Jonathan D. Ostry*Research Department, IMF

United Nations Conference on Capital FlowsRio De Janeiro

August 23, 2011

*The views expressed in this presentation are those of the presenter and do not necessarily represent those of the IMF or IMF policy. This presentation draws on joint work with Atish Ghosh, Karl Habermeier, Luc Laeven, Marcos Chamon, Mahvash Qureshi, and Annamaria Kokenyne.

MANAGING CAPITAL INFLOWS: WHAT TOOLS TO USE?

There is a tendency to regard foreign exchange controls, or any interference with the free movement of funds as, ipso facto, bad ... [but] there are times when it is in the best economic interest of a country to impose restrictions on movements of capital…[and] there are periods when failure to impose controls…have led to serious economic disruption.

The task before us is not to prohibit instruments of control but to develop those measures of control, those policies of administering such control, as will be the most effective in obtaining the objectives of world-wide sustained prosperity

HARRY DEXTER WHITE

The advocacy of a control of capital movements must not be taken to mean that the era of international investment should be brought to an end. On the contrary, the system contemplated should greatly facilitate the restoration of international loans and credits for legitimate purposes.

JOHN MAYNARD KEYNES

Capital Controls: Not Just An EME Phenomenon

33

0.0

0.2

0.4

0.6

0.8

1.0

1960-1969 1970-1979 1980-1989 1990-1999 2000-2006

Capital Account Restrictiveness in Selected Advanced Economies 1960-2006

Canada France Germany UK US

0.0

0.2

0.4

0.6

0.8

1.0

1960 1965 1970 1975 1980 1985 1990 1995 2000 2005

Canada France Germany United Kingdom United States

UK: Purchases of foreign securities by UK residents required foreign exchange bought at a premium

Canada: Interest payments on inward financial credits, and inter -bank loans and deposits subject to a withholding tax

Germany: Ban on non-resident purchases of German bonds introduced in 1972 (and lifted in 1974). Discriminatory reserve requirements employed

France: "Devise-titre” system: residents couldonly reinvest the proceeds of sales of previously held foreign securities; could not increase their foreign asset holdings

Capital Account Restrictiveness in Advanced Economies: Some Examples 1960-2006

USA: Interest Equalization Tax (IET) on purchases of foreign, fixed interest securities; tax was abandoned in 1974 soon after the adoption of floating exchange rates

Note: Capital account restrictiveness index (0=highly restrictive; 1=fully liberalized).Source: Quinn and Toyoda (2008).

Note: Capital account restrictiveness index (0=highly restrictive; 1=fully liberalized).Source: Quinn and Toyoda (2008), and OECD (1982).

4

Context

44

4

Capital Inflows: Recovery or Historic Surge?

5

Net Quarterly Capital Flows into EMEs, 2006Q1-11Q1 (billions of US dollars)

Net Annual Capital Flows into EMEs,

2001-2016 (billions of US dollars)

-100

0

100

200

300

400

500

600

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

Emerging Europe Emerging Asia Excl. ChinaLatin America OtherNet Capital Flows WEO Projections 5

-250

-150

-50

50

150

250

2006

Q1

2006

Q2

2006

Q3

2006

Q4

2007

Q1

2007

Q2

2007

Q3

2007

Q4

2008

Q1

2008

Q2

2008

Q3

2008

Q4

2009

Q1

2009

Q2

2009

Q3

2009

Q4

2010

Q1

2010

Q2

2010

Q3

2010

Q4

2011

Q1

Emerging Europe Emerging AsiaLatin America OtherNet Capital Flows

Source: International Financial Statistics.

6

Strong Fundamentals in EMEs

Source: IMF’s WEO database.Note: Average year-on-year growth (in percent).

Annual Real GDP Growth (in percent) Public Debt to GDP (in percent)

Source: IMF’s WEO database.Note: Average gross general government debt to GDP ratio (in percent).

