You are on page 1of 34

Financial frictions and optimal monetary policy

in an open economy∗
Marcin Kolasa† Giovanni Lombardo‡
January 18, 2010

Abstract
This paper analyzes the optimal monetary policy in a two-country DSGE model with producer cur-
rency pricing and financial frictions. We show that if credit markets do not work perfectly, strict PPI
targeting is excessively procyclical in response to positive productivity shocks. The related welfare losses
are non-negligible, especially if financial imperfections interact with such frictions as nontradable produc-
tion and wage stickiness. Foreign currency debt denomination affects the optimal monetary policy: it
should be less contractionary in response to favourable domestic productivity disturbances and more so
if productivity shocks originate abroad. We also find that central banks should allow for deviations from
price stability to offset the effects of balance sheet shocks. While financial frictions substantially decrease
attractiveness of all price targeting regimes, they do not have a significant effect on the performance of
a monetary union agreement. Finally, nominal wage targeting is found to be fairly robust also in the
presence of credit frictions.

Keywords: financial frictions, open economy, New Keynesian model, optimal monetary policy
JEL Codes: E52, E61, E44, F36, F41

1 Introduction
The New Keynesian model, which can be considered as a workhorse model for modern monetary policy
analysis, assumes that financial markets work perfectly and so the interest rate set by central banks uniquely
determines the cost of credit for borrowers. The recent financial crisis has exposed the lack of realism of
this simplifying assumption and revived interest in business cycle models with financial frictions. Due to the
path-breaking contributions from the 1990s, of which the most frequently cited are Bernanke et al. (1999)
and Kiyotaki and Moore (1997), the quantitative framework was already at hand. What needed to be done
was a normative analysis of how financial frictions affect the optimal monetary policy.
A number of recent papers have addressed this issue in a closed economy setup. For instance, Curdia and
Woodford (2008) extend the basic New Keynesian monetary model to allow for a spread between interest
rates faced by savers and borrowers. They demonstrate that if spreads are purely exogenous, the optimal
policy conduct does not differ substantially from the frictionless case. Allowing for endogenous spreads (in
an ad hoc way, i.e. by making them dependent on borrowers’ debt) affects this conclusion only modestly. In
particular, complete price stabilization is still very close to the optimal policy. Furthermore, adjusting the
intercept in the Taylor rule by changes in credit spreads improves upon an unadjusted rule.
A more structural contribution is offered by Carlstrom et al. (2009), who incorporate agency costs into
a standard New Keynesian model. Since agency costs manifest themselves as endogenous mark-up shocks,
maintaining price stability is not optimal in response to productivity shocks. However, it is very close to
optimal even if agency costs are quite severe. A similar conclusion is reached by Demirel (2009) and De Fiore
∗ The views presented in this paper are those of the authors and not necessarily of the institutions they represent.
† NationalBank of Poland, e-mail: marcin.kolasa@nbp.pl.
‡ European Central Bank, e-mail: giovanni.lombardo@ecb.int.

1
et al. (2009), who introduce costly state verification into a model with a direct credit channel a’la Ravenna
and Walsh (2006), in which firms need to borrow in advance to finance production.
Overall, this line of the literature suggests that if financial markets do not work perfectly, the central
bank has an incentive to depart from full price stability in response to productivity shocks. However, the
marginal welfare gain of neutralizing the credit friction distortion is rather low, so strict inflation targeting
is not far from optimal.
While the big advantage of the literature surveyed above is that it offers an analytical characterization of
the results, its focus is on models that are very simple. They abstract from endogenous capital formation,
which may have nontrivial consequences, given that financial frictions are considered to be particularly
relevant for investment decisions. Also, they do not address open economy issues and other potentially
important frictions. On the other hand, there is a number of papers incorporating financial frictions into
a more sophisticated framework, but not discussing what the optimal policy should do. Instead, they offer
welfare-based comparisons of alternative simple policy regimes.
For instance, building on Faia and Monacelli (2007), Faia (2008) considers a general class of Taylor
rules, with strict inflation and exchange rate targeting as extremes, in a two-country sticky price model
with financial accelerator as in Bernanke et al. (1999). Using the welfare rankings that ignore the effect of
volatilities on mean welfare, she finds that the presence of credit frictions strengthens the case for floating
exchange rate regimes in economies facing external shocks.
A related line of papers consider a small open economy model with a similar financial frictions block
and foreign denomination of debt. Gertler et al. (2007) find that a fixed exchange regime exacerbates the
contraction caused by an adverse risk premium shock. According to Devereux et al. (2006), financial frictions
magnify volatility but do not affect the ranking of alternative policy rules. Finally, Elekdag and Tchakarov
(2007) show that at a certain level of leverage the peg starts to dominate the float if shocks originate abroad.
The aim of this paper is to fill the gaps in the literature by providing a qualitative and quantitative
characterization of the optimal monetary policy conduct in an open economy facing financial frictions. To
this end, we consider a medium-size two-country New Keynesian DSGE model with producer currency
pricing, augmented by the financial accelerator mechanism. Having defined the optimal policy as a Ramsey
cooperative equilibrium, we show how it is affected by fixing the exchange rate and how its outcomes deviate
from those obtained for a set of standard simple targeting rules. Contrary to the existing literature, focusing
on very simple models, we discuss how financial market imperfections interact with such frictions as the
presence of nontradable goods and wage stickiness. We argue that a more complex model is a necessary step
forward as the policy implications are sensitive to the types of frictions and, though to lesser extent, to the
type of shocks. At the same time, to build intuition for the main results, our strategy is to start with a
simple New Keynesian framework with capital accumulation and then build it up towards its fully-fledged
version, explaining the impact of each extension for the policy prescriptions.
Our main results can be summarized as follows. First, we find that if credit markets do not work perfectly,
strict PPI targeting leads to overexpansion (overcontraction) in economic activity in response to positive (neg-
ative) productivity shocks. The related welfare losses are non-negligible, especially if financial imperfections
interact with such frictions as nontradable production and wage stickiness. Second, monetary policy should
try to offset the effects of balance sheet shocks, thus allowing for deviations from price stability. Third,
foreign currency debt denomination affects the optimal monetary policy: it should be less contractionary in
response to positive domestic productivity disturbances and more so if productivity shocks originate abroad.
Fourth, financial frictions substantially decrease attractiveness not only of PPI targeting, but also of other
price targeting regimes. In contrast, they do not have a significant effect on the performance of a monetary
union agreement. Fifth, while credit frictions make nominal wage targeting deviate from the optimal policy
more than in the perfect financial markets case, it can be considered robust compared to other standard
simple policies.1
The paper proceeds as follows. Section 2 lays out the structure of our model. Section 3 discusses its
calibration. The welfare-based framework for evaluating alternative policies is presented in section 4. Section
5 discusses our main results. Section 6 concludes.
1 This last result can be seen as complementing the findings of Levin et al. (2005), who document a remarkable robustness

of a simple wage stabilization rule to model uncertainty.

2
2 Structure of the model
There are two countries in the world: Home (H) and Foreign (F ). Each country is inhabited by a continuum of
infinite-lived households, who consume a homogeneous consumption good and supply labour to a continuum
of firms. A perfectly competitive sector of capital producers combines the existing capital with investment
flows to produce the installed capital stock. Capital is managed and rented to firms by a continuum of
entrepreneurs, who use their net worth and a bank loan to finance the capital expenditures. Productivity
of each entrepreneur is subject to an idiosyncratic shock, not observed by the bank. This creates agency
problems and so interest charged by the banking sector is subject to a premium over the risk-free rate paid
by banks on households’ deposits.
There are two types of firms in each economy, each using capital and labour as inputs. Nontradable goods
producers sell their output only domestically, while tradable goods firms produce both for the local market
and for exports. Prices are denominated in the producer currency and set in a monopolistically competitive
fashion. Nontradable and tradable goods produced at home are combined with goods imported from abroad
into final consumption and investment goods in a perfectly competitive environment.
Fiscal authorities finance their expenditures on nontradable goods by collecting lump sum taxes from the
households.
Since the general setup of the Foreign country is similar to that for the Home economy, in the following
and more detailed exposition we focus on the latter. To the extent needed, variables and parameters referring
to foreign agents are marked with an asterisk. Unless stated otherwise, all variables in the derivations below
are expressed in per capita terms. Whenever aggregation across countries is needed, we make use of the
normalization of the world population to one, so that the size of Home is n and that of Foreign is 1 − n.

2.1 Households
Households in a given country are assumed to be homogenous, i.e. they have the same preferences and endow-
ments. Each household has access to complete markets for state-contingent claims, traded domestically and
providing insurance against individual (but not aggregate) income risk. This implies that any idiosyncratic
shocks among the households do not result in heterogeneity of their behaviour. Hence, we can focus on the
optimization problem of a representative household.
A typical household maximizes the following lifetime utility function:
(∞ ¸)
X · εd,t+k κ 1+ϕ
k 1−σ
Ut = Et β C − L (1)
1 − σ t+k 1 + ϕ t+k
k=0

where Et is the expectation operator conditional on information available at time t, β is the discount
rate, σ is the inverse of the elasticity of intertemporal substitution, κ is the weight of leasure in utility and
ϕ denotes the inverse of the Frisch elasticity of labour supply. The instantaneous utility is thus a function
of a consumption bundle Ct , to be defined below, and labour effort Lt . The utility is also affected by a
consumption preference shock εd,t , common to all households in a given country.
The maximization of (1) is subject to a sequence of intertemporal budget constraints of the form:

PC,t Ct + Rt−1 Dt+1 + Rt−1 Bt+1 = Wt Lt + RK,t Kt + DivH,t + DivN,t + Tt + T RE,t + Dt + Bt (2)

where PC,t is the price of the consumption bundle Ct , Wt is the nominal wage rate, RK,t denotes house-
holds’ income from renting a unit of capital Kt , DivH,t and DivN,t are dividends from tradable and non-
tradable goods producers, respectively, Tt stands for lump sum government transfers net of lump sum taxes
and T RE,t denotes wealth received from exiting (net of transfers to surviving and entering) entrepreneurs.
Households hold their financial wealth in form of bank deposits Dt , paying the risk-free (gross) rate Rt . As
in Chari et al. (2002), we assume complete markets for state-contingent claims. This means that households
have also access to state-contingent bonds Bt , paying the expected gross rate of return Rt .

