International Macroeconomics: Slides For Chapter 8: International Capital Market Integration

Slides for Chapter 8:
International Capital Market Integration
International Macroeconomics
Schmitt-Grohé Uribe Woodford
Columbia University
May 1, 2016
1
International Macroeconomics, Chapter 8 Schmitt-Grohé, Uribe, Woodford
Idea: Infer the degree of international capital market integration

from savings - investment correlations?
Motivation: In a closed economy those correlations must be 1,

but in an open economy they need not. Thus, if we observe low
correlations, then we can conclude that that there is capital mobility
across countries.
Q: Suppose we find a near perfect correlation between S and I. Does

this necessarily mean that there is no capital mobility?
2
Measuring Capital Mobility I:

Observed Savings-Investment Correlations
1. Across countries
(a) The Feldstein Horioka Regression: 1960-1974
(b) Updates of the Feldstein Horioka Regression: 1960 - 2003
2. Overtime within the same country
(a) The United States, 1928-2015.
(b)
3
Saving and Investment Rates for 16 Industrial Countries

(1960-1974 Averages)
0.4
JAP
0.35 0.035 + 0.887 (S/GDP)→
FIN
0.3
I/GDP
AUSTRIA
AUS
NET
GER
0.25 GRE NEW

SWE
CAN BEL
DEN ITA
IRE
0.2 U.K
USA
0.15
0.15 0.2 0.25 0.3 0.35 0.4
S/GDP
Source: M. Feldstein and C. Horioka, “Domestic Saving and International Capital Flows,” Economic
Journal 90, June 1980, 314-29.
4
Feldstein and Horioka estimated the following equation:
! !
I S
= 0.035 + 0.887 + νi; R2 = 0.91
Q i (0.07) Q i
where (I/Q)i and (S/Q)i are, respectively, the average investment-

to-GDP and savings-to-GDP ratios in country i over the period
1960-74. They have 16 observations. The coefficient on (S/Q)i
is 0.887 with a standard error of 0.07, making it highly unlikely that
the true coefficient is zero.
5
More recent evidence:
Sample of 64 countries with data from 1960 to 2003 find
! !
I S
= constant + 0.52 + νi
Q i (0.06) Q i
where (I/Q)i and (S/Q)i are, respectively, the average investment-

to-GDP and savings-to-GDP ratios in country i over the period
1960-2003.
Source: Yan Bai and Jing Zhang “Solving the Feldstein-Horioka Puzzle with Financial Frictions,”
Econometrica 78, March 2010, 603-632.
6
U.S. National Saving and Investment, 1928-2015

26
Investment/GDP
Savings/GDP
24
22
20
18
Percent of GDP
16
14
12
10
6
1930 1940 1950 1960 1970 1980 1990 2000 2010
Year
Data Source: Bureau of Economic Analysis, www.bea.gov.
corr(s/y, i/y) = 0.98 pre 1972 and 0.76 post 1972
7
But do findings of high savings-investment correlations either across

countries or across time, necessarily imply imperfect capital mobility?
No. Even under perfect capital market integration one could observe
a high S-I correlation.
Give some examples [to be completed]
Therefore, we conclude that high S-I correlations are not evidence

of low capital mobility.
8
Measuring Capital Mobility II:

Interest Rate Differentials
1. Why are interest rate differentials informative about the degree

of capital mobility?
2. Empirical Evidence on the size of interest rate differentials
9
Asset Pricing in a 2-period small open economy

model.
Two important interest rate parity conditions are
• CIRP= covered interest rate parity
• UIRP= uncovered interest rate parity
In this lecture we will derive the CIRP condition and ask under what
conditions the UIRP condition holds.
Let i denote the domestic, nominally risk free, interest rate on

domestic bonds held from period 1 to period 2, and i∗ the foreign,
nominally risk-free, interest rate on foreign currency bonds held from
period 1 to period 2. The domestic currency price of one unit of
foreign currency in period t is denoted St .
10
Notation:
π = probability economy is in the good state in period 2.

