The Relationship Between Energy Consumption and Economic Growth

Using a neo-classical one-sector aggregate production where energy is used as one inputs, the paper develops a vector error-correction (VEC) model to test for the existence and direction of causality between output growth and energy consumption in Malaysia. Using the Johansen co-integration technique, the empirical findings indicate that the long-run movements of output and energy do not exist by one co-integrating vector. Then using a VEC specification, the short-run dynamics of the variables indicate that Granger-causality is not running in both directions between output growth and energy consumption. Hence, those findings conform the presence of “neutral hypothesis theory” which imply Malaysia’s economic growth is not dependent highly upon energy consumption, and energy cannot be considered as a limiting factor to output growth in Malaysia.


INTRODUCTION
The Government Transformation Programme (GTP) was officially designed to be an ambitious programme of change aiming to transform the very foundation of Malaysian Government.GTP is derived into seven Numbers Key Result Areas (NKRA), namely (1) addressing the rising cost of living, (2) reducing crime, (3) fighting corruption, (4) improving student outcomes, (5) raising living standards of low income households, (6) improving rural basis infrastructure, and ( 7) improving public transport.The implementation of the GTP is in part a response to request from the people, who were demanding more from the government.In addition, it is vividly a recognition of the fact that the impetus to become a high-income nation by 2020 is fast approaching.Given such policy, for sure, Malaysian government would be pushing the income generated over all citizens in order to accelerate and signify with the GTP aspiration.
Economically, it is very interesting to tie up GTP goal with current Malaysian Economy.According to quarterly update on Malaysian Economy in 4 th quarter 2012, published by Ministry of Finance, the Malaysian economy registered a robust growth of 6.4% in the fourth quarter of 2012 (Q3 2012: 5.3%) led by strong performance in all sectors.The services and manufacturing sectors remained the key drivers of growth.The services sector recorded a growth of 6.3% (Q3 2012: 7%) spurred by the finance and insurance, wholesale and retail trade as well as communication subsectors.In line with higher industrial production index (IPI), the manufacturing sector expanded 5.8% (Q3 2012: 3.3%) driven by the stronger performance of the transport equipment, petroleum and chemical products as well as electrical and electronic subsectors.The agricultural sector made up a higher growth of 5.6% (Q3 2012 : 0.5%) and eventually mainly attributed to strong growth in the oil palm, livestock and fishing subsectors.Meanwhile, growth in the mining sector rebounded strongly by 4.3% (Q3 2012: -1.2%) on account of increased production of crude oil and natural gas.For whole year, the economy expanded 5.6% surpassing the forecast of 4.5-5% in October 2012.In other words, the pro-growth policies were pushed by government in the sake for inclining towards long term goal (Table 1).
Meanwhile, in terms of energy formation, Jacobsen (2007) did the input-output tables for Malaysia in 2000 and the energy goods domestically supplied as well as imported.His study took three energy commodities/sectors in the Malaysia's inputoutput tables.According to Table 2, we extract that crude petrol, natural gas, and coal as the main contributors for output formation, accounted for RM 46 billion, meanwhile electricity and gas is a small amount, around RM 15 billion.Interestingly, those three energy commodities/sectors are mainly impetus for Malaysian growth since they are used to supply either for domestic intermediate or domestic final goods.In other words, Malaysian economic growth is highly dependence on the energy consumption, particularly crude oil, natural gas, coal, petrol, and coal product.This phenomenon was explained by Ang (2008) who explores long run relationship and causality among output, energy consumption and pollutant emissions for Malaysia over the period 1971-1999.He found pollution and energy consumption positively affect output in the long run.The causality runs from economic growth to energy consumption growth, both in the short and the longrun.More specifically, Shaari et al (2012) concluded that there is a long-run relationship between energy consumption and GDP.However, once the granger causality model is used, the oil and coal consumption do not granger cause economic growth and vice versa.A unidirectional relationship exists between gas and economic growth in Malaysia.
Therefore, looking at the fact that composition of energy towards output formation is attributable a large portion in Malaysian economic growth, it is relevant to look into detail empirically to what extent its contribution.However, although past study mentioned there existed a long term relationship between economic growth and energy consumption, the study which is concern with direction whether economic growth affects energy consumption affects economic growth is no consensus judgment or not clearly stated.As noted by Jumbe (2004), amongst others, if causality runs from energy consumption to GDP then it implies that an economy is energy dependent and hence energy is a stimulus to growth implying that a  shortage of energy may negatively affect economic growth or may cause poor economic performance, leading to a fall in income and employment.In other words, energy is a limiting factor in economic growth (Stern 2000).Whereas, if causality only runs from GDP to energy consumption this implies that an economy is not energy dependent hence, as noted by Masih and Masih (1997) amongst others, energy conservation policies may be implemented with no adverse effect on growth and employment.If, on the other hand, there is no causality in either direction (referred to be as the "neutrality hypothesis"), it implies that energy consumption is not correlated with GDP, so that energy conservation policies may be pursued without adversely affecting the economy (Jumbe 2004).
The aim of this paper is to empirically investigate the causal interactions between energy consumption and output growth in the case of Malaysia.More specifically, the paper tries to trace the direction of causality within the neo-classical aggregate production model.Thus the significant contribution of this study is to enrich the literature particularly related to energy growth nexus in the case of developing countries and favor Malaysian Government in achieving a high income nation policy by 2020 via analyzing the dynamic between the economic growth and the source of growth (energy consumption).Thus, it might provide policy makers some more complete plan for ensuring such noble goal in track.Table 3. shows the overview of selected studies which emanates numerous methods of related study on energy consumption-growth nexus.

