علیت بین مصرف انرژی و رشد اقتصادی: یک تحلیل چند بخشی با استفاده از داده های پانل غیر ثابت هم انباشته

The increasing attention given to global energy issues and the international policies needed to reduce greenhouse gas emissions have given a renewed stimulus to research interest in the linkages between the energy sector and economic performance at country level. In this paper, we analyse the causal relationship between economy and energy by adopting a Vector Error Correction Model for non-stationary and cointegrated panel data with a large sample of developed and developing countries and four distinct energy sectors. The results show that alternative country samples hardly affect the causality relations, particularly in a multivariate multi-sector framework.

The increasing attention given to global energy issues and the international policies needed to reduce greenhouse gas (GHG) emissions have given a renewed stimulus to research interest in the linkages between the energy sector and economic performance at country level. The empirical analyses and the adopted models for investigating these linkages highly depend on the development level and economic structure of the countries considered. Toman and Jemelkova (2003) argue that most of the literature on energy and economic development discusses how development affects energy use rather than vice versa. This strand of literature considers economic growth as the main driver for energy demand and only advanced economies with a high degree of innovation capacity can decrease energy consumption without reducing economic growth. Stern and Cleveland (2004), on the other hand, have stressed the importance of considering the effect of changes in energy supply on economic growth in both developed and developing countries. When energy supply is considered a homogenous input for the production function, economic development is harmed if policy constraints affect energy supply. When energy services are differentiated, emphasizing the existence of higher and lower-quality forms of energy, society should make a choice in terms of an optimal energy mix, considering that higher-quality energy services could produce increasing returns to scale. This means that energy regulation policies could provide impulse to economic growth rather than be detrimental to the development process, since they support the shift from lower-quality (typically less efficient and more polluting) to higher-quality energy services.These alternative views have important policy implications concerning, for example, aspects such as the development level of the considered country or the distributive effects related to the introduction of stringent energy (and environmental) regulations. If we consider highly industrialized countries, total energy use has increased, energy efficiency has improved and energy intensity has steadily fallen, especially in the industrial sector. Stabilization of greenhouse gas concentrations requires reductions in fossil fuel energy use which is a major essential input throughout all modern economies. If energy conservation and a switch from fossil fuels to alternative energy sources can be affected using new energy-efficient technologies, the trade-off between energy and growth becomes less severe. Moreover, if the development process is in the deindustrialization phase, the increasing importance of value-added produced by the service sector could lead to a global reduction in energy consumption due to a minor weight represented by energy-intensive industrial sectors. Nonetheless, empirical analysis has shown that energy regulations and the shifting in production structure do not necessarily lead to a consistent reduction in global energy consumption. This evidence is explained as a “rebound effect”, postulated first by Brookes, 1990 and Khazzoom, 1980. In some cases, energy-saving technical innovations tend to introduce more energy-using appliances to households and industries causing even more energy consumption as the money saved is spent on other goods and services which require energy to be produced. A stronger implication of the rebound effect is related to a reduction in energy prices that occurs when energy efficiency leads to a reduction in the energy demand (Binswanger, 2001). An innovation that reduces the amount of energy required to produce a unit of energy services lowers the effective price of energy services, resulting in an increase in their demand. The lower price of energy also results in an income effect that increases demand for all goods in the economy and therefore the energy required to produce them (Lovins, 1988 and Newell et al.,1999; Popp, 2002). Therefore, if delinking between economic growth and energy consumption is the aim of energy policies, policy makers should consider some form of energy regulation (taxes, price cap or other) that allows cost of energy services to remain unchanged provided that technological innovation lowers effective energy prices (Bentzen, 2004). Not many empirical studies have analysed this phenomenon by considering different economic sectors: a large part of the literature has investigated energy efficiency only at a general level. This has important policy implications. One of the most accurate contributions is the analysis by Zachariadis (2007) for G-7 countries where energy–economy causality for four sectors (industry, service, residential and transport) is analysed, using alternative estimation methods for each country. If declining energy intensity is observed only for specific sectors and not for the whole economy, differentiated policy measures are required in order to obtain the best results in terms of decoupling economic growth from energy consumption. There are many studies that investigate the strength of the structural linkage between energy and growth using time series analysis for single countries and, more recently, panel datasets, but at the best of our knowledge there are no contributions which adopt a panel approach for analysing energy–economic growth causality at the sectoral level. The purpose of this paper is to provide empirical evidence on the better performance of panel sectoral datasets in explaining the causal linkages between the economy and energy consumption. Moreover, by using energy prices for each specific sector, we can estimate the elasticity parameters related to energy demand changes induced by public regulation, expressed as energy taxes and empirically represented by energy prices. This paper is different from previous contributions in several aspects. The sample adopted for the dataset is rather wider than other contributions based on the panel approach and includes 71 countries, thus allowing a number of considerations on different results emerging from alternative sub-samples consisting of developed and developing countries. The analysis is carried out on the whole economy and on four distinct end-use sectors, industry, service, transport and residential, allowing for specific considerations to be made for each sector divided into the sub-samples examined in this paper. Comparing results from different sectors reinforces the need for a multivariate model that accounts for structural peculiarities of both sectors and countries. A first attempt is provided by including specific energy prices for each end-use sector for OECD countries and the results offer strong advice in favour of multivariate multi-sector models. The rest of the paper is structured as follows. Section 2 provides the methodological strategy for addressing Granger causality in the energy sector with particular emphasis on contributions dealing with non-stationary and cointegrated panel dataset, Section 3 gives a description of the data used in the empirical analysis, Section 4 describes the econometric strategy and presents the empirical results and Section 5 concludes with some policy implications.

