55 
6 
UDK: 336.74:338.26(497.5:497.4) 
Članek poskuša pojasniti vzročno 
odvisnost med izvozom in realnim 
deviznim tečajem. V članku avtor razvije 
enostaven ekonomski model, ki temelji 
na odnosu med izvozom in razliko v 
obrestnih merah, povpraševanjem iz 
tujine po domačem blagu in realnim 
deviznim tečajem. Model je empirično 
preverjen za Slovenijo in Hrvaško. 
Empirična analiza kaže, da je odvisnost 
med razlikami v realnih obrestnih 
merah, tujim povpraševanjem in 
izvozom, če sploh obstaja, zelo šibka. 
Hkrati pa obstaja močna odvisnost med 
izvozom in realnim deviznim tečajem. 
Ključna ugotovitev je, da je bila 
slovenska politika drsečega deviznega 
tečaja bistveno uspešnejša od politike 
trdnega deviznega tečaja na Hrvaškem. 
Ključne besede: realni tečaj, denarna 
politika, izvoz
I z v l e č e k 
JEL: F41, E42, O24 
Neven Vidaković, BA 
Privredna Banka Zagreb and 
Business School »Libertas« 
Empirična analiza denarne politike na Hrvaškem in 
v Sloveniji 
EMPIRICAL ANALYSIS OF MONETARY POLICY: 
CROATIA VS. SLOVENIA 
VIDAKOVIĆ: EMPIRICAL ANALYSIS OF MONETARY POLICY: CROATIA VS. SLOVENIA 
UDC: 336.74:338.26(497.5:497.4) 
This paper tries to explain empirical causation 
between exports and the real exchange 
rate. The paper develops a 
simple model based on the relationship 
of exports to real interest rate differential, 
foreign demand for domestic goods, 
and real exchange rate. The paper then 
tests the model with empirical data from 
Croatia and Slovenia. The empirical 
analysis in the paper finds that there is 
a very limited relationship between real 
interest rate differentials, foreign demand 
and exports. However, there is a 
strong relationship between exports and 
the real exchange rate. When empirically 
tested, the model confirms that the 
monetary policy of sliding exchange rate 
in Slovenia was vastly superior to the 
monetary policy of fixed exchange rate 
in Croatia. 
Key words: real exchange rate, monetary 
policy, exports
A b s t r a c t 
1 The whole process of transition and monetary adjustment is still not over in all countries 
of ex-Yugoslavia. 
1 Introduction 
When Yugoslavia fell apart, what was once a big closed economy became 
five small open economies. Once the military operations and initial instances of 
hyperinflation stopped (hyperinflation in Croatia ended in 1994), the succeeding 
countries started to build their own economies.1 
The process of transformation from a closed socialist system to an open 
economy was complex and challenging. Even fifteen years after the fall of 
communism, some ex-Yugoslavian republics, such as Bosnia or Serbia and 
Montenegro, have not made much headway in the transition process. Others, like 
Slovenia, are economically closer to western European countries than to other 
transition countries. The explanation of these differences presents a challenging 
task for any economics research paper. 
This paper does not look at all of the aspects of the transition process; it only 
looks at the results of the monetary policies implemented. In effect, the paper is 
empirically evaluating two different monetary policies implemented, one in Croatia 
and the other in Slovenia, with a special focus on exports. 
But before we look at the specific history, we should look at the big picture. 
The main economic problems for the newly formed countries can be formulated 
in the following way: 
a) How to transition from the socialist fiscal policy (the government owns 
everything) to the capitalist fiscal policies of limited (if any) government 
involvement? This process had to be done, at least in theory, with the minimum 
of social cost and maximum social benefit. 
b) How to formulate a monetary policy and set up a central monetary authority 
with clearly defined objectives and methods? The main problem with the setup 
of the monetary policy was the choice of the optimal monetary policy tool. 
This paper does not look into the fiscal policy, privatization, or any other kind 
of development of the capitalistic free market economies that countries of ex- 
Yugoslavia have undertaken. Instead, this paper takes a look at two opposing 
monetary regimes (Slovenia and Croatia) and tries to create, on an empirical 
level, a study of the real effects of the monetary policy choices. 
The fundamental question about monetary policy is the issue of inflation, 
more specifically the control of inflation. In small open economies with free 
fluctuation of capital, the control of inflation translates into the control of 
expectations, as presented in Sargent (1992). 
The newly founded countries lived under closed socialistic systems with fixed 
exchange rates (or exchange rates strictly determined by the National Bank of 
Yugoslavia). Due to the high inflation in the 1980s and the general instability of 
the dinar, most people in ex-Yugoslavia preferred to keep their income and savings
56 
NG, ŠT. 1–2/2007 IZVIRNI ZNANSTVENI ČLANKI/ORIGINAL SCIENTIFIC PAPERS 
in foreign currency, mostly Deutsch marks. Unlike other exsocialist 
countries which had very limited access to hard 
currency, Yugoslavia was a special case. There was a constant 
inflow of hard currency from tourism and from Yugoslavs 
who worked in Western Europe. 
When Yugoslavia fell apart, the main question for both 
the government and the people was: what to do with the 
exchange rate? 
