economicoutcomes will be explored in Financial Deepening in EmergingMarkets (IMF, forthcoming).

2Empirically, Alfaro, Kalemli-Ozcan, and Volosovych (2007);Chapter 3 of the April 2007 GFSR; and Broto, Daz-Cassou,

More liquid domestic markets can be expected tocontribute to stability by reducing the price impactof capital flows.

Financial deepening and capital flow sensitivity to global financial conditions: Improved local institu-tions, enhanced market transparency, a broaderinvestor base, and increasingly sophisticated localinvestors are likely to promote price discovery andreduce herding, thereby making flows less suscep-tible to global conditions.3 However, these marketscould experience more volatile flowsthough withlower price volatilityif global investors, facingdistress, prefer to unwind their positions in deepermarkets first, where the price impact is expected tobe smaller.4

and Erce (2011) report that more developed nancial sectors areempirically associated with less volatile portfolio ows. Tesestudies, however, do not relate nancial deepening to the sensi-tivity to global factors.

3Merton (1987) originally proposed the investor-base-broad-ening hypothesis. Wang (2007) extends the setting to discuss therole of foreign investors after nancial liberalization. Umutlu, Akdenizb, and Altay-Salih (2010) nd empirical support for thishypothesis.

4Broner and Ventura (2010) develop a model in whichcountries with deeper nancial markets experience more volatilecapital ows due to changes in investor sentiment.

Box 2.2(continued)


11/34



Over the past decade, institutional investors have beenallocating more funds to emerging markets. Despitedifferences in mandates ( able 2.3.1), all types of

institutional investors are attracted to emerging marketassets by their relatively high returns. Economic growthtrends, real currency appreciation, and deepening capitalmarkets in emerging market economies have spurredthe demand for emerging market assets. For pension funds, the decline in funded ratios led

them to pursue higher returns by relaxing invest-ment constraints. hey have been diversifying assetclasses, in particular to include local currency debtin emerging markets (J.P. Morgan Asset Manage-ment, 2009). heir current allocations to emergingmarket assets are still low, however. Allocation toemerging markets is expected to rise to 10 to 20percent over the longer term (OECD, 2013).

Insurance firms have increased their exposures toemerging markets since 2008 (despite a minorsetback in 201213; Figure 2.3.1, panel 1), and

interest continues to grow (Financial imes, 2013).In a recent survey of investment officers (Siegel andMorbi, 2013), more than 40 percent of insurance

companies intended to increase their allocationsto equity and to hard currency corporate debt inemerging Europe, the Middle East and Africa, andemerging Asia. Investment officers expect a 30percent increase for emerging markets.

Central bank reserve managers, who collectivelyhandle $11 trillion in assets, tend to be conservativeinvestors. Nonetheless, they do invest in emergingmarkets, and the most popular of those destinationshave been Brazil, China, Mexico, Poland, South

Africa, Korea, and urkey. Reserve managers arealso raising their allocation to emerging markets inline with their economic size and diversifying awayfrom hard currencies (Figure 2.3.1, panel 2).

Sovereign wealth funds have progressively expandedtheir exposure to emerging markets, especially toBrazil, China, India, Russia, and urkey. otal dealflows, concentrated in equity acquisition, peaked in2010 at $20 billion, then receded to $10 billion in2013 (Figure 2.3.1, panel 3).

Box 2.3. Investment Strategies of Institutional Investors

Table 2.3.1. Investment Constraints of Institutional Investors

Investor Type Risk Tolerance Time Horizon Need for Liquid AssetsRegulatoryConstraints

Private Pension Plan

(defined-benefit)

Determined by surplus, age of

workers, balance sheet

Long Depends on age of workers and

percent of retirees to totalworkforce

High

Life Insurance Fixed-income conservativeSurplus aggressive

Medium to long Fixed-income high Surplus low

High

NonLife Insurance Fixed-income conservativeSurplus aggressive

Short Fixed-income high Surplus low

Moderate

Central Bank ReserveFunds

Depends on international reserveamount and adequacy

Short Medium to high Moderate

Sovereign Wealth Funds Fiscal Stabilization

FundDepends on fiscal budget,

conservativeShort Mostly government bonds with

high liquidityLight

Savings Fund High risk-return profile Long Primarily equity and alternativeswith low liquidity

Light

Public Pension Fund Medium, high allocation to equityto hedge wage growth

Long Depends on immediacy ofcontingent claims, mediumto low

High

Sovereign WealthReserve Fund

Higher risk-return profile Long Low Light

Sources: Al-Hassan and others (2013); Chartered Financial Analyst Institute Curriculum; Papaioannou and others (2013); and Morahan and Mulder(2013).Note: The insurance surplus is assets above the reserves set aside for future insurance payout and is used to develop new business; it has a higherrisk-return prole than the reserves that are usually invested in xed-income assets.

Te author of this box is Soya Avramova.


12/34



Box 2.3(continued)

0

2

4

6

0

5

10

15

20

25

2002 03 0 4 0 5 06 07 08 0 9 10 11 12 13

EquityBonds (right scale)

0

2

4

6

8

0

100

200

300

400

1999 2001 03 05 07 09 11

Other currencies, including emerging markets AEs (right scale)EMs (right scale)

