آیا مصرف انرژی بیشتر رشد اقتصادی را تقویت می کند؟کاربرد مدل رگرسیون آستانه ای غیر خطی

This paper separates data extending from 1971 to 2002 into the energy crisis period (1971–1980) and the post-energy crisis period (1981–2000) for 82 countries. The cross-sectional data (yearly averages) in these two periods are used to investigate the nonlinear relationships between energy consumption growth and economic growth when threshold variables are used. If threshold variables are higher than certain optimal threshold levels, there is either no significant relationship or else a significant negative relationship between energy consumption and economic growth. However, when these threshold variables are lower than certain optimal levels, there is a significant positive relationship between the two. In 48 out of the 82 countries studied, none of the four threshold variables is found to be higher than the optimal levels. It is inferred that these 48 countries should adopt a more aggressive energy policy. As for the other 34 countries, at least one threshold variable is higher than the optimal threshold level and thus these countries should adopt energy policies with varying degrees of conservation based on the number of threshold variables that are higher than the optimal threshold levels.

On February 2, 2007, a report from the Intergovernmental Panel on Climate Change (IPCC) indicated that global atmospheric concentrations of carbon dioxide (CO2), methane, and nitrous oxide have increased markedly as a result of human activities since 1750 and now far exceed pre-industrial levels determined from ice cores spanning many thousands of years. The increase in CO2 has mainly originated from the use of fossil fuels and partly from the changes in land use. The emissions of CO2 and other gases from burning fossil fuels and other processes trap heat from the sun in the atmosphere, much like a greenhouse. Global warming, of course, raises the temperature in the air and in the ocean, and raises the sea level as well. Without properly controlling for the phenomenon, serious consequences on the eco-system and the extinction of many creatures will be unavoidable. The IPCC report places the blame for global warming mainly on the release of CO2 from the use of fossil fuels. Fossil fuels are a major source of energy for industrial production, residential consumption, and transportation.1 It is obvious that energy use plays an important role in our daily lives and economic activities. Since the fossil fuels are relatively cheap and readily available, part of the benefit from a greater use of fossil fuels is the increase in production and living standards. These benefits, however, must be offset, to some extent, by a negative externality that arises from an increase in global warming and environmental pollution. Whether or not the increased economic benefits from energy consumption will outweigh the negative externality depends on the empirical evidence of a positive causal relationship between energy consumption and economic growth. If the empirical evidence indicates that energy consumption (in terms of the growth rate) Granger-causes economic growth, it is suggested that a more aggressive energy policy should be followed. On the other hand, if economic growth Granger-causes energy consumption or there is no causal relationship between energy consumption growth and economic growth, the policy-maker should implement a more conservative energy policy since energy consumption may not bring about economic growth but may increase CO2 emissions into the atmosphere and accelerate global warming.2 Granger causality has been widely used in the literature in analyzing the relationship between energy consumption and economic growth. Kraft and Kraft (1978) discovered that economic growth leads energy consumption. Employing the same US data, however, Stern (2000) found that energy consumption leads economic growth. Therefore, two conflicting empirical results for the same country using the same data were found. Furthermore, Akarca and Long (1980), Yu and Jin (1992), and Soytas and Sari (2003) discovered that there is no causal relationship between energy consumption and economic growth. In addition, Lee (2005) showed that a feedback relationship existed between the two.3 Such inconsistent results were also found in emerging countries such as Taiwan and South Korea. The main reason for the inconsistencies may involve using different periods in time series data, obtaining an insufficient sample, or failing to take into account the nonlinearity due to certain country-specific factors. In addition, using the Granger-causality approach to investigate the causal relationship between energy consumption and economic growth in the previous literature led to three possible problems: (a) whether or not the yearly data were sufficient to represent the long-term relationship between the two; (b) the inability of the yearly data to eliminate the problems of short-term fluctuations due to business cycles and structural change; and (c) the failure to delineate countries with special features in terms of different causal relationships. Since the relationship between energy consumption and economic growth is inherently a long-term one, a biased estimate may be the result of an insufficiently large sample size in the time series, the existence of structural changes, or short-term economic fluctuations. Another reason for the inconsistent empirical results may emanate from the omission of specific characteristics in certain countries affecting the relationship between energy consumption and economic growth (Soytas and Sari, 2006). For example, high CO2 emission countries may be characterized by an overuse of energy or a lack of regulations to enforce proper energy consumption. As a result, the environmental damage from energy use may outweigh the benefits from economic growth. To tackle the insufficient sample size problem, many researchers have used the panel data approach. Lee and Chang (2007a) separated the data used into 18 developing countries and 22 developed countries and employed a dynamic panel data (DPD) approach to test the causal relationship between energy consumption and economic growth. He discovered that economic growth leads energy consumption growth in the developing countries, while in the developed countries there is a feedback relationship between the two. Huang et al. (2008) used panel data for 82 countries and grouped the data into four categories based on the income levels defined by the World Bank: low-income group, lower middle-income group, upper middle-income group, and high-income group. They employed the DPD approach to investigate Granger causality. They discovered that (a) in the lower and upper