Faculty of Economics & Business BSc Economics & Business Specialization Economics & Finance
Bachelor thesis
The effect of recent low oil prices on the growth of real
GDP of Norway
Name: T. van den IJssel Student number: 10452303 Supervisor: O. Furtuna
Statement of Originality
This document is written by Tom van den IJssel, 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.
Abstract
This paper aims to investigate whether the recent low oil prices significantly affect the growth of real GDP of Norway. The time–series model used in the paper is built on the simple linear model of Hamilton (1983), who analyzed the effects of oil price increases on U.S. recessions. The dependent variable used in the model is the growth rate of real GDP of Norway, with the growth rate of the real oil price as independent variable. A set of control variables is added including lagged GDP growth rates, as well as dummy variables to adjust for seasonality because of the quarterly frequency of the data used. The results don’t show any significant effects of oil price changes on the growth of real GDP of Norway.
Table of contents Page
Abstract 3
1. Introduction 5 2. Literature review 10
2.1 Transmission channels 11
2.2 Linear model – symmetry 13
2.3 Non-linear models – asymmetry 14
2.4 Norway 18 3. Methodology 20 4. Data 23 5. Results 25 6. Conclusion 29 7. Discussion 30 References 31 Appendices 34
1. Introduction
In June 2014, the price per barrel of North Sea Brent crude oil - considered a measure for the global price of oil - was almost 115 US dollars as shown in figure 1. In December 2014 the price of a barrel of crude oil was worth just over 60 US dollars (Economist, 2014). This implies a drop in oil prices of about 55 dollars or 45% in just six months time. In July 2008, oil prices dropped from approximately 120 US dollars to 40 US dollars five months later. That decline of 80 US dollars or 66% was of a larger magnitude compared with the decline in 2014, and almost entirely due to a collapse in demand after the financial crisis. The current price of a barrel of Brent crude oil is close to 55 US dollars according to the U.S. Energy Information Administration (2016).
Figure 1: Crude oil prices since 2005 (US Dollars per barrel)
Source: U.S. Energy Information Agency (2016) The oil price is determined by actual supply – and demand, and by
speculation (Economist, 2014). Speculation implies that oil producers expect the oil price to stay high, after which they will invest, which leads to an increase in oil supply. The opposite occurs when prices are low (Economist, 2014).
0 20 40 60 80 100 120 140 Q1 2005 Q3 2005 Q1 2006 Q3 2006 Q1 2007 Q3 2007 Q1 2008 Q3 2008 Q 1 200 9 Q3 2009 Q1 2010 Q3 2010 Q1 2011 Q3 2011 Q1 2012 Q3 2012 Q1 2013 Q3 2013 Q1 2014 Q3 2014 Q1 2015 Q3 2015 Q1 2016 Q3 2016 $ p er ba rr el Brent oil price
The Organization of the Petroleum Exporting Countries (OPEC) is also able to affect the oil price by the use of their supply management tool, which implies a decrease or increase in levels of oil production (Kilian, 2006). The OPEC is an organization consisting of 13 oil-exporting member countries1, mainly from the Middle East.
According to the OPEC Annual Statistic Bulletin (2016), the organization owns more than 80% of the total world’s oil reserves, and therefore it has a significant influence on the price of oil. According to the Economist (2014), the oil price decline in 2014 was the result of both lower demand and unchanged supply. The demand for oil was low because of weak economic activity, increased efficiency and the increased demand for other types of fuel instead of oil
(Economist, 2014). Furthermore, the oversupply has led to even lower oil prices since the OPEC decided to keep their production quotas unchanged (Baumeister & Kilian, 2016).
In this paper, I will examine the effect of the recent low oil prices on the growth of the real GDP of Norway. The effect of oil price fluctuations on the growth of real GDP depends on whether the country is net oil exporting – or importing. An oil-price decline usually implies a transfer of wealth from oil exporting to oil importing countries. Norway is chosen as country of interest because it has been an oil-exporting country for more than 30 years from now and the country’s oil-producing sector is large relative to the economy as a whole (Bjørnland, 2009). As reported by the Statbank Norway (2016), the national export volume of Norway consists by approximately 25% of crude oil export as can be seen in figure 2.