35

55

75

95

115

135

2005 2007 2009 2011 2013 2015

BRICSEMEsG7

Projection

-4

-2

0

2

4

6

8

10

2005 2007 2009 2011 2013 2015

BRICSEMEsG7

Projection

7

The Search for Yield…

Gold

GBI-EM Global div

US High Grade

US High Yield

EMBIG

CEMBI Broad

ELMI+

EM equities

Commodities

UST

S&P 500

136.4

54.6

42.5

40.2

40.0

36.0

30.1

10.2

9.9

1.7

-10.5

Returns on Assets (in percent)

Source: J.P. Morgan.Note: Returns as of June 29, 2007 to May 31, 2011 (in percent).

Interest Rate Differential (in basis points)

Source: Bloomberg.Note: 10-year government bond yield minus 3-month US T-bill rate in basis points.

Jan-0

7

May-0

7

Sep-0

7

Jan-0

8

May-0

8

Sep-0

8

Jan-0

9

May-0

9

Sep-0

9

Jan-1

0

May-1

0

Sep-1

0

Jan-1

1

May-1

1

-300

-100

100

300

500

700

900

1100

1300 Brazil

South Africa

Korea

China

Mexico

8

Improved Sovereign Credit Ratings

Sovereign Rating Upgrades & DowngradesEMBI Global Average Credit Rating

EMBIG Rating

BBB-/Baa3

BB+/Ba1

BB/Ba2

BBB/Baa2

BB-/Ba3

Jan-94 Jan-96 Jan-98 Jan-00 Jan-02 Jan-04 Jan-06 Jan-08 Jan-10

EMBIG Moody’s

EMBIG S&P

EMBIG Avg

Source: Chang (2011).Note: EM fixed income indices are now investment grade. (> BBB-).

2007 2008 2009 2010 2011YTD

1 0 0 0 00

7

10 11 12

35

19

14

28

65

28 28

4

14

Developed Up Developed Down

Emerging Up Emerging Down

Source: Chang (2011).Note: the total number of upgrades and downgrades includes both S&P and Moody’s actions. Last developed market sovereign rating upgrade occurred in 2007.

Is it Push or Pull Factors?

9

Large literature on determinants of capital flows to EMEs identifies: Push factors—Interest rate in advanced economies, commodity

prices, investor risk appetite, regional contagion Pull factors—Macroeconomic performance, trade and financial

openness, institutional quality, external vulnerability Ghosh et al. (2011) examine inflow surges to EMEs over 1980-2009

Surges are highly synchronized across regions; but Considerable heterogeneity in the amount of inflow in a surge Both push and pull factors matter—but in different ways

For surge occurrence, push factors matter (notably, the real US interest rate; international market uncertainty)

For surge magnitude, domestic conditions are important (e.g. macroeconomic performance, trade & financial openness, exchange rate flexibility)

10

Inflation and Credit Growth: Selected Cases

Magnitude of Net Inflows1

Maximum Net Inflow2

Composition of Gross Inflows3 Inflation4 Real Credit

Growth5

(In percent of GDP) (In percent of GDP) (In percent; y/y) (In percent; y/y)

(2009Q1-2011Q1) (2009Q1-2011Q1) (2009M1-2011M4) (2009M1-2011M2)

Brazil 8.16 14.61 5.14 10.35(2010Q4)

Indonesia 5.95 12.18 5.25 8.78(2009Q3)

Korea 7.69 15.06 3.08 1.86(2009Q3)

Peru 7.28 12.50 2.30 10.78(2009Q4)

South Africa 6.02 13.02 5.46 -0.68(2010Q3)

Thailand 9.06 19.00 1.53 4.84(2010Q4)

Turkey 3.28 11.35 6.74 13.39(2010Q4)

Source: IMF's IFS, INS, and WEO databases.1/ Average net f inancial f low to GDP (in percent). For Peru and Thailand, data end in 2010Q2 and 2010Q4, respectively.

3/ Composition of gross inf low s in 2010.4/ Average year-on-year inf lation over 2009M1 to 2011M4.5/ Average year-on-year real credit grow th over 2009M1 to 2011M2.

Red=Portfolio Yellow=Other

Investment Green=FDI

2/ Maximum net f inancial f low to GDP (in percent) in 2009Q1 and 2011Q1. For Peru and Thailand, data end in 2010Q2 and 2010Q4, respectively. Quarters in parentheses refer to the quarter in w hich net capital inf low w as the largest.