3
2.1.1 Consumption choice
The first order conditions to the representative consumer maximization problem imply the following conven-
tional stochastic Euler equation:
½ ¾
λC,t+1 Rt
βEt =1 (3)
λC,t ΠC,t+1
where ΠC,t denotes CPI inflation (expressed in gross terms) and λC,t is the marginal utility of consumption,
defined as:

λC,t = εd,t Ct−σ (4)
The consumption bundle Ct consists of final tradable goods CT,t and nontradable goods CN,t , aggregated
according to:
γc 1−γc
CT,t CN,t
Ct = (5)
γcγc (1 − γc )1−γc
where γc is the share of tradable goods in total consumption.
The index of tradable goods is defined by:
α 1−α
CH,t CF,t
CT,t = (6)
αα (1 − α)1−α
where CH,t is the bundle of home-made tradable goods consumed at home, CF,t is the bundle of foreign-
made tradable goods consumed at home and α denotes the share of home goods in the home basket of
tradable goods.
The indices of nontradable and both types of tradable goods are in turn given by the following aggregators
of individual varieties:
·Z 1 φN −1
¸ φφN−1
N
CN,t = Ct (zN ) φN
dzN (7)
0

·Z 1 φH −1
¸ φφH−1
H
CH,t = Ct (zH ) φH
dzH (8)
0

·Z 1 φF −1
¸ φφF−1
F
CF,t = Ct (zF ) φF
dzF (9)
0

where φN , φH , and φF are the elasticities of substitution across varieties of a given type.
The sequence of intratemporal optimization problems implies the following demand functions for each
variety of goods:
µ ¶−φN µ ¶−1
Pt (zN ) PN,t
Ct (zN ) = (1 − γc ) Ct (10)
PN,t PC,t
µ ¶−φH µ ¶−1 µ ¶−1
Pt (zH ) PH,t PT,t
Ct (zH ) = γc α Ct (11)
PH,t PT,t PC,t
µ ¶−φF µ ¶−1 µ ¶−1
Pt (zF ) PF,t PT,t
Ct (zF ) = γc (1 − α) Ct (12)
PF,t PT,t PC,t
where Pt (zj ) is the price of variety zj , while the composite price indexes are defined as follows:
·Z 1 ¸ 1−φ
1
N
PN,t = Pt (zN )1−φN dzN (13)
0

4
·Z 1 ¸ 1−φ
1
H
1−φH
PH,t = Pt (zH ) dzH (14)
0

·Z 1 ¸ 1−φ
1
F
1−φF
PF,t = Pt (zF ) dzF (15)
0

α 1−α
PT,t = PH,t PF,t (16)

γc 1−γc
PC,t = PT,t PN,t (17)

2.1.2 Wage setting
Each household j supplies monopolistically one distinctive type of labour Lt (j), which is aggregated with
labour services of other households in a given country into a homogenous labour input according to the
formula:
·Z 1 φW −1
¸ φφW−1
W
Lt = Lt (j) φW
dj (18)
0

We follow Erceg et al. (2000) and assume that only a fraction 1 − θW of households can renegotiate
their wage contracts in each period, while wages of the remaining households are indexed to the steady-state
consumer price (CPI) inflation.
Households that are allowed to reset their wages take into account that they may not be allowed to do so
for some time, so they solve the following maximization problem:
(∞ · ¸)
X κ Wt (j) k
k k 1+ϕ
Et θW β − Lt+k (j) + λC,t+k Π̄ Lt+k (j) (19)
1+ϕ PC,t+k C
k=0

subject to the sequence of labour demand constraints:
· ¸−φW
Wt (j) k
Lt+k (j) = Π̄ Lt+k (20)
Wt+k C
where the aggregate wage index is given by:
·Z 1 ¸ 1−φ1
W
1−φW
Wt = Wt (j) dj (21)
0

The first-order condition associated with the optimization problem (19) can be written as:
(∞ · ¸ )
X W t (j) φW
k
Et θW βk Π̄k − M RSt+k (j) λC,t+k Lt+k (j) = 0 (22)
PC,t+k C φW − 1
k=0

where M RSt is the marginal rate of substitution between consumption and labour defined as:

κLt (j)ϕ
M RSt (j) = (23)
λC,t
Since all households that can renegotiate their wage contracts face an identical optimization problem,
they set the same optimal wage W̃t , which allows us to rewrite the first-order condition (22) in a recursive
way as:
à !1+ϕφW
W̃t φW ΦW,t
= (24)
PC,t φW − 1 ΨW,t

5
where the auxiliary state variables ΦW,t and ΨW,t are defined as:
µ ¶φW (1+ϕ) (µ ¶φ (1+ϕ) )
Wt 1+ϕ ΠC,t+1 W
ΦW,t = κ Lt + βθW Et ΦW,t+1 (25)
PC,t Π̄C
µ ¶φW (µ ¶φ −1 )
Wt ΠC,t+1 W
ΨW,t = λC,t Lt + βθW Et ΨW,t+1 (26)
PC,t Π̄C
Given the wage setting scheme described above, the aggregate wage index evolves according to the fol-
lowing formula:
h ¡ ¢1−φW i 1−φ1
Wt = θW Wt−1 Π̄C + (1 − θW )W̃t1−φW W
(27)

2.2 Capital producers
There is a continuum of perfectly competitive capital producers, owned by the households. At the end of each
period, they buy capital from entrepreneurs and combine it with investment goods to produce new installed
capital, which is then sold to entrepreneurs.
Consistently with the market clearing on the capital market, the total amount of capital purchased by
capital producers must be equal to total undepreciated capital stock in the economy. The economy-wide
capital available for production Kt+1 evolves then according to the formula:

Kt+1 = (1 − τ )Kt + εi,t (1 − ΓI,t ) It (28)
where It is investment and τ is the depreciation rate. As in Christiano et al. (2005), capital accumulation
is subject to investment-specific technological progress εi,t and adjustment cost represented by a function
ΓI,t , defined as:
µ ¶2
ςI It
ΓI,t = −1 (29)
2 It−1
The optimization problem of a representative capital producer is to maximize the present discounted
value of future profits:

( ∞
)
X λC,t+k
Et βk [QT,t+k PC,t+k ((1 − τ )Kt+k + εi,t+k (1 − ΓI,t+k ) It+k − Kt+k ) − PI,t+k It+k ] (30)
PC,t+k
k=0

where PI,t is the price of investment goods It and QT,t is the real price of installed capital (Tobin’s Q).
The first order condition to this optimization problem yields the following investment demand equation:
½ 2 ¾
PI,t ¡ 0
¢ λC,t+1 It+1 0
= εi,t 1 − ΓI,t − It ΓI,t QT,t + βEt εi,t+1 Γ QT,t+1 (31)
PC,t λC,t It I,t+1
The final investment good is produced in a similar fashion as the final consumption good, which implies
the following definitions:
γI 1−γI
IT,t IN,t
It = (32)
γiγi (1 − γi )1−γI
α 1−α
IH,t IF,t
IT,t = (33)
αα (1 − α)1−α
γi 1−γi
PI,t = PT,t PN,t (34)
Hence, while we allow for differences in the tradable-nontradable composition between the final consump-
tion basket and the investment basket (i.e. γc need not be equal to γi ), we assume for simplicity that the
structure of the purely tradable component is identical for both types of goods.

6
2.3 Entrepreneurs and banks
Capital services to firms are supplied by a continuum of risk-neutral entrepreneurs, indexed by zE . At the
end of period t, each entrepreneur purchases installed capital Kt+1 (zE ) from capital producers, partly using
its own financial wealth Nt+1 (zE ) and financing the remainder by a bank loan BE,t+1 (zE ):

BE,t+1 (zE ) = QT,t PC,t Kt+1 (zE ) − Nt+1 (zE ) ≥ 0 (35)
After the purchase, each entrepreneur experiences an idiosyncratic productivity shock, which converts its
capital to aE (zE )Kt+1 (zE ), where aE is a random variable, distributed independently over time and across
entrepreneurs, with a cumulative density function F (aE ) and a unit mean. Following Christiano et al. (2003),
we assume that this distribution is log normal, with a time-varying standard deviation of log aE equal to
εe,t σE , known to entrepreneurs before their capital decisions.
Next, each entrepreneur rents out capital services, treating the rental rate RK,t+1 as given. Since the
mean of an idiosyncratic shock is equal to one, the average rate of return on capital earned by entrepreneurs
can be written as:

RK,t+1 + (1 − τ )QT,t+1 PC,t+1
RE,t+1 = (36)
QT,t PC,t
and the rate of return earned by an individual entrepreneur is aE (zE )RE,t+1 .
Idiosyncratic shocks are observed by entrepreneurs but not by banks, so lending involves agency costs,
reflected in a debt contract between these two parties. The contract specifies the size of the loan BE,t+1 (zE )
and the gross non-default interest rate RB,t+1 (zE ) charged by the bank. The solvency criterion can also
be defined in terms of a cutoff value of idiosyncratic productivity, denoted as ãE,t+1 (zE ), such that the
entrepreneur has just enough resources to repay the loan:2

ãE,t+1 RE,t+1 QT,t PC,t Kt+1 (zE ) = RB,t+1 BE,t+1 (zE ) (37)
Entrepreneurs with aE below the threshold level go bankrupt. Their all resources are taken over by the
banks, after they pay proportional and nontradable monitoring costs µ.
Banks finance their loans by issuing time deposits to households at the risk-free interest rate Rt . The
banking sector is assumed to be perfectly competitive and owned by risk-averse households. This together
with risk-neutrality of entrepreneurs implies a financial contract insulating the lender from any aggregate
risk.3 Hence, interest paid on a bank loan by entrepreneurs is state contingent and guarantees that banks
break even in every period. The aggregate zero profit condition for the banking sector can be written as:

(1 − F1,t+1 ) RB,t+1 BE,t+1 + (1 − µ) F2,t+1 RE,t+1 QT,t PC,t Kt+1 = Rt BE,t+1 (38)
or equivalently (using (37)):

RE,t+1 QT,t PC,t Kt+1 [ãE,t+1 (1 − F1,t+1 ) + (1 − µ) F2,t+1 ] = Rt BE,t+1 (39)
where
Z ãE,t
F1,t = dF (aE ) (40)
0
Z ãE,t
F2,t = aE dF (aE ) (41)
0

and the analytical formulas for F1,t and F2,t , making use of the log-normal assumption for F (aE ), are
given in the appendix.
The equilibrium debt contract maximizes welfare of each individual entrepreneur. We define it in terms
of expected end-of-contract net worth relative to the risk-free alternative, which is holding a domestic bond:
2 In order to save on notation, in what follows we use the result established later on, according to which the cutoff productivity

ãE (zE ) and the non-default interest paid on a bank loan RB,t+1 (zE ) is identical across entrepreneurs.
3 Given the infinite number of entrepreneurs, the risk arising from idiosyncratic shocks is fully diversifiable.

