Topics in International Macroeconomics

Luca Fornaro

Fall 2021

1
 Aim of the course

This course will:

• Introduce you to the literature on international

macroeconomics.

• Teach you some tools to start doing research in the field.

Two main topics:

• Capital flows

• Monetary policy in open economies

2
 Outline

• Capital flows and the business cycle

• International financial crises and regulation of capital flows

• Exchange rates and monetary policy

• The economics of monetary unions

• The world in secular stagnation

• The role of the dollar in the international monetary system

• Global imbalances and the productivity growth slowdown

3
 Practical information
Contact: lfornaro@crei.cat, twitter: @lucafornaro3
Office hours: by appointment.
Reading list: see syllabus.
Evaluation: The evaluation is based on a term paper (80%) and
class participation (20%). The term paper consists in extending
in some interesting direction the model of one of the papers listed
at the end of the syllabus. For example, you could modify some
assumption to think about different scenarios, consider alterna-
tive policy interventions, provide a quantitative analysis or rein-
terpret some historical event. You can work on the term paper
in pairs, unless you plan to use the material of the term paper
as basis for your master thesis. Topics for the term paper need
to be discussed with the instructor before XXX. The final paper
will have to be handed before XXX.
4
Capital flows and the business cycle

5
 Motivation

• Since the early 1970s, dramatic reduction in barriers to

international capital flows

6
 Capital account openness index

Source: Chinn and Ito (2008)

7
 Motivation

• Since the early 1970s, dramatic reduction in barriers to

international capital flows

• Large and volatile international capital flows

• Research questions
• How do capital flows correlate with the business cycle?
• Can capital flows be a source of economic fluctuations?
• Shall governments regulate capital flows?

8
 Spain - Net exports/GDP

-5 NX/GDP

1970 1975 1980 1985 1990 1995 2000 2005 2010 2015

Source: OECD
9
 Spain - GDP per capita

-3.4 GDP per capita

linear trend
-3.5

-3.6

-3.7

-3.8

-3.9

-4

-4.1

-4.2

1970 1975 1980 1985 1990 1995 2000 2005 2010 2015

Source: OECD
10
 Spain - detrended GDP & Net exports/GDP

-5

-10
detrended GDP per capita
NX/GDP

1970 1975 1980 1985 1990 1995 2000 2005 2010 2015

Correlation = -0.8
11
Business cycle facts (Schmitt-Grohé and Uribe, 2015)

Denote yt as the log of Yt , we are interested in the

decomposition
yt = ytc + yts
|{z} |{z}
cycle trend

Log-quadratic detrending

yt = a + bt + ct 2 + t

ytc = t
yts = a + bt + ct 2

12
 Volatility of selected variables

Source: Schmitt-Grohé and Uribe (2015)

13
Correlations with GDP: small developed economies

14
Correlations with GDP: small emerging economies

15
 Summing up

Net exports are typically countercyclical

• Booms accompanied by net capital inflows

• Recessions characterized by net capital outflows

This pattern is particularly pronounced in emerging countries,

but is also present in some advanced economies
Note: different detrending methods (HP filter, growth rates)
have a significant impact on magnitudes, but leave these results
unchanged.

16
Intertemporal approach to the current account

17
 A small open endowment economy

• Infinite horizon, discrete time

• Continuum of measure one of identical households

• Single tradable consumption good

• World interest rate taken as given

• We start by considering an economy under perfect foresight

and then introduce uncertainty

18
 Households
Maximize lifetime utility
∞
X
β t u (ct )
t=0

Subject to
bt+1 + ct = yt + bt (1 + r )
bt+j
lim ≥0
j →∞ (1 + r )j

u(·) twice continuously differentiable, u 0 > 0, u 00 < 0

bt+1 bonds purchased at time t
yt endowment of tradable good
ct consumption of tradable good

19
 Optimality conditions

1) Euler equation

u 0 (ct ) = β(1 + r )u 0 (ct+1 ) for all t ≥ 0


 =1 implies constant consumption
β(1 + r ) = <1 implies decreasing consumption
>1 implies increasing consumption


2) No Ponzi game constraint binds with equality

20
 Solution

Remember the budget constraint

bt+1 + ct = yt + bt (1 + r )

