Leyes de Botaderos

Download as pdf or txt
Download as pdf or txt
You are on page 1of 39

The euro area periphery and the rest-of-the-world: a DSGE

approach.
Leonor Coutinho

University of Cyprus and Europrism


March 2013 (Draft)
Abstract
This research builds upon recent work undertaken for other small open economies to
construct a DSGE model for an euro area peripheral economy. The model takes into account
of euro area membership and departs from the literature by considering two dierent sectors:
a goods and other services sector and a tourism sector. The tourism sector focuses on
exports, is more labour intensive, and has a signicant share of its market located outside
the euro zone. Some parameters of the model are calibrated and others estimated using
Bayesian techniques to match the case of Cyprus, but can also be applied to other countries.
The model properties are validated through statistical tests, impulse-response analysis, and
variance decompositions.
Keywords: DSGE models, small open economy, bayesian methods
JEL Classication: E30, F41, C11

Corresponding author: University of Cyprus and Europrism Research Centre, Oce, B21, Thessalonikis 5,
Nicosia 2113, CYPRUS, [email protected], www.europrism.eu. This research is funded under a con-
sortium with the University of Cyprus and co-funded by the Cyprus Research and Promotion Foundation, grant
agreement ANTROPISTIKES/OIKON/0609(BIE)10. I acknowledge the advice and encouragement of Professor
Fabio Canova, and also the excellent contribution of Charalambos Karagiannakis with the data collection and
the statistical treatment of the data. The views expressed are mine solely along with any remaining errors or
omissions.
1 Introduction
DSGE models have become the workhorse of modern policy analysis in recent years. They
combine the rigorous optimization analysis and rational expectations of real business cycle
models with nominal rigidities, giving a non-trivial role to monetary policy. Due to the fact
that they are constructed from optimizing conditions for all agents in the economy, their reduced
form can be linked to the structural parameters of the model, making them less vulnerable to
the Lucas critique, and therefore more suitable for analyzing the relative importance of shocks,
simulating the eects of dierent policies, and undertaking forecasting exercises.
Following the seminal work of Smets and Wouters (2003), who developed a new Keynesian
dynamic stochastic general equilibrium (DSGE) model for the euro area, similar work has been
undertaken for several European countries, including Austria, Belgium, Luxembourg, Sweden,
and Spain, with promising results (see Breuss and Rabitsh, 2009; Jenls and Burggrave, 2007;
Deak et al., 2011; Adolfson et al., 2008; and Burriel et al., 2010).
Among country models, two groups can be identied: (i) small open economy models, which
treat the country as a small-open economy taking developments in the rest of the euro area as
exogenous; and (ii) two-country models which model the countrys economy alongside the euro
area. Two-country models have been used even in the analysis of small countries like Austria
(Breuss and Rabitsh, 2009), the reason being that such countries are part of the euro area and
want to be able to make their own projections to contribute to the euro area monetary policy
debate.
The broad set up of a DSGE models tends to consider four main blocks: households, rms,
scal policy, and monetary policy. In addition, a range of shocks are also scattered in the
model to confer it its stochastic nature and enable to address specic policy questions. Labour
and product markets are often assumed to operate under monopolistic competition, resulting in
price and wage markups often assumed to be subject to shocks. Smets and Wouters (2003) show
that markup shocks when included explain a signicant percentage of the variation in ination.
In the household block, the shocks typically considered include a consumption preference shock,
and a labor supply shock (shock to the preferences for leisure). But so far small open economy
models for euro area countries have largely ignored the rest-of-the-world and shocks emanating
from it.
This project aims at developing a DSGE model in the style of new-Keynesian/New Open
Economy Macroeconomics for a peripheral small euro area economy exposed to euro exchange
rate shocks, and other shocks emanating from outside the euro area. The research builds
upon other small open economy models, in particular Cuche-Curti et al. (2008), Adolfson et
al (2007), and Cristofell at al. (2008); extended with a tourism sector, which exports to the
rest-of-the world and is also more labour intensive. This sector will be exposed to a range of
shocks originated in the rest of the world, and will also be exposed to euro nominal exchange
rate uctuations caused by euro area monetary policy. We show that this sector can be an
important channel for the transmission of shocks, a results which is in line with the ndings
1
of Canova and Dallari (2013), who show using data for a set of Mediterranean economies that
tourism can be an important channel for the transmission of business cycle uctiations across
countries, even though it has been largely disregarded in policy analysis.
The model is partially calibrated and partially estimated to match the behavior of the
Cypriot economy, and euro area peryphery with a signicant share of exports to non-Eurozone
countries. The parameter estimation uses Bayesian techniques, which are found to provide a
robust and ecient way of matching the models to actual data. The model is intended for
undertaking simulations of dierent policies and exogenous shocks, and can be applied to other
periphery countries.
Bayesian estimation basically boils down to treating the parameters of the model as random
variables, allowing the adoption of a prior probability distribution for each parameter which
summarizes any available information (see Carlin and Louis, 2009). Apart from simplifying the
use of priors in estimation, and therefore allowing for prior information to be used in models
with a relatively large number of parameters, Bayesian estimation also has the advantage that
the posterior inference does not depend on the model being correctly specied.
1
Estimating
a set of model parameters using Baeysian techniques also allows us to take into account of
parameter uncertainty when analysing, for instance, impulse response functions.
Overall we nd that the model has economically plausible properties, especially with regard
to the propagation of key economic shocks, and that the euro exchange rate can be an important
channel for the propagation of shocks in this type of economies.
2 The model
2.1 Households
Households are assumed to maximize utility over an innite life horizon. Their utility is as-
sumed to depend positively on consumption and leisure, with habit formation in consumption.
In terms of their budget constraints, we allow for households to be grouped into two categories:
saving and rule-of-thumb households. Rule-of- thumb households do not owe any assets and
just consume their current income ow, facing a period-by-period budget constraint. This as-
sumption, which can be justied by myopia or credit constrains, for instance, breaks down the
Ricardian equivalence, allowing a scal expansion to have a positive eect on private consump-
tion, as suggested by empirical evidence (see Gali et al., 2007). Saving households, on the other
hand can owe capital and assets, and receive dividends and transfers. These households decide
how much to save and invest so as to maximize their lifetime consumption and leisure. The
share of rule of thumb consumers in the economy is :
1
, while the share of optimizing consumers
is (1 :
1
).
1
Despite these advantages and the increasing popularity of Bayesian estimation of DSGE models, there are
still challenges in this area, relating in particular to the identication of structural parameters (see Canova,
2007).
2
2.1.1 Optimizing Consumers
Optimizing consumers on the other hand maximize the present value of their expected life-time
utility subject to an intertemporal budget constraint.
The present value of the expected life-time utility of optimizing household , is given by:
l
C,t
1
t
1

t=t
,
tt
_
-
C,t
ln
_
C
O
t
(,) H
O
t
_
-
|,t
|
O
t
(,)
1+i
1 i
_
(1)
where C
O
is the real consumption of optimizing households, and |
O
is the labour supplied
by optimizing households, -
C
and are -
|
are preference shocks to consumption and labour
respectively, and , is the time discount factor such that 0 < , < 1. H
O
t
(,) is the stock of
external habits given by H
O
t
= /C
O
t1
, where the parameter h (0 < / < 1) determines the
importance of external habit formation.
2
The presence of habit formation in consumption
means that an increase in consumption today will increase the marginal utility of consumption
in the following period (intuitively, as described in Schmitt-Grohe and Uribe, 2008, the more
the consumer eats today, the hungrier he wakes up tomorrow). As / tends to unity, households
act to smooth changes in consumption rather than the level of consumption. This assumption
improves in general the empirical t of the model, as it can generate the hump-shaped impulse
responses of output and consumption to demand and supply shocks, typically observed in VAR
models, since they provide an additional incentive for consumption smoothness on top of that
stemming from the concavity of the utility function (see Bergin and Tchakarov, 2003, and
references therein).
Optimizing households are assumed to supply dierentiated units of labour to labour unions.
Each union is a monopolistic supplier of a type labour and has some market power over wages
subject to Calvo-type restrictions (see Calvo, 1983), which imply that the optimal wage in
period t, \

t
(,) can only be negotiated with a probability (1
&
). Each period, wages that
are not reset are updated by a wage indexation rate \
Wt,tt
0
, where t
0
is the last period when
wages could be re-set. Hence \
t
= \
Wt,tt
0
\
t
0
with \
Wt,tt
0
= 1 for t = t
0
.
3
Households are
insured against labour income uncertainty through state-contingent securities, which pay a net
cash inow of 1
t

O
t
(,), such that (1t
|
t
)\
t
(,)|
O
t
(,)1
t

O
t
(,) is equal for households (where t
|
is the tax rate on labour), making the optimizing households maximization problem symmetric.
State contingent securities sum up to zero across optimizing households, that is
S
C
_
)=0

O
t
(,)d, = 0.
2
Notice that for the model to be consistent with a balanced-growth path it is necessary to assume either
non-separable preferences in consumption and leisure or logarithmic preferences over consumption (see Ireland,
2004).
3
Although the empirical relevance of wage rigidity has been questioned in the literature (since there is scope
for the stickiness in formal wages to be privately neutralized), a number of contributions have found that wage
rigidity helps DSGE models explain many features of the data, such as the persistence in the eects of monetary
shocks (see Christiano et al., 2005, as well as Canzoneri et al., 2007, and references therein).
3
Optimizing households can also invest in capital stock and rent it in the subsequent period
to domestic rms at the nominal rental price 7
1
t
. Investment is subject to adjustment costs,
hence the standard capital accumulation law of motion is modied as follows:
1
O
t
(,) = -
1,t
_
1 I
1
_
1
O
t
(,)
1
O
t1
(,)
__
1
O
t
(,) (1 c) 1
O
t1
(,) (2)
I
1


1
2
_
1
O
t
(,)
1
O
t1
(,)

