• No results found

The effect of the recent low oil price on the real GDP and inflation of the Netherlands

N/A
N/A
Protected

Academic year: 2021

Share "The effect of the recent low oil price on the real GDP and inflation of the Netherlands"

Copied!
28
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

UNIVERSITY OF AMSTERDAM

THE EFFECT OF THE RECENT LOW OIL PRICE

ON THE REAL GDP AND INFLATION OF THE

NETHERLANDS

BACHELOR THESIS ECONOMICS AND FINANCE

NAM E: ARTHUR ROKEBRAND STUDENT NUM BER: 10581308

SUPERVISOR: RON VAN M AURIK DATE: 28-06-2016

(2)

ABSTRACT

In this thesis, the effect of the current low oil prices on the real GDP and inflation of The Netherlands are studied. Two time-series models are used to estimate the direct effect and the lagged effect per quarter of an oil price shock. One model for the effect on real GDP and one model for the effect on inflation. According to the results, it seems that not only the direct effects are relevant but also the lagged effects of an oil price shock. Real GDP is negatively related to the oil price and inflation is positively related to the oil price according to the results. With the results of the regression on the time-series model, an estimation can be done by how much the real GDP and inflation of The Netherlands are effected after the recent oil price slump. Statement of Originality This document is written by Student Arthur Rokebrand who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

TABLE OF CONTENTS

INTRODUCTION ... 4 1. LITERATURE REVIEW ... 6 1.1 STUDIES ABOUT THE EFFECT ON THE REAL GDP AND INFLATION ... 6 1.2 STUDIES ABOUT THE EFFECT ON THE REAL GDP ... 9 1.3 STUDIES ABOUT THE EFFECT ON INFLATION ... 11 2. DATA ... 12 3. METHODOLOGY ... 15 4. RESULTS ... 18 4.1 MULTIPLE REGRESSION ON TIME-SERIES MODEL ... 18 4.2 ESTIMATION OF THE EFFECTS OF THE RECENT OIL PRICE SLUMP. ... 21 5. CONCLUSION ... 22 REFERENCES ... 24 APPENDICES ... 26

(4)

INTRODUCTION

According to the historical prices of crude oil, the oil price dropped with almost 70% in less than two years. On June the first 2014 the price per barrel of crude oil was around $105 and on January the first 2016 the price of oil was around $34 per barrel (investing.com, 2015). Almost all economic activities are based on crude oil, which supplies around 40% of the world’s total energy needs (Gómez-Loscos, Gadea and Montañés, 2012). Shocks in crude oil prices do not only affect energy markets but also have an effect on the rest of the economy. Production costs and, therefore, prices and the gross domestic product (GDP) are affected because of the high weight of oil inputs in production (Forni, Gerali, Notarpietro and Pisani, 2015). Oil price fluctuations can have different effects on real GDP growth and inflation. It depends whether the country is a net oil importer or exporter because an oil price shock redistributes income between net oil importing and net oil exporting counties. The Netherlands is a net oil importing country (Cuñado and Pérez de Gracia, 2003). The transmission channel through which oil price shocks affect real GDP includes both supply and demand channels. Crude oil is a basic input for production and lower oil prices are often associated with increased availability of oil, leading to higher output for firms and thus higher supply (Jiménez-Rodrigues and Sánchez, 2004). The demand side is affected because of the indirect effect on consumption. When oil prices decrease, disposable income is increased, which increases consumption and thus the demand (Jiménez-Rodrigues and Sánchez, 2004). According to Morana, it is unclear whether the recent oil price slump will stimulate economic growth in oil importing countries. On the one hand, energy costs are reduced which increases total factor productivity and that could stimulate economic growth (Morana, 2016). On the other hand, deflation might be a consequence which discourages economic growth (Morana, 2016). In this thesis, the direct and the lagged effects of crude oil price changes on the real GDP and inflation in The Netherlands are studied to make an estimation what the consequences are of the recent oil price drop on these macroeconomic variables. Two time-series models are used to make this estimation. The lagged effects of an oil price shock on the real GDP and inflation have been studied before but not for The

(5)

Netherlands and with different models. So this thesis is the first, analyzing the lagged effects of oil price shocks on the real GDP and inflation for The Netherlands. First the literature review is discussed. This chapter consists of three sections. In the first section, articles are discussed which investigated the effect on inflation and the real GDP. Next the articles which study the effect only on the real GDP are discussed and finally the articles which studied the effect on inflation will be discussed. The second chapter is about the data used in this thesis. Where the data is retrieved from and an explanation about the variables from which data is found. In the third chapter the methodology will be discussed. In this part, models are discussed that are used and some tests which are conducted in Stata are discussed. In the fourth chapter, the results will be discussed. This chapter is split up into two sections. In the first section the results of the time-series models used in this thesis will be presented and discussed. These results are compared with the results of other articles. In the second section an estimation will be done about the effects of the recent oil price slump. In the final chapter the conclusion is done.

