Abstract:
In the present paper, a case of fuzzy regression model was estimated for Iranian industrial energy intensity. To do so, at first the trapezoidal fuzzy values of energy intensity observations were calculated based on the Minimizing Entropy Principle Algorithm(MEPA) and then a tripled recursive model was estimated for fuzzified energy intensity. Because of application of the partial adjustment model, we explored the short-run and long-run membership functions for each of the explanatory variable fuzzy coefficients. The estimation results show that the lagged energy intensity values are the only factor which has a positive effect on the industrial energy intensity attitude, whereas other explanatory variables including energy price, value added share and technical efficiency score have negative effect on energy intensity trend in the period (1982-2006). Moreover the estimation results indicated the numerous potential energy saving in Iranian industrial sector which is mainly emerged from pure energy intensity in short-run
Machine summary:
"In this paper, we apply a least-squares approach fuzzy regression model which was introduced by D`Urso(2003) for the estimation of the Energy Intensity regression equation, to Iranian Industrial sector.
5. Specification of Data, Variables and model estimation As noted in previous sections, the purpose of the present paper is to estimate the fuzzy regression model based on the least squares approach for annual end-use energy intensity on its determinant factors for Iranian industrial activities.
Associated with these factors, we introduce, in the present study, explanatory variables such as the end-use energy average price, each sector’s share in industrial total value-added for the index of structural changes, the share of natural gas in total end-use carriers and technical efficiency score calculated by the DEA method 2 for carriers combinations and technical change progress factors, respectively.
3-5, our selected recursive model for estimating the fuzzy regression (2b) can be rewritten as bellow: (3b) (4b) (5b) where, eninmed is the med-point value of energy intensity(fuzzy number), spreleft and spreright are left and right spreads, eninmed(t-1) is the lagged value of eninmed, Enpric, valuads and effiscor are the energy average price, industry sector value added share and efficiency score, respectively, i and t are the time and the industry sector subscripts and Log is the natural logarithm assignment.
The MEPA procedure, with an energy minimizing screen process, subdivides the energy intensity data with threshold values, which allows us to partition the dataset into a number of fuzzy sets with associated membership functions (Tsai, 2006)."