Figure 2: exports Norway
Source: SSB (2016)
Despite the relative small size of the country, with a population of just over 5 million people, Norway is still one of the largest world oil exporters. In 2015, the country was ranked 11th on the list of the world’s largest oil exporters according to the OPEC Annual Statistic Bulletin (2016). As described in that paper, more than 1,2 million barrels of crude oil are exported per day as shown in figure 3. This means almost 3% of the global crude oil export, which is however not big enough to influence the oil price itself. Especially since Norway is not an OPEC member and countries such as Russia and some of the OPEC member countries daily export three – or four times as much as Norway does (OPEC Annual Statistic Bulletin, 2016). Crude oil (26%) Gas (22%) Chemicals (7%) Fish (12%) Manufactured goods (10%) Machines (13%) Other (10%)
Figure 3: World crude oil exports daily 2015 (1,000 barrels/d)
Source: OPEC Annual Statistic Bulletin (2016)
Since the share of exports as a percentage of GDP for Norway was close to 40% in 2015 according to the Worldbank (2017), an oil price decrease will be expected to negatively affect the Norwegian GDP. Based on the facts that the share of exports as percentage of GDP was close to 40%, and Norway’s national export volume consists by approximately 25% of oil exports, a simple calculation of the expected effect on GDP growth can be made. An oil price decline of 10% would lead to a decrease in GDP growth by 1% (0.25*0.40*-0.10).
As a result of all previously mentioned facts, oil price changes will
definitely affect the economic activity for a real oil-driven economy like Norway. One would easily expect that the effect of current low oil prices on the growth of real GDP of Norway is negative because of the country’s high dependency on oil exports. However, according to previous research this does not necessarily has to be the case. Possible explanations for this phenomenon could be more exports or another positive effect that arises from one or more of the transmission channels that will be discussed in the next section.
0 1000 2000 3000 4000 5000 6000 7000 8000 Saudi Arabia Russia Iraq UAE Canada Nigeria Venezuela Kuwait Angola Mexico Norway Iran Oman Algeria
This paper is organized as follows. In the next part, existing literature written on the effects of oil price shocks on GDP growth will be summarized. In section 3, the method of research will be described. Then the collected data used to test will be explained further, where after section 5 provides the results following the tests performed. In section six, a conclusion regarding the research question will be formulated based on the results from the fifth part.
2. Literature review
The related empirical studies on this subject started by finding a linear
relationship between oil prices and economic activity. Hamilton (1983) was the first who tried to estimate the change in real GDP after an oil price shock using a linear model. In later research, evidence of a non-linear impact of oil prices on real GDP is found. This implies that oil price increases are found to have an impact on GDP growth of a larger magnitude than that of oil price declines (Jimenez-Rodriguez and Sanchez, 2005). Thus, responses to price increases and decreases are allowed to be asymmetric (Mork and Olsen, 1994). Mork and Olsen (1994) tried to investigate whether the effect of oil price fluctuations on GDP showed evidence of asymmetric effects. Most countries actually show evidence, with Norway as an exception. They found that Norway benefits from oil price increases as an oil-exporting country, but the finding that it seems to be hurt by price declines is less significant. Furthermore, Jimenez-Rodriguez and Sanchez (2005) have shown that the effect of oil shocks on GDP growth differs between the two oil-exporting countries in their sample, namely the UK and Norway. This is related to the Dutch disease effect, which will be explained further in section 2.3.
A lot of research has been done about the effects of oil price shocks or fluctuations on economic activity. First of all, the transmission channels through which oil price changes may affect economic activity will be discussed. Second, the linear model of Hamilton (1983) will be analyzed and discussed. Thereafter, the non-linear models of Lardic and Mignon (2006) and Mork and Olsen (1994) will be discussed. In addition to the paper of Lardic and Mignon (2006), a part of Brown and Yücel (2002) their analysis on explanations for asymmetry will be described. Thirdly, the paper of Jimenez-Rodriguez and Sanchez (2005) will be analyzed, which made use of both linear – and non-linear models. Finally, the case for Norway will be discussed based on the estimates of Lardic and Mignon (2006). I try to forecast the effect of a decrease in oil prices on the growth of GDP of Norway.
2.1 Transmission channels
Lardic and Mignon (2006) discussed the six transmission channels through which oil prices may affect economic activity:
Firstly, the classic supply-side effect, which implies that rising oil prices lead to a reduced availability of a basic input to production, which lowers potential output directly (Brown and Yücel, 1999). As a result, the costs of
production will increase for many industries dependent on oil, and the growth of output and productivity are slowed (Lardic and Mignon, 2006). Consequently, the decline in growth of productivity lessens the growth of real wages and increases the unemployment rate (Brown and Yücel, 2002).
Second, an oil price increase worsens the terms of trade2 for oil importing countries because of the higher import prices and lower purchasing power. The value of imports increases because of higher oil prices, which results in a
detoriation of the terms of trade in case of oil importing countries. This typically leads to a transfer of wealth from oil importing countries to oil exporting
countries as described in the previous section.
Thirdly, an oil price increase would lead to a higher money demand due to the real balance effect. This implies that monetary authorities fail to meet the higher money demand with increased supply, which causes a rise in interest rates and a decrease in economic growth (Brown and Yücel, 2002).