_
2
(3)
where 1
O
t
(,) is the investment of household , in period t, -
1,t
is an investment shock, c is the
depreciation rate,
1
0 determines the costs of adjusting investment, and is the steady-state
growth rate of the economy. The inclusion of investment adjustment costs allows to map the
fall in real interest rates (liquidity eect) and the hump-shaped response of investment to a
monetary expansion, typically found in empirical VAR studies (see Christiano et al., 2010).
Investment adjustment costs are also said to generate other desirable dynamics to the models,
including more desirable responses of output, hours worked, and real wages to scal shocks (see
Burnside et al., 2004).
4
Households can also invest in both domestic government bonds and Eurozone (excluding the
domestic economy) bonds. It is assumed that households cannot borrow from the government,
hence the stock of domestic government bonds is non-negative. The stock of Eurozone bonds
can be positive (with the domestic economy being a net lender) or negative (with the domestic
economy being a net debtor). Domestic bonds yield the domestic nominal interest rate 1, while
Eurozone bonds yield the foreign interest rate 1
1Z
, adjusted by a risk premium O
_
/
1Z
t
, -
,t
_
=
-
,t
oxp
_
c

_
/
1Z
t
/
1Z
__
, where /
1Z
t
are the economys total real stationary holdings of
Eurozone bonds, such that /
1Z
t
=
1
1Z
t
I
t
1
t
, /
1Z
is the steady-state equilibrium real stationary
holdings of Eurozone bonds, c

is a parameter measuring the degree of capital mobility, and


-
,t
is a risk premium shock. This risk premium implies that the household has to pay a
remuneration higher than 1
1Z
when the economys net debt position is above its steady state
level or -
,t
is positive. This assumption is crucial to pin-down a well-dened steady-state
for consumption and assets (see Schmitt-Grohe and Uribe, 2003, for a discussion of the non-
stationarity problem). In steady-state -
,t
= 1 and O
_
/
1Z
, 1
_
= 1, hence 1 = 1
1Z
. Finally,
optimizing households also receive dividends from rms. It is assumed that all rms prots
are distributed equally among households in the form of dividends.
Given households resources and their uses described above, the optimizing household ,
intertemporal budget constraint can be written as:
4
Empirical evidence on the magnitude of investment adjustment costs is limited, with aggregate studies
typically nding costs more signicant than studies using disaggregate sectoral data (see Groth and Khan,
2010).
4
(1 t
C
t
)1
C,t
C
O
t
(,) 1
1,t
1
O
t
(,) 1
O
t
(,) 1
1Z,O
t
(,) (4)
1
t1
1
O
t1
(,) 1
1Z
t1
O
_
/
1Z
t1
, -
,t1
_
1
1Z,O
t1
(,) (1 t
|
t
)\
t
(,)|
O
t
(,) 1
t

O
t
(5)
(1 t
1
t
)7
1
t
1
O
t1
(,) 11\
O
t
T1
O
t
(6)
where t
C
, t
|
, and t
1
, are the economy-wide tax rates on consumption, labour, and capital,
and 1
C,t
and 1
1,t
are the price deators for consumption and investment, respectively, and
T1
O
t
(,) are net lum-sum transfers paid to optimizing households. 1
C,t
is assumed to be the
numeraire. In this set up, the no-Ponzi game condition for debt accumulation is given by:
lim
c!1
1
t
_
j
t+c
_
1
t+c
1
1Z
t+c
__
= 0 (7)
where j
t+c
=
c

I=0
1
1
t+I1
for s0, and j
t+c
= 1, for : = 0.
The optimizing household j maximizes its welfare (1) with respect to consumption (C
O
t
),
domestic and Eurozone bond holdings (1
O
t
, 1
1Z,O
t
), investment (1
O
t
), and capital (1
O
t
), sub-
ject to the constraints imposed by (2) (4),and (7).Maximization with respect to consumption
requires:
l
C,t
-
C,t
_
C
O
t
(,) /C
O
t1
_
1
= `
O
t
(,)(1 t
C
t
)1
C,t
`
O
t
(,) =
-
C,t
_
C
O
t
(,) /C
O
t1
_
1
(1 t
C
t
)1
C,t
Maximization with respect to domestic bonds requires:
,1
t
_
`
O
t+1
(,)

1
t
= `
O
t
(,)
where l
C,t
is the life-time marginal utility of consumption in period t. Maximization with
respect to foreign bonds requires:
,1
t
_
`
O
t+1
(,)

1
1Z
t
O
_
/
1Z
t
, -
,t
_
= `
O
t
(,)
Maximization with respect to investment requires:
1
1,t
= 1
C,t
Q
t
(,)-
1,t
_
1 I
1
_
1
O
t
(,)
1
O
t1
(,)
_
I
10
_
1
O
t
(,)
1
O
t1
(,)
_
1
O
t
(,)
1
O
t1
(,)
_

,1
t
_
_
`
O
t+1
(,)
`
O
t
(,)
1
C,t+1
Q
t+1
-
1,t+1
I
10
_
1
O
t+1
(,)
1
O
t
(,)
__
1
O
t+1
(,)
1
O
t
(,)
_
2
_
_
where
I
10
_
1
O
t
(,)
1
O
t1
(,)
_
=
1
_
1
O
t
(,)
1
O
t1
(,)

_
I
10
_
1
O
t+1
(,)
1
O
t
(,)
_
=
1
_
1
O
t+1
(,)
1
O
t
(,)

_
5
and Q
t
is the "Tobins Q": Q
t


t
1
C,t
`
O
t
. Maximization with respect to capital (1
t
) requires:
1
C,t
Q
t
(,) = ,1
t
_
`
O
t+1
(,)
`
O
t
(,)
_
(1 t
1
t+1
)7
1
t+1
1
C,t+1
Q
t+1
(1 c)
_
_
2.1.2 Rule-of-Thumb or Liquidity Constrained Consumers
It is assumed that 1OT consumers have an utility function of a similar form as (1),but face a
period-by-period budget constraint of the form:
(1 t
C
t
)1
C,t
C
1
t
= (1 t
|
t
)\
t
|
1
t
T1
1
t
(8)
where \
t
is the average nominal wage rate negotiated by labour unions, |
1
is the amount of
labour supplied by rule of thumb consumers, and T
1
t
are net lum-sum tranfers received by
this type of households. These transfers are useful because with and appropriate distribution
of steady-state transfers between optimizing and rule of thumb consumers it is possible to
ensure that C
1
= C
O
= C in steady-state and this simplies the solution of the model. These
consumers will use all of their disposable income in consumption.
Total real consumption (C) is the weighted average of consumption of these two consumer
categories:
C
t
= :
1
C
1
t
(1 :
1
)C
O
t
(9)
|
t
= :
1
|
1
t
(1 :
1
)|
O
t
(10)
T1
t
= :
1
T1
1
t
(1 :
1
)T1
O
t
(11)
2.1.3 Wage-setting
Wages are set by a continuum of labour unions, each representing a specic type of labour
j. Rule-of-thumb and optimizing consumers are assumed to be uniformily distributed across
labour types. Hence, for each labour type j, its union chooses a wage that maximizes the
weighted average of the utility of optimizing and rule of thumb consumers subject to their
budget constraints and the labour demand function, which is equivalent to maximizing:
1
t
1

t=t
(,
&
)
tt
_
_
_
:
1
_
`
1
t
(1 t
|
t
)\
t
|
1
t
(,) -
1,t
|
T
I
())
1+
1+i
_

(1 :
1
)
_
`
O
t
(1 t
|
t
)\
t
|
O
t
(,) -
1,t
|
C
I
())
1+
1+i
_
_
_
_
subject to the labour demand function, which under monopolistic competition will take the
form:
|
t
(,) =
_
\
t
(,)
\
t
_

I
|
t
where c
|
is the elasticity of substitution between varieties of labour. Households take total
labour demand as given and there is perfect labour mobility across sectors, hence the same
6
wage prevails across sectors. It is also assumed that the union takes into account the fact that
rms allocate labor demand uniformly across dierent worker, independently of their household
type. It follows that, in the aggregate, we will have |
1
t
= |
O
t
= |
t
.
With probability (
&
)
tt
the wage \
Wt,tt
\(,)

t
will be in eect in period t, hence the
FOC for the unions maximization problem with respect to \

t
(,) is given by:
\

t
(,)
1+
I
i
1
t
1

t=t
(,
&
)
tt
_
`
t
(,)(1 t
|
t
)\
Wt,tt
(\
t
,\
Wt,tt
)

I
|
t
_
=
c
|
(c
|
1)
1
t
1

t=t
(,
&
)
tt
_
-
|,t
(\
t
,\
Wt,tt
)

I
(1+i)
|
(1+i)
t
_
Dening j
W


I
(
I
1)
and rearranging yields:
\

t
(,)
1+
I
i
= j
W
1
t

1
t=t
(,
&
)
tt
_
-
|,t
(\
t
,\
Wt,tt
)

I
(1+i)
|
(1+i)
t
_
1
t

1
t=t
(,
&
)
tt
_
`
t
(,)(1 t
|
t
)\
Wt,tt
(\
t
,\
Wt,tt
)

I
|
t
_
which will be equal for all optimizing households, hence \

t
(,) = \

t
, for all ,.
\
1+
I
i
t
= j
W
w
W
t
1
W
t
where
w
W
t
= ,
&
1
t
(1,\
Wt+1,1
)

I
(1+i)
w
W
t+1
-
|,t
(\
t
)

I
(1+i)
|
O(1+i)
t
1
W
t
= ,
&
1
t
\
Wt+1,1
(1,\
Wt+1,1
)

I
1
W
t+1
`
t
(1 t
|
t
) (\
t
)

I
|
O
t
The optimal wage results therefore in a markup over the expected marginal rate of substitution
between consumption and labour. We do not consider wage markup shocks because these
cannot be distinguished from the preference shock to the disutility of labour considered above
5
Notice that the aggregate wage in the economy can be written as:
\
t
=
_
(1
&
) \
1
I
t