-

1

2

3

4

5

M a l a y s i a

I n d o n e s i a

C h i n a

H u n g a r y

I n d i a

J o r d a n

T u r k e y

T h a i l a n d

I n d o n e s i a

I n d i a

T u r k e y

I n d i a

C h i n a

T h a i l a n d

M a l a y s i a

S o u t h A

f r i c a

B r a z i l

I n d i a

R u s s i a

M a l a y s i a

I n d o n e s i a

C h i n a

S o u t h A

f r i c a

V i e t n a m

R u s s i a

M a l a y s i a

I n d o n e s i a

C h i l e

I n d i a

B r a z i l

C h i n a

B r a z i l

R u s s i a

T h a i l a n d

S o u t h A

f r i c a

I n d i a

M a l a y s i a

C h i n a

R u s s i a

I n d i a

C h i n a

B r a z i l

T u r k e y

I n d o n e s i a

B r a z i l

P h i l i p p i n e s

R u s s i a

I n d i a

C h i n a

2005 06 07 08 09 10 11 12 13

China $10.7China $8.7Turkey $6.2

Figure 2.3.1. Investments of Institutional Investors in Emerging Markets

Source: SWF Institute.Note: AE = advanced economy; EM = emerging market; SWF = sovereign wealth fund. Minimum investment is set at$100 million. The SWF Institute data cover about 53 percent of the capital ows to EMs in 2012. The data cover mainlyinvestment in equity, real estate, and infrastructure.

3. Sovereign Wealth Funds' Capital Flows by Country(Billions of U.S. dollars)

Sources: EPFR Global; and IMF staff estimates.Note: The sample includes about 4 percent of insurancerms in Organization for Economic Cooperation andDevelopment economies.

Insurance companies are investing more in EMs,but they pulled back mildly around the taperingannouncement in 2013.

Brazil, China, India, Russia, and Turkey continue to attract sovereign wealth fund capital.

Foreign reserve managers have been cautiouslydiversifying out of hard currencies.

Source: IMF Currency Composition of Ofcial ForeignExchange Reserves database.Note: Currencies other than those of the G7, Australia,and Switzerland. Data up to 2012 classify Canadian and Australian dollars as other currencies. Data withunknown currencies are excluded.

1. Insurance Companies Investment inEmerging Markets(Billions of U.S. dollars)

2. Foreign Reserves Asset Allocation byCurrency(Billions of U.S. dollars, left scale;percent, right scale)


13/34



hold the largest share of emerging market bonds(Figure 2.1).

Tese analyses cover a wide range of emerging markets,including former emerging and frontier markets (able2.4 in Annex 2.1). Annex 2.1 describes the details ofthe data and empirical frameworks.

The Evolving Role of Global and Local Factors

As emerging markets have become increasingly inte-grated with global markets, global factors have increas-ingly driven emerging market asset returns (Figure2.6).14 Although the heightened correlation of local

asset returns with global market returns (beta) duringthe global nancial crisis may partly reect the effectsof higher asset volatility typical of weak markets, equitybeta has remained at high levels (above one) sincethen. Te beta for emerging market bonds (especially

14See Forbes (2012), Bekaert and others (2011) and Bekaert andHarvey (2000) on equity market integration. See Burger, Warnock,and Warnock (2011) and Miyajima, Mohanty, and Chan (2012) onbond market integration.

those denominated in the local currency) is muchlower than that for equities but is rising rapidly.

Although country-level macroeconomic conditionsin emerging markets (pull factors) matter for assetprice resilience, their role during distress episodes doesnot seem to have risen since the late 1990s. Looking atdistress episodes for emerging markets since the Asiancrisis, it does not seem that the relative role of macro-economic fundamentals in explaining contagion pat-terns has been rising over time (Box 2.4). Tis couldbe partly because macroeconomic vulnerability hasbeen reduced in many emerging markets in the past 15years, keeping them within a comfort zone for manyglobal investors despite global turbulences.15

At the same time, herding among internationalequity investors is on the rise (Figure 2.7). If interna-tional investors buy or sell assets simply because theyobserve other investors doing so, this can amplify

15Dynamics within volatility periods also change: after an initialgeneralized sell-off in May and June 2013, nancial assets of emerg-ing markets with better macroeconomic fundamentals recoveredmore strongly than those with weaker fundamentals in the followingmonths (Box 2.4).

Table 2.1. Size of Global and Local Institutional Investors and Mutual Funds(Trillions of U.S. dollars, unless indicated otherwise)

1995 2000 2005 2007 2009 2011 2012Assets under management of mutual funds and institutional investors

Selected advanced economies1,2Total assets 22 35 5 68 65 70 76

Total as percent of GDP 96 143 159 179 172 167 180Mutual funds 6 13 19 26 25 26 29Of which

Share of open-end funds in total mutual fund assets (percent) 3 94 97 96 97 97 97 97Institutional investors 16 23 34 41 40 44 47

Share of institutional investors in total assets (percent) 72 64 64 61 61 63 61Selected emerging market and other economies 2,4

Total assets . . . . . . 2.3 4.4 4.8 6.4 . . .Total as percent of GDP5 . . . . . . 32 36 37 36 . . .

Mutual funds . . . . . . 0.8 1.9 1.9 2.3 . . .Institutional investors . . . . . . 1.5 2.5 2.9 4.1 . . .

Share of institutional investors in total assets (percent) 5 . . . . . . 65 59 60 62 . . .International reserves, excluding gold

Advanced economies1 0.7 0.8 1.3 1.5 1.8 2.3 2.5Emerging market and other economies 4 0.4 0.7 2.0 3.6 4.7 5.9 6.2

Sources: IMF, International Financial Statistics and Word Economic Outlook databases; Organization for Economic Cooperation and Development; WorlFinancial Development database; and IMF staff estimates.1Including Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Japan, Luxembourg, Netherlands, Norway, Spain, United KUnited States.2These data may reflect some double-counting of assets, such as those owned by pension funds and managed by mutual funds.3The data include Australia, Finland, France, Greece, Spain, United Kingdom, and United States.4Including Argentina, Brazil, Chile, China, Colombia, Croatia, Czech Republic, Egypt, Hungary, India, Indonesia, Latvia, Lithuania, Malaysia, Mexico, Peru, Philippines, Poland, Russia, South Africa, Korea, Sri Lanka, Thailand, Turkey, Uruguay, and Vietnam. Mutual fund data for China start in 2007.5Excluding China.