7
(R ∞ )
ãE,t
(RE,t+1 QT,t PC,t Kt+1 (zE )aE (zE ) − RB,t+1 BE,t+1 (zE ))
Et (42)
Rt Nt+1 (zE )
The first-order condition to this optimization problem can be written as:
( RE,t+1 )
Rt [1³− ãE,t+1 (1 − F1,t+1 ) − F2,t+1 ] + ´
Et 1−F RE,t+1 =0 (43)
+ 1−F1,t+1 −µã1,t+1
E,t+1 F
0 Rt [ãE,t+1 (1 − F1,t+1 ) + (1 − µ) F2,t+1 ] − 1
1,t+1

As can be seen from (43), the ex ante external financing premium arises because of monitoring costs: if µ
is set to zero, the expected rate of return on capital is equal to the risk-free interest rate and so the financial
markets are perfect.
Equation (43), together with the bank zero profit constraint (39) defines the optimal debt contract in
terms of the cutoff value of the idiosyncratic shock ãE,t+1 and the leverage ratio %t , defined as:
QT,t PC,t Kt+1
%t = (44)
Nt+1
It is easy to verify that these two contract parameters are identical across entrepreneurs. There are
two important implications of this result, facilitating aggregation. First, the loan amount taken by each
entrepreneur is proportional to his net worth. Second, the rate of interest paid to the bank is the same for
each non-defaulting entrepreneur:
ãE,t+1 RE,t+1 %t
RB,t+1 = (45)
%t − 1
Proceeds from selling capital, net of interest paid to the bank, constitute end of period net worth. To
capture the phenomenon of ongoing entries and exits of firms and to ensure that entrepreneurs do not
accumulate enough wealth to become fully self-financing, we assume that each period a randomly selected
and time-varying fraction 1 − εν,t υ of them go out of business, in which case all their financial wealth is
rebated to the households. At the same time, an equal number of new entrepreneurs enters, so that the
total number of entrepreneurs is constant. Those who survive and enter receive a transfer TE,t from the
households. This ensures that both entrants and surviving bankrupt entrepreneurs have at least a small but
positive amount of wealth, without which they would not be able to buy any capital.
Aggregating across all entrepreneurs and using (39) yields the following law of motion for net worth in
the economy:
· µ ¶ ¸
µF2,t RE,t QT,t−1 PC,t−1 Kt
Nt+1 = εν,t υ RE,t QT,t−1 PC,t−1 Kt − Rt−1 + BE,t + TE,t (46)
BE,t
The term in the square brackets represents the total revenue from renting and selling capital net of interest
paid on bank loans, averaged over both bankrupt and non-bankrupt entrepreneurs. The second term in the
round brackets is the expected excess cost of borrowing from the bank over the risk-free rate, and so can be
interpreted as the ex post external finance premium.
While discussing our results, we also consider a situation in which bank loans taken by entrepreneurs in
the home country are denominated in foreign rather than domestic currency. The modifications needed to
implement this variant are presented in the appendix.

2.4 Firms
2.4.1 Production technology
There exist a continuum of identically monopolistic competitive firms in each of the nontradable and trad-
able sectors, owned by households and indexed by zN and zH , respectively. The production technology is
homogenous with respect to labour and capital inputs:

Yt (zN ) = εn,t Lt (zN )1−ηN Kt (zN )ηN (47)

8
Yt (zH ) = εt,t Lt (zH )1−ηH Kt (zH )ηH (48)
where ηN and ηH are sector-specific output elasticities with respect to capital input, while εn,t and εt,t
are sector-specific productivity parameters. The output indexes are given by the Dixit-Stiglitz aggregators:
·Z 1 φN −1
¸ φφN−1
N
YN,t = Yt (zN ) φN
dzN (49)
0

·Z 1 φH −1
¸ φφH−1
H
YH,t = Yt (zH ) φH
dzH (50)
0

Since all firms in a given sector operate technologies with the same relative intensity of productive factors
and face the same prices for labour and capital inputs (factor markets are homogeneous), cost minimization
implies the following sector-specific capital-labour relationships:
Wt LN,t 1 − ηN Wt LH,t 1 − ηH
= = (51)
RK,t KN,t ηN RK,t KH,t ηH

2.4.2 Price setting
Firms producing nontradable goods set their prices according to the Calvo (1983) staggering mechanism.
Only a fraction 1 − θN of them set their prices in a forward-looking manner, while the prices of firms that
do not receive a price signal are fully indexed to the steady-state inflation in the nontradable sector Π̄N .
Firms that are allowed to reoptimize realize that they may not be allowed to do so for some time, hence
their price-setting problem is to maximize the expected present discounted value of future profits:
(∞ )
X £ ¤
k k λC,t+k k
Et θN β Yt+k (zN ) Pt (zN )Π̄N − PN,t+k M CN,t+k (52)
PC,t+k
k=0

subject to the sequence of demand constraints:
· ¸−φN
Pt (zN ) k
Yt+k (zN ) = Π̄ YN,t+k (53)
PN,t+k N
where M CN,t is the real marginal cost (identical across nontradable goods firms) defined as:
µ ¶1−ηN µ ¶ηN
1 Wt RK,t
M CN,t = (54)
PN,t εn,t 1 − ηN ηN
The first-order condition associated with the profit-maximization problem faced by reoptimizing firms
can be written as:
(∞ · ¸ )
X
k k λC,t+k k φN
Et θN β Yt+k (zN ) Pt (zN )Π̄N − PN,t+k M CN,t+k = 0 (55)
PC,t+k φN − 1
k=0

There are no firm-specific shocks in the model, so all firms that are allowed to reset their price in a forward-
looking manner select the same optimal price P̃N,t , which implies the following recursive representation of
the first-order condition (55):

P̃N,t φN ΦN,t
= (56)
PN,t φN − 1 ΨN,t
where (µ ¶φN )
λC,t ΠN,t+1
ΦN,t = M CN,t PN,t YN,t + βθN Et ΦN,t+1 (57)
PC,t Π̄N

9
(µ ¶φN −1 )
λC,t ΠN,t+1
ΨN,t = PN,t YN,t + βθN Et ΨN,t+1 (58)
PC,t Π̄N
The expression for the evolution of the home nontradable goods price index can be written as follows:
h ¡ ¢1−φN i 1
1−φN 1−φN
PN,t = θN PN,t−1 Π̄N + (1 − θN )P̃N,t (59)

The price-setting problem solved by firms producing tradable goods is similar and leads to first-order
conditions and price indices analogous to equation (55) and (59), respectively. We assume that prices are
set in the producer currency and that the international law of one price holds for each tradable variety.
Therefore, the price of home goods sold abroad and that of foreign goods sold domestically are given by:

Pt∗ (zH ) = ERt−1 Pt (zH ) Pt (zF ) = ERt Pt∗ (zF ) (60)
where ERt is the nominal exchange rate expressed as units of domestic currency per one unit of foreign
currency.

2.5 Exchange rate dynamics
The perfect risk sharing condition implies (see Chari et al., 2002):
λ∗C,t
= Qt (61)
λC,t
where Qt is the real exchange rate defined as:

ERt PC,t
Qt = (62)
PC,t
The real exchange rate is allowed to deviate from the purchasing power parity (PPP) due to changes in
relative prices of tradable vs. nontradable goods in both countries (the internal exchange rates) and changes
in terms-of-trade, as long as there is some home bias in preferences (α 6= α∗ ). This can be demonstrated
using the price indices derived above and the law of one price conditions for tradable goods:
∗1−γ ∗
∗ Xt c

Qt = Stα−α 1−γc (63)
Xt
where the terms-of-trade St is defined as home import prices relative to home export prices:

ERt PF,t
St = (64)
PH,t
and the internal exchange rates Xt and Xt∗ are defined as:

PN,t PN,t
Xt = Xt∗ = ∗ (65)
PT,t PT,t

2.6 Monetary and fiscal authorities
We consider several variants of monetary policy regimes, including the Ramsey optimal policy. For calibration,
we assume that the monetary authority responds to the economic conditions through the following interest-
rate feedback rule:
" µ ¶φ µ ¶φdy µ ¶φdπ #1−ρ
ρ ΠC,t π Yt ΠC,t
Rt = Rt−1 R̄ εm,t (66)
Π̄C Yt−1 ΠC,t−1

10
where Yt is total output, Ȳ is its steady state level, R̄ is the steady state interest rate and εm,t is a
monetary policy shock.
The fiscal authority is modelled in a very simplistic fashion: government expenditures and transfers to
the households are fully financed by lump sum taxes, so that the state budget is balanced each period. The
government spending is fully directed at nontradable goods and is modelled as a stochastic process εg,t . Given
our representative agent assumption, Ricardian equivalence holds in the model.

2.7 Market clearing conditions
2.7.1 Goods markets
The model is closed by imposing the following market clearing conditions. Output of each firm producing
non-tradable goods is either consumed domestically, spent on investment, purchased by the government or
used by banks to cover monitoring costs. Similarly, all tradable goods are consumed or invested domestically
or abroad. Using these conditions, the demand functions (10), (11) and (12) together with their analogs for
investment and government goods, the output indexes given by (49) and (50), and taking into account the
size of both countries, one can write the aggregate output in the two sectors at home as:

PC,t PI,t
YN,t = (1 − γc ) Ct + (1 − γi ) It + Gt + (67)
PN,t PN,t
−1
+µF2,t RE,t QT,t−1 PC,t−1 Kt PN,t


PC,t 1 − n ∗ ∗ PC,t
YH,t = αγc Ct + α γc ∗ Ct∗ + (68)
PH,t n PH,t

PI,t 1 − n ∗ ∗ PI,t
+αγi It + α γi ∗ It∗
PH,t n PH,t

Total output Yt is the sum of output produced in the nontradable and tradable sectors:

Yt = YN,t + YH,t (69)
or in nominal terms:

Pt Yt = PN,t YN,t + PH,t YH,t (70)
where Pt is the implicit total output deflator. We use these definitions to construct the overall producer
price index (PPI) inflation as a weighted average of inflation in both sectors, with weights consistent with
the Laspeyres concept:
PN,t−1 YN,t−1 PH,t−1 YH,t−1
Πt = ΠN,t + ΠH,t (71)
Pt−1 Yt−1 Pt−1 Yt−1

2.7.2 Factor markets
Equilibrium in factor markets requires:
Z 1 Z 1
Lt = Lt (zN )dzN + Lt (zH )dzH (72)
0 0
Z 1 Z 1
Kt = Kt (zN )dzN + Kt (zH )dzH (73)
0 0

which can be rewritten using (47), (48), (51) and the demand sequences like in (53) as:

11
µ ¶η µ ¶η µ ¶η ¶ηH µ
1 − ηN N RK,t N YN,t 1 − ηH RK,t H YH,t
Lt = ∆N,t + ∆H,t (74)
ηN Wt εn,t ηH Wt εt,t
µ ¶1−ηN µ ¶1−ηN µ ¶1−ηH µ ¶1−ηH
ηN Wt YN,t ηH Wt YH,t
Kt = ∆N,t + ∆H,t (75)
1 − ηN RK,t εn,t 1 − ηH RK,t εt,t
where ∆N,t and ∆H,t are the measures of price dispersion in the nontradable and tradable sector:
Z 1 µ ¶−φN Z 1 µ ¶−φH
PN,t (zN ) PH,t (zH )
∆N,t = dzN ∆H,t = dzH (76)
0 PN,t 0 PH,t
The following laws of motion for the two dispersion indexes can be derived using (59):
à !−φN µ ¶−φN
P̃N,t Π̄N
∆N,t = (1 − θN ) + θN ∆N,t−1 (77)
PN,t ΠN,t
à !−φH µ ¶−φH
P̃H,t Π̄H
∆H,t = (1 − θH ) + θH ∆H,t−1 (78)
PH,t ΠH,t

2.7.3 Financial markets
Finally, in equilibrium, household deposits at banks must be equal to total funds lent to entrepreneurs:

Dt = BE,t (79)

2.8 Exogenous shocks
There are eight stochastic disturbances hitting each economy. These concern: productivity in the tradable
sector, productivity in the nontradable sector, consumption preferences, government spending, investment-
specific technology, survival of entrepreneurs, idiosyncratic riskiness and the monetary policy. The log of
each shock follows a linear first-order autoregressive process, except for the monetary policy shock, which is
assumed to be white noise.

