Modeling Energy Consumption for Growth in an Open Economy: ARDL and Causality Analysis for Turkey

Abstract

In a growing open economy, energy consumption plays a critical role. Insufficient energy may constrain output, while exogenous factors, such as shocks in world price, may limit the aggregate growth rate of the economy. This study investigates the long-run equilibrium of energy consumption in Turkey from a unique perspective, taking tourism expansion as a proxy for exogenous growth. The paper uses the Autoregressive Distributed Lag (ARDL) model to determine an econometric relationship between electricity consumption, foreign tourist arrivals, and the economic growth using data during 1960–2010. Granger causality test determines the direction of causality among these variables under conditional error correction model (CECM) supplemented, whenever required, by the Bounds test. Results reveal that foreign tourist arrivals and real gross domestic product (GDP) are significant determinants of a long-run equilibrium relationship with electricity consumption for Turkey. A positive significant relationship has been found between electricity consumption, foreign tourist arrivals, and economic growth in the long run. The estimated ARDL elasticity coefficients indicate that a 1% increase in the number of foreign tourist arrivals and GDP growth cause 0.045% and 0.74% increase in the total electricity consumption of Turkey, respectively, after adjustments consistent with standard procedures.

Keywords:

• Electricity consumption
• Economic growth – Turkey
• Bounds test
• ARDL modeling
• Conditional causality

Notes

¹ PP approach allows for the presence of unknown forms of autocorrelation with a structural break in the time series and conditional heteroscedasticity in the error term.

² The code we utilize here has only model options A and C excluding model B in the original work. Model B incorporates trend stationary alternative with change in trend slope. However, since model C is the most general, it encompasses both models of A and B.

³ For detailed information, see Pesaran, Shin and Smith (2001), pp. 295–296.