14/34



0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

2004 05 06 07 08 09 10 11 12

EM equity in percent of total U.S. equity holdingsEM bond in percent of total U.S. bond holdings

EPFR institutional3 EPFR retail3

0

20

40

60

80

100

120

140

0

2

4

6

8

10

12

14

2002 04 06 08 10 12

AUM of dedicated

EM HF in percentof AUM(left scale)

Number of dedicated EM HFin percent of total HF(left scale)

AUM of dedicatedEM HF (right scale)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

July1997

July2000

July03

July06

July09

July12

Aug.1997

Aug.2000

Aug.03

Aug.06

Aug.09

Aug.12

Equity Bonds

0.5

0

0.5

1

1.5

2

Bonds Equity Bonds Equity Bonds Equity Bonds EquityEmerging Asia Emerging Europe L atin America Others4

0

5

10

15

20

25

June2001

June03

June05

June07

June09

June11

June13

Equity funds Bond funds

Figure 2.5. Allocation to Emerging Market Assets

U.S. investors (including both mutual funds and institutional investors)have increased their allocations to emerging markets.

1. Share in U.S. Investors' Portfolios 1

(Percent)2. Emerging Market Assets in Global Investors' Portfolio Flows

(Cumulative ows to EMs; relative to cumulative ows to othereconomies, December 2010 = 1) 2

3. Bond Flows by Types of Investors (Net inows between October2008 and September 2013; percent of GDP)

4. Portfolio Inows, by Region and Investor Type(Net inows between October 2008 and September 2013; percent ofGDP)

5. Share in Global Mutual Funds' Assets(Percent)

6. Dedicated Emerging Market Hedge Funds(Percent, left scale; billions of U.S. dollars, right scale)

More recently, retail investors have also been increasing theirengagement in emerging markets.

Mutual funds and institutional investors have largely invested in thesame countries, although the relative size of both types of inows has

varied somewhat across countries.

In emerging Europe, institutional investors pulled back since the globalnancial crisis, while mutual funds continued to invest.

The role of emerging markets in global funds portfolios has becomemore important, despite a recent setback...

while hedge fund investments in emerging markets have stagnatedsince 2008.

Sources: Bank of New York Mellon; EPFR Global; Federal Reserve; Hedge Fund Research; U.S. Treasury International Capital System, and IMF st aff estimates.Note: AUM = assets under management; EM = emerging market; HF = hedge funds.1U.S. portfolios include both domestic and foreign securities.2Cumulative ows are calculated using monthly ows-to-assets under management in order to control for expanding coverage of the data. Data end in October 2013.3See Annex 2.1 for EPFR Global denitions for institutional and retail investors.4Others include Egypt, Israel, and South Africa.

0.0 0.5 1.0 1.5 2.0 2.5 3.0Malaysia

South Africa

Mexico

Colombia

Peru

PhilippinesTurkey

Romania

Brazil

Chile

Mutual fundsInstitutional investors

Mutual fundsInstitutional investors


15/34



boom-bust cycles in nancial markets (Box 2.5). Inprinciple, as information about emerging marketsbecomes more widely available and coverage by coun-try analysts increases, country fundamentals shouldbecome more important determinants of investmentdecisions, and herding among international investorsshould decline (Calvo and Mendoza, 2000). However,there is little evidence of such a shift over the past 15years. For equity funds, herding behaviormeasuredby the extent to which a certain group of inves-tors trades in the same direction more often thanone would expect if they traded independently andrandomlyweakened somewhat in the rst half of the2000s but has continuously climbed since then. Telow-interest environment since the mid-2000s mayhave contributed to this rise, as investors searched foryield, neglecting country-specic risks and followingother investors. Although there is no clear trend for

bond funds, herding tends to pick up during turbulenttimes, and has been rising over the past two years, as

well.16

Investor Characteristics and Portfolio Flows

Bond Fund Flows versus Equity Fund Flows

Bond ows are much more sensitive to global nancialconditions than equity ows. Separate economet-ric analyses for bonds (covering both sovereign andcorporate bonds) and equities (using country-levelEPFR Global data for mutual fund ows and BNY

data for institutional investors) reveal a stark contrastbetween bond and equity investors. Figure 2.8, panel1, compares the sensitivity of bond and equity owsfrom mutual funds and institutional investors to a onestandard deviation rise (about 8 percentage points) in

16Note that a common move by global funds to emerging markets with better fundamentals during a period of volatility would alsoshow up in our measure as a temporary spike in herding (see Box2.5).

1.0

0.5

0.0

0.5

1.0

1.5

2.0

1 9 9 5

9 6

9 7

9 8

9 9

2 0 0 0

0 1

0 2

0 3

0 4

0 5

0 6

0 7

0 8

0 9

1 0

1 1

1 2

1 3

Equity Hard currency bondLocal currency bonds (hedged)

Source: IMF staff estimates.Note: Betas are coefcients estimated by regressing returns of emerging marketsassets on global market returns, using a monthly panel model with country xedeffects and a ve-year rolling window. If beta is negative, then the return of theasset rises when the overall market is declining, and that asset provides betterdiversication benets for investors. If beta is above 1, then the asset is not onlyhighly correlated with the market return but also rises or falls more than themarket return. Global market returns are measured using the S&P 500 Index forequities and the City World Government Bond Index for both hard currency andlocal currency sovereign bonds. Both country and global market returns aremeasured in excess of the one-month Eurodollar deposit rate.

