Download L21-23. - Harvard Kennedy School

(VII) INTERNATIONAL INTEGRATION
OF FINANCIAL MARKETS
LECTURES 20 - 22
Question 1: What are the arguments
in favor of open financial markets?
Question 2:
Does it really work this way?
Question 3: How integrated are financial markets,
and what are the remaining barriers?
Advantages of financial opening
• For a successfully-developing country,
with high return to domestic capital,
investment can be financed more cheaply by
borrowing abroad than out of domestic saving alone.
• Symmetrically, investors in rich countries can earn
a higher return on their saving by investing in
emerging markets than they could domestically.
• Households can smooth consumption over time.
• In the presence of uncertainty, investors
can diversify away some risks.
Classic gains from trade
future
wine
In autarky, Portugal can only
consume what it produces.
•
Under free trade,
Portugal responds to
new relative prices by
shifting into wine,
where it has a
comparative
advantage….
(Price mechanism puts it on
full-employment PPF & at the point
maximizing consumers’ utility.)
Textiles are cheaper
on world markets.
•
•
…Portuguese consumption in
textiles rises, which
it imports, thereby reaching a
higher indifference curve.
textiles
today
Next, we do the gains from trade again, substituting
period 0 & period 1, in place of wine & textiles.
Intertemporal optimization
We will maximize the intertemporal utility function:
u[C0] +βu[C1] where C0 ≡ consumption today; C1 ≡ consumption tomorrow;
u'(C) > 0; u''(C) < 0;
β ≡ subjective discount factor, reflecting patience. 0<β<1.
Total resources available = Y0 +
1
Y
1+𝑟 1.
where Y0 ≡ income today; Y1 ≡ income tomorrow; r ≡ real interest rate.
Total spending discounted to today = C0 +
1
1+𝑟
C1 .
Budget constraint: C1 = (1+r)(Y0-C0 ) + Y1.
Intertemporal utility subject to budget constraint:
u[C0] + β u[(1+r)(Y0-C0 ) + Y1]
To maximize, differentiate with respect to C0:
 Euler equation: u'[C0] + β u'[C1](-(1+r)) = 0.

u'[C0]/u'[C1] = β (1+r)
A simple functional form for intertemporal utility
Let’s try the case of log utility: log[C0] + β log[C1]
(a special case of iso-elastic utility functions)
Then Euler equation u'[C0]/u'[C1] = β (1+r)
becomes [1/C0]/[1/C1] = β(1+r).
C1
=>
= β (1+r).
C0
Result: Agents choose higher consumption tomorrow
than today if r is high and/or they are patient.
Aggregating up, gives a theory to determine the interest rate:
C1
1+r = / β => r > 0 if opportunities grow over time
C0
and people are impatient.
Welfare gains from open
capital markets:
The intertemporal optimization
theory of the current account
1. Even without intertemporal
reallocation of output, Y0 & Y1,
consumers are better off
(borrowing from abroad
to smooth consumption).
2. In addition, firms
can borrow abroad
to finance investment.
WTP, 2007
Prof.J.Frankel
The intertemporal-optimization theory of the current account,
and welfare gains from international borrowing
1. Financial opening with fixed output
High interest rate encourages
agents to postpone consumption.
future
Y1
Assume interest rates in the
outside world are closer to 0
● than they were at home
●
=> domestic residents borrow
from abroad, so that they
can consume more in Period 0.
(the slope of the line
is closer to -1.0).
=> C0↑
Y0
Source: Caves, Frankel & Jones (2007) Chapter 21.5, World Trade & Payments, 10th ed.
Prof.J.Frankel
Welfare is
higher at
point B.
today
The intertemporal optimization theory of the current account,
and welfare gains from international borrowing, continued
2. Financial opening with elastic output
Assume interest rates in the outside
world are closer to 0 than
future
they were at home.
●
Shift production
from Period 0 to 1,
and yet consume
more in Period 0,
thanks to foreign
capital flows.
●
●
today
Welfare is higher
at point C.
Source: Caves, Frankel & Jones (2007) Chapter 21.5, World Trade &Payments.
Does this theory ever
work in practice?
•
Norway discovered
NorthSea oil in 1970s.