Soon two opposing schools developed. One school of 
thought considered it the best to keep the exchange rate fixed. 
The main argument was that through a stable or fixed 
exchange rate, the central bank would have an easy 
mechanism for control of the quantity of money in the 
economy and, as such, inflation would be contained. In 
addition, perceptions and expectations of a stable exchange 
rate would create an expectation of a stable currency with 
low inflation. This argument was very important in Croatia, 
where after the war there was a period of hyperinflation, as 
described in Rohatinski et al (1995). 
The second school wanted to constantly depreciate the 
exchange rate, or implement a “sliding” exchange rate. It 
was believed that such policies of constant depreciation 
would weaken the national currency and stimulate exports, 
an important part of GDP. For a small open economy, a (real) 
depreciation and growth of exports can be a considerable 
source of economic growth. From a mathematical stand 
point, this policy can also be explained as the dynamic 
programming problem of the real exchange rate as presented 
in Vidaković (2005b). 
In essence, a constant real exchange rate depreciation 
policy was a mercantile “beggar thy neighbor” type policy, 
but with one major advantage: there was no concern about 
counterparty depreciation or a trade war. The countries of 
the EU, where most of Croatian and Slovenian exports went, 
would never just depreciate their currencies as a response to 
Croatian or Slovenian monetary policy, especially with the 
Maastricht Treaty and the process of building a currency 
union. On the other hand, the WTO prevents any tariff and 
customs impositions, but does not stipulate anything about 
monetary policy. 
So from a game theory point of view, it was possible to 
implement either monetary policy. It was a unique period in 
time, when two countries had to decide which path to take. 
The choice was between a path of stability and security (fixed 
exchange rate); or a path of uncertainty, but with a possible 
large payoff.2 Many of these dilemmas were presented in 
Ribnikar (1999). 
The main problem with the exchange rate for both schools 
was the determination of the transmission of the exchange 
rate into inflation, meaning: will there be some real 
depreciation? And is depreciation going to translate into the 
inflation one-for–one, or will there be an overshooting effect? 
If overshooting was the case, then the argument for the 
sliding exchange rate would be void. But today we have an 
opportunity to empirically see the results of two different 
policies. 
This paper looks at a natural experiment which has 
occurred in real life. The two “test subjects” are Slovenia 
and Croatia, two similar countries which have taken two 
different paths. The most interesting aspect of this paper is 
not a development of a model in order to artificially test the 
model’s assumptions. Rather, the most interesting aspect of 
this paper is the explanation of the results of monetary 
policies in the last 10 years. 
One policy was applied in Slovenia, a sliding exchange 
rate. It was based on a simple rule: depreciate the exchange 
rate at the same rate as inflation or higher. The exchange rate 
was depreciated slowly over time. In essence, it was vector 
targeting of the real exchange rate, as we shall see later. The 
path of the vector was set in order to achieve two goals: 
a) Depreciate the real exchange rate and thus constantly 
make Slovenian products more competitive in the world 
market. 
b) Prevent large capital inflows in Slovenia and thus 
decrease foreign debt. 
The Croatian strategy was different. After the war in 
Croatia, there was a period of hyperinflation. At one point, 
inflation was at 30% per month. Then in 1994 came the 
stabilization program that ended inflation. In order to keep 
inflation under control, the HNB (Hrvatska Narodna Banka, 
Croatian National Bank) decided to keep the exchange rate 
approximately the same. HNB stated that its purpose was to 
keep the oscillation of the exchange rate to a minimum, 
reacting to shocks to the exchange rate. This in effect made 
the kuna exchange rate a mean reverting series, with 
oscillations up to 3% from the mean. So the only goal of the 
monetary policy was the stability of the exchange rate. 
Monetary policy did not concern itself with exports, imports, 
foreign debt or any other economic variables, just the stability 
of the exchange rate. 
The model in this paper is based on a simple dynamic 
programming optimization of export function in a small open 
economy. Special attention is devoted to the analysis of the 
real exchange rate and the impact of the real exchange rate 
on exports. The model is tested empirically, comparing the 
results of the two monetary policies. 
The paper is organized as follows: part two develops the 
model, part three compares the model and the empirical data 
with some interesting results, and part four concludes. 
2 The model 
When the central banks of Slovenia and Croatia were 
determining their respective policies regarding the exchange 
rate, the literature on the real exchange rate and open 
economies was relatively small. Up to that time, the most 
important research done on open economies was Robert 
2 In essence, we have an Elsberg paradox as presented in Sargent 
and Hansen (2000) as a choice between known and unknown 
distribution.
57 
Mundell’s (1968) seminal work, and research done by Rudy 
Dornbush (1988). Furthermore, some computational 
economics techniques (forward looking rational expectations 
models3) had not been developed at that point in time, and 
computer power was low. 
Today the state of economic theory regarding open 
economies is unrecognizable from the state of economic 
theory fifteen years ago. In the last fifteen years, there have 
been many successful attempts in the creation of a working 
small open economy model, most notably the efforts of Gali 
et al (2005), Ball (1999), and Rogoff and Obstfeld (2000, 
2002). There have also been huge advances in the 
development of computational techniques as presented in 
the works of Ljungqvist and Sargent (2004) and Hansen and 
Sargent (2006). 
