






































Microsoft Word - DEM_2014_71to91.docx


© 2014 Nicolaus Copernicus University. All rights reserved.  
http://www.dem.umk.pl/dem 

D Y N A M I C  E C O N O M E T R I C  M O D E L S  
DOI: http://dx.doi.org/10.12775/DEM.2014.004  Vol. 14 (2014) 71−91 

Submitted October 20, 2014  ISSN 
Accepted December 23, 2014 1234-3862 

Andrzej Geise, Mariola Piłatowska* 

Oil Prices, Production and Inflation in the Selected  
EU Countries: Threshold Cointegration Approach 

A b s t r a c t. This paper applies the threshold cointegration technique developed by Enders 
and Siklos (2001) to investigate the impact of an oil price changes on changes in production 
and inflation in the presence of structural break in seven European Union countries. This 
technique will allow for a different speed of adjustment to the long-run equilibrium depending 
on whether production in selected economies is above or below the long-run relationship. 
Given the presence of asymmetric cointegration between oil prices, production and inflation, 
we estimate threshold error correction models to examine long- and short-run Granger causal-
ity. We found evidence for cointegration with asymmetric adjustment in the case of France, 
Denmark and the total EU.  

K e y w o r d s: asymmetric adjustment, oil price shocks, threshold cointegration, non-
linearity, threshold error correction model. 

J E L Classification: C32, E23, E32, Q43. 

Introduction 

 The relationship between oil prices and macroeconomy has drawn atten-
tion in many recent studies. Oil shocks have important effects on economic 
activity and macroeconomic policy of many countries. Barsky and Kilian 
                                                 

* Correspondence to: Andrzej Geise, Nicolaus Copernicus University, Faculty of Eco-
nomic Sciences and Management, 13A Gagarina Street, 87-100 Toruń, Poland, e-mail: 
a.geise@doktorant.umk.pl; Mariola Piłatowska, Nicolaus Copernicus University, Faculty of 
Economic Sciences and Management, 13A Gagarina Street, 87-100 Toruń, Poland, e-mail: 
mariola.pilatowska@umk.pl. 

 This work was financed from Faculty of Economic Sciences and Management research 
grant no. 1844-E. 



Andrzej Geise, Mariola Piłatowska 

DYNAMIC ECONOMETRIC MODELS 14 (2014) 71–91 

72

(2004) shows that the economic slowdowns and increases of inflation rates 
can be leaded by huge and sudden increases in oil prices. Also, Brown and 
Yucel (2002) and Lardic and Mignon (2006) presented explanations of the 
oil price shocks on economic activity. The rises in the crude oil prices are 
transferred on the energy prices through the price of petroleum products. The 
rise in the energy price causes an decrease in productivity, which is passed 
on to real wages, unemployment, inflation, investment or stock prices. 
Therefore, it is important to investigate how the oil price shock are transmit-
ted into economic activity, both in the short-run and long-run perspective. 
 The main goal of this study is to examine the dynamics between Brent 
oil prices and economic activity (described by the production and inflation) 
in the presence of structural break due to financial crisis. The error correc-
tion model with threshold cointegration is employed for the analysis of 
short-run relationship with non-linear long-term adjustment between produc-
tion, inflation and oil prices. We assume two long-run relationships: first 
equation without taking into account a structural break and second equation 
including a structural break due to financial crisis in 2008.  The analysis uses 
the nonlinear threshold cointegration elaborated by Enders and Granger 
(1998) and Enders and Siklos (2001). This technique will allow for a differ-
ent speed of adjustment to the long-run equilibrium depending on whether 
production in selected economies is above or below the long-run relation-
ship. At the end, an error correction model with threshold cointegration is 
utilized to analyze the short-run and long-run Granger causality. In the anal-
ysis the monthly data of Brent crude oil prices from 1995:01 to 2014:04, 
industry production index and consumer price index for the European Union 
economy and six European countries: Germany, France, Denmark, Neder-
land, Poland and Czech Republic are used. The data were taken from the 
U.S. Energy Information Administration database, the OECD database and 
Eurostat database. These economies have been selected on the basis of three 
criteria: level of economic growth, structure of net imports of crude oil and 
gas, and structure of final energy consumption in industry. Characteristics of 
countries according to selected criteria is presented in Table 1.  
 Besides these six economies, the total EU was taken into account when 
investigating the relationship between oil prices, production and inflation. 
This will allow to compare the results for chosen six economies and total 
average for the EU.  

 

 



Oil Prices, Production and Inflation in the Selected EU Countries... 

DYNAMIC ECONOMETRIC MODELS 14 (2014) 71–91 

73

Table 1. Characteristics of countries according to selected criteria 
Country Description 

Germany (DE) 
France (FR) 

Two highly developed countries in the EU with the highest 
GDP in 2012, the highest net imports of crude oil and gas in 
2012 and the highest final energy consumption in industry 
(Germany 61.15 Mtoe; France 29.58 Mtoe). 

Nederland (NL) 
Denmark (DN) 

Two highly developed countries in the EU economies with 
different structure of imports and exports of crude oil and/or 
gas. Nederland is a  gas net exporter and crude oil net im-
porter. Denmark is a net exporter of both, crude oil and gas. 

Poland (PL) 
Czech Republic (CZ) 

Two developing countries in the EU with the highest econom-
ic growth in 2012 among Central and Eastern Europe coun-
tries (CEE). Poland and Czech Republic are countries with 
the highest net imports of crude oil and gas in 2012 and the 
highest final energy consumption in industry among the CEE 
countries. 

 

 The plan of this paper is as follows. Section 1 presents the empirical 
literature review. Section 2  reviews the methodology applied in this paper. 
Empirical results of threshold cointegration and Granger-causality test based 
on threshold error correction model are presented in Section 3 and Section 4. 
The last section draws some conclusions. 

1. A Brief Overview of the Empirical Studies 

 Many studies are available which offer different theoretical explanations 
for the relationship between oil price changes and the level of economic 
activity. Hamilton (1983) using post-war data found a statistically significant 
relationship between oil price changes and GDP in US economy. Gisser and 
Godwin (1986) and Burbridge and Harrison (1984), among others, confirm 
the Hamilton’s results. They identified the causal relationship between oil 
prices and economic variables, i.e. real GDP, general price level, rate of 
unemployment and real investment. Gisser and Goodwin (1986) indicated 
for the analyzed period from 1961 to 1982 that the oil prices had not lost its 
potential to predict GDP growth. They showed also, that oil shocks have an 
impact on production by other means than inflationary cost-push effects. 
They confirmed Hamilton’s (1983) results about negative correlation be-
tween oil prices and real output in United States. Burbidge and Harrison 
(1984) found similar correlations for other industrialized countries. 
 A number of authors have suggested that the relation between oil price 
shocks and economic variables is nonlinear. Linear approximation to the 



