Lectures in Open Economy Macroeconomics1

Martı́n Uribe2

This draft, November 24, 2007

1
Preliminary and incomplete. I would like to thank Javier Garcı́a-Cicco, Felix
Hammermann, and Stephanie Schmitt-Grohé for comments and suggestions. Newer
versions of these notes are available at www.econ.duke.edu/∼uribe. Comments
welcome.
2
Duke University and NBER. E-mail: [email protected].
ii
Contents

1 A First Look at the Data 1

2 An Endowment Economy 5
2.1 The Model Economy . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Response to Output Shocks . . . . . . . . . . . . . . . . . . . 11
2.3 Nonstationary Income Shocks . . . . . . . . . . . . . . . . . 13
2.4 Testing the Model . . . . . . . . . . . . . . . . . . . . . . . . 18

3 An Economy with Capital 23

3.1 The Basic Framework . . . . . . . . . . . . . . . . . . . . . . 23
3.1.1 A Permanent Productivity Shock . . . . . . . . . . . . 26
3.1.2 A Temporary Productivity shock . . . . . . . . . . . . 29
3.2 Capital Adjustment Costs . . . . . . . . . . . . . . . . . . . . 30
3.2.1 Dynamics of the Capital Stock . . . . . . . . . . . . . 32
3.2.2 A Permanent Technology Shock . . . . . . . . . . . . . 34

4 The Real Business Cycle Model 37

4.1 The Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.2 The Model’s Performance . . . . . . . . . . . . . . . . . . . . 46
4.2.1 The Role of Capital Adjustment Costs . . . . . . . . . 48
4.3 Alternative Ways to Induce Stationarity . . . . . . . . . . . . 49
4.3.1 External Discount Factor (EDF) . . . . . . . . . . . . 50
4.3.2 External Debt-Elastic Interest Rate (EDEIR) . . . . 52
4.3.3 Internal Debt-Elastic Interest Rate (IDEIR) . . . . . 54
4.3.4 Portfolio Adjustment Costs (PAC) . . . . . . . . . . . 56
4.3.5 Complete Asset Markets (CAM) . . . . . . . . . . . . 58
4.3.6 The Nonstationary Case (NC) . . . . . . . . . . . . . 61
4.3.7 Quantitative Results . . . . . . . . . . . . . . . . . . . 61
4.4 Appendix A: Log-Linearization . . . . . . . . . . . . . . . . . 65

iii
iv CONTENTS

4.5 Appendix B: Solving Dynamic General Equilibrium Models . 67

4.6 Local Existence and Uniqueness of Equilibrium . . . . . . . . 78
4.7 Second Moments . . . . . . . . . . . . . . . . . . . . . . . . . 79
4.8 Impulse Response Functions . . . . . . . . . . . . . . . . . . . 83
4.9 Matlab Code For Linear Perturbation Methods . . . . . . . . 83
4.10 Higher Order Approximations . . . . . . . . . . . . . . . . . . 84
4.11 Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
4.11.1 An RBC Small Open Economy with an internal debt-
elastic interest-rate premium . . . . . . . . . . . . . . 85

5 The Terms of Trade 87

5.1 Defining the Terms of Trade . . . . . . . . . . . . . . . . . . . 88
5.2 Empirical Regularities . . . . . . . . . . . . . . . . . . . . . . 88
5.2.1 TOT-TB Correlation: Two Early Explanations . . . . 90
5.3 Terms-of-Trade Shocks in an RBC Model . . . . . . . . . . . 99
5.3.1 Households . . . . . . . . . . . . . . . . . . . . . . . . 99
5.3.2 Production of Consumption Goods . . . . . . . . . . . 102
5.3.3 Production of Tradable Consumption Goods . . . . . 103
5.3.4 Production of Importable, Exportable, and Nontrad-
able Goods . . . . . . . . . . . . . . . . . . . . . . . . 104
5.3.5 Market Clearing . . . . . . . . . . . . . . . . . . . . . 106
5.3.6 Driving Forces . . . . . . . . . . . . . . . . . . . . . . 106
5.3.7 Competitive Equilibrium . . . . . . . . . . . . . . . . 107
5.3.8 Calibration . . . . . . . . . . . . . . . . . . . . . . . . 107
5.3.9 Model Performance . . . . . . . . . . . . . . . . . . . . 111
5.3.10 How Important Are the Terms of Trade? . . . . . . . 113

6 Interest-Rate Shocks 115

6.1 An Empirical Model . . . . . . . . . . . . . . . . . . . . . . . 118
6.2 Impulse Response Functions . . . . . . . . . . . . . . . . . . . 120
6.3 Variance Decompositions . . . . . . . . . . . . . . . . . . . . . 126
6.4 A Theoretical Model . . . . . . . . . . . . . . . . . . . . . . . 131
6.4.1 Households . . . . . . . . . . . . . . . . . . . . . . . . 132
6.4.2 Firms . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
6.4.3 Driving Forces . . . . . . . . . . . . . . . . . . . . . . 141
6.4.4 Equilibrium, Functional Forms, and Parameter Values 142
6.5 Theoretical and Estimated Impulse Responses . . . . . . . . . 145
6.6 The Endogeneity of Country Spreads . . . . . . . . . . . . . . 147
CONTENTS v

7 Sovereign Debt 151

7.1 Empirical Regularities . . . . . . . . . . . . . . . . . . . . . . 152
7.2 The Cost of Default . . . . . . . . . . . . . . . . . . . . . . . 157
7.3 A Reputational Model of Sovereign Debt . . . . . . . . . . . . 159
7.4 Saving and the Breakdown of Reputational Lending . . . . . 166
vi CONTENTS
Chapter 1

A First Look at the Data

In discussions of business cycles in small open economies, a critical distinc-

tion is between developed and emerging economies. The group of developed

economies is typically defined by countries with high income per capita, and

the group of emerging economies is composed of middle income countries.

Examples of developed small open economies are Canada and Belgium, and

examples of small open emerging economies are Argentina and Malaysia.

A striking difference between developed and emerging economies is that

observed business cycles in emerging countries are about twice as volatile

as in developed countries. Table 1.1 illustrates this contrast by displaying

key business-cycle properties in Argentina and Canada. The volatility of

detrended output is 4.6 in Argentina and only 2.8 in Canada. Another re-

markable difference between developing and developed countries suggested

by the table is that the trade balance-to-output ratio is much more coun-

tercyclical in emerging countries than in developed countries. Periods of

economic boom (contraction) are characterized by relatively larger trade

1
2 Martı́n Uribe

Table 1.1: Business Cycles in Argentina and Canada

Variable σx corr(xt , xt−1 ) corr(xt , GDPt )
GDP
Argentina 4.6 0.79 1
Canada 2.8 0.61 1
Consumption
Argentina 5.4 0.96
Canada 2.5 0.70 0.59
Investment
Argentina 13.3 0.94
Canada 9.8 0.31 0.64
TB/GDP
Argentina 2.3 -0.84
Canada 1.9 0.66 -0.13
Hours
Argentina 4.1 0.76
Canada 2.0 0.54 0.80
Productivity
Argentina 3.0 0.48
Canada 1.7 0.37 0.70

Source: Mendoza (1991), Kydland and Zarazaga (1997). For Ar-

gentina, data on hours and productivity are limited to the manufac-
turing sector.
Lectures in Open Economy Macroeconomics, Chapter 1 3

deficits (surpluses) in emerging countries than in developed countries. A

third difference between Canadian and Argentine business cycles is that in

Argentina consumption appears to be more volatile than output at business-

cycle frequencies, whereas the reverse is the case in Canada. Two additional

differences between the business cycle in Argentina and Canada are that in

Argentina the correlation of the components of aggregate demand with GDP

are twice as high as in Canada, and that in Argentina hours and productivity

are less correlated with GDP than in Canada.

One dimension along which business cycles in Argentina and Canada are

similar is the procyclicality of consumption, investment, hours, and produc-

tivity. In both countries, these variables move in tandem with output.

The differences between the business cycles of Argentina and Canada

turn out to hold much more generally between emerging and developed

countries. Table 1.2 displays average business cycle facts in developed and

emerging economies. The table averages second moments of detrended data

for 13 small emerging countries and 13 small developed countries (the list

of countries appears at the foot of the table). For all countries, the time

series are at least 40 quarters long. The data is detrended using a band-

pass filter that leaves out all frequencies above 32 quarters and below 6

quarters. The data shown in the table is broadly in line with the conclusions

drawn from the comparison of business cycles in Argentina and Canada. In

particular, emerging countries are significantly more volatile and display a

much more countercyclical trade-balance share than developed countries.

Also, consumption is more volatile than output in emerging countries but

less volatile than output in developed countries.

4 Martı́n Uribe

Table 1.2: Business Cycles: Emerging Vs. Developed Economies

Emerging Developed
Moment Countries Countries
σy 2.02 1.04
σ∆y 1.87 0.95
ρy 0.86 0.9
ρ∆y 0.23 0.09
σc /σy 1.32 0.94
σi /σy 3.96 3.42
σtb/y 2.09 0.71
ρtb/y,y -0.58 -0.26
ρc,y 0.74 0.69
ρi,y 0.87 0.75

Note: Average values of moments for 13 small emerging countries and

13 small developed countries. Emerging countries: Argentina, Brazil,
Ecuador, Israel, Korea, Malaysia, Mexico, Peru, Philippines, Slovak
Republic, South Africa, Thailand, and Turkey. Developed Countries:
Australia, Austria, Belgium, Canada, Denmark, Finland, Netherlands,
New Zealand, Norway, Portugal, Spain, Sweden, Switzerland. Data are
detrended using a band-pass filter including frequencies between 6 and
32 quarters with 12 leads and lags.
Source: Aguiar and Gopinath (2004).
Chapter 2

An Endowment Economy

The purpose of this chapter is to build a canonical dynamic, general equi-

librium model of the small open economy capable of capturing some of the

empirical regularities of business cycles in small emerging and developed

countries documented in chapter 1. The model developed in this chapter is

simple enough to allow for a full characterization of its equilibrium dynamics

using pen and paper.

2.1 The Model Economy

Consider an economy populated by a large number of infinitely lived house-

holds with preferences described by the utility function

∞
X
E0 β t U (ct ), (2.1)
t=0

5
6 Martı́n Uribe

where ct denotes consumption and U denotes the single-period utility func-

tion, which is assumed to be strictly increasing and strictly concave.

The evolution of the debt position of the representative household is

given by

dt = (1 + r)dt−1 + ct − yt , (2.2)

where dt denotes the debt position assumed in period t, r denotes the in-

terest rate, assumed to be constant, and yt is an exogenous and stochastic

endowment of goods. This endowment process represents the sole source of

uncertainty in this economy. The above constraint states that the change in

the level of debt, dt − dt−1 , has two sources, interest services on previously

acquired debt, rdt−1 , and excess expenditure over income, ct − yt . House-

holds are subject to the following borrowing constraint that prevents them

from engaging in Ponzi games:

dt+j
lim Et ≤ 0. (2.3)
j→∞ (1 + r)j

This limit condition states that the household’s debt position must be ex-

pected to grow at a rate lower than the interest rate r. The optimal allo-

cation of consumption and debt will always feature this constraint holding

with strict equality. This is because if the allocation {ct , dt }∞

t=0 satisfies

the no-Ponzi-game constraint with strict inequality, then one can choose an

alternative allocation {c0t , d0t }∞

t=0 that also satisfies the no-Ponzi-game con-

straint and satisfies c0t ≥ ct , for all t ≥ 0, with c0t0 > ct0 for at least one date

t0 ≥ 0. This alternative allocation is clearly strictly preferred to the original

one because the single period utility function is strictly increasing.

Lectures in Open Economy Macroeconomics, Chapter 2 7

The household chooses processes for ct and dt for t ≥ 0, so as to maxi-

mize (2.1) subject to (2.2) and (2.3). The optimality conditions associated

with this problem are (2.2), (2.3) holding with equality, and the following

Euler condition:

U 0 (ct ) = β(1 + r)Et U 0 (ct+1 ). (2.4)

The interpretation of this expression is simple. If the household sacrifices

one unit of consumption in period t and invests it in financial assets, its

period-t utility falls by U 0 (ct ). In period t + 1 the household receives the

unit of goods invested plus interests, 1 + r, yielding β(1 + r)Et U 0 (ct+1 ) utils.

At the optimal allocation, the cost and benefit of postponing consumption

must equal each other in the margin.

We make two additional assumptions that greatly facilitates the analysis.

First we require that the subjective and pecuniary rates of discount, β and

1/(1 + r), be equal to each other, that is,

β(1 + r) = 1.

This assumption eliminates long-run growth in consumption. Second, we

assume that the period utility index is quadratic and given by

1
U (c) = − (c − c̄)2 , (2.5)
2

with c < c̄.1 This particular functional form makes it possible to obtain

a closed-form solution of the model. Under these assumptions, the Euler

1
After imposing this assumption, our model becomes essentially Hall’s (1978) perma-
nent income model of consumption.
8 Martı́n Uribe

condition (2.4) collapses to

ct = Et ct+1 , (2.6)

which says that consumption follows a random walk; at each point in time,

households expect to maintain a constant level of consumption.

We now derive an intertemporal resource constraint by combining the

household’s sequential budget constraint (2.2) and the no-Ponzi-scheme con-

straint (2.3) holding with equality—also known as the transversality condi-

tion. Begin by expressing the sequential budget constraint in period t as

(1 + r)dt−1 = yt − ct + dt .

Lead this equation 1 period and use it to get rid of dt :

yt+1 − ct+1 dt+1

(1 + r)dt−1 = yt − ct + + .
1+r 1+r

Repeat this procedure s times to get

s
X yt+j − ct+j dt+s
(1 + r)dt−1 = j
+ .
(1 + r) (1 + r)s
j=0

Apply expectations conditional on information available at time t and take

the limit for s → ∞ using the transversality condition (equation (2.3) hold-

ing with equality) to get the following intertemporal resource constraint:

∞
X yt+j − ct+j
(1 + r)dt−1 = Et .
(1 + r)j
j=0
Lectures in Open Economy Macroeconomics, Chapter 2 9

Intuitively, this equation says that the country’s initial net foreign debt

position must equal the expected present discounted value of current and

future differences between output and absorption.

Now use the Euler equation (2.6) to deduce that Et ct+j = ct . Use this

result to get rid of expected future consumption in the above expression and

rearrange to obtain

∞
X yt+j
r
rdt−1 + ct = Et . (2.7)
1+r (1 + r)j
j=0

This expression states that the optimal plan allocates the annuity value of
P yt+j
the income stream 1+rr
Et ∞ j=0 (1+r)j to consumption, ct , and to debt ser-

vice, rdt−1 . To be able to fully characterize the equilibrium in this economy,

we assume that the endowment process follows an AR(1) process of the

form,

yt = ρyt−1 + t ,

where t denotes an i.i.d. innovation and the parameter ρ ∈ (−1, 1) defines

the serial correlation of the endowment process. The larger is ρ the more

persistent the endowment process. Then, the j-period-ahead forecast of

output in period t is given by

Et yt+j = ρj yt .

Using this expression to eliminate expectations of future income from equa-

10 Martı́n Uribe

tion (2.7), we obtain

∞  j
r X ρ
rdt−1 + ct = yt
1+r 1+r
j=0
r
= yt .
1+r−ρ

Solving for ct , we obtain

r
ct = yt − rdt−1 . (2.8)
1+r−ρ

Because ρ is less than unity, we have that a unit increase in the endowment

leads to a less-than-unity increase in consumption. Letting tbt ≡ y − ct and

cat ≡ −rdt−1 + tbt denote, respectively, the trade balance and the current

account in period t, we have

1−ρ
tbt = rdt−1 + yt
1+r−ρ

and
1−ρ
cat = yt .
1+r−ρ

Note that the current account inherits the stochastic process of the under-

lying endowment shock. Because the current account equals the change in

the country’s net foreign asset position, i.e., cat = −(dt − dt−1 ), it follows

that the equilibrium evolution of the stock of external debt is given by

1−ρ
dt = dt−1 − yt .
1+r−ρ
Lectures in Open Economy Macroeconomics, Chapter 2 11

According to this expression, external debt follows a random walk and is

therefore nonstationary. A temporary increase in the endowment produces

a gradual but permanent decline in the stock of foreign liabilities. The long-

run behavior of the trade balance is governed by the dynamics of external

debt. Thus, an increase in the endowment shock leads to a permanent

deterioration in the trade balance.

2.2 Response to Output Shocks

Consider the response of our model economy to an unanticipated increase

in output. Assume that 0 < ρ < 1, so that endowment shocks are positively

serially correlated. Two polar cases are of interest. In the first case, the

endowment shock is assumed to be purely transitory, ρ = 0. According to

equation (2.8), when innovations in the endowment are purely temporary

only a small part of the changes in income—a fraction r/(1 + r)—is allo-

cated to current consumption. Most of the endowment increase—a fraction

1/(1 + r)—is saved. The intuition for this result is clear. Because income is

expected to fall quickly to its long-run level households smooth consumption

by eating a tiny part of the current windfall and leaving the rest for future

consumption. In this case, the current account plays the role of a shock

absorber. Households borrow to finance negative income shocks and save in

response to positive shocks. It follows that the more temporary are endow-

ment shocks, the more volatile is the current account. In the extreme case

of purely transitory shocks, the standard deviation of the current account

is given by σy /(1 + r), which is close to the volatility of the endowment

12 Martı́n Uribe

shock itself for small values of r. More importantly, the current account is

procyclical. That is, it improves during expansions and deteriorates during

contractions. This prediction represents a serious problem for this model.

For, as documented in chapter 1, the current account is countercyclical in

small open economies, especially in developing countries.

The other polar case emerges when shocks are highly persistent, ρ →

1. In this case, households allocate all innovations in their endowments to

current consumption, and, as a result the current account is nil and the

stock of debt remains constant over time. Intuitively, when endowment

shocks are permanent, an increase in income today is not accompanied by

the expectation of future income contractions. As a result, households are

able to sustain a smooth consumption path by consuming the totality of the

current income shock.

The intermediate case of a gradually trend-reverting endowment process

(ρ ∈ (0, 1)) is illustrated in figure 2.1. In response to the positive endowment

shock, consumption experiences a once-and-for-all increase. This expansion

in domestic absorption is smaller than the initial increase in income. As a

result, the trade balance and the current account improve. After the initial

increase, these two variables converge gradually to their respective long-run

levels. Note that the trade balance converges to a level lower than the pre-

shock level. This is because in the long-run the economy settles at a lower

level of external debt, which requires a smaller trade surplus to be served.

Summarizing, in this model, which captures the essential elements of

what has become known as the intertemporal approach to the current ac-

count, external borrowing is conducted under the principle: ‘finance tem-

Lectures in Open Economy Macroeconomics, Chapter 2 13

Figure 2.1: Response to a Positive Endowment Shock

porary shocks, adjust to permanent shocks.’ A central failure of the model

is the prediction of a procyclical current account. Fixing this problem is at

the heart of what follows in this and the next two chapters.

2.3 Nonstationary Income Shocks

Suppose now that the rate of change of output, rather than its level, displays

mean reversion. Specifically, let

∆yt ≡ yt − yt−1
14 Martı́n Uribe

Figure 2.2: Stationary Versus Nonstationary Endowment Shocks

denote the change in endowment between periods t − 1 and t. Suppose that

∆yt evolves according to the following autoregressive process:

∆yt = ρ∆yt−1 + t ,

where t is an i.i.d. shock with mean zero and variance σ2 , and ρ ∈ [0, 1)

is a constant parameter. According to this process, the level of income is

nonstationary, in the sense that a positive output shock (t > 0) produces a

permanent future expected increase in the level of output. Faced with such

an income profile, consumption-smoothing households have an incentive to

borrow against future income, thereby producing a countercyclical tendency

in the current account. This is the basic intuition why allowing for a non-

stationary output process can help explain the behavior of the trade balance

and the current account at business-cycle frequencies. Figure 2.2 provides

a graphical expression of this intuition. The following model formalizes this

story.
Lectures in Open Economy Macroeconomics, Chapter 2 15

As before, the model economy is inhabited by an infinitely lived repre-

sentative household that chooses contingent plans for consumption and debt

to maximize the utility function (2.5) subject to the sequential resource con-

straint (2.2) and the no-Ponzi-game constraint (2.3). The first-order con-

ditions associated with this problem are the sequential budget constraint,

the no-Ponzi-game constraint holding with equality, and the Euler equa-

tion (2.6). Using these optimality conditions yields the expression for con-

sumption given in equation (2.7), which we reproduce here for convenience

∞
X yt+j
r
ct = −rdt−1 + Et .
1+r (1 + r)j
j=0

Using this expression and recalling that the current account is defined as

cat = yt − ct − rdt−1 , we can write

∞
X yt+j
r
cat = yt − Et .
1+r (1 + r)j
j=0

Rearranging, we obtain

∞
X ∆yt+j
cat = −Et .
(1 + r)j
j=1

This expression states that the current account equals the present discounted

value of future expected income decreases. According to the autoregressive

process assumed for the endowment, we have that Et ∆yt+j = ρj ∆yt . Using
16 Martı́n Uribe

this result in the above expression, we can write the current account as:

−ρ
cat = ∆yt .
1+r−ρ

According to this formula, the current account deteriorates in response to

a positive innovation in output. This implication is an important improve-

ment relative to the model with stationary shocks. Recall that when the

endowment level is stationary the current account increases in response to

a positive endowment shock.

We note that the countercyclicality of the current account in the model

with nonstationary shocks depends crucially on output changes being posi-

tively serially correlated, or ρ > 0. In effect, when ρ is zero (negative), the

current account ceases to be countercyclical (is procyclical). The intuition

behind this result is clear. For an unexpected increase in income to induce

an increase in consumption larger than the increase in income itself, it is

necessary that future income be expected to be higher than current income,

which happens only if ∆yt is positively serially correlated.

Are implied changes in consumption more or less volatile than changes

in output? This question is important because, as we saw in chapter 1,

developing countries are characterized by consumption growth being more

volatile than output growth. Formally, letting σ∆c and σ∆y denote the

standard deviations of ∆ct ≡ ct − ct−1 and ∆yt , respectively, we wish to find

2 can be higher than σ 2 in equilibrium.2 We
out conditions under which σ∆c ∆y

2
Strictly speaking, this exercise is not comparable to the data displayed in chapter 1,
because here we analyze changes in consumption and output, whereas in chapter 1 we
reported statistics pertaining to the growth rates of consumption and output.
Lectures in Open Economy Macroeconomics, Chapter 2 17

start with the definition of the current account

cat = yt − ct − rdt−1 .

Taking differences, we obtain

cat − cat−1 = ∆yt − ∆ct − r(dt−1 − dt−2 ).

Noting that dt−1 − dt−2 = −cat−1 and solving for ∆ct , we obtain:

∆ct = ∆yt − cat + (1 + r)cat−1

ρ ρ(1 + r)
= ∆yt + ∆yt − ∆yt−1
1+r−ρ 1+r−ρ
1+r ρ(1 + r)
= ∆yt − ∆yt−1
1+r−ρ 1+r−ρ
1+r
= t . (2.9)
1+r−ρ

2 (1−ρ2 ) =
It follows directly from the AR(1) specificaiton of ∆yt that σ∆y

σ2 . Then, we can write the standard deviation of consumption changes as

 p
σ∆c 1+r
= 1 − ρ2 .
σ∆y 1+r−ρ

The right-hand side of this expression equals unity at ρ = 0. This result

confirms the one obtained earlier in this chapter, namely that when the

level of income is a random walk, consumption and income move hand in

hand, so their changes are equally volatile. The right hand side of the above

expression is increasing in ρ at ρ = 0. It follows that there are values of

18 Martı́n Uribe

ρ in the interval (0, 1) for which the volatility of consumption changes is

indeed higher than that of income changes. This property ceases to hold as

∆yt becomes highly persistent. This is because as ρ → 1, the variance of

∆yt becomes infinitely large as changes in income become a random walk,

whereas, as expression (2.9) shows, ∆ct follows an i.i.d. process with finite

variance for all values of ρ ∈ [0, 1).

2.4 Testing the Model

Hall (1978) was the first to explore the econometric implication of the simple

model developed in this chapter. Specifically, Hall tested the prediction

that consumption must follow a random walk. Hall’s work motivated a

large empirical literature devoted to testing the empirical relevance of the

model described above. Campbell (1987), in particular, deduced and tested

a number of theoretical restrictions on the equilibrium behavior of national

savings. In the context of the open economy, Campbell’s restrictions are

readily expressed in terms of the current account. Here we review these

restrictions and their empirical validity.

We start by deriving a representation of the current account that involves

expected future changes in income. Noting that the current account in

period t, denoted cat , is given by yt − ct − rdt−1 we can write equation (2.7)

as
∞
X
−(1 + r)cat = −yt + rEt (1 + r)−j yt+j .
j=1
Lectures in Open Economy Macroeconomics, Chapter 2 19

Defining ∆xt+1 = xt+1 − xt , it is simple to show that

∞
X ∞
X
−yt + rEt (1 + r)−j yt+j = (1 + r) (1 + r)−j Et ∆yt+j .
j=1 j=1

Combining the above two expression we can write the current account as

∞
X
cat = − (1 + r)−j Et ∆yt+j . (2.10)
j=1

Intuitively, this expression states that the country borrows from the rest of

the world (runs a current account deficit) income is expected to grow in the

future. Similarly, the country chooses to build its net foreign asset position

(runs a current account surplus) when income is expected to decline in the

future. In this case the country saves for a rainy day.

Consider now an empirical representation of the time series ∆yt and cat .

Define  
 ∆yt 
xt =  .
cat

Consider estimating a VAR system including xt :

xt = Dxt−1 + t .

Let Ht denote the information contained in the vector xt . Then, from the

above VAR system, we have that the forecast of xt+j given Ht is given by

Et [xt+j |Ht ] = D j xt .
20 Martı́n Uribe

It follows that
 
∞
X  
 ∆yt 
(1 + r)−j Et [∆yt+j |Ht ] = 1 0 [I − D/(1 + r)]−1 D/(1 + r)  .
j=1 cat

 
Let F ≡ − 1 0 [I − D/(1 + r)]−1 D/(1 + r). Now consider running a

regression of the left and right hand side of equation (2.10) onto the vector

xt . Since xt includes cat as one element, we obtain that the regression coef-

ficient for the left-hand side regression is the vector [0 1]. The regression

coefficients of the right-hand side regression is F . So the model implies the

following restriction on the vector F :

F = [0 1].

Nason and Rogers (2006) perform an econometric test of this restriction.

They estimate the VAR system using Canadian data on the current account

and GDP net of investment and government spending. The estimation sam-

ple is 1963:Q1 to 1997:Q4. The VAR system that Nason and Rogers estimate

includes 4 lags. In computing F , they calibrate r at 3.7 percent per year.

Their data strongly rejects the above cross-equation restriction of the model.

The Wald statistic associated with null hypothesis that F = [0 1] is 16.1,

with an asymptotic p-value of 0.04. This p-value means that if the null hy-

pothesis was true, then the Wald statistic, which reflects the discrepancy of

F from [0 1], would take a value of 16.1 or higher only 4 out of 100 times.

Consider now an additional testable cross-equation restriction on the

Lectures in Open Economy Macroeconomics, Chapter 2 21

theoretical model. From equation (2.10) it follows that

Et cat+1 − (1 + r)cat − Et ∆yt+1 = 0. (2.11)

According to this expression, the variable cat+1 − (1 + r)cat − ∆yt+1 is

unpredictable in period t. In particular, if one runs a regression of this

variable on current and past values of xt , all coefficients should be equal to

zero.3

This restriction is not valid in a more general version of the model featur-

ing private demand shocks. Consider, for instance, a variation of the model

economy where the bliss point is a random variable. Specifically, replace c̄

in equation (2.5) by c̄ + µt , where c̄ is still a constant, and µt is an i.i.d.

shock with mean zero. In this environment, equation (2.11) becomes

Et cat+1 − (1 + r)cat − Et ∆yt+1 = µt .