Fourth, an oil price increase will lead to an increase in inflation. An increase in oil prices causes a shift in the production function, which leads to lower output (Brown and Yücel, 1999). This reduction in output, ceteris paribus, results in an excess demand for goods and a higher interest rate according to Brown and Yücel (1999).
The fifth transmission channel as described by Lardic and Mignon (2006) implies that an oil price increase may affect consumption, investment and stock prices negatively.
2 Represents the value of the exports of a country, relative to the value of its imports (Brown and Yücel, 2002)
Consumption is negatively affected through its positive relation with disposable income, and investment is negatively affected by the higher firm costs (Lardic and Mignon, 2006). The previous channels will negatively affect these 3 variables as well for oil importing countries when the oil price increases.
The sixth and last channel states that if an oil price increase is long lasting, it may change the production structure and have an impact on
unemployment (Lardic and Mignon, 2006). An oil price increase leads to higher costs for oil-intensive sectors, which gives firms an incentive to construct new production methods that are less intensive in oil inputs to reduce costs
(Loungani, 1986). According to Loungani (1986), such a change in production methods will generate capital – and labor reallocations across sectors that may affect long-run unemployment.
Through these six channels listed above, oil prices may affect economic activity. The expected effects from each channel for oil importing countries are expected to be the opposite in case for an oil exporting country as Norway when oil prices increase.
2.2 Linear model – symmetry
The first studies on the impact of oil price shocks on economic activity came from the oil crises in the 1970’s. Hamilton (1983) examined the relationship between economic recessions and increases in the oil price for the U.S.
He found that seven out of eight of the U.S. recessions after World War II were preceded, typically with a lag of around three-fourths of a year, by a significant increase in the oil price. This correlation between U.S. recessions and oil price increases has three possible explanations according to Hamilton (1983):
1. The correlation is a historical coincidence
2. The correlation arises from an endogenous explanatory variable
3. Some of the U.S. recessions were causally influenced by an exogenous oil price increase
Hamilton (1983) found that oil price increases over the period between 1948 and 1972 were followed by a lower real GNP growth. He used the six-variable VAR estimation, introduced by Sims (1980), to examine the role of oil for real macroeconomic variables. Hamilton (1983) concludes that there exists a
systematic relationship between oil prices and output, which leads to a rejection of the first hypothesis. Furthermore, he did not find a possible third explanatory variable that could be responsible for both the oil price increases and the
recessions.
This implies a rejection of the second hypothesis as well. Therefore, Hamilton concludes that there exists a significant linear relationship between oil prices and economic activity.
2.3 Non-linear models – asymmetry
During the 1980’s, the linear relationship predicted by Hamilton (1983) began to lose importance. That was because of the effects of oil price declines found to be different from those of oil price increases. Mork (1989) was the first one who analyzed the possible existence of asymmetric effects of oil price shocks on economic activity. He found evidence that oil price increases had a negative significant effect, similar in magnitude to Hamilton’s (1983) research, on the real GDP of the U.S. and OECD countries. However, oil price declines didn’t showed any significant effects at all (Mork, 1989).
Lardic and Mignon (2006) tried to investigate if there exists a stable long-term relationship between oil prices and GDP for 12 European countries. They found evidence that economic activity reacts asymmetrically to oil price shocks. They concluded that rising oil prices seem to retard economic activity by more than falling oil prices stimulate it. Explanation for the asymmetric responses could be monetary policy, adjustment costs and asymmetry in petroleum product prices, which will be explained hereafter.
Brown and Yücel (2002) also found that monetary policy is a possible explanation for the asymmetric response of the economy to oil price shocks. If wages are nominally sticky downward, monetary policy can have asymmetric effects. An increase in oil prices, when wages are sticky downward, leads to losses in GDP when the monetary authority is not able to hold nominal GDP constant through unexpected inflation (Brown and Yücel, 2002). However, when oil prices decline, real wages must rise to clear the market (Lardic and Mignon, 2006).
Hamilton (1988) was the first who stated that adjustment costs might lead to asymmetric responses in case of changes in oil prices. He argued that rising oil prices lower economic activity immediately, and falling oil prices boost economic activity directly. However, the costs of adapting to changed oil prices
This implies that a decrease in oil prices would cause both negative – and
positive effects on economic activity, but an increase in oil prices would have two negative effects according to Brown and Yücel (2002).
Brown and Yücel (2002) also found that petroleum product prices might contribute to the existence of an asymmetric relationship between oil prices and economic activity. Various researchers showed evidence that petroleum product prices respond asymmetrically to oil prices, which implies that gasoline prices rise more quickly when oil prices are rising than they fall when oil prices are decreasing (Lardic and Mignon, 2006).