Andrzej Geise, Mariola Piłatowska 

DYNAMIC ECONOMETRIC MODELS 14 (2014) 71–91 

74

relation between oil price shock and macroeconomic activity may appear 
unstable over time. It is caused by shifts in the process generating oil prices 
(Mork, 1989; Hamilton, 1996; Davis and Haltiwanger, 2001). This means 
that the economic activity may respond asymmetrically to positive and nega-
tive oil price shocks, i.e. the increases of oil prices slow the economy more 
than the decreases of oil prices stimulate it (Brown and Yucel, 2002). The 
impact of oil price decreases is not always positive, indeed these decreases 
may slow output growth down. Mork (1989) shows that the relationship 
between oil price changes and real GNP growth for the U.S. economy breaks 
down when the analysis is extended to include the oil price collapse of 1986. 
Hence, Mork (1989) decided to test the influence of declines and increases in 
oil price on the economic growth, he showed that the coefficient on oil price 
increases are highly significant and negative. In another work, Mork et al. 
(1994) showed that for the most of European countries a negative relation-
ship between oil price increases and GDP growth  occurs. However, evi-
dence for the causal relationship between oil prices and economic growth 
was not always confirmed in empirical studies.  For example Hooker (1996) 
found some evidence that oil prices are no longer Granger cause of GDP. In 
direct response to this empirical work, Hamilton (1996) showed new meas-
ure (net oil price increases-NOPI- difference between oil price level and the 
maximum price of the previous four quarters) which was able to detect 
a significant relationship between oil prices and real GDP. The NOPI-
measure from assumption is the measure of rapid and huge changes in oil 
prices. 
 According to most empirical studies the rising oil prices lead to real 
GDP loses and it is consistent with the economic theory, though, it is possi-
ble to find the positive response of production to changes in oil prices. This 
kind of findings, however, should not be treated as inconsistent with eco-
nomic theory since, as Bernanke et. al (1997) showed on the example of the 
U.S. economy, the response of production to an oil price shock may be dif-
ferent depending on the response of monetary policy to oil prices shocks (i.e. 
whether the federal funds rate is constrained to be constant than in the case 
in which monetary policy is unconstrained). If the federal funds rate is held 
constant, a positive oil price shock leads to an increase in real GDP. In the 
unconstrained case, a positive oil prices shock leads to a decline in real GDP. 
Similar findings concern the response of GDP to changes in inflation, i.e. 
although the economic theory indicates that movements in inflation cause 
a decrease in production, there are some empirical findings indicating the 
opposite direction of relationship. The reason why increases in inflation have 
positive influence on economic growth can be explained  in the framework 



Oil Prices, Production and Inflation in the Selected EU Countries... 

DYNAMIC ECONOMETRIC MODELS 14 (2014) 71–91 

75

of the optimal rate of inflation (Friedman, 1976; Belka, 1986; Barro, 1995). 
If the inflation is below the optimal level, then the reaction of production is 
positive (slight increase to the optimal level of inflation stimulates the eco-
nomic growth). While, the impact of inflation on production is negative (an 
increase in inflation causes a decrease in production) when consumer price 
level is higher than the optimal one. 
 Most of empirical literature on relationships between oil prices and eco-
nomic activity focuses on the US economy. Europe, despite of economic size 
and huge oil import, has received relatively less attention. Only few studies 
focus on this issue, e.g. Scholtens and Yurtsever (2012) and Arouri (2011) 
investigated how European industries responded to oil price shock. They 
found that the impact of oil price shocks substantially differed along the 
different industries. They also found that the significance of the result differs 
along the various oil price specifications. To studies considering this rela-
tionship for some European countries belong the papers by Jimenez-
Rodriguez and Sanchez (2005) and Papapetrou (2001). The former con-
firmed positive effects of oil price decreases on output for Japan, Germany, 
France, Canada, Norway and the United Kingdom. And the latter found 
causal relationship from oil prices to industrial production, employment and 
share prices in Greek economy. To our knowledge, there is no such study 
which considers the relationship between oil prices changes and economic 
activity using threshold cointegration methodology with non-linear adjust-
ment for the countries studied in this paper.  

2. Methodology 

 To test the relationship between Brent oil prices and economic activity 
(described by the production and inflation) in the asymmetric framework the 
threshold cointegration technique developed by Enders and Granger (1998) 
and Enders and Siklos (2001) is used. Similarly to Engle-Granger (1987) 
procedure, this is indeed a residual-based two-stage estimation. In the first 
stage the long-run relationship between production, inflation and oil prices is 
considered, e.g. in the following form: 

,210 tttt InBP    (1) 

where tP − stands for production, tB − for oil prices and tIn  − for inflation, 

t  − disturbance term. 
The second stage focuses on the coefficient estimates of 1  and 2  in the 
following regression: 



Andrzej Geise, Mariola Piłatowska 

DYNAMIC ECONOMETRIC MODELS 14 (2014) 71–91 

76

,)1(
11211 t

r

i itittttt II       (2) 

where t  is a white noise disturbance.  

Term It is the Heaviside indicator function tI  such that 1tI  if   1t   

and 0tI  if  ,1  t  where   is the threshold value. 
Then, equation (2) is a threshold autoregressive (TAR) model of the disequi-
librium error. If the Heaviside indicator depends not on levels but on chang-
es of t−1, then it is specified as 1tI  if    1t  and 0tI  if  .1   t  
This model is termed the momentum threshold autoregressive (M-TAR) 
model. The TAR framework is designed to capture potential asymmetric 
movements in residuals t when they are above or below the long-run equi-
librium, while the M-TAR framework − the direction of these movements, in 
other words their momentum (Fosten et. al, 2012). 
 The necessary and sufficient condition for t  to be stationary is: ,01   

,02   1)1)(1( 21    for any threshold value   (Petruccelli, Wool-
ford, 1984). If these conditions are satisfied and threshold value   is set to 
zero (as occurs in many economic applications), 0t  can be considered as 

the long-run equilibrium value of the sequence. If t  is higher than the long-

run equilibrium, the adjustment is ,11 t  but if t  is lower than the long-

run equilibrium, the adjustment is .12 t  In general, the threshold value   

has to be estimated along with the values of adjustment parameters 1  and 

.2  In our studies we follow Enders and Siklos (2001) and Yau and Nieh 
(2009) by employing Chan's (1993) methodology of searching the consistent 
estimates of threshold value. 
 Testing for threshold cointegration is performed in two steps. First, the 
null hypothesis of no cointegration 0: 210  H  is tested, and when it is 

rejected, then the null hypothesis of symmetric adjustment, ,: 210  H  is 
verified. To test the first hypothesis Enders and Siklos (2001) proposed the 

  statistics which under the null of no cointegration has a non standard 

distribution. The critical values for this non standard   statistics are tabu-

lated in their paper. The second null hypothesis of asymmetric adjustment is 
tested using the standard F statistics. Rejecting both the null hypotheses 
implies the existence of threshold cointegration with asymmetric adjustment.  
 Given the threshold cointegration is found, the next step proceeds with 
the Granger-causality test using threshold (or momentum-threshold) error 



Oil Prices, Production and Inflation in the Selected EU Countries... 

