Estimating energy efficiency using panel stochastic frontier approach: investigating its asymmetric impacts on employment in India

Abstract

Purpose This study aims to examine the asymmetric effects of energy efficiency on employment in India. Instead of relying on partial factor energy efficiency measures, this study uses a total factor energy efficiency (TFE) measure to estimate sector-specific energy efficiency for empirical investigation. Design/methodology/approach Multi-sectoral panel data for India from 2000 to 2014 are considered for empirical estimation. The sector-specific energy efficiency estimates (using the TFE measure) are estimated in the initial stage using the stochastic frontier approach (SFA). Then the asymmetric effect of energy efficiency on employment is investigated by using a non-linear panel autoregressive distributed lag model. Findings The estimates of energy efficiency display that there is not much significant change in the trend of average energy efficiency over the period. The negative and statistically significant value of the error-correction term confirms the existence of asymmetric cointegrating relationship between energy efficiency and employment in India. Moreover, in the empirical findings, the positive and negative shocks in energy efficiency provide a long-run asymmetric and short-run symmetric effect on employment in India. Originality/value Rather than depending on the absolute measure of energy efficiency (energy to output ratio), this study estimates the sector-specific energy efficiency for India using panel SFA, which provides a relative measure of energy efficiency. Moreover, to the best of the authors’ knowledge, it is the first empirical study investigating the asymmetric impact of energy efficiency on employment at an aggregate level in developing countries like India. By contrast, previous studies have either concentrated on the symmetric effect of energy efficiency on employment or primarily restricted to developed countries.

Investigating asymmetric impacts of total factor energy efficiency on carbon emissions in India

“…Estimates of ENEF have been largely based on DEA and SFA in empirical research based on the total factor energy e ciency (TFENEF) framework (Haider and Mishra 2021). It has been noted that the TFENEF indication is more reliable than the PFENEF indicator for measuring ENEF (Zhou et al 2012;Mahapatra and Irfan 2022). One reason could be that PFENEF ignores the role played by and substitution effect of other important inputs (such labour and capital) in the manufacturing process (Zhou et al 2012).…”

“…Both the nonparametric DEA and the parametric SFA can be used to calculate the TFENEF indicator, however according to Zhou et al (2012), the latter yields more accurate results. The most likely explanation is because non-parametric methods are more likely to produce spurious results due to their inability to correct for statistical noise, especially when dealing with data that contains measurement errors (Zhou et al 2012;Mahapatra and Irfan 2022). Therefore, the SFA technique of ENEF assessment is used in the present study, which lls in the gaps left by the DEA method by offering a parametric method for determining the TFENEF.…”

“…The main aim of this study is to explore the asymmetries of ENEF, and the positive and negative shocks in ENEF can be modi ed following Mahapatra and Irfan (2022). The positive and negative shocks in ENEF are estimated employing the partial sum decomposition method advocated by Shin et al (2014) and proposed in the study of , which can be stated as:…”

“…The PMGE assumes that the long-run dynamics are homogenous across cross-sections in the sample panel but the short-run dynamics are heterogeneous (Boufateh and Saadaoui 2020). The MGE assumes that both long-and short-run impacts are homogenous across cross-sections in the sample panel (Kouton 2019;Mahapatra and Irfan 2022). The Hausman speci cation test can be conducted to select the appropriate estimator between the MGE and PMGE (Salisu and Isah 2017).…”

“…Fifth, this study uses the novel nonlinear panel autoregressive distributed lag (NPARDL) modelling framework to inspect the asymmetric impacts of ENEF on the CAE. Sixth, the dynamics of short-and long-run effects can be examined for in a single model arrangement (Shin et al 2014;Mahapatra and Irfan 2022).…”