11

Are New Bubbles Emerging in EMs?

Note: Non-weighted averages of the real house price index. 2007Q3 is set to equal 100. Source: OECD, Global Property Data, Haver Analytics and national sources.

Note: Non-weighted averages of the annual growth of real private credit. (in percent). The group of “other emerging economies” lies below the 75th percentile of the distribution of the 2010Q1-2010Q4 average of the annual growth of real domestic credit to the private sector.Source: IMF IFS.

Real Credit to the Private Sector

-5

0

5

10

15

20

2006

Q2

2006

Q3

2006

Q4

2007

Q1

2007

Q2

2007

Q3

2007

Q4

2008

Q1

2008

Q2

2008

Q3

2008

Q4

2009

Q1

2009

Q2

2009

Q3

2009

Q4

2010

Q1

2010

Q2

2010

Q3

2010

Q4

2011

Q1

Top Quartile (Argentina, Bolivia, Brazil, China, Hong Kong, Indonesia, Sri Lanka, Turkey)

Other Emerging Economies

60

70

80

90

100

110

120

130

2006

Q1

2006

Q2

2006

Q3

2006

Q4

2007

Q1

2007

Q2

2007

Q3

2007

Q4

2008

Q1

2008

Q2

2008

Q3

2008

Q4

2009

Q1

2009

Q2

2009

Q3

2009

Q4

2010

Q1

2010

Q2

2010

Q3

2010

Q4

Top Quartile (China, Columbia, Hong Kong,

Israel, Malaysia )

Other EmergingEconomies

Central EasternEuropean Countries

Real House Prices

12

Policy Responses to Capital Inflows

Currency Appreciation1

Reserve Increase2 Policy Rate3 Fiscal

Tightening4Prudential Policies/

Capital Controls

(In percent) (In percentage points) (Structural Balance)(2009Q1-2011Q1) (2008Q4-2010Q4)

Brazil 29.91 2.10 Raised No Yes

Indonesia 16.41 3.58 Raised Yes Yes

Korea 15.84 7.90 Raised Yes Yes

Peru 4.06 4.40 Raised No Yes

South Africa 30.30 -0.05 Lowered Yes No*

Thailand 3.06 12.44 Raised No Yes

Turkey -2.32 1.64 Lowered No Yes

Source: IMF's INS and WEO databases, and national sources.1/ Cumulative percentage change in NEER from 2009Q1 to 2011Q1.2/ Change in reserves to GDP ratio over end-2008 to end-2010.3/ Monetary policy is the change in policy rates over 2009Q3 to 2011Q1.4/ Fiscal policy is the change in cyclically adjusted f iscal stance betw een 2009 and 2010.

*South Africa has liberalized capital controls on outf low s in response to the surge in capital inf low s.

13

Capital Controls, Macroeconomic and Prudential Risks

13

When are Capital Controls Appropriate?

14

IMF staff (Ostry et al., Feb. 2010) argued that capital controls appropriate for inclusion in the policy toolkit to address: Macroeconomic risks, when

Currency overvalued Further reserve accumulation undesirable Inflation/overheating concerns Limited scope for fiscal tightening

Financial-stability risks, when Prudential framework still leaves high risk of financial fragility

14

Key Questions to be Addressed

15

How macroeconomic and prudential rationales for capital controls fit together?

What are the main elements of the policy toolkit (once macro-policy space is exhausted)?

What combination of prudential measures and controls should be deployed to address inflow-induced risks?

How should capital controls be designed?

Ostry et al. (2011) examine:

15

How do Macro and Prudential Concerns Fit Together?

16

Prudential policies: Strengthen/introdu

ce prudential measures

Macroeconomic concerns

Financial-stability risks

Macro policies: exchange rate

appreciation, reserves accumulation, fiscal and

monetary policy mix

Impose/intensify capital controls (or measures that act like them) subject

to multilateral considerations and macro tests

Primary responses

Macro policy options

exhausted? Residual risks?

Capital inflow surge

16

How do Macro and Prudential Concerns Fit Together?