DYNAMIC ECONOMETRIC MODELS 14 (2014) 71–91 

77

correction model (TAR-ECM or M-TAR-ECM) (Enders, Granger, 1998; 
Enders, Siklos, 2001). This model is expressed as the following:  

t
q

i itji
q

i itji
q

i itjitjtjjt vBInPZZY     




  321

1111211 
          (3) 

where ),,,( tttjt BInPY  ,3,2,1j  11 

  ttt IZ   and ,)1( 11 


  ttt IZ    

tI  − Heaviside indicator function, tv  is a white noise disturbance. The long-

run causality is determined by the parameters 1  and .2  The short-run cau-

sality is governed by the parameters ,i  i  and i  and may come either 
from its own history of lagged dynamics or from the lagged effects of 
changes in real GDP (and square real GDP) and/or some additional explana-
tory variables (e.g. energy consumption). It is also desirable to check wheth-
er this two sources of causation are jointly significant.  

3. Threshold Cointegration – Empirical Analysis 

 In this section, an integration and threshold cointegration analysis be-
tween Brent oil prices, production and inflation in German, France, Neder-
land, Denmark, Poland, Czech Republic and European Union economies are 
discussed. The data consist of seasonally adjusted monthly industrial produc-
tion index at constant prices of 2010, consumer price index for the EU coun-
tries (corresponding month of previous year = 100) and Brent crude oil pric-
es from January 1995 to April 2014. For seasonal adjustment the TRA-
MO/SEATS procedure was used. Each time series has been transformed 
using the natural logarithm. Then empirical threshold error correction mod-
els for oil prices, inflation and production are constructed to study short and 
long-run causality. 
 To assess the time series properties of the data, we examine the order of 
integration for all variables involved. We run the conventional unit root test, 
the Augmented Dickey-Fuller test. 
 Findings from the ADF test, as shown in Table 2, reveal that each series 
is integrated of first order, I (1), at least at the 5% significance level. 
 Since the studied time span includes the period 2008-2009 when the 
financial crisis occurred1 and the presence of a structural break may change 
the nature of the long-run relationship between production, inflation and oil 
prices, from the very start we assume both the log-run relationship without 

                                                 
1 Visual inspections of production suggests that the shift in mean occurred about Septem-

ber 2008.  



Andrzej Geise, Mariola Piłatowska 

DYNAMIC ECONOMETRIC MODELS 14 (2014) 71–91 

78

taking into account a structural break (eq. (4)) and long-run relationship in 
the presence of structural break (eq. (5)). The first model takes the following 
form:  

tttt utimeInBP  3210  , (4) 

where tP , tI  denote respectively the industrial production index and con-

sumer price index in selected economies, tB  stands for Brent crude oil pric-

es; all variables are in natural logarithms;  
and the second model has the form:  

ttttt DTtimeInBP   )(*
43210 ,  (5) 

where )(* tDT  is the dummy variable for the break in constant term of pro-

duction from the chosen exogenously breakpoint   so 1)(* tDT  if 

2008:09t   and  0 otherwise. 

Table 2.  Results of the ADF test 

Country Variable 
 

ADF 
Levels First differences 

C only C and t C only 
– Brent Oil Price –1.04 [1] –3.23 [1]* –9.51 [1]*** 

Germany (DE) Production –0.96 [1] –2.18 [1] –9.89 [1]*** 
Inflation –3.03 [1]** –3.07 [1] –10.51 [1]*** 

France (FR) Production –1.14 [1] –1.58 [1] –11.12 [1]*** 
Inflation –3.02 [1]** –3.00 [1] –10.01 [1]*** 

Nederland (NL) Production –2.18 [1] –2.23 [1] –15.69 [1]*** 
Inflation –2.57 [1]* –2.63 [1] –10.54 [1]*** 

Denmark (DN) Production –2.51 [1] –2.43 [1] –13.34 [1]*** 
Inflation –2.82 [1]* –2.95 [1] –9.95 [1]*** 

Poland (PL) Production –0.80 [1] –2.19 [1] –11.52 [1]*** 
Inflation –3.13 [1]** –3.15 [1]* –7.71 [1]*** 

Czech Republic (CZ) Production –0.76 [1] –2.02 [1] –13.04 [1]*** 
Inflation –2.10 [1] –2.50 [1] –7.71 [1]*** 

European Union (EU) Production –1.12 [1] –1.69 [1] –6.09 [1]*** 
Inflation –2.97 [1]** –3.05 [1] –8.61 [1]*** 

Note:  asymptotical critical values for the ADFmax test statistics: –3.46 (1%);  –2.88 (5%); –2.57 (10%) 
(ADF regression with drift – C only); asymptotical critical values for the ADFmax test statistics: –3.99 
(1%);  –3.43 (5%); –3.13 (10%) (ADF regression with an intercept and trend – C and t); ***, ** and * 
denote significant at 1%, 5% and 10% levels, respectively. 

  



Oil Prices, Production and Inflation in the Selected EU Countries... 

DYNAMIC ECONOMETRIC MODELS 14 (2014) 71–91 

79

Table 3.  Estimated parameters of long-run equilibrium for production 

Country  Estimated parameters of long-run equation ADF 

Const. Bt Int time DTt(λ) 
Levels 
(C+t) 

Long-run relationship without structural break (4) 
Germany (DE) –17.02 

(–7.63)*** 
0.020 
(1.65)* 

4.616 
(9.48)*** 

0.001 
(10.27)*** 

– –3.29* 

France (FR) –10.839 
(–4.21)*** 

–0.022 
(–1.38) 

3.373 
(6.00)*** 

0.0001 
(19.8)*** 

– –2.05 

Nederland (NL) 0.936 
(0.74) 

0.053 
(6.49)*** 

0.718 
(2.62)*** 

0.001 
(5.99)*** 

– –4.64*** 

Denmark (DN) –1.965 
(–0.54) 

0.003 
(0.14) 

1.428 
(1.81)* 

0.0001 
(0.62) 

– –2.48 

Poland (PL) 5.083 
(15.31)*** 

0.110 
(9.20)*** 

–0.351 
(–4.85)*** 

0.003 
(21.92)*** 

– –2.78 

Czech Republic 
(CZ) 

1.605 
(2.12)** 

0.132 
(8.87)*** 

0.466 
(2.84)*** 

0.002 
(8.47)*** 

– –2.81 

European 
Union (EU) 

9.645 
(6.7)*** 

0.132 
(10.98)*** 

–1.172 
(–3.74)*** 

–0.001 
(–7.12)*** 

– –3.73** 

Long-run relationship with structural break (5)  
Germany (DE) –11.67 

(–5.97)*** 
0.017 
(1.67)* 

3.455 
(8.11)*** 

0.002 
(15.11)*** 

–0.077 
(–9.78)*** 

–3.65 *** 

France (FR) 0.509 
(0.40) 

–0.031 
(–4.13)*** 

0.909 
(3.27)*** 

0.001 
(11.65)*** 

–0.017 
(–28.3)*** 

–3.78*** 

Nederland (NL) 1.387 
(1.29) 