(6)

1. LITERATURE REVIEW There has been done a lot of research about the macroeconomic consequences of a low oil price and how economic growth and inflation are affected. Economic growth is measured as the percentage rate of increase in real GDP. Firstly, the papers which study the effect of an oil price shock on both the GDP and inflation will be discussed. After that, the papers which study the effect of an oil price shock only on the real GDP will be discussed. Finally, the articles about the effect of an oil price shock only on inflation will be discussed. Basnet and Upadhyaya (2015), Roeger (2005), Forni, Gerali, Notarpietro and Pisani (2015) and Gómez-Loscos, Gadea and Montañés (2012) evaluate the effect of an oil price shock on real GDP and inflation. Mork and Olsen (1994), Jiménez-Rodrigues and Sánchez (2004) and Lardic and Mignon (2005) study the effect of oil price increases and decreases on real GDP. Cuñado and Gracia (2003) and Castro, Jerez and Barge-Gil (2015) analyze the effect on inflation rate. 1.1 STUDIES ABOUT THE EFFECT ON THE REAL GDP AND INFLATION The primary purpose of the article of Basnet and Upadhyaya (2015) was to estimate and analyze the impact of an oil price shock on the real GDP, inflation and the exchange rate in the ASEAN-5 countries. For the data of oil, they used the spot market price of West Texas Intermediate crude oil, which is the most used measure for world oil prices (Basnet and Upadhyaya, 2015). They used quarterly data from 1970:1 to 2010:2. In their article a structural VAR model is devolved so that the effect of oil price fluctuations on the real GDP, inflation and the real exchange rate can be estimated. They also used a co-integration test to see whether the real GDP, inflation and the real exchange rate share a common trend in the long run. All countries except Indonesia are oil importing countries, so the other four countries: Malaysia, The Philippines, Singapore and Thailand are highly dependent on oil imports (Basnet and Upadhyaya, 2015). Before Basnet and Upadhyaya (2015) estimated the SVAR model, they first conducted two unit-root tests, namely the Augmented Dickey fuller test and a Phillips-Perron test, to analyze if the data of the variables are non-stationary. Data points of

(7)

macroeconomic time-series are almost always non-stationary, meaning that means, variances and covariance’s changes over time. Non-stationary data cannot be modelled and have to be transformed into stationary data, otherwise the results will be spurious (Basnet and Upadhyaya, 2015) Their test results of the Phillips-Perron test confirms that the data series are non-stationary and thus must be transformed into stationary data series. After that, they proved the long term relationship between the real GDP, inflation and the real exchange rate. They did this by running a Johansen’s co-integration test. This test ensured them to identify the number of common trends among the variables. Thereafter, they conducted an impulse response test of an oil price shock. With this test they were able to analyze the effect of an oil price shock on the real GDP, inflation and the real exchange rate. They found for the oil exporting country, Indonesia, that they experienced an expansionary effect because of the positive oil price shock. They also found that a positive oil price shock contributed to a temporary inflationary effect. They observe that the price level decreases after the first quarter, followed by a sustained rise. The effect of a positive oil price shock on the real GDP of Malaysia is almost the same as the effect on the real GDP of Singapore. They observe that the immediate response of the real GDP to a positive oil price shock is positive for both countries. The growth of the real GDP, however, will not hold in the longer period. After five quarters the effect is not significant anymore for both counties. In the case of The Philippines and Thailand they found a contrary effect relative to Malaysia and Singapore after a positive oil price shock on the output. The effect on the real GDP starts in the first quarter showing a short term positive effect. After the first quarter the shock generates a contractionary impact, which is negative and significant throughout the ten quarters. In the case of all the four oil importing countries a positive oil price shock has an inflationary effect but the effect begins to decrease after the first quarter. They finally conclude that an oil price shock is not the major impediment of economic growth and inflation in these five countries. Because their findings do only impact on the short run. Roeger (2005) uses a DSGE model to analyze the short and the long run effects of a permanent oil price increase on real GDP and inflation in oil importing countries of the Euro area. In 2004 the oil price increased with almost 50%. There were fears this shock

(8)

would cause stagflation. Stagflation is a situation in which the inflation rate is high, the economic growth slows down and unemployment stays high. In his paper he first analyzes and simulates the effects of a permanent oil price increase predicted by the QUEST model and then compare the results with other models. The Quest model is a global macroeconomic model in the New-Keynesian tradition which assumes that households and firms have rational expectations (Roeger, 2005). In the QUEST model a Cobb Douglas framework is used which allows to distinguish between the direct effects and the indirect effects of an oil price shock. The direct effects of an oil price shock are the passing on of oil prices to producer prices and the CPI deflator but also the markup adjustment of the supply and demand changes. The indirect effects result from the response of wages, the reaction of monetary policy and changes of the exchange rate after a change in prices. The Quest model was used to simulate a permanent oil price increase of 50% in the Euro area. According to the results, the GDP is reduced with 0.5% after one year. In the long run, the GDP is reduced with 0.9% relative to the GDP before the oil shock. Prices increased with 0.3% after one year and 1.2% in the long run. Roeger (2005) also simulated the same oil price shock on four other models. These were the results: Price Level Change (%) GDP change (%)

Year 1 Year 2 Year 3 Year 1 Year 2 Year 3

Quest 0.3 0.5 0.6 -0.5 -0.6 -0.7 AWM 0.5 0.9 1.0 -0.1 -0.3 -0.4 NIGEM 0.3 0.5 0.5 -0.8 -0.8 -0.7 Interlink 0.6 0.8 0.9 -0.4 -0.2 0.1 Multimot 1.6 2.7 3.2 -0.1 -0.4 -0.3 The results of other simulations are pretty much alike. Some of the differences are model specific and depend on how oil is treated in the models. According to Roeger (2015) some of these models more emphasize the fact that oil is an intermediate product for production and thus take into account the efficiency losses associated with substitution.