It temporarily ran
a large CA deficit,
to finance investment
}
(while the oil fields
were being developed)
•
& to finance consumption
(as was rational,
since Norwegians knew
they would be richer
in the future).
}
ITF220 Prof.J.Frankel
Subsequently,
Norway ran big
CA surpluses.
Effect when countries open their
stock markets to foreign investors,
on cost of capital.
Peter Henry (2007)
“Capital Account Liberalization:
Theory, Evidence, and Speculation,“
JEL, 45(4): 887-935.
Liberalization occurs in “Year 0.”
Cost of capital falls,
on average.
Effect when countries open their
stock markets to foreign investors,
on investment.
Peter Henry (2007)
“Capital Account Liberalization:
Theory, Evidence, and Speculation,“
JEL, 45(4): 887-935.
Liberalization occurs in “Year 0.”
Investment rises,
on average.
Indications that financial markets
do not always work as advertised
1) The Lucas Paradox
2) Pro-cyclical capital flows
3) Crises
Indications that financial markets do not always work as advertised
1) The Lucas paradox:
• Capital flows do not systematically go from rich
countries (high K/L) to poor (low K/L).
– Robert Lucas (1990), “Why Doesn’t Capital Flow from Rich
to Poor Countries?” AER.
– Capital “flows uphill.”
• Possible explanation: In many developing countries
investors cannot reap the potential returns to capital
due to inferior institutions, especially inadequate
protection of property rights.
• -- Alfaro, Kalemli-Ozcan & Volosovych (2008).
Indications that financial markets do not always work as advertised
2) Pro-cyclicality:
• Capital flows tend to be pro-cyclical, not counter-cyclical.
– E.g., Kaminsky, Reinhart & Végh (2005) “When it rains, it pours.”
• Possible explanations: In developing countries,
•
• (i) given imperfect creditworthiness, investors require
collateral, e.g., tangible foreign exchange earnings.
The value of the collateral is higher in booms than busts.
• (ii) Fluctuations that appear cyclical, in truth may signal
changes in long-run growth prospects.
• -- Aguiar & Gopinath (2007).
Indications that financial markets do not always work as advertised
3) Crises
• Debt crises, currency crises, banking crises
 The 1982 international debt crisis;
 1992-93 crisis in the European Exchange Rate Mechanism;
 EM currency crashes of the late 1990s:
1994-95 Mexico;
1997 E.Asia, esp. Thailand, Korea & Indonesia;
1998 Russia, 2000 Turkey, 2001 Argentina, 2002 Uruguay.
 2008-2015
 2008-09 GFC (U.S. & U.K.: “North Atlantic Financial Crisis” !)
 Iceland, Hungary, Latvia, Ukraine, Pakistan…;
 The 2010-15 euro crisis (Greece, Ireland, Portugal, Spain, Cyprus…).
Indications that financial markets do not always work as advertised, cont.
• Do investors punish countries when and only when
governments follow bad policies?
Large inflows often give way suddenly to large
outflows, with little news appearing in between
to explain the change in sentiment.
Contagion sometimes spreads to countries that are
unrelated, or where fundamentals appear stronger.
Recessions have been so big, it seems hard
to argue that the system works well.
Empirical studies of financial openness
and economic performance,
reviewed by Kose, Prasad, Rogoff & Wei (2009),
often find little systematic relationship, in either direction.
Some studies find that financial openness is helpful only
if countries have already attained an adequate level of:
• income -- Biscarri, Edwards, & Perez de Grarcia (2003);
Klein & Olivei (1999); Edwards (2001); Martin & Rey (2002);
Ranciere, Tornell & Westermann (2008);
• financial depth, institutional quality & other reforms
-- Kaminsky & Schmukler (2003); Chinn & Ito (2002); Klein (2003);
Obstfeld (2009); Kose, Prasad & Taylor (2009); Wei & Wu (2002);
Prasad, Rajan & Subramanian (2007).
• Or macroeconomic discipline.
-- Arteta, Eichengreen & Wyplosz (2001).
=> Conventional wisdom regarding sequencing:
it is better to liberalize financial markets only
after other reforms have been put in place.
-- McKinnon (1993), Edwards (1984, 2008), and Kaminsky & Schmukler (2003).