0.036 
(4.97)*** 

0.626 
(2.67)*** 

0.001 
(11.33)*** 

–0.063 
(–9.61)*** 

–5.94*** 

Denmark (DN) –6.748 
(–3.97)*** 

–0.069 
(–6.92)*** 

2.484 
(6.73)*** 

0.002 
(17.12)*** 

–0.245 
(–28.67) 

–6.82*** 

Poland (PL) 3.287 
(6.05)*** 

0.064 
(3.93)*** 

0.047 
(0.4) 

0.005 
(14.49)*** 

–0.07 
(–4.1)*** 

–2.73 

Czech Republic 
(CZ) 

–1.012 
(–1.84)* 

0.074 
(6.75)*** 

1.046 
(8.74)*** 

0.003 
(19.32)*** 

–0.154 
(–15.67)*** 

–4.48*** 

European 
Union (EU) 

1.771 
(1.95)* 

0.025 
(2.91)*** 

0.576 
(2.90)*** 

0.001 
(8.14)*** 

–0.142 
(–20.47)*** 

–4.37*** 

Note: ***, ** and * denote significance at 1%, 5% and 10% level respectively; in parentheses the t-
statistics are given. 

 Table 3 presents the estimation results of eq. (4) and eq. (5). It is worth 
noting that for the long-run equation without a structural break only in the 
case of France the response of production to the change in oil prices is nega-
tive ( 01  ). Having included the dummy variable for a structural break in 
eq. (5), except France, only in the case of Denmark the negative response of 
production is additionally observed and for the rest of countries (Germany, 
Nederland, Poland, Czech Republic and European Union) the positive re-
sponse of production to changes in oil prices is found. The response of GDP 
to changes in inflation is positive too. Positive impact of oil prices on pro-



Andrzej Geise, Mariola Piłatowska 

DYNAMIC ECONOMETRIC MODELS 14 (2014) 71–91 

80

duction can be explained by the monetary policy carried out by the European 
Central Bank (ECB). While the ECB reacting to the financial crisis in 2008 
increased the interest rates and then in recession phase reduced them, the 
interest rates before the financial crisis were held constant. This may explain 
why the response of production to oil price shocks (Table 3) is positive as 
suggested by Bernanke et. al (1997) − see comments in Section 1. 
But the positive response of GDP to changes in inflation can be explained by 
low inflation rate, near to optimal one (the general price level was compara-
tively stable, except 2008 when inflation increased near to 4 per cent (in-
creased by 4 percentage points in Euro Zone) – see comments in Section 1. 
 While in the case of long-run equilibrium relationship without a structur-
al break the cointegration is found for Germany, Nederland and the EU, in 
the case of  long-run relationship allowing for a structural break the cointe-
gration is found for Germany, France, Netherlands, Denmark, Czech Repub-
lic and the European Union. 
 However, the standard cointegration framework (Engle-Granger ap-
proach) assuming symmetric adjustment toward equilibrium is misspecified 
if the adjustment is asymmetric. Therefore, to test for cointegration with 
non-linear (asymmetric) adjustment, we use the threshold cointegration ap-
proach proposed by Enders and Siklos (2001). In further analysis we use 
only the results based on residuals from long-run relationship with a struc-
tural break in 2008. Table 4 contains cointegration test results  based on the 
long-run equilibrium relationships for production in selected countries in the 
form of equation (5), when considering threshold and momentum adjust-
ment. We consider threshold autoregressive (TAR) model and momentum 
threshold autoregressive (M-TAR) model with zero-value of threshold,  
 = 0, and also with non-zero value of threshold   0 which should be esti-
mated. The table reports values of the adjustment coefficients 1 and 2 , the 

 statistics for the null hypothesis of no cointegration against the alterna-

tive of cointegration with asymmetric adjustment, the F statistics for testing 
the symmetric adjustment, AIC and Ljung-Box statistics. 
 The results for the threshold cointegration tests present some interesting 
relations among oil prices, production and inflation in selected economies.  



Oil Prices, Production and Inflation in the Selected EU Countries... 

DYNAMIC ECONOMETRIC MODELS 14 (2014) 71–91 

81

Table 4.  Results of TAR and M-TAR test for threshold cointegration on the long-
run equation with one structural break in production 

 Threshold cointegration test 
 = 0   0 

TAR M-TAR TAR M-TAR 

Germany (DE)  
 0 0 –0.016 –0.013 
1 –0.123 (2.23)** –0.037 (0.69) –0.089 (1.65)* –0.100 (2.30)** 
2 –0.161 (3.06)*** –0.243 (4.64)*** –0.196 (3.68)*** –0.286 (3.66)*** 
 6.882** 10.865** 7.859** 9.057** 

F(1–2=0) 0.266 [0.61] 7.875 [0.01]** 2.109 [0.15] 4.447 [0.04]** 
Lag 1 1 1 1 
AIC –1207.7 –1215.2 –1209.5 –1211.9 

LB(4) 3.87 [0.42] 4.88 [0.30] 4.13 [0.39] 4.40 [0.36] 
France (FR) 

 0 0 –0.022 –0.008 
1 –0.114 (2.01)** –0.067 (1.07) –0.118 (2.19)** –0.096 (1.91)* 
2 –0.197 (3.05)*** –0.219 (3.76)*** –0.207 (2.94)*** –0.290 (3.72)*** 
 6.208** 7.423** 6.247* 8.17** 

F(1–2=0) 1.048 [0.31] 3.41 [0.07]* 1.121 [0.29] 4.834 [0.03]** 
Lag 2 2 2 2 
AIC –1281.3 –1283.7 –1281.4 –1285.1 

LB(4) 1.47 [0.83] 1.75 [0.78] 1.44 [0.84] 1.26 [0.87] 
Netherlands (NL) 

 0 0 0.011 0.022 
1 –0.337 (3.93)*** –0.278 (3.04)*** –0.334 (3.87)*** –0.373 (2.69)** 
2 –0.318 (3.54)*** –0.371 (4.36)*** –0.322 (3.61)*** –0.321 (4.52)*** 
 11.404** 11.715** 11.395** 12.191** 

F(1–2=0) 0.031 [0.86] 0.688 [0.41] 0.013 [0.91] 1.551 [0.21] 
Lag 2 2 2 2 
AIC –1090.3 –1091 –1090.3 –1090.5 

LB(4) 0.80 [0.94] 0.77 [0.94] 0.81 [0.94] 0.84 [0.93] 
Denmark (DN) 

 0 0 0.031 –0.020 
1 –0.38 (4.69)*** –0.47 (4.73)*** –0.386 (4.48)*** –0.357 (5.14)*** 
2 –0.458 (5.56)*** –0.394 (5.53)*** –0.443 (5.52)*** –0.572 (5.52)*** 
 23.671** 23.517** 23.524** 25.26** 

F(1-2=0) 0.517 [0.47] 0.431 [0.51] 0.272 [0.60] 3.323 [0.07]* 
Lag 1 1 1 1 
AIC –975 –974.9 –974.8 –977.8 

LB(4) 1.96 [0.74] 2.54 [0.64] 2.05 [0.73] 4.50 [0.83] 