Iterate forward
∞
X ct − yt
b0 (1 + r ) =
t=0
(1 + r )t
Difference between present value of consumption and present
value of income depends on initial bond position

21
 Some accounting
Trade balance
TBt = yt − ct
Net foreign assets (start of period)

NFAt = bt

Current account

CAt = NFAt+1 − NFAt = bt+1 − bt = TBt + rbt

Intertemporal budget constraint

∞
X TBt
−NFAt (1 + r ) =
t=0
(1 + r )t

An indebted economy runs, on average, trade surpluses to repay

the debt inherited from the past
22
 Back to the solution

We will assume β(1 + r ) = 1 → ct = c̄ ∀t

∞
X c̄ − yt
b0 (1 + r ) =
t=0
(1 + r )t

∞
r X yt
c̄ = + rb0
1 + r t=0 (1 + r )t

23
 Optimal consumption and trade balance

TBt = yt − c̄
Trade balance improves in good times and deteriorates in bad
times

CAt = yt − c̄ + rbt
Foreign assets increase in good times and decrease in bad times

24
 Constant output yt = ȳ

In any period t we have

c̄ = ȳ + rb0

economy rolls over its foreign asset position (bt = b0 for all t)

CAt = 0 for all t

Note: with trend growth the CA would not be zero in steady

state!
What happens if yt unexpectedly and permanently increases to
ȳ¯ > ȳ?

25
 Unanticipated temporary negative output shock
Suppose b0 = 0 and

yL

if t < T
yt =
ȳ if t ≥ T

Consumption follows a constant path

bt+1 = y L − c̄ + (1 + r )bt if t < T

bt+1 = ȳ − c̄ + (1 + r )bt if t ≥ T

P∞
Recall that since b0 = 0 then t=0 TBt /(1 + r )t = 0
Hence the country runs a trade balance deficit when output is
low and a trade balance surplus when output is high (consistent
with empirical evidence?)

26
Unanticipated temporary negative output shock

27
 A stochastic economy

Maximize lifetime utility

∞
X
E0 β t u (ct )
t=0

Subject to
bt+1 + ct = yt + bt (1 + r )
bt+j
Et lim ≥0
j →∞ (1 + r )j

yt is now a stochastic variable

28
 Technical digression: inducing stationarity

Euler equation

u 0 (ct ) = β(1 + r )Et u 0 (ct+1 )

In a deterministic economy we could assume β(1 + r ) = 1.

What about now?
Typically we solve models by linearizing around the
deterministic steady state
But here the steady state depends on initial conditions
Moreover, after a shock the economy will not go back to the
initial steady state

29
 How to deal with this?

Approach 1: use global solution methods. Start by defining

Mt ≡ β t (1 + r )t u 0 (ct )
Using Euler equation

Mt = Et Mt+1

implying that Mt is a martingale

Since Mt ≥ 0 , the supermartingale convergence theorem applies
(Ljunqqvist and Sargent, ch. 16). Then Mt converges a.s. to
M∞ such that Mt →a.s. M∞ and E(M∞ ) < ∞
We have 3 cases depending on the value of β(1 + r )

30
Case β(1 + r ) > 1: implies that for Mt to converge to a finite
value, u 0 (ct ) →a.s. 0 which implies ct →a.s. ∞
Case β(1 + r ) = 1: because of randomness, it is not possible for
Mt to converge to a finite positive value. Hence, Mt will
converge to zero and ct to infinity
Case β(1 + r ) < 1: ct can remain finite and continues to vary
randomly, while M converges to zero because
limt→∞ (β(1 + r ))t = 0
→ β(1 + r ) < 1 is a necessary condition for stationarity
Hence, one possibility is to assume β(1 + r ) < 1, solve the
model using global solution methods and derive a stochastic
steady state

31
 Alternative approaches

Modify the model so that a non-stochastic steady state exists

and local solution methods can be applied

• Endogenous discount factor (internal or external)

• Debt-elastic interest rate (internal or external)

• Portfolio adjustment costs

These methods are reviewed in Schmitt-Grohe and Uribe (JIE,

2003)

32
 Endogenous discount factor
We consider an external discount factor, which depends on
aggregate variables
∞
X
Et θt u(ct )
t=0
θt+1 = f (ct )θt where fc < 0
Euler equation is now

u 0 (ct ) = f (ct )(1 + r )Et u 0 (ct+1 )