&
(\
Wt,1
\
t1
)
1
I
_ 1
1
I
2.2 Firms
2.2.1 Retail Sector (Final Goods: Private Consumption, Public Consumption,
Investment)
In the retail sector rms produce three types of nal goods under perfect competition: a
private consumption good (C), a public consumption good (G), and an investment good (1).
For the production of each one of the three nal goods, the following inputs are used: domestic
5
It is only possible to distinguish between these two shocks if the labour market is modeled explicitly to
account for unemployment (see Gali et al., 2011).
7
intermediate goods (A
jo
), imported foreign intermediate goods (A
jn
) and energy (1). The
nal goods are produced using CES production function. Moreover, nal goods are non-tradable
(i.e. cannot be exported). The nal good destined for sector o, o = C, G, 1, is given by:
o
t
=
_
(1 .
1
)
1j
T
(o
j
t
)
j
T
.
1j
T
1
(1
S
t
)
j
T
_
1j
T
(12)
where,
o
j
t
=
_
(1 .
A,S
)
1j
L,S
(A
jo,S
t
)
j
L,S
.
1j
L,S
A,S
(A
jn,S
t
)
j
L,S
_
1j
L,S
(13)
The parameters j
1
and j
A,S
determine the elasticities of substitution between intermediate
goods and energy and between domestic goods and imported goods, which are given by
1
1j
T
and
1
1j
L,S
, respectively, with j
1
< 1 and j
A,S
< 1. The nested structure assumed in the aggrega-
tion allows for dierent elasticities of substitution between domestic and imported intermediate
goods and energy. This can be important for matching CPI dynamics (see Cuche-Curtis et al.,
2009), and allows to compute a model consistent measure of core ination.
The demand of retail rms for intermediate goods and energy is determined by the mini-
mization of the expenditure needed to produce a given amount of nal good:
min1
jo
t
A
jo,c
t
1
jn
t
A
jn,c
t
1
1
t
1
c
t
(14)
s.t. (12) and (13) for o = C, G, 1. This assumes that the price charged by intermediate goods
rms and the price of energy inputs is invariant to its nal use. Note that 1
S,t
= 'C
S,t
=
lagrange multiplier since the retailer is a competitive producer:
L =
_
1
j
t
A
jo,S
1
n
t
A
jn,S
1
1
t
1
S
_

1
S,t
_
o
t

_
(1 .
1
)
1j
T
(o
j
t
)
j
T
.
1j
T
1
(1
S
t
)
j
T
_
1j
T
_
FOCs with respect to A
jo,S
t
yield:
A
jo,S
t
=
_
1
j
t
1
S,t
_

1
1p
L,S
[(1 .
1
)o
t
[
1p
T
1p
L,S
(1 .
A,S
) (o
j
t
)
p
T
p
L,S
1p
L,S
and analogously for imports:
A
jn,S
t
=
_
1
n
t
1
S,t
_

1
1p
L,S
[(1 .
1
)o
t
[
1p
T
1p
L,S
.
A,S
(o
j
t
)
p
T
p
L,S
1p
L,S
The FOCs with respect to 1
S
t
yield:
1
S
t
=
_
1
1
t
1
S,t
_

1
1p
T
.
1
o
t
8
The price of the nal good produced for sector s, o = C, G, 1, can be obtained by substituting
the input demands into the bundleling function (12):
1
S,t
=
_

_
(1 .
1
)
_
(1 .
A,S
) (1
j
t
)
p
L,S
p
L,S
1
.
A,S
(1
n
t
)
p
L,S
p
L,S
1
_
(
1p
L,S)
p
T
(1p
T
)p
L,S
.
1
_
1
1
t
_
p
T
p
T
1
_

_
p
T
1
p
T
The bundle of intermediate domestic goods A
jo,S
t
is a budle of varieties produced by each
individual intermediate domestic goods rm ), such that:
A
jo,S
t
=
__
1
0
A
jo,S
t
())
0

d)
_
1
0

The parameter 0
jo
determine the elasticity of substitution between varieties of domestic in-
termediate goods, which is given by
1
10
u
, with 0
jo
< 1. Given this bundle, the demand for
individual varieties which minimizes expenditures for a given amount of domestic intermediate
good is given by:
A
jo,S
t
()) =
_
1
j
t
())
1
j
t
_

10

A
jo,S
t
with the corresponding price indexes being:
1
j
t
=
_
_
1
0
1
j
t
())
0

1
d)
_
0

1
0

2.2.2 Imports
It is assumed that all of domestic imports come from the Eurozone and that the domestic
demand for Eurozone imports does not aect the price of imports, hence 1
n
t
(expressed in
euros) is taken as given.
2.2.3 Intermediate Sectors
It is assumed that there are two intermediate sub-sectors in the economy. One sub-sector pro-
duces intermediates goods and services that can be either used domestically (A
jo
) or exported
(A
j)
) to the Eurozone. The other sub-sector produces intermediate tourism services (A
c
) that
can only exported, either to the Eurozone (A
c,1Z
) or to the rest of the world (A
c,1cW
). Both
types of goods are produced combining three inputs: Capital (1), Labour (|) and Oil (1), but
the technology can dier across sectors.
The total output of rm ) will be A
o
()) = A
jo
(i) A
j)
(i) A
c
(i).
The goods production function is given by:
A
j
t
()) =
j
t
(I
t
|
j
t
()))
c
1
(1
j
t
()))
c
1
(1
j
t
()))
1c
1
c
1
c/
j
I
t
(15)
9
where
j
t
is a good specic stationary shock, c/
j
is a xed cost parameter. I
t
is a unit root
technology level, which grows at rate
t
=
I
t
I
t1
, and c/
j
is a xed cost parameter. Fixed costs
grow at the same rate as technology to ensure zero prots in steady-state.
The production function for tourism services is given, in an analogous way, by:
A
c
t
()) =
c
t
(I
t
|
c
t
()))
c
1s
(1
c
t
()))
c
1s
(1
c
t
()))
1c
1s
c
1s
c/
c
I
t
(16)
Factor Demands Each period rm ) in sector , solves the cost minimization problem to
choose the optimal amount of each input given the respective prices.
minTC
t
()) = \
t
|
)
t
()) 7
1
t
1
)
t
()) 1
1
t
1
)
t
()) (17)
s.t producing given amounts of goods and services using technology (15) for , = q, and () for
, = :.
L = \
t
|
)
t
()) 7
1
t
1
)
t
()) 1
1
t
1
)
t
())
1
)
t
_
A
)
t
())
)
t
_
I
t
|
)
t
())
_
c
I
_
1
)
t
())
_
c
1
_
1
)
t
())
_
1c
I
c
1
c/
)
I
t
_
The FOCs for good , imply that:
|
)
t
()) =
c
|)
c
1)
7
1
t
\
t
1
)
t
())
1
)
t
()) =
1 c
|)
c
1)
c
1)
7
1
t
1
1
t
1
)
t
())
Given the optimality conditions and the production functions, maginal costs for sector j can
be written as:
'C
)
t
=
0TC
)
t
())
0A
)
t
())
=
_
\
t
c
|)
I
t
_
c
I
_
7
1
t
c
1)
_
c
1
_
1
1
t
1 c
|)
c
1)
_
(1c
I
c
1)
1

)
t
(18)
Price Setting Intermediate goods rms in each sector also have to decide on the prot
maximizing prices they will charge for their output. It is assumed that rm ) in sector , is able
to set new prices with probability (1
)
). The fraction of rms with contracts set t periods ago
is (1
)
)
t
)
. Intermediate goods prices are set considering the total demand for intermediate
goods dened in terms of consumption units, regardless of their nal use (this simplies the
model considerably). Firms, therefore, set an optimal price in period t that maximizes the
present value of all future expected real prots to which this price may apply. The price of
intermediate tourism services is set in a similar way, so as to maximize the present value of
10
expected tourism prots.When setting prices, intermediate rms in sector ,, choose a price
1
)
t
()) which maximize the market value of the rm:
'\
)
t
()) = 1
t
1

t=t
,
tt
`
t
F
)
t
())
where
F
)
t
()) = 1
)
t
())A
)
t
()) TC
)
t
(A
)
t
()))
s.t to the goods demand:
A
)
t
()) =
_
1
)
t
())
1
)
t
_

1
10
,I
A
)
t
Prices that cannot be re-set in period t are indexed by the indexation rate \
1,tt
0
, where
t
0
is the last period when prices could be re-set, hence:
1
)
t
()) = \
)
1,tt
0
1
)
t
0
())
where \
)
1,tt
0
= 1 fot t = t
0
, with probability
)
. Therefore, with probability
_

)
_
tt
the
price \
)
1t,tt
1
)
t
()) will be in eect in period t. Maximizing the value of the rm under this
assumption yields an optimal re-setting price 1
)
t
()), given by:
1
)
t
=
w
)
t
1
)
t
for, , = q, :, where
w
)
t
=
)
,1
t
w
)
t+1
_
1,\
)
1t+1,1
_
1
10

`
t
_
1
)
t
())
_

0
,I
10
,I
_
1
1 0
),t
_
'C
)
t
_
1
)
t
_
1
10

A
)
t
1
)
t
=
)
,1
t
_
1,\
)
1t+1,1
_
1
10

1
1
)
t+1
`
t
_
1
)
t
())
_

0
,I
10
,I
_
0
),t
1 0
),t
_
_
1
)
t
_
1
10

A
)
t
Notice that the ggregate price 1
)
t
can be written as:
1
)
t
=
_
_
1
)
_
1
)10
,I
t

)
_
\
)
1t,1
1
)
t1
_
10
,I
_ 1
10
,I
2.3 Energy Inputs
Energy input prices (oil and gas) aect Cypriot ination both directly through its direct im-
pact on the CPI (nal goods production), and indirectly via changes in the price of domestic
intermediate goods which are a function of the marginal cost (18). It is assumed that an energy
importing rm, imports all the quantities of energy inputs (oil and gas) demanded by rms at
the international market price (1
1
t
), expessed in euros, and sells these quantities to rms at
marginal cost (1
1
t
), independently of their use.
1
t
= 1
C
t
1
G
t
1
1
t

_
1
0
1
j
t
())d)
_
1
0
1
c
t
())d)
11
2.4 Foreign Demand
Foreign aggregate demand consists of demand for Cypriot goods by the Eurozone (A
j)
), and of
demand for tourism services by both the Eurozone (A
c,1Z
) and the rest-of-the-world (A
c,1cW
).
For consistency with the model it is assumed that these foreign demands are exogenously given
by:
A
j)
t
=
_
c
1Z,j
_
_
1
j
t
1
Y,1Z
t
_

1
1j
T2


1Z
t

t
1
1Z
t
where A
j)
t
is expressed in per head of Cypriot population. The parameter c
1Z,j
gives the
quasi-share of Cypriot goods in the production of Eurozone nal goods;
1
1j
T2

is the elasticity
of substitution of Cypriot goods in the Eurozone; 1
Y,1Z
t
is the Eurozone GDP deator; and
1
1Z
t
is Eurozone per-capita output (excluding the domestic economy). The denitions are
analogous for tourism services:
A
c,1Z
=
_
c
1Z,c
_
_
1
c
t
1
Y,1Z
t
_