Clearly, because in general µt is correlated with cat , the orthogonality con-

dition stating that cat+1 − (1 + r)cat − ∆yt+1 be orthogonal to variables

dated t or earlier, will not hold. Nevertheless, in this case we have that

cat+1 − (1 + r)cat − ∆yt+1 should be unpredictable given information avail-

able in period t − 1 or earlier.4 Both of the orthogonality conditions dis-

cussed here are strongly rejected by the data. Nason and Rogers (2006) find

that a test of the hypothesis that all coefficients are zero in a regression of
3
Consider projecting the left- and right-hand sides of this expression on the information
set Ht . This projection yields the orthogonality restriction [0 1][D −(1+r)I]−[1 0]D =
[0 0].
4
In particular, one can consider projecting the above expression onto ∆yt−1 and cat−1 .
This yields the orthogonality condition [0 1][D − (1 + r)I]D − [1 0]D2 = [0 0].
22 Martı́n Uribe

cat+1 − (1 + r)cat − ∆yt+1 onto current and past values of xt has a p-value of

0.06. The p-value associated with a regression featuring as regressors past

values of xt is 0.01.
Chapter 3

An Economy with Capital

A theme of chapter 2 is that the simple endowment economy fails to predict

the countercyclicality of the trade balance. In this chapter, we show that

allowing for capital accumulation can help resolve this problem.

3.1 The Basic Framework

Consider a small open economy populated by a large number of infinitely

lived households with preferences described by the utility function

∞
X
β t U (ct ), (3.1)
t=0

where ct denotes consumption, β ∈ (0, 1) denotes the subjective discount

factor, and U denotes the period utility function, assumed to be strictly

increasing, strictly concave, and twice continuously differentiable. House-

holds seek to maximize this utility function subject to the following three

23
24 Martı́n Uribe

constraints:

bt = (1 + r)bt−1 + θt F (kt ) − ct − it , (3.2)

kt+1 = kt + it ,

and
bt+j
lim ≥ 0, (3.3)
j→∞ (1 + r)j

where bt denotes real bonds bought in period t yielding the constant interest

rate r > 0, kt denotes the stock of physical capital, and it denotes invest-

ment. The function F describes the production technology and is assumed

to be strictly increasing, strictly concave, and to satisfy the Inada conditions.

The variable θt denotes an exogenous nonstochastic productivity factor. For

the sake of simplicity, we assume that the capital stock does not depreciate.

We relax this assumption later.

We wish to characterize the response of the economy to a permanent

increase in θt .

The Lagrangian associated with the household’s problem is

∞
X
L= β t {U (ct ) + λt [(1 + r)bt−1 + kt + θt F (kt ) − ct − kt+t − bt+1 ]} .
t=0

The first-order conditions corresponding to this problem are

U 0 (ct ) = λt ,

λt = β(1 + r)λt+1 ,
Lectures in Open Economy Macroeconomics, Chapter 3 25

λt = βλt+1 [1 + θt+1 F 0 (kt+1 )],

bt = (1 + r)bt−1 + θt F (kt ) − ct − kt+1 + kt ,

and the transversality condition

bt
lim = 0.
t→∞ (1 + r)t

As in the endowment-economy model of chapter 2, we assume that β(1+r) =

1, to avoid inessential long-run dynamics. This assumption together with

the first two of the above optimality conditions implies that consumption is

constant over time.

The above optimality conditions can be reduced to the following two

expressions:

r = θt+1 F 0 (kt+1 ) (3.4)

and
∞
r X θt+j F (kt+j ) − kt+j+1 + kt+j
ct = rbt−1 + , (3.5)
1+r (1 + r)j
j=0

for t ≥ 0. The first of these equilibrium conditions states that households

invest in physical capital until the marginal product of capital equals the

rate of return on foreign bonds. It follows from (3.4) that next period’s level

of physical capital, kt+1 , is an increasing function of the future expected

level productivity, θt+1 , and a decreasing function of the opportunity cost

of holding physical capital, r. Formally,

kt+1 = κ(θt+1 ; r); κ1 > 0, κ2 < 0.

26 Martı́n Uribe

The second equilibrium condition, equation (3.5), says that consumption

equals the interest flow on a broad definition of wealth, which includes not

only initial financial wealth, b−1 , but also the present discounted value of the

differences between output and investment. To obtain this expression, follow

the same steps as in the derivation of its counterpart for the endowment

economy studied in chapter 2.

3.1.1 A Permanent Productivity Shock

Suppose that up until period -1 inclusive the technology factor θt was con-

stant and equal to θ̄. Suppose further that in period 0 there is a permanent,

unexpected increase in the technology factor to θ 0 > θ̄. That is, θt = θ̄ for

t < 0 and θt = θ 0 for t ≥ 0.

Consumption experiences a permanent increase in response to the per-

manent technology shock. That is, ct = c0 > c−1 for all t > 0. To

see this, consider the suboptimal paths for consumption and investment

cst = θ 0 F (k̄) + rb−1 and ist = 0 for all t, for t ≥ 0, where k̄ denotes the

initial level of capital. Clearly, because θ 0 > θ̄, the consumption path cst is

strictly preferred to the pre-shock path, given by θ̄F (k̄) + rb−1 . To show

that the proposed allocation is feasible, let us plug the consumption and

investment paths cst and ist into the sequential budget constraint (3.2) to

obtain the sequence of asset positions bst = b−1 for all t ≥ 0. Obviously,

limt→∞ b−1 /(1 + r)t = 0, so the proposed suboptimal allocation satisfies the

no-Ponzi-game condition (3.3). We have shown the existence of a feasible

consumption path that ! is strictly preferred to the pre-shock consump-

tion allocation. It follows that the optimal consumption path must also be
Lectures in Open Economy Macroeconomics, Chapter 3 27

strictly preferred to the pre-shock consumption path. This result together

with the fact that the optimal consumption path is constant over time im-

plies that consumption must jump up once and for all in period 0.

Because k0 was chosen in period −1, when households expected θ0 to be

equal to θ̄, we have that k0 = k̄, where k̄ is given by k̄ = κ(θ̄, r). In period

0, investment experiences a once-and-for-all increase that brings the level of

capital up from k̄ to a level k ∗ ≡ κ(θ 0 , r) > k̄. Thus, kt = k ∗ for t ≥ 1.

Plugging this path for the capital stock into equation (3.5) and evaluating

that equation at t = 0 we get

r  0  1 0
c0 = rb−1 + θ F (k̄) − k ∗ + k̄ + θ F (k ∗ ).
1+r 1+r

The trade balance is given by tbt = θt F (kt ) − ct − it . Thus, we have

1  0 
tb0 = −rb−1 − θ F (k ∗ ) − θ 0 F (k̄) + (k ∗ − k̄) .
1+r

Before period zero, the trade balance is simply equal to −rb−1 . This together

with the fact that the expression within square brackets in the above equa-

tion is unambiguously positive, implies that in response to the permanent

technology shock the trade balance deteriorates in period zero. In period

1 the trade balance improves. To see this, note that output increases from

θ 0 F (k0 ) in period 0 to θ 0 F (k ∗ ) in period 1. On the demand side, in period

1 investment falls to zero, while consumption remains unchanged. Thus,

the trade balance, given by output minus investment minus consumption,

necessarily goes up in period 1. The trade balance remains constant after

28 Martı́n Uribe

period 1. The current account deteriorates in period zero and is nil from

period 1 onward.

The initial deteriorations of the trade balance and the current account

implied by the model represent a significant improvement with respect to

the endowment economy studied in chapter 2. For the initial worsening of

the external accounts in a context of expansion in aggregate activity is in

line with the empirical evidence presented in chapter 1.

In obtaining a deterioration of the trade balance the assumed persistence

of the productivity shock plays a significant role. A permanent increase in

productivity induces a strong response in domestic absorption. To see this,

note first that from the vintage point of period zero, households face an

increasing path for output net of investment. As a result, the propensity

to consume out of current income is high. At the same time, because the

productivity of capital, θt , is expected to stay high in the future, it pays to

increase investment spending.

Another important factor in generating a decline in the trade balance in

response to a positive productivity shock is the assumed absence of capital

adjustment costs. Note that in response to the increase in future expected

productivity, the entire adjustment in investment occurs in period zero.

Indeed, investment falls to zero in period 1 and remains nil thereafter. In

the presence of costs of adjusting the stock of capital, investment spending

is spread over a number of periods, dampening the increase in domestic

absorption in the date the shock occurs. We will study the role of adjustment

costs more closely shortly.

Lectures in Open Economy Macroeconomics, Chapter 3 29

3.1.2 A Temporary Productivity shock

To stress the importance of persistence in productivity movements in in-

ducing a deterioration of the trade balance in response to a positive output

shock, it is worth analyzing the effect of a purely temporary shock. Specif-

ically, suppose that up until period -1 inclusive the productivity factor θt

was constant and equal to θ̄. Suppose also that in period -1 people assigned

a zero probability to the event that θ0 would be different from θ̄. In period

0, however, a zero probability event happens. Namely, θ0 = θ 0 > θ̄. Fur-

thermore, suppose that everybody correctly expects the productivity shock

to be purely temporary. That is, everybody expects θt = θ̄ for all t > 0. In

this case, equation (3.4) implies that the capital stock, and therefore also

investment, are unaffected by the productivity shock. That is, kt = k̄ for all

t ≥ 0, where k̄ is the level of capital inherited in period 0. This is intuitive.

The productivity of capital unexpectedly increases in period zero. As a re-

sult, households would like to have more capital in that period. But k0 is

fixed in period zero. Investment in period zero can only increase the future

stock of capital. But agents have no incentives to have a higher capital stock

in the future, because its productivity is expected to go back down to its

historic level θ̄ right after period 0.

The positive productivity shock in period zero does produce an increase

in output in that period, from θ̄F (k̄) to θ 0 F (k̄). That is

y0 = y−1 + (θ 0 − θ̄)F (k̄),

where y−1 ≡ θ̄F (k̄) is the pre-shock level of output. This output effect
30 Martı́n Uribe

induces higher consumption. In effect, using equation (3.5) we have that

r
c0 = c−1 + (θ 0 − θ̄)F (k̄),
1+r

where c−1 ≡ −rb−1 + θ̄F (θ̄) is the pre-shock level of consumption. Basically,

households invest the entire increase in output in the international financial

market and increase consumption by the interest flow associated with that

financial investment.

Combining the above two expressions and recalling that investment is

unaffected by the temporary shock, we get that the trade balance in period

0 is given by

1
tb0 − tb−1 = (y0 − y−1 ) − (c0 − c−1 ) − (i0 − i−1 ) = (θ 0 − θ̄)F (k0 ) > 0.
1+r

This expression shows that the trade balance improves on impact. The

reason for this counterfactual response is simple: ‘Firms have no incentive

to invest, as the increase in the productivity of capital is short lived, and

consumers save most of the purely temporary increase in income in order to

smooth consumption over time.

3.2 Capital Adjustment Costs

Consider now an economy identical to the one described above but in which

changes in the stock of capital come at a cost. Capital adjustment costs

are typically introduced, in different forms, in small open economy models

to dumpen the volatility of investment over the business cycle (see, e.g.,
Lectures in Open Economy Macroeconomics, Chapter 3 31

Mendoza, 1991; and Schmitt-Grohé, 1998). Suppose that the sequential

budget constraint is of the form

1 i2t
bt = (1 + r)bt−1 + θt F (kt ) − ct − it − .
2 kt

Here, capital adjustment costs are given by i2t /(2kt ) and are a strictly con-

vex function of investment. Moreover, both the level and the slope of this

function vanish at the steady-state value of investment, it = 0.

As in the economy without adjustment costs, the law of motion of the

capital stock is given by

kt+1 = kt + it .

Households seek to maximize the utility function given in (3.1) subject to

the above two restrictions and the no-Ponzi-game constraint (3.3). The

Lagrangian associated with this optimization problem is

∞
X   
1 i2
L= β t
U (ct ) + λt (1 + r)bt−1 + θt F (kt ) − ct − it − t − bt + qt (kt + it − kt+1 ) .
2 kt
t=0

The variables λt and qt denote Lagrange multipliers on the sequential budget

constraint and the law of motion of the capital stock, respectively. We

continue to assume that β(1 + r) = 1. The first-order conditions associated

with the household problem are:

it = (qt − 1)kt (3.6)

1
(1 + r)qt = θt+1 F 0 (kt+1 ) + (it+1 /kt+1 )2 + qt+1 (3.7)
2
32 Martı́n Uribe

kt+1 = kt + it
∞
r X θt+j F (kt+j ) − it+j − 2 (it+j /kt+j )
1 2
ct = rbt−1 + .
1+r (1 + r)j
j=0

The Lagrange multiplier qt represents the shadow relative price of capital

in terms of consumption goods, and is known as Tobin’s q. According to

equation (3.6), as qt increases agents have incentives to allocate goods to the

production of capital, thereby increasing it . Equation (3.7) compares the

rates of return on bonds and physical capital: Adding one unit of capital to

the existing stock costs qt . This unit yields θt+1 F 0 (kt+1 ) units of output next

period. In addition, an extra unit of capital reduces tomorrow’s adjustment

costs by (it+1 /kt+1 )2 /2. Finally, the unit of capital can be sold at a price qt+1

next period. The sum of these three sources of income form the right hand

side of (3.7). Alternatively, instead of using qt units of goods to buy one

unit of capital, the agent can engage in a financial investment by purchasing

qt units of bonds in period t with a gross return of (1 + r)qt . This is the

left-hand side of equation (3.7). At the optimum both strategies must yield

the same return.

3.2.1 Dynamics of the Capital Stock

Eliminating it from the above equations we get the following two first-order,

nonlinear difference equations in kt and qt :

kt+1 = qt kt (3.8)

θt+1 F 0 (qt kt ) + (qt+1 − 1)2 /2 + qt+1

qt = . (3.9)
1+r
Lectures in Open Economy Macroeconomics, Chapter 3 33

Figure 3.1: The Dynamics of the Capital Stock

q
Q

a
•
K K′
1 •

S′

Q′

k0 k* k

The perfect foresight solution to these equations is depicted in figure 3.1.

The horizontal line KK 0 corresponds to the pairs (kt , qt ) for which kt+1 = kt

in equation (3.8). That is,

q = 1. (3.10)

Above the locus KK 0 , the capital stock grows over time and below KK 0 the

capital stock declines over time. The locus QQ0 corresponds to the pairs

(kt , qt ) for which qt+1 = qt in equation (3.9). That is,

rq = θF 0 (qk) + (q − 1)2 /2. (3.11)

34 Martı́n Uribe

Jointly, equations (3.10) and (3.11) determine the steady-state value of the

capital stock, which we denote by k ∗ , and the steady-state value of the

Tobin’s q, 1. The value of k ∗ is implicitly determined by the expression r =

θF 0 (k), which is the same value obtained in the economy without adjustment

costs. This is not surprising, because, as noted earlier, adjustment costs

vanish in the steady state. For qt near unity, the locus QQ0 is downward

sloping. Above and to the right of QQ0 , q increases over time and below and

to the left of QQ0 , q decreases over time.

The system (3.8)-(3.9) is saddle-path stable. The locus SS 0 represents

the converging saddle path. If the initial capital stock is different from its

long-run level, both q and k converge monotonically to their steady states

along the saddle path.

3.2.2 A Permanent Technology Shock

Suppose now that in period 0 the technology factor θt increases permanently

from θ̄ to θ 0 > θ̄. It is clear from equation (3.10) that the locus KK 0 is not

affected by the productivity shock. On the other hand, it is clear from

equation (3.11) that the locus QQ0 shifts up and to the right. It follows

that in response to a permanent increase in productivity, the long-run level

of capital experiences a permanent increase. The price of capital, qt , on the

other hand, is not affected in the long run.

Consider now the transition to the new steady state. Suppose that the

steady-state value of capital prior to the innovation in productivity is k0

in figure 3.1. Then the new steady-state values of k and q are given by k ∗

and 1. In the period of the shock, the capital stock does not move. The
Lectures in Open Economy Macroeconomics, Chapter 3 35

price of installed capital, qt , jumps to the new saddle path, point a in the

figure. This increase in the price of installed capital induces an increase

in investment, which in turn makes capital grow over time. After the ini-

tial impact, qt decreases toward 1. Along this transition, the capital stock

increases monotonically towards its new steady-state k ∗ .

The equilibrium dynamics of output, investment, and capital in the pres-

ence of adjustment costs are quite different from those arising in the absence

thereof. In the frictionless environment, investment, output, and the capi-

tal stock all reach their long-run steady state one period after the produc-

tivity shock. Under capital adjustment costs, by contrast, these variables

adjust gradually to the unexpected increase in productivity. The more pro-

nounced are adjustment costs, the more sluggish is the response of invest-

ment, thereby making it less likely that the trade balance deteriorates in

response to a positive technology shock as required for the model to be in

line with the data.1

This observation opens the question of what would the model predict

for the behavior of the trade balance in response to output shocks when one

introduces a realistic degree of adjustment costs. We address this issue in

the next chapter.

1
It is straightforward to see that the response of the model to a purely temporary pro-
ductivity shock is identical as that of the model without adjustment costs. In particular,
capital and investment display a mute response.
36 Martı́n Uribe
Chapter 4

The Real Business Cycle

Model

In the previous two chapters, we arrived at the conclusion that a model

driven by productivity shocks can explain the observed countercyclicality

of the current account. We also established that two features of the model

are important in making this prediction possible. First, productivity shocks

must be sufficiently persistent. Second, capital adjustment costs must not

be too strong. In this chapter, we extend the model of the previous chapter

by allowing for three features that add realism to the model’s implied dy-

namics. Namely, endogenous labor supply and demand, uncertainty in the

technology shock process, and capital depreciation. The resulting theoret-

ical framework is known as the Real Business Cycle model, or, succinctly,

the RBC model.

37
38 Martı́n Uribe

4.1 The Model

Consider an economy populated by an infinite number of identical house-

holds with preferences described by the utility function

∞
X
E0 θt U (ct , ht ), (4.1)
t=0

θ0 = 1, (4.2)

θt+1 = β(ct , ht )θt t ≥ 0, (4.3)

where βc < 0, βh > 0. This preference specification was conceived by Uzawa

(1968) and introduced in the small-open-economy literature by Mendoza

(1991). The reason why we adopt this type of utility function here is that it

gives rise to a steady state of the model that is independent of initial condi-

tions. In particular, under these preferences the steady state is independent

of the initial net foreign asset position of the economy. This property is

desirable from a purely technical point of view because it makes it possible

to rely on linear approximations to characterize equilibrium dynamics.

The period-by-period budget constraint of the representative household

is given by

dt = (1 + rt−1 )dt−1 − yt + ct + it + Φ(kt+1 − kt ), (4.4)

where dt denotes the household’s debt position at the end of period t, rt

denotes the interest rate at which domestic residents can borrow in period

t, yt denotes domestic output, ct denotes consumption, it denotes gross

Lectures in Open Economy Macroeconomics, Chapter 4 39

investment, and kt denotes physical capital. The function Φ(·) is meant to

capture capital adjustment costs and is assumed to satisfy Φ(0) = Φ0 (0) =

0. As pointed out earlier, small open economy models typically include

capital adjustment costs to avoid excessive investment volatility in response

to variations in the foreign interest rate. The restrictions imposed on Φ

ensure that adjustment costs are nil in the steady state and that in the

steady state the interest rate equals the marginal product of capital net of

depreciation.

Output is produced by means of a linearly homogeneous production func-

tion that takes capital and labor services as inputs,

yt = At F (kt , ht ), (4.5)

where At is an exogenous stochastic productivity shock. The stock of capital

evolves according to

kt+1 = it + (1 − δ)kt , (4.6)

where δ ∈ (0, 1) denotes the rate of depreciation of physical capital.

Households choose processes {ct , ht , yt , it , kt+1 , dt , θt+1 }∞

t=0 so as to max-

imize the utility function (4.1) subject to (4.2)-(4.6) and a no-Ponzi con-

straint of the form

dt+j
lim Et Qj ≤ 0. (4.7)
j→∞
s=0 (1 + rs )

Letting θt ηt and θt λt denote the Lagrange multipliers on (4.3) and (4.4), the

first-order conditions of the household’s maximization problem are (4.3)-

40 Martı́n Uribe

(4.7) holding with equality and:

λt = β(ct , ht )(1 + rt )Et λt+1 (4.8)

λt = Uc (ct , ht ) − ηt βc (ct , ht ) (4.9)

ηt = −Et U (ct+1 , ht+1 ) + Et ηt+1 β(ct+1 , ht+1 ) (4.10)

−Uh (ct , ht ) + ηt βh (ct , ht ) = λt At Fh (kt , ht ) (4.11)

 
λt [1+Φ0 (kt+1 −kt )] = β(ct , ht )Et λt+1 At+1 Fk (kt+1 , ht+1 ) + 1 − δ + Φ0 (kt+2 − kt+1 )

(4.12)

These first-order conditions are fairly standard, except for the fact that the

marginal utility of consumption is not given simply by Uc (ct , ht ) but rather

by Uc (ct , ht )−βc (ct , ht )ηt . The second term in this expression reflects the fact

that an increase in current consumption lowers the discount factor (βc < 0).

In turn, a unit decline in the discount factor reduces utility in period t by ηt .

Intuitively, ηt equals the present discounted value of utility from period t + 1

onward. To see this, iterate the first-order condition (4.10) forward to ob-
P  
θt+j
tain: ηt = −Et ∞j=1 θt+1 U (ct+j , ht+j ). Similarly, the marginal disutility

of labor is not simply Uh (ct , ht ) but instead Uh (ct , ht ) − βh (ct , ht )ηt .

We assume free capital mobility. The world interest rate is assumed to

be constant and equal to r. Equating the domestic and world interest rates,

yields

rt = r. (4.13)
Lectures in Open Economy Macroeconomics, Chapter 4 41

The law of motion of the productivity shock is given by:

ln At+1 = ρ ln At + t+1 ; t+1 ∼ N IID(0, σ2 ); t ≥ 0. (4.14)

A competitive equilibrium is a set of processes {dt , ct , ht , yt , it , kt+1 , ηt , λt , rt }

satisfying (4.4)-(4.14), given the initial conditions A0 , d−1 , and k0 and the

exogenous process {t }.

We parameterize the model following Mendoza (1991), who uses the

following functional forms for preferences and technology:

 1−γ
c − ω −1 hω −1
U (c, h) =
1−γ

 −ψ1
β(c, h) = 1 + c − ω −1 hω

F (k, h) = k α h1−α

φ 2
Φ(x) = x ; φ > 0.
2

The assumed functional forms for the period utility function and the dis-

count factor imply that the marginal rate of substitution between consump-

tion and leisure depends only on labor. In effect, combining equations (4.9)

and (4.11) yields

hω−1
t = At Fh (kt , ht ). (4.15)

The right-hand side of this expression is the marginal product of labor,

which in equilibrium equals the real wage rate. The left-hand side is the

marginal rate of substitution of leisure for consumption. The above expres-

42 Martı́n Uribe

Table 4.1: Calibration of the Small Open RBC Economy

γ ω ψ1 α φ r δ ρ σ
2 1.455 .11 .32 0.028 0.04 0.1 0.42 0.0129

sion thus states that the labor supply depends only upon the wage rate and

in particular that it is independent of the level of wealth.

We also follow Mendoza (1991) in assigning values to the structural pa-

rameters of the model. Mendoza calibrates the model to the Canadian econ-

omy. The time unit is meant to be a year. The parameter values are shown

on table 4.1. All parameter values are standard in the real-business-cycle

literature. It is of interest to review the calibration of the parameter ψ1

defining the elasticity of the discount factor with respect to the composite

c − hω /ω. This parameter determines the stationarity of the model and the

speed of convergence to the steady state. The value assigned to ψ1 is set so

as to match the average Canadian trade-balance-to-GDP ratio. To see how

in steady state this ratio is linked to the value of ψ1 , use equation (4.12) in

steady state to get

  1
k α 1−α
= .
h r+δ

It follows from this expression that the steady-state capital-labor ratio is

independent of the parameter ψ1 . Given the capital-labor ratio, equilib-

rium condition (4.15) implies that the steady-state value of hours is also

independent of ψ1 and given by

  α  1
k ω−1
h = (1 − α) .
h
Lectures in Open Economy Macroeconomics, Chapter 4 43

Given the steady-state values of hours and the capital-labor ratio, we can

find directly the steady-state values of capital, investment (i = δk), and

output (y = k α h1−α ), independently of ψ1 . Now note that in the steady state

the trade balance, tb, is given by y − c − i. This expression and equilibrium

condition (4.8) imply the following steady-state condition relating the trade

balance to ψ1 : [1 + y − i − tb − hω /ω]−ψ1 (1 + r) = 1, which uses the specific

functional form assumed for the discount factor. The above expression can

be solved for the trade balance-to-output ratio to obtain:

 hω

tb i (1 + r)1/ψ1 + ω −1
=1− − .
y y y

Recalling that y, h, nd i are independent of ψ1 , it follows that this expression

can be solved for ψ1 given tb/y, given α, r, δ, and ω. Clearly, the larger is

the trade-balance-to-output ratio, the larger is ψ1 .

Approximating Equilibrium Dynamics

We look for solutions to the equilibrium conditions (4.4)-(4.14) where the

vector xt ≡ {dt−1 , ct , ht , yt , it , kt , ηt , λt , rt , At } fluctuates in a small neigh-

borhood around its nonstochastic steady-state level. Because in any such

solution the stock of debt is bounded, we have that the transversality con-

dition limj→∞ Et dt+j /(1 + r)j = 0 is always satisfied. We thus focus on

bounded solutions to the system (4.4)-(4.6) and (4.8)-(4.14) of ten equa-

tions in ten variables given by the elements of the vector xt . The system
44 Martı́n Uribe

can be written as

Et f (xt+1 , xt ) = 0.

This expression describes a system of nonlinear stochastic difference equa-

tions. Closed form solutions for this type of system are not typically avail-

able. We therefore must resort to an approximate solution.

There are a number of techniques that have been devised to solve dy-

namic systems like the one we are studying. The technique we will employ

here consists in applying a first-order Taylor expansion (i.e., linearizing) the

system of equations around the nonstochastic steady state. The resulting

linear system can be readily solved using well-established techniques.

Before linearizing the equilibrium conditions, we introduce a convenient

variable transformation. It is useful to express some variables in terms

of percent deviations from their steady-state value. This is the case, for

instance, with output or investment. For any such variable, say wt , we define

ŵt ≡ log(wt /w), where w denotes the steady-state value of wt . Note that

bt ≈ (wt − w)/w. Some

for small deviations of wt from w it is the case that w

other variables are more naturally expressed in levels. This is the case, for

instance, with net interest rates or variables that can take negative values,

such as the trade balance. For this type of variable, we define ŵt ≡ wt − w.

The linearized version of the equilibrium system can then be written as

Ab
xt+1 = Bb
xt ,

where A and B are square matrices conformable with xt . Appendix A

displays the linearized equilibrium conditions of the RBC model studied

Lectures in Open Economy Macroeconomics, Chapter 4 45

here. The vector x̂t contains 10 variables. Of these 10 variables, 3 are

state variables, namely, b

kt , dbt−1 , and A
bt . State variables are variables whose

values in any period t ≥ 0 are either predetermined (i.e., determined before

t) or determined in t but in an exogenous fashion. In our model, k̂t and

dˆt−1 are endogenous state variables and Ât is an exogenous state variable.

