Abstract
There is a burgeoning literature based on using panel cointegration techniques to
study the relationship between energy consumption and GDP. Most panel cointegration
tests employed take no cointegration as the null hypothesis. The current paper illustrates
how a rejection by such a test cannot be taken as evidence of cointegration for the panel
as a whole, a fact that seems to have gone largely unnoticed in the literature. Hence,
even if the no cointegration null is rejected, this evidence not enough to ensure that the
relationship can be meaningfully estimated, as most (if not all) estimators in the literature
require that the panel is cointegrated as a whole.
study the relationship between energy consumption and GDP. Most panel cointegration
tests employed take no cointegration as the null hypothesis. The current paper illustrates
how a rejection by such a test cannot be taken as evidence of cointegration for the panel
as a whole, a fact that seems to have gone largely unnoticed in the literature. Hence,
even if the no cointegration null is rejected, this evidence not enough to ensure that the
relationship can be meaningfully estimated, as most (if not all) estimators in the literature
require that the panel is cointegrated as a whole.
| Original language | English |
|---|---|
| Pages (from-to) | 359-363 |
| Journal | Energy Economics |
| Volume | 50 |
| DOIs | |
| Publication status | Published - 2015 |
Subject classification (UKÄ)
- Economics
Free keywords
- Panel data
- Cointegration
- Energy consumption.