1
1j
T2
s

1Z
t

t
1
1Z
t
A
c,1cW
=
_
c
1OW,c
_
_
1
c
t
,11
e/ROW
t
1
Y,1OW
t
_

1
1j
TCV
s

1OW
t

t
1
1OW
t
where 11
e/ROW
t
is the nominal exchange rate measured as euros per ROW currency and
A
c,)
, , = 17, 1O\ are expressed in per head of local population, and and 1
1OW
t
is the
rest-of-the-world per-capita output .
2.5 Indexation
It is assumed that economic agents take as an indexation factor the average of last periods gross
CPI ination rate and the economys steady-state gross CPI ination rate:
\
j
1t,tt
0
= \
c
1t,tt
0
= \
t,tt
0
=
_
t1

t=t
0
H
t
_

_
H
tt
0
_
1
Nominal wages are assumed to be indexed also to steady-state labour productivity growth.
\
Wt,tt
0
=
__
t1

t=t
0
H
t
_

_
H
tt
0
_
1
_
_
_
t

t=t
0+1

t
_
_
\
j
1t,0
= \
Wt,0
= 1
This is equivalent to assuming that a fraction g of non-adjusting rms index their prices acording
to last periods gross ination rate, while the other fraction uses the steady-state gross ination
rate (see also Del Negro, 2007).
12
2.6 Fiscal Authority
The government nances government consumption, debt payments, and net transfers with
revenues obtained from taxation and new debt issuance, hence the government budget constraint
is given by:
1
G,t
G
t
1
t1
1
t1

.
t
T1
t
= t
C
t
1
C
t
C
t
t
1
t
\
t
1
t
t
1
t
7
1
t
1
t1

.
t
1
t
where
.
t
is the growth rate of the population, since variables are dened in per-capita terms.
Dening current government expenditures and revenues:
1rj
G,t
1
G,t
G
t
(1
t1
1)
1
t1

.
t
T1
t
Io
G,t
t
C
t
1
C
t
C
t
t
1
t
\
t
1
t
t
1
t
7
1
t
1
t1

.
t
Dening the government budget decit 11
t
as:
11
t
1rj
G,t
Io
G,t
= 1
G,t
G
t
(1
t1
1)
1
t1

.
t
T1
t
t
C
t
1
C
t
C
t
t
1
t
\
t
1
t
t
1
t
7
1
t
1
t1

.
t
1
t
=
1
t1

.
t
11
t
Dening primary expenditures and the primary budget decit, that is the government expen-
ditures and decit net of interest payments:
1rj
jvin
G,t
1
G,t
G
t
T1
t
11
jvin
t
1rj
jvin
G,t
Io
G,t
= 1
G,t
G
t
T1
t
t
C
t
1
C
t
C
t
t
1
t
\
t
1
t
t
1
t
7
1
t
1
t1

.
t
11
t
= 11
jvin
t
(1
t1
1)
1
t1

.
t
To prevent an explosive debt path a scal rule is imposed on the primary decit, and net
transfers adjust automatically to fulll this rule, dened in terms of the log-linearized variables.
2.7 Monetary Authority
In DSGE models it is typical to assume that the central bank sets the nominal interest rate
according to a Taylor type rule. In its simplest form the Taylor rule indicates that interest
rates should react to ination and output gaps, but various generalizations have been proposed
in the literature. Two additional elements for instance are often considered (see Orphanides,
2007). The rst of this elements is the lagged interest rate, capturing interest rate smoothing,
which Woodford (2003) shows to be particularly important in models with strong expectations
channels. The other element are deviations of output growth (rather than level) from potential
output growth. A Taylor rule model including all four elements (ination gap, output gap,
13
lagged interest rate, and output growth gaps) has the empirical advantage that it can be shown
to nest a variety of monetary policy strategies, including money growth targeting, reformulated
in terms of an interest rate instrument (see Orphanides, 2007). This can be particularly useful
if there are breaks in monetary policy regimes. For some small open economies an exchange
rate objective can also be embedded in the Taylor rule (see, for instance, Adolfson et al., 2007).
For countries that belong to a monetary union monetary policy is decided by the central bank
of the union. When the union as a whole is modeled alongside the individual country, a Taylor
rule for the union is considered. When the country is small and modeled on its own as a small
open economy and the country is too small to inuence the common central banks decisions,
interest rates may be assumed exogenous (see Almeida, 2009), but in most cases some weight
for the country in the common central bank utility function is allowed for and this improves
the convergence properties of the mode (see Burriel at al., 2010). In the case of this model we
use a simple interest rate rule for the ECB of the form:
1
1Z
t
1
1Z
=
_
1
1Z
t1
1
1Z
_
j
T
_
_
H
1Z
t
H
1Z
_

r
_
1
1Z
t
1
1Z
_

_
(1j
T
)
-
11Z,t
(19)
where barred variables indicate steady-state values. We allow for some weight for the local
economy in this monetary policy function. The ECBs monetary policy will also have an
impact on the the euros exchange rate which is relevant for trade with the rest of the world.
For determining this eect it will be exogenously assumed that UIP holds between the euro
and ROW currencies:
1
1Z
t
1
1OW
t
= 1
t
_
11
e/ROW
t+1
11
e/ROW
t
_
2.8 Market Clearing Conditions
In this economy the market clearing conditions for labour capital and energy are given by the
following equations:
1
t
= 1
C
t
1
G
t
1
1
t
1
j
t
1
c
t
(20)
1
j
t
=
_
1
0
1
j
t
())d) (21)
1
c
t
=
_
1
0
1
c
t
())d) (22)
1
t1
= 1
j
t
1
c
t
(23)
1
j
t
=
_
1
0
1
j
t
())d) (24)
1
c
t
=
_
1
0
1
c
t
())d) (25)
14
|
t
= |
j
t
|
c
t
(26)
|
j
t
=
_
1
0
|
j
t
())d) (27)
|
c
t
=
_
1
0
|
c
t
())d) (28)
In domestic intermediate market the market clearing condition guarantees that the supply
of domestic intermediate goods for domestic use must be equal to the total demand for them,
from nal good producers:
A
jo
t
= A
jo,C
t
A
jo,G
t
A
jo,1
t
(29)
Equivalently for the supply of imported intermediate goods:
A
jn
t
= A
jn,C
t
A
jn,G
t
A
jn,1
t
(30)
Considering also exports, the market clearing conditions for total domestic production is
given by:
A
j
t
= A
jo
t
A
j)
t
A
c
t
= A
c,1Z
t
A
c,1OW
t
Notice that A
j)
t
and A
c,1Z
t
, and A
c,1OW
t
are expressed in domestic per capita terms.
2.8.1 GDP and Balance of Payments
Aggregating across households, yields the aggregate resource constraint for the economy:
G11
t
1
Y
t
1
t
= 1
C,t
C
t
1
G,t
G
t
1
1,t
1
t
1
1Z
t
1
1Z
t1
O
_

/
1Z
t1
, -
,t1
_
1

.
t
1
1Z
t1
where
.
t
=
.
I
.
I1
is the gross rate of growth of the population, and:
1
t
A
j
t
A
c
t
1
j
t
1
c
t
1
Y
t
=
A
j
t
1
t
1
j
t

A
c
t
1
t
1
c
t

1
j
t
1
c
t
1
t
1
1
t
The aggregare resource constraint equation uses the balance of payments equilibrium condition:
1
j
t
A
j)
t
1
c
t
A
c
t
1
n
t
A
jn
t
1
1
t
1
t
= 1
1Z
t
1
1Z
t1
O
_

/
1Z
t1
, -
,t1
_
1

.
t
1
1Z
t1
15
Figure 1: Main Agents and Flows
Cyprus
Financial Markets
net foreign assets
C(X
gd
,X
gm
,E)
I(X
gd
,X
gm
,E)
G(X
gd
,X
gm
,E)
X(K,l,E) = X
gd
+ X
gf
+ X
sf
K
l
E
Households
ECB
Monetary
Policy
Government
X
gm
Shocks
Trade
Imports
Exports:
Goods&Services
and Tourism
Energy
Final Goods Firms
Intermediate Goods Firms
Production Factors
euro area
ROW
euro
exchange
rate
2.9 Shocks
The stochastic behaviour of the model is given by a series of shocks that can be represented as
follows:
-
t
=
.
-
t1
o
.
j
.,t
j
.,t
s i.i.d (0, 1)
where - =
_
-
C
, -
|
, -
1
, -
,t
, j
j
, j
c
,
j
,
c
, j
n
, j
1
, q, j
1OW
,
1OW
, 1
1OW
, j
1Z
,
1Z
, -
1
T2
_
,
and hats represent deviations from steady state. Figure 1 gives a summary of the model
relations.
3 Steady-State Calibration
The solution strategy used to nd the steady-state of the model follows closely the solution
strategy described in Christoel et al. (2008). The steady-state of the model has been solved
and the steady-state parameters have been calculated to match the long-term properties (long-
term average shares relative to output) observed in the available data for Cyprus. The steady-
state parameters chosen to match the properties of aggregate available data are listed in Table
1.
For some of the parameters, the calibration has followed the literature. The long-run growth
rate of the economy and the steady-state ination rate have been both set at annualized rate of
2%, in line with the euro area potential output growth and with the ECBs target for ination.
The parameter , has been calibrated to yield an annualized long-term nominal interest rate of
16
Table 1: Calibration of steady-state parameters.
Calibrated steady-state parameters
Decription Parameter Value
utility discount factor 0.999
rate of capital depreciation 0.02
share of imports in final consumption basket
M,C
0.45
share of imports in final investment basket
M,I
0.45
share of imports in final government consumption basket
M,G
0.45
share of energy in final baskets
E
0.02
labour share in the domestic goods sector
lg
0.59
capital share in the domestic goods sector
Kg
0.38
labour share in the tourism sector
ls
0.69
capital share in the tourism sector
Ks
0.28
inverse of the price markup for the goods sector (OECD)
g
0.74
steady-state growth rate of productivity 1.005
steady-state population growth rate