1 − π = probability economy is in the bad state in period 2.
Nominal Endowments:
Period 1: Q1
g
Period 2, good state: Q2
Period 2, bad state: Qb2
Consumption:
Period 1: C1
g
Period 2, good state: C2
Period 2, bad state: C2b
11
Notation (ctd.):
Exchange rates:
Period 1: S1 , spot exchange rate
Period 1: F1, forward exchange rate
g
Period 2, good state: S2, spot exchange rate in good state
Period 2, bad state: S2b , spot exchange rate in bad state
Domestic Price Level:

Period 1: P1
g
Period 2, good state, P2
Period 2, bad state, P2b
g
Expectations operator: E1x2 = πx2 +(1−π)xb2 denotes the expected
value of the variable x2 given information in period 1.
12
Covered Interest Rate Parity (CIRP):

F
1 + i = (1 + i∗) 1
S1
Uncovered Interest Rate Parity (UIRP):
∗ S2
1 + i = (1 + i )E1
S1
As we will show below, under free capital mobility, CIRP must always
hold. At the same time uncovered interest rate parity typically fails.
For it to hold it would have to be the case that the forward rate is
equal to the expected future spot rate, that is, it would have to be
true that
F1 = E1S2
But this is not true in the data!
13
Deriving CIRP...
Household buys 3 types of bonds:
• B1 domestic currency bonds, pay interest i

• B1∗ foreign currency bonds, pay interest i∗, and buy forward cover
• B̃1∗ foreign currency bonds, pay interest i∗, without forward cover
14
Households
Expected utility:
g
U = U (C1) + πU (C2) + (1 − π)U (C2b ) (1)
Budget Constraints:
Period 1:
Q1 = P1C1 + B1 + S1 B1∗ + S1B̃1∗ (2)
Period 2, good state:
g g g g
Q2 + (1 + i)B1 + (1 + i∗)F1B1∗ + (1 + i∗)S2 B̃1∗ = P2 C2 (3)
Period 2, bad state:
Qb2 + (1 + i)B1 + (1 + i∗)F1B1∗ + (1 + i∗)S2b B̃1∗ = P2b C2b (4)
15
How to choose B1 , B1∗ , and B̃1∗ ? To maximize utility.
g
Household problem: Pick C1, C2, C2b , B1, B1∗ , and B̃1∗ to maximize
(1) subject to (2)-(4).
To make the problem easier to characterize solve (2) for C1, (3) for
g
C2, and (4) for C2b and use the resulting expressions to eliminate C1,
g
C2, and C2b from (1). Then we have a single objective function in
three unknowns, namely, B1, B1∗ , and B̃1∗ , which we pick to maximize
expected utility.
16
Solving the period-1 budget constraint for C1 yields:

Q1 − B1 − S1B1∗ − S1 B1∗
C1 =
P1
g
Solving the period-2, good state, budget constraint for C2 yields:
g g ∗

∗ ∗
Q2 + (1 + i)B1 + (1 + i ) F1B1 + S2 B1
g
C2 =
P2g
Solving the period-2, bad state, budget constraint for C2b yields:

Q b + (1 + i)B + (1 + i∗) F B ∗ + S b B ∗
b 2 1 1 1 2 1
C2 =
P2b
17
Now take the first order condition w.r.t. B1, this yields:
1 g 1+i 1+i
U 0(C1) = πU 0 (C2) g + (1 − π)U 0 (C2b ) b
P1 P2 P2
Rewrite this expression as:

g
U 0 (C2) P1 U 0(C2b ) P1
" #
1 = (1 + i) π 0 g + (1 − π) 0
U (C1) P2 U (C1) P2b
Letting E1 denote the expectations operator, we have:
U 0(C2) P1
( )
1 = (1 + i)E1
U 0(C1) P2
U 0(C2 ) P1