In a non-stochastic steady state with constant c

f (c)(1 + r ) = 1

→ steady state consumption is uniquely pinned down

If the discount factor is internal the households take into
account the impact of their consumption and labor decisions on
the discount factor
33
 External debt elastic interest rate
Interest rate depends on net foreign assets

r (bt ) = r + g(Bt )

where B represents the aggregate level of assets and r 0 < 0

Euler equation

u 0 (ct ) = β(1 + r + g(Bt ))Et u 0 (ct+1 )

Steady state assets pinned down by β(1 + r (B )) = 1

suppose g(Bt ) = ψe Bt −B̄ in ss B = B̄

Here households do not internalize the impact of their

borrowing decisions on r , but if the interest rate depends on
individual assets they will do so

34
 Back to the model

We will solve the model numerically employing a global solution

method

• Write equilibrium recursively

• Choose functional forms and parameters

• Form grid for state variables bt , yt

• Solve model iterating on Euler equation

• Use policy functions to simulate model and compute

statistics

35
 The dynamic programming problem

Maximize with respect to c and b 0

V (b, y) = max u(c) + βEV (b 0 , y 0 )

Subject to
b 0 + c = y + b(1 + r )

b 0 ≥ −κ, c ≥ 0
κ is an ad-hoc borrowing limit

36
 Solution method

1. Form discretized grid for bt and yt

2. Make a guess for c 0 (b, y) (linearly interpolate between grid
points for b)
3. Solve for c(b, y)
4. If max |c(b, y) − c 0 (b, y)| is small stop
5. Otherwise set c 0 (b, y) = λc(b, y) + (1 − λ)c 0 (b, y) and go
back to 3

37
 Functional forms

CRRA utility function

c 1−σ − 1
u(c) =
1−σ
yt evolves according to

log(yt ) = ρ log(yt−1 ) + t

where t ∼ N (0, σ  )

38
 Parameters
Value Source/Target
Risk aversion σ=2 Standard value
Interest rate r = 4% Standard value
Discount factor β = 0.96 avg. NFA/GDP ≈ −40%
Borrowing limit κ = 1.5 max debt/output = 150%
Standard deviation y σy = 0.063 Estimate for Spain
Persistence y ρ = 0.87 Estimate for Spain

39
 Policy function - bonds

-1
avg. output
low output
-1.1

-1.2

-1.3

-1.4

-1.5
-1.5 -1.4 -1.3 -1.2 -1.1 -1

40
 Stationary bond distribution

0.7
0.6
0.5
0.4
0.3
0.2
0.1

0 2 4 6

41
 Policy function - trade balance

0.08

0.06

0.04

0.02

avg. output
-0.02
low output

-1.5 -1.4 -1.3 -1.2 -1.1 -1

42
 Simulated data

15

10
% dev. from mean

-5

-10 Y
C
-15
1 1.0002 1.0004 1.0006 1.0008 1.001
105

43
 Spain - GDP and absorption

Y
10 C

-5

-10

-15

1970 1975 1980 1985 1990 1995 2000 2005 2010 2015

44
Spain - tradable output and absorption

30 Y
C
20

10

-10

-20

-30

-40

1970 1975 1980 1985 1990 1995 2000 2005 2010 2015

45
 Business cycle statistics
Data Model
Standard deviation y 0.063 0.063
Standard deviation c 0.086 0.067
Standard deviation tb/y 0.026 0.062
Correlation y, tb/y -0.799 0.405
Standard deviation yT 0.087 0.063
Standard deviation cT 0.204 0.067

Notes: c is proxied by domestic absorption = y − tb

46
 Lessons from endowment economy

• Economy uses international credit markets to smooth

consumption over time

• Procyclical trade balance (counterfactual)

• We will explore two possible reasons why net capital

inflows are procyclical
• Investment in physical capital
• Financial frictions (next lecture)

47
Capital flows and investment

48
 Investment and the current account

Besides consumption smoothing, another reason why a country

might want to borrow from abroad is to invest in capital
accumulation
Investment is high when productivity is high, so a country
might want to borrow in good times to finance investment
This channel might give rise to a countercyclical current account

