If Notes PDF
If Notes PDF
LECTURE NOTES ON
I N T E R N AT I O N A L F I N A N C E
T Y P E S E T B Y V U T. C H A U
This document contains lecture notes from ECON 2530B - International Finance, taught by Professor Gita
Gopinath at Harvard University, Fall 2015. Any error found in this document is presumably ours.
Contents
2 Introduction 15
10 Portfolio Bias 35
Part I
Non-neutrality of Nominal
Exchange Rate
1 Non-neutrality of Nominal Exchange Rate
1.1 Definitions
Definition (NER). The nominal exchange rate Ein,t is the price of country i’s
currency in terms of country n’s currency.
Definition (RER). The real exchange rate RERin,t is the price of country i’s
consumption basket in terms of country n’s consumption basket. Effectively, this
means:
P
RERin,t = i · Ein,t
Pn
Remark:
2. The behavior of RER over time depends much on what prices are used
to calculate RER. Common choices are CPI (consumer price index), PPI
(producer price index), and IPI (import, or also called border prices).
3. We can also rewrite the definition in log form:
rerin,t = pi − pn + ein,t
The concept of neutrality of nominal exchange rate is the idea that fluc-
tuations in relative national price ratio (pi − pn ) will be fully compensated
for by fluctuations in the NER, leaving the RER unchanged. For example,
higher inflation in country i should be accompanied by a depreciation of
the NER to allow for the real relative price of consumption baskets between
the two countries unchanged. This is also called the relative purchasing
power parity (relative PPP) hypothesis:
However, [?] found serious deviation from PPP, and in fact the RER
fluctuates a lot with long half-lives,2 which suggests non-neutrality of 2
Her actual finding was that the RER
of country with fixed exchange rate
NER.
fluctuates much less compared to RER
Finally, for the next section, we need to define exchange rate pass- of country with floating exchange rate.
through (ERPT), a concept that talks about sensitivity of price to move-
ments of the nominal exchange rate:
8 lecture notes on international finance
Here are some stylized facts about NER and RER (from 3 ): 3
Stylized Fact 1.2.1 (Deviation from PPP). CPI RER co-moves closely with NER
at short- and medium-horizons. The persistence of these RERs is large with long
half-lives.
Stylized Fact 1.2.2 (RER for tradable goods). Movements in the RER for
tradable goods are roughly as large as overall CPI RER when calculated using CPI
or PPI, but much smaller when using border prices.
cpi
The correlation between ∆rertr and ∆rer cpi are close to 1, but this corre-
lation is only 0.3 when using border prices. That is, relative prices at the
consumer level co-move more closely with the NER and are more volatile
than when using border prices.
Stylized Fact 1.2.3 (ERPT). ERPT into consumer prices is lower than into border
prices. ERPT into border prices is typically incomplete.
Stylized Fact 1.2.4. Border prices, in whatever currency they are set in, respond
partially to exchange rate shocks at most horizons, even when conditional on a
price change.4 4
‘Respond partially’ here means that
the elasticity is less than 1. ‘Conditional
Stylized Fact 1.2.5 (International Price System). A large fraction of ex- on a price change’ means we limit our
sample to only goods that have had
ports/imports around the world nowadays are priced in dollar (even when the a price change. We do so out of the
US is not involved in the trade). Countries with higher share of imports invoived concern that the results might have
in a foreign currency have higher short-run and long-run pass through. been biased down to goods that never
change prices at all.
Think about three countries: US, Japan, and Turkey. The US and Japan
are developed markets, while Turkey is an emerging market. ERPT into im-
port prices for the US is about .3, while it is above .8 for Japan and Turkey.
non-neutrality of nominal exchange rate 9
This says that the price of a good imported from country i into coun-
try n (expressed in country n’s currency), pin , is a markup exp(µin ) over
marginal cost exp(mcin ).6 The markup then is a function on the relative 6
Every variable is expressed in log
forms.
price ratio between the import price and the general price index in country
n. Intuitively, if the imported price is too high compared to the general
price level in country n, markup cannot be too high. The marginal cost
of production in country i depends on the quantity of exports qin , factor
price of production (here is labor) in country i (here is wage wi ), and the
nominal exchange rate between the two countries, since this marginal cost
is expressed in local currency of country n.