After the Asian, Mexican, Russian and Argentinean 
currency crises, it became apparent that economics as a 
science does not fully understand the workings of the 
mechanisms involving the real exchange rate and the 
transition from fixed exchange rate to flexible exchange rate. 
Recently the research of Aizenman and Glick (2005) studies 
the behavior of fixed exchange rates and the transition from 
a fixed exchange rate to a flexible exchange rate. Aizenman 
and Glick analyzed the cost of a switch from a fixed regime 
exchange to a flexible regime exchange rate and came to a 
stunning conclusion. Out of the 63 instances of currency 
regime switch, 32 were considered disorderly. However, the 
duration of the regime plays a considerable role. There were 
20 cases where a fixed exchange regime was longer then 
200 months. Out of the 20 instances of regime switching, 16 
had a negative rate of growth and a fall in real output once 
the move to the flexible exchange regime was made, 4 had a 
positive rate of real output, and one instance was neutral. 
These results offer a powerful empirical argument against a 
fixed exchange rate regime,4 especially in the long run. The 
conclusion of Aizenman and Glick’s paper can be summed 
up as follows: the longer the fixed exchange rate regime, the 
larger is the cost of switching to the flexible regime. 
There are several models that try to portray the behavior 
of a small open economy. One has been presented in Ball 
(1999), and many have been presented in the works of 
Obstfeld and Rogoff. In this paper we will work with a 
modification of the Mundell-Fleming model explained in 
Vidakovic (2005a). This paper is in effect the continuation 
of the theoretical base made in Vidakovic (2005a and 2005b). 
The main model can be shown as follows: 
y= c + s + g + (ex – im) (1) 
e = c + I + g + (f – d) (2) 
c = c°(-r,y) + A + c(y,- r, -E/P, W/P); A= z + .(y, E/P) (3) 
I = i(-r)° + i(CM) – iA*(CM, E/P) = i° + p (4) 
im = c(y) + im° ± .(CM, E/P) = im° + m (5) 
ex = ex°(E/P) + A*(y*) ± . (CM, E/P) = ex° + x (6) 
g= g° (7) 
y – GDP 
y* – GDP of the rest of the world 
e – expenditures 
c – consumption 
s – savings 
g – government expenditures 
ex – exports 
im – imports 
I – investments 
r – real interest rates 
f – Croatian investment in the rest of the world 
d – foreign investment in Croatia 
c° – autonomous consumption 
A – Croatian demand for imports 
W – wages 
P – price level 
CM– world interest rates 
g° – autonomous government expenditures 
i° – autonomous investments 
ex° – autonomous exports 
im° – autonomous imports 
A* – foreign demand for Croatian imports 
From the above equations, we can get the IS curve for a 
small open economy: 
y = c°(-r,y) + c(y,-r, -E/P, W/P) + i°(-r) + g° + 
+ im° + ex° + [ A + p + m + x] (8) 
For the LM curve, we will use a standard Keynesian LM 
function: 
l 
hy x l 
r 
- - . 
= (9) 
Once we graph the two equations, we will get a static 
Keynesian small open economy model: 
VIDAKOVIĆ: EMPIRICAL ANALYSIS OF MONETARY POLICY: CROATIA VS. SLOVENIA 
3 Here the author is mostly considering the rational expectations 
model presented in Sargent and Hansen (2006) 
4 Naturally this argument does not hold if a country has a fixed 
exchange rate in order to enter a currency union. 
This model, in effect, is the Mundell-Fleming version of 
the IS-LM closed economy model. It is a standard Keynesian 
static macroeconomic model. It is assumed that the labor 
market clears and that all changes are instantaneous. The
58 
expectations do not enter into the model and time variant 
processes do not play any role, so there are no lag operators. 
Obviously this kind of model is not an appropriate or 
modern tool in today’s economic theory. Today we have 
mathematical tools to develop more dynamic models, so we 
shall work with an updated version of the model. 
The theoretical background from Vidaković (2005b) tries 
to create an optimization of the export function. The basis 
for this optimization is the mathematics of dynamic 
programming. The central focus of Vidaković (2005b) is on 
the functions: 
im = c(y) + im° ± .(CM, E/P) = im° + m (10) 
ex = ex°(E/P) + A*(y*) ± .(CM, E/P) = ex° + x (11) 
These two functions represent the static functions for 
exports and imports in a small open economy. Previous work 
has theoretically transformed these functions from static to 
dynamic forms. This transformation has ushered in a model 
able to give optimal dynamic account of the export function 
for a small open economy. 
2.1 Dynamic model of exports. 
In order to develop the model, we will have to make 
some assumptions about the model. The first assumption 
will be that there are n households in economy. Although 
the number of households is large, there is one representative 
agent that resembles the rest of the households. The main 
problem the household faces is how to maximize utility over 
an infinite period of time. Utility comes from consumption: 
dt 
c 
t .. 
- 
- 
- 
. . 
.
. 
. . 
.
. 
0 - 
1 1 
1 
max 
. 
. 
.
. 
ß (12) 
Subject to: 
.