Finally, let M2 ≡ U 0(C1 ) P2
denote the nominal mrs between period
2 and period 1, to arrive at the following asset pricing condition:
18
1 = (1 + i)E1 {M2 } (5)
And the first-order condition w.r.t. to B1∗ plus forward cover is:
S F F
U 0(C1) 1 = π(1 + i∗)U 0(C2g ) 1g + (1 − π)(1 + i∗)U 0(C2b ) 1b
P1 P2 P2
Rewrite this expression as
0 (C g ) P 0 (C b ) P
" #
F 1 U U
1 = (1 + i∗) π 0 2 g + (1 − π) 0 2 1b
1
S1 U (C1) P2 U (C1) P2
Using the expectations operator notation we have:

U 0(C2) P1
( )
∗ F1 ∗ F1
1 = (1 + i ) E1 = (1 + i ) E1 {M2 } (6)
S1 U 0(C1) P2 S1
19
Combining (5) and (6) we obtain:
(1 + i) = (1 + i∗) F
S
1 (7)
1
which is the covered interest rate parity condition, we had set

out to derive.
20
What about Uncovered Interest Rate Parity?
Consider the FOC w.r.t B̃1∗ .
g b
0 S1 ∗ 0 g S2 ∗ 0 b S2
U (C1) = π(1 + i )U (C2) g + (1 − π)(1 + i )U (C2) b
P1 P2 P2
Rewrite this expression as
g 0
" g b 0 b
#
S U (C ) P S U (C ) P
1 = (1 + i∗) π 2 0 2 1g + (1 − π) 2 0 2 1b
S1 U (C1) P2 S1 U (C1) P2
Using the expectations operator notation we have:

U 0(C2) P1
( ! !) ( ! )
S2 S2
1 = (1 + i∗)E1 = (1 + i∗)E1 M2 (8)
S1 U 0(C1) P2 S1
21
Combining the asset pricing equations (8) and (6) we obtain
F1E1M2 = E1S2 M2
but this expression does in general not imply that the forward rate,
F1, is equal to the expected future spot rate, S2. That is, it does
not follow from here that
F1 = E1S2
Hence, under free capital mobility, uncovered interest rate parity fails
in general.
This was the second result we had set out to show.
22
Now we want to ask under what condition would UIRP hold.
Recall that
cov(a, b) = E (a − E(a)) (b − E(b))

= E (ab) − E(a)E(b)
or
E (ab) = cov(a, b) + E(a)E(b)
We then can express E1M2 (S2 /S1) as

! ! !
S2 S2 S2
E M2 = cov , M2 + E E(M2 )
S1 S1 S1
and rewrite (8) as
" ! ! #
S2 S2
1 = (1 + i∗) cov , M2 + E E(M2)
S1 S1
23
Suppose now that the depreciation rate, S2 /S1, is uncorrelated with
the pricing kernel, M2, that is, assume that
!
S2
cov , M2 =0
S1
then equation (8) becomes

!
S2
1 = (1 + i∗)E1 E1 (M2)
S1
combine this expression with equation (6) to obtain
F1 = E1S1 (9)
If the depreciation rate is uncorrelated with the pricing kernel, M2,

then the forward rate equals the expected future spot rate.
And further if we combine the above expression with (5) we have
S2
n o
(1 + i) = (1 + i∗)E1 S1
or Uncovered Interest Rate Parity holds.

Let’s review what we have shown:
1.) Under free capital mobility, CIRP must hold.
2.) When we observe violations of UIRP, we cannot conclude that

this is evidence against free capital mobility. For even under free
capital mobility UIRP need not hold, that is, the forward rate need
not be the expected future spot rate.
24
Empirical Evidence of Covered Interest Rate Parity
1. CIRD 1982-1988
2. Onshore-Offshore Differentials, 1982 - 1993
3. China
4. Brazil
5. UK - Germany 1870-2000
25
Covered interest rate differentials for selected countries

September 1982-January 1988 (in percent)
F
Covered Interest Rate Differential = (1 + i) − (1 + i∗) .
S
i − i∗ − f s
Mean Std. Dev.
Germany 0.35 0.03
Switzerland 0.42 0.03
Mexico -16.7 1.83
France -1.74 0.32
The covered interest rate differential is measured by the domestic 3-month interest rate minus the
3-month Euro-dollar interest rate minus the forward discount. Source: J. Frankel, “Quantifying
International Capital Mobility in the 1980s.”