Beta of Emerging Market Assets vis--vis Global Market Returns(Five-year rolling estimation of emerging market returns on global marketreturns)

Figure 2.6. Integration of Emerging Market Assets into GlobalMarkets

M a r . 1 9 9 7

J a n

. 9 8

N o v . 9 8

S e p

. 9 9

J u l y 2 0 0 0

M a y . 0 1

M a r . 0 2

J a n

. 0 3

N o v . 0 3

S e p

. 0 4

J u l . 0 5

M a y . 0 6

M a r . 0 7

J a n

. 0 8

N o v . 0 8

S e p

. 0 9

J u l . 1 0

M a y . 1 1

M a r . 1 2

J a n

. 1 3

EM equity EM bond

More herding

Less herding

Sources: EPFR Global; and IMF staff estimates.Note: EM = emerging market. The herding measure is that proposed by LakonishokShleifer, and Vishny (1992). It assesses the strength of correlated trading amongmutual funds investing in EMs, controlling for their overall trade trends (see Box 2The measure is 0 when there is no sign of herding. It is calculated every month,looking at the fund-level activity in each country, and then averaged across EMs. Tmeasure is computed when there are at least six funds in a month.

Figure 2.7. Herding among Equity and Bond Funds Investingin Emerging Markets(Percent)


16/34



Over the past 15 years, the impact of crises was notuniform across emerging markets, and the literature hassought to identify the macroeconomic conditions that

determine the susceptibility of countries to shocks.1 Weassess whether variations in domestic macroeconomicfundamentals across emerging markets are increasinglyinuencing market participants in their investmentdecisions during crises.

Method

Our study compares the cross-country pattern of assetprice movements during six crises affecting emerging mar-kets. For each event, we estimate the relationship betweenmarket pressure and macroeconomic variables acrossemerging markets, then compare the explanatory power(measured by theR squared) of the macroeconomic vari-ables across episodes to see whether markets have becomemore discriminating over time. Crises: he analysis covers the hai, Russian, and

Brazilian crises; the global financial crisis; the Euro-pean crisis; and the 2013 sell-off episode owing toconcerns over U.S. monetary policy.

Market pressure: Market pressure is measured bychanges in exchange rates, an index of exchangemarket pressure, and bond and equity pricesbetween the beginning and the end of an episode.

Macroeconomic variables: Given the small samplesize for each stress episode, we use only a few keymacroeconomic variables from the literature: theinflation rate, the current account balance, a mea-sure of trade linkages with the country where thecrisis originated, and a measure of financial open-ness. All macroeconomic variables are taken prior tothe crisis episode to reduce endogeneity concerns.

Te limited sampling and highly varied crises mean thatthe results should be interpreted as only indicative.

ResultsTe role of individual macroeconomic variablesappears to be tied both to specic markets and tospecic crises. rade linkages and ination have played

a signicant role across multiple types of crises andfor several types of assets. Countries with a strongertrade connection with the shock-originating economyexperience higher market pressure on asset prices.Markets also seem to exert more pressure on countries with higher ination. Te current account balance, which has been agged as an important determinant

Te author of this box is Nicols Arregui.1See Aizenman and Pasricha (2012); Feldkircher, Horvath, and

Rusnak (2014); and Eichengreen and Gupta (2013).

of pressure on emerging market asset prices, does notappear to be a robust inuence.2

Te explanatory power of fundamentals across crisesdoes not suggest that investors are becoming more discrim-inating among emerging markets according to differencesin their macroeconomic fundamentals (Figure 2.4.1).3 However, within the tapering-related sell-off episode in2013, the R squared of macroeconomic fundamentals roseover time. Market participants agree that investors startedto differentiate more across countries over the summer2013 after the initial generalized sell-off in May and June.

2See Chapter 1 of the October 2013 GFSR and IMF (2014).3Te results are robust to the use of additional controls,

including the real effective exchange rate appreciation, real GDPgrowth, the scal balance in percent of GDP, total foreign debtin percent of GDP, an indicator for the exchange rate regime,and the size of the economy.

Box 2.4. Are Investors Differentiating among Emerging Markets during Stress Episodes?

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Thailand Apr.

Aug. 1997

RussiaJun.

Sep. 98

BrazilDec. 98Jan. 99

GFCSep.

Dec. 2008

EuropeJul.

Sep. 11

TaperingMay

Sep. 13

EquitiesBondsExchange rateExchange market pressure

Figure 2.4.1. Role of MacroeconomicFundamentals over Time

Macroeconomic Fundamentals Explanatory Poweracross Stress Episodes(R squared of separate cross-country regressions byepisode)

Source: IMF staff calculations.Note: GFC = global nancial crisis. The dependentvariables are changes in exchange rates, bond and equityprices, and an exchange market pressure index over thecrisis periods. Explanatory variables are trade linkages,consumer price ination, ratio of current account balanceto GDP, and nancial openness. Models a re estimated byordinary least squares with robust standard errors. SeeTable 2.4 for sample economies and Table 2.5 for variabledenitions.


17/34



the VIX (the Chicago Board Options Exchange Mar-ket Volatility Index).17 In the case of mutual funds,bond ows are clearly more sensitive to the VIX thanequity funds. For institutional investors, however, thedifference between the sensitivity of bond and equity

funds is not statistically signicant.18

17Te results are based on panel regressions for bond and equity

ows from mutual funds and institutional investors on various globaland domestic factors. See Annex 2.1 on details of the mutual fundow data calculations.