The remaining 7 elements of x̂t that is, ĉt , ĥt , λ̂t , η̂t , rt , bit , and yt , are

co-state variables. Co-states are endogenous variables whose values are not

predetermined in period t. All the coefficients of the linear system, that

is, the elements of A and B, are known functions of the deep structural

parameters of the model to which we assigned values when we calibrated

the model. The linearized system has three known initial conditions b
k0 , db−1 ,
b0 . To determine the initial value of the remaining seven variables, we
and A

impose a terminal condition requiring that at any point in time the system

be expected to converge to the nonstochastic steady state. Formally, the

terminal condition takes the form

lim |Et xt+j | = 0.

j→∞

Appendix B shows in some detail how to solve linear stochastic systems like

the one describing the dynamics of our linearized equilibrium conditions.

That appendix also shows how to compute second moments and impulse

response functions.
46 Martı́n Uribe

Table 4.2: Empirical and Theoretical Second Moments

Variable Canadian Data Model

σx t ρxt ,xt−1 ρxt ,GDPt σx t ρxt ,xt−1 ρxt ,GDPt
y 2.8 0.61 1 3.1 0.61 1
c 2.5 0.7 0.59 2.3 0.7 0.94
i 9.8 0.31 0.64 9.1 0.07 0.66
h 2 0.54 0.8 2.1 0.61 1
tb
y 1.9 0.66 -0.13 1.5 0.33 -0.012
ca
y 1.5 0.3 0.026

Note. Empirical moments are taken from Mendoza (1991). Stan-

dard deviations are measured in percentage points.

4.2 The Model’s Performance

Table 4.2 displays some unconditional second moments of interest implied

by our model. It should not come as a surprise that the model does very well

in replicating the volatility of output, the volatility of investment, and the

serial correlation of output. For we picked values for the parameters σ , φ,

and ρ so as to match these three moments. But the model performs relatively

well along other dimensions. For instance, it correctly implies a volatility

ranking featuring investment above output and output above consumption.

Also in line with the data is the model’s prediction of a countercyclical

trade balance-to-output ratio. On the downside, the model overestimates

the correlations of hours and consumption with output. Note in particular

that the implied correlation between hours and output is exactly unity. This

prediction is due to the assumed functional form for the period utility index.

In effect, equilibrium condition (4.15), equating the marginal product of

Lectures in Open Economy Macroeconomics, Chapter 4 47

Figure 4.1: Responses to a Positive Technology Shock

Consumption Output
2 1.5

1.5
1
1
0.5
0.5

0 0
0 2 4 6 8 10 0 2 4 6 8 10
Investment Hours
10 1.5

5 1

0 0.5

−5 0
0 2 4 6 8 10 0 2 4 6 8 10
Trade Balance / GDP Current Account / GDP
1 1

0.5 0.5

0 0

−0.5 −0.5

−1 −1
0 2 4 6 8 10 0 2 4 6 8 10

labor to the marginal rate of substitution between consumption and leisure,

can be written as hωt = (1−α)yt . The log-linearized version of this condition

is ωb
ht = ybt , which implies that b
ht and ybt are perfectly correlated.

Figure 4.1. displays the impulse response functions of a number of vari-

ables of interest to a technology shock of size 1 in period 0. The model

predicts an expansion in output, consumption, investment, and hours. The

increase in domestic absorption is larger than the increase in output, result-

ing in a deterioration of the trade balance.

48 Martı́n Uribe

Figure 4.2: Response of the Trade-Balance-To-Output Ratio to a Positive

Technology Shock
0.6
benchmark
high φ
low ρ
0.4

0.2

−0.2

−0.4

−0.6

−0.8
0 1 2 3 4 5 6 7 8 9

4.2.1 The Role of Capital Adjustment Costs

In previous chapters, we deduced that the negative response of the trade

balance to a positive technology shock was not a general implication of

the neoclassical model. In particular, two conditions must be met for the

model to generate a deterioration in the external accounts in response to

a mean-reverting improvement in total factor productivity. First, capital

adjustment costs must not be too stringent. Second, the productivity shock

must be sufficiently persistent. To illustrate this conclusion, figure 4.2 dis-

plays the impulse response function of the trade balance-to-GDP ratio to a

technology shock of unit size in period 0 under three alternative parameter

Lectures in Open Economy Macroeconomics, Chapter 4 49

specifications. The solid line reproduces the benchmark case from figure 4.1.

The broken line depicts an economy where the persistence of the produc-

tivity shock is half as large as in the benchmark economy (ρ = 0.21). In

this case, because the productivity shock is expected to die out quickly, the

response of investment is relatively weak. In addition, the temporariness

of the shock induces households to save most of the increase in income to

smooth consumption over time. As a result, the expansion in aggregate

domestic absorption is modest. At the same time, because the size of the

productivity shock is the same as in the benchmark economy, the initial

responses of output and hours are identical in both economies (recall that,

by equation (4.15), ht depends only on kt and At ). The combination of a

weak response in domestic absorption and an unchanged response in output,

results in an improvement in the trade balance when productivity shocks are

not very persistent.

The crossed line depicts the case of high capital adjustment costs. Here

the parameter φ equals 0.084, a value three times as large as in the bench-

mark case. In this environment, high adjustment costs discourage firms from

increasing investment spending by as much as in the benchmark economy.

As a result, the response of aggregate domestic demand is weaker, leading

to an improvement in the trade balance-to-output ratio.

4.3 Alternative Ways to Induce Stationarity

In the RBC model analyzed thus far households have endogenous discount

factors. We will refer to that model as the ‘internal discount factor model,’
50 Martı́n Uribe

or IDF model. The inclusion of an endogenous discount factor responds to

the need to obtain stationary dynamics up to first order. Had we assumed

a constant discount factor, the log-linearized equilibrium dynamics would

have contained a random walk component. Two problems emerge when

the linear approximation possesses a unit root. First, one can no longer

claim that the linear system behaves like the original nonlinear system—

which is ultimately the focus of interest—when the underlying shocks have

sufficiently small supports. Second, when the variables of interest contain

random walk elements, it is impossible to compute unconditional second

moments, such as standard deviations, serial correlations, correlations with

output, etc., which are the most common descriptive statistics of the business

cycle.

In this section, we analyze and compare alternative ways of inducing

stationarity in small open economy models. Our analysis follows closely

Schmitt-Grohé and Uribe (2003), but expands their analysis by including a

model with an internal interest-rate premium.

4.3.1 External Discount Factor (EDF)

Consider an alternative formulation of the endogenous discount factor model

where domestic agents do not internalize the fact that their discount fac-

tor depends on their own levels of consumption and effort. Alternatively,

suppose that the discount factor depends not upon the agent’s own con-

sumption and effort, but rather on the average per capita levels of these
Lectures in Open Economy Macroeconomics, Chapter 4 51

variables. Formally, preferences are described by (4.1), (4.2), and

θt+1 = β(c̃t , h̃t )θt t ≥ 0, (4.16)

where c̃t and h̃t denote average per capital consumption and hours, which

the individual household takes as given.

The first-order conditions of the household’s maximization problem are

(4.2), (4.4)-(4.7), (4.16) holding with equality and:

λt = β(c̃t , h̃t )(1 + rt )Et λt+1 (4.17)

λt = Uc (ct , ht ) (4.18)

−Uh (ct , ht ) = λt At Fh (kt , ht ) (4.19)

 
λt [1+Φ0 (kt+1 −kt )] = β(c̃t , h̃t )Et λt+1 At+1 Fk (kt+1 , ht+1 ) + 1 − δ + Φ0 (kt+2 − kt+1 )

(4.20)

In equilibrium, individual and average per capita levels of consumption and

effort are identical. That is,

ct = c̃t (4.21)

and

ht = h̃t . (4.22)

A competitive equilibrium is a set of processes {dt , ct , ht , c̃t , h̃t , yt , it ,

kt+1 , λt , rt , At } satisfying (4.4)-(4.7), (4.13), (4.14), (4.17)-(4.22) all hold-

ing with equality, given A0 , d−1 , and k0 and the stochastic process {t }.
52 Martı́n Uribe

Note that the equilibrium conditions include one Euler equation less, equa-

tion (4.10), and one variable less, ηt , than the standard endogenous-discount-

factor model of subsection 4.1. These fewer elements facilitate the compu-

tation of the equilibrium dynamics using perturbation methods.1

We evaluate the model using the same functional forms and parameter

values as in the IDF model.

4.3.2 External Debt-Elastic Interest Rate (EDEIR)

Under an external debt-elastic interest rate, stationarity is induced by as-

suming that the interest rate faced by domestic agents, rt , is increasing in

the aggregate level of foreign debt, which we denote by d˜t . Specifically, rt is

given by

rt = r + p(d˜t ), (4.23)

where r denotes the world interest rate and p(·) is a country-specific interest

rate premium. The function p(·) is assumed to be strictly increasing.

Preferences are given by equation (4.1). Unlike in the previous model,

preferences are assumed to display a constant subjective rate of discount.

Formally,

θt = β t ,

where β ∈ (0, 1) is a constant parameter.

1
It is remarkable that the degree of computational complexity is reversed when the
computational technique consists in iterating a Bellman equation over a discretized state
space. The economy with a noninternalized discount factor features an externality, which
complicates significantly the task of computing the equilibrium by value function iter-
ations. A similar comment applies to the computation of equilibrium in a model with
an interest-rate premium that depends on the aggregate level of extern! al debt to be
discussed in the next subsection.
Lectures in Open Economy Macroeconomics, Chapter 4 53

The representative agent’s first-order conditions are (4.4)-(4.7) holding

with equality and

λt = β(1 + rt )Et λt+1 (4.24)

Uc (ct , ht ) = λt , (4.25)

−Uh (ct , ht ) = λt At Fh (kt , ht ). (4.26)

 
λt [1+Φ0 (kt+1 −kt )] = βEt λt+1 At+1 Fk (kt+1 , ht+1 ) + 1 − δ + Φ0 (kt+2 − kt+1 ) .

(4.27)

Because agents are assumed to be identical, in equilibrium aggregate per

capita debt equals individual debt, that is,

d˜t = dt . (4.28)

A competitive equilibrium is a set of processes {dt , d˜t+1 , ct , ht , yt , it , kt+1 , rt , λt }∞

t=0

satisfying (4.4)-(4.7), and (4.23)-(4.28) all holding with equality, given (4.14),

A0 , d−1 , and k0 .

We adopt the same forms for the functions U , F , and Φ as in the IDF of

subsection 4.1. We use the following functional form for the risk premium:

 ¯

p(d) = ψ2 ed−d − 1 ,

where ψ2 and d¯ are constant parameters.

We calibrate the parameters γ, ω, α, φ, r, δ, ρ, and σ using the values

shown in table 4.1. We set the subjective discount factor equal to the world
54 Martı́n Uribe

Table 4.3: Model 2: Calibration of Parameters Not Shared With Model 1

β d¯ ψ2
0.96 0.7442 0.000742

interest rate; that is,

1
β= .
1+r

The parameter d¯ equals the steady-state level of foreign debt. To see this,

note that in steady state, the equilibrium conditions (4.23) and (4.24) to-

gether with the assumed form of the interest-rate premium imply that
h  i
¯
1 = β 1 + r + ψ2 ed−d − 1 . The fact that β(1 + r) = 1 then implies
¯ If follows that in the steady state the interest rate premium is
that d = d.

nil. We set d¯ so that the steady-state level of foreign debt equals the one

implied by Model 1. Finally, we set the parameter ψ2 so as to ensure that

this model and Model 1 generate the same volatility in the current-account-
¯ and ψ2 are given in Table 4.3.
to-GDP ratio. The resulting values of β, d,

4.3.3 Internal Debt-Elastic Interest Rate (IDEIR)

The model with an internal debt-elastic interest rate assumes that the in-

terest rate faced by domestic agents is increasing in the individual debt

position, dt . In all other aspects, the model is identical to the model featur-

ing an external debt-elastic interest rate. Formally, in the IDEIR model the

interest rate is given by

rt = r + p(dt ), (4.29)
Lectures in Open Economy Macroeconomics, Chapter 4 55

where, as before, r denotes the world interest rate and p(·) is a household-

specific interest-rate premium. In the household’s problem the only opti-

mality condition that changes relative to the case with an external premium

is the Euler equation for debt accumulation, which now takes the form

λt = β[1 + r + p(dt ) + p0 (dt )dt ]Et λt+1 . (4.30)

This expression features the derivative of the premium with respect to debt

because households internalize the fact that as they increase their debt po-

sitions, so does the interest rate they face in financial markets.

A competitive equilibrium is a set of processes {dt , ct , ht , yt , it , kt+1 , rt , λt }∞

t=0

satisfying (4.4)-(4.7), (4.25)-(4.27), (4.29), and (4.30), all holding with equal-

ity, given (4.14), A0 , d−1 , and k0 .

We assume the same functional forms and parameter values as in the

model with an external interest-rate premium. We note that in the model

analyzed in this subsection the steady-state level of debt is no longer equal

¯ Recalling that β(1 + r) = 1, the steady-state version of equation (4.30)
to d.

imposes the following restriction on d,

¯
(1 + d)ed−d = 1,

which, given d¯ = 0.7442, yields d = 0.4045212.

The fact that the steady-state debt is lower than d¯ implies that the

country premium is negative in the steady state. However, the marginal

country premium, given by ∂[ρ(dt )dt ]/∂dt , is nil in the steady state. An
56 Martı́n Uribe

¯ and adjust β to ensure

alternative calibration strategy is to impose d = d,

that equation (4.30) holds in steady state. In this case, the country premium

vanishes in the steady state, but the marginal premium is positive and equal
¯
to ψ2 d.

4.3.4 Portfolio Adjustment Costs (PAC)

In this model, stationarity is induced by assuming that agents face con-

vex costs of holding assets in quantities different from some long-run level.

Preferences and technology are as in the EDEIR model of section 4.3.2. In

contrast to what is assumed in the EDEIR model, here the interest rate at

which domestic households can borrow from the rest of the world is con-

stant and equal to the world interest, that is, equation (4.13) holds. The

sequential budget constraint of the household is given by

ψ3 ¯ 2,
dt = (1 + rt−1 )dt−1 − yt + ct + it + Φ(kt+1 − kt ) + (dt − d) (4.31)
2

where ψ3 and d¯ are constant parameters defining the portfolio adjustment

cost function. The first-order conditions associated with the household’s

maximization problem are (4.5)-(4.7), (4.25)-(4.27), (4.31) holding with

equality and
¯ = β(1 + rt )Et λt+1
λt [1 − ψ3 (dt − d)] (4.32)

This optimality condition states that if the household chooses to borrow an

additional unit, then current consumption increases by one unit minus the
¯ The value of this increase in
marginal portfolio adjustment cost ψ3 (dt − d).

consumption in terms of utility is given by the left-hand side of the above

Lectures in Open Economy Macroeconomics, Chapter 4 57

equation. Next period, the household must repay the additional unit of debt

plus interest. The value of this repayment in terms of today’s utility is given

by the right-hand side. At the optimum, the marginal benefit of a unit debt

increase must equal its marginal cost.

A competitive equilibrium is a set of processes {dt , ct , ht , yt , it , kt+1 , rt , λt }∞

t=0

satisfying (4.5)-(4.7), (4.13), (4.25)-(4.27), (4.31), and (4.32) all holding with

equality, given (4.14), A0 , d−1 , and k0 .

Preferences and technology are parameterized as in the EDEIR model.

The parameters γ, ω, α, φ, r, δ, ρ, and σ take the values displayed in ta-

ble 4.1. As in model 2, the subjective discount factor is assumed to satisfy

β(1+r) = 1. This assumption and equation (4.32) imply that the parameter

d¯ determines the steady-state level of foreign debt (d = d).

¯ We calibrate d¯ so

that the steady-state level of foreign debt equals the one implied by models

IDF, EDF, and EDEIR (see table 4.3). Finally, we assign the value 0.00074

to ψ3 , which ensures that this model and the IDF model of section 4.1 gener-

ate the same volatility in the current-account-to-GDP ratio. This parameter

value is almost identical to that assigned to ψ2 in the EDEIR model. This

is because the log-linearized versions of models 2 and 3 are almost identi-

cal. Indeed, the models share all equilibrium conditions but the resource

constraint (equations (4.4) and (4.31)), the Euler equations associated with

the optimal choice of foreign bonds (equations (4.24) and (4.32)), and the

interest rate faced by domestic households (equations (4.13) and (4.23)).

The log-linearized versions of the resource constraints are the same in both

models.! The log-linear approximation to the domestic interest rate is given

+ rt = ψ2 d(1 + r)−1 dbt in the EDEIR and by 1\

by 1\ + rt = 0 in the PAC
58 Martı́n Uribe

model. In turn, the log-linearized versions of the Euler equation for debt are
bt = ψ2 d(1 + r)−1 dbt + Et λ
λ bt+1 in model 2 and λ bt+1 in Model
bt = ψ3 ddbt + Et λ

3. It follows that for small values of ψ2 and ψ3 satisfying ψ2 = (1 + r)ψ3 the

EDEIR and PAC models imply similar dynamics.

4.3.5 Complete Asset Markets (CAM)

All model economies considered thus far feature incomplete asset markets.

In those models agents have access to a single financial asset that pays a

risk-free real rate of return. In the model studied in this subsection, agents

have access to a complete array of state-contingent claims. This assumption

per se induces stationarity in the equilibrium dynamics.

Preferences and technology are as in model 2. The period-by-period

budget constraint of the household is given by

Et rt+1 bt+1 = bt + yt − ct − it − Φ(kt+1 − kt ), (4.33)

where rt+1 is a stochastic discount factor such that the period-t price of

a random payment bt+1 in period t + 1 is given by Et rt+1 bt+1 . Note that

because Et rt+1 is the price in period t of an asset that pays 1 unit of good

in every state of period t + 1, it follows that 1/[Et rt+1 ] denotes the risk-free

real interest rate in period t. Households are assumed to be subject to a

no-Ponzi-game constraint of the form

lim Et qt+j bt+j ≥ 0, (4.34)

j→∞
Lectures in Open Economy Macroeconomics, Chapter 4 59

at all dates and under all contingencies. The variable qt represents the

stochastic discount factor between periods 0 and t, such that the period-0

price of a random payment bt in period t is given by E0 qt bt . We have that

qt satisfies

q t = r 1 r2 . . . r t ,

with q0 ≡ 1. The first-order conditions associated with the household’s max-

imization problem are (4.5), (4.6), (4.25)-(4.27), (4.33), and (4.34) holding

with equality and

λt rt+1 = βλt+1 . (4.35)

A difference between this expression and the Euler equations that arise in

the models with incomplete asset markets studied in previous sections is that

under complete markets in each period t there is one first-order condition

for each possible state in period t + 1, whereas under incomplete markets

the above Euler equation holds only in expectations.

In the rest of the world, agents have access to the same array of financial

assets as in the domestic economy. Consequently, one first-order condition

of the foreign household is an equation similar to (4.35). Letting starred

letters denote foreign variables or functions, we have

λ∗t rt+1 = βλ∗t+1 . (4.36)

Note that we are assuming that domestic and foreign households share the

same subjective discount factor. Combining the domestic and foreign Euler
60 Martı́n Uribe

equations—equations (4.35) and (4.36)—yields

λt+1 λ∗
= t+1 .
λt λ∗t

This expression holds at all dates and under all contingencies. This means

that the domestic marginal utility of consumption is proportional to its

foreign counterpart. Formally,

λt = ξλ∗t ,

where ξ is a constant parameter determining differences in wealth across

countries. We assume that the domestic economy is small. This means that

λ∗t must be taken as an exogenous variable. Because we are interested only

in the effects of domestic productivity shocks, we assume that λ∗t is constant

and equal to λ∗ , where λ∗ is a parameter. The above equilibrium condition

then becomes

λ t = ψ4 , (4.37)

where ψ4 ≡ ξλ∗ is a constant parameter.

A competitive equilibrium is a set of processes {ct , ht , yt , it , kt+1 , λt }∞

t=0

satisfying (4.5), (4.6), (4.25)-(4.27), and (4.37), given (4.14), A0 , and k0 .

The functions U , F , and Φ are parameterized as in the previous models.

The parameters γ, β, ω, α, φ, δ, ρ, and σ take the values displayed in

tables 4.1 and 4.3. The parameter ψ4 is set so as to ensure that the steady-

state level of consumption is the same in this model as in the IDF, EDF,

EDEIR, and PAC models.

Lectures in Open Economy Macroeconomics, Chapter 4 61

4.3.6 The Nonstationary Case (NC)

For comparison with the models considered thus far, in this section we de-

scribe a version of the small open economy model that displays no station-

arity. In this model (a) the discount factor is constant; (b) the interest

rate at which domestic agents borrow from the rest of the world is constant

(and equal to the subjective discount factor); (c) agents face no frictions

in adjusting the size of their portfolios; and (d) markets are incomplete in

the sense that domestic households have only access to a single risk-free

international bond. This specification of the model induces a random walk

component in the equilibrium marginal utility of consumption and the net

foreign asset position.

A competitive equilibrium in the nonstationary model is a set of processes

{dt , ct , ht , yt , it , kt+1 , rt , λt }∞
t=0 satisfying (4.4)-(4.7), (4.13), and (4.24)-

(4.27) all holding with equality, given (4.14), A0 , d−1 , and k0 . We calibrate

the model using the parameter values displayed in tables 4.1 and 4.3.

4.3.7 Quantitative Results

Table 4.4 displays a number of unconditional second moments of interest

implied by the IDF, EDF, EDEIR, IDEIR, PAC, CAM, and NC models.2

For all models, we compute the equilibrium dynamics by solving a log-

linear approximation to the set of equilibrium conditions. The appendix

shows the log-linear version of the IDF model of subsection 4.1. The Matlab

computer code used to compute the unconditional second moments and

2
Model 5 is nonstationary, and therefore does not have well defined unconditional
second moments.
62 Martı́n Uribe

impulse response functions for all models presented in this section is available

at www.econ.duke.edu/∼uribe.

The main result of this section is that regardless of how stationarity is

induced, the model’s predictions regarding second moments are virtually

identical. This result is evident from table 4.4. The only noticeable differ-

ence arises in the CAM model, the complete markets case, which as expected

predicts less volatile consumption. The low volatility of consumption in the

complete markets model introduces an additional difference between the pre-

dictions of this model and the IDF, EDF, EDEIR, IDEIR, and PAC models.

Because consumption is smoother in the CAM model, its role in determin-

ing the cyclicality of the trade balance is smaller. As a result, the CAM

model predicts that the correlation between output and the trade balance is

positive, whereas the models featuring incomplete asset markets imply that

it is negative.

Figure 4.3 demonstrates that all of the models being compared imply

virtually identical impulse response functions to a technology shock. Each

panel shows the impulse response of a particular variable in the six models.

For all variables but consumption and the trade-balance-to-GDP ratio, the

impulse response functions are so similar that to the naked eye the graph

appears to show just a single line. Again, the only small but noticeable

difference is given by the responses of consumption and the trade-balance-

to-GDP ratio in the complete markets model. In response to a positive

technology shock, consumption increases less when markets are complete

than when markets are incomplete. This in turn, leads to a smaller decline

in the trade balance in the period in which the technology shock occurs.
Lectures in Open Economy Macroeconomics, Chapter 4 63

Table 4.4: Implied Second Moments

IDF EDF IDEIR EDEIR PAC CAM

Volatilities:
std(yt ) 3.1 3.1 3.1 3.1 3.1 3.1
std(ct ) 2.3 2.3 2.5 2.7 2.7 1.9
std(it ) 9.1 9.1 9 9 9 9.1
std(ht ) 2.1 2.1 2.1 2.1 2.1 2.1
std( tbytt ) 1.5 1.5 1.6 1.8 1.8 1.6
std( ca
yt )
t
1.5 1.5 1.4 1.5 1.5
Serial Correlations:
corr(yt , yt−1 ) 0.61 0.61 0.62 0.62 0.62 0.61
corr(ct , ct−1 ) 0.7 0.7 0.76 0.78 0.78 0.61
corr(it , it−1 ) 0.07 0.07 0.068 0.069 0.069 0.07
corr(ht , ht−1 ) 0.61 0.61 0.62 0.62 0.62 0.61
corr( tbytt , tbyt−1
t−1
) 0.33 0.32 0.43 0.51 0.5 0.39
cat cat−1
corr( yt , yt−1 ) 0.3 0.3 0.31 0.32 0.32
Correlations with Output:
corr(ct , yt ) 0.94 0.94 0.89 0.84 0.85 1
corr(it , yt ) 0.66 0.66 0.68 0.67 0.67 0.66
corr(ht , yt ) 1 1 1 1 1 1
corr( tbytt , yt ) -0.012 -0.013 -0.036 -0.044 -0.043 0.13
corr( ca t
yt , yt ) 0.026 0.025 0.041 0.05 0.051

Note. Standard deviations are measured in percent per year.

IDF = Internal Discount Factor; EDF = External Discount Fac-
tor; IDEIR = Internal Debt-Elastic Interest Rate; EDEIR = Ex-
ternal Debt-Elastic Interest Rate; PAC = Portfolio Adjustment
Costs; CAM = Complete Asset Markets. NC = Nonstationary
Case.
64 Martı́n Uribe

Figure 4.3: Impulse Response to a Unit Technology Shock in Models 1 - 5

Output Consumption
2 1.5

1.5
1
1
0.5
0.5

0 0
0 2 4 6 8 10 0 2 4 6 8 10
Investment Hours
10 1.5

5 1

0 0.5

−5 0
0 2 4 6 8 10 0 2 4 6 8 10
Trade Balance / GDP Current Account / GDP
1 1

0.5 0.5

0 0

−0.5 −0.5

−1 −1
0 2 4 6 8 10 0 2 4 6 8 10

Note. Solid line: Endogenous discount factor model; Squares:

Endogenous discount factor model without internalization;
Dashed line: Debt-elastic interest rate model; Dash-dotted line:
Portfolio adjustment cost model; Dotted line: complete asset
markets model; Circles: Model without stationarity inducing el-
ements.
Lectures in Open Economy Macroeconomics, Chapter 4 65

4.4 Appendix A: Log-Linearization

bt ≡ log(xt /x̄) denote the log-deviation of xt from its steady-state value.

Let x

Then, taking a first-order log-linearization of the model we get:

r
stb dbt = stb rbt−1 + stb (1 + r)dbt−1 − r[b ct − sibit ]
y t − sc b
1+r

ybt = A kt + (1 − α)b
bt + αb ht

b
kt+1 = (1 − δ)b
kt + δbit

bt = r
λ ct + βhb
rbt + βc b bt+1
ht + E t λ
1+r

bt = (1 − β)c ββc
λ ct + chb
[ccb ht ] − [b ct + βchb
ηt + βccb ht ]
(1 − β)c − ββc (1 − β)c − ββc

ct+1 + h Et b
ηbt = (1 − β)[c Et b ct + βhb
ht+1 ] + β[Et ηbt+1 + βc b ht ]

(1 − β)h ββh
ct +hhb
[hc b ht ]+ ct +βhhb
ηt +βhcb
[b bt +A
ht ] = λ bt +αb
kt −αb
ht
(1 − β)h + ββh (1 − β)h + ββh

bt + φk b
λ kt+1 − φk b ct + βhb
kt = βc b bt+1 + β(β −1 + δ − 1)[Et A
ht + E t λ bt+1

+(1 − α)Et b kt+1 + βφkEt b

ht+1 − (1 − α)b kt+2 − βφk b
kt+1

rbt = 0

Ât = ρÂt−1 + t ,

where βc ≡ cβc /β, βh ≡ hβh /β, βcc ≡ cβcc /βc , βch ≡ hβch /βc , c ≡

cUc /U , cc = cUcc /Uc , ch = hUch /Uc , stb ≡ tb/y, sc ≡ c/y, si = i/y.

In the log-linearization we are using the particular forms assumed for the
66 Martı́n Uribe

production function and the capital adjustment cost function.