(9)

Forni, Gerali, Notarpietro and Pisani (2015) evaluated the macroeconomic effects of oil price shocks on the Euro Area by developing and estimating with Bayesian methods an open economy DSGE model. They used quarterly data for the period 1995:1-2012:4 and found that a 10% increase in the international price of oil generates a decrease of 0.1% of the GDP in the Euro Area as a whole. The Euro area is considered as an oil importing Area. Forni, Gerali, Notarpietro and Pisani (2015) did not only evaluate the effect of an oil price shock on the GDP but also on inflation. They used the same model and found that a 10% increase in the international price of oil generates an increase of 0.1 annualized percentage points in the EA consumer price index inflation. The article of Gómez-Loscos, Gadea and Montañés (2012) analyses the responses of an oil price shock on the real GDP and inflation of the G7 countries between 1970:1 and 2008:4, identifying different periods. They proved the existence of three breaks and thus identified four different periods for their sample. For these four periods they estimated the influence of an oil price shock on the real GDP and inflation. They found that the evidence of a temporary reduction of the real GDP and prices becomes weaker from 1970 to the 1990s. In the last period, from 2000, they found that the impact of oil prices on the GDP and inflation recovers some of its initial importance. They could also confirm that the variability of oil prices of the last period they investigated had only a minimum effect on prices. 1.2 STUDIES ABOUT THE EFFECT ON THE REAL GDP In the paper of Mork and Olsen (1994), they investigate the correlations between oil price movements and GDP fluctuations for seven OECD countries. They found that the correlations between oil price increases are negative and significant for most countries but positive for the oil exporting country, Norway. The correlations with oil price decreases and the real GDP are mostly positive but only significant for the United States and Canada. In their paper they seek to answer three questions: 1. Does the negative correlation persist in data series extending through 1992? 2. Is the correlation pattern the same for price decreases as for increases? 3. Does the correlation pattern vary from country to country?

(10)

Their first step in the paper was to analyze the bivariate correlations between the real GDP and the oil price. They did that by estimating a regression equation with GDP growth as independent variable and lagged values of GDP growth and oil price as dependent variables. Their estimation period was between 1967:3 to 1992:4. They found out that the first question seems to be true and thus persist in data series extending through 1992. The correlations seem not to be significant for all countries but they think this is because of measurement errors. They answered the second question by comparing the correlations of oil price increases and decreases and concluded that for most countries the coefficients of oil price increases seem to be the opposite sign of oil price decreases. The last question seems to be true because of the differences in the countries net export position which influences the oil price - real GDP correlation. The paper of Jiménez-Rodrigues and Sánchez (2004) assesses empirically the effect of oil price increases and decreases on the real GDP in the main industrialized countries. They used multivariate vector auto regression analysis and found that among most industrialized OECD countries, oil price increases, negatively affect the real GDP. Both linear and non-linear models were used and they distinguished between oil importing and oil exporting countries. The next step in this paper was to conclude that the interaction between the oil price and macroeconomic variables was significant. To conclude this, they used the Granger Causality type test. They found that the effect of an oil price increase on the real GDP differs from the effect of an oil price decrease. This means that oil prices do not have symmetric effects on the real GDP which could be evidence against the linear approach. In their paper they compared different models and they finally found that there is evidence of a non-linear effect of oil price shocks on the real GDP in oil importing and exporting countries. They used three non-linear models and found that the scaled specification is better than the other two models, namely the asymmetric and the net specification. With the scaled specification it is possible to compare the effects of an oil price rise and fall. For the multivariate vector auto-regression model used in the paper of Jiménez-Rodrigues and Sánchez (2004) they used quarterly data of seven variables: real GDP, Real effective exchange rate, real oil price, real wage, inflation and short and long term interest rates where real GDP, real effective exchange rate, real oil price and real wage

(11)

are expressed in logs. In most cases they found that oil price increases have a significant negative effect on the real GDP of oil importing counties. The main goal of the paper of Lardic and Mignon (2005) was to investigate if there is a long term relationship between real economic growth and the oil price in some European countries including The Netherlands. First they discussed the transmission channels through which oil prices can affect economic activity. Thereafter they applied two standard root test on all countries individually to test if there is a unit-root. Like expected they found that for all countries there is a root and so the process depends on time with an ! of 10%. 1.3 STUDIES ABOUT THE EFFECT ON INFLATION The aim of Cuñado and Gracia (2003) was to analyze the relationship between the oil price and some macroeconomic variables in 15 European countries including The Netherlands. They analyzed this by using the Granger causality and structural stability test on the relationship between the oil price and inflation and between the oil price and the production growth rate. They used quarterly data between 1960:1 and 1999:1. As a first step of their empirical analysis they did a unit-root test on all the variables. After that, they used a sequential augmented Dickey Fuller test to detect structural changes because the results of the first unit-root tests suggest that the null hypotheses can’t be rejected for all variables. After they tested for co-integration allowing for structural breaks, they obtained a long term relationship between the oil price and inflation rate for most countries. Unfortunately, they did not find this relationship in The Netherlands. Castro, Jerez and Barge-Gil (2015) also studied the relationship between the oil price and inflation in the Euro area with the focus on the recent oil price slump. With the falling oil prices in the past few years, the main concern in the Euro Area is its deflationary effect. In their paper they evaluated the effect of oil price changes on inflation under different oil price scenarios. The second contribution of their paper was to use a model-based indicator of inflation to track the risk of deflation. A time-series model was used with which twelve months ahead forecasts for inflation in the Euro area were computed for three different oil price scenarios. With this model they were also able to estimate which part of inflation fluctuations can be assigned to oil price fluctuations.

(12)

Castro, Jerez and Barge-Gil (2015) finally observed that negative inflation is not expected for the twelve months ahead forecast for any of the three scenarios. But the short time effect of an oil price shock is important. 25% of the variance of inflation changes is because of oil price fluctuations (Castro, Jerez and Barge-Gil, 2015). Deflation in the long term could only happen if consumers their expectations are affected after a long period of deflation so that economic activity is affected. 2. DATA In this thesis quarterly data is used between 1995:1 to 2015:4. So there are 83 observations of all variables, except for oil. For oil 92 observations are used, with the first observation starting in 1993:1. The data must be stationary, otherwise it cannot be modeled. In the paper of Basnet and Upadhyaya (2015), Lardic and Mignon (2005) and Cuñado and Gracia (2003) standard unit-root tests were applied to test whether the data are stationary or non-stationary. In this paper also unit-root tests are applied, namely Phillips Perron tests. This test is applied because a unit-root can cause problems in time-series models and with this test, the null-hypothesis is tested whether a unit-root is present in de model. If the null-hypotheses is not rejected, then the variables would be non-stationary with means, variances and covariance’s that changes over time. Non-stationary data have to be transformed into stationary data. In this thesis non-stationary data are transformed to stationary data by transforming the data into growth rates with the formula: "#− "#%& "#%&