METHOD
In many situations, some researchers are interested in the proximate determinants of growth.The Solow model is one of among growth theories in neo-classical economics school of thoughts which focus on several input variables, namely capital, labor, and knowledge.The central assumptions of the Solow model concern the properties of the production function and the evolution of the observable inputs into production over time.The model's critical assumption concerning the production function is that it has constant returns to scale in its two arguments.For instance, doubling capital and effective labor, with keeping technology constant, the amount of production doubles.The assumption of The Relationship Between Energy Consumption and Economic Growth constant returns theoretically can be thought of as a combination of two separate assumptions.The first is that the economy is big enough that the gains from specialization have been exhausted.In very small economy, there are probably enough possibilities for further specialization that doubling the amounts of capital and labor more than doubles output.The Solow model assumes, however, that the economy is sufficiently large that, if capital and labor double, the new inputs are used in essentially the same way as the existing inputs, and thus that output doubles.
The second assumption is that inputs other than capital, labor, and knowledge are relatively unimportant.In particular, the model neglects land and other natural resources.If natural resources are important doubling inputs could be less than double output.In practice, however, the availability of natural resources does not appear to be a major constraint on growth.Assuming constant returns to capital and labor alone therefore appears to be a reasonable approximation.
In addition, we often want to know how much of growth over some period is due to increases in various factors of production, and how much stems from other forces.Growth accounting, which was pioneered by Abramovitz (1956) and Solow (1957), provide a way of tackling this subject.To see how growth accounting works, consider again the production function: Where a dot on the top of a variable means that the variable is now in a growing rate from.The constant parameter a is representing the elasticity of output with respect to energy.The relationship between output and energy input described by production function in Eq. (1) suggests that their long-run movement may be related.Furthermore, if we allow for short-run dynamics in factor-input behavior, the analysis above would also suggest that past changes in energy could be a useful information for predicting the future change of current output, ceteris paribus.This implication can be easily exercised using tests for bivariate co-integration and granger causality.

Testing for Co-Integration
If we test for short run dynamic of the variable in Eq (1), then we can articulate the intertemporal interaction within a VAR specification as follows : Where l is the lag-length and Z t is a 2X 1 vector containing Y, and E. the two variables will be measured by their by their natural logarithm so that their first difference approximate their growth rates.
In addition, if these variables have unit roots, then we can express the idea that there may exist comovements in their behavior and possibilities that they will trend together towards a long run equilibrium state.Then, using the Granger representation theorem, we may posit the following testing relationships that constitute a VEC model for output growth : Where ΔZ t contains the growth rates of the variables.The τ's are estimable parameters, Δ is a difference operator, пt is a vector of impulses which represent the unanticipated movements in Z t with пt-niid (0,ε) and Π is the long run parameter matrix.With r cointegrating vectors (1 < r < 4), Π has rank r and can be decomposed as Π = αβ, with α and β both 2Xr matrices.On expanding out Eq (5), the model can be expressed as follows:  6) Where, p = l-1.The parameters αl.k are the adjustment coefficient, vk, t-p are the cointegrating vectors, and μ 1-2 are intercepts.

Dimas Bagus Wiranata Kusuma Masyhudi Muqorrobin
In addition to being consistent with the specifications in Eqs. ( 2) and ( 3), the model in Eqs. ( 6)-( 7) describes the intertemporal interaction between output and the factor input included in the production function.Once the equilibrium conditions represented by the cointegrating relations are imposed, the VEC model describes how, in each time period, output growth is adjusting towards its long-run equilibrium state.Since the variable are supposed to be cointegrated, then in the short term, deviation of output from its long-run equilibrium path will feed back on its future changes in order to force its movement towards the long-run equilibrium state.The cointegrating vectors from which the errorcorrection terms are derived are each indicating an independent direction where a stable, meaningful long run equilibrium state exists.The coefficients of the error correction terms, however, represent the proportion by which the long-run disequilibrium in the dependent variables is corrected in each shortterm period.
The Johansen (1988) approach estimates the long run or cointegrating relationships between the non stationary variables using a maximum likelihood procedure which tests for the cointegrating rank r and estimates the parameters α and β.

Data and Variables Definitions
Data used in the analysis are annual time series on real GDP and energy use for Malaysia during the period 1980-2010.The variables' notations and definitions are as follows.