1.002
steady-state capital tax rate
K
0.48
steady-state consumption tax rate
C
0.17
steady-state labour tax rate
l
0.25
4.5%, in line with that of the euro area (see Christoel et al., 2008). In line with the literature,
we set the capital depreciation rate at 2%, while the tax rate on capital has been calibrated
to yield a capital-to-output ratio of about 7.5. The import shares were calibrated so that the
imports to GDP ratio can match closely the data, and the shares of energy in both the retail
and the intermediate goods sector have been set so that the energy to output ratio is close to
its empirical counterpart of 5%. The labour and capital shares across sectors have calibrated
to approximate the labour ratios that could be observed in the data. The consumption and
labour tax rates were calibrated so that the government revenue to GDP ratios approximate
the ratios observed in the data. The inverse of the steady-state price markup in the goods
and other services sector has been calibrated yield a price markup of about 1.35, in line with
OECD estimates (see Martins et al., 1996). The inverse of the steady-state price markup in the
tourism sector and the steady-state TFP in the tourism sector are jointly calibrate so that the
steady-state ratio of tourism to consumer prices is one (this calibration considerably simplies
the solution). In Table 2 we show the steady-state solution that these parameters generate and
compare them with the available aggregate data.
17
Table 2: Steady-State Output Shares Model and Data.
Steady-state output shares
Variable description Name Model Data
capital-output ratio K/Y 7.578 -
investment-output ratio I/Y 0.188 0.19
home demand for domestic goods X
gd
/Y 0.539 -
foreign demand for domestic goods X
gf
/Y 0.344 0.33
total demand domestic goods X
g
/Y 0.883 -
home demand domestic goods, consumption X
gd,C
/Y 0.340 -
home demand domestic goods, investment X
gd,I
/Y 0.102 -
home demand domestic goods, government X
gd,G
/Y 0.097 -
demand for goods and services imports, consumption X
gm,C
/Y 0.278 -
demand for goods and services imports, investment X
gm,I
/Y 0.083 -
demand for goods and services imports, government X
gm,G
/Y 0.079 -
euro-area demand for domestic tourism services X
s,EZ
/Y 0.074 0.07
rest-of-the world demand for domestic tourism services X
s,ROW
/Y 0.074 0.07
total demand for domestic tourism services X
s
/Y 0.148 0.14
energy use, final consumption good E
C
/Y 0.013 -
energy use, final investment good E
I
/Y 0.004 -
energy use, final government consumption good E
G
/Y 0.004 -
energy use, goods and other services sector E
g
/Y 0.004 -
energy use, tourism sector E
s
/Y 0.031 -
energy use, total E/Y 0.051 0.05
capital-output ratio, goods and other services sector K
g
/Y 6.713 -
capital-output ratio, tourism sector K
s
/Y 0.826 -
labour-output ratio, goods and other services sector l
g
/Y 0.476 -
labour-output ratio, tourism sector l
s
/Y 0.093 -
labour-output ratio l/Y 0.569 -
government debt B/Y 0.600 -
government consumption G/Y 0.180 -
government transfers TR/Y 0.264 -
government expenditures Exp/Y 0.451 0.47
government primary expenditures Exp
prim
/Y 0.444 -
consumption taxes Tax
C
/Y 0.107 -
labour taxes Tax
l
/Y 0.156 -
capital taxes Tax
K
/Y 0.182 -
government revenue Rev/Y 0.445 0.44
government deficit BD/Y 0.006 -
government primary deficit BD
prim
/Y -0.001 -
private consumption C/Y 0.632 0.64
imports of goods and services X
gm
/Y 0.492 0.51
total exports of goods and services (X
gf
+X
s
)/Y 0.492 0.48
Table 3 provides additional information on labour ratios, by comparing the sectoral labour
shares generated by the model with those available in the data. In the model the labour share
of the tourism sector appears slightly overstated.
4 Bayesian Estimation
Using the steady state calibration described earlier, we used the log-linearized model equations
to estimate the dynamics of the model to match a sub-set of macroeconomic series.
6
The vari-
ables were selected based on data quality and convergence of the optimization algorithm. In
the nal specication we have retained seven variables: output (y); consumption (c); govern-
ment consumption as a share of GDP (gy); investment (inv); consumer price ination Pi; hours
6
The estimation and analysis of the model uses Dynare (see Adjemian et al., 2011).
18
Table 3: Sectoral Labour Shares
Labour Supply: Share of total
Description Variable model data
labour-output ratio, goods and other services sector l
g
/Y 0.84 0.86
labour-output ratio, tourism sector l
s
/Y 0.16 0.14
labour-output ratio l/Y 1 1
worked (l); the euro area interest rate (REZ) and tourism output proxied by estimated tourism
revenues (xs).
The data is quarterly and covers the period 1995Q1 to 2012Q2. The series were collected
from several sources, including Eurostat, the Statistical Service of Cyprus, the Central Bank of
Cyprus and the Euro Area Wide Model Database (for euro area variables). All the series are
adjusted for seasonality. Data on tourism revenues is limited, and in this case regression analysis
was used to obtain a longer time series. Series on tourism revenues for Cyprus are available from
the Statistical Service of Cyprus (CyStat); however the data starts from 2001. To estimate our
series we made use of three monthly series, namely the tourism arrivals, tourism revenues and
the real eective exchange rate for Cyprus. The exchange rate series were taken from Eurostat
while the rest were taken from CyStat. The tourism arrivals and the real exchange rate series
both start in 1995M1, but the tourism revenues were available only since 2001M1. To estimate
the tourism revenues for the period 1995M1 to 2000M12, we regressed the available real tourism
revenues (deated using the CPI deator), on tourism arrivals and the real eective exchange
rate of Cyprus, all in logarithms. The Adjusted-R2 turned out to be 0.99 (or 0.67 when we
include a constant in the model). All the time series were seasonally adjusted. The sample of
the regression spanned the period from 2001M1 to 2011M12. Given the availability of tourism
arrivals and the real exchange rate, we were then able to compute the tted values of tourism
revenues for the whole period including the missing period, i.e. 1995M1 to 2000M12. Since the
regression was estimated in logs, we took the exponential of the tted values as an estimate
of real tourism revenues for Cyprus. Finally, the estimated monthly tourism revenues were
converted into quarterly. The resulted estimated quarterly series for tourism revenues can be
seen in Figure 2. For comparison reasons, in the graph below we also have included the travel
credit series from the balance of payments, representing the exported travel products of Cyprus.
This series can be found in the Central Bank of Cyprus database and like the rest related series,
it is also short in terms of period covered, starting from 2001Q1.
Using these variables we were able to estimate the parameters determining the exogenous
shocks hitting the Cypriot economy as well as the parameters determining habit formation
(h), wage rigidity (
W
), price rigidity (
j
and
c
), and indexation (). Coecients which were
not estimated have been calibrated from the literature. Table 4 shows reports the calibration
choices for non-estimated parameters that aect only the dynamics of the model (are not used
to dene the steady-state of the model). In this version of the model the share of rule-of-thumb
consumers has been set to zero, since the performance of the model was worse for positive shares.
The risk parameter is a parameter that is typically dicult to identify and hence it is typically
19
Figure 2: Tourism Revenues in Real Terms, Data and Estimates
calibrated in the literature. Our calibration follows (Christoel et al., 2008) and the parameter
is set at 0.01 so that the evolution of the current account has a relatively small impact on the
risk premium in the short-run, while the net foreign asset position is stabilized at zero over
a reasonable period of time. Another parameter that is dicult to identify according to the
literature is the inverse of the Frisch elasticity of labour supply. This has been set equal to 2
in line with available estimates reported in the literature. Elasticities of substitution have also
been calibrated in this study. The elasticity of substitution between labour varieties has been
xed so as to yield a wage markup of markup of 1.3, which is consistent with OECD estimates
(see Jean and Nicoletti, 2002). The parameters determining the elasticities of substitution
between imports and domestic goods and services and between goods and services and energy
were taken from the calibrations of Cuche-Curtis et al. (2009) for Switzerland. The elasticities
of substitutions for exports were calibrated in line with that for imports. Finally, we also chose
to calibrate policy parameters. Monetary policy parameters were taken from Adolfson et al
(2007), while the scal rule parameters calibration follows Gali at al. (2007), and ensures that
government debt converges close to the long-term target within simulation periods.
For the parameter estimation we had to use priors which we initially obtained from previous
literature (Christoel et al., 2008) but subsequently adjusted to improve the estimation results.
The retained priors are summarized in Tables 5 to 7, along with the posterior estimates. The
posterior distributions reported in the table are based on a Markov chain with 150,000 draws
from which 75% have been discarded, and 5 parallel chains.
7
The parameter estimates obtained
are broadly consistent with the literature. For instance, wages are found to be stickier than
prices, a result which is commonly found in the literature. The relatively high stickiness of
wages is a fairly intuitive result for Cyprus where wage negotiations remain highly centralized.
7
According to Grioli (2011) this improves the computation of between group variance of the parameter
means, one of the key criteria to evaluate the e ciency of the Metropolis-Hastings to evaluate the posterior
distribution.
20
Table 4: Calibration of Remaining Parameter Values.
Cal i brated Dynami c Parameters
Descri pti on Parameter Val ue
Ri sk premi um sensi ti vi ty to CA posi ti on

0.01
El asti ci ty of substi tuti on vari eti es of l abour
l 4.35
Parameter determi ni ng the el asti ci ty of substi tuti on between imports and domesti c goods.
M 0.01
Parameter determi ni ng the el asti ci ty of substi tuti on between energy and intermedi ate goods
E -10.0
Parameter determi ni ng the Eurozone el asti ci ty of substi tuti on for domesti c goods
g,EZ 0.01
Parameter determi ni ng the Eurozone el asti ci ty of substi tuti on for touri sm services
s,EZ 0.01
Parameter determi ni ng the Rest-of-the-worl d el asti ci ty of substi tuti on for touri sm services
s,ROW 0.01
Inverse of the Fri sch el asti ci ty of l abour suppl y 2.00
Eurozone Taylor Rul e i nterest rate smoothi ng parameter
R 0.88
Eurozone Taylor Rul e i nflati on gap wei ghti ng parameter

