An Exploratory Analysis of Electricity Consumption and Stock Prices: Evidence From OECD Countries

Abstract

This study explores the impact of electricity consumption, as a proxy for economic activity, on stock prices for a panel of 23 Organization for Economic Cooperation and Development countries for the period 1992 to 2010. Heterogeneous panel cointegration tests reveal a long-run equilibrium relationship between stock prices and electricity consumption, while controlling for factors related to arbitrage pricing models and macroeconomic fundamentals. Finally, the panel error correction model shows bidirectional causality in both the short-run and long-run between electricity consumption and stock prices.

Keywords:

• electricity consumption
• OECD countries
• panel cointegration
• stock prices

Notes

¹ Ozturk (2010) and Payne (2010) provide extensive surveys of the electricity consumption-growth literature.

²Indeed, interest in generating low-carbon electricity production and mechanisms to substitute alternative energy sources for electricity generation may impact overall electricity consumption patterns over time (see Budzianowski and Budzianowska [2012] and Budzianowski and Gomes [in press] for low-carbon electricity production strategies).

³Annual data is used due to the unavailability of higher frequency data for electricity consumption. The study employs year-on-year electricity consumption data, December this year to December next year for the case of the US and April this year to April next year for the remaining OECD countries, which alleviates the impact of weather-induced within-year seasonal variation in electricity consumption.

⁴To proxy earnings, we use the definition of net operational income. Given our use of annual data, book value and earnings per share were simply calculated as the sum of individual items, e.g., firms listed on the stock market.

⁵RATS software (Version 7.0, Estima, Evanston, IL) is used in the empirical analysis.

⁶The Carrion-i-Silvestre et al. (2005) panel stationarity test did not reveal any statistically significant structural breaks.

⁷Preliminary tests for dynamics heterogeneity following Holtz-Eakin (1986) reveal heterogeneity in both the dynamics and error variance across groups. Results are available upon request.

⁸The heterogeneous panel cointegration test allows for cross-section interdependence with different individual effects.

* denotes statistical significance at 1%.

⁹Speed of adjustment is measured by the reciprocal of the absolute value of the coefficient on the error correction term.