Where "# is the value of the variable at time ' and "#%& is the value of a variable at time ' − 1, where t is in quarters. The results of the Phillips-Perron test can be found in the appendix table 3. The summary statistics of the absolute variables and the summary statistics of the growth values of the variables can be found in appendix table 1 and 2. Oil: The data of oil is retrieved from the database of investing.com (2016). West Texas intermediate crude oil prices are used and the monthly prices are converted into quarterly prices. These oil prices are considered as the benchmark of the world oil prices (Basnet and Upadhyaya, 2015).

(13)

The observations of the oil prices start two years before the observations of the other variables. This is necessary because the delayed effects of the crude oil price on the real GDP and inflation are analyzed. With the observations starting eight quarters before the observations of other variables, eight lags can be captured. The oil price shows an upward trend when time passes so the option trend must be used in STATA when conducting the Phillips Perron test. The observations of the oil prices are transformed from non-stationary data to stationary data because according to table 1 in de appendices, the null hypothesis of whether a unit-root is present in the absolute values of the oil price is not rejected. After transforming the data, the null hypothesis is rejected at a significance level of 1% and data are thus transformed to stationary data. GDP: The data of the gross domestic product of The Netherlands are retrieved from CBS statistics Netherlands (2016). Real GDP is recorded since 1995:1. That is the reason why the observations of the variables start then. The reason why quarterly data are used, is because data on the real GDP in The Netherlands are only available per quarter. Real GDP of The Netherlands also shows an upward trend when time passes, so for this variable the option trend is also used with the Phillips-Perron test. According to the test results of the Phillips-Perron test in the appendices, the observations of the real GDP are also non-stationary and must be transformed into growth rates. The growth rates are stationary at a significance level of 1%. Inflation: For the data on inflation in The Netherlands, the consumer price index is used and is recorded on monthly basis. This is transformed to quarterly data by using the data points of the last month of each quarter. The data on inflation are retrieved from CBS statistics Netherlands (2016). The consumer price index also shows an upward trend. According to the Phillips-Perron test results, the absolute values of the consumer price index were already stationary before transforming the values into growth rates. This was not expected because in the paper of Before Basnet and Upadhyaya (2015), Lardic and Mignon (2005) and Cuñado and Gracia (2003), data on inflation was non-stationary and had to be transformed to stationary data. For that reason, the values were still transformed into growth rates in this thesis. The reason of this unexpected result could be because maybe not enough observations were used.

(14)

Explanatory variables: For the regression on the real GDP of The Netherlands, the explanatory variables used beside oil are: consumption, investment, government spending, imports and exports. Consumption is the final consumption of households. The data of this variable are retrieved from CBS Statistics Netherlands (2016) and are seasonally adjusted. With investment, gross capital formation is meant. Gross fixed capital formation consists of producers’ acquisitions less disposals of fixed assets. These data are retrieved from CBS statistics Netherlands (2016) and are also seasonally adjusted. With government spending, final consumption expenditure by general government that results from the specific recording of government output is meant. These data are also seasonally adjusted and retrieved from CBS statistics Netherlands (2016). The data on imports and exports of The Netherlands are also retrieved from CBS Statistics Netherlands (2016) but are not seasonally adjusted. For the regression on inflation in The Netherlands, these are the variables used besides oil: unemployment rate, money supply, interest rate, wages and real GDP. The data of the harmonized unemployment rate are retrieved from Eurostat (2016) and are seasonally adjusted. The hourly wage is seen as gross hourly earnings in this paper. The data are retrieved from CBS Statistics Netherlands (2016) and are not seasonally adjusted. With money supply, M3 money is meant. The data are retrieved from DNB De Nederlandsche Bank (2016) and the prices are not seasonally adjusted. The interest rate is the interbank 3 month rates, retrieved from DNB (2016) and not seasonally adjusted. With wages, the gross hourly earnings are meant. The data are retrieved from CBS Statistics (2016) and prices are not seasonally adjusted. According to the test results of the Phillip-Perron test in table 1 in the appendices, all the absolute values of the variables were non-stationary and had to be transformed into stationary data. After the growth rates were calculated, the data seems to be stationary at a significance level of 1%. For all the explanatory variables, a trend was visible except for unemployment and the interest rate. Therefore, the option trend was not used in the Phillips-Perron test on these two variables.

(15)

3. METHODOLOGY In this thesis, the delayed effects of a changing oil price on the real GDP and inflation of The Netherlands will be analyzed with a time-series model. With this model future values can be predicted based on previously observed values. With the results of a multiple regression on these models, it will be possible to estimate by how much the real GDP and inflation are affected in The Netherlands after the recent oil price slump. before this model is used, the optimal number of lags have to be obtained and there have to be tested if the lagged values of oil provide significant information about the real GDP and inflation of The Netherlands. The optimal number of lags is obtained by comparing the Adjusted R-squares of models with a different number of lags for oil and control variables for inflation and GDP which are also used in the time-series models. The adjusted R-squares are compared for a model without lags to eight lags of oil. The values of the adjusted R-squares can be found in the appendix table 4. In the model of the real GDP, it seems that after adding the fourth lag together with the control variables, the adjusted R-squared is highest. For inflation the optimal number of lags is five because then the adjusted R-squared is highest. Next, a Granger Causality test is conducted. Jiménez-Rodrigues and Sánchez (2004) and Cuñado and Gracia (2003) also used this test to test whether a time-series is useful forecasting another time-series. Granger Causality tests are done to test whether the lagged values of oil provide statistically significant information about the future values of the real GDP and inflation (Cuñado and Pérez de Gracia, 2003). In this thesis the null hypotheses are tested if four lagged values oil do not cause GDP growth in The Netherlands and if five lagged values of oil do not cause inflation growth in The Netherlands. In STATA, Granger Causality Wald tests are used and gave the following results: GDP )ℎ+, 4 = 18.9760 5678 > )ℎ+, = 0.001 Inflation )ℎ+, 5 = 9.4314 5678 > )ℎ+, = 0.093 The tests use four and five lags and therefore four and five degrees of freedom are used. According to the test result of the GDP, the null-hypothesis is rejected at a