Measuring
International
Financial
Integration
I. Direct measures
of barriers,
19702004
e.g., IMF’s count of freedom
from KA restrictions.
II. “Price tests”
III. “Quantity tests”
e.g., intl. assets+liabilities/GDP
Source: Kose, Prasad, Rogoff & Wei (2009)
I. Direct Measure of Financial Barriers:
Chinn-Ito tally of capital controls, from IMF data
Figure 1: Development of KAOPEN for Different I ncome Groups
2
3
Development of KAOPEN for different income groups
-1
0
1
Rapid financial
liberalization in
1990s
1970
1980
1990
year
Industrial Countries
Emerging Markets
2000
Less Developed
Menzie Chinn & Hiro Ito, "A New Measure of Financial Openness"
(Journal of Comparative Policy Analysis, 2008), updated July 2010
http://web.pdx.edu/~ito/Chinn-Ito_website.htm.
2010
Chinn-Ito Measure of Financial Openness
The calculations are based on 4 categories in the IMF’s Annual Report on Exchange
Arrangements & Exchange Restrictions: multiple exchange rates, current account
restrictions, capital account restrictions, and required surrender of export proceeds.
Measuring International Financial Integration, cont.
II. “Price” tests
1.Uniform price of an asset across markets
E.g., gap between China’s A shares and off-shore shares.
2. Interest rate parity (IRP)
i) Covered interest parity (CIP)
ii) Uncovered interest parity (UIP)
iii) Real interest parity (RIP)
1. Price of the same asset across borders
Chinese firms’ stock prices remain higher onshore than offshore.
Premium of “A shares” (held in Shanghai),
over “H shares” (held in Hong Kong)
}
June 2014: After net inflows turn to outflows, reserves start to fall.
Nov. 17, 2014: Shanghai-HK Stock Connect goes into effect.
Nov. 21: PBoC starts cutting interest rates. Bubble follows.
Summer 2015: Shanghai stock market & RMB begin to fall;
China’s authorities pressure domestic institutions to buy stocks.
Prof.J.Frankel
Source: FT,
Feb.29, 2016
2. Interest Rate Parity:
Why does i not equal i* ?
I. Currency factors
• Expected currency depreciation
• Exchange risk premium
The currency premium can be measured
as the forward discount, or swap rate, or
differential between domestic & local $-linked bonds.
II. Country factors
…
Decomposition of the nominal interest differential
i – i* ≡ country premium + currency premium
e.g.,
≡
( i – i* - fd )
+
fd
fd ≡ (fd - Δse) + (Δse)
exchange + expected
risk
nominal
premium
depreciation
The country premium could be measured by the sovereign spread,
Credit Default Swap, or covered interest differential (i-i*-fd).
The currency premium could be measured by the forward discount (fd),
currency swap rate, or local spread of $-linked vs. domestic-currency bonds.
WHY DOES i NOT EQUAL i* ?
II. Country factors, continued
• Default risk –
• reflected in sovereign spreads or Credit Default Swaps
• Capital controls –
• reflected in covered interest differentials
• Taxes on cross-border investments
• Transaction costs
• Imperfect information
• Risk of future capital controls
Sovereign spreads
Brazilian interest rate decomposed
country premium + currency premium + LIBOR
}
}
Total spread (Brazil rate minus LIBOR) =
Currency premium (forward premium) + Country premium (spread)
1995-98
Sovereign spreads
Mexican spread decomposed: currency premium + country premium
2004-13
Total spread for Mexican sovereign bonds
over US Treasury bill interest rate
Currency premium
≡ pesos/$ swap rate
Country premium ≡ total spread
adjusted for currency premium
Total spread over US T bill rate
}
Country premium
Currency swap rate
Wenxin Du & Jesse Schreger, “Sovereign Risk, Currency Risk & Corporate Balance Sheets,” Oct. 2014
Sovereign spreads, 2003-06
650
Spreads were low for Emerging Market bonds in 2006,
and even lower for South Africa.