Mork and Olsen (1994) investigated the correlations between changes in oil prices and economic activity for seven OECD countries as mentioned before. They showed that the correlations between oil-price increases are negative and significant for most countries, but positive for Norway. The correlations between oil-price decreases are positive for most of the countries, but only significant for the U.S. and Canada (Mork and Olsen, 1994). They also found that most of the countries showed evidence of asymmetric effects, except for Norway. In their research, they tried to answer the following 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?
First of all, they analyzed the bivariate correlations between oil price changes and GDP growth. Mork and Olsen (1994) tried to estimate a regression equation with GDP growth as dependent variable, and lagged values of GDP growth, with lags of five quarters, and oil price changes as independent variables. Their estimation period was between the third quarter of 1967 and the fourth quarter of 1992.
They found a general pattern of negative correlations, mostly significant, between GDP growth rates and oil price increases. The low significance or non-significance for countries like Germany and Canada is caused by measurement errors of GDP according to Mork and Olsen (1994). Therefore, the answer to the first question is affirmative.
For most countries, the coefficients of oil price decreases appear to be the opposite sign of oil price increases (Mork and Olsen, 1994). However, they found that most of these correlations weren’t significant. Overall, they found evidence for asymmetry at different significance levels, except for Norway. The lack of significance for Norway may be due to measurement errors in quarterly GDP figures (Mork and Olsen, 1994). Concluding, the correlation patterns are not the same for price increases and decreases, so the second question can be answered in the affirmative.
The answer to the third question is affirmative as well, however not as strongly as could have been expected according to Mork and Olsen (1994). They conclude that a country’s net oil export position influences the oil price – GDP correlation.
Jiménez-Rodríguez & Sánchez (2005) examined the effect of oil price changes on the real economic activity of the main industrialized countries (OECD countries), including both net oil exporting – and oil importing countries. They used multivariate vector auto regression (VAR) analysis, with one linear and three non-linear models. They used quarterly data of the following variables: real GDP, real effective exchange rate, real oil price, real wage, inflation, and short and long-term interest rates.
First of all, their findings on net oil-importing countries show a negative relationship between oil price and real GDP, except for Japan. Thereafter,
Jiménez-Rodríguez & Sánchez (2005) found that oil prices have different impacts on economic activity when they increase than when they fall.
In this paper, evidence is showed of a non-linear impact, or asymmetric effects, of oil prices on real GDP using the Granger causality test.
Secondly, two net oil-exporting countries, Norway and the UK, show contrasting results. Oil prices found to affect positively Norwegian GDP growth, but found to have a negative impact on the GDP growth for the UK. This implies that the UK reacts similar to oil-importing countries, because an oil price
increase of 100% leads to a loss in GDP growth rate for the UK of more than 1% after the first year (Jiménez-Rodríguez & Sánchez, 2005).
This unexpected result has to do with the fact that oil price increases led to a large real exchange rate appreciation of the pound, which indicates the Dutch disease.
The Dutch disease had been first experienced in the Netherlands, where natural gas discoveries in the 1960’s led to adverse effects on the Dutch
manufacturing sector, mainly through a real exchange rate appreciation (Bjørnland, 1998). Thereafter, during the 1970’s the high income from gas resources fell and other industries weren’t able to compensate for these losses, and unemployment rose quickly (Bjørnland, 1998). For this reason, the negative consequence of a natural resource discovery is called the Dutch disease as
described by Bjørnland (1998).
In case for the UK, it implies that negative consequences arise from a large increase in the value of the pound, which reduces price competitiveness for exports of the UK’s manufactured goods (Jiménez-Rodríguez & Sánchez, 2005). As a result, there will be higher unemployment in the long run.
Jiménez-Rodríguez & Sánchez (2005) showed that Norway’s real
exchange rate appreciation after the first year was much weaker than in case of the UK. This implies that the effect on GDP growth of Norway had a greater positive effect after an oil price increase, due to the weaker real exchange rate appreciation compared to the effect on GDP growth of the UK. The difference in responses between Norway and the UK could also partly be explained by the difference in adjustments of inflation, interest rates and real wages (Jiménez-Rodríguez & Sánchez, 2005).