(16)

significance level of 1% because the p-value is smaller than 0.01. The null-hypothesis of the test on inflation can only be rejected at a level of 10% because the p-value is smaller than 0.10 but bigger than 0.01 and 0.05. So it seems that oil Granger Cause inflation and real GDP and thus provides information about future values of the real GDP and inflation in The Netherlands at a significance level of 10%. Predictions of the value of the real GDP and inflation based on its own values and on the past values of oil are thus better than predictions of real GDP and inflation based only on its own past value. Before introducing the time-series models, endogeneity must be tested and controlled. The variable oil and the lagged variables of oil are all treated as independent variables in the time-series models. If the variable oil seems to be endogenous, meaning that this variable is a function of other variables presented in the model, then the time-series models have to be adjusted because an instrumental variable is then necessary to use. In this thesis, it is tested whether the variable oil is determined by the lagged variables of oil en thus whether oil is endogenous. In STATA a Durbin-Wu-Hausman test for endogeneity is performed. This is the result: <(1, 75) = 0.99 5678 > < = 0.3229 According to this test result, the null hypothesis that oil is exogenous and no IV regression is necessary, is not rejected because the p-value is higher than 0.10. So it can be assumed that the variable oil is not endogenous. A hetroscedasticity test was also necessary to conduct to find out if the option robust must be used in the time-series models. The Breusch-Pagan test was used and tests whether the estimated variance of the residuals from a regression are dependent on the values of the independent variables. This test gave the following results: GDP )ℎ+, 6 = 26.67 5678 > )ℎ+, = 0.0002 Inflation )ℎ+, 6 = 2.06 5678 > )ℎ+, = 0.9142 Six degrees of freedom were used because six explanatory variables are used. According to these results, the null hypothesis of homoscedasticity can only be rejected in the first test. In the time-series model of GDP, hetroscedasticity is present so the option robust must be used in this case. In the time-series model of inflation, the null-hypothesis cannot be rejected. So the residuals are homoscedastic in this case.

(17)

In the articles of Basnet and Upadhyaya (2015), Roeger (2005) and Castro, Jerez and Barge-Gil (2015) the delayed effect of an oil price shock is also analyzed. Now that is proved that the data are stationary and that the optimal number of lagged values of oil provides information about future values of the real GDP and inflation, the time-series model can be introduced. The two models used to estimate the direct effect of an oil price shock on the real GDP and inflation and the effect after the optimal number of lags are as following:

AB5#= CD + C&G+H#+ C,I. G+H#%&+ CJI2. G+H#%,+ CKI3. G+H#%J+ CLI4. G+H#%K+ CMNOP# + CQAGP#+ CR)GO#+ CSNT5#+ C&DU"5#+ V#

NO<#= CD + C&G+H#+ C,I. G+H#%&+ CJI2. G+H#%,+ CKI3. G+H#%J+ CLI4. G+H#%K

+ CMI5. G+H#%L+ CQWOU#+ CRXYA#+ CSAB5#+ C&DNOZ#+ C&&TO[#+ V# The dependent variables, the real GDP of The Netherlands (GDP) and inflation in The Netherlands (INF) are explained by these variables: Oil = Crude oil return Ln.Oil = Crude oil return with n months lag INV = Investment growth in The Netherlands GOV = Government spending growth in The Netherlands CON = Consumption growth in The Netherlands IMP = Imports growth in The Netherlands EXP = Exports growth in The Netherlands UNE = Unemployment growth in The Netherlands HWG = Hourly wage growth in The Netherlands GDP = Real gross domestic product of The Netherlands MNS = Money supply growth in The Netherlands. In STATA, two regressions will be run on the two models. To conclude if the lags are significant, the p values are being sought. If the independent variables are significant, meaning that the p value is smaller than 0.10 in this thesis, then it will be able to estimate the lagged effect of an oil price shock on the real GDP and inflation.

(18)

4. RESULTS In this section the results are discussed. First the results of the multiple regressions on the time-series models will be analyzed and discussed. In the second section an estimation will be done about the effect of the recent oil price slump on inflation and the real GDP in The Netherlands. 4.1 MULTIPLE REGRESSION ON TIME-SERIES MODEL After having proved that all the transformed variables are stationary and that the variable oil, which evolves over time, Granger-Causes the real GDP and inflation, an estimation can be done about how big the effect of a changing oil price is, on the real GDP and inflation with the optimal level of lags. The two time-series models described in the methodology are used to make this estimation. Two multiple regressions were done in STATA and the option robust was used with the regression on GDP. These are the results:

GDP Coefficient (t-value) p-value Inflation Coefficient (t-value) p-value

Consumption 0.4739*** (5.65) 0.0000 Unemployment 0.0275*** (2.63) 0.0100 Investment 0.1175*** (4.70) 0.0000 Wage 1.4037*** (7.61) 0.0000 Gov Spending 0.2267*** (7.34) 0.0000 GDP 0.1315 (1.33) 0.1880 Imports -0.2008*** (-2.79) 0.0070 Interest 0.0000 (0.00) 0.9980 Exports 0.3212*** (4.68) 0.0000 Money supply -0.1272 (-1.52) 0.1320 Oil -0.0112*** (-3.39) 0.0010 Oil 0.0097* (1.74) 0.0850 L.Oil 0.0085*** (2.64) 0.010 L.Oil -0.0057 (-0.98) 0.3320 L2.Oil 0.0054* (1.90) 0.0620 L2.Oil 0.0012 (0.20) 0.8420 L3.Oil -0.0083** (-2.48) 0.0160 L3.Oil -0.0021 (-0.36) 0.7210 L4.Oil 0.0053 (1.41) 0.1640 L4.Oil 0.0151** (2.54) 0.0130 L5.Oil -0.0072 (-1.19) 0.2370 Notes: *, **, *** indicate significant at 10%, 5% and 1%

(19)

In the regression on the GDP, the fourth lag is not significant at a level of 10% because the p-value is higher than 0.10. In the regression on inflation, the first, second and third lags are not significant. According to these results, the direct effect of a positive oil price shock on the real GDP in The Netherlands is negative. After one and two quarters this same shock will increase GDP and after three quarters the effect will be negative again. The direct effect of a positive oil price shock on inflation in The Netherlands is positive. After four quarters this same shock will increase inflation again. Basnet and Upadhyaya (2015) analyzed the effect of a positive oil price shock on five different countries and found similar results on the GDP for two oil importing countries, The Philippines and Thailand. They observed that for both countries the first effect of a positive oil price shock on real GDP is negative. Then the effect will be positive for one quarter and next, the effect will be negative again. The effect will remain negative and significant for ten quarters. For the other two oil importing countries they found an opposite effect. They observe that the real GDP is positively affected after a positive oil price shock and this effect remains significant throughout six quarters. They argue that a possible reason for the temporary positive effect could be that companies increase the sales price before the production costs are increased because of higher oil prices. A possible reason why the direct effect on GDP and the effect after one, two and three quarters in this thesis are significant and not after the fourth quarter could be because a different model is used or that maybe not enough observations were used. Basnet and Upadhyaya (2015) found for all four oil importing countries a positive effect on inflation after a positive oil price shock but that prices begin to show lesser impact as time passes. The results in this paper also show a positive effect after a positive oil price shock but prices do not show less impact as time passes. Only the direct effect and the effect after four quarters are significant and the coefficient is higher after four quarters. A reason why the effect is not significant after the first three quarters could be because not enough independent variables or not enough observations are used. The adjusted R-squared of the regression on inflation is only 0.5070. This relative low R-squared could also be an explanation for the growing impact after four quarters instead of a decreasing impact. Forni, Gerali, Notarpietro and Pisani (2015) found that a 10% increase in the oil price decreases the real GDP with 0.10% in the Euro area. Even though a different model

(20)

is used, according to the results in this paper, real GDP also decreases but only with approximately 0.056% after summing up the total effect and the lagged effects. Roeger (2005) simulated an oil price shock of a 50% price increase and analyzed the effect per year. He found that an oil price increase of 50% should contribute to a decrease in the real GDP in the Euro area of 0.1%, which is lower than the results of this paper which state that real GDP in the Netherlands should decrease with 0.28% after this shock. Roeger (2005) compared his results with four other models. The results of a 50% oil price increase on real GDP diverge between -0.1% and -0.8% after one year according to the other four models. So the findings in this paper on the real GDP are similar to other papers. Forni, Gerali, Notarpietro and Pisani (2015) also found a positive relationship between the oil price and inflation. They found that a 10% increase in the oil price increases inflation with 0.10 percentage point in the Euro area. According to the results in this paper, inflation in The Netherlands increases with 0.248% after the same shock. Roeger (2005) found that after an oil price increase of 50%, the price level in the Euro area increased with 0.5% after one year. According to the results in this paper, the price level should have been increased with 1.24% one year after the same shock. Even though a same positive relationship was found in these papers, the effect on inflation in The Netherlands seems to be much higher compared to the inflation in the Euro Area as a whole. An explanation why the results are a bit different relative to other articles could be because the effect only on The Netherlands was investigated and not the effect on the Euro Area as a whole like in the articles of Forni, Gerali, Notarpietro and Pisani (2015) and Roeger (2005). The inflation and GDP growth differs over the countries in the Euro area. In almost every article a negative relation was found between the oil price and the GDP in oil importing countries. In this paper the effect on GDP after summing up the direct effect and the lagged effect was also negative after a positive oil price shock, even after the unexpected positive effect one and two quarters after the shock. A possible explanation for this unexpected temporary positive result was found in the article of Brown and Yücel (2008). They analyzed the relationship between the oil price and the gas price and found a positive relationship between these prices. Except that The

(21)

Netherlands is a net importer of oil, their net export position of gas can cause results that would be expected in oil exporting countries. In this thesis no distinction was made between oil price decreases and increases like in the paper of Jiménez-Rodrigues and Sánchez (2004). They found that the effect of an oil price increase on the real GDP differs from the effect of an oil price decrease. If no distinction was made in the models, results can be insignificant. 4.2 ESTIMATION OF THE EFFECTS OF THE RECENT OIL PRICE SLUMP. Between the start of third quarter of 2014 and the end of the fourth quarter of 2015, the oil price decreased sharply. In the article of Castro, Jerez and Barge-Gil (2015), they were able to estimate which part of inflation fluctuations can be assigned to oil price fluctuations. In this part of the thesis, an estimation is done which part of the real GDP and inflation changes can be assigned to the recent oil price shock. This estimation can be done by multiplying the oil price returns in a period with the coefficient of the direct effect and the coefficient of the relevant lag of the time-series models. These are the results of this simulation:

Quarter Oil price ($) Return GDP growth (%) Inflation growth (%)

2014q2 99.74 -0.0157 0.0176 -0.0153 2014q3 98.17 -0.1796 0.1878 -0.1742 2014q4 80.54 -0.4010 0.2880 -0.3890 2015q1 48.240 0.2361 -0.6892 0.2290 2015q2 59.63 -0.2098 0.3682 -0.2273 2015q3 47.12 -0.0112 0.2946 -0.2821 2015q4 46.59 -0.2784 -0.0070 -0.8756 2016q1 33.62 -0.0686 0.3565 2016q2 -0.1410 -0.3168 2016q3 0.2311 -0.0170 2016q4 - -0.4204 Total: 0.4814% -2.1320% The oil price is not the only transmission channel which affects GDP growth and inflation so it cannot be checked if these macroeconomic variables changes with exactly

(22)

this percentage. According to the results of the time-series model, it is significant at a level of 10% to conclude that the real GDP increases with 0.4814% and that prices decreases with 2.132% because of the oil price shock between the second quarter of 2014 and the fourth quarter of 2015. Because of the lagged effects, the real GDP and inflation doesn’t change at one time but changes gradually. 5. CONCLUSION This thesis analyzes the effect of a changing oil price on the real GDP and inflation in The Netherlands so that an estimation can be done how the real GDP and inflation are affected because of the recent oil price drop. According to most of the articles discussed in this thesis, a positive relation exists between the oil price and inflation and a negative relation exists between the oil price and the real GDP of oil importing countries. In previous studies it became clear that the effect of an oil price shock depends whether the country has a net importing or exporting position. The Netherlands is a net importing country and thus the real GDP should increase and inflation should decrease after the recent oil price shock. The two time-series models used in this thesis were able to estimate the direct and the lagged effects of an oil price shock. According to the Granger Causality test, the optimal number of lags that must be used was four in the model of the real GDP and five in the model of inflation. Unfortunately, only the first three lags of the regression on the real GDP were significant and only the fourth lag of the regression on inflation was significant. A possible reason why the other lags were not significant could be because not enough observations were used. Contradictive to some previous studies, a temporary positive relation between the oil price and the real GDP in The Netherlands was found one and two quarters after an oil price shock instead of a negative relation. A possible reason of this unexpected result could be that, even though The Netherlands is a net oil importing country, the net export position of gas could give unexpected results. The results of the reaction of inflation after an oil price shock is similar to previous studies except that the reaction is only significant after four quarters. An explanation why no effect was observed in the three quarters after an oil price shock could be because not enough explanatory variables were used which caused the

(23)

relatively low R-squared. Another reason for insignificance results could be because no distinction was made between oil price increases and decreases. Even though not all the lags were significant, the results are still pretty similar to other studies. With these results a significant estimation about the effect of the recent oil price shock on the real GDP and inflation in The Netherlands could still be done. According to the results, the real GDP increased with 0.4814% because of the recent oil price shock. Inflation was affected negatively according to the results. Consumer price index should have decreased with 2.1320% after the recent oil price shock. Even though the results are significant, it cannot be checked whether the real GDP and inflation are really affected this much because of other transmission channels which affects these variables.

(24)

REFERENCES Basnet, H., & Upadhyaya, K. (2015). Impact of oil price shocks on output, inflation and the real exchange rate: Evidence from selected ASEAN countries. Applied Economics, 47(29), 3078-3091. doi:10.1080/issn:00036846 Brown, S., & Yücel, M. (2008). What Drives Natural Gas Prices? The Energy Journal, 29(2), 45-60. doi:10.5547/issn:0195-6574 Castro, Jerez, & Barge-Gil. (2016). The deflationary effect of oil prices in the euro area. Energy Economics, 56, 389-397. Doi:10.1016/issn:0140-9883

CBS. (2016). CBS Statline - Consumer prices. Retrieved 21 may 2016, from statline.cbs.nl: http://statline.cbs.nl/Statweb/publication/?DM=SLEN&PA=83135eng&D1=0 2&D2=(l-39)-l&LA=EN&VW=T CBS. (2016). CBS Statline – Gross domestic product. Retrieved 21 may 2016, from statline.cbs.nl: http://statline.cbs.nl/Statweb/publication/?DM=SLEN&PA=82601eng&D1=0 71&D2=0,2&D3=100-105&LA=EN&VW=T Cuñado, & Pérez de Gracia. (2003). Do oil price shocks matter? Evidence for some European countries. Energy Economics, 25(2), 137-154. Doi: 10.1016/issn:01409883 DNB De Nederlansche Bank. (2016). Money supply, Euro, Netherlands. Retrieved 21 may 2016 from http://www.dnb.nl/home/index.jsp Eurostat. (2016). Netherlands, Harmonized Unemployed Rates. Retrieved 21 may 2016 from http://www.dnb.nl/home/index.jsp Forni, Gerali, Notarpietro, & Pisani. (2015). Euro area, oil and global shocks: An empirical model-based analysis. Journal of Macroeconomics, 46, 295. Doi:10.1016/issn:01640704 Gómez-Loscos, A., Gadea, M., & Montañés, A. (2012). Economic growth, inflation and oil shocks: Are the 1970s coming back? Applied Economics, 44(35), 4575-4589. Doi:10.1080/issn:00036846 Hamilton, J. (1983). Oil and the Macroeconomy since World War II. The Journal of Political Economy, 91(2), 228-248. 91(2), 228. Doi:10.1086/issn:00223808

(25)