3 Calibration
We calibrate our model to the euro area economy, setting its size in our two-country world to 0.25. The
parameters for the rest of the world are assumed to be identical to those in the euro area. Our calibration
proceeds in two steps. We first match the key steady-state ratios and set the other structural parameters so
that they are consistent with the estimated version of the New Area-Wide Model (NAWM), documented in
Christoffel et al. (2008). Parameterization of the financial frictions block is based on Bernanke et al. (1999)
and Christiano et al. (2007). In the next step, the inertia and volatility of stochastic disturbances are chosen
to match the moments of a standard set of euro area macroaggregates, augmented by two financial variables.
These are the debt of the enterprise sector and the spread between loans to firms and the risk-free rate. The
results of the calibration exercise are reported in Tables 1 to 4 and the resulting variance decomposition is
shown in Table 5.

Tables 1 to 5 about here
Our model replicates the standard deviations of GDP and its main components. It significantly underes-
timates the volatility of the short-term interest rate and roughly captures that of inflation. As regards our
two financial variables, there is a trade-off in matching the standard deviation of entrepreneurs’ debt and
that of the external financing premium. We chose to keep the former somewhat lower than observed in the
data, and the latter much higher. This choice is motivated by the fact that there may be some measurement
problems with the euro-wide data as the volatility of the corporate credit premium is an order of magnitude

12
smaller than that found in the US.4 Turning to other moment matching results, our model gets persistence
and cyclical behaviour of most of the variables of interest more or less right, although the fit for investment
can be seen as disappointing, given the model’s focus on frictions in financing capital expenditures. It is
also worth noting that while our model makes the premium less countercyclical than in the data, it some-
what exaggerates its negative correlation with investment. Clearly, better fit in this dimension would require
allowing for financial frictions also in the household sector.

4 Welfare-based evaluation of alternative policies
In our framework, a natural welfare criterion is the discounted sum of expected utility flows given by (1),
averaged over all households:

X · Z 1 ¸
k εd,t+k 1−σ κ
Ut = Et β C − Lt+k (j)1+ϕ dj (80)
1 − σ t+k 0 1+ϕ
k=0

Using a general form of labour demand sequences (20), our welfare criterion can be rewritten in a recursive
form:
εd,t 1−σ κ
Ut = C − ∆W,t L1+ϕ + βEt {Ut+1 } (81)
1−σ t 1+ϕ t

where the wage dispersion index ∆W,t is defined as:
Z 1 µ ¶−φW (1+ϕ)
Wt (j)
∆W,t = dj (82)
0 Wt
and (using (27)) subject to the following law of motion:
à !−φW (1+ϕ) µ ¶−φW (1+ϕ)
W̃t Π̄C Wt−1
∆W,t = (1 − θW ) + θW ∆W,t−1 (83)
Wt Wt

It is clear from (81) that in our model welfare depends positively on the level of consumption and negatively
on wage and price dispersion, with the latter relationship following from the market clearing condition for
the aggregate labour input (74).
We define the optimal policy in our model as the Ramsey cooperative equilibrium, in which both cen-
tral banks implement policies that maximize the weighted average of welfare (as defined above) of the two
regions, with the weights given by the population size.5 As in Woodford (2003), we consider policies under
commitment in a timeless perspective. Under these restrictions, our benchmark generates the best possible
allocation in our two-region world. In principle, there is no guarantee that this policy maximizes welfare of a
representative consumer in each region. Coenen et al. (2008) show that the Nash equilibrium, in which each
central bank maximizes welfare of its own country taking as given the other central bank’s action, might yield
higher welfare from an individual country’s perspective so that the gains from cooperation are negative. We
leave this more complex6 analysis of non-cooperative policies for future research and will refer henceforth to
the cooperative equilibrium as optimal.
In order to build intuition for the optimal policy outcomes, we compare them to those obtained under
simple policy variants. These include various forms of strict inflation targeting and, following the findings in
Levin et al. (2005), nominal wage targeting. We also consider the case of a full monetary integration, defined
as the same cooperative equilibrium of the benchmark case, except that the exchange rate between the two
regions is fixed.
4 For instance, Gilchrist et al. (2009) estimate the standard deviation of credit spreads in the US at about 300 bps., compared

to 43 bps. in our euro area data.
5 The first order conditions of the welfare maximization problem of the policy maker(s) are computed using G. Lombardo’s

lq solution routine, downloadable from http://home.arcor.de/calomba.
6 As it is well known, the Nash equilibrium depends on the choice of instrument defining the policy game and the concept of

equilibrium (open loop vs. closed loop).

13
We assess the welfare implications of the alternative monetary policy strategies by taking a second-order
approximation of all model equations, including the first-order conditions of the welfare maximization problem
of the policy maker.7 Such a numerical approach yields a correct ranking of alternative policies and has been
used in many analyses of optimal policy (see e.g. Schmitt-Grohe and Uribe, 2006; Coenen et al., 2008).8
We evaluate each policy by calculating the welfare loss, expressed in terms of the proportion of each
period’s consumption that the typical household in the home economy would need to give up in a deterministic
world so that its welfare is equal to the expected conditional utility in the stochastic world. More precisely,
we calculate Ω that satisfies the following equation:

∞ µ ¶ Ã ·µ ¶ ¸1−σ !
X εd,t+k 1−σ κ 1 1 Ω κ
Et βk C − ∆W,t+k L1+ϕ = 1− C − L1+ϕ (84)
1 − σ t+k 1+ϕ t+k
1−β 1−σ 100 1+ϕ
k=0

where variables without time subscripts denote their respective steady state values and the starting point
for the left hand side of (84) is the ergodic mean of the cooperative equilibrium.

5 Results
5.1 Optimal policy in a simple model with capital accumulation
Our main objective is to demonstrate how the presence of financial frictions changes the optimal policy
response. To make the exposition more transparent and comparable to the previous literature, we first
consider a simplified version of our model and then build it up, explaining the policy implications of each
addition. More specifically, our departure point is a perfectly symmetric two-country world with the following
features: perfect financial markets, flexible wages, only tradable goods, no home bias and no government
purchases. This special case can be easily obtained from the full specification described in section 2 by
setting a subset of parameters to appropriate values. In other words, we consider a standard two-country
New Keynesian model with capital accumulation and analyze how adding to it financial frictions changes the
optimal monetary policy.
Before we move on to the results, one remark is in order. Remember that we calibrate all parameters
(and shock volatilities in particular) using the fully-fledged version of our model. This means that its simple
variants do not necessarily retain a solid empirical basis if the parameters are kept unchanged. Therefore,
in order to facilitate comparisons, we normalize all welfare losses presented below by the ratio of output
variance in a given version (under the Taylor rule) to output variance in the full version of our model.9
Table 6 present the welfare losses (relative to the optimal cooperative policy) of a set of simple policies in
our benchmark model with perfect or imperfect financial markets. It is well known (see e.g. a recent review
by Engel, 2009) that in the standard open economy New Keynesian model with producer currency pricing,
PPI targeting replicates the optimal policy outcomes. As can be seen from Table 6, this is also the case
in the model with endogenous capital accumulation.10 The losses associated with keeping consumer prices
or the exchange rate stable are non-negligible but do not exceed 0.08% of steady-state consumption, while
the Taylor rule performs worst. These losses are almost entirely due to technology disturbances, while the
contribution of other shocks (preference and investment-specific in this simple model version) is very close to
zero.

Table 6 about here
7 Thecalculations are performed in Dynare 4, which can be downloaded from http://www.cepremap.cnrs.fr/dynare.
8 Analternative would be to use a linear-quadratic approximation described in Benigno and Woodford (2005), which is a
generalization to Rotemberg and Woodford (1998). As discussed by Benigno and Woodford (2006), a clear advantage of this
analytical approach is that it helps to gain insight into fully optimal policy. However, given the size and complexity of our
model, following this way seems to be of little use, so we opt for a more practical and less time-consuming method.
9 This is motivated by the fact that welfare losses are approximately proportional to the variance of stochastic disturbances.

Therefore, our normalization can be thought of as a proportional correction of all shock volatilities so that the standard deviation
of output in a given model version matches that observed in the data.
10 As demonstrated by Edge (2003), the utility-based loss function of a policy maker in a standard New Keynesian model is

modified if one allows for endogenous capital accumulation. Our estimates show that the effects of this modification for optimal
policy prescriptions can be ignored in practice.

14
If financial markets are imperfect, PPI targeting is no longer optimal. The welfare loss associated with
this policy amounts to less than 0.05% of steady-state consumption. This number may appear rather small.
However, as Table 7 reveals, the consequences of introducing financial frictions turn out to be an order of
magnitude larger than those related to home bias, habits, nontradable goods or government expenditures.
One can also note that the presence of financial frictions makes the monetary union or the Taylor rule
relatively more attractive, while the opposite holds true for CPI targeting.

Table 7 about here
In order to gain insight into an optimal monetary policy conduct when financial markets are imperfect,
we compare the impulse responses under the cooperative regime to those implied by PPI targeting. Figure
1 depicts the dynamic responses to a positive productivity shock in the home economy. If both central
banks target domestic producer prices, the shock is clearly expansionary, although there is a short-lived
contraction in foreign economy’s output. Since entrepreneurs’ balance sheets improve, the external finance
premium goes down, which leads to an acceleration of investment. No home bias in preferences implies that
the real exchange rate is constant (see equation (63)), so real interest rates in both countries must respond
symmetrically. However, nominal rates at home fall more than abroad, so the exchange rate depreciates.
Under the optimal cooperative regime, there is also an expansion in the economic activity, but it is more
moderate, especially on impact. To achieve this, both central banks actually tighten their policy before
letting the short-term real interest rates follow closely the path of PPI targeting. As a result, the decline
in the external financing premium of the home country is more than halved and its response abroad is even
slightly positive. The reason why the optimal policy attempts to dampen the movements in the risk premium
is that they act as a distortionary tax on investment, the excessive fluctuations of which are inefficient. In
other words, the policy maker faces a trade-off between the costs of price dispersion and inefficiencies related
to movements in the financing premium. It is also worth noting that the optimal policy delivers smaller
responses of the exchange rate and consumer price inflation in both countries.