- 
- - 
= - + + 
1 
0 
1 (1 ) 
t 
i 
t i 
t t t t i 
c w sw . r sw 
0 < ß < 1 
where w is the wage the household receives, s is constant 
savings rate, and . is the portion of savings households decide 
to liquidate in period t, 0 . . . 1. This number is stochastic 
for every time period. Function 12 gives us the optimization 
problem for the household given its consumption. But since 
we are dealing with a small open economy it is necessary to 
define consumption. 
The household has the opportunity to consume two kinds 
of goods, domestic goods and foreign goods. Equation 13 
defines such total consumption: 
h d f 
c = (1-.)c +.c (13) 
Where . represents the fraction of goods consumed, 
totaling 1; subscript d represents the domestic goods and 
subscript f represents the foreign goods; and c is the vector 
of goods consumed. 
If we take the utility function as: 
. 
. 
.
. 
- 
- 
. . 
.
. 
. . 
.
. 
- 
= 
1 1 
1 
( ) 
c 
u c
h (14) 
Combining equations 13 and 14 we get: 
( ) . 
. 
. 
.
. . - 
- 
. . 
.
. 
. . 
.
. 
-
- + 
= 
1 1 
1 
(1 ) 
( ) d f 
h 
c c 
u c (15) 
The cost of consumption can be defined as follows: 
max =. +.
n 
i i f 
m 
i i d 
P c c 
0 
, 
0 
, . (. *. ) (16) 
Subject to .
- 
- - 
= - + + 
1 
0 
1 (1 ) 
t 
i 
t i 
t t t t i 
c w sw . r sw , so c=P 
for every time period. 
The price (P) spent on goods is the sum total of prices 
paid for domestic goods plus prices paid for foreign goods; 
. in period t represents the exchange rate; and . is the price 
of the ith good. 
2.2 Factors that have an impact on exports 
In order to better understand the optimization of the 
export process, we have to define and investigate what factors 
have an impact on exports. 
According the Mundell-Fleming model presented at the 
beginning of the paper, the most important factors affecting 
exports in a small open economy are: 
1. Real exchange rate. 
2. Capital mobility, caused by the real interest rate differential. 
3. The economic condition of the country where most of 
the exports go. 
The reasons for naming each of these variables should 
be obvious. The real exchange rate presents the true value 
of the goods; capital mobility (real interest rate) will serve 
as an equalizer for the marginal productivity of capital; and 
the economic condition of the country that is importing the 
goods from a small open economy represents the demand 
factor for export products. 
The author’s main interest is the analysis of the time 
vector for each of the three variables presented. Some 
assumptions are in order. The first one is the assumption of 
rational expectations as presented in Muth (1961) and Lucas 
(1972). The second assumption is the assumption of time 
optimization consistent with the set-up of a dynamic 
programming problem as presented in Stokey and Lucas 
(1989) and Adda and Cooper (2003). The third assumption 
is the assumption of perfect substitution between domestic 
and foreign goods. This assumption states that a 
representative household is indifferent between the domestic 
and foreign good as long as the real price is the same. In 
case the domestic currency is undergoing a real appreciation 
over time, consumers will start to substitute for the cheaper 
NG, ŠT. 1–2/2007 IZVIRNI ZNANSTVENI ČLANKI/ORIGINAL SCIENTIFIC PAPERS
59 
foreign goods. This changes the relative values of the 
distribution of weight between domestic and foreign goods. 
The reader should notice here that there is no mention of 
the nominal exchange rate. This is a very important aspect 
of this paper. In essence the author of this paper is arguing 
that under the above assumptions the nominal exchange rate 
is irrelevant. 
Extrapolating from the argument of Lucas (1972), we see 
that any kind of announced exchange rate movements will 
not have any kind of real effect on the customer’s preferences. 
The nominal exchange rate in this paper is considered only 
in relation to the real exchange rate. The effect of changes in 
nominal exchange rate is understood to be neutral in the 
model. However, this assumption of neutrality of the nominal 
exchange rate is tested in the appendix. 
Capital mobility is an important factor for the standard 
of a small open economy. Better real return on capital 
invested can mean movement of jobs from one country to 
the next one, thus increasing employment and overall 
standard of living in the economy. 
With perfect capital mobility, as presented in the model, 
capital will move to countries where it can obtain a greater 
real interest rate return. The movement of capital in essence 
is the capital account balance. In case a country has large 
and persistent trade deficits, it will be necessary to finance 
those deficits. For a country to do that, it has to allow a 
counter balance in the capital account to offset the current 
account deficit. This can only be done by selling goods or 
by selling labor. For the scenarios of trade deficits in the 
case of Croatia, see Vidaković (2005a). 
The third factor is the demand for export goods, or we 
can say the current economic state of the country where the 
exports go. The rationale for this is simple: if we are 
exporting in a country with high economic growth, even if 
the percentage of the market held by the exporting country 
stays the same, economic growth will cause imports to 
increase in nominal quantities, although the percentage in 
the market might stay the same. This is the exogenous 
variable in the model. 
2.3 Real exchange index 
The following part of this paper develops an index of 
the real exchange rate and runs simple OLS regressions in 
order to establish which of the above noted variables have 
an impact on the exports and to what extent. The purpose of 
the OLS regression is not to be used as a forecasting 
mechanism, but to confirm or deny causation and connection 
between the variables. There will also be a separate test for 
the nominal exchange rate. 
