The puzzle of energy efficiency in Turkey: combining a multiple criteria decision making and the time series analysis

ABSTRACT

Enhancing energy efficiency is important for sustainable economic development and environmental-related issues such as carbon emissions. Therefore, measuring energy efficiency is crucial for countries to achieve economic and environmental objectives. In this study, we utilize the TOPSIS method with the intent of monitoring the energy efficiency of Turkey from the 1980 to 2018 period. Several variables such as renewable and nonrenewable energy consumption, GDP, labor force, and capital stock are employed in calculating energy efficiency scores. Thereafter, we analyze the long-term, symmetric and asymmetric effects of some contextual variables such as capital stock-labor force ratio, renewable energy consumption-total energy consumption ratio, and carbon emission-total energy consumption ratio on energy efficiency via ARDL and nonlinear ARDL models. Results from the linear model indicate that an increase in the renewable energy consumption-total energy consumption ratio and capital stock-labor force ratio increase the energy efficiency. In contrast, an increase in the carbon emission-total energy consumption ratio decreases it. The nonlinear ARDL model, on the other hand, signals asymmetric effects in the long-run but not in the short-run.

KEYWORDS:

• Energy efficiency
• TOPSIS
• ARDL
• nonlinear ARDL
• Turkey

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

¹ The data of nonrenewable energy consumption and renewable energy consumption collected from International U.S. Energy Administration (EIA 2021), CO₂ emissions, GDP and labor force extracted from World Bank database (World Bank 2022), and capital stock obtained from Penn World Table 10.0 (Groningen Growth and Development Centre 2021).

² Limitations on data availability in particular on renewable and nonrenewable energy do not allow to lengthen the sample size.

³ We are grateful to an anonymous referee for pointing out us to clarify this issue.

⁴ The asymmetric variables of a random variable X_t by partial sum process are.

$X_t^{+}={\sum }_{j=1}^t\mathrm{\Delta }{X}_j^{+}={\sum }_{j=1}^t\mathrm{m}\mathrm{a}\mathrm{x}\left(\mathrm{\Delta }{X}_j,0\right)$ for positive effects and

$X_t^{-}={\sum }_{j=1}^t\mathrm{\Delta }{X}_j^{-}={\sum }_{j=1}^t\mathrm{m}\mathrm{i}\mathrm{n}\left(\mathrm{\Delta }{X}_j,0\right)$ for negative effects.

⁵ ${\mathrm{H}}_0:{\alpha }^{+}={\alpha }^{-}$; ${\mathrm{H}}_0:{\lambda }^{+}={\lambda }^{-}$ and ${\mathrm{H}}_0:\hspace{0.17em}{\zeta }^{+}={\zeta }^{-}\hspace{0.17em}$denote the null hypotheses expressing long-run symmetry for REN, Co2 and C/L, respectively. Similarly, the null hypothesis of the short-run weak symmetries for REN, Co2 and C/L: ${\mathrm{H}}_0:\sum _{j=0}^{p_1}{\vartheta }_j^{+}=\sum _{j=0}^{p_1}{\vartheta }_j^{-}$; ${\mathrm{H}}_0:\sum _{j=0}^{p_2}{\delta }_j^{+}=\sum _{j=0}^{p_2}{\delta }_j^{-}$and ${\mathrm{H}}_0:\sum _{j=0}^{p_3}{\mathrm{\partial }}_j^{+}=\sum _{j=0}^{p_3}{\mathrm{\partial }}_j^{-}$.

⁶ For testing the existence of both the linear and nonlinear long – run relationship we keep the lag lengths fixed for all variables, then allow the lag lengths to differ once the cointegration relationship is established, following Pesaran, Shin, and Smith (2001).:

⁷ $\frac{\mathrm{\partial }lnEE}{\mathrm{\partial }t}=\frac{1}{EE}\frac{\mathrm{\partial }EE}{\mathrm{\partial }t}=0.016$. Solving for $\frac{\mathrm{\partial }EE}{\mathrm{\partial }t}$ and evaluating it at the mean value of EE (0.417615) gives us 0.00668184 ≈ 0.0067.

⁸ Correlation coefficients between trend and asymmetric variables varies between 92.3% 99.5% in magnitude.

⁹ Turkey’s openness (sum of exports and imports of goods and services as a percentage of GDP) increased from 5.7% in 1960 to 61.1% in 2020 (World Bank 2022). A particular increase should be emphasized with liberal economic policies after 1980 and Customs Union with the European Union (EU) in 1996. Additionally, in 2020 the EU’s share in Turkey’s exports and imports in goods are 41.3% and 33.4% respectively (European Commission 2021).