Professional Documents
Culture Documents
Chinn Lecture1 PDF
Chinn Lecture1 PDF
Chinn Lecture1 PDF
€ /$
1.1
1.0
0.9
0.8
0.7
1999 2000 2001 2002 2003 2004
1 /E E U
Example: €/¥
.011
.010
.009
€ /¥
.008
.007
1999 2000 2001 2002 2003 2004
E J P /E E U
The Value of the Dollar
(vs. Euro, Yen, and Trade Weighted basket)
1.1
$ in d e x
1.0
(F e d b ro a d )
0.9
0.8
¥ /$
€ /$
0.7
0.6
1999 2000 2001 2002 2003 2004
The institutional features of the
forex market
• Average turnover in April 2004 = USD1.9
trillion per day
• Most trading volume between dealers,
next most with financial institutions, next
most with non-financials.
• USD, EUR, JPY, GBP, SFR most
important in turnover.
• AUD moving up quickly.
Trends in FX Turnover
Definitions of Terms
Spot transaction: single outright transaction involving the
exchange of two currencies at a rate agreed on the date
of the contract for value or delivery.
Outright forward: transaction involving the exchange of
two currencies at a rate agreed on the date of the
contract for value or delivery at some time in the future
Foreign exchange swap: transaction which involves the
actual exchange of two currencies (principal amount
only) on a specific date at a rate agreed at the time of
the conclusion of the contract, and a reverse exchange
of the same two currencies at a date further in the future
at a rate agreed at the time of the contract.
Definition of Terms (cont’d)
Currency swap: contract which commits two
counterparties to exchange streams of interest payments
in different currencies for an agreed period of time and to
exchange principal amounts in different currencies at a
previously agreed exchange rate at maturity.
Currency option: Option contract that gives the right to
buy or sell a currency with another currency at a
specified exchange rate during a specified period.
Forward rate agreement (FRA): interest rate forward
contract in which the rate to be paid or received on a
specific obligation for a set period of time, beginning at
some time in the future, is determined at contract
initiation.
Definition of Terms (cont’d)
.4 London
.3
NY
.2 E u ro zo n e
c o u n try
c e n te rs
.1 T okyo
Z urich
.0
1990 1992 1994 1996 1998 2000 2002 2004
Central bank holdings of reserve currencies
.8
.7
USD
.6
.5
.4
.3
DEM EUR
.2
.1 JPY
GBP
.0 S FR
1975 1980 1985 1990 1995 2000
II
∆st = ∆p t − ∆p *t + ∆µ t
*
Relative PPP in rates
q = s− p+ p = µ
*
Equivalently, test to see if the
real exchange rate is
constant
Both are typically rejected, so
Instantaneous/SR relative
PPP rejected.
Relative PPP
• Empirical literature in the mid-1980’s to
1990’s moved to testing for long run
relative PPP
• Advent of unit root literature
• Subsequent cointegration literature
• If q reverts to its mean, then relative PPP
holds in the long run.
Econometrics of PPP: I
q t − q t −1 = (ϕ − 1)q t −1 + u t
q t = φq t −1 + u t ; φ < 1
∆q t = βq t −1 + u t ; β = ϕ − 1
Econometrics of PPP: II
• Cointegration: Test whether s, p, p* exhibit
a long run relationship, using:
- Engle-Granger (many)
- Johansen (many)
- Horvath-Watson (Chinn, 1998)
• Panel unit root test (Frankel and Rose)
• Panel cointegration test (Pedroni)
• Threshold autoregression (Obstfeld &
Taylor)
What’s the consensus?
• There is mean reversion.
• It is faster than Rogoff’s (1995) stylized
fact of 3-5 years for half life.
• It can be even be faster w/TAR
• But it is still slower than can be
rationalized by sticky prices.
• Absolute PPP cannot be found except for
country subgroups.
III
f t,t+k = s e
+ η t,t+k . Def’n of risk
t,t+k
premium
∆s e
t,t+k = ( it,k - i ) - η t,t+k , Combining
*
t,k
Testing for unbiasedness
st+k = s ret,t+k + ξ t,t+k , Rational exp.
Approach
-2 *
**
***
-4 /
1994- ***
*** 2000 ***
-6
**
***
-8
Can. Fr. Ger. Ita. J ap. UK Pooled
Some competing explanations
• Risk premium (but the “risk premium” isn’t
correlated with anything). See Engel and
Frankel.
• More exotic risk premia (consumption
CAPM, w/ or w/o habit formation, keeping
up with Joneses, loss aversion). See
Bekaert, Hodrick and Marshall.
Some more competing
explanations (econometric)
• Inappropriate coefficient restrictions induce
size distortions.
st + 1 = α + βf t ,1 + ε t + 1
st + 1 − st = α + βf t ,1 − st + ε t + 1
st + 1 − st = α + β ( f t ,1 − st ) + ε t + 1
Econometric issues (cont’d)
• Hence Fama regression makes sense if
β=1 exactly.
• However, if β≈1 and the forward discount
is highly serially correlated (or fractionally
integrated) then may reject (McNown and
Wallace; Zivot; Elliott; Baillie and
Bollerslev).
• Solution: Estimate cointegrating
relationship. (But why does β≠1?)
Yet more econometrics
• Extreme support: examine only large
deviations (Huisman, Kool and Nessen;
Flood and Rose)
• Thresholds/bands of inaction: (Sarno and
Taylor).