First we have to create an index of the real exchange 
rate for Croatia and Slovenia. The notation for the data used 
and the process of index creation are in the data appendix at 
the end of the paper. 
The main purpose of the index is to show in a simple 
and straightforward manner the movement of the real 
exchange though time. 
The index of the real exchange in the model is: 
( ) . 
- + 
. = . 
t 
im 
t 
ex 
t 
t 
e e 0 1 
1 
(17) 
. - constant, the beginning value of index 1994 = 100. 
eex - price change in Croatia or Slovenia (percentage change 
or inflation), plus exchange rate appreciation, minus 
the exchange rate depreciation in the period. 
eim - world inflation, in this case inflation in the EU. 
The index created here is very simple. If the index is going 
down, this means that the prices in the domestic country are 
increasing more then the prices in the rest of the world. This 
means that the real exchange rate is appreciating. Domestic 
goods are more expensive, foreign goods are cheaper, and 
the substitution effect takes place. On the other hand, if the 
index is going up, the prices in the rest of the world are 
increasing faster then the prices in the domestic country, the 
real exchange rate is depreciating, and households will start 
to substitute foreign goods for domestic goods. 
According to the basic theory, the decrease in this index 
should have a negative effect on exports in a small open 
economy. 
3 Empirical testing 
Before we start the regression analysis, let us look at 
the nominal exchange rate for a ten-year period for Croatia 
and Slovenia. The exchange rate used is the exchange rate 
of the HRK vs. the euro and the tolar vs. the euro in the 
period 01/95-01/05. 
As seen in the graph, it is clear that the kuna exchange 
rate has been in a very narrow range from the period of mid- 
1998 until today. The mean for the whole period is 7.28, 
with a standard deviation of 0.33. The minimum in the series 
is 6.61 and the maximum is 7.73. The lower bound is 0.67 
kuna away from the mean of about 9%. The upper bound is 
0.45 kuna away from the mean of 6%, essentially indicating 
an upward resistance. 
The graph shows that since the beginning of 1999 the 
kuna has been heavily controlled. The exchange rate is not 
fixed, but it has been kept in a very narrow band. Now given 
the theory of the model, such an economic behavior should 
be negative for the exports in Croatia if there exists a 
considerable price differential between Croatia and the 
country that imports Croatian goods. 
The same analysis can be done with the tolar. The mean 
of the series is 201.783. The highest point is 239.99 and the 
minimum of the series is 150.78. It should be noted that the 
minimum value occurs at the beginning of the series, and 
the highest value occurs towards the end of the series. 
It can be clearly seen that there is a persistent nominal 
depreciation of the exchange rate of the tolar vs. the euro. 
The first observation from these two graphs should be the 
way the graphs look. The kuna is a straight line, while the 
tolar is almost a linear function with a steady slope. The slope 
VIDAKOVIĆ: EMPIRICAL ANALYSIS OF MONETARY POLICY: CROATIA VS. SLOVENIA
60 
ends and levels off once Slovenia entered the EU and had to 
stabilize the exchange rate in order to prepare for the EMU. 
Let us now look at the real exchange index. The real 
exchange rate for Croatia is: 
As the graphs show, there has been a double appreciation. 
Over time both countries have experienced downward 
movement in the real exchange rate. Similar results can be 
found in Flere (2004) and Coricelli and Jazbec (2004). 
Here are the actual values for the index according to the 
author’s calculations from Equation 17: 
The Slovenian rate fell by 14%, and the Croatian rate 
fell by 25% from their peaks in 1995. Now given the fact 
that the index starts with 1994=100, there are several things 
that have to be said before we analyze the index. 
1. Slovenian policy has been constant. Monetary policy has 
attempted to prevent a drop in the real exchange rate. 
On average in the last ten years, the index depreciated 
less then 1% each year. 
2. The initial jump for Croatia in 1995 can be interpreted 
as an after effect of the hyperinflation period more than 
the effect of a radical shift in monetary policy. 
3. We can see two distinct shocks in the index in 1998 when 
the tolar and kuna dramatically appreciated. 
4. There was a switch in policy in Croatia. From 1994 to 
1996 the index is about even, but from 1997 the index 
appreciates dramatically until 2002, when it stabilizes. 
5. In the same time span (1995-2004), the index in Slovenia 
dropped by 13.44 points. In Croatia in the same time 
period, the index dropped by 27.91 points, more than 
twice the move in Slovenia. 
From this data we can conclude that the shock of the 
real exchange rate on exports should be much more adverse 
on Croatian exports then on Slovenian exports in the same 
NG, ŠT. 1–2/2007 IZVIRNI ZNANSTVENI ČLANKI/ORIGINAL SCIENTIFIC PAPERS 
Source: Author’s calculations 
Graph 1: Nominal exchange rate, Croatia 
Source: Author’s calculations 
Graph 2: Nominal exchange rate, Slovenia 
Source: Author’s calculation 
Table 1: Real exchange rate index
61 
time period, under the assumption that real exchange rate 
has an effect on exports. This assumption still has to be 
proven empirically. 