2.50
Eurozone Taylor Rul e output gap wei ghti ng parameter

0.15
Fi scal rul e parameter - debt
b 0.20
Fi scal rul e parameter - defi ci t
d 0.10
Investment adjutsment costs parameter
Inv 5.00
Additionally, the persistence of investment-specic shocks is estimated to be relatively low,
while the standard deviation of investment-specic shocks is estimated to be relatively large as
would be expected from investment data. Also consistent with the literature is the fact that
the persistence of tax rate shocks is estimated to be higher than the persistence of government
spending shocks. Plots of the data used in estimation, together with plots of prior and posterior
distribution and smoothed innovations are reported in the Appendix. The estimated innovations
appear stationary but may exhibit some autocorrelation in a few cases which could be resolved
with the inclusion of more autoregressive terms in the shock equations, at the cost of an increase
in the number of parameters.
Table 5: Prior and Posterior Distributions of Estimated Parameters.
Shape Mean St. Dev. Mode Mean 5% 95%
Habits
h beta 0.70 0.20 0.7167 0.7551 0.6660 0.8469
Wage and Price Setting
w beta 0.70 0.10 0.7411 0.7034 0.5866 0.8227
g beta 0.70 0.10 0.5874 0.5682 0.3987 0.7383
s beta 0.70 0.10 0.6098 0.5647 0.3817 0.7453
beta 0.70 0.10 0.4286 0.4441 0.2749 0.6152
Pri or Posteri or
Parameter
5 Diagnosis
Using both the calibrations and the estimated parameters, we analyze the impulse responses of
the model to a range of shocks to evaluate the properties of the model. The simulation horizon
corresponds to 20 quarters. Figure 4 (displayed at the end) shows impulse-response functions
to a set of selected structural shocks: (i) a consumption preference shock; (ii) an investment
specic shock; (iii) a transitory technology shock in the goods and other services sector; (iv)
a transitory technology shock in the tourism sector; (v) a euro area ination shock; and (vi)
a rest-of-the-world output shock. All impulse responses are reported as percentage deviations
from the non-stochastic steady state, except for the impulse responses of the ination and
21
Table 6: Prior and Posterior Distributions of Estimated Parameters - Autoregressive parameters
for Shocks.
Shape Mean St. Dev. Mode Mean 5% 95%
Autoregressive parameters - shock persistence
beta 0.70 0.10 0.5048 0.4951 0.3413 0.6469
C beta 0.60 0.10 0.5591 0.5003 0.3021 0.6893
l beta 0.60 0.10 0.6090 0.5992 0.4400 0.7591
Inv beta 0.20 0.10 0.0644 0.0972 0.0169 0.1726
G beta 0.50 0.10 0.3988 0.4031 0.2803 0.5273
beta 0.80 0.10 0.7287 0.7058 0.5686 0.8418
g beta 0.20 0.10 0.0625 0.0937 0.0144 0.1720
s beta 0.50 0.10 0.4889 0.4821 0.3243 0.6420
pm beta 0.70 0.10 0.6072 0.6018 0.4438 0.7635
pE beta 0.70 0.10 0.7215 0.7026 0.5463 0.8653
pyEZ beta 0.50 0.10 0.6645 0.6507 0.5029 0.7969
pyROW beta 0.50 0.10 0.7206 0.6824 0.5132 0.8507
RROW beta 0.70 0.10 0.6703 0.6671 0.5244 0.8262
yEZ beta 0.50 0.10 0.4887 0.4915 0.3362 0.6449
yROW beta 0.50 0.10 0.4981 0.4903 0.3309 0.6593
C beta 0.70 0.10 0.7202 0.7013 0.5461 0.8650
l beta 0.70 0.10 0.7221 0.7011 0.5399 0.8683
K beta 0.70 0.10 0.7212 0.6982 0.5473 0.8629
EZ beta 0.50 0.10 0.4562 0.4647 0.3062 0.6149
ROW beta 0.50 0.10 0.5021 0.5066 0.3511 0.6675
Ag beta 0.70 0.10 0.7131 0.6984 0.5680 0.8294
As beta 0.70 0.10 0.7345 0.7121 0.5642 0.8637
REZ beta 0.70 0.10 0.4088 0.4159 0.2936 0.5370
Parameter
Pri or Posteri or
interest rates which are reported as percentage-point deviations.
A consumption preference shock increases consumption. The highest impact of the shock
is felt with a delay due to the consumption habits inbuilt in the model. Investment falls and
only starts to recover after 5 quarters. Goods and services production and GDP increase
together with the increase in consumption demand, and so does import demand. Wages and
employment increase together with ination. Higher imports deteriorate the current account
and put upward pressure on interest rates. Pressure on marginal costs induces a contraction of
the tourism sector.
An investment shock increases investment on impact and lowers consumption temporarily.
Output increases to face the higher investment demand. There is a short-lived fall in tourism
output as resources are diverted to the goods and services sector, but in the long-run the sector
benets from lower marginal costs. As production progressively shifts towards capital, employ-
ment falls after the rst quarter and so does wage ination. This shock puts also downward
pressure on ination.
A transitory technology shock in the goods and other services sector triggers a decline in real
marginal cost. This decline in marginal cost causes prices to fall, as the prices of intermediate
goods are set as a markup on marginal cost. With domestic demand adjusting only sluggishly
to the shift in supply, both employment and wages go down. The supply shock also causes a
temporary increase in investment. In the tourism sector the real depreciation of the domestic
currency leads to expenditure switching away from foreign towards domestic goods, thereby
boosting exports in the long-run although there is a short-term diversion of resources away
from the tourism sector.
The transmission of a transitory productivity shock in the tourism sector is similar to that
22
Table 7: Prior and Posterior Distributions of Estimated Parameters - Standard Deviations of
Shocks.
Shape Mean St. Dev. Mode Mean 5% 95%
Standard deviations of shocks
i nv.gamma 0.05 i nf 0.0100 0.0105 0.0078 0.0130
C i nv.gamma 0.05 i nf 0.0222 0.0447 0.0109 0.0867
l i nv.gamma 0.05 i nf 0.0230 0.0454 0.0116 0.0884
Inv i nv.gamma 0.05 i nf 0.1826 0.1742 0.1349 0.2191
G i nv.gamma 0.05 i nf 0.0261 0.0267 0.0229 0.0304
i nv.gamma 0.05 i nf 0.0288 0.0318 0.0149 0.0479
g i nv.gamma 0.05 i nf 0.0244 0.0260 0.0165 0.0350
s i nv.gamma 0.05 i nf 0.0219 0.0380 0.0113 0.0683
pm i nv.gamma 0.05 i nf 0.0130 0.0136 0.0104 0.0167
pE i nv.gamma 0.05 i nf 0.0226 0.0356 0.0119 0.0595
pyEZ i nv.gamma 0.05 i nf 0.0188 0.0206 0.0125 0.0286
pyROW i nv.gamma 0.05 i nf 0.0231 0.0673 0.0118 0.1578
RROW i nv.gamma 0.05 i nf 0.0693 0.0482 0.0147 0.0815
yEZ i nv.gamma 0.05 i nf 0.0256 0.0249 0.0156 0.0338
yROW i nv.gamma 0.05 i nf 0.0234 0.0592 0.0111 0.1361
C i nv.gamma 0.05 i nf 0.0229 0.0412 0.0123 0.0793
l i nv.gamma 0.05 i nf 0.0230 0.0397 0.0127 0.0707
K i nv.gamma 0.05 i nf 0.0230 0.0360 0.0124 0.0623
EZ i nv.gamma 0.05 i nf 0.0146 0.0161 0.0110 0.0211
ROW i nv.gamma 0.05 i nf 0.0234 0.0609 0.0116 0.1254
Ag i nv.gamma 0.05 i nf 0.0097 0.0099 0.0080 0.0117
As i nv.gamma 0.05 i nf 0.0210 0.0250 0.0132 0.0366
REZ i nv.gamma 0.05 i nf 0.0063 0.0066 0.0059 0.0073
Parameter
Pri or Posteri or
of a shock to the goods and other services sector. Since this sector is more labour intensive,
the decline in employment and the increase in investment are less pronounced. Improvement
in the countrys current account position leads to a reduction of interest rate risk premium and
a small reduction of the domestic interest rate.
An euro area ination shock prompts the European central bank to increase euro area
interest rates. Domestic demand is negatively aected with investment falling relatively more
that consumption. The increase in euro area interest rates prompts a nominal appreciation of
the euro which translates into a real appreciation for Cyprus, given also the expected ination.
This real appreciation has a signicant negative impact on the tourism sector, and output is
this sector is falls relatively more than in the remaining sectors. Following the fall in demand,
rms reduce their demand for labour and employment falls along with nominal wages.
An output shock in the rest-of-the -world increases demand for tourism services signicant,
and has also a positive eect on employment and output. The eect on the goods and other
services sector, as well as the eect on comsumption and investment is positive but more dicult
to quantify.There is a positive but small eect on ination, and a real appreciation, consistent
with the revenue inow from an increase in tourism receipts.
In the DSGE-CY model the contribution of the various shocks to the uctuation of en-
dogenous variables can be analyzed through variance-decompositions. Table 8 shows the vari-
ance decompositions of selected variable to shocks. To facilitate the analysis the shocks have
been grouped into technology shocks (productivity and investment shocks); demand shocks
(consumption preference, consumption tax rate, and government consumption shocks); supply
shocks (price markup and capital tax rate shocks); labour market shocks (labour preference