18Te results are generally robust when other measures of globalfactors are used, including the volatility of the two-year Eurodollarinterest rate future; the Merrill Lynch Option Volatility Estimate(MOVE) index; and the reasury and Eurodollar ( ED) spread. Ourresults and other research (see for example Gonzlez-Hermosillo,2008; Fratzscher, 2012; and Rey, 2013) suggest that different sets of

Institutional Investors versus Mutual Funds

As expected, global mutual funds react more stronglyto global nancial shocks than do large global institu-tional investors (Figure 2.8). Te results conrm thatows from retail-oriented mutual funds (EPFR Global

data) are signicantly more sensitive to the VIX thanows from the institutional investors (BNY data) forboth bonds and equities. Tis may reect the fact that

global factors are relevant for emerging market portfolio ows andasset returns. Similarly, IMF (2013d and 2013e) and Chapter 1 ofthe April 2013 GFSR nd that global factors such as the VIX andto a lesser extentgovernment bond yields have played a signicantrole in explaining swings in portfolio ows to emerging markets. Ingeneral, however, the VIXoften interpreted as a measure of globalrisk aversionis the factor that plays a signicant and robust role.

Herding in nancial markets emerges when investorsmimic other investors. Such behavior can destabilizenancial markets, aggravate shocks, and lead to mispric-

ing or asset price bubbles. While herding can be theresult of cognitive biases or of heuristic-based decisionmaking, it can also be rational. For instance, herdingmay emerge if less-informed asset managers followtheir possibly better-informed peers instead of relyingon their own assessments (Bikhchandani, Hirshleifer,and Welch, 1992). Herding may also be rational forasset managers if they are evaluated against each other(Scharfstein and Stein, 1990) or vis--vis similar bench-marks (Maug and Naik, 2011).

We use a herding measure that quanties comove-ments in trading patterns for a subgroup of inves-torshere, international funds investing in emergingmarkets. Tis measure, originally introduced byLakonishok, Shleifer, and Vishny (1992), assesses whether funds move in the same direction more oftenthan one would expect if they traded independentlyand randomly. Te herding measure (HM ) is the aver-age across countries of the following country-specicherding metric:

HM c ,t = | pc ,t pt | AF c ,t , (2.5.1)

where pc ,t is the proportion of all funds active in coun-try c and month t that are net buyers (1 pc ,t is theproportion of net sellers), pt is its expected value, and

AF c ,t is an adjustment factor to ensure thatHM c ,t is zeroif there is no herding.1 pt is approximated by the shareof funds that are net buyers across all emerging markets

Te authors of this box are Johannes Ehrentraud and HibikiIchiue.

in our sample, and is allowed to be time-varying tocontrol for common trends across countries, such asswings in aggregate inows to emerging markets due to

marketwide developments.2SinceHM c ,t measures the correlation in trading

patterns, it gives only indicative evidence of trueherding.3 A positive value of the measure in a givenperiod may also reect, for instance, the inclusionof a country in a benchmark index or regulatorychanges affecting this subgroup of investors in speciccountries. However, a generalized market reaction tofundamental news should not necessarily result in spu-rious positive herding values with this measure, since,for example, not everybody can react to bad news byselling: there must be a buyer for every seller. Con-sequently, forHM to misclassify a reaction to newsabout fundamentals as herding, news must either (1)fundamentally affect the group of mutual funds stud-ied here in a different manner than other investors or(2) propagate slowly across different types of investorgroups (which should not be an important issue at themonthly frequency (see Cipriani and Guarino, 2014).

1Te adjustment factor is equal to the expected value of therst term under the null hypothesis that there is no herding. Itis needed since the distribution of the rst term is not centeredaround zero.

2In this chapter, we show that mutual funds react more toglobal nancial shocks than do other investors. o the extentthat this results in uniform relative changes in emerging marketallocation across countries, this effect is controlled for by pt .

3See the discussions in Bikhchandani and Sharma (2001) andCipriani and Guarino (2014).

Box 2.5. Measuring Herding


18/34



institutional investors have limited redemption pres-

sures and that they allocate assets following long-terminvestment strategies.19Mutual funds are also more likely to engage in

return chasing, creating more procyclical ows. Tesignicant, positive coefficient in Figure 2.8, panel 2,for recipient economies asset returns indicates that

19For instance,Financial imes (2014a) reports intense redemp-tions by retail investors in January 2014.

bond mutual funds and, to a lesser extent, equitymutual funds favor countries with high recent returns.Such momentum trading amplies cyclical swings ofportfolio ows and can be destabilizing. Institutionalinvestors, on the other hand, do not engage in this

type of behavior.Institutional investors, however, are not always morestable: they pull back more strongly from bond mar-kets than do mutual funds when faced with extremeshocks (Figure 2.9). Institutional investors bond owsdropped more appreciably than those of mutual fundsafter the September 2008 Lehman Brothers shock,although their ows have been more resilient thanmutual fund ows and have even grown during otherepisodes of distress. Moreover, institutional investorsreduced their bond exposure more persistently thandid mutual funds when a country was downgraded to

below investment grade. One factor in this behavior isthat institutional investors typically face tighter limitson the ratings of the securities they can hold thando mutual funds (Box 2.3). In contrast, institutionalinvestors equity allocations were broadly unaffected bysovereign downgrades or the Lehman Brothers shock.Tese investors typically do not change their invest-ment strategies frequently, irrespectively of short-termmarket uctuations. However, once a strategy to shiftaway from certain emerging markets is adopted, theeffects can be persistent.

Hedge funds have shown a mix of behaviors during

stress episodes. In principle, hedge funds can trade in adestabilizing manner. But they can also behave as con-trarians and thus smooth market turbulences becausethey face fewer portfolio restrictions (see Ilyina, 2006).Hedge funds, especially those that are leveraged, pulledback substantially during the global nancial crisis,although they maintained their exposures in 2013(Figure 2.9). Moreover, market participants suggestthat some hedge funds are becoming similar to mutualfunds in terms of transparency and investment strate-gies because many of them now serve more conserva-tive institutional investors, such as pension funds. Among dedicated emerging market hedge funds, about40 percent are leveraged, down from about 50 percentin 2008.