Lectures in Open Economy Macroeconomics, Chapter 4 67

4.5 Appendix B: Solving Dynamic General Equi-

librium Models

The equilibrium conditions of the simple real business cycle model we stud-

ied in the previous chapter takes the form of a nonlinear stochastic vector

difference equation. Reduced forms of this sort are common in Macroeco-

nomics. A problem that one must face is that, in general, it is impossible

to solve such systems. But fortunately one can obtain good approximations

to the true solution in relatively easy ways. In the previous chapter, we

introduced one particular strategy, consisting in linearizing the equilibrium

conditions around the nonstochastic steady state. Here we explain in detail

how to solve the resulting system of linear stochastic difference equations.

In addition, we show how to use the solution to compute second moments

and impulse response functions.

The equilibrium conditions of a wide variety of dynamic stochastic gen-

eral equilibrium models can be written in the form of a nonlinear stochastic

vector difference equation

Et f (yt+1 , yt , xt+1 , xt ) = 0, (4.38)

where Et denotes the mathematical expectations operator conditional on

information available at time t. The vector xt denotes predetermined (or

state) variables and the vector yt denotes nonpredetermined (or control)

variables. The initial value of the state vector x0 is an initial condition

for the economy. (Beyond the initial condition, the complete set of equi-
68 Martı́n Uribe

librium conditions also includes a terminal condition, like a no-Ponzi game

constraint. We omit such a constraint here because we focus on approx-

imating stationary solutions.) The state vector xt can be partitioned as

xt = [x1t ; x2t ]0 . The vector x1t consists of endogenous predetermined state

variables and the vector x2t of exogenous state variables. Specifically, we

assume that x2t follows the exogenous stochastic process given by

x2t+1 = h̃(x2t , σ) + η̃σt+1 ,

where both the vector x2t and the innovation t are of order n × 1.3 The

vector t is assumed to have a bounded support and to be independently

and identically distributed, with mean zero and variance/covariance matrix

I. The eigenvalues of the Jacobian of the function h̃ with respect to its first

argument evaluated at the non-stochastic steady state are assumed to lie

within the unit circle.

The solution to models belonging to the class given in equation (4.38) is

of the form:

yt = ĝ(xt ) (4.39)

and

xt+1 = ĥ(xt ) + ησt+1 . (4.40)

The vector xt of predetermined variables is of size nx × 1 and the vector yt

of nonpredetermined variables is of size ny × 1. We define n = nx + ny . The

function f then maps Rny × Rny × Rnx × Rnx into Rn .

3
It is straightforward to accommodate the case in which the size of the innovations
vector t is different from that of x2t .
Lectures in Open Economy Macroeconomics, Chapter 4 69

The matrix η is of order nx × n and is given by

 
 ∅ 
η =  .
η̃

The shape of the functions ĥ and ĝ will in general depend on the amount

of uncertainty in the economy. The key idea of perturbation methods is to

interpret the solution to the model as a function of the state vector xt and

of the parameter σ scaling the amount of uncertainty in the economy, that

is,

yt = g(xt , σ) (4.41)

and

xt+1 = h(xt , σ) + ησt+1 , (4.42)

where the function g maps Rnx × R+ into Rny and the function h maps

Rnx × R+ into Rnx .

Given this interpretation, a perturbation methods finds a local approx-

imation of the functions g and h. By a local approximation, we mean an

approximation that is valid in the neighborhood of a particular point (x̄, σ̄).

Taking a Taylor series approximation of the functions g and h around the

point (x, σ) = (x̄, σ̄) we have (for the moment to keep the notation simple,
70 Martı́n Uribe

let’s assume that nx=ny=1)

g(x, σ) = g(x̄, σ̄) + gx (x̄, σ̄)(x − x̄) + gσ (x̄, σ̄)(σ − σ̄)

1
+ gxx (x̄, σ̄)(x − x̄)2 + gxσ (x̄, σ̄)(x − x̄)(σ − σ̄)
2
1
+ gσσ (x̄, σ̄)(σ − σ̄)2 + . . .
2
h(x, σ) = h(x̄, σ̄) + hx (x̄, σ̄)(x − x̄) + hσ (x̄, σ̄)(σ − σ̄)
1
+ hxx (x̄, σ̄)(x − x̄)2
2
+hxσ (x̄, σ̄)(x − x̄)(σ − σ̄)
1
+ hσσ (x̄, σ̄)(σ − σ̄)2 + . . . ,
2

The unknowns of an nth order expansion are the n-th order derivatives of

the functions g and h evaluated at the point (x̄, σ̄).

To identify these derivatives, substitute the proposed solution given by

equations (4.41) and (4.42) into equation (4.38), and define

F (x, σ) ≡ Et f (g(h(x, σ) + ησ0 , σ), g(x, σ), h(x, σ) + ησ0 , x) (4.43)

= 0.

Here we are dropping time subscripts. We use a prime to indicate variables

dated in period t + 1.

Because F (x, σ) must be equal to zero for any possible values of x and

σ, it must be the case that the derivatives of any order of F must also be

equal to zero. Formally,

Fxk σj (x, σ) = 0 ∀x, σ, j, k, (4.44)

Lectures in Open Economy Macroeconomics, Chapter 4 71

where Fxk σj (x, σ) denotes the derivative of F with respect to x taken k times

and with respect to σ taken j times.

As will become clear below, a particularly convenient point to approxi-

mate the functions g and h around is the non-stochastic steady state, xt = x̄

and σ = 0. We define the non-stochastic steady state as vectors (x̄, ȳ) such

that

f (ȳ, ȳ, x̄, x̄) = 0.

It is clear that ȳ = g(x̄, 0) and x̄ = h(x̄, 0). To see this, note that if σ = 0,

then Et f = f . The reason why the steady state is a particularly convenient

point is that in most cases it is possible to solve for the steady state. With

the steady state values in hand, one can then find the derivatives of the

function F .

We are looking for approximations to g and h around the point (x, σ) =

(x̄, 0) of the form

g(x, σ) = g(x̄, 0) + gx (x̄, 0)(x − x̄) + gσ (x̄, 0)σ

h(x, σ) = h(x̄, 0) + hx (x̄, 0)(x − x̄) + hσ (x̄, 0)σ

As explained earlier,

g(x̄, 0) = ȳ

and

h(x̄, 0) = x̄.

The remaining unknown coefficients of the first-order approximation to g

72 Martı́n Uribe

and h are identified by using the fact that, by equation (4.44), it must be

the case that:

Fσ (x̄, 0) = 0.

and

Fx (x̄, 0) = 0

To find those derivatives let’s repeat equation (4.43)

F (x, σ) ≡ Et f (g(h(x, σ) + ησ0 , σ), g(x, σ), h(x, σ) + ησ0 , x)

= 0.

Taking derivative with respect to the scalar σ we find:

Fσ (x̄, 0) = Et fy0 [gx (hσ + η0 ) + gσ ] + fy gσ + fx0 (hσ + η0 )

= fy0 [gx hσ + gσ ] + fy gσ + fx0 hσ

This is a system of n equations. Then imposing

Fσ (x̄, 0) = 0.

one can identify gσ and hσ :

 
   hσ 
fy0 gx + fx0 fy 0 + f y  =0
gσ
Lectures in Open Economy Macroeconomics, Chapter 4 73

This equation is linear and homogeneous in gσ and hσ . Thus, if a unique

solution exists, we have that

hσ = 0.

and

gσ = 0.

These two expressions represent an important theoretical result. They show

that in general, up to first order, one need not correct the constant term of

the approximation to the policy function for the size of the variance of the

shocks.

This result implies that in a first-order approximation the expected val-

ues of xt and yt are equal to their non-stochastic steady-state values x̄ and

ȳ. In this sense, we can say that in a first-order approximation the certainty

equivalence principle holds, that is, the policy function is independent of

the variance-covariance matrix of t . This is an important limitation of

first-order perturbation techniques. Because in many economic applications

we are interested in finding the effect of uncertainty on the economy. For

example, up to first-order the mean of the rate of return of all all assets

must be same. Thus, first-order approximation techniques cannot be used

to study risk premia. Another important question that can in general not be

addressed with first-order perturbation techniques is how uncertainty affects

welfare. This question is at the heart of the recent literature on optimal fiscal

and monetary stabilization policy. Because in a first-order accurate solution

the unconditional expectation of a variable is equal to the non-stochastic

steady state, any two policies that give rise to the same steady state yield,
74 Martı́n Uribe

up to first-order the same level of welfare.

To find gx and hx differentiate (4.43) with respect to x to obtain the

following system

Fx (x̄, 0) = fy0 gx hx + fy gx + fx0 hx + fx

Note that the derivatives of f evaluated at (y 0 , y, x0 , x) = (ȳ, ȳ, x̄, x̄) are

known. The above expression represents a system of n × nx quadratic equa-

tions in the n × nx unknowns given by the elements of gx and hx . Imposing

Fx (x̄, 0) = 0

the above expression can be written as:

   
 I   I 
[fx0 fy 0 ]   hx = −[fx fy ]  
gx gx

Let A = [fx0 fy0 ] and B = −[fx fy ]. Note that both A and B are known.

Let x̂t ≡ xt − x̄, then postmultiplying the above system equation ??) by

x̂t we obtain:    
 I   I 
A  hx x̂t = B   x̂t
gx gx

Consider for the moment, a perfect foresight equilibrium. In this case,

hx x̂t = x̂t+1 .    
 I   I 
A  x̂t+1 = B   x̂t
gx gx
Lectures in Open Economy Macroeconomics, Chapter 4 75

We are interested in solutions in which

lim |x̂t | < ∞

t→∞

We will use this limiting conditions to find the matrix gx . In particular,

we will use the Schur decomposition method.

To solve the above system, we use the generalized Schur decomposition

of the matrices A and B.4 The generalized Schur decomposition of A and

B is given by upper triangular matrices a and b and orthonormal matrices

q and z satisfying:5

qAz = a

and

qBz = b.

Let

st ≡ z 0 [I; gx ]x̂t .

Then we have that

ast+1 = bst

Now partition a, b, z, and st as

       
 a11 a12   b11 b12   z11 z12    s1t
a= , b =  ; z =   ; st =  ,
0 a22 0 b22 z21 z22 2
st

4
More formal descriptions of the method can be found in Sims (1996) and Klein (2000).
5
Recall that a matrix a is said to be upper triangular if elements aij = 0 for i > j. A
matrix z is orthonormal if z 0 z = zz 0 = I.
76 Martı́n Uribe

where a22 and b22 are of order ny × ny , z12 is of order nx × ny , and s2t is of

order ny × 1. Then we have that

a22 s2t+1 = b22 s2t ,

or

b−1 2 2
22 a22 st+1 = st .

Assume, without loss of generality, that the ratios abs(aii /bii ) are decreasing

in i. Suppose further that the number of ratios less than unity is exactly

equal to the number of control variables, ny , and that the number of ratios

greater than one is equal to the number of state variables, nx . By construc-

tion, the eigenvalues of b−1 6

22 a22 are all less than unity in modulus. Thus,

the requirement limj→∞ |s2t+j | < ∞ is satisfied only if s2t = 0. In turn, by

the definition of s2t , this restriction implies that

0 0
(z12 + z22 gx )x̂t = 0.

Because this condition has to hold for any value of the state vector, x̂t , it

follows that it must be the case that

0 0
z12 + z22 gx = 0.

6
Here we are applying a number of properties of upper triangular matrices. Namely,
(a) The inverse of a nonsingular upper triangular matrix is upper triangular. (b) the
product of two upper triangular matrices is upper triangular. (c) The eigenvalues of an
upper triangular matrix are the elements of its main diagonal.
Lectures in Open Economy Macroeconomics, Chapter 4 77

Solving this expression for gx yields

0 −1 0
gx = −z22 z12 .

The fact that s2t = 0 also implies that

a11 s1t+1 = b11 s1t ,

or

s1t+1 = a11 −1 b11 s1t

Now

s1t = (z11
0 0
+ z21 gx)x̂t .

Replacing gx , we have

−1 0
s1t = [z11
0 0 0
− z21 z22 z12 ]x̂t .

Combining this expression with the equation describing the evolution of st

shown two lines above, we get

0 0 0 −1 0 −1 −1 0
x̂t+1 = [z11 − z21 z22 z12 ] a11 −1 b11 [z11
0 0 0
− z21 z22 z12 ]x̂t ;

so that

0 0 0 −1 0 −1 −1 0
hx = [z11 − z21 z22 z12 ] a11 −1 b11 [z11
0 0 0
− z21 z22 z12 ].
78 Martı́n Uribe

We can simplify this expression for hx by using the following restrictions:

 
0 z
z11 + 0 z
z21 0 z
z11 + 0 z
z21
 11 21 12 22 
I = z0z =  
0 z + z0 z
z12 0 0
11 22 21 z12 z12 + z22 z22

to write:7

hx = z11 a−1 −1
11 b11 z11 .

4.6 Local Existence and Uniqueness of Equilib-

rium

In the above discussion, we assumed that the number of eigenvalues of D

with modulus less than unity is exactly equal to the number of control

variables, ny , and that the number of eigenvalues of D with modulus greater

than one is equal to the number of state variables, nx . In this case there is a

unique local equilibrium. But not for every economy this is the case. Let’s

first consider the case that the number of eigenvalues of D with modulus

greater than unity is equal to m < ny, which is less than the number of

control variables. Then the requirement that we wish to study equilibria

in which limj→∞ Et |x̂t+j | < ∞ will only yield m restrictions, rather than

ny restrictions. It follows that one can choose arbitrary initial values for

ny − m elements of y0 and the resulting first order solution will still be

7
To obtain this simple expression for hx , use element (2, 1) of z 0 z to get z12 0
z11 =
0 0 −1 −1 0 −1 0
−z22 z21 . Premultiply by z22 and post multiply by z11 to get z22 z12 = −z21 z11 −1 .
0 −1 0
Use this expression to eliminate z22 z12 From the square bracket in the expression for
0 0
hx . Then this square bracket becomes [z11 + z21 z21 z11 −1 ]. Now use element (1, 1) of z 0 z
0 0 0
to write z21 z21 = I − z11 z11 . Using this equation to eliminate z21 z21 from the expression
0 0 −1
in square brackets, we get [z11 + (I − z11 z11 )z11 −1 ], which is simply z11 .
Lectures in Open Economy Macroeconomics, Chapter 4 79

expected to converge back to the steady state. In this case the equilibrium

is indeterminate.

On the other hand, if the number of eigenvalues of D with modulus

greater than unity is greater than the number of control variables, ny , then

no local equilibrium exists. Let again m denote the number of eigenvalues of

D greater than unity in modulus and assume that m > ny . Then in order to

ensure that lim Et |x̂t+j | < ∞ we must set m elements of [x0 y0 ] equal to zero.

This implies that m − ny elements of x0 must be functions of the remaining

nx − (m − ny ) elements. But this can never be the case, because x0 is a

vector of predetermined or exogenous variables and therefore its elements

can take arbitrary values. In this case, we say no local equilibrium exists.

4.7 Second Moments

Start with the equilibrium law of motion of the deviation of the state vector

with respect to its steady-state value, which is given by

x̂t+1 = hx x̂t + σηt+1 , (4.45)

Covariance Matrix of xt

Let

Σx ≡ E x̂t x̂0t

denote the unconditional variance/covariance matrix of x̂t and let

Σ ≡ σ 2 ηη 0 .
80 Martı́n Uribe

Then we have that

Σx = hx Σx h0x + Σ .

We will describe two numerical methods to compute Σx .

Method 1

One way to obtain Σx is to make use of the following useful result. Let A,

B, and C be matrices whose dimensions are such that the product ABC

exists. Then

vec(ABC) = (C 0 ⊗ A) · vec(B),

where the vec operator transforms a matrix into a vector by stacking its

columns, and the symbol ⊗ denotes the Kronecker product. Thus if the vec

operator is applied to both sides of

Σx = hx Σx h0x + Σ ,

the result is

vec(Σx ) = vec(hx Σx h0x ) + vec(Σ )

= F vec(Σx ) + vec(Σ ),

where

F = hx ⊗ hx .
Lectures in Open Economy Macroeconomics, Chapter 4 81

Solving the above expression for vec(Σx ) we obtain

vec(Σx ) = (I − F)−1 vec(Σ )

provided that the inverse of (I −F) exists. The eigenvalues of F are products

of the eigenvalues of the matrix hx . Because all eigenvalues of the matrix hx

have by construction modulus less than one, it follows that all eigenvalues

of F are less than one in modulus. This implies that (I − F) is nonsingular

and we can indeed solve for Σx . One possible drawback of this method is

that one has to invert a matrix that has dimension n2x × n2x .

Method 2

The following iterative procedure, called doubling algorithm, may be faster

than the one described above in cases in which the number of state variables

(nx ) is large.

Σx,t+1 = hx,t Σx,t h0x,t + Σ,t

hx,t+1 = hx,t hx,t

Σ,t+1 = hx,t Σ,t h0x,t + Σ,t

Σx,0 = I

hx,0 = hx

Σ,0 = Σ
82 Martı́n Uribe

Other second moments

Once the covariance matrix of the state vector, xt has been computed, it

is easy to find other second moments of interest. Consider for instance the

covariance matrix E x̂t x̂0t−j for j > 0. Let µt = σηt .

j−1
X
E x̂t x̂0t−j = E[hjx x̂t−j + hkx µt−k ]x̂0t−j
k=0
j 0
= hx E x̂t−j x̂t−j

= hjx Σx

Similarly, consider the variance covariance matrix of linear combinations of

the state vector xt . For instance, the co-state, or control vector yt is given

by yt = ȳ + gx (xt − x̄), which we can write as: ŷt = gx x̂t . Then

E ŷt ŷt0 = Egx x̂t x̂0t gx0

= gx [E x̂t x̂0t ]gx0

= gx Σx gx0

and, more generally,

0
E ŷt ŷt−j = gx [E x̂t x̂0t−j ]gx0

= gx hjx Σx gx0 ,

for j ≥ 0.
Lectures in Open Economy Macroeconomics, Chapter 4 83

4.8 Impulse Response Functions

The impulse response to a variable, say zt in period t + j to an impulse in

period t is defined as:

IR(zt+j ) ≡ Et zt+j − Et−1 zt+j

The impulse response function traces the expected behavior of the system

from period t on given information available in period t, relative to what

was expected at time t − 1. Using the law of motion Et x̂t+1 = hx x̂t for the

state vector, letting x denote the innovation to the state vector in period 0,

that is, x = ησ0 , and applying the law of iterated expectations we get that

the impulse response of the state vector in period t is given by

IR(x̂t ) ≡ E0 x̂t − E−1 x̂t = htx [x0 − E−1 x0 ] = htx [ησ0 ] = htx x; t ≥ 0.

The response of the vector of controls yˆt is given by

IR(ŷt ) = gx htx x.

4.9 Matlab Code For Linear Perturbation Meth-

ods

Stephanie Schmitt-Grohé and I have written a suite of programs that are

posted on the courses webpage: www.econ.duke.edu/~uribe/2nd_order.htm.

The program gx_hx.m computes the matrices gx and hx using the Schur de-
84 Martı́n Uribe

composition method. The program mom.m computes second moments. The

program ir.m computes impulse response functions.

4.10 Higher Order Approximations

In this chapter, we focused on a first-order approximation to the solu-

tion of a nonlinear system of stochastic difference equations of the form

Et f (xt+1 , xt ) = 0.. But higher order approximations are relatively easy to

obtain. Indeed, there is a sense in which higher order approximations are

simpler than the first order approximation. Namely, obtaining a higher-

order approximation to the solution of the non-linear system is a sequential

procedure. Specifically, the coefficients of the ith term of the jth-order ap-

proximation are given by the coefficients of the ith term of the ith order

approximation, for j > 1 and i < j. So if the first-order approximation

to the solution is available, then obtaining the second order approximation

requires only to compute the coefficients of the quadratic terms, since the

coefficients of the linear terms are those of the first order approximation.

More importantly, obtaining the coefficients of the ith order terms of the

appro! ximate solution given all lower-order coefficients involves solving a

linear system of equations.

Schmitt-Grohé and Uribe (2004) describe in detail how to obtain a

second-order approximation to the solution of the nonlinear system and

provide MATLAB code that implements the approximation.

Lectures in Open Economy Macroeconomics, Chapter 4 85

4.11 Exercise

4.11.1 An RBC Small Open Economy with an internal debt-

elastic interest-rate premium

Consider RBC open economy model with a debt elastic interest rate pre-

mium studied in Schmitt-Grohé and Uribe (SGU) (JIE, 2003, section 3,

model 2). Modify the model by assuming that agents internalize the depen-

dence of the interest rate premium on the level of debt. Specifically, suppose

that the function p(·) depends upon the individual debt position, dt , rather

than on the aggregate per capita level of debt, d˜t .

1. Derive the model’s equilibrium conditions.

2. Use the same forms for the functions U , F , Φ, and p as in SGU.

¯ and ψ2 using
Calibrate the parameters γ, ω, α, φ, r, δ, ρ, σ , β, d,

tables 1 and 2 in SGU. Calculate the model’s nonstochastic steady

state and compare it to that of model 2 in SGU.

3. Compute the unconditional standard deviation, serial correlation, and

correlation with output of output, consumption, investment, hours, the

trade balance-to-output ratio, and the current account-to-output ratio

implied by the model. Compare these statistics to those associated

with model 2 in SGU (reported in their table 3).

4. Compute the impulse response functions of output, consumption, in-

vestment, hours, the trade balance-to-output ratio, and the current

account-to-output ratio implied by the model. Compare these im-

86 Martı́n Uribe

pulse responses to those associated with model 2 in SGU (shown in

their figure 1).

Hint: You might find it convenient to use as a basis the matlab code associ-

ated with the SGU paper, located at www.econ.duke.edu/~uribe/closing.htm

Chapter 5

The Terms of Trade

Three key stylized facts documented in chapter 1 are: (1) that emerging

market economies are about twice as volatile as developed economies; (2)

that private consumption spending is more volatile than output in emerg-

ing countries, but less volatile than output in developed countries; and (3)

that the trade-balance-to-output ratio is significantly more countercyclical

in emerging markets than it is in developed countries. Explaining this strik-

ing contrast between emerging and industrialized economies is at the top

of the research agenda in small-open-economy macroeconomics. Broadly,

the available theoretical explanations fall into two categories: One is that

emerging market economies are subject to more volatile shocks than are

developed countries. The second category of explanations argues that in

emerging countries government policy tends to amplify business-cycle fluc-

tuations whereas in developed countries public policy tends to mitigate ag-

gregate instability. This and the following two chapters provide a progress

report on the identification and quantification of exogenous sources of busi-

87
88 Martı́n Uribe

ness cycles in small open economies. The present chapter concentrates on

terms-of-trade shocks.

5.1 Defining the Terms of Trade

The terms of trade are defined as the relative price of exports in terms of

imports. Letting Ptx and Ptm denote indices of world prices of exports and

imports for a particular country, the terms of trade for that country are

given by tott ≡ Ptx /Ptm .

Typically, emerging countries specialize in exports of a few primary com-

modities, such as metals, agricultural products, or oil. At the same time,

emerging countries are normally small players in the world markets for the

goods they export or import. It follows that for many small countries, the

terms of trade can be regarded as an exogenous source of aggregate fluctu-

ations. Because primary commodities display large fluctuations over time,

the terms of trade have the potential to be an important source of business

cycles in developing countries.

5.2 Empirical Regularities

Table 5.1 displays summary statistics relating the terms of trade to output,

the components of aggregate demand, and the real exchange rate in the

postwar era. In the table, the real exchange rate (rer) is defined as the

relative price of consumption in terms of importable goods. Specifically, let

Ptc denote a domestic CPI index. Then the real exchange rate is given by

Ptc /Ptm . A number of empirical regularities emerge from the table:

Lectures in Open Economy Macroeconomics, Chapter 5 89

Table 5.1: The Terms of Trade and Business Cycles

Summary Developed Developing Oil Exporting
Statistic Countries Countries Countries
σ(tot) 4.70 10.0 18.0
ρ(tott , tott−1 ) 0.47 0.40 0.50
σ(tot)/σ(y) 0.52 0.77 1.40
ρ(tot, y) 0.78 0.39 0.30
ρ(tot, c) 0.74 0.34 0.19
ρ(tot, i) 0.67 0.38 0.45
ρ(tot, tb) 0.24 0.28 0.33
ρ(tot, rer) 0.70 0.07 0.42

Source: Mendoza (1995), tables 1 and 3-6.

Note: tot, y, c, i, and tb denote, respectively, the terms of trade, out-
put, consumption, investment, and the trade balance. The sample is
1955 to 1990 at annual frequency. The terms of trade are measured
as the ratio of export to import unit values with 1900=100. All other
variables are measured per capita at constant import prices. All vari-
ables are expressed in percent deviations from a HP trend constructed
using a smoothing parameter of 100. The group of developed coun-
tries is formed by the US, UK, France, Germany, Italy, Canada, and
Japan. The group of developing countries is formed by Argentina,
Brazil, Chile, Mexico, Peru, Venezuela, Taiwan, India, Indonesia, Ko-
rea, Philippines, and Thailand. The group of oil-exporting countries is
formed by Mexico, Venezuela, Saudi Arabia, Algeria, Cameroon, and
Nigeria.
90 Martı́n Uribe

1. The terms of trade are twice as volatile in emerging countries as in de-

veloped countries, and they are almost twice as volatile in oil-exporting

countries as in developing countries.

2. The terms of trade are half as volatile as output in developed coun-

tries, 75 percent as volatile as output in developing countries, and 150

percent as volatile as output in oil-exporting countries.

3. The terms of trade are procyclical. They are twice as procyclical in

developed countries as in developing countries.

4. The terms of trade display positive but small serial correlation.

5. The correlation between the terms of trade and the trade balance is

positive but small.

6. The terms of trade are positively correlated with the real exchange

rate. This correlation is high for developed countries but almost nil

for less developed countries.

The information provided in table 5.1 is mute on the importance of terms

of trade shocks in explaining movements in aggregate activity. Later in this

chapter, we attempt to answer this question by combining the empirical

information contained in table 5.1 with the theoretical predictions of a fully

specified dynamic general equilibrium model of the open economy.

5.2.1 TOT-TB Correlation: Two Early Explanations

The effects of terms-of-trade shocks on the trade balance is an old subject of

investigation. More than half a century ago, Harberger (1950) and Laursen
Lectures in Open Economy Macroeconomics, Chapter 5 91

and Metzler (1950) formalized, within the context of a keynesian model, the

conclusion that rising terms of trade should be associated with an improving

trade balance. This conclusion became known as the Harberger-Laursen-

Metzler (HLM) effect. This view remained more or less unchallenged until

the early 1980s, when Obstfeld (1982) and Svensson and Razin (1983), using

a dynamic optimizing model of the current account, concluded that the

effect of terms of trade shocks on the trade balance depends crucially on the

perceived persistence of the terms of trade. In their model a positive relation

between terms of trade and the trade balance (i.e., the HLM effect) weakens

as the terms of trade become more persistent and may even be overturned

if the terms of trade are of a permanent nature. This view became known

as the Obstfeld-Razin-Svensson (ORS) effect. Let us look at the HLM and

ORS effects in some more detail.