and labour tax rate shocks); foreign shocks (foreign output, foreign ination, risk premium,
23
Figure 3: Sources of Fluctuations in GDP - Historical Decompositions of Shocks
and foreign interest rates shocks). The results show that foreign shocks explain the largest
percentage of Cyprus macroeconomic uctuations.
Table 8: Variance decompositions, in percent.
Vari abl es
Technol ogy Demand Suppl y Labour Forei gn Total
Output 0.6 2.3 2.3 0.0 94.8 100
Consumpti on 0.9 13.8 1.3 0.0 83.9 100
i nv 37.3 0.1 1.8 0.0 60.8 100
xg 1.0 4.7 0.8 0.0 93.6 100
xs 0.5 0.0 7.8 0.0 91.7 100
l 14.0 0.1 47.5 0.1 38.3 100
Shocks
The relative weight of foreign shocks in the model can also be observed with the historical
decomposition of observed variables used in the estimation into the contributions of its struc-
tural shocks (see Figure 3). Here we have focused on the decompositions of GDP cycles (the
decomposition of consumption and investment in shown in the Appendix). Shocks have been
once more grouped into the ve categories mentioned above. As it can be observed, foreign
shocks tend to drag GDP in most of the cycles. Domestic demand shocks have some weight
in explaining uctuations in consumption, but are less signicant for determining GDP cycles,
where technology shocks have more weight (technology shocks appear to have had a signicant
weight in the 2004 downturn). It is important to notice that the 2010-2011 downturn cannot
be fully captured within the model as this would require extending the model with a banking
sector and credit supply shocks, but the dominance of negative technology shocks in this period
(which include investment specic shocks) is consistent with the negative shock to investment
felt during this period.
24
6 Conclusions and Further Research
In this study, a New-Keynesian DSGE model for a small open economy integrated in a monetary
union has been developed and estimated for the Cypriot economy using a Bayesian approach. In
addition, the study has also provided a survey of the literature associated with DSGE models,
and a description of a set of small open economy models recently produced by other studies.
This is the rst eort to build a DSGE model for Cyprus which can be used for policy
analysis. The modeling strategy is ambitious in the sense that it deviates from the literature
by incorporating two external blocks instead of one (the euro area and the rest of the world),
and because it considers two intermediate sectors. A subset of model parameters has been
estimated to allow the model to better match the moments observed in macroeconomic data
for Cyprus, while the remaining sub-set of parameters was calibrated with values obtained from
the literature.
We have presented the estimation results obtained by employing Bayesian methods. The
obtained estimates for the parameters of interest are generally in line with the DSGE literature.
Among them, some are particularly noteworthy. Wages stickiness is found to be higher than
price stickiness, a result commonly found in the literature. The relatively high stickiness of
wages is a fairly intuitive result for Cyprus where wage negotiations remain highly centralized.
In addition, wages and prices exhibit a considerable degree of indexation to past ination,
a result that is in line with the formal link of wages to ination still in place. Also, the
persistence of investment-specic shocks is estimated to be relatively low, while the standard
deviation of investment-specic shocks is estimated to be relatively large as would be expected
from investment data. Equally consistent with the literature is the fact that the persistence of
tax rate shocks is estimated to be higher than the persistence of government spending shocks.
We have further examined the empirical properties of the model by studying its impulse-
response functions and variance decompositions. Overall, the results indicate that the estimated
DSGE-CY has economically plausible properties, especially with regard to the propagation of
key economic shocks.
Some problems were however encountered, which should be mentioned. In particular, the
treatment of the data was a complex task, and we opted for the use of hp-ltered data, which
is recognized to be subject to a number of caveats. Also, some identication problems may
exist, as this is a common problem in the estimation of medium-large scale DSGE models.
Although the data was informative in the majority of the cases, in some cases the prior and
posterior distributions overlap, and some estimates seem to be signicantly inuenced by the
chosen priors, an inuence that ideally should have been minimal. It also appears that some of
the estimated innovations may exhibit autocorrelation. In addition, this version of the model
does not incorporate many of the adjustment costs, typically found in the literature, needed
to produce smoother and well-shaped impulse responses. This simplication compensates the
more structural complexity of the model, limiting the number of parameters, and therefore
increasing the probability of convergence of the optimization algorithms.
25
The existence of these caveats indicates that this work can be improved in some dimensions.
The estimated version of the DSGE-CY can be subject to further renements in the light of
the practical experience gained with this exercise. As mentioned the model contains a number
of simplifying assumptions which can be relaxed. For instance, the number of autocorrelation
terms in some of the shock equations could be increased to deal with autocorrelation. In
addition, some of the building blocks of the model can be modeled in further detail, like energy
markets. The interactions among foreign variables can also be improved with the use of VARs
(e.g. in the current version import prices are unrelated to energy prices). The scal and foreign
variables of the model can also be modeled outside the model, following Adolfson et al. (2007).
This strategy would allow for a more realistic treatment of these variables and a considerably
reduction of the "estimation burden" currently imposed.
Finally, a better description of the 2010-2011 downturn which is still unfolding in Cyprus
as a consequence of the Greek debt crisis would require extending the model with a banking
sector, and exercise which we leave to further research.
26
Figure 4: Impulse Responses
5 10 15 20
0
2
4
x 10
- 3
y
5 10 15 20
0
5
10
15
x 10
- 3
c
5 10 15 20
-10
-8
-6
-4
-2
x 10
- 3
i nv
5 10 15 20
0
2
4
6
x 10
- 3
xg
5 10 15 20
-10
-5
0
x 10
- 4
xs
5 10 15 20
0
5
10
x 10
- 4
Pi
5 10 15 20
0
2
4
6
x 10
- 3
l
5 10 15 20
0
5
10
x 10
- 4
Pi W
5 10 15 20
-2
-1
0
x 10
- 3
rerROW
IRFs to a consumption preference shock
5 10 15 20
1
1.5
2
2.5
x 10
- 3
y
5 10 15 20
-4
-2
0
2
4
x 10
- 3
c
5 10 15 20
0
0.01
0.02
0.03
i nv
5 10 15 20
0.5
1
1.5
2
2.5
x 10
- 3
xg
5 10 15 20
0
2
4
6
x 10
- 3
xs
5 10 15 20
-4
-3
-2
-1
x 10
- 3
Pi
5 10 15 20
-2
0
2
x 10
- 3
l
5 10 15 20
-3
-2
-1
x 10
- 3
Pi W
5 10 15 20
0
5
10
15
x 10
- 3
rerROW
IRFs to an investment specic shock
27
5 10 15 20
0
1
2
3
x 10
- 3
y
5 10 15 20
0
2
4
x 10
- 3
c
5 10 15 20
0
5
10
x 10
- 3
i nv
5 10 15 20
0
1
2
3
x 10
- 3
xg
5 10 15 20
-2
0
2
x 10
- 3
xs
5 10 15 20
-8
-6
-4
-2
0
x 10
- 3
Pi
5 10 15 20
-15
-10
-5
0
x 10
- 3
l
5 10 15 20
-4
-2
0
x 10
- 3
Pi W
5 10 15 20
5
10
15
x 10
- 3
rerROW
IRFs to a transitory productivity shock - goods and services sector
5 10 15 20
0
1
2
x 10
- 3
y
5 10 15 20
-1
0
1
2
3
x 10
- 4
c
5 10 15 20
-2
0
2
4
6
x 10
- 4
i nv
5 10 15 20
0
10
20
x 10
- 5
xg
5 10 15 20
0
5
10
15
x 10
- 3
xs
5 10 15 20
-6
-4
-2
0
x 10
- 4
Pi
5 10 15 20
-4
-2
0
x 10
- 3
l
5 10 15 20
-4
-3
-2
-1
0
x 10
- 4
Pi W
5 10 15 20
-5
0
5
x 10
- 4
rerROW
IRFs to a transitory technology shock - tourism sector
28
5 10 15 20
-10
-8
-6
-4
-2
x 10
- 3
y
5 10 15 20
-15
-10
-5
x 10
- 3
c
5 10 15 20
-0.03
-0.02
-0.01
0
i nv
5 10 15 20
-8
-6
-4
-2
x 10
- 3
xg
5 10 15 20
-0.025
-0.02
-0.015
-0.01
-0.005
xs
5 10 15 20
-6
-4
-2
0
2
x 10
- 3
Pi
5 10 15 20
-10
-5
0
x 10
- 3
l
5 10 15 20
-4
-2
0
2
x 10
- 3
Pi W
5 10 15 20
-0.05
-0.04
-0.03
-0.02
-0.01
rerROW
Figure 4: IRFs to an euro area ination shock
References
[1] Adjemian, S., Bastani, H., Juillard, M., Karame, F., Mihoubi, F., Perendia, G., et al.
(2011). "Dynare: Reference Manual, Version 4". Dynare Working Papers, 1. CEPREMAP.
[2] Adolfson, M., Laseen, S., Linde, J., & Villani, M. (2007). "Bayesian estimation of an open
economy DSGE model with incomplete pass through". Journal of International Economics,
72(2), 481 - 511.
[3] Adolfson, M., Laseen, S., Linde, J., & Villani, M. (2008). "Evaluating an estimated new
Keynesian small open economy model". Journal of Economic Dynamics & Control 3, 32,