Flows from Different ypes of Mutual Funds

Different types of mutual funds show distinctive sensi-tivities to global nancial shocks. Te key results from

45

40

35

30

25

20

15

10

5

0

Institutionalbond

Mutual fundbond

Institutionalequity

Mutual fundequity

Institutionalbond

Mutual fundbond

Institutionalequity

Mutual fundequity

0.0000.0010.0020.0030.0040.0050.0060.0070.0080.0090.010

Signi cant Not s igni cant

1. Sensitivity of Portfolio Flows to Emerging Markets to the VIX by Type ofInvestors and Assets(Changes of ows when the VIX increases by one standard deviation;percent of standard deviation of ows)

2. Evidence for Momentum Trading(Estimated coefcient on lagged country index returns)

Figure 2.8. Mutual Fund and Institutional Investor Flows

Sources: Bank of New York Mellon; EPFR Global; and IMF staff estimates.Note: ICRG =International Country Risk Guide; VIX = Chicago Board of OptionsExchange Market Volatility Index. Each type of monthly ow is regressed on the

VIX, the lagged change in the ICRG country risk index, the real interest ratedifferential, and the lagged country index return. Estimation period is December2000October 2013. One standard deviation of the VIX is about 8 percentagepoints. See Annex 2.1 for the details of method, Table 2.4 for sample economies,and Table 2.5 for the denition of variables.


19/34



Figure 2.9. Cumulative Monthly Portfolio Flows to Emerging Markets from Different Types of Investors duringDistress Episodes

Institutional investors continued to add money to emergingmarkets while retail investors pulled back in 2013

1. May 2013 Sell-Off(May 2013 = 100)

2. European Sovereign Debt Crisis(July 2011 = 100)

3. Global Financial Crisis(September 2008 = 100)

4. Earlier Emerging Market Crises(Event month = 100) 1

5. Emerging Market Sovereign Downgrade Events 2 6. Dedicated Emerging Market Hedge Funds around Recent Global Crises(Event month = 100) 3

and during the European sovereign debt crisis

but institutional investors withdrew more than retail investors frombonds after the Lehman Brothers shock. Institutional investor ows were more resilient and often continued torise during past emerging market crises.

However, institutional investors withdrew more persistently thanretail-oriented mutual funds from sovereign bonds downgradedbelow investment grade.

Hedge funds, especially leveraged ones, pulled back from emergingmarkets in 2008 and to a lesser extent in 2011, but their exposuresremained unchanged in 2013.

Sources: Bank of New York Mellon (BNY); EP FR Global; Eurekahedge; and IMF staff calculations.1See Annex 2.1 for the denition of EPFR retail and EPFR institutional investors.2The data are average ows for ve episodes in which sovereigns were downgraded to below investment grade between 2000 and 2013 when institutionalinvestor data are available: Croatia, 2012; Egypt, 2002; Hungary, 2011; Latvia, 2009; and Romania, 20 08.3Flows are estimated by adjusting changes in assets under management with returns at the fund level. In distress periods, funds tend to increase their cashholdings, so the outows from emerging markets may then be even greater.

60

70

80

90

100

110

120

130

654321 0 1 2 3 4 5 6 654321 0 1 2 3 4 5 6Equity (May 2013 = 0) Bond (May 2013 = 0)

EPFR retailBNY institutional

60

70

80

90

100

110

May 2013 sell-off(May 2013 = 0)

European debt crisis(July 2011 = 0)

Global nancial crisis(September 2008 = 0)

LeveragedWithout leverage

6 4 2 0 2 4 6 6 4 2 0 2 4 6 6 4 2 0 2 40.4

0.3

0.2

0.1

0.0

0.1

16141210

8642

024

6420 2 4 6 810121416186420 2 4 6 81012141618Equity

(Month from event date = 0)Bond

(Month from event date = 0)

P e r c e n t o

f r e c i p i e n t e c o n o m y

G D P

P e r c e n t o

f m u t u a l

f u n d

a s s e t s u n d e r m a n a g e m e n t

EPFR mutual funds (left scale)BNY institutional investors (right scale)

60

70

80

90

100

110

120

130

6 4 2 0 2 4 6 6 4 2 0 2 4 6 6 4 2 0 2 4 6 Asian crisis

(April 1997 = 0)Russian crisis

(July 1998 = 0)Brazil crisis

(January 1999 = 0)

EPFR retailEPFR institutional

60

70

80

90

100

110

120

130

65432 1 0 1 2 3 4 5 6 65432 1 0 1 2 3 4 5 6

Equity (September 2008 = 0) Bond (September 2008 = 0)


60

70

80

90

100

110

120

130

65432 1 0 1 2 3 4 5 6 654321 0 1 2 3 4 5 6

Equity (July 2011 = 0) Bond (July 2011 = 0)



20/34



a comparison of their ows sensitivity to the VIX(Figure 2.10, panels 1 and 3) are as follows:20 For fixed-income funds, active funds that are not

exchange-traded funds (E Fs) (mostly activelymanaged open-end funds, which are the majorityof mutual funds; see Box 2.1), are more sensitive toglobal financial conditions than are passive E Fs.

20Te differences across various types of mutual funds highlightedhere are statistically signicant. Te results are generally robust whensubsamples (before and after the global nancial crisis) are used; when multiple fund characteristics are examined at the same time;and when alternative global factors (such as the ED spread and the volatility of the U.S. federal funds futures rate) are used.

Closed-end funds, especially for equity, seem to beless reactive to global financial conditions. his sug-gests that redemption pressures by funds ultimate

investors play an important role in mutual fundinvestment strategies. News reports around the JanuaryFebruary 2014 volatility episode are in line with this interpretation (Financial imes, 2014b).