2.4 Norway
As discussed in section 2.2, Lardic and Mignon (2006) investigated whether there exists a long-term relationship between oil prices and GDP for 12
European countries, including Norway. They found that economic activity reacts asymmetrically to oil price changes using the following models:
(1) GDPt− + ΔGDPt+ = α− + β− OIL− + εi (2) GDPt+ + ΔGDPt− = α + + β+ OIL+ + εi
The first equation shows the effect on GDP of an oil price decrease. The second equation shows the effect on GDP after an oil price increase. The values of alpha and beta are given for Norway in the research of Lardic and Mignon (2006). Based on the coefficients obtained in Lardic and Mignon (2006), we obtain the following formulas:
(1) GDPt− + ΔGDPt+ = -0,071 + 0,258 OIL− + εi (2) GDPt+ + ΔGDPt− = -0,052 + 0,466 OIL+ + εi
The value for beta determines the slope in this case, and alpha is the intersect term. As immediately can be seen, the effect of an oil price increase has a larger effect on the GDP for Norway compared to an equally sized decrease in oil price. This proves the asymmetric reaction of oil price changes on real economic activity as Lardic and Mignon (2006) mentioned.
According to the U.S. Energy Information Agency (2016), the annual average price per barrel of crude oil was $52,32 for 2015. In 2016, the average price was $43,55. Now we are able to calculate the predicted change in GDP of Norway following the model of Lardic and Mignon (2006). We assume an oil price decrease of $8,77:
(1) GDPt− + ΔGDPt+ = -0,071 + 0,258 * 8,77 GDPt− + ΔGDPt+ = 2,192%
Using the model of Lardic and Mignon (2006), we expect an increase of GDP of Norway by 2,192% due to the oil price decrease. This result is somehow
surprising since Norway is an oil exporting country, and a negative GDP growth is expected after an oil price decrease because of lower oil export revenues, as described in the introduction.
3. Methodology
The effect of recent low oil prices on the growth of real GDP of Norway will be tested using a linear time-series model based on the papers of Hamilton (1983) and Brown and Yücel (1999). For all the variables except the dummy variables, growth rates are used.
The model used to estimate the effect of a decrease in oil prices on the growth of real GDP of Norway has the following form:
GDPNWi = β0 + β1 * RGOPi + β2 * GDPUKi + β3 * growthREERi + β4 * inflationi +β5 *
growthUi + β6 * LGDPNWi-4 + β7 * LGDPUKi-4+ β8 * Q_1 + β9 * Q_2 + β10 * Q_3 + εi
GDPNWi: The growth rate of the real GDP of Norway is the dependent variable for this model. The expected sign of the coefficient is negative because of lower oil prices, which are expected to negatively affect the GDP of Norway.
RGOPi: The real growth of the oil price is the independent variable. The expected sign of the coefficient is negative because of decreasing oil prices within the timeframe.
In addition to the two quantities of interest, we also include a set of control variables, which are based on the variables used in studies of Hamilton (1983), Mork and Olsen (1994) and Jiménez-Rodríguez & Sánchez (2005)
GDPUKi: The growth rate of the real GDP of the UK is added to compensate for economic fluctuations. The UK is chosen because it’s Norway’s most important trade partner (globaledge.msu.edu). The expected sign of UK growth rates is difficult to estimate because the country is currently not highly dependent on oil exports, so lower oil prices won’t harm the country’s GDP directly.
GrowthREERi: The growth rate of the real exchange rate. Measured in terms of the weighted average of the Krone against a basket of other currencies. Real effective exchange rates are the same weighted averages of bilateral exchange rates adjusted by relative consumer prices (bis.org, 2016).
Bjørnland (1998) found that the Norwegian real effective exchange rate
depreciates after a decline in oil prices, which implies relatively cheaper exports to other countries. This implies an expected positive sign after an oil price decline.
Inflationi: The growth rate of inflation. Measured by consumer price index (CPI), which is defined as the change in de prices of a basket of goods and services for Norway. The expected sign of this variable is positive because lower oil prices imply higher inflation for an oil exporting country, as described in section 2.1 on transmission channels.
GrowthUi: The growth rate of unemployment of Norway is added as
independent variable as well. As described in Jiménez-Rodríguez & Sánchez (2005), a change in oil prices might affect unemployment in the long run. The expected sign of this coefficient is positive because lower oil prices will higher unemployment rates for an oil exporting country.
LGDPNWi-4 / LGDPUKi-4: The lagged growth rates of GDP for both countries are
added because a growth rate is highly determined by its previous level. I decided to choose a lag of four periods, based on earlier literature on this subject
(Hamilton, 1983). According to Hamilton (1983), it seems that adding four lags leads to the highest adjusted R-squared. Adding fewer lags would lead to less reliable results, because it typically requires a lag of around three-fourth of a year, for the oil price to adjust. If these lagged variables would not be added to the model, it could lead to omitted variable bias.
Q_1/Q_2/Q_3: Added dummy variables to correct for the presence of seasonality because of quarterly data. Seasonality is a time-series characteristic in which data shows a regular and predictable pattern, repeated every year. Such a predictable yearly repeated pattern can be considered as seasonal. The value of the dummy variable can only be 1 or 0. I included 3 dummies in the model instead of 4 to avoid multicollinearity while doing regressions.