E : Total Energy consumption in Kj Y : Real GDP in US Dollars
All variables are transformed into their natural logarithm so that their first differences approximate their growth rates.The data were retrieved from World Bank Database and Energy Information Administration from US Government.

Test Results for Unit Roots
Since, the VEC specification in Eqs. ( 6) to (7) requires that some or all the variables are integrated of order one, we herein investigate the stationarity status of the variables using both the augmented Dickey-Fuller (ADF) and the Philips-Perron (PP) tests for unit roots.The null hypothesis tested is that the variable under investigation has a unit root against the alternative that it does not.In terms of lag length, we choose the Akaike Information Criterion (AIC), Schwarz Information Criterion (SC), and Hannan-Quinn Information Criterion (HQ) to determine the optimal lags after testing for first and higher order serial correlation in the residuals.Table 4. reports the results of testing for unit roots in the level variables as well as in their first difference.In the first half of the table 4.1, the null hypothesis that each variable has a unit root cannot be rejected by both tests.However, after applying the first difference, both tests reject the null hypothesis.Since the data appear to be stationary in first differences, no further tests are performed.We, therefore, maintain the null hypothesis that each variable is integrated of order one.

Test Results for Cointegration
Given the results of unit roots, we now use the Johansen (1988), Johansen (1991), Johansen (1992) and Johansen and Juselius (1990) techniques to test for cointegration between the variables within a VEC model as specified in eqs ( 6) and ( 7).The results of testing for the number of cointegrating vectors are reported in Table 4. which presents both the maximum eigenvalue (λmax) and the trace statistics, 1% and 5% critical values as the corresponding λ values.
According to Table 5, we can see that both tests suggest there is no cointegration between observed variables at both 1% and 5% levels.Thus, there is no evidence for the presence of a long run relationship between the variables in all system we estimate.

Test Results for Granger-Causality
Table 6.reports the results of the granger causality tests.These tests are conducted using a joint F-statistic for the exclusion of one variable from one equation as illustrated above.The results of these tests indicate that granger causality is no longer running in both directions between output growth and energy use.According to Table 7., the results suggest that the proportion of energy use attributable to output formation is increasing.In the second period, its contribution is around 5.7% and becomes 32.67% in the 30 period.Meanwhile, the contribution of output formation towards energy consumption is roughly 4.9% in the second period and jumps to 25.77% in the 30 period.

DISCUSSION
The results of the granger causality tests of this research are in line with findings by Erol and Yu (1987), Yu and Jin (1992), and Masih and Masih (1996), who obtained similar results on other countries as well as Malaysia case.The granger causality tests conducted above indicate only the existence of causality.They do not, however, provide any indication on how important is the causal impact that energy has on output formation.For example, when there is a shock to energy supply, it would also be interesting to gauge by how much this shock will affect the output formation.In addition, it is very important to know how long the effect of such a shock will last.In order to provide such explanation, we decompose the variance of the forecast-error of output into proportions attributable to innovations in variable energy in the system including its own.Impulse response analysis is taken into account to further capture temporal responses of a variable to its own innovation and innovations in other variables in the system.The function can observe whether the response of observable variables is temporal or persistent in nature.
Even though there is no longer co-integration between two variables, the Table 5. indicates either energy or future growth of output in Malaysia has moderate impact each other.At least, in the short run, the shock both in energy use and output formation would create instability or disturbances on demand for energy and generated output.Stabilization policy through fiscal policy could be one proposed exercise for maintaining and monitoring the level of stability in the economy.Those findings ultimately confirm on what assumptions posed by Solow model that in the case of small economy, natural resources (energy consumption) is not a substantial factor led to output growth and therefore Malaysia cannot gain much on energy sector for driving its output.Given such condition and looking at mentioned assumptions, specialization idea in energy formation should be then put as a negligible portion upon GTP roadmap since it does really hamper the long run growth in the country.

CONCLUSION
This paper attempted to analyse the causal relationship between energy and output growth in Malaysia.Based on the neo-classical one sector aggregate production.By using time series data, the findings suggest that no cointegration exists over the observation, particularly between energy use and output growth.In addition, the short-run dynamics of the variables show that the flow of causality is not running in both directions between observable variables.With this, our results seem to significantly comply the neo-classical assumption that energy is neutral to growth.Consequently, we deduce that energy is not a limiting factor to output growth in Malaysia, and hence, shocks to energy supply will not have a negative effect to output.In addition, due to unbirectional between growth to energy use exists, it implies the output growth is not only dependent upon energy consumption.All in all, Malaysia government has a wide range of creative policy measure to respond the dynamic of energy prices with respect to growth.The vision 2020 achievement is finally and totally in the hand of government regardless energy use volatility.

Table 3 .
Overview of Selected Studies

Table 4 .
Test Results for Unit Roots *** denotes for 1% significant level.

Table 5 .
Testing the Rank of Cointegration

Table 6 .
Test Results for Granger Causality

Table 7 .
Results of Variance Decomposition