This paper provides new empirical insights into the analysis of the causal relationship between energy consumption and economic growth when considering a large sample of developed and developing countries and a sectoral specification. Standard results for non-stationarity and panel cointegration analysis have been found for both economic and energy variables in the period 1960–2005, both for the whole sample and for the two sub-samples considered here. The presence of non-stationary and cointegrated time series in a panel context makes more complex econometric estimates necessary using recent models such as the FMOLS developed by Pedroni (2000). The possible existence of mutual causal relationships between economic and energy variables must be considered in a Granger causality framework by using a Vector Error Correction Model that includes the long-run cointegrating relationship obtained by the FMOLS. The empirical analysis carried out on the full sample and on separate sub-samples, on the whole economy and at disaggregated level has shown a number of interesting results. These implications should be considered when such models are used to calculate income elasticity or when assisting policy makers in energy policy design. Differences in the causality direction have been detected in sub-samples of countries, particularly in the specific sector analysis. In the industrial sector, there is a converging trend in the short-run but the causality directions diverge when a strong causality hypothesis is tested separately for the two sub-samples. For the transport sector, all three kinds of causality show contrasting results for OECD and non-OECD countries revealing that the application of similar energy policies in structurally divergent countries could bring to contrasting effects. On the contrary, when considering the residential sector, it is clear that there are no univocal causality relationships in both developed and developing countries, meaning that policy evaluations and model settings should be performed with caution accounting for endogeneity and mutual causality. These results cast some doubt on the capacity of bivariate models to shape causal relationships in the energy–economy binomial especially when different sectors are investigated. While Zachariadis (2007) has shown that there are divergent results when using alternative estimators or datasets for single countries, we have shown that the same scepticism on bivariate models applies even in a panel context. Working with specific sectors allows the existence of divergent trends to be considered even in a quite homogeneous country sample such as the OECD one. A strong policy advice should come from these first results, when the international community try to involve developing countries in virtuous energy-saving actions, without an explicit effort in shaping policy design appropriately for underdeveloped countries. Looking at the industry and transport sectors, it is worth noting that the causality direction changes when different time horizons are accounted for. In the short-run, it is the economic growth process that determines the energy consumption trend so that it is mainly driven by production demand, and policies oriented towards promoting energy-saving do not seem to affect economic development negatively. On the contrary, long-run causality is bi-directional, showing that changes in energy consumption could influence economic performance and vice versa. When energy prices are included, the picture becomes much clearer, thus stimulating further research in multivariate sectoral energy models. Far from being conclusive, this study allows us to open new research directions in the assessment of public policies and technological innovation in the energy sector. Future research should consider the capital/labour ratio, the role of energy prices and taxes and energy regulation on the economic system more appropriately by adopting an induced technical change framework and focusing on a homogeneous country sample such as OECD or the European Union. Further applications of this empirical framework could be the estimation of short and long-run elasticities of energy services related to more disaggregated sectors, in order to calibrate the matrix used by energy models thus producing scenarios on the basis of relationships estimated from observed behaviours.