49
 Model with capital and perfect foresight

∞
X
β t u(ct )
t=0

bt+1 + ct + it = yt + bt (1 + r )

yt = At F (kt )

kt+1 = (1 − δ)kt + it
Optimality conditions

u 0 (ct ) = β(1 + r )u 0 (ct+1 )

u 0 (ct ) = β(1 − δ + At+1 F 0 (kt+1 ))u 0 (ct+1 )

50
Assume β(1 + r ) = 1 we have ct = c̄ and

1 + r = 1 − δ + At+1 F 0 (kt+1 )
∞
r X At+j F (kt+j ) − kt+j +1 + (1 − δ)kt+j
ct = rbt−1 +
1+r (1 + r )j
j =0

• kt+1 is increasing in next period A and decreasing in the

opportunity cost of capital r
• Consumption equals annuity value of wealth

→ separation between consumption and investment decisions

51
 Current account

CAt ≡ bt+1 − bt
Use resource constraint

CAt = rbt + yt − ct − it

Define domestic savings as

St ≡ rbt + yt − ct

Current account allows for separation between saving and

investment decisions
CAt = St − it

52
 Unanticipated permanent increase in TFP

Start from steady state

c̄ = rb0 + ȳ

¯ = −rb0
TB
¯ =0
CA
The economy experiences an unanticipated permanent increase
in TFP (↑ A)

53
Unanticipated permanent increase in TFP (δ = 0)

54
 Unanticipated permanent increase in TFP

• Current output increases because of higher TFP

• Investment temporarily rises because of higher return from
capital
• Future output increases by more than current output
because of capital accumulation
• Consumption increases more than output in anticipation of
future increase in output
• Current account and trade balance deteriorate on impact

55
Temporary increase in TFP (δ = 0)

56
 Temporary increase in TFP

• Investment does not change

• Output increases more than consumption on impact
• Trade balance moves to surplus on impact and then to
deficit
• Current account is initially positive and then zero

In general, the more persistent are productivity shocks, the

more likely is an initial deterioration of the trade balance

57
 Capital adjustment costs

Empirically, investment is much less volatile than what our

simple model would predict
The presence of investment adjustment cost can dampen the
volatility of investment over the business cycle
Suppose that δ = 0 and the budget constraint is

it2
At F (kt ) + (1 + r )bt = bt+1 + ct + it +
2kt

The term it2 /(2kt ) captures the presence of investment

adjustment costs

58
 ∞
1 it2
X   
t
L= β U (ct ) + λt At F (kt ) + (1 + r )bt − bt+1 − ct − it − +
t=0
2 kt

+λt qt (kt + it − kt+1 ))

Optimality conditions for investment and capital
it
1+ = qt
kt
"  2 #
1 it+1
λt qt = βλt+1 qt+1 + At+1 F 0 (kt+1 ) +
2 kt+1
qt is the shadow relative price of capital (Tobin’s q)
The first optimality condition tells us that investment is
increasing in qt
The second optimality condition gives us qt as a function of the
future marginal product of capital and qt+1
59
Assume β(1 + r ) = 1 so that λt and ct are constant over time
We can then write the optimality condition for capital as
 

1 it+1 2
 
1 
0

qt = At+1 F (kt+1 ) + + qt+1 
 
1+r | {z } 2 kt+1 |{z} 
dividends | {z } resale price
future adj. costs

Consumption is now given by

2
∞ A 1 it+j
r X t+j F (kt+j ) − it+j − 2 kt+j
ct = rbt +
1+r (1 + r )j
j =0

60
 Dynamics of the capital stock

Combine law of motion for capital with optimality conditions

kt+1 = qt kt

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

qt =
1+r
This system describes the dynamics of kt and qt .
Assume At = Ā and define

1 = qt (KK’)

ĀF 0 (qt kt ) + (qt − 1)2 /2 + qt

qt = (QQ’)
1+r

61
Dynamics of the capital stock

62
 Capital adjustment costs and capital flows

• The more pronounced capital adjustment costs are the less

likely that the trade balance deteriorates after a persistent
positive productivity shock

• Mendoza (1991, AER) is a classic example of quantitative

small open economy model with capital accumulation (see
also USG chapter 4)

• Key result: investment in physical capital can make the

trade balance mildly countercyclical, but cannot capture
the extreme countercyclicality observed in countries such as
Spain

63