26
Observations on the table:

Germany and Switzerland had small CIRD: less than 50 basis points on average.
Thus, Germany and Switzerland appeared to be relatively open to international
capital flows in the early 1980s.
By contrast, Mexico had an enormous negative CIRD of over 16 percent. The

period 1982-1988 corresponds to the post debt crisis period, when the financial
sector in Mexico was nationalized and deposits were frozen. During that period,
investors wanted to take their capital out of Mexico, but were impeded by financial
regulations.
In France barriers to the movement of capital were in place until 1986, which
explains the large average deviations from covered interest rate parity vis-a-vis
the two other industrialized countries shown in the table.
The fact that the country risk premia of France and Mexico are negative indicates
that capital controls were preventing capital from flowing out of these countries.
27
2.) Onshore Offshore Differentials
International capital mobility in the 1990s
28
alternative approach to measuring the degree of capital market

integration.
interest rate differentials between domestic deposit rates and Eurocurrency

deposit rates. For example, compare the interest rate on a French
franc deposit in France to the interest rate on a French franc deposit
outside France, say in London. Since both deposits are in French
francs the exchange rate plays no role in comparing the two interest
rates.
29
Domestic Interbank minus Eurocurrency 3-month interest rates:

(in percent)
1/1/82- 2/1/87- 7/1/90- 6/1/92-

Country 1/31/87 6/30/90 5/31/92 4/30/93
France -2.27 -0.11 0.08 -0.01
Italy -0.50 0.29 0.56 0.36
Germany 0.17 0.05 -0.05 0.07
Japan -0.07 -0.60 0.09 0.17
Source: M. Obstfeld, “International Capital Mobility in the 1990s,” in Kenen, Understanding
Interdependence: The Macroeconomics of the Open Economy, Princeton University Press, 1995,
table 6.1.
30
Observations on the table:
The table provides further evidence suggesting that the presence of capital controls
leads to deviations from covered interest rate parity.
It shows differences between domestic interbank and the corresponding Euro

currency interest rate for France, Italy, Germany, and Japan from 1982 to 1993.
In general, interest rate differentials are lower after 1987.
This is most evident for France, where important capital market deregulation took
place in 1986.
In Italy, the high differential observed between 1990 and 1992 reflects market
fears that capital controls might be imposed to avoid realignment of the lira,
as an attempt to insulate the lira from speculative attacks, like the one that
took place in August/September 1992. These violent speculative attacks, which
affected a number of European economies, particularly, France, Sweden, Italy,
and England, led to exchange rate realignments and a temporary suspension of
the European Exchange Rate Mechanism (ERM) in September 1992. Once the
ERM was reestablished, the lira interest rate differential falls as fears of capital
controls vanish.
Japan had large onshore/offshore differentials between February 1987 and June
1990, which were the result of the Bank of Japan’s heavy use of administrative
guidelines to hold interbank rates below offshore rates.
31
[International Capital Mobility: 1870 to 2000]
32
Forward exchange contracts of the kind common after 1920 were