Global funds are more stable sources of capital flows.21 he evidence suggests that this may be because they

21Tis is contrary to the perception that crossover funds (those notdedicated to emerging market assets but that opportunistically investin them) are more return sensitive and volatile. Te average share of

0.3 0.2 0.1 0 0.1

Small

Large

Active, non-ET F

Passive, ETF

Closed-end

U.S.

Offshore

GlobalEM regional

Small

Large

Active, non-ET F

Passive, ETF

Closed-end

U.S.

Offshore

Global

EM regional

Geography

Domicile

Structure

Size

More sensitive

0.10 0.08 0.06 0.04 0.02 0

Geography

Domicile

Structure

Size

Directeffect

of the VIX

Directeffect

of the VIX

More sensitive

0 10 20

Active, non -ETF

Passive, ETF

Closed-end

U.S.

Offshore

EM regionalGeography

Domicile

Structure

Global

20 10

20 10 0 10 20

Active, non-ET F

Passive, ETF

Closed-end

U.S.

Offshore

EM regionalGeography

Domicile

Structure

Global

1. Sensitivity of Flows from Bond Funds to the VIX(The signicant coefcient on the interaction term of the VIX and afund characteristic plus the signicant coefcient on the VIX)

3. Sensitivity of Flows from E quity Funds to the VIX(The signicant coefcient on the interaction term of the VIX a fundcharacteristic plus the signicant coefcient on the VIX)

2. Changes in the Share in Total Bond Funds between 2005 and 2 013(Percentage points)

4. Changes in the Share in Total Equity Funds between 2005 and 2013(Percentage points)

Source: IMF staff estimates.Note: EM= emerging market; ETF = exchange-traded funds; VIX = Chicago Board of Options Exchange Market Volatility Index. Panels 1 and 3 summarize the resultsof panel regressions with country-fund xed effects. A value of 1 means that a 1 percentage-point increase in the VIX reduces ows by 1 percent of assets undermanagement per month. The dependent variables are monthly equity and bond ows of individual funds into individual countries as a proportion of the funds assetsallocated to the country. The independent variables consist of the VIX, its interaction with one dummy variable representing a fund characteristic, and controlvariables. All independent variables are demeaned. Estimation periods are December 2003September 2013 for bond funds and March 1996October 2013 forequity funds. See Annex 2.1 for details, Table 2.4 for sample economies, and Table 2.5 for denitions of variables. Box 2.1 provides the denition of fund characteris-tics. Active non-ETF funds are mostly actively managed open-end funds. Panels 1 and 3 show the sum of the two coefcients only when they are signicant at the 5percent level. Panels 2 and 4 do not show the share of small and large funds because size indicators are relative to the total sample.

Figure 2.10. Flow Sensitivity to Global Financial Conditions by Fund Characteristics


21/34


22/34



portfolio ows, the focus is on differences in the reac-tion to global nancial conditions, here depending onthe degree of nancial sector deepening. In our study,most of the dimensions of nancial deepening are asso-ciated with a lower sensitivity to global shocks for equitymarkets as well as for markets of bonds denominated inforeign or local currencies; the results for the exchangerate market are somewhat weaker ( able 2.2).26

Having a larger local investor base has a stabilizingeffect. A larger nancial sector (banks and nonbankssuch as mutual funds, pension funds, and insurancecompanies) signicantly helps limit the effects of globalnancial shocks. Moreover, these effects are quantita-tively large (Figure 2.12). Some of the effects of a largerlocal investor base are sufficient to offset the unfavorabledirect impact from the increase in the VIX. Tese resultsare consistent with the literature stressing the counter-

26Tese results are generally robust when the estimation sample isseparated into the periods before and after the global nancial crisis.

cyclical nature of capital ows of domestic investors (seeBroner and others, 2013; and IMF, 2013c).27

Similarly, capital market development generallylowers the sensitivity of asset returns to global nancialconditions. A higher stock market capitalization con-tributes to the stability of bond, equity, and currencymarkets. Large and liquid stock markets also mitigatethe sensitivity of equity returns to global nancialconditions. Similarly, bond markets with higher liquid-ity (that is, with lower bond bid-ask spreads) are lessreactive to VIX shocks.

Terefore, the recent decline in liquidity in someemerging markets appears to have contributed to mak-ing local bond yields more sensitive to the VIX in thesemarkets (Figure 2.13). Market participants attribute thisto reduced market making by global banks operating inemerging markets. Tis could be partly due to tighter

27 A robustness check showed little evidence of nonlinear effects(when returns are large or small and when they are negative orpositive.)

Table 2.2. Role of Financial Deepening in Dampening the Impact of Global Financial Shocks on Asset Prices(Estimated coefficients on the interaction terms of the VIX and respective financial development measure)

Equity Excess ReturnsForeign Currency

Sovereign Bond SpreadsLocal Currency Sovereign

Bond Yields Currency Excess ReturnsExpected

Sign EstimateExpected



Sign Estimate

Financial DepthBank Assets + 0.001** 0.001*** 0.002*** + 0.001**Nonbank Financial Institution Assets + 0.001*** 0.000 0.001*** + 0.019*Domestic Bonds + 0.000 0.001 0.002*** + 0.001**Stock Market Capitalization + 0.001** 0.001*** 0.001*** + 0.001***

Investor BaseMutual Fund Assets + 0.003** 0.000 0.003** + 0.000Insurance Company Assets + 0.005*** 0.004*** 0.005*** + 0.002Pension Fund Assets + 0.004*** 0.001*** 0.001*** + 0.003**