Furthermore, I have to test whether the following Ordinary Least Squares (OLS) assumptions hold in order for the coefficients to be unbiased:
1.
ε
iis a random variable with an expected value - given the value of theindependent variable (RGOPi) – equal to zero.
2. The growth of real GDP of Norway (GDPNWi) and the real growth of the oil price (RGOPi) are independently and identically distributed.
3. Large outliers are unlikely
A regression output will be generated using the previously described model. The coefficients for the variables will be stated as significant using a significance level of 10%, or a P-value smaller than 0.01. Furthermore, a joint significance test will be executed to test the significance of the coefficients.
Thereafter, an Augmented Dickey-Fuller test will determine whether there exists a unit root in the model. A unit root is a stochastic trend or random walk in a time series, which implies that the trend in a time series is not predictable. If the model has a unit root, it shows an unpredictable systematic pattern. The Breusch-Godfrey test will be done to test for the presence of serial correlation inside the model. Finally, a Breusch-Pagan test will determine whether the error term of the model is assumed to be hetero – or homoscedastic.
4. Data
The analysis is carried out using quarterly data from the first quarter of 2000 up to and including the fourth quarter of 2016, so there are 68 observations for each variable. The GDP growth rates are already seasonally adjusted. Since the GDP growth rates are only given quarterly, data for the other variables are converted into quarterly rates as well.
Growth real GDP Norway
The growth rates of the real GDP of Norway are retrieved from the OECD database (data.oecd.org, 2016). The data is given quarterly. This indicator is seasonally adjusted and measured in percentage change from previous quarter and from same quarter previous year.
Real oil price
The Brent crude oil prices are retrieved from the US Energy Information Agency (eia.gov, 2016). The price per barrel is converted from daily into quarterly prices. The quarterly inflation rate is deducted to obtain the real oil price and finally, the prices are converted into growth rates.
Inflation
The inflation rate is also retrieved from the OECD database (data.oecd.org, 2016). Quarterly inflation rates are given for Norway and converted into growth rates.
Unemployment
Unemployment rates for Norway are obtained from the OECD database as well (data.oecd.org). Rates are given quarterly and converted into growth rates to add them to the model.
Growth real GDP UK
The growth rates of the real GDP of the UK are retrieved from the OECD database (data.oecd.org, 2016). The data is given quarterly.
Growth real effective exchange rate
The data for real effective exchange rates is retrieved from the Bank for
International Settlements (bis.org, 2016). The data is converted from monthly into quarterly and also converted into growth rates.
Lagged growth real GDP Norway and UK
The lag for this variable is four periods or quarters and retrieved from the OECD database (data.oecd.org, 2016).
5. Results
In this section, the results regarding the performed regressions will be shown. The regression results for the model including all control variables are
summarized in the following table:
Table 1: Multiple regression results (linear* and robust**)
GDPNW Coefficient (p-value) Coefficient (p-value) RGOP GDPUK growthREER LGDPNW LGDPUK inflation growthU Q_1 Q_2 Q_3 _cons -0.008357 (0.728) 0.2545889 (0.582) -0.0154214 (0.906) -0.0286741 (0.841) -0.0515841 (0.889) -0.3988205 (0.094) 0.0077519 (0.882) -0.0067473 (0.362) -0.0053428 (0.494) -0.0029312 (0.698) 0.0134161 (0.077) -0.008357 (0.710) 0.2545880 (0.355) -0.0154214 (0.865) -0.0286741 (0.663) -0.0515841 (0.864) -0.3988205 (0.170) 0.0077519 (0.806) -0.0067473 (0.428) -0.0053428 (0.252) -0.0029312 (0.596) 0.0134161 (0.040)
*R-squared = 0.0810 **R-squared = 0.0810 *Prob > F = 0.8812 **Prob > F = 0.6466
The first column shows us the results of the linear regression with all the variables included. The second column shows us the results of the robust linear regression.
According to the results from table 1, the negative coefficient of the RGOP variable implies a negative effect of an oil price increase on the real growth of GDP of Norway. However, the variable for real oil price growth is far from significant with a significance level of 10% (P=0.728 and 0.710 > 0.100).
This means that no statistically significant linear dependence of the real growth of GDP of Norway on the growth of the real oil price was detected.
However, these results match with the findings of Mork and Olsen (1994), who stated that the lack of significance might be due to the substantial residual
variance for Norway, which they attributed partly to measurement errors in GDP figures. Mork (1989) also did not found any significant effects on economic activity during oil price declines.