The Harberger-Laursen-Metzler Effect

A simple way to obtain a positive relation between the terms of trade and

the trade balance in the context of a Keynesian model is by starting with

the national accounting identity

y t = ct + gt + it + xt − m t ,

where yt denotes output, ct denotes private consumption, gt denotes public

consumption, it denotes private investment, xt denotes exports, and mt

denotes imports. Consider the following behavioral equations defining the

dynamics of each component of aggregate demand. Public consumption and

92 Martı́n Uribe

private investment are assumed to be independent of output. For simplicity,

we will assume that these two varibles are constant over time and given by

gt = ḡ

and

it = ī,

respectively, where ḡ and ī are parameters. Consumption is assumed to be

an increasing linear function of output

ct = c̄ + αyt ,

with α ∈ (0, 1) and c̄ > 0 are parameters. Imports are assumed to be

proportional to output,

mt = µyt ,

with µ ∈ (0, 1). In the jargon of the 1950s, the parameters α and µ are

referred to as the marginal propensities to consume and import, respectively,

whereas the term c̄ + ḡ + ī is referred to as the autonomous component of

domestic absorption. Output as well as all components of aggregate demand

are expressed in terms of import goods. The quantity of goods exported in

period t is denoted by qt . Thus, the value of exports in terms of importables,

xt , is given by

xt = tott qt ,
Lectures in Open Economy Macroeconomics, Chapter 5 93

where tott denotes the terms of trade. The terms of trade are assumed to

evolve exogenously, and the quantity of goods exported, qt , is assumed to

be constant and vigen by

qt = q̄,

where q̄ is a positive parameter. Using the behavioral equations to eliminate

ct , it , gt , xt , and mt from the national income identity, and solving for output

yields
c̄ + ḡ + ī + tott q̄
yt = .
1+µ−α

Letting tbt ≡ xt − mt denote the trade balance, we can write

1−α µ(c̄ + ḡ + ī)

tbt = tott q̄ − .
1+µ−α 1+µ−α

Clearly, this theory implies that an improvement in the terms of trade (an

increase in tott ) gives rise to an expansion in the trade surplus. This positive

relation between the terms of trade and the trade balance is stronger the

larger is the volume of exports, q̄, the smaller is the marginal propensity

to import, µ, and the smaller is the marginal propensity to consume α.

The reason why µ increases the TOT multiplier is that a higher value of

µ weakens the endogenous expansion in aggregate demand to an exogenous

increase in exports, as a larger fraction of income is used to buy foreign

goods. Similarly, a larger value of α reduces the TOT multiplier because it

exacerbates the endogenous response of aggregate demand to a TOT shock

through private consumption.

It is worth noting that in the context of this model, the sign of the effect
94 Martı́n Uribe

of a TOT shock on the trade balance is independent of whether the terms

of trade shocks are permanent or temporary in nature. This is the main

contrast with the Obstfeld-Razin-Svensson effect.

The Obstfeld-Razin-Svensson Effect

The ORS effect is cast within the dynamic optimizing theoretical frame-

work that differs fundamentally from the reduced-form Keynesian model we

used to derive the HLM effect. Consider the small, open, endowment econ-

omy studied in chapter 2. This is an economy inhabited by an infinitely

lived representative household with preferences described by the intertem-

poral utility function given in (2.5). Suppose that the good the household

consumes is different from the good it is endowed with. The household,

therefore exports the totality of its endowment and imports the totality of

its consumption. Let tott denote the relative world price of exported goods

in terms of imported goods, or the terms of trade. Assume for simplicity

that the endowment of exportable goods is constant and normalized to unity,

yt = 1 for all t. The resource constraint is then given by

dt = (1 + r)dt−1 + ct − tott .

The borrowing constraint given in (2.3) prevents the household from engag-

ing in Ponzi games. The economy is small in world product markets, so it

takes the evolution of tott as exogenous. The model is therefore identical

to the stochastic-endowment economy studied in chapter 2, with tott taking

the place of yt . We can then use the results derived in chapter 2 to draw the
Lectures in Open Economy Macroeconomics, Chapter 5 95

following conclusion: if the terms of trade are stationary then an increase in

the terms of trade produces an improvement in the current account. Agents

save in order to ensure higher future consumption. When terms of trade

are nonstationary, an improvement in the terms of trade induces a trade

balance deficit. In this case, the value of income is expected to grow over

time, so agents can afford assuming higher current debts without sacrificing

future expenditures.

This conclusion can be extended to a model with endogenous labor sup-

ply and capital accumulation. A simple way to do this is to modify the

RBC model of chapter 4 by assuming again that households do not con-

sume the good they produce. In this case, the productivity shock At can

be interpreted as a terms-of-trade shock. An increase in the terms of trade

produces an improvement in the trade balance if the terms of trade shock is

transitory, but as the serial correlation of the terms of trade shock increases,

an improvement in the terms of trade can lead to a deterioration in the

current account driven by investment expenditures.

Is the ORS effect borne out in the data? If so, we should observe that

countries experiencing more persistent terms-of-trade shocks should display

lower correlations between the terms of trade and the trade balance than

countries facing less persistent terms of trade shocks. Figure 5.1 plots the

serial correlation of the terms of trade against the correlation of the trade

balance with the terms of trade for 30 countries, including the G7 countries

and 23 selected developing countries from Latin America, Africa, East Asia,

and the Middle East. The 30 observations were taken from Mendoza (1995),

table 1. The cloud of points, shown with circles, displays no pattern. The
96 Martı́n Uribe

Figure 5.1: TOT Persistence and TB-TOT Correlations

0.8

0.6
Argentina

0.4
TB−TOT Correlation

0.2

−0.2

−0.4

−0.6
−0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
TOT Serial Correlation

Source: Mendoza (1995), table 1.

Note: Each point corresponds to one country. The TOT ser-
ial correlation and the TB-TOT correlation are computed over
the period 1955-1990. The sample includes the G-7 coun-
tries (United States, United Kingdom, France, Germany, Italy,
Canada, and Japan), 6 countries from Latin America, (Ar-
gentina, Brazil, Chile, Mexico, Peru, Venezuela), 3 countries
from the Middle East (Israel, Saudi Arabia, and Egypt), 6 coun-
tries from Asia (Taiwan, India, Indonesia, Korea, Philippines,
and Thailand), and 8 countries from Africa (Algeria, Cameroon,
Zaire, Kenya, Morocco, Nigeria, Sudan, and Tunisia). The
solid line is the OLS fit and is given by corr(T B, T OT ) =
0.35−0.14ρ(T OT ). The dashed line is the OLS fit after eliminat-
ing Argentina from the sample and is given by corr(T B, T OT ) =
0.23 + 0.12ρ(T OT ). The dashed-dotted line is the OLS fit after
eliminating the G7 countries, Saudi Arabia, and Argentina from
the sample and is given by corr(T B, T OT ) = 0.28+0.03ρ(T OT ).
Lectures in Open Economy Macroeconomics, Chapter 5 97

OLS fit of the 30 points, shown with a solid line, displays a small negative

slope of -0.14. The sign of the slope is indeed in line with the ORS effect: As

the terms of trade shocks become more persistent, they should be expected

to induce a smaller response in the trade balance. It is apparent in the

graph, however, that the negative slope in the OLS regression is driven

by a single observation, Argentina, the only country in the sample with a

negative serial correlation of the terms of trade. Eliminating Argentina from

the sample one obtains a positive OLS slope of 0.12.1 The corresponding

fitted relationship is shown with a broken line on figure 5.1.

A number of countries in figure 5.1 are likely to be large players in

the world markets for the goods and services they import and/or export.

Countries in this group would include all of the G7 nations, the largest

economies in the world, and Saudi Arabia, a major oil exporter. For these

countries, the terms of trade are not likely to be exogenous. Eliminating

these 8 countries (as well as the outlier Argentina) from the sample, gives us

a better idea of what the relation between the TB-TOT correlation and the

TOT persistence looks for small emerging countries that take their terms of

trade exogenously. The fitted line using this reduced sample has a negligible

slope equal to 0.03 and is shown with a dash-dotted line in figure 5.1. We

conclude that the observed relationship between the TB-TOT correlation

and the persistence of TOT is close to nil.

Does this conclusion suggest that the empirical evidence presented here

is against the Obstfeld-Razin-Svensson effect? Not necessarily. The ORS

1
Indeed, Argentina is the only country whose elimination from the sample results in a
positive slope.
98 Martı́n Uribe

effect requires isolating the effect of TOT shocks on the trade balance. The

raw data is in principle driven by a multitude of shocks, of which the terms

of trade is just one. Moreover, some of these shocks may directly affect

both the trade balance and the terms of trade. Not controlling for these

shocks may result in erroneously attributing part of their effect on the trade

balance to the terms of trade. A case in point is given by world-interest-rate

shocks. High world interest rates may be associated with depressed economic

activity in developed and emerging economies alike. In turn, low levels of

economic activity in the developed world are likely to be associated with a

weak demand for primary commodities, and, as a result, with deteriorated

terms of trade for the emerging countries producing those commodities. At

the same time, high world interest rates are associated with contractions

in aggregate demand and improvements in the trade balance in emerging

countries. Under this scenario, the terms of trade and the trade balance are

moving at the same time, but attributing all of the movement in the trade

balance to changes in the terms of trade would be clearly misleading. As

another example, suppose that domestic technology shocks are correlated

with technology shocks in another country or set of countries. Suppose

further that this other country or set of countries generates a substantial

fraction of the demand for exports or the supply of imports of the country

in question. In this case, judging the empirical validity of the ORS effect

only on the grounds of raw correlations would be misplaced.

An important step in the process of isolating terms-of-trade shocks—or

any kind of shock, for that matter—is identification. Data analysis based

purely on statistical methods will in general not result in a successful iden-

Lectures in Open Economy Macroeconomics, Chapter 5 99

tification of technology shocks. Economic theory must be used be at center

stage in the identification process. The following exercise, which follows

Mendoza (1995), represents an early step in the task of identifying the ef-

fects of terms-of-trade shocks on economic activity in emerging economies.

5.3 Terms-of-Trade Shocks in an RBC Model

Consider expanding the real-business-cycle model of chapter 4 to allow for

terms-of-trade shocks. In doing this, we follow the work of Mendoza (1995).

The household block of the model is identical to that of the standard RBC

model studied in chapter 4. The main difference with the model of chap-

ter 4 is that the model studied here features three sectors: a sector producing

importable goods, a sector producing exportable goods, and a sector produc-

ing nontradable goods. An importable good is either an imported good or a

good that is produced domestically but is highly substitutable with a good

that is imported. Similarly, an exportable good is either an exported good

or a good that is sold domestically but is highly substitutable with a good

that is exported. A nontradable good is a good that is neither exportable

nor importable.

5.3.1 Households

This block of the model is identical to that of the RBC model studied in

chapter 4. The economy is populated by a large number of identical house-

100 Martı́n Uribe

holds with preferences described by the utility function

∞
X
E0 θt U (ct , ht ), (5.1)
t=0

where ct denotes consumption, ht denotes labor effort, and U is a period

utility function taking the form

[c (1 − h)ω ]1−γ
U (c, h) = .
1−γ

The variable θt /θt−1 is a time-varying discount factor and is assumed to

evolve according to the following familiar law of motion:

θt+1 = θt β(ct , ht ), (5.2)

where the function β is assumed to take the form

β(c, h) = [1 + c(1 − h)ω ]−β .

We established in chapter 4 that the endogeneity of the discount factor serves

the purpose of rendering the deterministic steady state independent of the

country’s initial net foreign asset position.

Households offer labor services for a wage wt and own the stock of capital,

kt , which they rent at the rate ut . The stock of capital evolves according to

the following law of motion:

kt+1 = (1 − δ)kt + it − Φ(kt+1 − kt ), (5.3)

Lectures in Open Economy Macroeconomics, Chapter 5 101

where it denotes gross investment, which is assumed to be an importable

good. The parameter δ ∈ [0, 1] denotes the capital depreciation rate. The

function Φ introduces capital adjustment costs, and is assumed to satisfy

Φ(0) = Φ0 (0) = 0 and Φ00 > 0. Under these assumptions, the steady-state

level of capital is not affected by the presence of adjustment costs. As

discussed in chapter 4, capital adjustment costs help curb the volatility of

investment in small open economy models like the one studied here.

Households are assumed to be able to borrow or lend freely in interna-

tional financial markets by buying or issuing risk-free bonds denominated in

units of importable goods and paying the constant interest rate r ∗ . Letting

dt denote the debt position assumed by the household in period t and pct

denote the price of the consumption good, the period budget constraint of

the household can be written as

dt = (1 + r ∗ )dt−1 + pct ct + it − wt ht − ut kt . (5.4)

The relative prices pct , wt , and ut are expressed in terms of importable goods,

which serve the role of numeraire. Households are subject to a no-Ponzi-

game constraint of the form

Et dt+j
lim ≤ 0. (5.5)
j→∞ (1 + r ∗ )j

The household seeks to maximize the utility function (5.1) subject to (5.2)-

(5.5). Letting θt ηt and θt λt denote the Lagrange multipliers on (5.2) and

(5.4), the first-order conditions of the household’s maximization problem

102 Martı́n Uribe

are (5.2), (5.4), (5.5) holding with equality, and

Uc (ct , ht ) − ηt βc (ct , ht ) = λt pct (5.6)

−Uh (ct , ht ) + ηt βh (ct , ht ) = λt wt (5.7)

λt = β(ct , ht )(1 + rt )Et λt+1 (5.8)

 
λt [1 + Φ0 (kt+1 − kt )] = β(ct , ht )Et λt+1 ut+1 + 1 − δ + Φ0 (kt+2 − kt+1 )

(5.9)

ηt = −Et U (ct+1 , ht+1 ) + Et ηt+1 β(ct+1 , ht+1 ) (5.10)

5.3.2 Production of Consumption Goods

The consumption good, ct , is produced by domestic firms. These firms

operate a CES production function that takes tradable consumption goods,

cTt , and nontradable consumption goods, cN

t , as inputs. Formally,

ct = [χ(cTt )−µ + (1 − χ)(cN

t )
−µ −1/µ
] , (5.11)

with µ > −1. Firms operate in perfectly competitive product and input

markets. They choose output and inputs to maximize profits, which are

given by

pct ct − pTt cTt − pN N

t ct ,

where pTt and pN

t denote, respectively, the relative prices of tradable and

nontradable consumption goods in terms of importable goods. The first-

order conditions associated with this profit-maximization problem are (5.11)

Lectures in Open Economy Macroeconomics, Chapter 5 103

and
  1   1+µ
1
cN
t 1−χ 1+µ pTt
= , (5.12)
cTt χ pN
t

  1  T  1+µ
1
ct 1 1+µ pt
= . (5.13)
cTt χ pct

It is clear from the first of these efficiency conditions that the elasticity of

substitution between tradable and nontradable goods is given by 1/(1 + µ).

From the second optimality condition, one observes that if the elasticity of

substitution between tradables and nontradables is less than unity (or µ >

0), then the share of tradables in total consumption, given by pTt cTt /(pct ct ),

increases as the relative price of tradables in terms of consumption, pTt /pct ,

increases.

5.3.3 Production of Tradable Consumption Goods

Tradable consumption goods, denoted cTt , are produced using importable

consumption goods, cM X
t , and exportable consumption goods, ct , via a Cobb-

Douglas production function. Formally,

cTt = (cX α M 1−α

t ) (ct ) , (5.14)

where α ∈ (0, 1) is a parameter. Firms are competitive and aim at maxi-

mizing profits, which are given by

pTt cTt − pX X M
t ct − c t ,
104 Martı́n Uribe

where pX
t denotes the relative price of exportable goods in terms of im-

portable goods, or the terms of trade. Note that because the importable

good plays the role of numeraire, we have that the relative price of importa-

bles in terms of the numeraire is always unity (pM

t = 1). The optimality

conditions associated with this problem are (5.14) and:

pX
t ct
X
=α (5.15)
pTt cTt

cM
t
= 1 − α. (5.16)
pt cTt
T

These optimality conditions state that the shares of consumption of exporta-

bles and importables in total consumption expenditure in tradable goods are

constant and equal to α and 1 − α, respectively. This implication is a con-

sequence of the assumption of Cobb-Douglas technology in the production

of tradable consumption.

5.3.4 Production of Importable, Exportable, and Nontrad-

able Goods

Exportable and importable goods are produced with capital as the only

input, whereas nontradable goods are produced using labor services only.

Formally, the three production technologies are given by

ytX = AX X αX
t (kt ) , (5.17)

ytM = AM M αM
t (kt ) , (5.18)
Lectures in Open Economy Macroeconomics, Chapter 5 105

and

ytN = AN N αN
t (ht ) , (5.19)

where ytX denotes output of exportable goods, ytM denotes output of im-

portable goods, and ytN denotes output of nontradable goods. The factors Ait

denote exogenous and stochastic technology shocks in sectors i = X, M, N .

The variable kti denotes the capital stock in sector i = X, M , and the vari-

able hN
t denotes labor services employed in the nontradable sector. Firms

demand input quantities to maximize profits, which are given by

pX X M N N N X M
t yt + yt + pt yt − wt ht − ut (kt + kt ).

The optimality conditions associated with this problem are

ut ktX
= αX , (5.20)
pX
t yt
X

ut ktM
= αM , (5.21)
ytM

and
wt hN
t
= αN . (5.22)
pN y
t t
N

According to these expressions, and as a consequence of the assumption of

Cobb-Douglas technologies, input shares are constant.

106 Martı́n Uribe

5.3.5 Market Clearing

In equilibrium, the markets for capital, labor, and nontradables must clear.

That is,

kt = ktX + ktM , (5.23)

ht = hN
t , (5.24)

and

cN N
t = yt . (5.25)

Also, in equilibrium the evolution of the net foreign debt position of the

economy is given by

dt = (1 + r ∗ )dt−1 − pX X X M M
t (yt − ct ) − yt + ct + it . (5.26)

5.3.6 Driving Forces

There are four sources of uncertainty in this economy: One productivity

shock in each of the three sectors (importable, exportable, and nontradable),

and the terms of trade. We assume that all shocks follow autoregressive

processes of order one. Mendoza (1995) imposes four restrictions on the

joint distribution of the exogenous shocks: (1) all four shocks share the

same persistence. (2) The sectorial productivity shocks are assumed to be

perfectly correlated. (3) The technology shocks affecting the production of

importables and exportables are assumed to be identical. (4) Innovations to

productivity shocks and terms-of-trade shocks are allowed to be correlated.

Lectures in Open Economy Macroeconomics, Chapter 5 107

These assumptions give rise to the following laws of motion:

p
ln pX X
t = ρ ln pt−1 + t ; pt ∼ N (0, σ2p ).

ln AX X T
t = ρ ln At−1 + t .

ln AM M T
t = ρ ln At−1 + t .

ln AN N N
t = ρ ln At−1 + t .

Tt = ψT pt + νtT ; νtT ∼ N (0, σν2T ), E(pt , νtT ) = 0.

N T
t = ψ N t .

We are now ready to define a competitive equilibrium.

5.3.7 Competitive Equilibrium

A stationary competitive equilibrium is a set of stationary processes {ct , cTt ,

cX M N N X M N X M c N T
t , ct , ct , ht , ht , yt , yt , yt , kt , kt , kt , it , dt , pt , pt , pt , wt , ut , ηt ,

λt }∞
t=0 satisfying equations (5.3) and (5.6)-(5.26), given the initial conditions

k0 and d−1 and the exogenous processes {AX M N X ∞

t , At , At , pt }t=0 .

5.3.8 Calibration

Mendoza (1995) presents two calibrations of the model, one matching key

macroeconomic relations in developed countries, and the other matching key

macroeconomic relations in developing countries.

In calibrating the driving forces of the developed-country version of the

model, the parameters ρ and σp are set to match the average serial cor-
108 Martı́n Uribe

relation and standard deviation of the terms of trade for the group of G7

countries. Using the information presented in table 1 of Mendoza (1995)

yields

ρ = 0.473,

and
p
σp = 0.047 1 − ρ2 .

Using estimates of productivity shocks in five industrialized countries by

Stockman and Tesar (1995), Mendoza (1995) sets the volatility of produc-

tivity shocks in the importable and exportable sectors at 0.019 and the

volatility of the productivity shock in the nontraded sector at 0.014. This

implies that
q p
ψT2 σ2p + σν2T = 0.019 1 − ρ2 ,

and
q p
ψN ψT2 σ2p + σν2T = 0.014 1 − ρ2 .

Based on correlations between Solow residuals and terms of trade in five

developed countries, Mendoza (1995) sets the correlation between the pro-

ductivity shock in the exportable sector and the terms of trade at 0.165.

This implies that

ψT σp
q = 0.165.
ψT2 σ2p + σν2T

Table 5.2 displays the parameter values implied by the above restrictions.

This completes the calibration of the parameters defining exogenous driving

forces in the developed-country model.

Lectures in Open Economy Macroeconomics, Chapter 5 109

Table 5.2: Calibration

Parameter Developed Developing
Country Country
σp 0.041 0.011
σν T 0.017 0.032
ρ 0.47 0.41
ψT 0.067 -0.156
ψN 0.74 0.74
r∗ 0.04 0.04
αX 0.49 0.57
αM 0.27 0.70
αN 0.56 0.34
δ 0.1 0.1
φ 0.028 0.028
γ 1.5 2.61
µ 0.35 -0.22
α 0.3 0.15
ω 2.08 0.79
β 0.009 0.009
110 Martı́n Uribe

In calibrating the driving forces of the developing-country model, the

parameter ρ and σp are picked to match the average serial correlation and

standard deviation of the terms of trade for the group of developing countries

reported in table 1 of Mendoza (1995). This implies:

ρ = 0.414,

and
p
σp = 0.12 1 − ρ2 .

Mendoza (1995) assumes that the standard deviation of productivity shocks

in the traded sector are larger than in the nontraded sector by the same

proportion as in developed countries. This means that the parameter ψN

takes the value 0.74 as in the developed-country model. Mendoza sets the

standard deviation of productivity shocks in the traded sectors at 0.04 and

their correlation with the terms of trade at -0.46 to match the observed

average standard deviation of GDP and the correlation of GDP with TOT

in developing countries. This yields the restrictions:

q p
ψT2 σ2p + σν2T = 0.04 1 − ρ2 ,

and
ψT σp
q = −0.46.
ψT2 σ2p + σν2T

The implied parameter values are shown in table 5.2. This completes the

calibration of the exogenous driving forces for the developing-country version

Lectures in Open Economy Macroeconomics, Chapter 5 111

Table 5.3: Data and Model Predictions

σx
Variable σT OT ρxt ,xt−1 ρx,GDP ρx,T OT
G7 DCs G7 DCs G7 DCs G7 DCs
TOT
Data 1 1 .47 .41 .78 .25 1 1
Model 1 1 .47 .41 .78 .32 1 1
TB
Data 1.62 1.60 .33 .38 .18 -.17 .34 .32
Model 3.5 .86 .37 .69 -.11 -.45 .19 .08
GDP
Data 1.69 1.30 .49 .49 1 1 .78 .25
Model .86 .47 .68 .82 1 1 .78 .32
C
Data 1.59 1.27 .44 .42 .96 .89 .74 .18
Model 1.01 1.32 .85 .99 .81 .75 .39 .03
I
Data 1.90 1.62 .51 .49 .84 .72 .66 .26
Model 2.0 .94 .14 .11 .67 .39 .70 .28
RER
Data 1.44 1.23 .38 .43 .58 .52 .70 .12
Model .57 .60 .79 .95 .80 .71 .57 .25
Source: Mendoza (1995).

of the model. Table 5.2 also displays the values assigned to the remaining

parameters of the model.2

5.3.9 Model Performance

Table 5.3 presents a number of data summary statistics from developed (G7)

and developing countries (DCs) and their theoretical counterparts. The

following list highlights a number of empirical regularities and comments on

the model’s ability to capture them.

2
See Mendoza, 1995 for more details.
112 Martı́n Uribe

1. In the data, the terms of trade are procyclical, although much less so

in developing countries than in G7 countries. The model captures this

fact relatively well.

2. The observed terms of trade are somewhat persistent. This fact is

matched by construction; recall that the parameter ρ is set to pin

down the serial correlation of the terms of trade in developing and G7

countries.

3. The terms of trade are less volatile than GDP. The model fails to

capture this fact.

4. The terms of trade are positively correlated with the trade balance.

The model captures this empirical regularity, but underestimates the

TB-TOT correlation, particularly for developing countries.

5. The trade balance is countercyclical in DCs but procyclical in G7 coun-

tries. In the model, the trade balance is countercyclical in both, de-

veloped and developing countries. The failure of the model to capture

the procyclicality of the trade balance in developed countries should be

taken with caution, for other authors estimate negative TB-GDP cor-

relations for developed countries. The model appears to overestimate

the countercyclicality of the trade balance.

6. In the data, the real exchange rate (RER) is measured as the ratio of

the domestic CPI to an exchange-rate-adjusted, trade-weighted aver-

age of foreign CPIs. In the model, the RER is defined as the relative

price of consumption in terms of importables and denoted by pct . In the

Lectures in Open Economy Macroeconomics, Chapter 5 113

data, the RER is procyclical. The model captures this fact, although

it overestimates somewhat the RER-GDP correlation.

7. The RER is somewhat persistent (with a serial correlation of less than

0.45 for both developing and G7 countries). In the model, the RER

is highly persistent, with an autocorrelation above 0.75 for both types

of country.

5.3.10 How Important Are the Terms of Trade?

To assess the contribution of the terms of trade to explaining business cy-

cles in developed and developing countries, one can run the counterfactual

experiment of computing equilibrium dynamics after shutting off all sources

of uncertainty other than the terms of trade themselves. In the context of

the model of this section, one must set all productivity shocks at their deter-

ministic steady-state values. This is accomplished by setting σν T = ψT = 0.

Mendoza (1995) finds that when the volatility of all productivity shocks

is set equal to zero in the developed-country version of the model, the volatil-

ity of output deviations from trend measured at import prices falls from 4.1

percent to 3.6 percent. Therefore, in the model the terms of trade explain

about 88 percent of the volatility of output. When output is measured in

terms of domestic prices, the terms of trade explain about 66 percent of

output movements.

When the same experiment is performed in the context of the developing-

country version of the model, shutting off the variance of the productivity

shocks results in an increase in the volatility of output. The reason for this
114 Martı́n Uribe

increase in output volatility is that in the benchmark calibration the terms of

trade are negatively correlated with productivity shocks—note in table 5.2

that ψT < 0 for developing countries. Taking this result literally would lead

to the illogical conclusion that terms of trade explain more than 100 percent

of output fluctuations in the developing-country model. What is wrong?

One difficulty with the way we have measured the contribution of the

terms of trade is that it is not based on a variance decomposition of output.

A more satisfactory way to assess the importance of terms-of-trade and

productivity shocks would be to define the terms of trade shock as pt and

the productivity shock as νtT . One justification for this classification is that

pt affects both the terms of trade and sectoral total factor productivities,

while νtT affects sectoral total factor productivities but not the terms of

trade. Under this definition of shocks, the model with only terms of trade

shocks results when σν T is set equal to zero. Note that the parameter ψT

must not be set to zero. An advantage of this approach is that, because the

variance of output can be decomposed into a nonnegative fraction explained

by pt and a nonnegative part explained by νtT , the contribution of the terms

of trade will always be a nonnegative number no larger than 100 percent.

Exercise 5.1 Using the model presented in this section, compute a variance

decomposition of output. What fraction of output is explained by terms-of-

trade shocks in the developed- and developing-country versions of the model?