Investing.com. (2016). Crude oil historical data (dollars per barrel. Retrieved 21 may 2016, from http://www.investing.com/commodities/crude-oil-historical-data J. H. Stock, & M. W. Watson. (2011). Introduction to Econometrics (Vol. 3). Pearson. Jiménez-Rodrı́guez, R., & Sánchez, M. (2005). Oil price shocks and real GDP growth: empirical evidence for some OECD countries. Applied Economics, 37(2), 201-228. doi:10.1080/0003684042000281561 Lardic, & Mignon. (2006). The impact of oil prices on GDP in European countries: An empirical investigation based on asymmetric cointegration. Energy Policy, 34(18), 3910-3915. Doi:10.1016/issn: 0301-4215 Lee, J. & Song, J. (2009). Nature of oil price shock and monetary policy. Cambridge, M: National Bureau of Economic Research Morana, C. (2016). Macroeconomic and Financial Effects of Oil Price Shocks: Evidence for the Euro Area. ESP: Energy Scenarios and Policy Mork, K., & Olsen, O. (1994). Macroeconomic responses to oil price increases and decreases in seven OECD countries. Energy Journal, 15(4), 19-35. Issn: 0195-6574 Naifar, & Al Dohaiman. (2013). Nonlinear analysis among crude oil prices, stock markets' return and macroeconomic variables. International Review of Economics and Finance, International Review of Economics and Finance. Doi: 10.1016/issn: 1059-0560 Pradhan, P., Arvin, B. & Ghoshray, A. (2015). The dynamics of economic growth, oil prices, stock market depth, and other macroeconomic variables: Evidence from the G-20 countries. SSRN Electronic Journal. Doi: 10.1016/issn. 10575219* Roeger, W. (2005). International oil price changes: Impact of oil prices on growth and inflation in the EU/OECD. International Economics and Economic Policy, 2(1), 15 32. Doi: 10.1007/issn. 16124804 Sadorsky, P. (1999). Oil price shocks and stock market activity. SSRN Electronic Journal. Doi: 10.1016/issn. 01409883* Sommer, M. (2015) Global Implications of Lower Oil Prices. International Monetary Fund 2015 RePEc

(26)

APPENDICES

Table 1:

Summery statistics absolute values of variables:

Variable Obs Mean Std. Dev. Min Max

GDP 84 133240.6 28499.04 80016.2 171123 Inflation (CPI) 84 84.3858 10.1755 67.4550 100.5 Oil 92 50.2396 31.7396 12.5633 126.9367 Investment 84 27736.49 4982.806 17302 35953 Gov spending 84 31599.36 9208.931 17877 43327 Consumption 84 62591.19 10950.68 38915.3 75864.7 Imports 84 81790.75 25699.59 40615 123765 Exports 84 92847.6 30008.39 46587 142188 Hourly wage 84 89.0499 11.9052 69.247 106.5 Interest 84 2.4818 1.5413 -0.09 5.13 Unemployment 84 5.3949 1.5107 2.833 8.7 Table 2: Summary statistics growth values of variables:

Variable Obs Mean Std. Dev. Min Max

GDP 83 0.0092 0.0088 -0.0263 0.0286 Inflation (CPI) 83 0.0047 0.0093 -0.0132 0.0270 Oil 91 0.0182 0.1393 -0.5096 0.3763 Investment 83 0.0081 0.0088 -0.0163 0.0354 Gov spending 83 0.0087 0.0296 -0.0578 0.1100 Consumption 83 0.0106 0.0137 -0.0337 0.0733 Imports 83 0.0139 0.0268 -0.0936 0.0638 Exports 83 0.0138 0.0264 -0.1041 0.0643 Hourly wage 83 0.0052 0.0042 0.0000 0.0146 Interest 83 -0.0022 0.3821 -1.2000 2.0000 Unemployment 83 -0.0001 0.0800 -0.1448 0.1918 Money supply 83 0.0131 0.0096 -0.0152 0.0374

(27)

Table 3:

Phillips-Perron tests

Test statistics Z(t)

Variable: Absolute values Growth rate

GDP -0.856 -6.736*** Oil -1.243 -7.568*** Consumption -1.357 -7.455*** Government spending -0.314 -8.412*** Investment -1.926 -9.465*** Imports -3.020 -5.653*** Exports -3.046 -5.223*** Inflation -4.241*** -19.850*** Exchange rate -1.844 -6.421*** Interest rate -1.535 -3.89*** Wage -0.749 -9,909*** Unemployment rate -2.151 -6,928*** Note: *** indicate significant at 1%. Table 4: Number of lags GDP Inflation Lags Adjusted R-Squared Adjusted R-squared 0 0.7460 0.4857 1 0.7684 0.4858 2 0.7677 0.4789 3 0.7751 0.4734 4 0.7779 0.5041 5 0.7768 0.5070 6 0.7739 0.5037 7 0.7738 0.497 8 0.7749 0.4937 The underlined is the highest value, the optimal number of lags.

(28)

Referenties

GERELATEERDE DOCUMENTEN

As Altides (2013) argues, national organisations gain easier access to their members and as the national media, in the absence of European mass media, inform people on the

The relationship between content style and engagement is mediated by trust, credibility and visibility as literature research showed that these factors influence the usage

Om de tweede hypothese te toetsen, of volwassenen die hoog betrokken zijn bij gezondheid en/of gewicht, een sterkere intentie hebben om gezond te eten en het gezonde gedrag zal

In the Thusa Mama Study, 98 pregnant black women were included and of these 98 women, two women had miscarriages and five women were lost during the follow-up visits. They

When differences in income per capita are controlled the results show that economic freedom (proxy variable for democracy) increases happiness in poor countries with a

In 2001 the Central Bank of Iceland stopped using the exchange rate as its main monetary policy instrument and adopted inflation targeting in order to increase

The results show that the cultural variables, power distance, assertiveness, in-group collectivism and uncertainty avoidance do not have a significant effect on the richness of the

For every episode (stable, stop, and surge), this model is calculated two times: for the indirect effect of an increase in total gross capital inflows on the central