middle-income groups, economic growth leads energy consumption positively; (b) in the high-income group, economic growth leads energy consumption negatively; and (c) in the low-income group, no causal relationship exists between energy consumption and economic growth. Their conclusion—a reduction in energy use in the high-income group—was consistent with the policy of decreasing the use of fossil fuels (e.g., extending the summer daylight time and the more extensive use of fluorescent lighting). By grouping on the basis of the difference in the degree of economic development in their research, Lee and Chang (2007a) and Huang et al. (2008), in effect, assume a nonlinear relationship between energy consumption and GDP. Moon and Sonn (1996) employed an endogenous growth model to infer that the economic growth rate rises initially with productive energy expenditure but subsequently declines. In other words, there is an inverse U-shaped nonlinear relationship between energy consumption and economic growth as was evidenced by their empirical results from the yearly data extending from 1968 to 1989 in Korea. Lee and Chang (2007b) used the level of total energy consumption as a threshold variable to investigate the existence of a nonlinear relationship under the one-sector and two-sector growth models. The empirical result from the 1955–2003 annual data in Taiwan indicates that there is an inverse U-shaped relationship between energy consumption growth and economic growth. That is, the relationship between energy consumption and economic growth indicated above is nonlinear and the traditional linear model is no longer appropriate. Although the DPD approach may overcome the shortcoming of insufficient sample size, the problem of biased estimates remains unresolved. The biasness may arise from the inadequate representation of a long-run relationship, the inability to allow for structural change, and the presence of short-term fluctuations from business cycles when using annual data. Last, but not least, it can emanate from failure to segment the data into different panel groups based on a country's characteristics. In order to tackle these problems, many researchers employ average values over time for each cross-sectional unit.4 The advantage of this is that the cross-sectional data are separated into different regimes based on certain threshold values in order to investigate different relationships between energy consumption and economic growth in each regime. This approach helps us calculate the conditions for the existence of different relationships between the growth of energy consumption and economic growth. It is widely recognized that global warming will have a great impact on the welfare of future generations.5 If we can identify the conditions under which a significant relationship exists or does not exist between energy consumption and economic growth, more relevant energy policies can be formed. If energy consumption cannot bring about economic growth in a country, the best policy may be to reduce energy consumption with a view of avoiding a negative impact on the environment. The major contribution of this research is to apply 82-country cross-sectional data to investigate the relationship between energy consumption and GDP growth based on four energy-related threshold variables. These four variables include CO2 emissions (CO2), the efficiency of energy use represented by the total primary energy supply to production per $1 of GDP (tpes/Y), the ratio of industrial energy consumption to total energy consumption (ind/tfc), and per capita energy consumption (ec). Among those four variables, the ec variable was used in the past by Moon and Sonn (1996) and Lee and Chang (2007b). As far as we know, this paper is the first to use the other three threshold variables. Although the cross-sectional data analysis has advantages as indicated above, it also has certain disadvantages. To test the causal relationship using cross-sectional data, the potential relationship between consumption and economic growth is assumed a priori in the growth equation. In other words, the energy consumption growth variable is assumed to be an explanatory variable for explaining economic growth as was indicated in prior research. This paper uses relevant data covering the period from 1972 to 2002 for 82 countries. Since the energy crisis led to structural change, the data are separated into two periods: the energy crisis period (1972–1980), or Period-I, and the post-energy crisis period (1981–2002), or Period-II. First, we take the yearly averages in these two periods to establish the 82-country cross-sectional data set. These variables are then tested statistically and grouped into different regimes based on certain energy-related features (called threshold variables, qi). Under different regimes, the threshold regression model is employed to investigate the relationships between energy consumption and economic growth with different qi's. In the nonlinear test, the candidates for the threshold variables in our regression model are CO2 emissions (CO2), the efficiency of energy use (tpes/yr), the ratio of industrial energy consumption to total energy consumption (ind/tfc), and per capita energy consumption (ec). From the different regimes classified by these threshold variables, we discover that the relationship between energy consumption and economic growth is rather different. Furthermore, irrespective of whether a linear or nonlinear structure is used, the relationship between energy consumption and economic growth also differs between the energy crisis period and the post-energy crisis period. The major finding of this paper is that it distinguishes countries with strong relationships between energy consumption and economic growth from countries without such relationships. We also discover that, in Period-II, 34 of the 82 countries exhibit a positive relationship between energy consumption and economic growth, regardless of which threshold variable is used. As such, these countries should adopt a more aggressive energy policy. For the remaining 48 countries, no such relationship can be found between the two using either of the threshold variables. These 48 countries should, to different extents, adopt conservative energy policies depending on the number of threshold variables through which no such a relationship is found. This paper is organized as follows. Following Section 1, which discusses the motivation behind the paper and provides a review of the related literature, Section 2 introduces the econometric models and data. Section 3 then analyzes and discusses the empirical results. The final section, Section 4, provides the concluding remarks and the policy implications derived from this research.