17

Both macroeconomic and prudential considerations suggest that capital controls are appropriate

No real conflict—but possible design issues Macro considerations say yes, but prudential ones

say no No conflict of principle, but again possible conflict of design Controls as transitional measure given macro policy

implementation lags? Macro considerations say no, prudential ones say yes

Genuine conflict Multilaterally-consistent approach implies the bar is much higher

for the use of capital controls—especially broad-based controls Exhaust the available macro policy space and allow exchange rate

appreciation before tightening capital controls on inflows for prudential risks

17

18

The Policy Toolkit

18

What’s in the Toolbox?

19

FX-related prudential measures Discriminate according to the currency, not the residency, of the flow Applied to regulated financial institutions, primarily banks Examples: limits on banks’ open FX position (as a proportion of their

capital), and limits on FX lending by domestic banks (or higher capital requirements)

Other prudential measures Reduce systemic risk without discriminating based on

residency/currency Examples: LTV ratios, limits on credit growth and sectoral lending,

dynamic loan-loss provisions, and counter-cyclical capital requirements Capital controls

Discriminate between residents and non-residents in cross-border capital movements (OECD Code of Liberalization of Capital Movements, 2009)

Economy-wide or sector specific (usually the financial sector) or industry specific

Cover all flows, or target specific types (debt, equity, FDI; short vs. long-term)

Examples: taxes, URRs, licensing requirements, and outright limits or bans

19

How Common are the Measures?

2020

21

•Tax on equity and bond inflows (Brazil)•Fee on NR purchases of central bank paper (Peru)•Reserve requirements on NR deposits (Peru)

Capital controls

•Reserve requirements on foreign currency deposits (Peru)

•Limits on banks FX derivative positions in percent of bank capital (Korea)

•Capital requirements for FX loans (Peru)•Limits on banks net open FX positions (Peru)•Limits on ratio of banks FX loans and securities to FX

borrowing (Korea)

FX-related measures•Reserve requirements for local currency deposits (Brazil,

Turkey)•LTV ratios (Korea, Peru, Thailand, Turkey)•Levy on interest from consumer loans (Turkey)•Capital requirements for specific loans (Brazil)

Other prudential measures

Recent Examples of Measures

Issues in Classifying Instruments

22

De jure prudential tools may operate like capital controls A regulation differentiating based on the currency of denomination may

operate like a capital control to the degree that most FX liabilities are to nonresidents

A measure that requires banks to pay a tax on their non-core liabilities could well in practice operate just like a capital control if most of the funding that banks receive comes from abroad

A regulation discouraging FX lending to unhedged borrowers may act as a capital control (reduce inflow) or prudential measure (change currency composition of foreign liabilities). Difficult to tell at implementation stage

De jure capital controls may have primarily prudential intent (e.g. differential reserve requirements by residence of liability)

Fine line between FX-related and other prudential measures (e.g. differential LTV ratio by currency of denomination)

22

23

Alternative Classification

Capital Flow Management Measures (CFMs)—measures designed to influence capital flows

Residency-based—commonly referred to as capital controls Other—measures that do not discriminate on the basis of residency,

but are nonetheless designed to influence capital inflows (including a subset of prudential measures that discriminate on the basis of currency)

Non-CFMs—structural and prudential policies not designed to influence capital flows. Include measures that do not discriminate by residency and typically, but not always, do not differentiate by currency

24

Matching Risks and Tools

24

25

Ceilings on banks’ foreign derivative positions/Capital

controls on banks (esp. short-term debt), e.g., taxes/reserve

requirements

Open FX limits/higher capital requirements on

loans to unhedged borrowers

Cyclical capital requirements, LTV limits

Legal or other

impediments to capital controls?

FX-related prudential

Capital controls

Fragile external liability structure (maturity

mismatch/sudden-stop risk)

Currency risk (due to open FX position) or credit risk

(due to unhedged borrower)

Credit boom/asset price bubble

FX-related prudential1/

Other prudential

Flows to domestic banks

Concerns about

access to finance/

distortions?

FX-related prudential/

Capital controls1/

1/ Once macro policy space exhausted, and taking due account of multilateral considerations.