26902721.
[4] Almeida, V. (2009). "Bayesian Estimation of a DSGE Model for the Portuguese Economy".
Working Paper 14/2009. Bank of Portugal.
[5] Bergin, P., & Tchakarov, I. (2003). "Does Exchange Rate Risk Matter for Welfare? A
Quantitative Investigation". NBER Working Paper No.9900.
[6] Breuss, F., & Rabitch, K. (2009). "An estimated two country DSGE model of Austria and
the Euro Area". Empirica, 36, 123-158.
[7] Burnside, C., Eichenbaum, M., & Fisher, J. (2004). "Fiscal Shocks and Their Conse-
quences". Journal of Economic Theory, 115, 89117.
29
5 10 15 20
0
2
4
6
8
x 10
- 3
y
5 10 15 20
0
2
4
6
x 10
- 4
c
5 10 15 20
0
5
10
x 10
- 4
i nv
5 10 15 20
0
1
2
3
x 10
- 4
xg
5 10 15 20
0
0.02
0.04
0.06
xs
5 10 15 20
0
5
10
15
x 10
- 4
Pi
5 10 15 20
0
5
10
x 10
- 3
l
5 10 15 20
0
5
10
x 10
- 4
Pi W
5 10 15 20
-4
-3
-2
-1
x 10
- 3
rerROW
Figure 5: IRFs to a ROW output shock
[8] Burriel, P., Fernandez-Villaverde, J., & Rubio-Ramirez, J. (2010). "MEDEA: a DSGE
model forthe Spanish economy". Journal of the Spanish Economic Association, 1, 175
243.
[9] Calvo, G. (1983). "Staggered Prices in a Utility Maximizing Framework" . Journal of
Monetary Economics, 12, 38398.
[10] Canova, F. (2007). Methods for Applied Macroeconomic Research. NJ: Princeton Univer-
sity Press.
[11] Canova, F., and Dallari, P. (2013), "How important is tourism for the international trans-
mission of cyclical uctuations? Evidence from the Mediterranean.", working paper, Eu-
ropean University Institute.
[12] Canzoneri, M., Cumby, R., & Diba, B. (2007). "The Cost of Nominal Rigidity in NNS
Models". Journal of Money, Credit and Banking, 39, 15631586.
[13] Carlin, B., & Louis, T. (2009). Bayesian Methods for Data Analysis. Chapman &
Hall/CRC.
[14] Christiano, L., Trabandt, M., & Walentin, K. (2010). "DSGE Models for Monetary Policy
Analysis". In a. M. Benjamin M. Friedman (Ed.), Handbook of Monetary Economics, Vol.
3A. The Netherlands: North-Holland.
30
[15] Christoel, K., Kuester, K., & Linzert, T. (2009). "The Role of Labor Markets for Euro
Area Monetary Policy". European Economic Review, 53, 908-936.
[16] Cuche-Curti, N., Dellas, H., & Natal, J.-M. (2008). "DSGE-CH: A DSGE Model for
Switzerland". Swiss National Bank Economic Studies, 5.
[17] Deak, S., Fontagne, L., Maezzoli, M., & Marcellino, M. (2011). "LSM: A DSGE model
for Luxembourg". Economic Modelling, 28, 28622872.
[18] Faccini, R., Millard, S., & Zanetti, F. (2011). "Wage Rigidities in an Estimated DSGE
Model of the UK Labour Market" . Working Paper No. 408. Bank of England.
[19] Fernandez-Villaverde, J. (2010). "The Econometrics of DSGE Models". Journal of the
Spanish Economic Association, 2010(1), 3-49.
[20] Gali, J., Lopez-Salido, J. D., & Valles, J. (2007). "Understanding the Eects of Government
Spending on Consumption". Journal of the European Economic Association, 5, 227270.
[21] Gali, J., Smets, F., & Wouters, R. (2011). "Unemployment in an Estimated New Keynesian
Model". In NBER Macroeconomics Annual 2011, Volume 26. National Bureau of Economic
Research, Inc.
[22] Groth, C., & Khan, H. (2005). "Investment Adjustment Costs: and Empirical Assessment".
Journal of Money, Credit and Banking, 42(8), 1470-1494.
[23] Guerrieri, L. (2006). "The Ination Persistence of Staggered Contracts". Journal of Money,
Credit and Banking, 38(2)(2), 483-494.
[24] Harrison, R., Nikolov, K., Quinn, M., Ramsay, G., Scott, A., & Thomas, R. (2005). The
Bank of England Quarterly Model . Bank of England.
[25] Ireland, P. (2004). "Moneys Role in the Monetary Business Cycle". Journal of Money,
Credit, and Banking, 36(6), 970-983.
[26] Jean, S., & Nicoletti, G. (2002). "Product Market Regulation and Wage Premia in Europe
and North America". ECO WKP(2002)4. OECD.
[27] Jeanls, P., & Burggraeve, K. (2008). "NONAME: A new quarterly model for Belgium".
Economic Modelling, 25, 118127.
[28] King, R., Plosser, C., & Rebelo, S. (1988). "Production,Growth and Business Cycles: I.
The Basic Neoclassical Model". Journal of Monetary Economics, 21(January), 195232.
[29] Kydland, F., & Prescott, E. (1982). "Time to Build and Aggregate Fluctuations". Econo-
metrica, 50, 13451370.
31
[30] Martins, J., Scarpetta, S., & Pilat, D. (1996). "Mark-up Pricing, Market Structure, and
Business Cycle". OECD Economic Studies, 27, 71-106.
[31] Orphanides, A. (2007). "Taylor Rules". Finance and Economics Discussion Series 18. Fed-
eral Reserve Board, Washington, D.C.
[32] Rotemberg, J., & Woodford, M. (1998). "An optimization-based econometric framework
for the evaluation of monetary policy". In B. S. Bernanke, & J. Rotemberg, NBER Macro-
economics Annual 1997 (pp. 297346). Cambridge, MA: MIT Press.
[33] Schmitt-Grohe, S., & Uribe, M. (2003). "Closing small open economy models". Journal of
International Economics, 61(1), 163-85.
[34] Smets, F., & Wouters, R. (2007). "Shocks and Frictions in US Business Cycles". American
Economic Review, 97(3), 586-606.
[35] Woodford, M. (2003). Interest and Prices. Princeton, NJ: Princeton University Press
32
Appendix: Data and Estimation Results
20 40 60
-0.04
-0.02
0
0. 02
0. 04
y
20 40 60
-0.05
0
0. 05
0. 1
0. 15
c
20 40 60
-0.1
-0.05
0
0. 05
0. 1
gy
20 40 60
-0.1
0
0. 1
0. 2
0. 3
inv
20 40 60
-0.04
-0.02
0
0. 02
0. 04
Pi
20 40 60
-0.02
-0.01
0
0. 01
0. 02
l
20 40 60
-0.02
0
0. 02
0. 04
REZ
20 40 60
-0.4
-0.2
0
0. 2
0. 4
xs
Observed Variables
2 4 6 8 10 12 14
x 10
4
5
10
15
20
Interval
2 4 6 8 10 12 14
x 10
4
0
20
40
60
m2
2 4 6 8 10 12 14
x 10
4
0
200
400
600
m3
Multivariate MH Convergence Diagnosis
33
0.05 0.1 0.15 0.2 0.25
0
100
200
SE_ita_zeta
0 0.1 0.2
0
10
20
30
SE_ita_C
0 0.1 0.2
0
10
20
30
SE_ita_l
0 0.1 0.2 0.3
0
10
20
30
SE_ita_Inv
0 0.1 0.2
0
10
20
30
SE_ita_tauC
0 0.1 0.2
0
10
20
30
SE_ita_taul
0 0.1 0.2
0
10
20
30
SE_ita_tauK
0.05 0.1 0.15 0.2 0.25
0
50
100
150
SE_ita_gy
0 0.1 0.2
0
10
20
30
40
SE_ita_risk
Prior and Posterior Distributions
0 0.1 0.2
0
20
40
60
80
SE_ita_miug
0 0.1 0.2
0
10
20
30
SE_ita_mius
0.05 0.1 0.15 0.2 0.25
0
50
100
150
200
SE_ita_pm
0 0.1 0.2
0
10
20
30
SE_ita_pE
0 0.1 0.2
0
20
40
60
80
SE_ita_pYEZ
0 0.1 0.2 0.3
0
10
20
30
SE_ita_pYROW
0 0.1 0.2
0
50
100
SE_ita_PiEZ
0 0.1 0.2 0.3
0
10
20
30
SE_ita_PiROW
0 0.1 0.2
0
10
20
30
SE_ita_RROW
Prior and Posterior Distributions (cont.)
34
0 0.1 0.2
0
20
40
60
SE_ita_YEZ
0 0.1 0.2 0.3
0
10
20
30
SE_ita_YROW
0.05 0.1 0.15 0.2 0.25
0
100
200
300
SE_ita_Ag
0 0.1 0.2
0
20
40
SE_ita_As
0.05 0.1 0.15 0.2 0.25
0
200
400
600
800
SE_ita_REZ
0.2 0.4 0.6 0.8 1
0
2
4
6
8
h
0 0.5 1
0
2
4
6
ksi_w
0 0.5 1
0
1
2
3
4
ksi_g
0 0.5 1
0
1
2
3
4
ksi_s
Figure 6: Prior and Posterior Distributions (cont.)
0 0.5 1
0
1
2
3
4
gamma
0 0.5 1
0
1
2
3
4
gam_zeta
0 0.5 1
0
1
2
3
4
gam_C
0 0.5 1
0
1
2
3
4
gam_l
0 0.2 0.4
0
2
4
6
8
gam_Inv
0 0.2 0.4 0.6 0.8
0
2
4
gam_gy
0.2 0.4 0.6 0.8 1
0
2
4
gam_risk
0 0.2 0.4
0
2
4
6
8
gam_miug
0 0.5 1
0
1
2
3
4
gam_mius
Prior and Posterior Distributions (cont.)
35
0.2 0.4 0.6 0.8 1
0
1
2
3
4
gam_pm
0.2 0.4 0.6 0.8 1
0
1
2
3
4
gam_pE
0.2 0.4 0.6 0.8 1
0
2
4
gam_pYEZ
0 0.5 1
0
1
2
3
4
gam_pYROW
0.2 0.4 0.6 0.8 1
0
1
2
3
4
gam_RROW
0 0.5 1
0
1
2
3
4
gam_YEZ
0 0.5 1
0
1
2
3
4
gam_YROW
0 0.5 1
0
1
2
3
4
gam_tauC
0.2 0.4 0.6 0.8 1 1.2
0
1
2
3
4
gam_taul
Prior and Posterior Distributions (cont.)
0 0.5 1
0
1
2
3
4
gam_tauK
0 0.5 1
0
1
2
3
4
gam_PiEZ
0 0.5 1
0
1
2
3
4
gam_PiROW
0.2 0.4 0.6 0.8 1
0
2
4
gam_Ag
0.2 0.4 0.6 0.8 1
0
1
2
3
4
gam_As
0 0.2 0.4 0.6 0.8
0
2
4
gam_REZ
Prior and Posterior Distributions (cont.)
36
20 40 60
-0.02
0
0. 02
0. 04
ita_C
20 40 60
-2
0
2
4
x 10
-3
ita_l
20 40 60
-1
-0.5
0
0. 5
1
ita_Inv
20 40 60
-0.04
-0.02
0
0. 02
0. 04
ita_miug
20 40 60
-4
-2
0
2
x 10
-3
ita_mius
20 40 60
-0.1
-0.05
0
0. 05
0. 1
ita_risk
20 40 60
-4
-2
0
2
4
x 10
-3
ita_tauC
20 40 60
-1
0
1
2
x 10
-3
ita_taul
20 40 60
-1
-0.5
0
0. 5
1
x 10
-3
ita_tauK
Estimated Innovations
20 40 60
-0.1
-0.05
0
0.05
0.1
ita_gy
20 40 60
-0.02
-0.01
0
0.01
0.02
ita_Ag
20 40 60
-0.1
-0.05
0
0.05
0.1
ita_As
20 40 60
-0.01
-0.005
0
0.005
0.01
ita_zeta
20 40 60
-0.01
0
0.01
0.02
it a_YROW
20 40 60
-0.01
-0.005
0
0.005
0.01
ita_pYROW
20 40 60
-0.02
-0.01
0
0.01
ita_PiROW
20 40 60
-0.2
-0.1
0
0.1
0.2
ita_RROW
20 40 60
-0.1
-0.05
0
0.05
0.1
ita_YEZ
Estimated Innovations (cont.)
37
10 20 30 40 50 60
-0.04
-0.02
0
0.02
ita_pm
10 20 30 40 50 60
-0.04
-0.02
0
0.02
ita_pYEZ
10 20 30 40 50 60
-0.02
-0.01
0
0.01
0.02
ita_PiEZ
10 20 30 40 50 60
-0.02
-0.01
0
0.01
ita_REZ
10 20 30 40 50 60
-4
-2
0
2
4
x 10
-3
ita_pE
Estimated Innovations (cont.)
Sources of Fluctuations in Private Consumption - Historical Decompositions of Shocks
Sources of Fluctuations in Investment - Historical Decompositions of Shocks
38

You might also like