Log-differentiating (1.1) gives us
where:
−∂µin
Γin ≡ : Elasticity of markup to relative price
pin − pn
∂mcin
mcq ≡ : Sensitivity of marginal cost on quantity imported
∂qin
∂mcin
αin ≡ : Sensitivity of marginal cost (in destination’s currency) to NER
∂ein
Denote Φin = mcq ein as the total effects coming from decreasing returns
to scale. Re-arranging gives us a formula for ERPT:
The direct ERPT is defined as the ERPT when ∆pn = ∆qn = ∆wi = 0,
that is when there is no change in the general price or quantity level in
the sector of imported goods, nor wage in the exporting country. This
allows us to isolate the sole effect of exchange rate on domestic price of
destination country. Regression effectively estimates overall ERPT without
appropriate controls.
It is important to remember the formula for direct ERPT:
∆pin αin
= (1.4)
∆ein 1 + Γin + Φin
(3) Φin : The role of decreasing returns to scale (∆ein → ∆qin → ∆mcin ).
Intuitively, a nominal depreciation of country n’s currency makes
imports into n more expensive, this forces exporters in i to cut back
production. With decreasing returns to scale, this lowers average (and
possibly marginal) cost, making the goods cheaper, working against the
effect of a nominal depreciation. Therefore, the higher Φin , the lower
ERPT.
Ignore currency choice for now and assume LCP (local currency pricing).
The FOC for the reset price p̄in,t is simply the derivative of the fraction
of discounted stream of profits in every period in the future that would
prevail had firms not been able to change the price at all:8 8
Note that the remaining part of
the objective function when firms
∞
are allowed to change price again is
∑ κ l Et Θt+l Π p ( p̄in,t |st+l ) = 0 (1.6) orthogonal to this problem, hence
l =0 derivative of that part would be zero.
We first-order approximate marginal profit state-by-state around the
desired-price in that state:
e pp (t + l )[ p̄in,t − pein,t+l ] + O( p̄in,t − pein,t+l )2
Π p ( p̄in,t |st+l ) = Π (1.7)
Π pp (t + l ) = Π pp (t) + O(kst+l − st k)
See 9 for a complete proof. Re-arranging and ignore high order terms gives 9
us the approximation for optimal reset price (in terms of desired prices):
∞
p̄in,t = (1 − βκ ) ∑ ( βκ )l Et pein,t+l (1.8)
l =0
(2) If the NER is a random walk, then nominal rigidity does not matter,
and the ERPT in this case is the same as flexible price pass-through.
10.1.2 Equilibrium
Optimization: Complete market implies Backus-Smith condition
u0 (C )/u0 (C ∗ ) = k · P/P∗ = k
C
⇒ = k−1/σ
C∗
Intra-temporal choice of X versus Y in the case of Cobb-Douglas gives
x x∗ θ py
= ∗ =
y y (1 − θ ) p x
Now, consider portfolio autarky, i.e. countries cannot smooth out con-
sumption across time and states of nature anymore. However, there can
still be balanced trade:
PC = p x x + py y = p x X, PC ∗ = p x x ∗ + py y∗ = py Y
36 lecture notes on international finance
PC C px X θ yX θ ωY X
∗
= ∗ = = =
PC C py Y 1−θ x Y 1 − θ ωX Y
So:
C θ
=
C∗ 1−θ
and the financial autarky is Pareto efficient!
In this case, even with complete absence of the financial market, each
country’s level of consumption remains efficient, and there is zero gains
from financial liberalization.