A possible explanation for these non-significant could be the high amount of measurement errors in the quarterly GDP growth rates. Another explanation could be the small dataset with only 68 observations, which possibly causes the non-significance of most of the coefficients. The low R-squared value of
approximately 8% indicates that the model explains little of the variability of the real GDP growth of Norway. This could be due to non-included explanatory variables.
Thereafter, the significance of the coefficients is tested using a joint-significance test. The null hypothesis of this test assumes that at the same time, the coefficients for the variables are equal to zero. The joint-significance test has led to the following results:
F (9, 58) = 0.57 Prob > 0.8197
Because of the small F-statistic and large P-value, we cannot reject the H0, which implies no evidence for joint significance. This result is not surprising because of the mostly non-significant outcomes for the single coefficients.
The Augmented Dickey-Fuller (ADF) test is used to test whether a unit root is present in the model. Testing for unit roots helps to determine whether a
process is stationary. The null hypothesis assumes a non-stationary process or a random walk. The ADF test has led to the following results:
T-statistic = -3.067
1% critical value = -3.562 5% critical value = -2.920 10% critical value = -2.595 P-value for Z(t) = 0.0291
At a 5 of 10% significance level, we can reject our null hypothesis the T-statistic is smaller than these critical values. That implies that H0 is rejected and the process is assumed to be stationary, without a unit root at a 5 or 10% significance level.
The Breusch-Godfrey test is used to test for the presence of serial
correlation that has not been included in the model. The chosen lag for this test is 1 period, which implies 1 degree of freedom. The Breusch-Godfrey test has led to the following results:
Chi2 (1) = 2.535 Prob > Chi2 = 0.1113
The null hypothesis assumes no serial correlation. The Breusch-Godfrey test showed a P-value of 0.1113. That implies that with a significance level of 10%, we don’t reject H0.
Furthermore, a Breusch-Pagan test is executed to test for heteroscedasticity of the error term in our linear regression model. Heteroscedasticity of the error term implies a non-constant variance. The Breusch-Pagan test has led to the following results:
Chi2 (10) = 87.95 Prob > Chi2 = 0.0000
The null hypothesis assumes homoscedasticity. The Breusch Pagan test showed a P-value of 0.0000. Therefore, we can reject our null hypothesis and assume evidence of heteroscedasticity of the error term in this model.
6. Conclusion
In conclusion, the paper aims to investigate the effect of recent low oil prices on the growth of the real GDP of Norway. The first studies on this subject assumed a linear relationship between oil prices and GDP. More recent literature showed evidence of asymmetric responses, which implies that oil price increases are found to have an impact on GDP growth of a larger magnitude than that of oil price declines.
The effect of changes in oil prices on the growth of real GDP depends on whether the country is net oil exporting – or importing. An oil-price decline usually implies a transfer of wealth from oil exporting to oil importing countries. Norway is an oil exporting country for more than 30 years from now, and since the share of oil exports equals approximately one fourth of the country’s national export volume, a negative impact on real GDP growth is expected after a decline in oil prices.
A linear model was used to test whether the effect of recent low oil prices on real GDP growth of Norway was actually negative. The dependent variable in this model was the real growth of GDP of Norway. The independent variable was the real growth of Brent crude oil prices. A set of control variables were added based on the variables used in previous literature. The model included the following control variables: real growth of GDP of the UK, growth of the real effective exchange rate, lagged growth rates of GDP for the UK and Norway, growth rates of inflation and the growth of unemployment rate for Norway. Furthermore, I added 3 dummy variables for the first three quarters to adjust for seasonality, because all dates are quarterly rates from the first quarter in 2000 up to and including the fourth quarter of 2016.
The results from the regression showed no significant coefficients, except for the inflation growth at a 10% significance level, which however corresponds with previous research on this subject related to Norway. This means that based on these results, no binding conclusions can be formed about the effect of recent low oil prices on the growth of real GDP of Norway and further investigation is suggested.
7. Discussion
For any further research on the effects of oil price changes on the growth of GDP of Norway, a larger dataset with longer time-series is recommendable. Furthermore, there should be dealt with the lack of significance, possibly partly due to measurement errors in quarterly GDP rates. Another possible solution would be to split up the GDP in components, and examine the effect of an oil price change on the growth of one specific GDP component, for example net exports.
In this research, a linear model is used to test the relationship between oil price changes and real GDP growth. However, it should be interesting to use a non-linear regression method, like the most commonly used multivariate vector auto regression (VAR) analysis. Further research should also consider that a significant coefficient for the oil price variable not directly implies that the oil price influences GDP at that time. This has to do with the fact that there are possible other non-included variables, which may affect real GDP growth via any other transmission channel as described.
References
Baumeister, C., & Kilian, L. (2016). Lower oil prices and the US economy: is this time different. Brookings papers on economic activity.
https://www.brookings. edu/bpea-articles/lower-oil-prices-and-the-us-economy-is-this-time-different.