not prevalent before then. However, there was another widely traded
instrument, called the long bill of exchange. Long bills could be
used to cover the exchange risk that might otherwise be involved
in interest-rate arbitrage. Let bt denote the long bill rate, which is
defined as the date-t dollar price in New York of £1 deliverable in
London after ninety days. (Note that bt is paid 90 days prior to the
date of delivery of the £.) Let i∗t denote the 90-day deposit rate
in London, it the 90-day deposit rate in New York, and St the spot
exchange rate, that is, the dollar price of one British pound.
How can you construct a test of free capital mobility between the
United States and Great Britain in the period prior to 1920.
There are two ways to get 1 pound in 90-days from now. Way 1:
put 1/(1 + i∗) on deposit in London. Way 2: take bt/St pounds
today, exchange them U.S. dollars. You get bt dollars. With those
33
you can buy 1 British pound to be delivered in London 90 days from
now. Under free capital mobility it should be true that
1 bt
=
1 + i∗ St
So we can construct the following long-bill interest rate differential:
1
LBIRD = St − bt
(1 + i∗t )
compute the mean or the mean of the absolute value of those to
find the deviations from free capital mobility.
The next figure shows annualized covered interest rate differentials

between Germany and the United Kingdom over the period 1870 to
2000. What can you deduce from the figure about the degree of
international capital mobility between these two countries over time.
Source: This is figure 3.5 of Maurice Obstfeld and Alan M. Taylor, ‘Globalization and Capital Markets,’ in “Globalization in Historical Perspective,” Michael D. Bordo,
Alan M. Taylor and Jeffrey G. Williamson, editors, University of Chicago Press, January 2003.
The figure shows that the mean annual interest rate differential
was below 1 percent from 1870 to 1914, suggesting that there
was relatively high international capital mobility. Then in 1914 the
34
mean differential suddenly becomes significantly negative. We see
UK rates being more than 1 percent below German rates, in some
years the differential is as high as 4 percent. Only in 1980 do we
see these interest rate differentials disappear, reaching consistently
levels below 0.5 percent. In fact, Great Britain abandoned capital
controls only in 1980. So a rough description of the history of
free capital mobility between the UK and Germany is, free capital
mobility between 1870 and 1914, then rather limited international
capital mobility between 1914 and 1980, and from 1980 onwards we
see very low CIRD indicating free capital mobility.
Brazil 2009-2012
Between October 2009 and March 2012 Brazil imposed a number

of capital control taxes to reduce capital inflows into Brazil. After
March 2012 those restrictions were removed. The cupom cambial,
cupom
it , is the 360-day interest rate of U.S. dollar deposits inside
Brazil. It is defined as
cupom St
1 + it = (1 + it) ,
Ft
where St is the spot exchange rate (that is, the reais price of one
U.S. dollar), Ft is the 360-day forward exchange rate of U.S. dollars,
and it the 360-day nominal reais interest rate in Brazil. Let i∗t denote
the 360-day U.S. dollar LIBOR rate and define the spread as
cupom
spreadt = it − i∗t .
a strategy to use observations on the variable spreadt to test the
effectiveness of the Brazilian capital controls in reducing capital
inflows.
35
If the inflow controls were successful we should see that (1) the
spread is nonzero between 2009 and 2012 and (2) and more importantly,
that the spread becomes positive during that time. Meaning dollar
interest rates in Brazil can be higher than in London. This interest
rate differential cannot be arbitraged away because the capital controls
prevent investors from engaging in a trading strategy that consists
in borrowing dollars in London at i∗t and putting it on deposit in
cupom
Brazil at it .
The next figure shows the variable spreadt from June 2009 to December
2012.
The figure plots the variable spreadt. Ideally one would like to see
a value of zero until October 2009 and again starting in March
2012 and significantly positive in the meantime. The figure shows
that spreadt was higher during 2011. This suggests that the capital
contorls were not that effective until the year 2011. In that year the
spread was consistently higher than pre Oct 2009 and post March
2012 reaching up to 4 percent. The figure suggests that in the
periods without capital controls the spread is below one percent.
So I would conclude that yes at least in 2011 the controls lead to
non-zero CIRD.
The “Forward Premium Puzzle”
Recall CIRP
∗ Ft
(1 + it) = (1 + it )
St
it = domestic interest rate

i∗t = foreign interest rate
Ft = forward exchange rate (domestic price of foreign )
St = spot exchange rate (domestic price of foreign)
Low interest rate currencies are at a premium in the forward market:

1 + it Ft
∗ <1 by CIRP ⇒ <1
1 + it St
If Ft < St , then the domestic currency is said to be at a premium in

the forward market.
36
What is the “Forward Premium Puzzle”?
If Ft < St , that is, when the domestic currency is at a premium in

the forward market
then one might expect that
St+1 < St
(that is that the domestic currency will appreciate and the foreign
currency will depreciate)
But in the data the opposite is true:
Currencies that trade at a forward premium (Ft < St ), tend to

appreciate by less than the interest rate differential or even to depreciate
(St+1 > St )
Clearly, this empirical finding implies that Uncovered Interest Rate

Parity does not hold in the data.
37
Carry Trade
This observation suggests the following investment strategy:
Borrow in the low interest rate currency, invest in the high interest
rate currency, and do not hedge the exchange rate risk.
This investment strategy is known as carry trade and is widely used

by practitioners.
38
Burnside, Eichenbaum, Kleshchelski, and Rebelo∗ document that

the returns to carry trade have been on average positive.
The payoff from a carry trade.
Suppose it < i∗t . Then borrow y at the rate it and invest in foreign
country at rate i∗t .
" #
∗ St+1
Payoff from Carry Trade: = (1 + it ) − (1 + it ) y
St
∗ THE RETURNS TO CURRENCY SPECULATION Burnside, Eichenbaum,

Kleshchelski, and Rebelo, NBER WP 12489, August 2006.
39
Burnside et al. use monthly data from 1976:1 to 2005:12. In their

study the domestic country is the UK, so it is either the 1-month
or the 3-month pound sterling interest rate. They collect data on
1-month and 3-month forward exchange rates of the pound, Ft,
on spot exchange rates, St , and on foreign 1-month and 3-month
interest rates. i∗t . Then they compute the average payoff from carry
trade — taking into account that there are some transaction costs.
And they find that the average return to carry trade is positive.
The following table is taken from their paper.
40
41
Observations on Table 4 of Burnside et al. (2006)
1. Consider the case with corrections for transactions cost, columns

5, 6, and 7 of the table. The average return to carry trade for
an equally-weighted portfolio of the 10 currencies considered is
0.0029 per unit of currency invested for one month.
2. To generate substantial profits, speculators must wager very

large sums of money. For example, suppose y = 1, 000, 000, 000,
that is, you invest one billion pounds in carry trade, then after
one month the carry trade had, over the sample period, an
average payout of 2.9 million pounds per month.
3. The fact that the average payoff from carry trade is non-zero
implies that UIRP fails empirically.
42
4. Columns 4 and 7 of the table report the Sharpe Ratio, which is
defined as
mean(payoff)
Sharpe ratio =
std of payoff
The Sharpe ratio is a measure of risk. The higher the Sharpe
ratio, the higher the risk adjusted return. For comparision, note
that the Sharpe ratio of investing in the S&P 500 index over
the sample period was 0.14, which is comparable to the Sharpe
ratio of the carry trade.
5. Q: Suppose a speculator wants to generate a payoff of 1 million

pounds on average per year. How large a carry trade must he
engage in? A: He needs GBP 28.7 million each month.
6. Burnside et al. also compute covered interest rate differentials

(not shown) and find that CIRP parity holds in their sample.
Carry Trade returns are thought to have crash risk. The Economist
article refers to carry trade returns as “picking up nickels in front of
steamrollers.”
Example: large surprise appreciation of the Japanese Yen against

the U.S. dollar on October 6-8, 1998. The Yen appreciated by 14
percent (or equivalently the U.S. dollar depreciated by 14 percent).
Suppose that you were a carry trader with 1 billion dollars short in
Yen and long in U.S. dollars. The payoff of that carry trade in the
span of 2 days was -140 million dollars — that is, the steamroller
caught up with the carry trader.
43
Real interest rate differentials as indicators of the