Figure 1 about here
As in the frictionless case, the welfare implications of other shocks (preference and investment-specific
shocks, as well as two shocks related to entrepreneurs, i.e. survival rate and riskiness) lumped together are
an order of magnitude smaller than those of productivity shocks. It is interesting to note, however, that in
this case the fixed exchange rate regime performs better than strict PPI targeting. This observation applies
for any of these shocks considered individually.
We take a closer look at a negative survival rate shock as this shock is sometimes argued to reflect net
worth destruction observed in the recent financial turmoil. The dynamic responses are shown in Figure 2. By
raising the external financing premium, this shock acts like a cost-push shock, so keeping prices stable requires
tightening of the monetary policy. The optimal response tries to strike a balance between the negative effects
of price dispersion and an inefficient rise in the external financing premium. As a result, central banks try
to offset net worth destruction resulting from the shock with an initial easing of the monetary policy, which
allows to dampen the response in the premium at the expense of a rise in inflation and large swings in the
nominal exchange rate. If both countries agree to fix their exchange rate, they come closer to the optimum
than under PPI targeting. The reason is that the union case allows for some increase in inflation and a
short-run expansion, which helps to limit an increase in the premium.

Figure 2 about here
While the general prescriptions for the optimal policy facing financial frictions and net worth shocks
developed above are broadly consistent with the Carlstrom et al. (2009) analysis abstracting away from
capital accumulation, one remark is in order. In their model, optimal policy is expansionary in response to
a negative net worth shock in terms of cumulative real rates, but they actually increase on impact only to
decline below levels implied by price stability, so the initial rise of the risk premium is higher under optimal
policy. This results in a rather unintuitive conclusion that introducing risk premia will lead the central
bank to magnify their movements compared to the strict inflation targeting regime. The authors conjecture
that a more elaborate model, featuring demand-side effects via endogenous capital accumulation, is going to

15
preserve this result. In contrast, our model implies that the initial response of optimizing policy makers to
net worth destruction will be easing on impact, which we consider more realistic.

5.2 Debt denomination
In the simple model considered so far, the two economies were perfectly symmetric. In this section we revisit
the case when one of the country’s entrepreneurial debt is denominated in the other country’s currency. We
will refer to this case as debt euroization and contrast its welfare implications with the baseline variant, i.e.
domestic currency debt denomination. The results are reported in Table 8, which splits the shocks by their
origin.

Table 8 about here
The most striking result is that if domestic entrepreneurs’ debt is denominated in the foreign currency,
PPI targeting nearly replicates the optimal response to home productivity shocks, while the other regimes
somewhat lose in attractiveness. In contrast, strict producer price inflation stability performs significantly
worse than CPI targeting, and even more so compared to the union case, if productivity shocks originate
abroad. On balance, if productivity volatility is equal in both economies, the welfare ranking of alternative
regimes remains intact compared to the non-euroized case. However, if productivity shocks abroad are on
average sufficiently larger than at home, then the euroized economy may find itself better-off having the
exchange rate fixed.
The intuition for this result is as follows. We have seen in our baseline variant that a positive productivity
shock at home leads to an expansion in economic activity and depreciation of the exchange rate, both under
the optimal policy and PPI targeting. If entrepreneurs’ debt is denominated in the foreign currency, however,
the depreciation has also an adverse effect on their balance sheets, which cools down the boom. This is
confirmed by the impulse responses presented in Figure 3. Comparing them to those shown in Figure 1 for
the case of domestic currency debt denomination reveals that the drop in the external financing premium is
now substantially smaller. Interestingly, this exchange rate related additional balance sheet channel implies
that the optimal policy no longer tightens on impact but rather lets the real interest rates fall, like under PPI
targeting. By construct, this channel does not operate if the exchange rate is fixed, so the responses for the
union case in the euroized and non-euroized cases are the same. Since keeping the exchange rate constant
requires initial tightening, the responses under this regime are now further away from the optimum.

Figure 3 about here
An analogous reasoning can be used to see why PPI targeting deviates more from the optimal policy if
entrepreneurs’ debt is denominated in the foreign currency and productivity shocks originate abroad. Since
in this case the exchange rate appreciates, the balance sheets of entrepreneurs improve and so the economic
expansion at home is magnified. Therefore, as can be seen from Figure 4, the optimal policy now tightens
much more on impact and so moves further away from the easing policy required to stabilize PPI inflation.
With our parametrization, this effect is strong enough to make the fixed exchange rate a relatively more
attractive option.

Figure 4 about here
More generally, the policy ranking obtained in the euroized case for foreign productivity shocks depends
on the extent of financial market imperfections and the size of leverage. If financial frictions are substantial
and entrepreneurs run sufficiently high debt denominated in the foreign currency, the balance sheet effects
related to exchange rate movements will be important and the fixed exchange rate regime will yield higher
welfare than PPI targeting. If on the other hand financial markets are close to perfect and leverage small,
stabilizing PPI inflation will be preferred. More detailed experiments reveal that our model parametrization
is quite close to this threshold: if about 55% or more of capital purchased by entrepreneurs is financed by
their net worth (compared to 50% in our baseline calibration), the welfare ranking of policies is the same as
in the non-euroized case.11
11 Alternative values of leverage were obtained by playing around with various combinations of parameters, but mainly with
the share of transfers received by entrepreneurs from the households.

16
5.3 Nontradable production
We next relax the assumption that all goods are tradable and consider a more general version of our model,
with the share of nontradable inputs in consumption and investment as in our baseline parametrization. The
welfare losses of alternative monetary regimes are reported in Table 9.
Our findings for a model with perfect capital markets are consistent with the previous literature. In
particular, PPI targeting no longer replicates the optimal policy, even though the losses are very small
in practice.12 Also, in line with Duarte and Obstfeld (2008), the presence of nontradable goods clearly
strengthens the case for exchange rate flexibility.
Adding financial frictions makes the losses from PPI targeting non-negligible. This effect is substantially
stronger than in a model where all goods are tradable. As before, the monetary union somewhat gains relative
to other regimes, but this time, at least under our parametrization, falls short of CPI targeting.
Taking a closer look at the decomposition of welfare losses by shocks when entrepreneurs’ debt is de-
nominated in the domestic currency, at least one observation warrants a comment. Contrary to the model
without nontradables, PPI targeting performs slightly worse than CPI targeting and substantially worse than
the monetary union in response to productivity shocks originating in the domestic tradable sector. However,
in this very case it is actually targeting PPI inflation in the nontradable sector that comes closest to the
optimal policy.

Table 9 about here
In order to shed some light on why the policy ranking changes if we allow for nontradable production,
we use the impulse response analysis. Figure 5 shows the dynamic responses to a home tradable sector
productivity shock under various regimes. The outcomes under optimal policy are qualitatively very similar
to those obtained in the fully tradable version of our model presented in Figure 1. Compared to strict
PPI targeting, the optimal cooperative policy tries to dampen the boom fuelled by the financial accelerator
mechanism. Importantly, however, the difference between these two policies is now more pronounced in
relative terms. In particular, the optimal policy designs nearly twice lower depreciation and a three times
lower decrease in the external finance premium. The reason why PPI targeting overexpands relatively more
than we have seen in our simple model is that the presence of the nontradable sector makes stabilizing the
overall producer inflation more difficult. This is because nontradable goods prices are less flexible, which
follows from our calibration (see Table 1), but also from the fact that they are insulated from direct effects of
exchange rate movements.13 As a result, keeping PPI constant now requires more policy easing (in relative
terms, i.e. after correcting for the fact that an increase in productivity affects only one sector of the economy),
the side effect of which is an excessive decrease in external financing premium.
If the exchange rate is pegged, the effect of the financial accelerator is nearly eliminated and so the
economic expansion is dampened too much. However, the deviations from the optimal policy turn out to
be smaller than in the PPI targeting case, hence the union case ranks better. Targeting nontradable goods
prices comes closest to optimal and so outperforms all other simple regimes. It completely eliminates price
dispersion in the nontradable sector and lets the average level of producer prices drop. Hence, it does not
require as much easing as PPI targeting and so leads to a boom that is only slightly excessive.

Figure 5 about here
Naturally, the ranking of regimes established for a model without nontradables remains valid in the
model with nontradable production as long as the share of the latter in output is sufficiently small. As
our experiments show, however, the union case dominates PPI targeting in response to foreign tradable
productivity shocks already when the share of nontradables is around 10%, so our finding about the change
in the policy ranking can be treated as robust.
If both some goods are nontradable and domestic entrepreneurs’ debt is denominated in the foreign
currency, our results are qualitatively similar to those obtained for the fully tradable and euroized case
12 Interestingly,if productivity shocks originate abroad, PPI targeting yields marginally higher welfare for the home economy
than the cooperative equilibrium. This means that in some cases implementing the cooperative policy might be problematic.
13 If the Calvo probabilities in the tradable and nontradable sectors are equal, monetary union still generates higher welfare in

response to a home tradable sector productivity shock than PPI targeting. Naturally, the difference between the performance
of these two regimes is then much smaller.

17
presented in Table 8. In particular, PPI targeting performs closest to optimal in response to domestic tradable
sector productivity shocks. This means that, under our parameterization, the presence of nontradables is not
enough to offset the stabilizing effect of euroized liabilities discussed in the previous section. In principle, this
result depends on the size of the nontradable sector. We find, however, that the fixed exchange rate regime
yields higher welfare than PPI targeting in response to domestic tradable sector productivity disturbances
in a euroized economy only if the share of nontradable production in total output is at least 80%, which is
more than observed in the data (see Lombardo and Ravenna, 2009).
Finally, we note that, as before, if shocks originate in the tradable sector abroad, PPI targeting performs
worse than CPI targeting and the union. However, these two rules are beaten by nontradable goods inflation
stabilization. More generally, while targeting nontradable sector prices seems to be an attractive alternative
to stabilizing the weighted average of inflation in both sectors whenever the latter policy performs worse than
the union case, it is not so in general. Taking all shocks into account, targeting PPI in the nontradable sector
is inferior to total PPI targeting.

5.4 The role of wage stickiness
As the last step in our analysis, we extend the models discussed in the previous section for sticky wages. The
welfare losses of alternative monetary regimes are reported in Table 10.
In line with the studies using a simpler closed economy setup (see e.g. Erceg et al., 2000; Benigno and
Woodford, 2004), we find that adding wage stickiness to the frictionless version of our model makes strict
inflation stabilization policies clearly suboptimal. A particularly strong increase in welfare losses can be
observed for both variants of PPI targeting, which become inferior to CPI stabilization. On the contrary,
the performance of the fixed exchange rate and Taylor rules hardly change compared to the flexible wages
case. Both of these policies clearly dominate any of the price targeting regimes. Consistently with the results
obtained by Levine et al. (2005), the best performer is now nominal wage targeting.
When we relax the assumption on perfect financial markets, these findings become even more pronounced.
The welfare losses related to any of the three strict price stabilization policies increase by a factor of two or
more and can be considered very substantial. This can be contrasted with a very moderate change in the
performance of the monetary union and Taylor rules, particularly if entrepreneurs’ debt is denominated in
the foreign currency. Finally, even though the size of welfare losses associated with wage inflation targeting
increases by a factor of three or four, this policy remains superior to any of the alternatives considered.