Market LiquidityStock Market Total Value Traded + 0.001*** 0.001*** 0.002*** + 0.001**Bond Bid-Ask Spreads 0.463 + 0.614*** + 0.467*** 0.559

Debt StructureOriginal Sin Index nil 0.050 + 0.268*** + 0.159*** 0.118Foreign Holdings of Sovereign Debt 0.233*** + 0.098* + 0.209*** 0.232**Foreign Share in LC Sovereign Debt 0.447* + 0.100 + 0.202*** 0.644

Institutional QualityRule of Law + 0.044** 0.083*** 0.088*** + 0.011Accounting Standards + 0.077*** 0.036*** 0.125*** + 0.066**Transparency of Government Policy + 0.122*** 0.054 0.150*** + 0.077

Source: IMF staff estimates.Note: LC = local currency; TED = Treasuries and Eurodollar; VIX = Chicago Board Options Exchange S&P 500 Implied Volatility Index. Equity returns are on a dollar bacurrency sovereign bond yields are without hedging. Each model is estimated using country fixed effects; global factors as controls (including the TED spread, credit spreaspread, and global market returns); country-specific factors (including dividend yield differentials, interest differentials, currency returns, exchange rate regimes, sovereign rating, and the forecasts of GDP growth, inflation, and the current account balance, depending on asset types); a global risk factor (a measure of foreign exchange risk for excess returns and the VIX for all others) and the global risk factors interaction with one measure of financial deepening, debt structure, or institutional quality at a time. Eperiods are May 1995August 2013 for equities; May 1995August 2013 for foreign currency sovereign bonds; January 2001August 2013 for local currency bonds; and August 2013 for currencies. Results are robust to other measures of financial deepening (including bank private credit, nonbank private credit, stock market turnover, and btraded) and institutions (quality of government regulation, control of corruption, government effectiveness, and political stability, among others). See Annex 2.1 for detailsfor sample economies, and Table 2.5 for definitions of variables. ***, **, and * indicate statistical significance at the 1, 5, and 10 percent levels, respectively.


23/34



regulation set by supervisors in their home countries andalso to changes in bank business models following theglobal nancial crisis.28 Market participants believe thatlocal banks have helped ll this liquidity gap somewhat,but not fully. Expanding local institutional investors cancreate demand in primary markets but do not necessar-

28See Chapter 1 of the October 2013 GFSR for details.

ily help improve secondary market liquidity, as they tendto buy and hold.

Overcoming original sin has reduced the sensi-tivity to global factors of both foreign and domesticcurrency bond prices. In principle, reducing the shareof foreign currency debt in total external debt lessensissuers credit risk, thereby reducing the price sensitiv-ity to VIX shocks.

20 10 0 10 20

Bank assetsDomestic bonds

Stock market capitalizationMutual fund assets

Insurance assetsPension assets

Stock value tradedBond bid-ask spreads

Original sin indexForeign holdings of sov. debt

Rule of law Accounting standards

Signicant Not sign icant VIX

Depth

Liquidity

LocalInvestor

LocalInvestor

Debt structure

Institutions

More sensitive Less sensitive

Direct effect of the VIX 1

201001020

Bank assetsDomestic bondsStock market capitalizationMutual fund assetsInsurance assetsPension assetsStock value tradedBond bid-ask spreadsOriginal sin indexForeign holdings of sov. debtRule of law

Accounting standa rds

Depth

Liquidity

Local Investor

Institutions

Debt structure



Bank assetsDomestic bondsStock market capitalizationMutual fund assetsInsurance assetsPension assetsStock value tradedBond bid-ask spreadsOriginal sin index

Foreign holdings of sov. debtRule of law

Accounting standa rds

Depth

Liquidity

Local Investor

Institutions

Debt structure


1001020 20

Bank assetsDomestic bonds

Stock market capitalizationMutual fund assets

Insurance assetsPension assets

Stock value tradedBond bid-ask spreads

Original sin index

Foreign holdings of sov. debtRule of law

Accounting standards

Depth

Liquidity

Debt structure

Institutions



20020 10 10


1. Equity Excess Returns over U.S. Treasury Yield

(Annualized; percentage points)

3. Local Currency Sovereign Bond Yields(Basis points)

2. Foreign Currency Sovereign Bond Spreads over U.S. Treasury Yield

(Basis points)

4. Currency Excess Returns over Interest Rate Differential(Annualized; percentage points)

Source: IMF staff estimates.Note: VIX = Chicago Board of Options Exchange Market Volatility Index. With the estimation results presented in Table 2.2, the panels illustrate how much a countrycan mitigate the negative effect of a 10 percentage point increase in the VIX by having more developed (deeper) domestic nancial systems or better institutions. Forexample, the effect of a 10 percentage point shock to the VIX on stock prices for economies with the largest 75th percentile ratio of stock market capitalization toGDP is 4 percentage points smaller than those with the lowest 25th percentile ratio of market capitalization to GDP (panel 1, stock market capitalization bar).Percentile data are taken from the whole country-month sample from 2005 to 2013 (some variables, such as insurance and pension fund assets, start for most of thecountries only around 200305). For bond bid-ask spreads and original sin index, the best 75 percent means those with lowest 25th percentile values. See Table2.4 for sample economies and Table 2.5 for denitions of variables.1The direct effect of the VIX (red bar) is the average effect of a 10 percentage point increase of the VIX, without controlling for the level of nancial development orinstitutional quality.

Figure 2.12. The Effects of Financial Deepening on the Sensitivities of Asset Returns to Global Risk Factor(Estimated coefcients on the interaction terms of the VIX and one nancial deepening variable best 75th worst 25th percentile of the nancialdeepening variable 10 percentage point change in the VIX)


24/34


24 International Mo

c025b97a5b6a2388ef7b7ecdda1e286d

Documents