Going back to Equation 15, we see utility is based on 
consumption, and from equation 16 we see that if domestic 
goods are becoming cheaper, households will substitute 
foreign goods for domestic goods. From the results in the 
real exchange rate movement, we can conclude the 
substitution effect in Croatia should be much more adverse 
to domestic production than in Slovenia. 
3.1 Regressions 
Now we will run regressions on several variables to see 
the overall impact of real exchange rate, interest rate 
differential, and foreign demand for exports. We will be 
dealing with the basic one step ordinary least squares 
regression of the form: 
ex = . + ß1. + ß2. + ß3. 
. - log value A* 
. - log value . 
. - log value . 
The purpose of this regression is not to be able to predict 
the future movement of the real exchange rate or to serve as 
a model. Much better and more accurate results for prediction 
can be obtained using VAR methods as in Echebaum and 
Christiano (2005) or a recently developed FAVAR procedure 
as proposed by Bernanke et al (2005). Rather OLS 
regressions here are being used to be able to determine with 
statistical significance which of the variables have an impact 
on exports and which variables do not have impact on 
exports. The three main variables that we are focusing on 
are: real exchange rate (as presented in the index above), 
interest rate differential, and growth in the country where 
Croatia and Slovenia export, in this case growth in the EU. 
The regression results here are for Croatia. The regression 
results for Slovenia are in the appendix. 
The most striking result of regression was the fact that 
the interest rate differential and growth in the EU were not 
statistically significant. 
ex = 10.37 – 5.72. – 4.188. + 0.536. 
(13.32) (-0.52) (-1.01) (3.59) 
The t-statistic values are in brackets. 
As we can see, neither interest rate differential nor growth 
is significant. The only statistically significant element of 
the equation is the real exchange rate index. From this the 
conclusion follows that in order to increase exports, a country 
should depreciate the real exchange rate. 
VIDAKOVIĆ: EMPIRICAL ANALYSIS OF MONETARY POLICY: CROATIA VS. SLOVENIA 
Source: Author’s calculations 
Graph 3: Real exchange rate, Croatia 
Source: Author’s calculations 
Graph 4: Real exchange rate, Slovenia
62
There is also a possible problem with the small sample 
size, but since the time period considered is small, the 40 
observations are the only observations possible. 
From this regression we can move to the single factor 
regression, and regressing only the values of exports on the 
index of real exchange rate we get: 
ex = 10.1625 + 0.5514. 
(13.84) (3.79) 
Again the real exchange rate index is extremely 
significant and confirms the assumptions of the model. 
In fact it is very significant and R square is at 26%, a 
relatively high R square for only a one variable regression. 
Also keep in mind that these are log values, so a positive 
sign in front of â makes sense. As the index goes up by 1 
percent, the exports will go up by 0.55 percent. 
The results for Slovenia are very similar to those obtained 
for Croatia and are shown in the appendix with the rest of 
the results. 
3.2 Discussion 
The testing of data rejects the theoretical implications in 
Vidakovic (2005b). Vidakovic (2005b) states that there 
should be three main variables for the stimulation of exports; 
however, empirical testing rejects that hypothesis and leaves 
us with only one variable: real exchange rate. 
The only realistic policy for a small open economy is to 
keep depreciating the real exchange rate. Through real 
depreciation of the exchange rate, a small open economy in 
essence forces households to substitute domestic for foreign 
goods. 
But there are two problems with this logic: 
1. The first problem is the fact that the real exchange rate 
cannot be constantly depreciated. The reason for this is 
the autonomous imports. We are dealing with a small open 
economy and there are some imports that a small open 
economy needs in order to function properly, and there 
are some goods where the substitution effect is impossible. 
The autonomous imports are part of eEquation 5. 
2. The second reason is the nature of the problem. It is not 
possible to set a goal for the depreciation of the real exchange 
rate forever. Such constant depreciation of real exchange 
rate in a linear or on an exponential (by constant percentage) 
basis seems highly implausible in the real world. 
But there is a solution to the problem. A small open 
economy cannot constantly depreciate its currency ad 
infinitum, but a small open economy can optimize the real 
depreciation/appreciation, which we have empirically seen 
in the case of Slovenia. The real exchange index is 
appreciating, but through proper monetary policy the real 
depreciation has been slowed and put under control. On the 
other hand, in the case of Croatia we are seeing a lack of 
defined monetary policy. There is only preservation of the 
status quo: a fixed exchange rate no matter what the cost. In 
Croatia’s case, the real exchange rate has been left to drift 
widely, and due to the considerable interest rate differential, 
this has caused massive foreign debt and a huge current 
account deficit. 
3.3 Solution 
As we can see from the regressions, the most important 
factor for growth in imports is the real exchange rate. So in 
order for an economy to have growth in exports, it is 
necessary to optimize the real exchange rate. 
However, the real exchange rate is not just a variable 
that can be easily changed. If we look at equation 17, we see 
that prices in the exporting country are beyond our control. 
In essence, controlling the real exchange rate is a stochastic 
dynamic programming problem. Such a problem can be 
represented in the value equation 18. 