Table 10 about here
As before, we build intuition for our results by looking at how different shocks contribute to the welfare
losses discussed above, starting first with domestic currency debt denomination. It is clear from Table 10
that the large welfare losses of price stabilization strategies are mainly due to productivity shocks, while the
relative decrease in performance of the remaining regimes can be mainly, or in some cases even exclusively,
attributed to other shocks. We illustrate these results in Figures 6 to 10, plotting the impulse responses to
the main shocks of interest.
We have seen in the previous sections that strict PPI targeting required monetary policy easing that was
too excessive. Looking at Figure 6, showing the responses to a tradable sector productivity shock at home,
it is clear that now this effect is much stronger. This is because if wages are sticky, their contribution to
mitigate deflationary pressures caused by an increase in productivity is reduced for a given interest rate path.
Therefore, the monetary expansion needed to keep PPI inflation unchanged is bigger and the resulting boom
magnified by the financial accelerator. At the same time, fluctuations in nominal wages are now undesired
from the welfare perspective. This means that the optimal policy has an incentive to reduce them, so it is
actually more restrictive on impact compared to the model with flexible wages. As a result, PPI targeting
and the cooperative equilibrium deviate from each other more than before. Similar mechanisms deteriorate
the performance of CPI stabilization. This regime, however, is now superior to PPI targeting as it allows for
a fall in producer prices and so does not require so much monetary expansion. The remaining policies imply
significantly more moderate wage adjustment, hence the outcomes they yield do not differ that much from
those obtained under flexible wages. In particular, the dampening of economic expansion obtained with the
wage targeting regime turns out to be close to that designed by the optimal policy.

18
Figure 6 about here
Considerations analogous to those described above also help to understand the weak performance of price
stabilization regimes in response to home nontradable sector shocks (see Figure 7). This time, however, it is
nontradable sector inflation stabilization that leads to largest welfare losses as it requires very strong policy
easing to offset deflationary pressures in this sector.

Figure 7 about here
If productivity shocks originate abroad, producer price stabilization policies in the home economy do not
require so much monetary expansion. However, the strong appreciation pressure caused by a large decrease
in interest rates abroad implies the need for excessive policy easing at home as well (see Figure 8 and 9).
Therefore, similarly to what we have seen for the shocks of a domestic origin, overall PPI targeting does
badly in response to tradable productivity shocks in the foreign economy, while nontradable PPI targeting
generates large welfare losses in response to foreign nontradable sector shocks. Contrary to producer price
stabilization, the relative attractiveness of CPI targeting declines significantly if productivity shocks come
from abroad. This is because, given deflationary pressures from the foreign economy, inflation needs to be
generated in domestic sectors to keep the price of a home consumption basket stable, which implies monetary
easing and an inefficiently large drop in the risk premium.

Figure 8 and 9 about here
We note that while wage targeting outperforms all other simple regimes in response to productivity shocks
when wages are sticky and financial markets are imperfect, it is not always the case for other shocks. In
particular, fixing the exchange rate turns out to be a better option in response to a destruction in net worth.
Even though the implications of this disturbance for welfare evaluation are very small compared to produc-
tivity shocks, it might be instructive to examine the impulse responses to this shock under different regimes,
which we plot in Figure 10. Similarly to a model with flexible wages and no nontradable production (see
Figure 2), the optimal policy is expansionary following an unexpected decrease in net worth. A qualitatively
similar response is also possible if the exchange rate is fixed. In contrast, due to a cost-push nature of the
shock, a tightening is required to keep nominal wages (and even more so prices) constant.

Figure 10 about here
Finally, we discuss how the results presented above are affected by denomination of entrepreneurs’ debt.
The most significant observation that can be taken from Table 10 is that debt euroization makes the two
PPI targeting regimes deviate less (more) from the optimal policy if productivity shocks originate at home
(abroad). This result can be understood by looking at the exchange rate movements presented for the non-
euroized case, using the intuition we have built for the simple model with flexible wages and tradable goods
only. It is clear from Figures 6 to 9 that PPI stabilization policies lead to depreciation of the exchange rate
for domestic productivity shocks, helping to dampen the boom and inefficient movements in the external
financing premia through an adverse effect on entrepreneurs’ balance sheets, while the opposite holds true
for shocks coming from abroad.

6 Conclusions
In this paper, we have analyzed and quantified how frictions in financing capital expenditures affect the
optimal monetary policy conduct in a two-country DSGE setup. Consistently with the earlier literature
using more simple and closed-economy models, we find that financial market imperfections generate a trade-off
between inflation and external financing premium stabilization, making strict inflation targeting suboptimal.
We show, however, that the welfare implications of this trade-off are non-negligible. In particular, financial
frictions substantially magnify the incentives to deviate from price stability created by such frictions if we
allow for nontradable goods and wage stickiness.
In contrast, financial market imperfections considered in our paper do not have a significant effect on the
performance of the monetary union. This means that the presence of financial frictions strengthens the case

19
for such an arrangement if cooperation between countries under flexible exchange rate regimes is difficult to
implement.
There is a number of potentially fruitful future research directions, of which we will name only two. First,
it might be interesting to revisit the literature on international monetary policy cooperation. According to
our preliminary (and not reported) calculations, gains from cooperation after introducing financial frictions
remain small, especially if mark-up shocks are absent. However, this may change if one allows for international
financial market integration, where firm balance sheets depend on foreign assets, as in Dedola and Lombardo
(2009). Second, some of the issues addressed in our paper, like debt euroization and monetary integration,
may be particularly relevant for small open economies. A realistic investigation of such cases would call
for an asymmetric setup, especially while constructing monetary policy games. We leave these interesting
extensions for future research.

References
Benigno, P., and M. Woodford (2004): “Optimal Stabilization Policy When Wages and Prices are Sticky:
The Case of a Distorted Steady State,” NBER Working Papers 10839, National Bureau of Economic
Research, Inc.
(2005): “Inflation Stabilization And Welfare: The Case Of A Distorted Steady State,” Journal of
the European Economic Association, 3(6), 1185–1236.
(2006): “Linear-Quadratic Approximation of Optimal Policy Problems,” NBER Working Papers
12672, National Bureau of Economic Research, Inc.
Bernanke, B. S., M. Gertler, and S. Gilchrist (1999): “The financial accelerator in a quantitative
business cycle framework,” in Handbook of Macroeconomics, ed. by J. B. Taylor, and M. Woodford, vol. 1
of Handbook of Macroeconomics, chap. 21, pp. 1341–1393. Elsevier.
Calvo, G. A. (1983): “Staggered prices in a utility-maximizing framework,” Journal of Monetary Eco-
nomics, 12(3), 383–398.
Carlstrom, C. T., T. S. Fuerst, and M. Paustian (2009): “Optimal Monetary Policy in a Model with
Agency Costs,” mimeo.
Chari, V. V., P. J. Kehoe, and E. R. McGrattan (2002): “Can Sticky Price Models Generate Volatile
and Persistent Real Exchange Rates?,” Review of Economic Studies, 69(3), 533–63.
Christiano, L. J., M. Eichenbaum, and C. L. Evans (2005): “Nominal Rigidities and the Dynamic
Effects of a Shock to Monetary Policy,” Journal of Political Economy, 113(1), 1–45.
Christiano, L. J., R. Motto, and M. Rostagno (2003): “The Great Depression and the Friedman-
Schwartz hypothesis,” Proceedings, pp. 1119–1215.
(2007): “Financial Factors in Business Cycles,” mimeo.
Christiano, L. J., M. Trabandt, and K. Walentin (2008): “Introducing Financial Frictions and Un-
employment into a Small Open Economy Model,” Discussion Paper 214, Sveriges Riksbank (Central Bank
of Sweden).
Christoffel, K., G. Coenen, and A. Warne (2008): “The new area-wide model of the euro area -
a micro-founded open-economy model for forecasting and policy analysis,” Working Paper Series 944,
European Central Bank.
Coenen, G., G. Lombardo, F. Smets, and R. Straub (2008): “International transmission and monetary
policy cooperation,” Working Paper Series 858, European Central Bank.
Curdia, V., and M. Woodford (2008): “Credit Frictions and Optimal Monetary Policy,” Discussion
Papers 0809-02, Columbia University, Department of Economics.

20
Dedola, L., and G. Lombardo (2009): “Financial frictions, financial integration and the international
propagation of shocks,” mimeo, European Central Bank.
Demirel, U. D. (2009): “Optimal Monetary Policy in a Financially Fragile Economy,” The B.E. Journal
of Macroeconomics, 9(1).
Devereux, M. B., P. R. Lane, and J. Xu (2006): “Exchange Rates and Monetary Policy in Emerging
Market Economies,” Economic Journal, 116(511), 478–506.

Duarte, M., and M. Obstfeld (2008): “Monetary policy in the open economy revisited: The case for
exchange-rate flexibility restored,” Journal of International Money and Finance, 27(6), 949–957.
Edge, R. M. (2003): “A utility-based welfare criterion in a model with endogenous capital accumulation,”
Finance and Economics Discussion Series 2003-66, Board of Governors of the Federal Reserve System
(U.S.).
Elekdag, S., and I. Tchakarov (2007): “Balance sheets, exchange rate policy, and welfare,” Journal of
Economic Dynamics and Control, 31(12), 3986–4015.
Erceg, C. J., D. W. Henderson, and A. T. Levin (2000): “Optimal monetary policy with staggered
wage and price contracts,” Journal of Monetary Economics, 46(2), 281–313.
Faia, E. (2008): “Financial Frictions and The Choice of Exchange Rate Regimes,” mimeo, Goethe University
Frankfurt.
Faia, E., and T. Monacelli (2007): “Optimal interest rate rules, asset prices, and credit frictions,” Journal
of Economic Dynamics and Control, 31(10), 3228–3254.

Fiore, F. D., and O. Tristani (2009): “Optimal monetary policy in a model of the credit channel,”
Working Paper Series 1043, European Central Bank.
Gertler, M., S. Gilchrist, and F. M. Natalucci (2007): “External Constraints on Monetary Policy
and the Financial Accelerator,” Journal of Money, Credit and Banking, 39(2-3), 295–330.

Gilchrist, S., V. Yankov, and E. Zakrajsek (2009): “Credit market shocks and economic fluctuations:
Evidence from corporate bond and stock markets,” Journal of Monetary Economics, 56(4), 471–493.
Kiyotaki, N., and J. Moore (1997): “Credit Cycles,” Journal of Political Economy, 105(2), 211–48.

Levin, A. T., A. Onatski, J. C. Williams, and N. Williams (2005): “Monetary Policy Under Uncer-
tainty in Micro-Founded Macroeconometric Models,” NBER Working Papers 11523, National Bureau of
Economic Research, Inc.
Lombardo, G., and F. Ravenna (2009): “Trade and Optimal Monetary Policy,” mimeo.