Chapter 6

Interest-Rate Shocks

Business cycles in emerging market economies are correlated with the in-

terest rate that these countries face in international financial markets. This

observation is illustrated in figure 6.1, which depicts detrended output and

the country interest rate for seven developing economies between 1994 and

2001. Periods of low interest rates are typically associated with economic

expansions and times of high interest rates are often characterized by de-

pressed levels of aggregate activity.1

Data like those shown in figure 6.1 have motivated researches to ask what

fraction of observed business cycle fluctuations in emerging markets is due

to movements in country interest rate. This question is complicated by the

fact that the country interest rate is unlikely to be completely exogenous to

the country’s domestic conditions.2 To clarify ideas, let Rt denote the gross
1
The estimated correlations (p-values) are: Argentina -0.67 (0.00), Brazil -0.51 (0.00),
Ecuador -0.80 (0.00), Mexico -0.58 (0.00), Peru -0.37 (0.12), the Philippines -0.02 (0.95),
South Africa -0.07 (0.71).
2
There is a large literature arguing that domestic variables affect the interest rate
at which emerging markets borrow externally. See, for example, Edwards (1984), Cline

115
116 Martı́n Uribe

Figure 6.1: Country Interest Rates and Output in Seven Emerging Countries
Argentina Brazil
0.15
0.15
0.1
0.1
0.05
0 0.05
−0.05
0
−0.1

94 95 96 97 98 99 00 01 94 95 96 97 98 99 00 01

Ecuador Mexico
0.4
0.15
0.3
0.1
0.2
0.05
0.1
0
0
−0.05
94 95 96 97 98 99 00 01 94 95 96 97 98 99 00 01

Peru Philippines
0.1
0.1

0.05
0.05
0

−0.05 0

94 95 96 97 98 99 00 01 94 95 96 97 98 99 00 01

South Africa

0.06

0.04

0.02

94 95 96 97 98 99 00 01

Output Country Interest Rate

Note: Output is seasonally adjusted and detrended using a log-linear
trend. Country interest rates are real yields on dollar-denominated
bonds of emerging countries issued in international financial markets.
Data source: output, IFS; interest rates, EMBI+.
Source: Uribe and Yue (2006).
Lectures in Open Economy Macroeconomics, Chapter 6 117

interest rate at which the country borrows in international markets, or the

country interest rate. This interest rate can be expressed as Rt = Rtus St .

Here, Rus denotes the world interest rate, or the interest rate at which de-

veloped countries, like the U.S., borrow and lend from one another, and St

denotes the gross country interest-rate spread, or country interest-rate pre-

mium. Because the interest-rate premium is country specific, in the data we

find an Argentine spread, a Colombian spread, etc. If the country in ques-

tion is a small player in international financial markets, as many emerging

economies are, it is reasonable to assume that the world interest rate Rtus ,

is completely exogenous to the emerging country’s domestic conditions. We

can’t say the same, however, about the country spread St . An increase in

output, for instance, may induce foreign lenders to lower spreads on believes

that the country’s ability to repay its debts has improved.

Interpreting the country interest rate as an exogenous variable when in

reality it has an endogenous component is likely to result in an overstatement

of the importance of interest rates in explaining business cycles. To see

why, consider the following example. Suppose that the interest rate Rt is

purely endogenous. Thus, its contribution to generating business cycles is

nil. Assume, furthermore, that Rt is countercyclical, i.e., foreign lenders

reduce the country spread in response to expansions in aggregate activity.

The researcher, however, wrongly assumes that the interest rate is purely

exogenous. Suppose now that a domestic productivity shock induces an

expansion in output. In response to this output increase, the interest rate

falls. The researcher, who believes Rt is exogenous, erroneously attributes

(1995), and Cline and Barnes (1997).

118 Martı́n Uribe

part of the increase in output to the decline in Rt . The right conclusion,

of course, is that all of the movement in output is due to the productivity

shock.

It follows that in order to quantify the macroeconomic effects of interest

rate shocks, the first step is to identify the exogenous components of coun-

try spreads and world interest rate shocks. Necessarily, the identification

process must combine statistical methods and economic theory. The partic-

ular combination adopted in this chapter draws heavily from Uribe and Yue

(2006).

6.1 An Empirical Model

Our empirical model takes the form of a first-order VAR system:

     
ŷt ŷt−1 yt
     
     i 
 ı̂t   ı̂t−1   t 
     
     
A
 tbyt
 = B  tby
  t−1
 +  tby
  t

 (6.1)
     
 us   us   rus 
 R̂t   R̂t−1   t 
     
R̂t R̂t−1 rt

where yt denotes real gross domestic output, it denotes real gross domestic

investment, tbyt denotes the trade balance to output ratio, Rtus denotes the

gross real US interest rate, and Rt denotes the gross real (emerging) country

interest rate. A hat on yt and it denotes log deviations from a log-linear

trend. A hat on Rtus and Rt denotes simply the log. We measure Rtus as the

3-month gross Treasury bill rate divided by the average gross US inflation
Lectures in Open Economy Macroeconomics, Chapter 6 119

over the past four quarters.3 We measure Rt as the sum of J. P. Morgan’s

EMBI+ stripped spread and the US real interest rate. Output, investment,

and the trade balance are seasonally adjusted.

To identify the shocks in the empirical model, Uribe and Yue (2006)

impose the restriction that the matrix A be lower triangular with unit diag-

onal elements. Because Rtus and Rt appear at the bottom of the system, this

identification strategy presupposes that innovations in world interest rates

(rus r
t ) and innovations in country interest rates (t ) percolate into domestic

real variables with a one-period lag. At the same time, the identification

scheme implies that real domestic shocks (yt , it , and tby
t ) affect financial

markets contemporaneously. This identification strategy is a natural one,

for, conceivably, decisions such as employment and spending on durable

consumption goods and investment goods take time to plan and implement.

Also, it seems reasonable to assume that financial markets are able to react

quickly to news about the state of the business cycle.4

An additional restriction imposed on the VAR system, is that the world

interest rate Rtus follows a simple univariate AR(1) process (i.e., A4i = B4i =

0, for all i 6= 4). Uribe and Yue (2006) adopt this restriction primarily

because it is reasonable to assume that disturbances in a particular (small)

emerging country will not affect the real interest rate of a large country like

the United States.

The country-interest-rate shock, rt , can equivalently be interpreted as a

3
Using a more forward looking measure of inflation expectations to compute the US
real interest rate does not significantly alter our main results.
4
Uribe and Yue (2006), di! scuss an alternative identification strategy consisting in
placing financial variables ‘above’ real variables first in the VAR system.
120 Martı́n Uribe

country spread shock. To see this, consider substituting in equation (6.1) the

country interest rate R̂t using the definition of country spread, Ŝt ≡ R̂t −R̂tus .

Clearly, because Rtus appears as a regressor in the bottom equation of the

VAR system, the estimated residual of the newly defined bottom equation,

call it st , is identical to rt . Moreover, it is obvious that the impulse response

functions of ŷt , ı̂t , and tbyt associated with st are identical to those associated

with rt . Therefore, throughout the paper we indistinctly refer to rt as a

country interest rate shock or as a country spread shock.

After estimating the VAR system (6.1), Uribe and Yue use it to address

a number of questions central to disentangle the effects of country-spread

shocks and world-interest-rate shocks on aggregate activity in emerging mar-

kets: First, how do US-interest-rate shocks and country-spread shocks affect

real domestic variables such as output, investment, and the trade balance?

Second, how do country spreads respond to innovations in US interest rates?

Third, how and by how much do country spreads move in response to in-

novations in emerging-country fundamentals? Fourth, how important are

US-interest-rate shocks and country-spread shocks in explaining movements

in aggregate activity in emerging countries? Fifth, how important are US-

interest-rate shocks and country-spread shocks in accounting for movements

in country spreads? We answer these questions with the help of impulse

response functions and variance decompositions.

6.2 Impulse Response Functions

Figure 6.2 displays with solid lines the impulse response function implied
Lectures in Open Economy Macroeconomics, Chapter 6 121

Figure 6.2: Impulse Response To Country-Spread Shock

Output Investment

0 0

−0.05 −0.2

−0.1 −0.4

−0.15 −0.6
−0.2 −0.8
−0.25 −1
−0.3
−1.2
5 10 15 20 5 10 15 20

Trade Balance−to−GDP Ratio World Interest Rate

1

0.4

0.5
0.3

0.2 0

0.1
−0.5

0
−1
5 10 15 20 5 10 15 20

Country Interest Rate Country Spread

1 1

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0

5 10 15 20 5 10 15 20

Notes: (1) Solid lines depict point estimates of impulse responses,

and broken lines depict two-standard-deviation error bands. (2)
The responses of Output and Investment are expressed in percent
deviations from their respective log-linear trends. The responses
of the Trade Balance-to-GDP ratio, the country interest rate,
the US interest rate, and the country spread are expressed in
percentage points. The two-standard-error bands are computed
using the delta method.
122 Martı́n Uribe

by the VAR system (6.1) to a unit innovation in the country spread shock,

rt . Broken lines depict two-standard-deviation bands.5 In response to an

unanticipated country-spread shock, the country spread itself increases and

then quickly falls toward its steady-state level. The half life of the coun-

try spread response is about one year. Output, investment, and the trade

balance-to-output ratio respond as one would expect. They are unchanged

in the period of impact, because of our maintained assumption that external

financial shocks take one quarter to affect production and absorption. In the

two periods following the country-spread shock, output and investment fall,

and subsequently recover gradually until they reach their pre-shock level.

The adverse spread shock produces a larger contraction in aggregate domes-

tic absorption than! in aggregate output. This is reflected in the fact that

the trade balance improves in the two periods following the shock.

Figure 6.3 displays the response of the variables included in the VAR sys-

tem (6.1) to a one percentage point increase in the US interest rate shock,

rus
t . The effects of US interest-rate shocks on domestic variables and coun-

try spreads are measured with significant uncertainty, as indicated by the

width of the 2-standard-deviation error bands. The point estimates of the

impulse response functions of output, investment, and the trade balance,

however, are qualitatively similar to those associated with an innovation

in the country spread. That is, aggregate activity and gross domestic in-

vestment contract, while net exports improve. However, the quantitative

effects of an innovation in the US interest rate are much more pronounced

than those caused by a country-spread disturbance of equal magnitude. For

5
These bands are computed using the delta method.
Lectures in Open Economy Macroeconomics, Chapter 6 123

Figure 6.3: Impulse Response To A US-Interest-Rate Shock

Output Investment

0 0
−1

−0.5 −2
−3

−1 −4
−5
−6
−1.5
5 10 15 20 5 10 15 20

Trade Balance−to−GDP Ratio World Interest Rate

1
2

0.8
1.5
0.6
1
0.4
0.5
0.2

0
0

5 10 15 20 5 10 15 20

Country Interest Rate Country Spread

3 2.5

2.5 2

2 1.5

1.5 1

1 0.5

0.5 0

0 −0.5

−0.5 −1
5 10 15 20 5 10 15 20

Notes: (1) Solid lines depict point estimates of impulse responses,

and broken lines depict two-standard-deviation error bands. (2)
The responses of Output and Investment are expressed in percent
deviations from their respective log-linear trends. The responses
of the Trade Balance-to-GDP ratio, the country interest rate,
and the US interest rate are expressed in percentage points.
124 Martı́n Uribe

instance, the trough in the output response is twice as large under a US-

interest-rate shock than under a country-spread shock.

It is remarkable that the impulse response function of the country spread

to a US-interest-rate shock displays a delayed overshooting. In effect, in the

period of impact the country interest rate increases but by less than the

jump in the US interest rate. As a result, the country spread initially falls.

However, the country spread recovers quickly and after a couple of quarters

it is more than one percentage point above its pre-shock level. Thus, country

spreads increase significantly in response to innovations in the US interest

rate but with a short delay. The negative impact effect is in line with the

findings of Eichengreen and Mody (1998) and Kamin and Kleist (1999).

We note, however, that because the models estimated by these authors are

static in nature, by construction, they are unable to capture the rich dynamic

relation linking these two variables. The overshooting of country spreads is

responsible for the much larger response of domestic variables to an innovat!

ion in the US interest rate than to an innovation in the country spread of

equal magnitude.

We now ask how innovations in output, yt , impinge upon the variables of

our empirical model. The model is vague about the precise nature of output

shocks. They can reflect variations in total factor productivity, the terms-of-

trade, etc. Figure 6.4 depicts the impulse response function to a one-percent

increase in the output shock. The response of output, investment, and the

trade balance is very much in line with the impulse response to a positive

productivity shock implied by the small open economy RBC model (see

figure 4.1). The response of investment is about three times as large as that
Lectures in Open Economy Macroeconomics, Chapter 6 125

Figure 6.4: Impulse Response To An Output Shock

Output Investment
1
3
0.8
2.5

0.6 2

1.5
0.4
1
0.2
0.5

0 0
5 10 15 20 5 10 15 20

Trade Balance−to−GDP Ratio World Interest Rate

1
0

−0.1 0.5

−0.2
0
−0.3

−0.4
−0.5
−0.5

−1
5 10 15 20 5 10 15 20

Country Interest Rate Country Spread

0 0

−0.2 −0.2

−0.4 −0.4

−0.6 −0.6

−0.8 −0.8

5 10 15 20 5 10 15 20

Notes: (1) Solid lines depict point estimates of impulse response

functions, and broken lines depict two-standard-deviation er-
ror bands. (2) The responses of Output and Investment are
expressed in percent deviations from their respective log-linear
trends. The responses of the Trade Balance-to-GDP ratio, the
country interest rate, and the US interest rate are expressed in
percentage points.
126 Martı́n Uribe

of output. At the same time, the trade balance deteriorates significantly

by about 0.4 percent and after two quarters starts to improve, converging

gradually to its steady-state level. More interestingly, the increase in output

produces a significant reduction in the country spread of about 0.6 percent.

The half life of the country spread response is about five quarters. The

countercyclical behavior of the country spread in response to output shocks

suggests that country interest rates behave in ways that exacerbates the

business-cycle effects of output shocks.

6.3 Variance Decompositions

Figure 6.5 displays the variance decomposition of the variables contained in

the VAR system (6.1) at different horizons. Solid lines show the fraction

of the variance of the forecasting error explained jointly by US-interest-rate

shocks and country-spread shocks (rus

t and rt ). Broken lines depict the

fraction of the variance of the forecasting error explained by US-interest-

rate shocks (rus rus and r are orthogonal disturbances, the

t ). Because t t

vertical difference between the solid line and the broken line represents the

variance of the forecasting error explained by country-spread shocks at dif-

Lectures in Open Economy Macroeconomics, Chapter 6 127

Figure 6.5: Variance Decomposition at Different Horizons

Output Investment

0.3
0.25
0.25
0.2
0.2
0.15
0.15

0.1 0.1

0.05 0.05

0 0
5 10 15 20 5 10 15 20
quarters quarters

Trade Balances−to−GDP Ratio World Interest Rate

2
0.4

1.5
0.3

1
0.2

0.1 0.5

0 0
5 10 15 20 5 10 15 20
quarters quarters

Country Interest Rate Country Spread

0.8 0.8
0.7 0.7
0.6 0.6
0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1

5 10 15 20 5 10 15 20
quarters quarters

rus rus + r

Note: Solid lines depict the fraction of the variance of the k-

quarter-ahead forecasting error explained jointly by rus
t and rt
at different horizons. Broken lines depict the fraction of the
variance of the forecasting error explained by rus
t at different
horizons.
128 Martı́n Uribe

ferent horizons.6,7 Note that as the forecasting horizon approaches infinity,

the decomposition of the variance of the forecasting error coincides with the

decomposition of the unconditional variance of the series in question.

For the purpose of the present discussion, we associate business-cycle

fluctuations with the variance of the forecasting error at a horizon of about

five years. Researchers typically define business cycles as movements in

time series of frequencies ranging from 6 quarters to 32 quarters (Stock and

Watson, 1999). Our choice of horizon falls in the middle of this window.

According to our estimate of the VAR system given in equation (6.1),

innovations in the US interest rate, rus

t , explain about 20 percent of move-

ments in aggregate activity in emerging countries at business cycle frequency.

At the same time, country-spread shocks, rt , account for about 12 percent of

6
These forecasting errors are computed as follows. Let xt ≡ [ŷt ı̂t tbyt R̂tus R̂t ] be the
vector of variables included in the VAR system and t ≡ [yt it tby !rus rt ] the vector of
P
t t
disturbances of the VAR system. Then, one can write the MA(∞) representation of xt
as xt = ∞
P
−1
j=0 Cj t−j , where Cj ≡ (A B)j A−1 . The error in forecasting xt+h at time t

matrix of this h-step-ahead forecasting error is given by Σx,h ≡

P
for h > 0, that is, xt+h − Et xt+h , is given by hj=0 Cj t+h−j . The variance/covariance
h 0
j=0 Cj Σ Cj , where
Σ is the (diagonal) variance/covariance matrix of t . Thus, the variance of the h-step-
ahead forecasting error of xt is simply the vector containing the diagonal elements of
Σx,h . In turn, the variance of the error of the h-step-ahead forecasting error of xt due to
P
rus
a particular shock, say rus t , is given by the diagonal elements of the matrix Σx, ,h ≡
!h 0
j=0 (Cj Λ4 )Σ (Cj Λ4 ) , where Λ4 is a 5×5 matrix with all elements equal to zero except
element (4,4), which takes the value one. Then, the broken lines in figure 6.5 are given by
rus
the element-by-element ratio of the diagonal elements of Σx, ,h to the diagonal elements
x,h
of the matrix Σ for different values of h. The difference between the solid lines and
the broken lines (i.e., the fraction of the variance of the forecasting error due to rt ) is
computed in a similar fashion but using the matrix Λ5 .
7
We observe that the estimates of yt , it , tby r
t , and t (i.e., the sample residuals of
the first, second, third, and fifth equations of the VAR system) are orthogonal to each
other. But because ŷt , ît , and tbyt are excluded from the Rtus equation, we have that
the estimates of rust will in general not be orthogonal to the! estimates of yt , it , or tby
t .
However, under our maintained specification assumption that the US real interest rate
does not systematically respond to the state of the business cycle in emerging countries,
this lack of orthogonality should disappear as the sample size increases.
Lectures in Open Economy Macroeconomics, Chapter 6 129

aggregate fluctuations in these countries. Thus, around one third of business

cycles in emerging economies is explained by disturbances in external finan-

cial variables. These disturbances play an even stronger role in explaining

movements in international transactions. In effect, US-interest-rate shocks

and country-spread shocks are responsible for about 43 percent of move-

ments in the trade balance-to-output ratio in the countries included in our

panel.

Variations in country spreads are largely explained by innovations in US

interest rates and innovations in country-spreads themselves. Jointly, these

two sources of uncertainty account for about 85 percent of fluctuations in

country spreads. Most of this fraction, about 60 percentage points, is at-

tributed to country-spread shocks. This last result concurs with Eichengreen

and Mody (1998), who interpret this finding as suggesting that arbitrary

revisions in investors sentiments play a significant role in explaining the

behavior of country spreads.

The impulse response functions shown in figure 6.4 establish empirically

that country spreads respond significantly and systematically to domestic

macroeconomic variables. At the same time, the variance decomposition

performed in this section indicates that domestic variables are responsible

for about 15 percent of the variance of country spreads at business-cycle

frequency. A natural question raised by these findings is whether the feed-

back from endogenous domestic variables to country spreads exacerbates

domestic volatility. Here we make a first step at answering this question.

Specifically, we modify the R̂t equation of the VAR system by setting to

zero the coefficients on ŷt−i , ît−i , and tbyt−i for i = 0, 1. We then compute
130 Martı́n Uribe

Table 6.1: Aggregate Volatility With and Without Feedback of Spreads from
Domestic Variables Model
Variable Feedback No Feedback
Std. Dev. Std. Dev.
ŷ 3.6450 3.0674
ı̂ 14.1060 11.9260
tby 4.3846 3.5198
R 6.4955 4.7696

the implied volatility of ŷt , ît , tbyt and R̂t in the modified VAR system

at business-cycle frequency (20 quarters). We compare these volatilities to

those emerging ! from the original VAR model. Table 6.1 shows that the

presence of feedback from domestic variables to country spreads significantly

increases domestic volatility. In particular, when we shut off the endogenous

feedback, the volatility of output falls by 16 percent and the volatility of in-

vestment and the trade balance-to-GDP ratio fall by about 20 percent. The

effect of feedback on the cyclical behavior of the country spread itself is even

stronger. In effect, when feedback is negated, the volatility of the country

interest rate falls by about one third.

Of course, this counterfactual exercise is subject to Lucas’ (1976) cele-

brated critique. For one should not expect that in response to changes in

the coefficients defining the spread process all other coefficients of the VAR

system will remain unaltered. As such, the results of table 6.1 serve solely

as a way to motivate a more adequate approach to the question they aim

to address. This more satisfactory approach necessarily involves the use of

a theoretical model economy where private decisions change in response to

alterations in the country-spread process. We follow this route next.

Lectures in Open Economy Macroeconomics, Chapter 6 131

6.4 A Theoretical Model

The process of identifying country-spread shocks and US-interest-rate shocks

involves a number of restrictions on the matrices defining the VAR sys-

tem (6.1). To assess the plausibility of these restrictions, it is necessary to

use the predictions of some theory of the business cycle as a metric. If the

estimated shocks imply similar business cycle fluctuations in the empirical

as in theoretical models, we conclude that according to the proposed theory,

the identified shocks are plausible.

Accordingly, we will assess the plausibility of our estimated shocks in

four steps: First, we develop a standard model of the business cycle in small

open economies. Second, we estimate the deep structural parameters of the

model. Third, we feed into the model the estimated version of the fourth and

fifth equations of the VAR system (6.1), describing the stochastic laws of

motion of the US interest rate and the country spread. Finally, we compare

estimated impulse responses (i.e., those shown in figures 6.2 and 6.3) with

those implied by the proposed theoretical framework.

The basis of the theoretical model presented here is the standard neoclas-

sical growth model of the small open economy (e.g., Mendoza, 1991). We de-

part from the canonical version of the small-open-economy RBC model along

four dimensions. First, as in the empirical model, we assume that in each pe-

riod, production and absorption decisions are made prior to the realization

of that period’s world-interest-rate shock and country-spread shock. Thus,

innovations in the world interest rate or the country spread are assumed

to have allocative effects with a one-period lag. Second, preferences are as-
132 Martı́n Uribe

sumed to feature external habit formation, or catching up with the Joneses

as in Abel (1990). This feature improves the predictions of the standard

model by preventing an excessive contraction in private non-business ab-

sorption in response to external financial shocks. Habit formation has been

shown to help explain asset prices and business fluctuations in both devel-

oped economies (e.g., B! oldrin, Christiano, and Fisher, 2001) and emerging

countries (e.g., Uribe, 2002). Third, firms are assumed to be subject to a

working-capital constraint. This constraint introduces a direct supply side

effect of changes in the cost of borrowing in international financial markets,

and allows the model to predict a more realistic response of domestic output

to external financial shocks. Fourth, the process of capital accumulation is

assumed to be subject to gestation lags and convex adjustment costs. In

combination, these two frictions prevent excessive investment volatility, in-

duce persistence, and allow for the observed nonmonotonic (hump-shaped)

response of investment in response to a variety of shocks (see Uribe, 1997).

6.4.1 Households

Consider a small open economy populated by a large number of infinitely

lived households with preferences described by the following utility function

∞
X
E0 β t U (ct − µc̃t−1 , ht ), (6.2)
t=0

where ct denotes consumption in period t, c̃t denotes the cross-sectional

average level of consumption in period t, and ht denotes the fraction of

time devoted to work in period t. Households take as given the process

Lectures in Open Economy Macroeconomics, Chapter 6 133

for c̃t . The single-period utility index U is assumed to be increasing in its

first argument, decreasing in its second argument, concave, and smooth. The

parameter β ∈ (0, 1) denotes a subjective discount factor, and the parameter

µ measures the intensity of external habit formation.

Households have access to two types of asset, physical capital and an

internationally traded bond. The capital stock is assumed to be owned

entirely by domestic residents. Households have three sources of income:

wages, capital rents, and interest income bond holdings. Each period, house-

holds allocate their wealth to purchases of consumption goods, purchases of

investment goods, and purchases of financial assets. The household’s period-

by-period budget constraint is given by

dt = Rt−1 dt−1 + Ψ(dt ) + ct + it − wt ht − ut kt , (6.3)

where dt denotes the household’s debt position in period t, Rt denotes the

gross interest rate faced by domestic residents in financial markets, wt de-

notes the wage rate, ut denotes the rental rate of capital, kt denotes the

stock of physical capital, and it denotes gross domestic investment. We as-

sume that households face costs of adjusting their foreign asset position. We

introduce these adjustment costs with the sole purpose of eliminating the

familiar unit root built in the dynamics of standard formulations of the small

open economy model. The debt-adjustment cost function Ψ(·) is assumed

¯ = Ψ0 (d)
to be convex and to satisfy Ψ(d) ¯ = 0, for some d¯ > 0. Earlier in

chapter 4, we compared a number of standard alternative ways to induce

stationarity in the small open economy framework, including the one used
134 Martı́n Uribe

here, and conclude that they all produce virtually identical implications for

business fluctuations! .8

The process of capital accumulation displays adjustment costs in the

form of gestation lags and convex costs as in Uribe (1997). Producing one

unit of capital good requires investing 1/4 units of goods for four consecutive

periods. Let sit denote the number of investment projects started in t − i

for i = 0, 1, 2, 3. Then investment in period t is given by

3
1X
it = sit . (6.4)
4
i=0

In turn, the evolution of sit is given by

si+1t+1 = sit . (6.5)

The stock of capital obeys the following law of motion:

 
s3t
kt+1 = (1 − δ)kt + kt Φ , (6.6)
kt

where δ ∈ (0, 1) denotes the rate of depreciation of physical capital. The

8
The debt adjustment cost can be decentralized as follows. Suppose that financial
transactions between domestic and foreign residents require financial intermediation by
domestic institutions (banks). Suppose there is a continuum of banks of measure one that
behave competitively. They capture funds from foreign investors at the country rate Rt
and lend to domestic agents at the rate Rtd . In addition, banks face operational costs,
Ψ(dt ), that are increasing and convex in the volume of intermediation, dt . The problem
of domestic banks is then to choose the volume dt so as to maximize profits, which are
given by Rtd [dt − Ψ(dt )] − Rt dt , taking as given Rtd and Rt . It follows from the first-order
condition associated with this problem that the interest rate charged to domestic residents
is given by Rtd = 1−ΨR0t(dt ) , which is precisely the shadow interest rate faced by domestic
agents in the centralized problem (see the Euler condition (6.10) below). Bank profits are
assumed to be distributed to domestic households in a lump-sum fashion. This digression
will be of use later in the paper when we analyze the firm’s problem.
Lectures in Open Economy Macroeconomics, Chapter 6 135

process of capital accumulation is assumed to be subject to adjustment costs,

as defined by the function Φ, which is assumed to be strictly increasing, con-

cave, and to satisfy Φ(δ) = δ and Φ0 (δ) = 1. These last two assumptions

ensure the absence of adjustment costs in the steady state and that the

steady-state level of investment is independent of Φ. The introduction of

capital adjustment costs is commonplace in models of the small open econ-

omy. As discussed in chapters 3 and 4, adjustment costs are a convenient

and plausible way to avoid excessive investment volatility in response to

changes in the interest rate faced by the country in international markets.