Andrzej Geise, Mariola Piłatowska 

DYNAMIC ECONOMETRIC MODELS 14 (2014) 71–91 

82

Table 4 (continued) 

Poland (PL) 
 0 0 -0.012 0.015 
1 –0.126 (2.67)*** –0.136 (2.84)*** –0.134 (2.85)*** –0.124 (1.67)* 
2 –0.061 (1.26) –0.053 (1.13) –0.052 (1.08) –0.086 (2.27)** 
 4.256 4.546 4.546 3.846 

F(1–2=0) 0.972 [0.33] 1.572 [0.21] 1.533 [0.22] 0.212 [0.65] 
Lag 1 1 1 1 
AIC –1136.2 –1136.8 –1136.8 –1135.4 

LB(4) 0.05 [0.99] 0.08 [0.99] 0.05 [0.99] 0.05 [0.99] 
Czech Republic (CZ) 

 0 0 0.03 –0.017 
1 –0.184 (2.79)*** –0.24 (3.48)*** –0.203 (2.87)*** –0.192 (3.65)*** 
2 –0.192 (2.95)*** –0.148 (2.37)** –0.179 (2.88)*** –0.173 (1.68)* 
 7.437** 8.092** 7.408** 7.353** 

F(1–2=0) 0.005 [0.94] 1.296 [0.26] 0.074 [0.79] 0.032 [0.86] 
Lag 2 2 2 2 
AIC –1026.2 –1010.1 –1026.3 –1026.2 

LB(4) 0.11 [0.99] 0.07 [0.99] 0.10 [0.99] 0.11 [0.99] 
European Union (EU) 

 0 0 0.003 –0.006 
1 –0.155 (2.84)*** –0.061 (1.07) –0.153 (2.79)*** –0.082 (1.75)* 
2 –0.200 (3.44)*** –0.274 (5.21)*** –0.203 (3.48)*** –0.368 (5.61)*** 
 9.901** 14.117** 9.942** 17.037** 

F(1–2=0) 0.333 [0.56] 8.126 [0.004]*** 0.407 [0.52] 13.465 [<0.01]*** 
Lag 1 1 1 1 
AIC –1238 –1245.6 –1238.1 –1250.7 

LB(4) 2.76 [0.60] 2.59 [0.63] 2.47 [0.65] 3.30 [0.51] 
Note: ***, ** and * denote significance at 1%, 5% and 10% level respectively; in parentheses the t-
statistics are given; in brackets the p-values are given. 

 It can be seen that both the null hypotheses of 021    and 

021    (the null of no cointegration and the null of symmetric adjust-
ment respectively) are rejected for Germany, France, Denmark and European 
Union. This finding implies the existence of threshold cointegration with 
asymmetric adjustment for these countries and the EU what means that the 
adjustment back to equilibrium between production, inflation and oil prices 
is non-linear. To select the best adjustment mechanism (TAR or M-TAR) the 
Akaike information criterion (AIC) is used (the minimum AIC is reported in 
bold in Table 4). In case of Germany the M-TAR model with zero-value of 
threshold is selected, and in case of France, Denmark and the EU − the M-
TAR model with non-zero threshold value. Selection of M-TAR model indi-
cates that the direction in which production is moving (its momentum) mat-



Oil Prices, Production and Inflation in the Selected EU Countries... 

DYNAMIC ECONOMETRIC MODELS 14 (2014) 71–91 

83

ters more than whether production is above or below the equilibrium (as in 
TAR model). It is worth noting that all adjustment coefficients 1( and )2  
have negative signs acting to eliminate deviations from the long-run rela-
tionship and are significant for either or both regimes. The point estimates 

1 and 2  in the case of above mentioned countries indicate that deviations 
below the threshold adjust faster toward the long-run equilibrium than the 
deviations above the threshold (since ).21    In other words, deviations 

below the threshold from the long-run equilibrium resulting from decreases 
in production or increases in inflation and oil prices are corrected (eliminat-
ed) more quickly than deviations above the threshold). These models differ 
with regard to the magnitude of   terms in regimes (below and above the 
threshold value. The largest discrepancy between the elimination of below 
and above threshold deviations occurs for Germany, France and the EU, e.g. 
for the EU the deviations below the threshold are eliminated at 36.8% rate 
per month, while deviations above the threshold are eliminated only at a rate 
of 8.2%. 
 Although for Nederland and Czech Republic the null hypothesis of no 
cointegration is rejected, no support is found for cointegration with asym-
metric adjustment (since the null of symmetric adjustment is not rejected, 
see F statistics in Table 4). The results for Polish economy show that both 
the null hypothesis of no cointegration and symmetric adjustment are not 
rejected, suggesting oil prices, production and inflation in Poland are not 
cointegrated. 
 These results showed that the threshold cointegration occurs only for the 
developed countries (i.e. Germany, France, Denmark) those contribution to 
the total production of the EU is the highest. This is the reason why for the 
total EU the threshold cointegration has been also confirmed. Since the 
speed of adjustment is faster in the lower regime (deviations in production 
below the threshold), these countries and the EU as a whole have the mecha-
nism that enable them to quickly revert to equilibrium after moving away 
from it or in other words, economies remain shorter period of time in the 
regime of slower economic activity. For Poland and Czech Republic belong-
ing to the developing countries with lower level of GDP than averagely for 
the total EU or developed countries no evidence in favor of the threshold 
cointegration with asymmetric adjustment has been found. 
 
 
 
 



Andrzej Geise, Mariola Piłatowska 

DYNAMIC ECONOMETRIC MODELS 14 (2014) 71–91 

84

4. Granger Causality Analysis form Threshold Error Correction 
Model 

 Given the threshold cointegration results found in the previous Section, 
the next step proceeds with Granger causality test using the momentum 
threshold error correction model (M-TAR-ECM). In our case, the M-TAR-
ECM models take the following forms: 

ttitit
i

ittt BInPZZP   






  11111

3

1
1112111

,  (6) 

ttitit
i

ittt BInPZZIn   






  12121

3

1
2122121

,  (7) 

ttitit
i

ittt BInPZZB   






  13131

3

1
3132131

,  (8) 

where tP , tIn  denote log first differences of production and inflation in 

Germany (DE), France (FR), Denmark (DN) and European Union (EU) re-
spectively, tB  stands for the log differences of Brent crude oil prices. Here, 

11 

  ttt IZ   and ,)1( 11 


  ttt IZ   given 1tI  if ,1   t  and 0tI  

if .1   t  Furthermore, t  is the residual from long-run equation (5) for 

a given country and ttt v,,  are the white noise disturbance. 

 This approach allows us to distinguish between short-run and long-run 
Granger causality. Since Granger causality tests are very sensitive to the 
selection of lag length, we apply the AIC criterion to determinate the appro-
priate lag lengths. It is found that the lag length of production equals 3 and 
for both inflation and Brent oil prices is equal to one for all the economies.  
 The Wald F-statistics for Granger causality is employed to examine 
whether all the coefficients of a given first differenced variable are jointly 
statistically different from zero (short-run causality) and t-statistics for the 

21,  coefficients of error correction term to test their significance (long-run 
causality). Moreover, the jointly significance of the   coefficients and all 
the coefficients of a given explanatory variable is tested in order to indicate 
which variables bear the burden of short-run adjustment to restore the long-
run equilibrium given a shock to the system.  
 Tables 5 and 6 present the results of the Granger-causality test based on  
M-TAR-ECM models for oil prices, inflation and production.   