Baumeister, C., & Kilian, L. (2016). Understanding the Decline in the Price of Oil since June 2014. Journal of the Association of Environmental and Resource
Economists, 3(1), 131-158.
Bis.org. (2016). BIS effective exchange rate indices. Retrieved 18 January 2017, from http://www.bis.org/statistics/eer/
Bjørnland, H. C. (2009). Oil price shocks and stock market booms in an oil exporting country. Scottish Journal of Political Economy, 56(2), 232-254.
Bjørnland, H. (1998). The economic effects of North Sea oil on the manufacturing sector. Scottish Journal of Political Economy, 45(5), 553-585.
Brown, S. P., & Yücel, M. K. (2002). Energy prices and aggregate economic activity: an interpretative survey. The Quarterly Review of Economics and
Finance, 42(2), 193-208.
Brown, S. P., & Yücel, M. K. (1999). Oil prices and US aggregate economic activity: a question of neutrality. Economic & Financial Review, 16.
Data.oecd.org. (2016). Data Norway. Retrieved 18 January 2017, from
https://data.oecd.org/norway.htm
Eia.gov. (2016). U.S. Energy Information Administration (EIA). Retrieved 18 January 2017, from
https://www.eia.gov/dnav/pet/hist/LeafHandler.ashx?n=PET&s=rbrte& f=D
Globaledge.msu.edu. (2015). Norway: Trade Statistics. Retrieved 17 February 2017, from https://globaledge.msu.edu/Countries/Norway/TradeStats
Hamilton, J. D. (1988). A neoclassical model of unemployment and the business cycle. Journal of Political Economy, 96(3), 593-617.
Hamilton, J. D. (1983). Oil and the macroeconomy since World War II. The
Journal of Political Economy, 228-248.
Jiménez-Rodríguez*, R., & Sánchez, M. (2005). Oil price shocks and real GDP growth: empirical evidence for some OECD countries. Applied economics,
37(2), 201-228.
Kilian, L. (2006). Not all oil price shocks are alike: Disentangling demand and supply shocks in the crude oil market.
Lardic, S., & Mignon, V. (2006). The impact of oil prices on GDP in European countries: An empirical investigation based on asymmetric cointegration.
Energy policy, 34(18), 3910-3915.
Loungani, P. (1986). Oil price shocks and the dispersion hypothesis. The
Review of Economics and Statistics, 536-539.
Mork, K. A., Olsen, Ø., & Mysen, H. T. (1994). Macroeconomic responses to oil price increases and decreases in seven OECD countries. The Energy
Journal, 19-35.
Mork, K. A. (1989). Oil and the macroeconomy when prices go up and down: an extension of Hamilton's results. Journal of political Economy, 97(3), 740-744.
Opec.org. (2016). Annual Statistical Bulletin. Retrieved 20 December 2016, from
http://www.opec.org/opec_web/static_files_project/media/downloads/ publications/ASB2016.pdf
Opec.org. (2017). Member countries. Retrieved 15 January 2017, from http://www.opec.org/opec_web/en/about_us/25.htm
Sims, C. A. (1980). Macroeconomics and reality. Econometrica: Journal of the
Econometric Society, 1-48.
Ssb.no. (2016). StatBank Norway. Retrieved 20 December 2016, from http://www.ssb.no/en/statistikkbanken
The Economist. (2014). Why the oil price is falling. Retrieved 20 December 2016, from
http://www.economist.com/blogs/economistexplains/2014/12/econom ist-explains-4
WorldBank. (2016). Norway trade statistics. Retrieved February 12, 2017, from http://wits.worldbank.org/CountryProfile/en/NOR
Appendices
Appendix 1
Table 1: summary statistics data
Variable Obs Mean Std. Dev. Min Max
GDPNW 68 .0021044 .0195854 -.131 .0351 RGOP 68 .0223661 .1442361 -.5040568 .3392087 GDPUK 68 .0044824 .0062361 -.0226 .0136 growthREER 68 -.0003172 .0252955 -.0932136 .0475545 LGDPNW 68 .002575 .0197106 -.131 .0351 LGDPUK 68 .0047485 .0065629 -.0226 .0182 inflation 68 .0206324 .0111929 -.014 .047 growthU 68 .0067247 .052202 -.1529052 .126935 Q_1 68 .25 .4362322 0 1 Q_2 68 .25 .4362322 0 1 Q_3 68 .25 .4362322 0 1 Appendix 2
Graph 1: correlation %Δ oil price / %Δ GDP
-. 1 5 -. 1 -. 0 5 0 .0 5 G D PN W -.6 -.4 -.2 0 .2 .4