( ) max ( ) ( ') ' V ex u c E V ex 
ex ex = + ß (18) 
Empirically, if we look at the exchange rate indices for 
Croatia and Slovenia, we see that the real exchange index 
for Slovenia is a solution to Equation 18.5 Although the index 
is not moving in the desired direction (the index is 
appreciating instead of depreciating), we see that the index 
is smooth with very small volatility. Croatia’s real exchange, 
on the other hand, is wild and volatile. 
3.4 Results 
This is an empirically oriented research paper and now we 
shall look at the empirical results of the monetary policy chosen. 
Looking at the model, the main prediction is that due to 
oscillations in the real exchange rate, a substitution effect will 
take place. If the real exchange rate is appreciating, domestic 
goods will increase in price and force rational households to 
substitute domestic for foreign goods. This substitution effect 
will cause large persistent trade deficits and in case there is are 
no alternate domestic means to finance the deficit, the country 
will have a large increase in foreign debt. 
From the data we have seen on real exchange rate, the 
model tells us that Slovenia should have a small trade deficit 
and small foreign debt. On the other hand, Croatia, according 
to the model, should have a large trade deficit and large 
foreign debt. Now we shall look at the data and see if the 
model’s predictions are correct. 
In the last ten years, Croatia has undergone a persistent 
trade deficit and foreign debt has exploded, as can be seen 
from the data below. During the same time period, Slovenia 
has had a stable balance of payment and relatively benign 
foreign debt. 
Thus, the model is absolutely correct in its predictions 
about the effect of the real exchange rate on the trade balance 
and foreign debt. 
As we see, the empirical data has shown the wrong 
orientation of Croatian monetary policy and the correct 
NG, ŠT. 1–2/2007 IZVIRNI ZNANSTVENI ČLANKI/ORIGINAL SCIENTIFIC PAPERS 
5 By solution, the author here means smoothness of the curve 
and small standard deviation of the index, not the mathematical 
properties of the dynamic programming value equation solution.
63 
orientation of Slovenian monetary policy. The wrong choice 
of monetary policy has caused Croatia to have a large debt 
and persistently inept economy; on the other hand, such 
economic doom has been averted in the case of Slovenia. 
4 Conclusion 
The purpose of this paper was to look at the development 
of the monetary policies in two countries: Slovenia and Croatia. 
Instead of just a comparative analysis of the economies, the 
paper tests a model based on the dynamic optimization of the 
real exchange rate. Once the model was developed, the 
empirical analysis brings two points to light. Foreign demand 
for domestic goods and the real interest rate differential do not 
play a major role in the level of exports for a small open 
economy. Thus, the two variables predicted to be important by 
the model are dismissed. The only variable left to be statistically 
important for the level of exports in a small open economy is 
the real exchange rate. Simple OLS regression confirms that 
there is a strong statistical relationship between exports and 
real exchange rate as the model originally predicted. 
VIDAKOVIĆ: EMPIRICAL ANALYSIS OF MONETARY POLICY: CROATIA VS. SLOVENIA 
After statistical tests, the empirical data is analyzed and 
empirical data fully confirms the model. A strong real 
currency with the tendency for appreciation will cause a large 
trade imbalance and persistent growth of foreign debt. As 
we see in the data for Croatia, a constantly appreciating 
currency in real terms has caused a large and ever increasing 
trade deficit and exponential growth of foreign debt. 
If we look at the research paper as an empirical study, it 
is not very hard to conclude which policy turned out to be 
correct and what the cost (benefit) was of the policy chosen. 
Appendix 
Here are the results for some computations run and 
mentioned in the text. 
We see when we run regression with the Slovenian data 
that only the real exchange index is significant. What is 
strange is the fact that the real exchange index has a negative 
number. This implies that a fall in the real exchange index 
(real appreciation) will cause exports to go up; economically, 
Source: Slovenian Central Bank 
Table 2: Economic Indicators, Slovenia 
Source: Croatian Central Bank 
Table 3: Economic Indicators, Croatia 
6 In millions of tolars, constant prices. 
7 In millions of euros 
8 As of January 2006, foreign debt is 25.5 billion euros, around 
87% of GDP. 
Source: Author’s calculations 
Table 4: Regression results for Slovenia
64 
this does not make any sense. This can be explained in two 
ways: 
1. Statistically the regression is correct. In the time period 
tested, the index is falling and exports are rising, so the 
regression coefficient should be negative. 
2. The nominal value of the exchange rate is important, 
although in the paper we have assumed households only 
care about the real value of variables, not just nominal 
values. 
Running a one variable regression, we again obtain a 
negative coefficient on the long index, but this time the 
coefficient is much smaller (-0.39 versus -2.16 from the 
previous regression). Also in this regression the t-statistic is 
much larger. 
The next regression is the regression of nominal value 
of exports on the nominal value of the exchange rate. As we 
can see, regression is statistically significant and the values 
of the t statistic are extremely large in the case of Slovenia, 
but in the case of Croatia we do not get this result. 
In the case of Croatia, the same regression is not valid, 
the t statistic is small, and the nominal exchange rate is 
insignificant even at the 10% level test. The p value is 13%. 
So we can conclude that the nominal exchange rate and 
exports are not correlated. This is an extremely powerful 
conclusion that completely supports the main argument of 
the paper: only the real exchange rate matters. 
