The Causal Dynamics Between US Renewable Energy Consumption, Output, Emissions, and Oil Prices

Abstract

This article examines the causal dynamics between renewable energy consumption, real gross domestic product (GDP), carbon emissions, and real oil prices using the Toda-Yamamoto long-causality test procedure over the period 1949 to 2009. The results indicate that renewable energy legislation and policies since 1978 had a positive and statistically significant impact on renewable energy consumption. Though the results suggest that real GDP, carbon emissions, and real oil prices did not have a causal impact on renewable energy consumption, unexpected shocks to real GDP and carbon emissions yielded a positive impact on renewable energy consumption over time.

Keywords:

• causality
• emissions
• oil prices
• renewable energy

Notes

¹ Ozturk (2010) and Payne (2010a,b) provide exhaustive surveys of the empirical literature on the causal relationship between energy consumption and economic growth inclusive of renewable energy.

²PURPA introduced the use of “feed-in” policy which required utilities to purchase power from small renewable generators. Subsequent renewable energy legislation and policies included a myriad of initiatives from state renewable portfolio standards, financial incentives (tax credits, rebates, low-interest loans), net metering policies, and voluntary green power programs, to name a few policies.

³Test statistics associated with Granger-causality have the standard asymptotic distribution to draw valid inferences.

⁴Details of the unit root tests are omitted to conserve space, but are available upon request.

Note: All unit root tests include an intercept and trend.

Critical values for the respective unit root tests are as follows: ADF and PP: ^a(1%) −4.12. DF-GLS: ^a(1%) −3.73.

^a(1%),

^b(5%),

and ^c(10%). +/− denote whether the sum of the lagged coefficients (k + d _max) are positive or negative, respectively. The model is free of serial correlation and autoregressive conditional hetereoscedasticity.

⁵Orthogonalized and generalized impulse responses will be identical only in the case when the covariance matrix is diagonal (see Koop et al., 1996 and Pesaran and Shin, 1998).