Choice of Instruments: Flows Intermediated through the Financial Sector

25

Direct flows or through unregulated financial

sector

Fragile external liability structure (debt,

especially short-term)

Currency risk (due to lack of natural or financial hedge)

Asset price bubble

Capital controls1/

Capital controls to discourage debt

instruments

Capital controls to discourage FX borrowing

by unhedged entitiesBroad-based capital

controls

Capital controls1/

Capital controls1/

Borrower-based FX-measures

Legal or other

impediments to capital

controls?

261/ Once macro policy space exhausted, and taking due account of multilateral considerations

Choice of Instruments: Flows Not Intermediated through the Financial Sector

Exceptions to Flow Chart

27

Playing field for access to credit of large firms vs. SMEs

Prudential regulations may cause flows to be intermediated through the unregulated financial sector (e.g. Croatia)

- Extend the perimeter of regulation? Not easy in short run

- Regulatory arbitrage more likely in countries with weak supervision, sophisticated financial institutions, and deep capital markets

International obligations may prohibit or constrain the use of capital controls (e.g., the EU treaty, the GATS, the OECD code, or various bilateral investment treaties)

27

Policy Measures and Financial Stability Risks: Some Suggestive Evidence

28

Empirical Evidence: External Liability Structure

Debt in proportion to total external liabilities Cross section (2007; measures in 2003-05, 38 countries)

Economy-wide, and financial sector capital controls significantly associated with lower debt (controlling for external vulnerability)

FX regulations significantly associated with lower debt (though FX regulations lose significance when included together with economy-wide capital controls)

Moving from 25th to 75th percentile of economy-wide capital controls or FX-regulation lowers debt share by almost 10 ppt

Panel data (1995-2008) Economy-wide, and financial sector capital controls significantly

associated with lower debt External vulnerability index, region and time effects, and per capita

income included as additional regressors29

Capital controls and FX-related prudential measures associated with smaller proportionof debt in external liabilities …

30

External Liability Structure: Cross-Sectional Evidence*

Source: Authors’ estimates.*Sample: 38 EMEs over 2003-07. Debt liabilities is the residual (including constant) obtained after regressing the share of debt liabilities in total external liabilities in 2007 (in percent) on a (lagged) composite external vulnerability index.

Debt Liabilities

0

10

20

30

40

50

60Below mean indexAbove mean index

Economywide capital controls

Financial sector capital controls

Forexregulations

Macroprudential measures

***** *

Deb

tlia

bilit

ies

in t

otal

ext

erna

l lia

bilii

ties

(in p

erce

nt)

Empirical Evidence: FX Lending

Foreign currency loans in total domestic credit Cross section (2007; measures in 2003-05, 28 countries)

Economy-wide, and financial sector capital controls significantly associated with lower proportion of FX credit (controls: initial private credit ratio; exchange rate regime)

FX regulations significantly associated with lower FX credit Both economy-wide capital controls and FX regulations significant

when included together Moving from 25th to 75th percentile of economy-wide capital controls

and FX regulations lowers proportion of FX credit by 20-25 ppt Panel data (1995-2008)

Results similar to those above Private credit (lagged, in % of GDP); exchange rate regime; per capita

income; institutional quality index; region and time dummies included as additional regressors 31

0

5

10

15

20

25Below mean indexAbove mean index

Fore

xcr

edit t

o G

DP

(in p

erce

nt)

Economy wide capital controls

Financial sector capital controls

Forexregulations

Macroprudential measures

Forex credit

*** ***

* Capital controls and FX-related prudential measures associated with lower foreign currency denominated lending by domestic banks

32

FX Lending: Cross-Sectional Evidence*

Source: Authors’ estimates.*Sample: 28 EMEs over 2003-07. Forex credit is the residual (including constant) obtained after regressing forex credit to GDP in 2007 on private credit to GDP in 2005 and a binary variable (=1) if fixed exchange rate regime in place.

