Is there any relationship between energy and TFP (total factor productivity)? A panel cointegration approach for Italian regions Maria Gabriela Ladu 1 , Marta Meleddu * Department of Economics and Business (DISEA), University of Sassari and CRENoS, Via Torre Tonda, 34, 07100 Sassari, Italy article info Article history: Received 27 March 2014 Received in revised form 1 July 2014 Accepted 4 August 2014 Available online 22 August 2014 Keywords: Total factor productivity Energy use Dynamic panel abstract The aim of this work is to examine the long run relationship between TFP (total factor productivity) and energy consumption in Italy from 1996 to 2008 at a regional level. Differently from previous studies, TFP is employed as a measure of economic growth and therefore of technological change. Dynamic Panel estimation technique is used to investigate whether a connection between TFP and energy use exists and its direction. Results show a bi-directional causality for the Italian regions implying that those regions characterized by a higher TFP would invest more in research rather than in energy intensive activities, improving in that way the use of scarce resources and favoring a sustainable growth. © 2014 Elsevier Ltd. All rights reserved. 1. Introduction A great attention toward the relationship between energy con- sumption and GDP (gross domestic product) has been devoted by the literature in the last few decades for developed and developing countries [1,3,15,24,28,30,38,42]. Stern [39] provides an exhaustive literature review where he highlights that energy use and output are, theoretically and empirically, closely connected. More recent contributions highlight also the relationship between GDP and electrical energy consumption [6,7]. In particular, they find that developed economies requires less electrical energy consumption if compared to developing economies. Operationally, different econometric methods have been employed and following Costantini and Martini [15] and Belke et al. [10] four generations of contributions are identified. The first generation studies employ VAR (vector autoregression), assuming stationarity in time series as in Kraft and Kraft [23] who studied the relationship between energy use and income in the United States for the time span 1947e1974. More recently Fallahi [19] uses MS- VAR (Markov-switching vector auto regressive) models, rather than VAR. MS-VAR models are capable of detecting changes in the relationship between variables. The second generation contributions introduced non- stationarity, cointegration analysis and ECM (error correction model) to test for Granger causality [13]. However Beaudreau [9] argues that Granger energy-GDP causality tests remain inconclu- sive. The third generation analysis applied Johansen methodology to dynamic cointegration model as in Stern [38] that studied the relation among GDP, quality-weighted energy, labor, and capital. The author highlighted a cointegrating relation between the four variables and that energy Granger causes GDP. Fourth generation studies exploit panel cointegration and causality techniques to study GDP and energy consumption short and long run relationship in more than one country [1,3,10,15,24,28,30,42]. The above contributions studied the causality direction between growth and energy consumption employing GDP as a measure of economic growth. It has been implied that whenever a high level of innovation is present in the economy the energy use decreases but the economic growth does not decline [15]. The findings highlight miscellaneous results due to the different econometric methods employed, the miscellaneous time periods and countries heterogeneity. As Dobnik [17] underlines, the outcome of the causality analysis in this field has four implications. The first implication is connected to the growth hypothesis that considers energy consumption as a key element for growth. In this case, the economy is called ‘energy dependent’ and a decrease in energy consumption causes a decrease in real GDP. The second implication, that is the conser- vation hypothesis, asserts that policies focused towards lower en- ergy consumption may have little or no adverse impact on real GDP. * Corresponding author. Tel.: þ39 079 2017323. E-mail addresses: mmeleddu@uniss.it, martameleddu@gmail.com (M. Meleddu). 1 Tel.: þ39 079 2017323. Contents lists available at ScienceDirect Energy journal homepage: www.elsevier.com/locate/energy http://dx.doi.org/10.1016/j.energy.2014.08.018 0360-5442/© 2014 Elsevier Ltd. All rights reserved. Energy 75 (2014) 560e567