550
EMBI+
450
350
250
EMBI+
RSA EMBI+
150
50
2Jun03
}
30- 26- 26- 28- 26- 25- 23- 21- 19- 20- 21- 18- 18- 14- 14- 13- 15- 12- 10Jul- Sep- Nov- Jan- Mar- May- Jul- Sep- Nov- Jan- Mar- May- Jul- Sep- Nov- Jan- Mar- May- Jul03
03
03
04
04
04
04
04
04
05
05
05
05
05
05
06
06
06
06
Global investors were under-pricing risk
-- as also reflected in US corporate spreads, options prices, etc.
All of them shot back up in 2008.
Sovereign spreads for 5 euro countries
shot up in the 1st half of 2010
Snapshot of spot & forward exchange rates on Aug. 14, 2014
Spot
rate
Transaction
cost
Forward rates
*
*
* Value of ₤ and € are shown in terms of $. All other exchange rates are local per $.
Selling at
a forward
discount
vs. the $
Selling at
a forward
premium
Spot and forward exchange rates on Aug. 14, 2014, continued
Spot
rate
Transaction
cost
Forward rates
Selling at
a forward
discount
vs. the $
$ pegs
COVERED INTEREST PARITY
( 1 + iTurkey )
=
(1/S) ( 1 + iUS ) F
where S is the spot rate in TL/$ and F is the forward rate.
Forward discount fd  (F-S)/S
=> 1 + fd  F/S
=>
(1 + iTurkey ) = (1 + fd) (1 + iUS).
= (1 + fd + iUS + fd iUS).
Because (fd iUS) is small, iTurkey ≈ fd + iUS .
=> If the Turkish nominal interest rate exceeds the U.S. rate,
then the lira sells at a discount in the forward exchange market.
Liberalization in a country that had controls on capitalinflows.
Domestic & offshore interest rates,
Germany, 1973-74
}
From: Marston (1989)
Liberalization in a country that had controls on outflows.
France, 1973-1993: Domestic and Offshore Interest Rates
{
From: Mussa and Goldstein,(1993).
France kept its controls on capital outflows until the late 1980s.
They produced an offshore-onshore differential, which
shot up whenever there was speculation of a franc devaluation.
Again, the differential disappeared after controls were removed.
ITF220 Prof.J.Frankel
In late 2008 Covered Interest Parity surprisingly failed,
in the Global Financial Crisis rush to the $ as safe haven.
Covered interest differentials, using Overnight Index Swap interest rates, 2003-2011
Significant determinants are apparently counterparty risk & liquidity,
proxied by financial stock CDS, VIX, implied fx volatility, OIS bid-ask spreads & Fed swap lines.
Inês Isabel Sequeira de Freitas Serra, ”Covered Interest Parity,” NOVA – School of Business & Economics, Lisbon, Jan. 2012
http://run.unl.pt/handle/10362/9528
THREE INTEREST RATE PARITY CONDITIONS
Investors decide
whether to hold:
Arbitrage=>
parity
Does it hold
condition.
in practice?
CIP
$ deposits in New i$NY - i£L =
Covered York vs. covered £
fd.
interest parity deposits in London
Yes, if default risk
& capital controls
arelow & liquidity high.
$ deposits in NY vs. i$NY - i£L =
UIP
Uncovered £ deposits in
Δse
interest parity London uncovered.
If risk is
unimportant.
Hard to tell in
practice.
Real
Arbitrage is not
RIP
interest parity directly relevant
i$NY - i£L =
e
e
πUS - πUK
No, not in short
run.
Summary of Interest Rate Parity conditions
to be used in L24-26: Exchange Rate Models
Covered interest parity
i – i* = fd
+
No risk premium
fd =
Δse
}
=>
Uncovered interest parity
i – i* = Δse,
+
Ex ante Relative
Purchasing Power Parity
Δse = πe – π*e
=>
i – i* = πe – π*e .
Real interest parity
}
III QUANTITY TESTS: some show rising integration
IMF
Quantity tests
point to surprisingly low international integration
1. Home bias in portfolios:
Do citizens of each country hold a basket of assets
that is optimally diversified internationally?
No
2. Consumption risk-sharing:
Are countries’ consumption levels correlated
with each other more than country incomes?
No
3. Feldstein-Horioka test:
Do countries’ Investment rates vary independently
of their National Saving rates?
No
Feldstein-Horioka test of capital mobility
Regression:
(I/GDP) = α + β (NS/GDP) + v.