Oil Prices, Production and Inflation in the Selected EU Countries... 

DYNAMIC ECONOMETRIC MODELS 14 (2014) 71–91 

85

Table 5. Estimates of threshold ECM models and Granger-causality analysis (case of 
Germany and France) 

 

Germany (DE) 
M-TAR-ECM ( = 0) 

France (FR) 
M-TAR-ECM ( = –0.008) 

Bt Int Pt Bt Int Pt 

1 Z+t-1 0.027 
(0.099) 

0.0257 
(2.979)*** 

–0.158 
(–3.549)*** 

–0.242 
(–0.793) 

0.0042 
(0.462) 

–0.052 
(–1.269) 

2 Z-t-1 –0.2363 
(–0.863) 

0.0051 
(0.596) 

–0.101 
(–2.325)*** 

–0.651 
(–0.443) 

0.005 
(0.38) 

–0.244 
(–4.104)*** 

       
H0: 1=2=0 0.3821 

[0.683] 
4.544 

[0.012]** 
8.6267 

[<0.01]*** 
1.398 
[0.249] 

0.677 
[0.509] 

9.246 
[<0.01]*** 

H0: 1=2 0.479 
[0.49] 

3.039 
[0.08]* 

0.853 
[0.357] 

0.573 
[0.45] 

0.0023 
[0.96] 

7.028 
[<0.01]*** 

H0: 1=2=3=0 0.931 
[0.336] 

0.0275 
[0.869] 

 
2.973 

[0.086]* 
0.0303 
[0.862] 

 

H0: 1=2=3=1=0 0.5061 
[0.604] 

5.383 
[<0.01]*** 

 
1.615 
[0.201] 

0.673 
[0.511] 

 

H0: 1=2=3=2=0 0.6792 
[0.508] 

0.276 
[0.759] 

 
2.588 

[0.077]* 
0.123 
[0.884] 

 

H0: 1=0 0.0026 
[0.96] 

 
3.286 

[0.071]* 
0.025 
[0.874] 

 
0.006 
[0.938] 

H0: 1=1=0 0.0056 
[0.99] 

 
9.066 

[<0.01]*** 
0.3187 
[0.728] 

 
0.806 
[0.448] 

H0: 1=2=0 0.378 
[0.69] 

 
4.583 

[0.011]** 
1.115 
[0.329] 

 
8.496 

[<0.01]*** 
H0: 1=0  

17.378 
[<0.01]*** 

4.589 
[0.033]** 

 
8.35 

[<0.01]*** 
7.017 

[<0.01]*** 
H0: 1=1=0  

12.062 
[<0.01]*** 

9.393 
[<0.01]*** 

 
4.19 

[0.016]** 
4.697 

[0.011]** 

H0: 1=2=0  
8.697 

[<0.01]*** 
5.609 

[<0.01]*** 
 

4.189 
[0.018]** 

12.808 
[<0.01]*** 

LB(4) 
0.91 

[0.92] 
1.84 
[0.77] 

3.31 
[0.51] 

0.85 
[0.93] 

5.28 
[0.26] 

1.22 
[0.75] 

Note: *** (**) (*) indicate significance at 1% (5%) (10%) level. t-statistics for 1, 2 in parentheses.  
p-values for Wald statistics in brackets. LB(4) for Ljung-Box statistics along with p-values. 

 It is clear (Table 5 and 6) that the point estimates of 1  and 2  in all M-

TAR-ECM models for tP  have a negative sign but only for Germany and 

Denmark both are significant (for France and the EU only the coefficient 2  
is significant). In the case of France, Denmark and the EU, similarly to the 
findings in Table 4, the production deviations below the threshold adjust 
faster toward the long-run relationship than the deviations above  the  thresh- 



Andrzej Geise, Mariola Piłatowska 

DYNAMIC ECONOMETRIC MODELS 14 (2014) 71–91 

86

Table 6. Estimates of threshold ECM models and Granger-causality analysis (case of 
Denmark and the total EU) 

 
Denmark (DN) 

M-TAR-ECM ( = –0.020) 
European Union (EU) 

M-TAR-ECM ( = –0.006) 
Bt Int Pt Bt Int Pt 

1 Z+t-1 –0.79 
(–3.58)*** 

–0.013 
(–2.044)** 

–0.23 
(–3.466)*** 

0.481 
(1.251) 

0.016 
(1.645) 

–0.044 
(–1.388) 

2 Z-t-1 –0.167 
(–0.514) 

–0.016 
(–1.707)* 

–0.513 
(–5.242)*** 

–0.992 
(–1.943)* 

–0.0089 
(–0.686) 

–0.2701 
(–6.48)*** 

       
H0: 1=2=0 6.4086 

[<0.01]*** 
3.073 

[0.048]** 
17.318 

[<0.01]*** 
2.594 

[0.077]* 
1.553 
[0.214] 

22.277 
[<0.01]*** 

H0: 1=2 2.9376 
[0.088]* 

0.0823 
[0.774] 

6.708 
[0.011]** 

5.781 
[0.017]** 

2.574 
[0.11] 

16.528 
[<0.01]*** 

H0: 1=2=3=0 1.835 
[0.177] 

5.426 
[0.021]** 

 
0.893 
[0.346] 

0.571 
[0.451] 

 

H0: 1=2=3=1=0 5.804 
[0.003]*** 

3.097 
[0.047]** 

 
2.006 
[0.137] 

1.904 
[0.152] 

 

H0: 1=2=3=2=0 0.9423 
[0.391] 

3.095 
[0.047]** 

 
2.564 

[0.079]* 
0.704 
[0.496] 

 

H0: 1=0 0.171 
[0.697] 

 
2.228 
[0.137] 

0.172 
[0.679] 

 
0.1897 
[0.664] 

H0: 1=1=0 7.299 
[<0.01]*** 

 
6.226 

[<0.01]*** 
0.872 
[0.42] 

 
1.054 
[0.351] 

H0: 1=2=0 0.248 
[0.781] 

 
14.104 

[<0.01]*** 
1.996 
[0.139] 

 
21.175 

[<0.01]*** 
H0: 1=0  

6.127 
[0.014]** 

1.512 
(0.22) 

 
5.732 

[0.018]** 
5.435 

[0.021]** 
H0: 1=1=0  

5.523 
[<0.01]*** 

6.504 
[<0.01]*** 

 
4.06 

[0.019]** 
3.825 

[0.023]** 

H0: 1=2=0  
4.98 

[<0.01]*** 
14.001 

[<0.01]*** 
 

3.534 
[0.031]** 

27.508 
[<0.01]*** 

LB(4) 
2.38 

[0.67] 
1.85 
[0.76] 

0.54 
[0.97] 

0.66 
[0.96] 

2.68 
[0.61] 

4.36 
[0.36] 

Note: see Table 5.  

old (since ).21    We can see that respectively  24.4%,  51.3%  and 

27.01%  of  the production deviations from equilibrium is corrected in the 
next period when they are below the threshold, but only 5.2%, 23% and 
4.4% when they are above the threshold. In other words, deviations below 
the threshold from the long-run equilibrium resulting from decreases in pro-
duction or increases in inflation and oil prices are corrected more quickly 
than deviations above the threshold. However, for Germany, unlike the pre-
vious results in Table 4, the speed of adjustment is faster in the regime with 
deviations in production above the threshold but on the other hand both -



Oil Prices, Production and Inflation in the Selected EU Countries... 