With weather patterns becoming more erratic, there is increasing evidence to suggest that CO2 emissions from the use of fossil fuels is the main source of global warming. The major purpose of energy use is to increase economic growth. However, the use of energy also has a negative impact on global warming, environmental pollution, and resource exhaustion. Whether or not the use of energy can bolster economic growth has been a focus of attention for academics and policy-makers alike. Unfortunately, there have not been consistent findings in the literature based on the results of the Granger-causality tests. One important reason for this involves the use of time series data. Problems of small sample size, structural change, and/or short-term variations due to the business cycle may render the time series estimates invalid. Biased results may also ensue. To overcome the above-mentioned pitfalls as a result of using time series data, this paper separates the data covering the period from 1971 to 2002 into the energy crisis period (1971–1980) and the post-energy crisis period (1981–2000) for 82 countries in order to accommodate the problem of structural change. The cross-sectional data for yearly averages in these two periods are used to smooth out short-term variations. To identify the conditions under which economic growth is related to energy consumption, we selected CO2 emissions (CO2), energy efficiency (tpes/Y), the ratio of industrial energy consumption to total energy consumption (ind/tfc), and per capita energy consumption (ec) as threshold variables. Our nonlinear test results support the fact that these variables be used as threshold variables because the energy consumption–economic growth equation clearly exhibits a nonlinear relationship. Moreover, the estimated results from these two periods indicate that the relationship between energy consumption and economic growth differs between the energy crisis period and the post-energy crisis period. During the post-energy crisis period, the nonlinear results indicate that, when the CO2, tpes/Y, ind/tfc, and ec threshold variables are higher than their optimal threshold values, there is no significant positive relationship (or even a presence of negative relationship) between energy consumption and economic growth. Conversely, there exists a significant positive relationship between the two if these threshold values are below the estimated optimal values. We also find that, among the 82 countries, 48 countries have no single threshold variable higher than the estimated optimal values. Therefore, these 48 countries may increase their use of energy consumption to promote economic growth. In addition, 17 countries have one threshold variable that is higher than the optimal value (only one country has CO2 emissions higher than the optimal value) and these countries may still use energy consumption to help promote economic growth if CO2 emissions are appropriately controlled. There are eight countries with two threshold variables higher than the estimated optimal values (of which five have a common threshold variable: CO2 emissions). These eight countries should adopt a more conservative energy policy compared with the 17-country group with only one threshold variable. Finally, nine countries have at least three threshold variables higher than the estimated optimal values. Among them, eight countries have CO2 emissions higher than the optimal value. For this nine-country group, the negative externality outweighs the benefits brought about by the energy consumption and as such the energy policy adopted should be to reduce energy consumption and CO2 emissions, and to increase energy efficiency. The policy implications derived from this study need to be clarified. In order to design a sound energy consumption policy, each country needs to take into consideration information on the CO2 emission (CO2), the efficiency of energy use (tpes/Y), the share of industrial energy consumption to total energy consumption (ind/tfc), and per capita energy consumption (ec). It is evident that the CO2 emission from the use of fossil fuels is a major source of global warming. When CO2 emission is higher than a certain threshold level, energy consumption can no longer bring about significant output increase because of the nonlinear relationship between energy consumption and economic growth. The CO2 emission in 14 out of 82 countries is higher than the threshold level estimated from our threshold regression. For such high CO2 emission countries, the use of fossil fuels needs to be reduced. Other energy policies need to be implemented such as increasing the efficiency of energy use, more stringent regulations of CO2 emission, and/or the replacement of fossil-type fuels by cleaner energy sources. For 20 countries with at least one threshold variable higher than the estimated optimal threshold levels, appropriate reduction of energy use, the use of cleaner energy sources, and more efficient use of energy to avoid unnecessary CO2 emission could be a good energy policy to follow. Lastly, for the 48 countries without a threshold variable higher than the estimated level, though energy consumption will bring about output growth, more efficient use of energy and the use of cleaner energy sources are encouraged to combat global warming. Though the use of cross-sectional data is able to identify the possible existence of a correlation between energy consumption growth and economic growth via certain threshold variables, it is nevertheless unable to test the Granger causality between the two. Future analysis on the nonlinear Granger causality between energy consumption growth and economic growth across various economies over the long run requires the use of a dynamic panel data (DPD) model developed by Arellano and Bond (1991). However, one of the problems in applying the DPD model is how to segment the data into different regimes via threshold variable(s) in order to decipher the relationship in detail.14