Feldstein (1980) argued that if capital
were perfectly mobile, we would find β = 0:
countries with good investment opportunities
could borrow abroad to finance them.
Instead, β was much closer to 1:
Countries are apparently savings-constrained.
The Feldstein-Horioka, still as high as 0.7
in the 1980s, declined in the 90s and until 2007.
Kristin Forbes, “Financial “deglobalization”?: Capital flows, banks, and the Beatles,” Bank of England, 18 Nov., 2014
Appendices: Country risk
• Appendix 1: Inter-shuffling of credit-worthiness
between advanced & developing countries
– Recent credit rating rankings
– The end of “original sin”?
• Appendix 2: EM Sovereign Spreads
– More examples
– “Risk on – risk off”
Appendix 1: The blurring of lines between debt
of advanced countries and developing countries
• 1) Since the crisis of the euro periphery began in Greece
in 2010, we have become aware that “advanced”
countries also have sovereign default risk.
• 2) After 2000, Emerging Market Countries increasingly
became able to borrow in their own currencies, so their
debt carries currency risk (not just default risk).
1) Country creditworthiness became inter-shuffled
“Advanced” countries EM & “Developing” countries
AAA Germany, UK
Singapore, Hong Kong
AA+
US, France
AA
Belgium
Chile
AAJapan
China
A+
Korea
A
Malaysia, South Africa
ABrazil, Thailand, Botswana
BBB+ Ireland, Italy, Spain
BBB- Iceland
Colombia, India
BB+
Indonesia, Philippines
BB Portugal
Costa Rica, Jordan
B
Burkina Faso
SD
Greece
S&P ratings, 2012
Spreads for Italy, Greece, & other Mediterranean members
of € were near zero, from 2001 until 2008,
and then shot up in 2010.
Market Nighshift Nov. 16, 2011
46
2) The end of Original Sin?
After 2000, Emerging Markets successfully issued more debt
in their own local currencies (LC), instead of $-denominated (FC).
Fig. 2 from Jesse Schreger & Wenxin Du
“Local Currency Sovereign Risk,” HU, March 2013
Turkey is able to borrow in local currency (lira),
but has to pay a high currency premium to do so.
{
Total premium on
Turkey’s lira debt
over US treasuries
Pure default risk premium on lira debt
Fig. 5 from Schreger & Du, “Local Currency Sovereign Risk,” HU, March 2013
{
Appendix 2: EM sovereign spreads
Spreads shot up in 1990s crises
EMBI, 1994-2001
Sovereign spreads
Sovereign spreads
Sovereign spreads on South African Dollar Debt
Downtrend in SA country risk premium,
to below 100 basis points by 2006,
in tandem with upgrades by rating agencies
Source: SA Treasury
700.00
S&P Upgrade (BB+ to BBB-)
S&P Upgrade (BBB- to BBB)
Moody's upgrade
(Baa3 to Baa2)
S&P Upgrade (BBB to BBB+)
600.00
Moody's upgrade
(Baa2 to Baa1)
500.00
400.00
300.00
200.00
100.00
Global 06
Global 09
Global 14
Global 17
Global 12
1996-2006
6/15/2006
2/15/2006
10/15/2005
6/15/2005
2/15/2005
10/15/2004
6/15/2004
2/15/2004
10/15/2003
6/15/2003
2/15/2003
10/15/2002
6/15/2002
2/15/2002
10/15/2001
6/15/2001
2/15/2001
10/15/2000
6/15/2000
2/15/2000
10/15/1999
6/15/1999
2/15/1999
10/15/1998
6/15/1998
2/15/1998
10/15/1997
6/15/1997
2/15/1997
10/15/1996
-
EM sovereign spreads
Sovereign spreads
Spreads fell to low levels by 2007.
WesternAsset.com
Sovereign spreads
EM sovereign spreads
Spreads rose again
in Sept. 2008,
• especially on $denominated debt
Bpblogspot.com
• particularly in
Eastern Europe.
World Bank
What determines spreads?
Sovereign spreads
EMBI is correlated with risk perceptions
risk off
“risk on”
Data sources: Bloomberg and Federal Reserve
Laura Jaramillo & Catalina Michelle Tejada, IMF Working Paper, March 2011