DYNAMIC ECONOMETRIC MODELS 14 (2014) 71–91 

87

coefficients are similar in magnitude (the hypothesis of symmetric adjust-
ment is not rejected, F=0.853 with p-value=0.357). This rather symmetric 
adjustment of production may indicate that Germany in general is not nega-
tively influenced by higher energy prices. Although it is one of the important 
oil importing countries, at the same time it is the leading goods exporting 
economy with a strong investment goods industry and is currently becoming 
the leader for some renewable energies and energy efficiency goods. This 
allows Germany to level off the negative impact of increasing oil prices  − 
see Lutz, Meyer (2009).   
 The M-TAR-ECM results for Denmark show that the deviation away 
from the long-run relationship is corrected not only by movements in pro-
duction, but also by movements in inflation and oil prices when the produc-
tion in Denmark is temporarily above the long-run equilibrium. This can be 
seen by the negative signs and significance of the error correction parameter 

1  in the M-TAR-ECMs for tP , tIn  and tB . When the production in 

Denmark is temporarily below the long-run equilibrium, the deviation away 
from long-run relation is corrected by movements in production and inflation 

(significance of 2  in equation for tP  and tIn ). The M-TAR-ECM mod-

els for tIn  and tB  in Germany, France and EU adjust to the ‘wrong’ di-

rection (the   coefficients have positive signs) in either or both regimes, and 

additionally parameters  1  and 2  are not always significant – see Table 5. 
 Summing up, the results from threshold error correction model con-
firmed the previous findings (Table 4) with regard to cointegration with 
asymmetric adjustment for deviations in production in the case of France, 
Denmark and the total EU. This can be seen by statistical significance of 
both or either adjustment coefficients () which are properly signed (are neg-
ative). While in the case of Germany the cointegration between production, 
oil prices and inflation occurs, production adjustments toward equilibrium 
seem to be symmetric. 
 The results of Granger-causality test, based on the M-TAR-ECM models 
for all economies show that there is short-run unidirectional causal relation-
ship running from oil prices )( tB  to  production )( tP  and inflation )( tIn  

(rejection of 0: 10 H  in equations for tP  and tIn  with exception of 

Denmark for which unidirectional short-run causal relationship running from 
)( tB  to )( tP  was not confirmed). Bidirectional short-run causal relationship 

between oil prices and production exists for France (the rejection of 
0: 10 H  in equation for tP  and 0: 3210  H  in equation for 



Andrzej Geise, Mariola Piłatowska 

DYNAMIC ECONOMETRIC MODELS 14 (2014) 71–91 

88

,tB  but the latter only at the 10% significance level).  In terms of long-run 
situation, a unidirectional strong causal relationship running from oil prices 

tB  to production tP  is found for both regimes in the case of Germany and 

Denmark, and in lower regime (below the threshold) in the case of France 
and the EU (rejection of 0: 110  H  and/or 0: 210  H ). This 
means that oil prices, among explanatory variables, contributes most to the 
short-run adjustment to re-establish the long-run equilibrium in the case of 
all analyzed economies.  

Conclusions  

 This paper investigated the relationship between oil prices, production 
and inflation in an asymmetric framework in selected European Union econ-
omies We tested for the threshold cointegration assuming two types of long-
run relationship: firstly, the long-run equation without a structural break and 
then the long-run relationship taking into account a structural break in 2008 
due to financial crisis. While the in the case of long-run equilibrium relation-
ship without a structural break the cointegration was not found (except Ger-
many, Nederland and the total EU), we proceeded with testing threshold 
cointegration based only on the long-run relationship with a structural break. 
The results provided evidence that production, oil prices and inflation are 
cointegrated with non-linear (asymmetric) adjustment process in the case of 
Germany, Denmark, France and the EU, i.e. for developed countries with the 
high level of economic growth. Adjustments toward the long-run relation-
ship revert faster when production is below the threshold value and tend to 
persist more when production is above the threshold. This indicate that all 
these economies have the mechanism allowing them to quickly revert to 
equilibrium after moving away from it or in other words, economies remain 
shorter period of time in the regime of slower economic activity. However, 
when estimating the threshold error correction models, the findings of 
threshold cointegration with asymmetric adjustment was fully confirmed in 
the case of France, Denmark and the total EU.  
 The estimated threshold error correction models and Granger causality 
tests provide evidence that there is unidirectional short run causality running 
form oil prices to production and inflation in the case of Germany, Denmark, 
France and the EU. Besides, the support for unidirectional long-run causal 
relationship from oil prices to production is found for Germany, Denmark 
(in both regimes) and for  France and the EU. This evidence suggests that oil 
prices play an important role in driving economic growth.  



Oil Prices, Production and Inflation in the Selected EU Countries... 

DYNAMIC ECONOMETRIC MODELS 14 (2014) 71–91 

89

 In this paper we focused on testing causality using threshold error cor-
rection model. In further studies we will concentrate on investigating the 
relationship between production, oil prices and inflation in the framework of 
threshold vector error correction model.  

References  

Albiński, P. (2014), Kryzys a polityka stabilizacyjna w Unii Europejskiej (Crisis and Stabili-
zation Policy in European Union), Oficyna Wydawnicza SGH, Warszawa. 

Arouri, M. (2011), Does Crude Oil Move Stock Markets in Europe? A Sector Investigation, 
Economic Modelling, 28, 1716–1725, 

 DOI: http://dx.doi.org/10.1016/j.econmod.2011.02.039. 
Barro, R. (1995), Inflation and Economic Growth, NBER Working Paper, No. 5326, 

www.nber.org (15.01.2015). 
Barsky, R. B., Kilian, L. (2004), Oil and Macroeconomy Since the 1970s., Journal of Eco-

nomic Perspectives, 18 (4), 115–134,  
 DOI: http://dx.doi.org/10.1257/0895330042632708. 
Belka, M. (1986), Doktryna społeczno-ekonomiczna Miltona Freadmana (Milton’s Friedman 

Socio-economic Doctrine), PWN, Warszawa. 
Bernanke, B. S., Gertler, M., Watson, M. (1997), Systematic Monetary Policy and the Effects 

of Oil Price Shocks, Brooking Papers on Economic Activity, Vol. 1997, No. 1, 91–157, 
 DOI: http://dx.doi.org/10.2307/2534702.  
Brown, S. P. A., Yucel, M. K. (2002) Energy Prices and Aggregate Economic Activity and 