Ravenna, F., and C. E. Walsh (2006): “Optimal monetary policy with the cost channel,” Journal of
Monetary Economics, 53(2), 199–216.

Rotemberg, J. J., and M. Woodford (1998): “An Optimization-Based Econometric Framework for the
Evaluation of Monetary Policy: Expanded Version,” NBER Technical Working Papers 0233, National
Bureau of Economic Research, Inc.
Schmitt-Grohe, S., and M. Uribe (2006): “Optimal Simple and Implementable Monetary and Fiscal
Rules: Expanded Version,” NBER Working Papers 12402, National Bureau of Economic Research, Inc.
Woodford, M. (2003): Interest and prices. Princeton University Press.

21
Tables and figures

Table 1. Structural parameters

Parameter Value Description
Households
β 0.994 discount rate
σ 2.0 inverse of intertemporal elasticity of substitution
κ 160 weight on disutility of labour
ϕ 2.0 inverse of Frisch elasticity of labour supply
φW 4.33 elasticity of substitution between labour varieties
θW 0.75 Calvo probability for wages
γc 0.3 share of distributed tradables in consumption
α 0.6 home bias (consumption and investment goods)
Capital production and financial frictions
τ 0.025 depreciation rate
ςI 5.2 investment adjustment costs
γi 0.6 share of tradables in investment
µ 0.1 monitoring costs
ν 0.977 survival rate for entrepreneurs
σE 0.27 steady-state standard deviation of idiosyncratic productivity
Intermediate goods firms
ηN 0.38 capital share in nontradable production
ηH 0.38 capital share in tradable production
φN 3.50 elasticity of substitution between intermediate nontradable varieties
φH 5.76 elasticity of substitution between intermediate tradable varieties
θN 0.9 Calvo probability for nontradables
θH 0.75 Calvo probability for tradables
Monetary authority
ρ 0.85 interest rate smoothing
φπ 1.90 long-run response to inflation
φ∆y 0.15 response to output growth
φ∆π 0.19 response to change in inflation

22
Table 2. Stochastic processes

Parameter Value Description
Autoregressive coefficients
ρt 0.85 productivity shock in tradable sector
ρn 0.85 productivity shock in nontradable sector
ρd 0.85 consumption preference shock
ρg 0.96 government spending shock
ρi 0.75 investment-specific technology shock
ρν 0.5 financial wealth shock
ρe 0.75 riskiness shock
Standard deviations
σt 0.031 productivity shock in tradable sector
σn 0.022 productivity shock in nontradable sector
σd 0.006 consumption preference shock
σg 0.0045 government spending shock
σi 0.018 investment-specific technology shock
σν 0.01 financial wealth shock
σe 0.06 riskiness shock
σm 0.001 monetary policy shock

Table 3. Steady-state ratios

Variable Value
Consumption share in GDP 58.5
Government expenditures share in GDP 21.0
Investment share in GDP 20.5
Exports share in GDP 12.0
Net exports share in GDP 0.0
Net worth share in capital 50.0
External financing premium (RE − R, annualized) 1.64
External financing premium (RB − R, annualized) 0.50
Bankruptcy rate (per quarter) 0.73
Bankruptcy costs share in output 0.27
Share of transfers to entrepreneurs in output 4.1

23
Table 4. Moment matching for the euro area

Variable model data
Standard deviations
GDP 0.48 0.48
Consumption 0.48 0.48
Investment 1.31 1.31
Government spending 1.61 1.60
Inflation 0.29 0.36
Short-term interest rate 1.04 2.81
Entrepreneurs’ debt 1.16 1.53
External financing premium 1.37 0.43
Autocorrelations
GDP 0.28 0.24
Consumption 0.06 0.06
Investment 0.73 0.16
Government spending 0.96 0.96
Inflation 0.65 0.70
Short-term interest rate 0.94 0.98
Entrepreneurs’ debt 0.52 0.18
External financing premium 0.86 0.81
Correlations with GDP
Consumption 0.70 0.65
Investment 0.41 0.80
Government spending 0.01 -0.21
Inflation -0.34 -0.04
Short-term interest rate 0.01 -0.04
Entrepreneurs’ debt 0.12 0.26
External financing premium -0.05 -0.22
Other correlations
External premium-investment -0.19 -0.12
Notes: GDP components and entrepreneurs’ debt are de-
fined in log differences.

Table 5. Variance decomposition

Shock GDP Cons. Invest. Inflation Interest Entrepr. Ex. fin.
rate debt premium
Productivity (T) 28.0 13.9 13.7 40.0 12.3 1.2 3.0
Productivity (NT) 28.6 30.1 4.0 38.0 32.4 3.0 4.8
Preference 7.1 35.5 0.0 0.6 2.5 0.0 0.0
Gov. spending 1.0 1.1 0.0 0.4 1.1 0.0 0.0
Inv. specific 8.5 0.7 37.3 1.7 16.3 3.9 2.7
Monetary 18.2 15.3 7.8 14.6 5.7 0.6 1.4
Entrepr. survival 2.0 3.3 32.4 4.5 29.0 90.9 39.7
Entrepr. riskiness 6.7 0.1 4.7 0.1 0.9 0.4 48.4
Notes: GDP components and entrepreneurs’ debt are defined in log differences.

24
Table 6. Welfare costs of alternative regimes: simple model

PPI targ. CPI targ. Mon. union Taylor rules
No financial frictions
All shocks 0.000 0.076 0.077 0.197
Productivity shocks 0.000 0.076 0.077 0.197
Financial frictions
All shocks 0.045 0.093 0.063 0.172
Productivity shocks 0.040 0.088 0.062 0.166
Notes: All numbers are relative to the cooperative equilibrium. Welfare losses are
expressed in percent of steady-state consumption. The normalization factors (vari-
ance of output relative to the fully-fledged version of the model) are 6.7 (no financial
frictions) and 7.1 (financial frictions).

Table 7. Welfare costs of PPI targeting: various frictions

Welfare losses
Baseline 0.0000
Home bias 0.0000
Consumption habits 0.0005
Nontradable goods 0.0036
Government 0.0001
Financial frictions 0.0453
Notes: All numbers are relative to the cooperative equilibrium. Welfare
losses are expressed in percent of steady-state consumption. Baseline refers
to the simple New Keynesian model. Other rows show the consequences of
augmenting the baseline with various frictions (one at a time). The nor-
malization factors (variance of output relative to the fully-fledged version
of the model) are: 6.7 (baseline), 7.8 (home bias), 5.4 (consumption habits,
with persistence 0.57), 1.5 (nontradable goods), 4.0 (government spending)
and 7.1 (financial frictions).

Table 8. Welfare costs: the role of debt denomination

PPI targ. CPI targ. Mon. union Taylor rules
Domestic currency debt denomination
All shocks 0.045 0.093 0.063 0.172
Productivity (H) 0.022 0.043 0.030 0.084
Productivity (F) 0.017 0.045 0.032 0.082
Foreign currency debt denomination
All shocks 0.056 0.100 0.069 0.179
Productivity (H) 0.000 0.054 0.040 0.093
Productivity (F) 0.052 0.042 0.029 0.083
Notes: All numbers are relative to the cooperative equilibrium. Welfare losses
are expressed in percent of steady-state consumption. The normalization factors
(variance of output relative to the fully-fledged version of the model) are 7.1
(domestic currency denomination) and 7.0 (foreign currency denomination).

25
Table 9. Welfare costs: the role of nontradables

PPI CPI Mon. Taylor ntPPI
targ. targ. union rules targ.
No financial frictions
All shocks 0.004 0.074 0.132 0.269 0.047
Trad. productivity (H) 0.005 0.013 0.019 0.074 0.008
Nontrad. productivity (H) 0.003 0.019 0.043 0.083 0.003
Trad. productivity (F) -0.002 0.036 0.044 0.087 0.024
Nontrad. productivity (F) -0.001 0.006 0.024 0.020 0.012
Domestic currency debt denomination
All shocks 0.091 0.127 0.139 0.293 0.119
Trad. productivity (H) 0.047 0.020 0.017 0.073 0.008
Nontrad. productivity (H) 0.007 0.017 0.045 0.085 0.040
Trad. productivity (F) 0.005 0.044 0.036 0.072 0.015
Nontrad. productivity (F) 0.004 0.017 0.029 0.022 0.027
Foreign currency debt denomination
All shocks 0.111 0.126 0.129 0.279 0.145
Trad. productivity (H) 0.004 0.023 0.021 0.070 0.009
Nontrad. productivity (H) 0.005 0.020 0.048 0.091 0.002
Trad. productivity (F) 0.051 0.045 0.034 0.073 0.024
Nontrad. productivity (F) 0.033 0.021 0.023 0.019 0.091
Notes: All numbers are relative to the cooperative equilibrium. Welfare losses
are expressed in percent of steady-state consumption. The normalization factors
(variance of output relative to the fully-fledged version of the model) are: 1.5
(no financial frictions), 1.6 (domestic currency denomination) and 1.7 (foreign
currency denomination).

26
Table 10. Welfare costs: the role of sticky wages

PPI CPI Mon. Taylor ntPPI Wage
targ. targ. union rules targ. targ.
No financial frictions
All shocks 1.171 0.646 0.133 0.264 1.106 0.023
Trad. productivity (H) 0.782 0.225 0.020 0.085 0.001 0.005
Nontrad. productivity (H) 0.063 0.044 0.043 0.080 0.768 0.008
Trad. productivity (F) 0.303 0.313 0.028 0.048 0.003 0.007
Nontrad. productivity (F) 0.029 0.066 0.039 0.031 0.338 0.003
Domestic currency debt denomination
All shocks 2.551 1.633 0.138 0.311 2.217 0.065
Trad. productivity (H) 1.652 0.630 0.016 0.069 0.023 0.005
Nontrad. productivity (H) 0.163 0.099 0.044 0.078 1.512 0.008
Trad. productivity (F) 0.661 0.735 0.022 0.040 0.011 0.005
Nontrad. productivity (F) 0.075 0.147 0.039 0.029 0.656 0.003
Foreign currency debt denomination
All shocks 2.699 1.560 0.114 0.279 2.409 0.091
Trad. productivity (H) 0.693 0.606 0.019 0.070 0.001 0.007
Nontrad. productivity (H) 0.055 0.081 0.041 0.081 0.675 0.013
Trad. productivity (F) 1.715 0.713 0.019 0.043 0.052 0.017
Nontrad. productivity (F) 0.260 0.153 0.030 0.025 1.689 0.027
Notes: All numbers are relative to the cooperative equilibrium. Welfare losses are
expressed in percent of steady-state consumption. The normalization factors (variance of
output relative to the fully-fledged version of the model) are: 1.0 (no financial frictions),
1.1 (domestic currency denomination) and 1.1 (foreign currency denomination).