Households choose contingent plans {ct+1 , ht+1 , s0,t+1 , dt+1 }∞

t=0 so as to

maximize the utility function (6.2) subject to the budget constraint (6.3),

the laws of motion of total investment, investment projects, and the capital

stock given by equations (6.4)-(6.6), and a borrowing constraint of the form

dt+j+1
lim Et Qj ≤0 (6.7)
j→∞
s=0 Rt+s

that prevents the possibility of Ponzi schemes. The household takes as given

the processes {c̃t−1 , Rt , wt , ut }∞

t=0 as well as c0 , h0 , k0 , R−1 d−1 , and sit for

i = 0, 1, 2, 3. The Lagrangian associated with the household’s optimization

problem can be written as:

∞
( " 3
#
X 1X
L = E0 β t U (ct − µc̃t−1 , ht ) + λt dt − Rt−1 dt−1 − Ψ(dt ) + wt ht + ut kt − sit − ct
t=0
4
i=0
    2
)
s3t X
+ λt qt (1 − δ)kt + kt Φ − kt+1 + λt νit (sit − si+1t+1 ) ,
kt
i=0

where λt , λt νit , and λt qt are the Lagrange multipliers associated with con-
136 Martı́n Uribe

straints (6.3), (6.5), and (6.6), respectively. The optimality conditions as-

sociated with the household’s problem are (6.3), (6.4)-(6.7) all holding with

equality and

Et λt+1 = Uc (ct+1 − µc̃t , ht+1 ) (6.8)

Et [wt+1 λt+1 ] = −Uh (ct+1 − µc̃t , ht+1 ) (6.9)

 
λt 1 − Ψ0 (dt ) = βRt Et λt+1 (6.10)

1
Et λt+1 ν0t+1 = Et λt+1 (6.11)
4
β
βEt λt+1 ν1t+1 = Et λt+1 + λt ν0t (6.12)
4
β
βEt λt+1 ν2t+1 = Et λt+1 + λt ν1t (6.13)
4
  
0 s3t+1 β
βEt λt+1 qt+1 Φ = Et λt+1 + λt ν2t (6.14)
kt+1 4
      
s3t+1 s3t+1 0 s3t+1
λt qt = βEt λt+1 qt+1 1 − δ + Φ − Φ + λt+1 ut+1 .
kt+1 kt+1 kt+1
(6.15)

It is important to recall that, because of our assumed information structure,

the variables ct+1 , ht+1 , and s0t+1 all reside in the information set of period

t. Equation (6.8) states that in period t households choose consumption and

leisure for period t + 1 in such as way as to equate the marginal utility of

consumption in period t + 1 to the expected marginal utility of wealth in

that period, Et λt+1 . Note that in general the marginal utility of wealth will

differ from the marginal utility of consumption (λt 6= Uc (ct − µc̃t−1 , ht )), be-

cause current consumption cannot react to unanticipated changes in wealth.

Equation (6.9) defines the household’s labor supply schedule, by equating

Lectures in Open Economy Macroeconomics, Chapter 6 137

the marginal disutility of effort in period t + 1 to the expected utility value

of the wage rate in that period. Equation (6.10) is an asset pricing relation

equating the intertemporal marginal rate of substitution in ! consumption

to the rate of return on financial assets. Note that, because of the pres-

ence of frictions to adjust bond holdings, the relevant rate of return on this

type of asset is not simply the market rate Rt but rather the shadow rate

of return Rt /[1 − Ψ0 (dt )]. Intuitively, when the household’s debt position
¯ we have that Ψ0 (dt ) > 0 so that the
is, say, above its steady-state level d,

shadow rate of return is higher than the market rate of return, providing fur-

ther incentives for households to save, thereby reducing their debt positions.

Equations (6.11)-(6.13) show how to price investment projects at different

stages of completion. The price of an investment project in its ith quarter

of gestation equals the price of a project in the i-1 quarter of gestation plus

1/4 units of goods. Equation (6.14) links the cost of producing a unit of

capital to the shadow price of installed capital, or Tobin’s Q, qt . Finally,

equation (6.15) is a pricing condition for physical capital. It equates the

revenue from selling one unit of capital today, qt , to the discounted value of

renting the unit of capital for one period and then selling it, ut+1 + qt+1 , net

of depreciation and adjustment costs.

6.4.2 Firms

Output is produced by means of a production function that takes labor

services and physical capital as inputs,

yt = F (kt , ht ), (6.16)
138 Martı́n Uribe

where the function F is assumed to be homogeneous of degree one, increas-

ing in both arguments, and concave. Firms hire labor and capital services

from perfectly competitive markets. The production process is subject to a

working-capital constraint that requires firms to hold non-interest-bearing

assets to finance a fraction of the wage bill each period. Formally, the

working-capital constraint takes the form

κt ≥ ηwt ht ; η ≥ 0,

where κt denotes the amount of working capital held by the representative

firm in period t.

The debt position of the firm, denoted by dft , evolves according to the

following expression

dft = Rt−1
d
dft−1 − F (kt , ht ) + wt ht + ut kt + πt − κt−1 + κt ,

Rt
where πt denotes distributed profits in period t, and Rtd ≡ 1−Ψ0 (dt ) is the

interest rate faced by nonfinancial domestic agents, as shown in footnote 8.

The interest rate Rtd will in general differ from the country interest rate

Rt —the interest rate that domestic banks face in international financial

markets—because of the presence of debt-adjustment costs.

Define the firm’s total net liabilities at the end of period t as at =

Rtd dft − κt . Then, we can rewrite the above expression as

 
at Rtd − 1
= at−1 − F (kt , ht ) + wt ht + ut kt + πt + κt .
Rt Rtd
Lectures in Open Economy Macroeconomics, Chapter 6 139

We will limit attention to the case in which the interest rate is positive at all

times. This implies that the working-capital constraint will always bind, for

otherwise the firm would incur in unnecessary financial costs, which would

be suboptimal. So we can use the working-capital constraint holding with

equality to eliminate κt from the above expression to get

  d 
at Rt − 1
= at−1 − F (kt , ht ) + wt ht 1 + η + u t kt + π t . (6.17)
Rtd Rtd

It is clear from this expression that the assumed working-capital constraint

increases the unit labor cost by a fraction η(Rtd − 1)/Rtd , which is increasing

in the interest rate Rtd .

The firm’s objective is to maximize the present discounted value of the

stream of profits distributed to its owners, the domestic residents. That is,

∞
X λt
max E0 βt πt .
t=0
λ0

We use the household’s marginal utility of wealth as the stochastic discount

factor because households own domestic firms. Using constraint (6.17) to

eliminate πt from the firm’s objective function the firm’s problem can be

stated as choosing processes for at , ht , and kt so as to maximize

∞
X    d  
λt
t at Rt − 1
E0 β − at−1 + F (kt , ht ) − wt ht 1 + η − u t kt ,
t=0
λ 0 Rtd Rtd

subject to a no-Ponzi-game borrowing constraint of the form

at+j
lim Et Qj ≤ 0.
j→∞ d
s=0 Rt+s
140 Martı́n Uribe

The first-order conditions associated with this problem are (6.10), (6.17),

the no-Ponzi-game constraint holding with equality, and

  
Rtd − 1
Fh (kt , ht ) = wt 1 + η (6.18)
Rtd

Fk (kt , ht ) = ut . (6.19)

It is clear from the first of these two efficiency conditions that the working-

capital constraint distorts the labor market by introducing a wedge between

the marginal product of labor and the real wage rate. This distortion is

larger the larger the opportunity cost of holding working capital, (Rtd −

1)/Rtd , or the higher the intensity of the working capital constraint, η.9 We

also observe that any process at satisfying equation (6.17) and the firm’s

no-Ponzi-game constraint is optimal. We assume that firms start out with

no liabilities. Then, an optimal plan consists in holding no liabilities at all

times (at = 0 for all t ≥ 0), with distributed profits given by

  
Rtd − 1
πt = F (kt , ht ) − wt ht 1 + η − u t kt
Rtd

In this case, dt represents the country’s net debt position, as well as the

amount of debt intermediated by local banks. We also note that the above

three equations together with the assumption that the production technol-

ogy is homogeneous of degree one imply that profits are zero at all times

(πt = 0 ∀ t).
9
The precise form taken by this wedge depends on the particular timing assumed in
modeling the use of working capital. Here we adopt the shopping-time timing. Alternative
assumptions give rise to different specifications of the wedge. For instance, under a cash-
in-advance timing the wedge takes the form 1 + η(Rtd − 1).
Lectures in Open Economy Macroeconomics, Chapter 6 141

6.4.3 Driving Forces

One advantage of our method to assess the plausibility of the identified US-

interest-rate shocks and country-spread shocks is that one need not feed into

the model shocks other than those whose effects one is interested in studying.

This is because we empirically identified not only the distribution of the two

shocks we wish to study, but also their contribution to business cycles in

emerging economies. In formal terms, we produced empirical estimates of

the coefficients associated with rt and rus

t in the MA(∞) representation

of the endogenous variables of interest (output, investment, etc.). So using

the economic model, we can generate the corresponding theoretical MA(∞)

representation and compare it to its empirical counterpart. It turns out that

up to first order, one only needs to know the laws of motion of Rt and Rtus

to construct the coefficients of the theoretical MA(∞) representation. We

therefore close our model by introducing the law of motion of the country

interest rate Rt . This process is the estimate of the bottom equation of the

VAR system (6.1) and is given by

R̂t = 0.63R̂t−1 + 0.50R̂tus + 0.35R̂t−1

us
− 0.79ŷt + 0.61ŷt−1 + 0.11ı̂t − 0.12ı̂
(6.20)
t−1

+ 0.29tbyt − 0.19tbyt−1 + rt ,

where r is an i.i.d. disturbance with mean zero and standard deviation

0.031. As indicated earlier, the variable tbyt stands for the trade balance-
142 Martı́n Uribe

to-GDP ratio and is given by:10

yt − ct − it − Ψ(dt )
tbyt = . (6.21)
yt

Because the process for the country interest rate defined by equation (6.20)

involves the world interest rate Rtus , which is assumed to be an exogenous

random variable, we must also include this variable’s law of motion as part

of the set of equations defining the equilibrium behavior of the theoretical

model. Accordingly, we estimate Rtus as follows an AR(1) process and obtain

R̂tus = 0.83R̂t−1
us
+ rus
t , (6.22)

where rus
t is an i.i.d. innovation with mean zero and standard deviation

0.007.

6.4.4 Equilibrium, Functional Forms, and Parameter Values

In equilibrium all households consume identical quantities. Thus, individual

consumption equals average consumption across households, or

ct = c̃t ; t ≥ −1. (6.23)

An equilibrium is a set of processes ct+1 , c̃t+1 , ht+1 , dt , it , kt+1 , sit+1 for

i = 0, 1, 2, 3, Rt , Rtd , wt , ut , yt , tbyt , λt , qt , and νit for i = 0, 1, 2 satisfying

10
In an economy like the one described by our theoretical model, where the debt-
adjustment cost Ψ(dt ) are incurred by households, the national income and product ac-
counts would measure private consumption as ct + Ψ(dt ) and not simply as ct . However,
because of our maintained assumption that Ψ0 (d) ¯ = 0, it follows that both measures of
private consumption are identical up to first order.
Lectures in Open Economy Macroeconomics, Chapter 6 143

conditions (6.3)-(6.16), (6.18)-(6.21), and (6.23), all holding with equality,

given c0 , c−1 , y−1 , i−1 , i0 , h0 , the processes for the exogenous innovations

rus
t and rt , and equation (6.22) describing the evolution of the world interest

rate.

We adopt the following standard functional forms for preferences, tech-

nology, capital adjustment costs, and debt adjustment costs,

 1−γ
c − µc̃ − ω −1 hω −1
U (c − µc̃, h) = ,
1−γ

F (k, h) = k α h1−α ,

φ
Φ(x) = x − (x − δ)2 ; φ > 0,
2
ψ ¯ 2.
Ψ(d) = (d − d)
2

In calibrating the model, the time unit is meant to be one quarter. Following

Mendoza (1991), we set γ = 2, ω = 1.455, and α = .32. We set the steady-

state real interest rate faced by the small economy in international financial

markets at 11 percent per year. This value is consistent with an average

US interest rate of about 4 percent and an average country premium of 7

percent, both of which are in line with actual data. We set the depreciation

rate at 10 percent per year, a standard value in business-cycle studies.

There remain four parameters to assign values to, ψ, φ, η, and µ. There is

no readily available estimates for these parameters for emerging economies.

We therefore proceed to estimate them. Our estimation procedure follows

Christiano, Eichenbaum, and Evans (2001) and consists in choosing values

144 Martı́n Uribe

for the four parameters so as to minimize the distance between the estimated

impulse response functions shown in figure 6.2 and the corresponding im-

pulse responses implied by the model.11 In our exercise we consider the first

24 quarters of the impuls! e response functions of 4 variables (output, in-

vestment, the trade balance, and the country interest rate), to 2 shocks (the

US-interest-rate shock and the country-spread shock). Thus, we are setting

4 parameter values to match 192 points. Specifically, let IRe denote the

192×1 vector of estimated impulse response functions and IRm (ψ, φ, η, µ)

the corresponding vector of impulse responses implied by the theoretical

model, which is a function of the four parameters we seek to estimate. Then

our estimate of (ψ, φ, η, µ) is given by

argmax{ψ,φ,η,µ} [IRe − IRm (ψ, φ, η, µ)]0 Σ−1 e m

IRe [IR − IR (ψ, φ, η, µ)],

where ΣIRe is a 192×192 diagonal matrix containing the variance of the im-

pulse response function along the diagonal. This matrix penalizes those ele-

ments of the estimated impulse response functions associated with large er-

ror intervals. The resulting parameter estimates are ψ = 0.00042, φ = 72.8,

η = 1.2, and µ = 0.2. The implied debt adjustment costs are small. For

example, a 10 percent increase in dt over its steady-state value d¯ main-

tained over one year has a resource cost of 4 × 10−6 percent of annual GDP.

On the other hand, capital adjustment costs appear as more significant. For

11
A key difference between the exercise presented here and that in Christiano et al. is
that here the estimation procedure requires fitting impulse responses to multiple sources
of uncertainty (i.e., country-interest-rate shocks and world-interest-rate shocks, whereas in
Christiano et al. the set of estimated impulse responses used in the estimation procedure
are originated by a single shock.
Lectures in Open Economy Macroeconomics, Chapter 6 145

Table 6.2: Parameter Values

Symbol Value Description
β 0.973 Subjective discount factor
γ 2 Inverse of intertemporal elasticity of substitution
µ 0.204 Habit formation parameter
ω 1.455 1/(ω − 1) = Labor supply elasticity
α 0.32 capital elasticity of output
φ 72.8 Capital adjustment cost parameter
ψ 0.00042 Debt adjustment cost parameter
δ 0.025 Depreciation rate (quarterly)
η 1.2 Fraction of wage bill subject to working-capital constraint
R 2.77% Steady-state real country interest rate (quarterly)

instance, starting in a steady-state situation, a 10 percent increase in invest-

ment for one year produces an increase in the capital stock of 0.88 percent.

In the absence of capital adjustment costs, the capital stock increases by

0.96 percent. The estimated value of η implies that firms maintain a level of

working capital equivalent to about 3.6 months of wage payments. Finally,

the estimated degree of habit formation is modest compared to the values

typically used to explain asset-price regularities in closed economies (e.g.,

Constantinides, 1990). Table 6.2 gathers all parameter values.

6.5 Theoretical and Estimated Impulse Responses

Figure 6.6 depicts impulse response functions of output, investment, the

trade balance-to-GDP ratio, and the country interest rate.12 The left col-
12
The Matlab code used to produce theoretical impulse response functions is available
on line at http://www.econ.duke.edu/∼uribe/uribe yue jie/uribe yue jie.html.
146 Martı́n Uribe

Figure 6.6: Theoretical and Estimated Impulse Response Functions

Response of Output to εrus Response of Output to εr

0
0

−0.1
−0.5

−1 −0.2

−1.5 −0.3
0 5 10 15 20 0 5 10 15 20

Response of Investment to εrus Response of Investment to εr

0
0

−2 −0.5

−4
−1
−6
0 5 10 15 20 0 5 10 15 20

Response of TB/GDP to εrus Response of TB/GDP to εr

2
0.4
1.5
0.3
1
0.2
0.5
0.1
0
0
0 5 10 15 20 0 5 10 15 20

Response of Country Interest Rate to εrus Response of Country Interest Rate to εr

1
3
0.8
2 0.6

1 0.4

0.2
0
0
0 5 10 15 20 0 5 10 15 20

Estimated IR -x-x- Model IR 2-std error bands around Estimated IR

Note: The first column displays impulse responses to a US in-

terest rate shock (rus ), and the second column displays impulse
responses to a country-spread shock (r ).
Lectures in Open Economy Macroeconomics, Chapter 6 147

umn shows impulse responses to a US-interest-rate shock (rus

t ), and the

right column shows impulse responses to a country-spread shock (rt ). Solid

lines display empirical impulse response functions, and broken lines depict

the associated two-standard-error bands. This information is reproduced

from figures 6.2 and 6.3. Crossed lines depict theoretical impulse response

functions.

The model replicates three key qualitative features of the estimated im-

pulse response functions: First, output and investment contract in response

to either a US-interest-rate shock or a country-spread shock. Second, the

trade balance improves in response to either shock. Third, the country inter-

est rate displays a hump-shaped response to an innovation in the US interest

rate. Fourth, the country interest rate displays a monotonic response to a

country-spread shock. We therefore conclude that the scheme used to iden-

tify the parameters of the VAR system (6.1) is indeed successful in isolating

country-spread shocks and US-interest-rate shocks from the data.

6.6 The Endogeneity of Country Spreads

According to the estimated process for the country interest rate given in

equation (6.20), the country spread Ŝt = R̂t − R̂tus moves in response to four

types of variable: its own lagged value St−1 (the autoregressive component),

the exogenous country-spread shock rt (the sentiment component), current

and past US interest rates Rtus and Rt−1

us ), and current and past values of a

ˆ , tby
set of domestic endogenous variables, ŷt , ŷt−1 , ı̂t , ı̂t−1 , tby ˆ
t t−1 . A natural

question is to what extent the endogeneity of country spreads contributes

148 Martı́n Uribe

to exacerbating aggregate fluctuations in emerging countries.

We address this question by means of two counterfactual exercises. The

first exercise aims at gauging the degree to which country spreads amplify the

effects of world-interest-rate shocks. To this end, we calculate the volatility

of endogenous macroeconomic variables due to US-interest-rate shocks in a

world where the country spread does not directly depend on the US interest

rate. Specifically, we assume that the process for the country interest rate

is given by

R̂t = 0.63R̂t−1 + R̂tus − 0.63R̂t−1

us
− 0.79ŷt + 0.61ŷt−1 + 0.11ı̂t − 0.12ı̂
(6.24)
t−1

+ 0.29tbyt − 0.19tbyt−1 + rt .

This process differs from the one shown in equation (6.20) only in that the

coefficient on the contemporaneous US interest rate is unity and the coef-

ficient on the lagged US interest rate equals -0.63, which is the negative

of the coefficient on the lagged country interest rate. This parametrization

has two properties of interest. First, it implies that, given the past value
us , the current country spread,
of the country spread, Ŝt−1 = R̂t−1 − R̂t−1

St , does not directly depend upon current or past values of the US interest

rate. Second, the above specification of the country-interest-rate process

preserves the dynamics of the model in response to country-spread shocks.

The process for the US interest rate is assumed to be unchanged (i.e., given

by equation (6.22)). We note that in conducting this and the next counter-

factual exercises we do not reestimate the VAR system. The reason is that

doing so would alter the estimated process of the country spread shock rt .
Lectures in Open Economy Macroeconomics, Chapter 6 149

Table 6.3: Endogeneity of Country Spreads and Aggregate Instability

Std. Dev. due to rus Std. Dev. due to r
Baseline No ŷt Baseline No ŷt
Variable Model No R us ı̂, or tby Model No R us ı̂, or tby
ŷ 1.110 0.420 0.784 0.819 0.819 0.639
ı̂ 2.245 0.866 1.580 1.547 1.547 1.175
tby 1.319 0.469 0.885 0.663 0.663 0.446
R 3.509 1.622 2.623 4.429 4.429 3.983
S 2.515 0.347 1.640 4.429 4.429 3.983

Note: The variable S denotes the country spread and is defined

as S = R/Rus . A hat on a variable denotes log-deviation from
its non-stochastic steady-state value.

This would amount to introducing two changes at the same time. Namely,

changes in the endogenous and the sentiment components of the country

spread process.

The precise question we wish to answer is: what process for R̂t induces

higher volatility in macroeconomic variables in response to US-interest-rate

shocks, the one given in equation (6.20) or the one given in equation (6.24)?

To answer this question, we feed the theoretical model first with equation

(6.20) and then with equation (6.24) and in each case compute a variance

decomposition of output and other endogenous variables of interest. The

result is shown in table 6.3. We find that when the country spread is assumed

not to respond directly to variations in the US interest rate (i.e., under the

process for Rt given in equation (6.24)) the standard deviation of output and

the trade balance-to-output ratio explained by US-interest-rate shocks is

about two thirds smaller than in the baseline scenario (i.e., when Rt follows
150 Martı́n Uribe

the process given in equation (6.20)). This indicates that the aggregate

effects of US-interest-rate shocks are strongly amplified by the dependence

of country spreads on US interest rates.

A second counterfactual experiment we wish to conduct aims to assess

the macroeconomic consequences of the fact that country spreads move in

response to changes in domestic variables, such as output and the external

accounts. To this end, we use our theoretical model to compute the volatility

of endogenous domestic variables in an environment where country spreads

do not respond to domestic variables. Specifically, we replace the process

for Rt given in equation (6.20) with the process

R̂t = 0.63R̂t−1 + 0.50R̂tus + 0.35R̂t−1

us
+ rt . (6.25)

Table 6.3 displays the outcome of this exercise. We find that the equilib-

rium volatility of output, investment, and the trade balance-to-output ratio

explained jointly by US-interest-rate shocks and country-spread shocks (rus

t

and rt ) falls by about one fourth when the feedback from endogenous do-

mestic variables to country spreads is shut off.13 We conclude that the fact

that country spreads respond to the state of domestic business conditions

significantly exacerbates aggregate instability in emerging countries.

13
Ideally, this particular exercise should be conducted in an environment with a richer
battery of shocks capable of explaining a larger fraction of observed business cycles than
that accounted by rus
t and rt alone.
Chapter 7

Sovereign Debt

Why do countries pay their international debts? This is a fundamental

question in open-economy macroeconomics. A key distinction between in-

ternational and domestic debts is that the latter are enforceable. Countries

typically have in place domestic judicial systems capable of punishing de-

faulters. Thus, one reason why residents of a given country honor their debts

with other residents of the same country is because creditors are protected

by a government able and willing to apply force against delinquent debtors.

At the international level the situation is quite different. For there is no such

a thing as a supernational authority with the capacity to enforce financial

contracts between residents of different countries. Defaulting on interna-

tional financial contracts appears to have no legal consequences. If agents

have no incentives to pay their international debts, then lenders should have

no reason to lend internationally to begin with. Yet, we do observe a sig-

nificant amount of borrowing and lending across nations. It follows that

international borrowers must have reasons to repay their debts other than

151
152 Martı́n Uribe

pure legal enforcement.

Two main reasons are typically offered for why countries honer their

international debts: economic sanctions and reputation. Economic sanctions

may take many forms, such as seizures of debtor country’s assets located

abroad, trade embargoes, import tariffs and quotas, etc. Intuitively, the

stronger is the ability of creditor countries to impose economic sanctions,

the smaller the incentives for debtor countries to default.

A reputational motive to pay international debts arises when creditor

countries have the ability to exclude from international financial markets

countries with a reputation of being defaulters. Being isolated from interna-

tional financial markets is costly, as it precludes use of the current account to

smooth out consumption in response to aggregate domestic income shocks.

As a result, countries may choose to repay their debts simply to preserve

their reputation and thereby maintain access to international financing.

This chapter investigates whether the existing theories of sovereign debt

are capable of explaining the observed levels of sovereign debt. Before plung-

ing into theoretical models of country debt, however, we will present some

stylized facts about international lending and default that will guide us in

evaluating the existing theories.

7.1 Empirical Regularities

Table 7.1 displays average debt-to-GNP ratios over the period 1970-2000

for a number of emerging countries that defaulted upon or restructured

their external debt at least once between 1824 and 1999. The table also
Lectures in Open Economy Macroeconomics, Chapter 7 153

Table 7.1: Debt-to-GNP Ratios and Country Premiums Among Defaulters

Average Debt-to-GNP
Debt-to-GNP Ratio at Year Average
Country Ratio of Default Country Spread
Argentina 37.1 54.4 1756
Brazil 30.7 50.1 845
Chile 58.4 63.7 186
Colombia 33.6 649
Egypt 70.6 112.0 442
Mexico 38.2 46.7 593
Philippines 55.2 70.6 464
Turkey 31.5 21.0 663
Venezuela 41.3 46.3 1021
Average 44.1 58.1 638

Notes: The sample includes only emerging countries with at least one
external-debt default or restructuring episode between 1824 and 1999.
Debt-to-GNP ratios are averages over the period 1970-2000. Coun-
try spreads are measured by EMBI country spreads, produced by J.P.
Morgan, and expressed in basis points, and are averages through 2002,
with varying starting dates as follows: Argentina 1993; Brazil, Mexico,
and Venezuela, 1992; Chile, Colombia, and Turkey, 1999; Egypt 2002;
Philippines, 1997. Debt-to-GNP ratios at the beginning of a default
episodes are averages over the following default dates in the interval
1970-2002: Argentina 1982 and 2001; Brazil 1983; Chile 1972 and 1983;
Egypt 1984; Mexico 1982; Philippines 1983; Turkey 1978; Venezuela
1982 and 1995. Colombia did not register an external default or re-
structuring episode between 1970 and 2002.
Source: Own calculations based on Reinhart, Rogoff, and Savastano
(2003), tables 3 and 6.
154 Martı́n Uribe

displays average debt-to-GNP ratios at the beginning of default or restruc-

turing episodes. The data suggest that at the time of default debt-to-GNP

ratios are significantly above average. In effect, for the countries considered

in the sample, the debt-to-GNP ratio at the onset of a default or restruc-

turing episode was on average 14 percentage points above normal times.

The information provided in the table is silent, however, about whether the

higher debt-to-GDP ratios observed at the brink of default episodes obey to

a contraction in aggregate activity or to a faster-than-average accumulation

of debt in periods immediately preceding default.

Table 7.1 also shows the country premium paid by the 9 emerging coun-

tries listed over a period starting on average in 1996 and ending in 2002.

During this period, the interest rate at which these 9 countries borrowed in

the international financial market was on average about 6 percentage points

above the interest rate at which developed countries borrow from one an-

other. There is evidence that country spreads are higher the higher the

debt-to-GNP ratio. Akitoby and Stratmann (2006), estimate a semielastic-

ity of the spread with respect to the debt-to-GNP ratio of 1.3 (see their

table 3, column 4). That is, they estimate

∂ log country spread

= 1.3.
∂debt-to-GNP ratio

Thus, if, as documented in table 7.1, the debt-to-GNP ratio is 14 percentage

points higher at the beginning of a debt default or restructuring episode

than during normal times, and the average country spread during normal

times is about 640 basis points, it follows that at the beginning of a default
Lectures in Open Economy Macroeconomics, Chapter 7 155

episode the country premium increases on average by 1.3 × 640 × 0.14 = 116

basis points, or 1,16 percent. This increase in spreads might seem small

for a country that is at the brink of default. We note, however, that this

increase in the country premium is only the part of the total increase in

country spreads that is attributable to changes in the debt-to-GNP ratio.

Country spreads are known to respond systematically to other variables that

may take different values during normal and default times. For instance,

Akitoby and Stratmann (2006) and others have documented that spreads

increase significantly with the rate of inflation and decrease significantly with

the foreign reserve-to-GDP ratio. Because around default periods inflation

tends to be high and foreign reserves tend to be low, these two factors

contribute to higher country spreads in periods immediately preceding a

default episode.

Table 7.2 displays empirical probabilities of default for 9 emerging coun-

tries over the period 1824-1999. On average, the probability of default is

about 3 prevent per year. That is, countries defaulted on average once every

33 years. Table 7.2 also reports the average number of years countries are

in state of default or restructuring after a default or restructuring episode.