What these two regressions tell us is that Slovenian 
exports are growing not because the exchange rate is falling, 
but because the real exchange rate is being controlled through 
depreciation. On the other hand, in Croatia there is no 
exchange rate movement due to any factor, so the relationship 
between the exchange rate and exports is not statistically 
significant. 
Source: Author’s calculations 
Table 4: Regression results for Slovenia 
Source: Author’s calculations 
Table 5: Regression results for Slovenia 
Source: Author’s calculations 
Table 6: Regression results for Croatia 
NG, ŠT. 1–2/2007 IZVIRNI ZNANSTVENI ČLANKI/ORIGINAL SCIENTIFIC PAPERS
65 
Bibliography 
1. Adda, Jerome and Cooper, Russell (2003). Dynamic 
Economics: Quantitative Methods and Applications. MIT 
Press. 
2. Aizenman, Joshua and Glick, Reuven (2005). Pegged 
Exchange Rate Regimes - A Trap? NBER Working Paper 
w11652. 
3. Ball, Laurence (1999). Policy Rules for Open Economies. 
In Monetary Policy Rules, ed. John Taylor. University 
of Chicago Press. 
4. Ben S. Bernanke, Jean Boivin and Piotr Eliasz (2005). 
Measuring the Effects of Monetary Policy: A Factor- 
Augmented Vector Autoregressive (FAVAR) Approach. 
Quarterly Journal of Economics 120 (1): 387-422. 
5. Christiano, Lawrence, Eichenbaum, Martin and 
Vigfusson, Robert (2005). Assessing Structural VARs. 
Working paper, Northwestern University. 
6. Coricelli, Fabrizio and Jazbec, Bostjan (2004). Real 
exchange rate dynamics in transition economies. Structural 
Change and Economic Dynamics 15 (1): 83-100. 
7. Dornbush, Rudy (1988). Open Economy 
Macroeconomics: New Directions. In Misalignment of 
Exchange Rates, ed T.R. Marston , University of Chicago 
Press. 
8. Flere, Luka (2004). Realni Efektivni Devizni Tečaj 
Tolarja in Konkurenčnost. Prikazi in analize XII/1. 
Central Bank of Slovenia Ljubljana. 
9. Gali, Jordi and Monacelli, Tommaso (2005). Monetary 
Policy and Exchange Rate Volatility in a Small Open 
Economy. Review of Economic Studies, 72 (3): 707-734. 
10. Lucas, Robert E. Jr. and Sargent, Thomas J (1981). After 
Keynesian Economics. In Rational Expectations and 
Econometric Practice, eds Robert E. Lucas, Jr. and Thomas 
J. Sargent. Minneapolis: University of Minnesota Press. 
11. Lucas, Robert E., Jr. (1972). Expectations and the 
Neutrality of Money. Journal of Economic Theory 4 (1): 
103–124. 
12. Mundell, Robert (1968). International Economics. New 
York : Macmillan. 
13. Muth, John F. (1961). Rational Expectations and the Theory 
of Price Movements. Econometrica 29 (6): 315–335. 
14. Obstfeld, Maurice and Rogoff, Kenneth (2000). New 
Directions for Stochastic Open Economy Models. 
Journal of International Economies 50 (2): 117-153. 
15. Obstfeld, Maurice and Rogoff, Kenneth (2002). The Six 
Major Puzzles in International Macroeconomics: Is There 
a Common Cause? NBER Working Paper w 7777. 
16. Ribnikar, Ivan (1999). Monetary arrangements and 
exchange rate regime in a small transitional economy 
(Slovenia). In Inclusion of Central European countries 
in the European Monetary Union, eds, Paul de Grauwe 
and Vladimir Lavrač. Boston: Dordrecht; London: 
Kluwer Academic Publishers. 
17. Rohatinski, Željko; Anušić, Zoran; Šonje, Velimir (1995). 
Put u Nisku Inflaciju. Vlada Republike Hrvatske. 
18. Rogoff, Kenneth, Reinhart, Carmen, and Savastano, 
Miguel (2003). Debt Intolerance. Brookings Papers on 
Economic Activity 1: 1-62. 
19. Stokey, Nancy, and Lucas, Jr., Robert E. (with Edward 
C. Prescott) (1989) Recursive Methods in Economic 
Dynamics. Cambridge, Mass.: Harvard University Press. 
20. Sargent, Thomas (1992). Rational Expectations and 
Inflation. New York: Harper and Row, second edition. 
21. Sargent, Thomas and Ljungqivst, Lars (2004). Recursive 
Macroeconomic Theory. Second Edition, MIT Press. 
22. Sargent, Thomas and Hansen, Lars (2000). Wanting 
Robustness in Macroeconomics. June unpublished. 
23. Sargent, Thomas; Hansen, Lars (2006). Robustness. 
unpublished book manuscript. 
24.Vidaković, Neven (2005a). Mundell-Fleming Model and 
its Applicability on a Small Open Economy. Ekonomija 
11 (3): 392-424. 
25. Vidaković, Neven (2005b). Theory of Rational Expectations 
and Exports. Ekonomija 12 (1): 153-173. 
VIDAKOVIĆ: EMPIRICAL ANALYSIS OF MONETARY POLICY: CROATIA VS. SLOVENIA