27
Figure 1. Home productivity shock - simple model

Output (H) Ext. financing premium (H) Real interest rate (H)
3 0 1

−0.2 0.5
2
−0.4 0
1
−0.6 −0.5

0 −0.8 −1
0 20 40 0 20 40 0 20 40

Output (F) Ext. financing premium (F) Real interest rate (F)
0.4 0.1 1

0.2 0.05 0.5

0 0 0

−0.2 −0.05 −0.5

−0.4 −0.1 −1
0 20 40 0 20 40 0 20 40

PPI inflation (H) CPI inflation (H) Terms of trade
0.4 2 4

0.2 3
1
0 2
0
−0.2 1

−0.4 −1 0
0 20 40 0 20 40 0 20 40

PPI inflation (F) CPI inflation (F) Exchange rate depreciation
0.06 1 4

0.04
0 2
0.02
−1 0
0

−0.02 −2 −2
0 20 40 0 20 40 0 20 40

Optimal PPI targeting

Note: The financing premium, inflation and the interest rate are expressed as percentage point
deviations from their steady-state levels. All remaining variables are reported as percentage devi-
ations.

28
Figure 2. Negative home net worth shock - simple model

Output (H) Ext. financing premium (H) Real interest rate (H)
0.2 0.4 0.1

0.1 0.3 0

0 0.2 −0.1

−0.1 0.1 −0.2

−0.2 0 −0.3
0 20 40 0 20 40 0 20 40

PPI inflation (H) CPI inflation (H) Exchange rate depreciation
0.15 0.2 0.4

0.1 0.1 0.2

0.05 0 0

0 −0.1 −0.2

−0.05 −0.2 −0.4
0 20 40 0 20 40 0 20 40
Optimal PPI targeting Union

Note: The financing premium, inflation and the interest rate are expressed as percentage point
deviations from their steady-state levels. All remaining variables are reported as percentage devi-
ations.

Figure 3. Home productivity shock - dollarized debt

Output (H) Ext. financing premium (H) Real interest rate (H)
3 0.1 1

0.05 0.5
2
0 0
1
−0.05 −0.5

0 −0.1 −1
0 20 40 0 20 40 0 20 40

PPI inflation (H) CPI inflation (H) Exchange rate depreciation
0.5 2 4

0 1 2

−0.5 0 0

−1 −1 −2
0 20 40 0 20 40 0 20 40
Optimal PPI targeting Union

Note: The financing premium, inflation and the interest rate are expressed as percentage point
deviations from their steady-state levels. All remaining variables are reported as percentage devi-
ations.

29
Figure 4. Foreign productivity shock - dollarized debt

Output (H) Ext. financing premium (H) Real interest rate (H)
0.5 0 2

−0.2
0 1
−0.4
−0.5 0
−0.6

−1 −0.8 −1
0 20 40 0 20 40 0 20 40

PPI inflation (H) CPI inflation (H) Exchange rate depreciation
0.6 1 2

0.4
0 0
0.2
−1 −2
0

−0.2 −2 −4
0 20 40 0 20 40 0 20 40
Optimal PPI targeting Union

Note: The financing premium, inflation and the interest rate are expressed as percentage point
deviations from their steady-state levels. All remaining variables are reported as percentage devi-
ations.

Figure 5. Home tradable sector productivity shock

Output (H) Ext. financing premium (H) Real interest rate (H) Real exchange rate
1.5 0.1 1 1.5

0 0.5 1
1
−0.1 0 0.5
0.5
−0.2 −0.5 0

0 −0.3 −1 −0.5
0 20 40 0 20 40 0 20 40 0 20 40

Trad. PPI inflation (H) Nontrad. PPI inflation (H) CPI inflation (H) Exchange rate depreciation
0.5 0.15 0.6 2

0.1 0.4
0 1
0.05 0.2
−0.5 0
0 0

−1 −0.05 −0.2 −1
0 20 40 0 20 40 0 20 40 0 20 40
Optimal PPI targeting Union ntPPI targeting

Note: The financing premium, inflation and the interest rate are expressed as percentage point deviations from their
steady-state levels. All remaining variables are reported as percentage deviations.

30
Figure 6. Home tradable sector productivity shock - model with sticky wages

Output (H) Ext. financing premium (H) Real interest rate (H) Wage inflation (H)
6 0.5 1 0.6

0 0 0.4
4
−0.5 −1 0.2
2
−1 −2 0

0 −1.5 −3 −0.2
0 20 40 0 20 40 0 20 40 0 20 40

Trad. PPI inflation (H) Nontrad. PPI inflation (H) CPI inflation (H) Exchange rate depreciation
0.2 0.15 1 6

0 0.1 4
0.5
−0.2 0.05 2
0
−0.4 0 0

−0.6 −0.05 −0.5 −2
0 20 40 0 20 40 0 20 40 0 20 40
Optimal PPI targ. Union ntPPI targ. CPI targ. Wage targ.

Note: The financing premium, inflation and the interest rate are expressed as percentage point deviations from their
steady-state levels. All remaining variables are reported as percentage deviations.

Figure 7. Home nontradable sector productivity shock - model with sticky wages

Output (H) Ext. financing premium (H) Real interest rate (H) Wage inflation (H)
6 0.5 1 0.6

0 0 0.4
4
−0.5 −1 0.2
2
−1 −2 0

0 −1.5 −3 −0.2
0 20 40 0 20 40 0 20 40 0 20 40

Trad. PPI inflation (H) Nontrad. PPI inflation (H) CPI inflation (H) Exchange rate depreciation
0.6 0.05 1 6

0.4 0 4
0.5
0.2 −0.05 2
0
0 −0.1 0

−0.2 −0.15 −0.5 −2
0 20 40 0 20 40 0 20 40 0 20 40
Optimal PPI targ. Union ntPPI targ. CPI targ. Wage targ.

Note: The financing premium, inflation and the interest rate are expressed as percentage point deviations from their
steady-state levels. All remaining variables are reported as percentage deviations.

31
Figure 8. Foreign tradable sector productivity shock - model with sticky wages

Output (H) Ext. financing premium (H) Real interest rate (H) Wage inflation (H)
2 0.5 1 0.3

0 0.2
1 0
−1 0.1
0 −0.5
−2 0

−1 −1 −3 −0.1
0 20 40 0 20 40 0 20 40 0 20 40

Trad. PPI inflation (H) Nontrad. PPI inflation (H) CPI inflation (H) Exchange rate depreciation
0.3 0.06 0.5 2

0.2 0.04 0
0
0.1 0.02 −2
−0.5
0 0 −4

−0.1 −0.02 −1 −6
0 20 40 0 20 40 0 20 40 0 20 40
Optimal PPI targ. Union ntPPI targ. CPI targ. Wage targ.

Note: The financing premium, inflation and the interest rate are expressed as percentage point deviations from their
steady-state levels. All remaining variables are reported as percentage deviations.

Figure 9. Foreign nontradable sector productivity shock - model with sticky wages

Output (H) Ext. financing premium (H) Real interest rate (H) Wage inflation (H)
1 0.1 0.5 0.15

0 0 0.1
0.5
−0.1 −0.5 0.05
0
−0.2 −1 0

−0.5 −0.3 −1.5 −0.05
0 20 40 0 20 40 0 20 40 0 20 40

Trad. PPI inflation (H) Nontrad. PPI inflation (H) CPI inflation (H) Exchange rate depreciation
0.15 0.03 0.5 2

0.1 0.02 0
0
0.05 0.01 −2
−0.5
0 0 −4

−0.05 −0.01 −1 −6
0 20 40 0 20 40 0 20 40 0 20 40
Optimal PPI targ. Union ntPPI targ. CPI targ. Wage targ.

Note: The financing premium, inflation and the interest rate are expressed as percentage point deviations from their
steady-state levels. All remaining variables are reported as percentage deviations.

32
Figure 10. Negative home net worth shock - model with sticky wages

Output (H) Ext. financing premium (H) Real interest rate (H) Wage inflation (H)
0.6 0.4 0.2 0.02

0.4 0.3 0
0.01
0.2 0.2 −0.2
0
0 0.1 −0.4

−0.2 0 −0.6 −0.01
0 20 40 0 20 40 0 20 40 0 20 40

Trad. PPI inflation (H) Nontrad. PPI inflation (H) CPI inflation (H) Exchange rate depreciation
0.02 0.02 0.1 0.5

0.05
0.01 0.01
0 0
0 0
−0.05

−0.01 −0.01 −0.1 −0.5
0 20 40 0 20 40 0 20 40 0 20 40
Optimal PPI targ. Union ntPPI targ. CPI targ. Wage targ.

Note: The financing premium, inflation and the interest rate are expressed as percentage point deviations from their
steady-state levels. All remaining variables are reported as percentage deviations.

33
Appendix
A.1 Foreign currency denomination of entrepreneurs’ debt
If loans taken by entrepreneurs are denominated in the foreign currency, the amount borrowed in the domestic
currency can be written as BE,t+1 ERt . The principal due, however, is equal to BE,t+1 ERt+1 , so that
entrepreneurs are exposed to exchange rate risk. The zero profit condition (39) thus becomes:

RE,t+1 QT,t PC,t Kt+1 [ãE,t+1 (1 − F1,t+1 ) + (1 − µ) F2,t+1 ] = Rt∗ BE,t+1 ERt+1 (A.1)
Similarly, the first order condition defining the optimal debt contract (43) is now given by:

 RE,t+1


 ERt+1 [1 − ãE,t+1 (1 − F1,t+1 ) − F2,t+1 ] + 

Rt∗
Et
ERt µ ¶ =0 (A.2)

 + 1−F1,t+11−F 1,t+1 RE,t+1
[ãE,t+1 (1 − F1,t+1 ) + (1 − µ) F2,t+1 ] − 1 

−µãE,t+1 F 0 1,t+1 ∗ ERt+1
Rt ERt

Finally, the law of motion for aggregate net worth (46) can be rewritten as:
" #
³ RE,t QT,t−1 PC,t−1 Kt − ´
Nt+1 = εn,t υ ∗ ERt µF2,t RE,t QT ,t−1 PC,t−1 Kt + TE,t (A.3)
− Rt−1 ERt−1 + BE,t ERt−1 BE,t ERt−1

A.2 Probability distributions
In this section we show the analytical formulas for functions of entrepreneurs’ idiosyncratic productivity
distribution. Detailed derivations can be found in Christiano et al. (2008).
If aE has a log normal distribution F with mean equal to 1, then log aE has a normal distribution with
σ2 2
mean equal to − 2E , where σE is the variance of log aE . This observation leads to the following formulas,
which we use in the derivations presented in section 2:
Z ãE,t µ 2 ¶
log ãE,t + 12 σE
F1,t = dF (aE ) = cdf (A.4)
0 σE
Z ãE,t µ ¶
log ãE,t + 12 σE2
F2,t = aE dF (aE ) = cdf − σE (A.5)
0 σE
µ 2 ¶
0 1 log ãE,t + 21 σE
F1,t = pdf (A.6)
ãE,t σE σE
where pdf (cdf ) is probability density function (cumulated distribution function) of a standard normal
distribution.

34