Interpretative Survey, The Quarterly Review of Economic and Finance, 42, 193–208, 
 DOI: http://dx.doi.org/10.1016/S1062-9769(02)00138-2. 
Chan, K. S. (1993), Consistency and Limiting Distribution of the Least Squares Estimator of a 

Threshold Autoregressive Model, The Annals of Statistics, 21, 520–533, 
 DOI: http://dx.doi.org/10.1214/aos/1176349040. 
Davis, S. J., Halitwanger, J. (2001), Sectoral Job Creation and Destruction Responses to Oil 

Price Changes, Journal of Monetary Economics, 48, 465–512, 
 DOI: http://dx.doi.org/10.1016/S0304-3932(01)00086-1. 
EBC (2008), 10 rocznica EBC, Biuletyn miesięczny, Europejski Bank Centralny, Frankfurt 

am Main, 
 http://www.ecb.int/pub/pdf/other/10thanniversaryoftheecbmb200806pl.pdf 

(15.01.2015). 
Enders, W., Granger, C. W. J. (1998), Unit-Root Tests and Asymmetric Adjustment With an 

Example Using the Term Structure of Interest Rates, Journal of Business and Econom-
ic Statistics, 16, 304–311, DOI: http://dx.doi.org/10.2307/1392506. 

Enders, W., Siklos, P. (2001), Cointegration and Threshold Adjustment, Journal of Business 
and Economic Statistics, 19 (2), 166–176,  

 DOI: http://dx.doi.org/10.1198/073500101316970395. 
Engle, R. F., Granger, C. W. J. (1987) Cointegration and Error Correction Representation. 

Estimation and Testing, Econometrica, 55, 251–276, 
 DOI: 10.1111/j.1368-423X.2005.00149.x. 
Fosten, J., Morley, B., Taylor, T. (2012), Dynamic Misspecification in the Environmental 

Kuznets Curve: Evidence From CO2 and SO2 Emissions in the United Kingdom, Eco-
logical Economics, 76, 25–33,  

 DOI: http://dx.doi.org/10.1016/j.ecolecon.2012.01.023. 



Andrzej Geise, Mariola Piłatowska 

DYNAMIC ECONOMETRIC MODELS 14 (2014) 71–91 

90

Friedman, M. (1976), Inflation and Unemployment, Nobel Memorial Lecture, 
http://www.nobelprize.org/nobel_prizes/economic-sciences/laureates/1976/friedman-
lecture.pdf (15.01.2015). 

Hamilton, J. D. (1983), Oil and the Macroeconomy Since World War II,  Journal of Political 
Economy, 91, 228–248, DOI: http://dx.doi.org/10.1086/261140. 

Hamilton, J. D. (1996), This Is What Happened to Oil Price-Macroeconomy Relationship, 
Journal of Monetary Economics, 38, 215–220, 

 DOI: http://dx.doi.org/10.1016/S0304-3932(96)01282-2. 
Hooker, M. (1996), What happened to the oil price-macroeconomy relationship?, Journal of 

Monetary Economics, 38, 195–213, 
 DOI: http://dx.doi.org/10.1016/S0304-3932(96)01281-0. 
Jimenez-Rodriguez, R., Sanchez, M. (2005), Oil Price Shocks and Real GDP Growth: Empir-

ical Evidance For Some OECD Countries, Applied Economics, 37, 201–228, 
 DOI: http://dx.doi.org/10.1080/0003684042000281561. 
Lardic, S., Mignon, V. (2006), The Impact of Oil Prices on GDP in European Countries: An 

Empirical Invetigation Based on Asymmetric Cointegration, Energy Policy, 34, 3910–
3915, DOI: http://dx.doi.org/10.1016/j.enpol.2005.09.019. 

Lutz, C., Mayer, B. (2009), Economic Impacts of Higher Oil and Gas Prices. The Role of 
International Trade for Germany, Energy Economics, 31, 882–887, 

 DOI: http://dx.doi.org/10.1016/j.eneco.2009.05.009. 
Mork, K. A. (1989) Oil and the Macroeconomy When Prices Go Up and Down: an Extention 

of Hamilton’s Results, Journal of Political Economy, 91, 740–744, 
 DOI: http://dx.doi.org/10.1086/261625. 
Mork, K.A., Olsen, O., Mysen, H.T. (1994), Macroeconomic Responses to Oil Price Increases 

and Decreases in Seven OECD Countries, Energy Journal, 15, 19–35, 
 DOI: http://dx.doi.org/10.5547/ISSN0195-6574-EJ-Vol15-No4-2. 
Papapetrou, E. (2001), Oil Price Shocks a, Stock Market, Economic Activity and Employ-

ment in Greece,  Energy Economics, 23 (5), 511–532, 
 DOI: http://dx.doi.org/10.1016/S0140-9883(01)00078-0. 
Petruccelli, J., Woolford, S. W. (1984), A Threshold AR(1) Model, Journal of Applied Prob-

ability, 21(2), 270–286, DOI: http://dx.doi.org/10.2307/3213639. 
Scholtens, B., Yurtsever, C. (2012), Oil Price Shocks and European Industries, Energy Eco-

nomics, 34(4), 1187–1195, DOI: http://dx.doi.org/10.1016/j.eneco.2011.10.012. 
Yau, H-Y., Nieh, C. (2009), Testing for Cointegration With Threshold Effect Between Stock 

Prices and Exchange Rates in Japan and Taiwan,  Japan and the World Economy, 
21(3), 292–300, DOI: http://dx.doi.org/10.1016/j.japwor.2008.09.001. 

Ceny ropy naftowej, produkcja oraz inflacja w gospodarkach  
Unii Europejskiej - podejście kointegracji progowej 

Z a r y s  t r e ś c i. W artykule wykorzystuje się podejście kointegracji progowej, które zosta-
ło opracowane przez Endersa i Siklosa (2001) w celu zbadania zależności między cenami 
ropy a aktywnością gospodarczą (wyrażoną przez produkcję i inflację) z uwzględnieniem 
wystąpienia zmiany strukturalnej dla sześciu wybranych państw UE. Podejście to pozwala na 
różną szybkość dostosowań do długookresowej równowagi w zależności, czy produkcja jest 
powyżej czy poniżej długookresowej zależności. Ponadto, dla zbadania krótko- i długookre-
sowej przyczynowości między produkcją, cenami ropy i inflacją, zostały oszacowane progo-



Oil Prices, Production and Inflation in the Selected EU Countries... 

DYNAMIC ECONOMETRIC MODELS 14 (2014) 71–91 

91

we modele korekty błędem. Otrzymane wyniki wskazują, że dla Francji, Danii i całej Unii 
Europejskiej zmienne te są skointegrowane, przy czym proces dostosowawczy ma asyme-
trycznych charakter.  

S ł o w a  k l u c z o w e: asymetryczne dostosowania, cenowe szoki naftowe, kointegracja 
progowa, nieliniowość, progowy model korekty błędem. 



 

 