What about the long horizon?
Ex Post Depreciation and 5-Year Government Bond Yields:
1980-2004
^ ^ Reject
" $ H0 : $ = 1 Adj-R2 N
DEM -0.005 0.902 0.07 100
(0.010) (0.532)
GBP 0.001 0.515 0.02 100
(0.009) (0.311)
CAD -0.007 0.512 0.02 100
(0.005) (0.332)
Constr. ... 0.709 0.08 300
Panel (0.404)
Long horizon (cont’d)
Ex Post Depreciation and 10-Year Government Bond Yields:
1980-2004
^ ^ Reject
" $ H0 : $ = 1 Adj-R2 N
.2 One year .2
interest
.1 differential
GRDEP1Y
.1
.0 .0
-.1 -.1
-.2 -.2
-.3 -.3
80 82 84 86 88 90 92 94 96 98 00 -.08 -.04 .00 .04 .08
GR1YDIFLAG4
$/DM, 10 year (1980-2000)
.10
.10
DM/$ 10 year
.08 depreciation .08
.06 .06
.04 .04
GRDEP10Y
.02 .02
.00 .00
/
0.8 U n b ia s e d n e s s
0.67 0.68
c o e ffic ie n t va lu e
0.4
0.09
0.0
-0.4
-0.54
-0.8 -0.76 -0.76
1.5
+ /- 1 s .e .
1.4
1.3
U S D /E u ro
1.2 e xc h a n g e
ra te (e o p )
1.1
1.0
0.9
0.8
1999 2000 2001 2002 2003 2004 2005
Source: FX4casts
$/¥
.0104
+ /- 1 s .e .
.0100
.0096
U S D /Y en
.0092
.0088
.0084
.0080
.0076
1999 2000 2001 2002 2003 2004 2005
Source: FX4casts
Consensus, IRP, historical
correlations ($/€)
1.5
C onsensus
1.4 \
1.3 /
IR P
1.2 and
interest
1.1 ra te
U SD /E uro imp lie d
1.0
0.9
0.8
1999 2000 2001 2002 2003 2004 2005
Consensus, IRP and historical
correlations ($/¥)
.0104
C onsensus
IR P implied \
.0100
\
.0096
USD/Yen
.0092
/
in te re s t
.0088 rate implied
.0084
.0080
.0076
1999 2000 2001 2002 2003 2004 2005
UIP regressions: 3 mo.
Constrained Unconstr. Unconstr., EMS
OLS 0.883 0.805 1.214
OLS SE (0.169) (0.209) (0.311)
GMM SE (0.327) (0.382) (0.346)
^ ^ Reject
" $ H0 : $ = 1 Adj-R2N
DEM -0.031 0.219 -0.11 10
(0.010) (0.731)
GBP -0.002 1.613 0.36 10
(0.011) (0.570)
CAD 0.007 0.724 0.13 10
(0.005) (0.378)
Constr. ... 0.737 0.37 30
Panel (0.384)
What about Emerging Markets?
Dates N β (S. E.) t: β=0 t: β=1 DW F Prob
Emerging and Newly Industrialized Economies
1. Czech Republic 12/96-4/04 88 0.4260 0.65 0.76 1.90 0.5206
(0.6604)
2. Hong Kong 12/96-4/04 88 -0.0439 -1.17 768 2.44 0.2468
(0.0376)
3. Hungary 10/97-4/04 78 0.7541 0.60 0.04 1.82 0.5511
(1.2594)
4. India 10/97-4/04 78 -0.6181 -0.72 3.53 1.43 0.4751
(0.8612)
5. Indonesia 12/96-12/02 73 0.1456 0.71 17.28 1.55 0.4807
(0.2055)
6. Kuwait 12/96-4/04 88 0.4050 0.43 0.40 1.89 0.6674
(0.9394)
7. Mexico 12/96-4/04 88 -0.6399 -1.57 16.16 1.99 0.1204
(0.4079)
Emerging Markets (cont’d)
1. Philippines 12/96-4/04 88 1.6770 0.98 0.16 1.87 0.3303
(1.7128)
2. Saudi Arabia 12/96-4/04 88 -0.0831 -1.00 168.17 2.94 0.3223
(0.0835)
3. Singapore 12/96-4/04 88 0.1911 0.15 0.39 1.86 0.8826
(1.2898)
4. South Africa 12/96-4/04 88 -3.2693 -1.78 5.38 1.74 0.0792
(1.8403)
5. Taiwan 12/96-4/04 88 0.1442 0.27 2.65 1.75 0.7842
(0.5252)
6. Thailand 12/96-4/04 88 0.9613 1.40 0.00 1.62 0.1643
(0.6853)
7. Turkey 12/96-4/04 88 -0.0031 -0.11 1241 1.54 0.9133
(0.0284)
Interpretation
• Unbiasedness is just as variable in
emerging markets.
• What seems to matter is type of exchange
rate regime.
• This is not so simple as one might think –
what determines exchange rate regimes
might jointly determine whether
unbiasedness holds.
IV
∆st = π t − π *
t Relative PPP in rates
of change
UIP
∆s e
t ,t + k = i t ,k − i *
t ,t + k
∆s e
t ,t + k −π e
t ,t + k +π *e
t ,t + k = (i t , k − π e
t ,t + k ) − (i *
t ,k −π *e
t ,t + k )