Empirical Evidence: Credit Booms

Change in private sector credit to GDP ratio Cross section (change in credit ratio 2003-07; measures in

2000-02, 28 countries) Prudential regulations significantly associated with smaller credit booms

(controls: real per capita income in 2005; political stability index; financial market development; exchange rate regime; and credit bureaus)

Prudential regulations significant if capital controls/FX regulations included

Moving from 25th to 75th percentile of prudential regulations lowers credit growth (03-07) by 1.5 percent per year

Panel data (1995-2008) Results similar to those above Region and time dummies, per capita income, political stability index,

stock market capitalization, exchange rate regime and credit bureaus included as additional regressors

33

-1

1

3

5

7

9

11

13

15

17 Below mean indexAbove mean index

Economy wide capital controls

Financial sector capital controls

Forexregulations

Macroprudential measures

Domestic private credit boom

Ch

ange

in p

rivat

e cr

edit

to G

DP

(in

per

cent

age

poi

nts)

**

Other prudential measures associated with lower lendingbooms by domestic banks

34

Credit Booms: Cross-Sectional Evidence*

Source: Authors’ estimates.*Sample: 28 EMEs over 2003-07. Private credit boom is the residual (including constant) obtained after regressing change in private credit to GDP over 2003-07 on private credit to GDP in 2003.

Empirical Evidence: Crisis Resilience

If policy measures reduce vulnerabilities through previous channels, then downturn in event of crisis should be smaller

Cross-section (change in growth 2008-09 relative to average 2003-07, 41 countries)

Economy-wide capital controls, FX and domestic prudential regulations associated with smaller decline (controlling for growth in trading partners, change in terms of trade)

Both economy-wide capital controls and domestic prudential regulations retain significance when included together

But economy-wide capital controls dominate when included with FX regulations

Moving from 25th to 75th percentile of economy-wide controls or FX regulations lowers growth decline by 3.5 and 2.5 ppt, respectively

Past Crises(1995-2008) Capital controls associated with smaller growth decline in past crises

35

Capital controls, FX-related andother prudential measuresassociated with greater crisis resilience

-4.5

-4.0

-3.5

-3.0

-2.5

-2.0

-1.5

-1.0

-0.5

0.0

Below mean indexAbove mean index

Cris

is g

row

thde

clin

e (in

per

cent

age

poin

ts)

**

**

Economy wide capital controls

Financial sector capital controls

Forex regulations

Macropudentialmeasures

Crisis Resilience

Crisis Resilience: Recent Crisis*

36

Source: Authors’ estimates.*Sample: 41 EMEs over 2003-07. Crisis resilience is the residual (including constant) obtained after regressing the difference between real GDP growth rates averaged over 2008-09 and 2003-07 on trading partner growth and terms of trade change.

Robustness

Numerous robustness tests Additional regressors to capture political stability,

financial market development, and overall regulatory quality

Alternative indices (use first principal component of sub-indices instead of averages)

Endogeneity—use bilateral investment treaty with US (which generally prohibit use of capital controls) as instrument for capital control

Key results are unaffected or strengthened37

38

Designing Capital Control Instruments

38

Designing Capital Controls: Some Considerations

39

Broad principles Effective: achieve intended aim; not easily circumvented

Efficient: minimize distortions and scope for non-transparent/arbitrary enforcement

But a number of questions… Permanent or temporary inflow?− Macroeconomic concerns: Controls for temporary, not permanent inflows− Prudential concerns: Controls could be imposed for persistent flows Broad-based or targeted controls?− Macroeconomic concerns: Broad based possibly with limited exemptions− Prudential concerns: Targeted but taking account of circumvention possibilities Price or quantity-based controls?− Macro concerns: Price-based measures easier to adjust cyclically, and simpler to administer− Prudential concerns: Quantitative measures more appropriate when authorities face

information asymmetries/uncertainty about private sector’s response Other considerations: Administrative and institutional capacity 39

40

Conclusions

40

Key Takeaways

41

Macro and prudential policies can go a long way to deal with inflow surges Use and strengthen orthodox toolkit before resorting to capital controls

There is strength in numbers—no measure is likely to work perfectly, so diversify and use more than one

Capital controls and prudential measures should target specific risks Prudential measures main instrument when flows are intermediated through the

banking sector Capital controls main instrument when flows by-pass the banking sector

In designing capital controls, Macro concerns imply broad and price-based controls for temporary surges Prudential concerns imply targeted on specific risks and possibly administrative

capital-control measures, even in case of persistent inflows Design should reflect administrative inheritance/apparatus 41