After a debt crisis, countries are in state of default for about 11 years on

average. If one assumes that while in state of default countries have little

access to fresh funds from international markets, one would conclude that

default causes countries to be in financial autarky for about a decade. But

the connection between state of default and financial autarky should not be

taken too far. For being in state of default with a set of lenders, doesn’t

necessarily preclude the possibility of obtaining new loans (possibly from

156 Martı́n Uribe

Table 7.2: Probability of Default and Length of Default State 1824-1999

Probability Years in State of
of Default Default per
Country per year Default Episode
Argentina 0.023 11
Brazil 0.040 6
Chile 0.017 14
Colombia 0.040 10
Egypt 0.011 11
Mexico 0.046 6
Philippines 0.006 32
Turkey 0.034 5
Venezuela 0.051 7
Average 0.030 11

Note: The sample includes only emerging countries with at least one
external-debt default or restructuring episode between 1824 and 1999.
Therefore, the average probability is conditional on at least one default
in the sample period.
Source: Own calculations based on Reinhart, Rogoff, and Savastano
(2003), table 1.
Lectures in Open Economy Macroeconomics, Chapter 7 157

other lenders) or issuing new external debt.

7.2 The Cost of Default

Default and debt restructuring episodes are typically accompanied by sig-

nificant declines in aggregate activity. Sturzenegger (2003) finds that after

controlling for a number of factors that explain economic growth, the cumu-

lative output loss associated with default and restructuring episodes in the

1980s was of about 4 percent over four yeas.

Default episodes are also associated with disruptions in international

trade. Rose (2005) investigates this issue empirically. The question of

whether default disrupts international trade is of interest because if for some

reason trade between two countries is significantly diminished as a result of

one country defaulting on its financial debts with other cuntries, then main-

taining access to international trade could represent a reason why countries

tend to honor their international financial obligations. Rose estimates an

equation of the form

M
X
Tijt = β0 + φijtm Rijt−m + βij Xijt + ijt ,
m=0

where Tijt is a measure of average bilateral trade between countries i and

j in period t. Rose identifies default with dates in which a country enters

a debt restructuring deal with the Paris Club. The Paris Club is an in-

formal association of creditor-country finance ministers and central bankers

that meets to negotiate bilateral debt rescheduling agreements with debtor-

158 Martı́n Uribe

country governments. The regressor Rijt proxies default and is a binary

variable equal to unity if countries i and j renegotiated debt in period t in

the context of the Paris Club and zero otherwise. The main focus of Rose’s

work is the estimation of the coefficients φijm .

Rose’s empirical model belongs to the family of gravity models. The

variable Xijt is a vector of regressors including (current and possibly lagged)

characteristics of the country pair ij at time t such as output, population,

distance, area, sharing of a common language, sharing of land borders, mem-

bership to the same free trade agreement, country pair-specific dummies, etc.

The vector Xijt also includes current and lagged values of a variable IM Fijt ,

that takes the values 0, 1, or 2, respectively, if neither, one, or both countries

i and j engaged in an IMF program at time t.

The data set used for the estimation of the model covers all bilateral

trades between 217 countries between 1948 and 1997 at an annual frequency.

The sample contains 283 Paris-Club debt-restructuring deals. Rose finds

sensible estimates of the parameters pertaining to the gravity model. Specif-

ically, countries that are more distant geographically trade less, whereas

high-income country pairs trade more. Countries that share a common cur-

rency, a common language, a common border, or membership in a regional

free trade agreement trade more. Landlocked countries and islands trade

less, and most of the colonial effects are large and positive. The inception

of IMF programs is associated with an accumulated contraction in trade of

about 10 percent over three years.

Default, as measured by the debt restructuring variable Rijt has a signif-

icant and negative effect on bilateral trade. Rose estimates the parameter
Lectures in Open Economy Macroeconomics, Chapter 7 159

φijm to be on average about 0.07 and the lag length, M , to be about 15

years. This means that entering in a debt restructuring agreement with a

member of the Paris Club leads to a decline in bilateral trade of about 7

percent per year for about 15 years. Thus, the cumulative effect of default

on trade is about one year worth of trade in the long run.

Based on this finding, Rose concludes that one reason why countries pay

back their international financial obligations is fear of trade disruptions in

the case of default.

7.3 A Reputational Model of Sovereign Debt

Eaton and Gersowitz (1981) pioneered a literature that explains sovereign

debt with a reputational argument. The central assumption behind this

theory is that somehow foreign lenders coordinate to exclude from inter-

national financial markets countries with a reputation for being defaulters.

Thus, countries that default are punished by having to operate in autarky.

Autarky is costly because it implies that countries cannot share systematic

income risks with the rest of the world. As a result, countries pay their

international debts to preserve their reputation and in that way maintain

access to external financing.

Our version of the Eaton-Gersowitz model follows Arellano (2005). Con-

sider a small open economy populated by a large number of identical indi-

viduals. Preferences are described by the utility function

∞
X
E0 β t u(ct ),
t=0
160 Martı́n Uribe

where ct denotes consumption in period t, u is a period utility function

assumed to be strictly increasing and strictly concave, and β ∈ (0, 1) is a

parameter denoting the subjective discount factor. Each period t ≥ 0, the

representative household is endowed with yt units of consumption goods.

This endowment is assumed to be exogenous, stochastic, and i.i.d., with a

distribution featuring a bounded support Y = [y, ȳ].

At the beginning of each period, the household can be either in good

financial standing or in bad financial standing If the household is in bad fi-

nancial standing, then it is prevented from borrowing or lending in financial

markets. When in autarky the household is forced to consume its endow-

ment. Formally, consumption of a household in bad financial standing is

given by

c = y.

We drop the time subscript in expressions where all variables are dated in

the current period. If the household is in good financial standing, it can

choose to default on its debt obligations or to honor its debt. If it chooses

to default, then it immediately acquires a bad financial status. If it chooses

to honor its debt, then it maintains its good financial standing until the

beginning of the next period. Households that are in good standing and

choose not to default face the following budget constraint:

c + d = y + q(d0 )d0 ,

where d denotes the household’s debt due in the current period, d0 denotes

the debt acquired in the current period and due in the next period, and
Lectures in Open Economy Macroeconomics, Chapter 7 161

q(d0 ) denotes the market price of the household’s debt. Note that the price

of debt depends on the amount of debt acquired in the current period and

due next period, d0 , but not on the level of debt acquired in the previous

period and due in the current period, d. This is because the default decision

in the next period depends on the amount of debt due then. Notice also

that q(·) is independent of the current level of output. This is because of

the assumed i.i.d. nature of the endowment, which implies that its current

value conveys no information about future expected endowment levels. If

instead we had assumed that y were serially correlated, then bond prices

would depend on the level of current endowment.

We assume that ‘bad financial standing’ is an absorbent state. This

means that once the household falls into bad standing, it remains in that

status forever. The household enters in bad standing when it defaults on

its financial obligations. The value function associated with bad financial

standing is denoted v b (y) and is given by

v b (y) = u(y) + βEv b (y 0 ).

Here, y 0 denotes next period’s endowment.

For an agent in good standing, the value function associated with con-

tinuing to participate in capital markets (i.e., not defaulting) is denoted by

v c (d, y) and is given by


v c (d, y) = max
0
u(y + q(d 0 0
)d − d) + βEv g 0 0
(d , y ) ,
d
162 Martı́n Uribe

subject to
¯
d0 ≤ d,

where v g (d, y) denotes the value function associated with being in good

financial standing, and is given by

v g (d, y) = max{v b (y), v c (d, y)}.

The parameter d¯ > 0 is a debt limit that prevents agents from engaging in

Ponzi games. In this economy, households choose to default when servicing

the debt entails a cost in terms of forgone current consumption that is larger

than the inconvenience of living in financial autarky forever. It is then

reasonable to conjecture that default is more likely the larger the level of

debt and the lower the current endowment. In what follows, we demonstrate

that this intuition is in fact correct. We do so in steps.

The Default Set

The default set contains all endowment levels at which a household chooses

to default given a particular level of debt. We denote the default set by

D(d). Formally, the default set is defined by

D(d) = {y ∈ Y : v b (y) > v c (d, y)}.

Because it is never in the agent’s interest to default when its asset position

is nonnegative (or d ≤ 0), it follows that D(d) is empty for all d ≤ 0.

The following proposition shows that at debt levels for which the default
Lectures in Open Economy Macroeconomics, Chapter 7 163

set is not empty, the economy must run trade surpluses.

¯
Proposition 7.1 If D(d) 6= ∅, then q(d0 )d0 − d < 0 for all d0 ≤ d.

 
Proof: Suppose that q db db − d ≥ 0 for some db ≤ d.
¯ Then,


v c (d, y) ≡ max u(y + q(d0 )d0 − d) + βEv g (d0 , y 0 )
d0 <d¯
 
≥ u(y + q db db − d) + βEv g (d,
b y0 )

≥ u(y) + βEv b (y 0 )

≡ v b (y),

for all y. The first inequality follows from the fact that db was picked arbi-
b db − d ≥ 0
trarily. The second inequality holds because, by assumption, q(d)
b y 0 ) ≥ v b (y 0 ). It follows that if q(d)
and because, by definition, v g (d, b db− d ≥ 0

for some db ≤ d,
¯ then D(d) = ∅.

This proposition states that if the household has a level of debt that puts

it in risk of default, then if it is to continue to participate in the financial

market, it must devote part of its current endowment to servicing the debt.

We now establish that in this economy households tend to default in bad

times. Specifically, we show that if a household with a certain level of debt

and income chooses to default then it will also choose to default at the same

level of debt and lower income. In other worlds, if the default set is not

empty then it is indeed an interval with lower bound given by the lowest

endowment set level y.

Proposition 7.2 If y2 ∈ D(d) and y1 < y2 , then y1 ∈ D(d).

164 Martı́n Uribe

Proof: Suppose D(d) 6= ∅. Consider any y ∈ Y such that y ∈ D(d). Let

vyb (y) ≡ ∂v b (y)/∂y and vyc (d, y) ≡ ∂v c (d, y)/∂y. By the envelope theorem,

vyb (y) = u0 (y) and vyc (d, y) = u0 (y + q(d0 )d0 − d). By proposition 7.1, we
¯ This implies, by strict concavity
have that q(d0 )d0 − d < 0 for all d0 ≤ d.

of u, that u0 (y + q(d0 )d0 − d) > u0 (y). It follows that vyb (y) − vyc (d, y) < 0,

for all y ∈ D(d). That is, v b (y) − v c (d, y) is a decreasing function of y for

all y ∈ D(d). This means that if v b (y2 ) > v c (d, y2 ) for y2 ∈ D(d), then

v b (y1 ) > v c (d, y1 ) for y1 < y2 . Equivalently, if y2 ∈ D(d), then y1 ∈ D(d)

for any y1 < y2 .

We have shown that the default set is an interval with a lower bound

given by the lowest endowment y. We now show that the default set D(d)

is a larger interval the larger the stock of debt. Put differently, the higher

the debt, the larger the probability of default.

Proposition 7.3 If D(d) 6= ∅, then D(d) is an interval, [y, y ∗ (d)], where

y ∗ (d) is increasing in d if y ∗ (d) < ȳ.

Proof: We already proved that the default set D(d) is an interval. By

definition, every y ∈ D(d) satisfies v b (y) − v c (d, y) > 0. At the same time,

we showed that vyb (y) − vyc (d, y) < 0 for all y ∈ D(d). It follows that y ∗ (d) is

given either by ȳ or (implicitly) by v b (y ∗ (d)) = v c (d, y ∗ (d)). Differentiating

this expression yields

dy ∗ (d) vdc (d, y ∗ (d))

,
dd vyb (y ∗ (d)) − vyc (d, y ∗ (d))

where vdc (d, y) ≡ ∂v c (d, y)/∂d. We have shown that vyb (y ∗ (d))−vyc (d, y ∗ (d)) <
Lectures in Open Economy Macroeconomics, Chapter 7 165

0. Using the definition of vdc (d, y) and applying the envelope theorem, it fol-

lows that vdc (d, y ∗ (d)) = −u0 (y ∗ (d) + q(d0 )d0 − d) < 0. We then conclude

that
dy ∗ (d)
> 0,
dd

as stated in the proposition.

Summarizing, we have obtained two important results: First, given the

stock of debt, default is more likely the lower the level of output. Second,

the larger the stock of debt, the higher the probability of default. These two

results are in line with the stylized facts presented earlier in this chapter. In

effect, table7.1 shows that at the time of default countries tend to display

above-average debt-to-GNP ratios.

Default Risk and the Country Premium

We now consider the behavior of the country interest-rate premium in this

economy. Let the world interest rate be constant and equal to r ∗ > 0. We

assume that foreign lenders are risk neutral and perfectly competitive. It

follows that the expected rate of return on the debt must equal r ∗ . If the

country does not default, foreign lenders receive 1/q(d0 ) units of goods per

unit lent. If the country DOES default, foreign lenders receive nothing.

Therefore, equating the expected rate of return on the domestic debt to the

risk-free world interest rate, one obtains

Prob {y 0 ≥ y ∗ (d0 )}
1 + r∗ = .
q(d0 )
166 Martı́n Uribe

The numerator on the right side of this expression is the probability that the

country will not default next period. Letting F (y) denote the cumulative

density function of the endowment shock, we can write

1 − F (y ∗ (d0 ))
q(d0 ) = .
1 + r∗

Taking derivative with respect to next period’s debt yields

0
dq(d0 ) −F 0 (y ∗ (d0 ))y ∗ (d0 )
= ≤ 0.
dd0 1 + r∗

The inequality follows because by definition F 0 ≥ 0 and because, by proposition7.3,

0
y ∗ (d0 ) ≥ 0. It follows that the country spread, given by the difference be-

tween 1/q(d0 ) and 1+r ∗ , is nondecreasing in the stock of debt. We summarize

this result in the following proposition:

Proposition 7.4 The country spread, given by 1/q(d0 ) − 1 − r ∗ is nonde-

creasing in the stock of debt.

7.4 Saving and the Breakdown of Reputational

Lending

A key assumption of the reputational model of sovereign debt is that when

a country defaults foreign lenders coordinate to exclude this country from

the possibility to borrow or lend in international financial markets. At a

first glance, it might seem that what is important is that defaulters be

precluded from borrowing in international financial markets. Bullow and

Lectures in Open Economy Macroeconomics, Chapter 7 167

Rogoff (1989) have shown, however, that the prohibiting defaulters to lend

(or save) to foreign agents is crucial for the reputational model to work.

If delinquent countries were not allowed to borrow but could run current

account surpluses, no lending at all could be supported on reputational

grounds alone.

To illustrate this insight in a simple setting, consider a deterministic

economy. Suppose that a reputational equilibrium supports a path for ex-

ternal debt given by {dt }∞

t=0 , where dt denotes the level of external debt

assumed in period t and due in period t + 1.1 .Assume that default is

punished with perpetual exclusion from borrowing in international financial

markets, but that lending in these markets is allowed after default. This

assumption and the fact that the economy operates under perfect foresight

imply that any reputational equilibrium featuring positive debt at at least

one date must be characterized by no default. To see this, notice that if

the country defaults at some date T > 0, then no foreign investor would

want to lend to this country in period T − 1, since default would occur for

sure one period later. Thus, dT −1 ≤ 0. In turn, if the country is excluded

from borrowing starting in period T − 1, then it will have no incentives to

honor any debts outstanding in that period. As a result, no foreign investor

will be willing to lend to the country in period T − 2. That is, dT −2 ≤ 0.

Continuing with this logic, we arrive at the conclusion that default in period

T implies no debt at any time. That is, dt ≤ 0 for all t ≥ 0.

It follows from this result that in an equilibrium with positive external

1
For an example of a deterministic model with sovereign debt supported by reputation,
see Eaton and Fernández (1995).
168 Martı́n Uribe

debt the interest rate must equal the world interest rate r ∗ > 0, because

the probability of default is nil. That is, the country premium is nil. The

evolution of the equilibrium level of debt is then given by

dt = (1 + r ∗ )dt−1 − tbt ,

for t ≥ 0, where tbt denotes the trade balance in period t. Let dT be

the maximum level of external debt in this equilibrium sequence. That is,

dT ≥ dt for all t ≥ −1. Does it pay for the country to honor this debt?

The answer is no. The reason is that the country could default on this debt

in T + 1—and therefore be excluded from borrowing internationally forever

thereafter—and still be able to run trade balances no larger than the ones

that would have obtained in the absence of default. To see this, let d˜t for

t > T denote the post-default path of external debt. Let the debt position

acquired in the period of default be

d˜T +1 = −tbT +1 ,

where tbT +1 is the trade balance prevailing in period T + 1 under the orig-

inal debt sequence {dt }. According to the economy’s no-default resource

constraint, we have that −tbT +1 = dT +1 − (1 + r ∗ )dT , which implies that

d˜T +1 = dT +1 − (1 + r ∗ )dT . (7.1)

Because by assumption dT ≥ dT +1 and r ∗ > 0, we have that d˜T +1 < 0.

That is, the country can generate the no-default level of trade balance in
Lectures in Open Economy Macroeconomics, Chapter 7 169

period T +1 without having to borrow internationally. Let the external debt

position in period T + 2 be

d˜T +2 = (1 + r ∗ )d˜T +1 − tbT +2 ,

where, again, tbT +2 is the trade balance prevailing in period T + 2 under

the original debt sequence {dt }. Using (7.1) and the fact that tbT +2 =

(1 + r ∗ )dT +1 − dT +2 , we obtain

d˜T +2 = dT +2 − (1 + r ∗ )2 dT < 0.

The inequality follows because by assumption dT +2 ≤ dT and r ∗ > 0. Con-

tinuing in this way, one obtains that the no-default sequence of trade bal-

ances, tbT +j for j > 0, can be supported by a debt path d˜T +j satisfying

d˜T +j = dT +j − (1 + r ∗ )j dT < 0,

for all j ≥ 1. It follows that it pays for the country to default immediately

after reaching the largest debt level dT . But we showed that default in this

perfect foresight economy implies zero debt at all times. It follows that no

external debt can be supported in equilibrium on reputational grounds.

For simplicity, we derived the breakdown of the reputation model under

saving using a model without uncertainty. But the result also holds in a

stochastic environment (see Bullow and Rogoff, 1989).

170 Martı́n Uribe
References

Abel, Andrew, “Asset Prices Under Habit Formation and Catching Up With

The Joneses,” American Economic Review (80), 1990, pp. 38-42.

Aguiar, Mark, and Gita Gopinath “Emerging Market Business Cycles: The

Cycle is the Trend,” NBER working paper No. 10734, August 2004.

Akitoby, B. and T. Stratmann “Fiscal Policy and Financial Markets,” IMF

working paper wp/06/16, January 2006.

Arellano, C. “Default Risk and Income Fluctuations in Emerging Markets,”

working paper, University of Minnesota, December 2005.

Arellano, M. and S. Bond, “Some Tests of Specification for Panel Data:

Monte Carlo Evidence and An Application to Employment Equations,”

Review of Economic Studies (58), 1991, 277-297.

Anderson, T. H. and Hsiao, C., “Estimation of Dynamic Models With Er-

ror Components,” Journal of the American Statistical Association (76),

1981, 598-606.

Boldrin, Michele; Lawrence J. Christiano, and Jonas Fisher, “Asset Pric-

ing Lessons for Modeling Business Cycles,” American Economic Review

(91), 2001, pp. 149-166.

171
172 Martı́n Uribe

Bullow, J., and K. Rogoff, “Sovereign Debt: Is to Forgive to Forget?,” Amer-

ican Economic Review (79), March 1989, 43-50.

Campbell, John Y., “Does Saving Anticipate Declining Labor Income? An

Alternative Test of the Permanent Income Hypothesis,” Econometrica

(55), November 1987, pp. 1249-1273.

Christiano, Lawrence J., Martin Eichenbaum, and Charles Evans “Nominal

Rigidities and the Dynamic Effects of a Shock to Monetary Policy,”

NBER working paper No. 8403, July 2001.

Chuhan, P. and F. Sturzenegger “Default Episodes in the 1980s and 1990s:

What have We Learned?,” manuscript, Universidad Torcuato Di Tella,

November 24, 2003.

Cline, William R., International Debt Reexamined, Washington, DC, Insti-

tute for International Finance/ 1995.

Cline, William R. and Kevin S. Barnes “Spreads and Risk in Emerging

Market Lending,” Research paper 97-1, Washington, DC, Institute for

International Finance, 1997.

Constantinides, George M., “Habit Persistence: A Resolution of the Equity

Premium Puzzle,” Journal of Political Economy (1990), 98, 519-543.

Correia, Isabel, João C.Neves, and Sergio Rebelo, “Business Cycles in a

Small Open Economy,” European Economic Review (39), 1995, 1089-

1113.

Eaton, J. and R. Fernández, “Sovereign Debt,” in G. Grossman and K.

Rogoff editors, Handbook of International Economics III, Amsterdam,

The Netherlands: Elsevier Science, 1995, chapter 39, pages 2031-2077.

Eaton, J. and M. Gersowitz, “Debt with Potential Repudiation: Theoretical

Lectures in Open Economy Macroeconomics, Chapter 7 173

and Empirical Analysis,” Review of Economic Studies (47), 1981, 289-

309.

Edwards, Sebastian, “LDC Foreign Borrowing and Default Risk: An Em-

pirical Investigation,” American Economic Review (74), 1984, 726-734.

Eichengreen, Barry and Ashoka Mody “What Explains Changing Spreads on

Emerging-Market Debt: Fundamentals or Market Sentiment?,” NBER

Working Paper No. 6408, February 1998.

Grossman, Herschel I. and John B. Van Huyck, “Sovereign Debt as a Contin-

gent Claim: Excusable Dafault, Repudiation, and Reputation,” Ameri-

can Economic Review (78), December 1988, 1088-1097.

Hall, Robert, “Stochastic Implications of the Life Cycle-Permanent Income

Hypothesis: Theory and Evidence,” Journal of Political Economy (86),

1978, 971-987.

Harberger, Arnold C., “Currency Depreciation, Income, and the Balance of

Trade,” Journal of Political Economy (58), 1950, 47-60.

Judson, Ruth A. and Ann L. Owen, “Estimating Dynamic Panel Data Mod-

els A Guide For Macroeconomists,” Economic Letters (65), 1999, pp.

9-15.

Kamin, Steven and Karsten von Kleist “The Evolution and Determinants

of Emerging Market Credit Spreads in the 1990s,” BIS Working Paper

No. 68, May 1999.

Kim, Sunghyun Henry and Ayhan Kose, “Dynamics of Open Economy Busi-

ness Cycle Models: Understanding the Role of the Discount Factor,”

Macroeconomic Dynamics (7), 2001, pp. 263-90.

Klein, Paul, “Using the Generalized Schur Form to Solve a Multivariate

174 Martı́n Uribe

Linear Rational Expectations Model,” Journal of Economic Dynamics

and Control (24), 2000, 1405-1423.

Krugman, Paul R., “A Model of Balance of Payments Crises,” Journal of

Money, Credit and Banking (11), 1979, 311-25.

Kydland, Finn E., and Carlos E. J. M. Zarazaga, “Is the Business Cycle

of Argentina Different?,” Economic Review (Federal Reserve Bank of

Dallas), 1997, 21-36.

Laursen, Svend and Lloyd A. Metzler, “Flexible Exchange Rates and the

Theory of Employment,” Review of Economics and Statistics (32), No-

vember 1950, 281-299.

Lucas, R.E., Jr., “Econometric Policy Evaluation: A Critique,” Carnegie-

Rochester Conference Series on Public Policy (1), 1976, 19-46.

Mendoza, Enrique, “Real Business cycles in a small-open economy,” Amer-

ican Economic Review (81), 1991, 797-818.

Mendoza, Enrique, “The Terms of Trade, the Real Exchange Rate, and

Economic Fluctuations,” International Economic Review (36), February

1995, 101-137.

Mendoza, Enrique and Martı́n Uribe, “Devaluation Risk and the Business-

Cycle Implications of Exchange-Rate Management,” Carnegie-Rochester

Conference Series on Public Policy (53), December 2000, 239-96.

Nason, James M. and John H. Rogers, “The Present-Value Model of the

Current Account Has Been Rejected: Round Up the Usual Suspects,”

Journal of International Economics (68), January 2006, 159-187.

Neumeyer, Pablo A. and Fabrizio Perri “Business Cycles in Emerging Mar-

kets:The Role of Interest Rates,” manuscript, New York University, No-

Lectures in Open Economy Macroeconomics, Chapter 7 175

vember 2001.

Obstfeld, Maurice, “Aggregate Spending and the Terms of Trade: Is There

a Laursen-Metzler Effect?,” Quarterly Journal of Economics (97), May

1982, 251-270.

Obstfeld, Maurice, “Intertemporal dependence, impatience, and dynamics,”

Journal of Monetary Economics (26), 1990, 45-75.

Obstfeld, Maurice, “Rational and Self-Fulfilling Balance of Payments Crises,”

American Economic Review (76), 1986, 72-81.

Obstfeld, Maurice, “The Logic of Currency Crises,” Cahiers Economiques

et Monetaires, Bank of France (43), 1994, 189-213.

Reinhart, Carmen M., Kenneth Rogoff, and Miguel Savastano, “Debt Intol-

erance,” Brookings Papers On Economic Activity (2003(1)), 2003, 1-74.

Rose, Andrew K., “One Reason Countries Pay Their Debts: Renegotiation

and International Trade,” Journal of Development Economics (77(1)),

2005, 189-206.

Schmitt-Grohé, Stephanie, “The international transmission of economic fluc-

tuations: Effects of U.S. business cycles on the Canadian economy,”

Journal of International Economics (44), April 1998, 257-287.

Schmitt-Grohé, Stephanie and Martı́n Uribe, “Stabilization Policy and the

Costs of Dollarization,” Journal of Money, Credit, and Banking (33),

May 2001, 482-509.

Schmitt-Grohé, Stephanie and Martı́n Uribe, “Closing Small Open Economy

Models,” Journal of International Economics (61), October 2003, pp.

163-185 .

Schmitt-Grohé, Stephanie and Martı́n Uribe, “Solving Dynamic General

176 Martı́n Uribe

Equilibrium Models Using a Second-Order Approximation to the Pol-

icy Function,” Journal of Economic Dynamics and Control (vol. 28),

January 2004, pp. 755-775.

Senhadji, Abdelhak S. “Adjustment of a Small Open Economy to External

Shocks,” Dissertation, University of Pennsylvania, 1994.

Sims, Christopher “Solving Linear Rational Expectations Models,” mimeo,

Yale University/ 1996.

Stock, James H. and Mark W. Watson, “Business Cycle Fluctuations in

US Macroeconomic Time Series,” in John B. Taylor and Michael Wood-

ford, eds., Handbook of Macroeconomics, Amsterdam, The Netherlands:

Elsevier Science, 1999, p. 3-64.

Stockman, Alan C. and Linda L. Tesar, “Tastes and Technology in a Two-

Country Model of the Business Cycle: Explaining International Comove-

ments,” American Economic Review (85), 1995, 168-85.

Svensson, Lars E.O., and Assaf Razin, “The Terms of Trade and the Current

Account: The Harberger-Laursen-Metzler Effect,” Journal of Political

Economy (91), February, 1983, 97-125.

Uribe, Martı́n, “Exchange rate based inflation stabilization: the initial real

effects of credible plans,” Journal of Monetary Economics (39), 1997,

197-221.

Uribe, Martı́n, “The Price-Consumption Puzzle of Currency Pegs,” Journal

of Monetary Economics (49), April 2002, 533-569.

Uribe, Martı́n and Z. Vivian Yue, “Country Spreads and Emerging Coun-

tries: Who Drives Whom?,” Journal of International Economics (forth-

coming), 2006, available on line at the authors’ websites.

Lectures in Open Economy Macroeconomics, Chapter 7 177

Uzawa, H., “Time Preference, the Consumption Function and Optimum

Asset Holdings,” in Wolfe, J.N. (Ed.), Value, Capital and Growth: Pa-

pers in Honor of Sir John Hicks, The University of Edinburgh Press

Edinburgh, 1968, pp. 485-504.