degree of international capital market integration
One might think that observed real interest rate differentials are a
good measure of capital market integration.
For example, in the model of chapter 3, we found that the domestic

real interest rate, r, was equal to the world real interest rate, r∗,
under free capital mobility. However, that model abstracts from (a)
uncertainty and (b) does not allow for nontraded good. As we will
see either uncertainty or the presence of nontraded goods can give
rise to non-zero real interst rate differentials even when international
capital markets are fully integrated.
44
Decomposing real interest rate differentials
Our starting point is the Fisher relation, according to which the real
interest rate equals the nominal interst rate minus expected inflation.
rt — real interest rate between period t and t + 1

it — nominal interst rate between period t and t + 1
Pt —
45

International Macroeconomics: Slides For Chapter 8: International Capital Market Integration

Uploaded by

Copyright:

Available Formats

International Macroeconomics: Slides For Chapter 8: International Capital Market Integration

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

International Macroeconomics: Slides For Chapter 8: International Capital Market Integration

Uploaded by

Copyright:

Available Formats

Slides for Chapter 8:

International Capital Market Integration

Schmitt-Grohé Uribe Woodford

Idea: Infer the degree of international capital market integration

Motivation: In a closed economy those correlations must be 1,

Q: Suppose we find a near perfect correlation between S and I. Does

Measuring Capital Mobility I:

(a) The Feldstein Horioka Regression: 1960-1974

(b) Updates of the Feldstein Horioka Regression: 1960 - 2003

2. Overtime within the same country

(a) The United States, 1928-2015.

Saving and Investment Rates for 16 Industrial Countries

0.35 0.035 + 0.887 (S/GDP)→

0.25 GRE NEW

Journal 90, June 1980, 314-29.

Feldstein and Horioka estimated the following equation:

where (I/Q)i and (S/Q)i are, respectively, the average investment-

More recent evidence:

Sample of 64 countries with data from 1960 to 2003 find

where (I/Q)i and (S/Q)i are, respectively, the average investment-

Econometrica 78, March 2010, 603-632.

U.S. National Saving and Investment, 1928-2015

Data Source: Bureau of Economic Analysis, www.bea.gov.

corr(s/y, i/y) = 0.98 pre 1972 and 0.76 post 1972

But do findings of high savings-investment correlations either across

Give some examples [to be completed]

Therefore, we conclude that high S-I correlations are not evidence

Measuring Capital Mobility II:

1. Why are interest rate differentials informative about the degree

2. Empirical Evidence on the size of interest rate differentials

Asset Pricing in a 2-period small open economy

Two important interest rate parity conditions are

• CIRP= covered interest rate parity

• UIRP= uncovered interest rate parity

Let i denote the domestic, nominally risk free, interest rate on

π = probability economy is in the good state in period 2.

Domestic Price Level:

Covered Interest Rate Parity (CIRP):

But this is not true in the data!

Household buys 3 types of bonds:

• B1 domestic currency bonds, pay interest i

Qb2 + (1 + i)B1 + (1 + i∗)F1B1∗ + (1 + i∗)S2b B̃1∗ = P2b C2b (4)

How to choose B1 , B1∗ , and B̃1∗ ? To maximize utility.

Solving the period-1 budget constraint for C1 yields:

Rewrite this expression as:

Letting E1 denote the expectations operator, we have:

Using the expectations operator notation we have:

Combining (5) and (6) we obtain:

which is the covered interest rate parity condition, we had set

What about Uncovered Interest Rate Parity?

Consider the FOC w.r.t B̃1∗ .

Using the expectations operator notation we have:

Combining the asset pricing equations (8) and (6) we obtain

This was the second result we had set out to show.

Now we want to ask under what condition would UIRP hold.

cov(a, b) = E (a − E(a)) (b − E(b))

We then can express E1M2 (S2 /S1) as

then equation (8) becomes

combine this expression with equation (6) to obtain