Scribd
Upload
22 views3 pages

Model

Uploaded by

KHIEM HUOL GIA
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
22 views3 pages

Model

Uploaded by

KHIEM HUOL GIA
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 3

The function used in this study includes the following variables: Carbon

dioxide emissions (CDE), Renewable energy consumption (REC), GDP per


capita growth (GDPG), Energy intensity level of primary energy (EI),
Agricultural productivity (AP), Total natural resources rent (NRR). which
leads to the following extended production function:
CDE = f(Renewable energy consumption, GDP per capita growth, Energy
intensity level of primary energy, Agricultural productivity, Total natural
resources rent)
The simplified Carbon dioxide emissions function is assumed to be Cobb-
Douglas form:
α1 α2 α3 α4 α5
CDE ¿= A ¿ GDP¿ REC ¿ EI ¿ AP¿ NRR ¿
When taking logs of variables without percentage units of the equation
above, the following linear multivariate regression is produced:
ln Y ¿ =α +α 1 GDP¿ + α 2 ln REC ¿ + α 3 ln EI ¿ +α 4 AP ¿ + α 5 NRR ¿

The specified equation is estimated employing a time series Autoregressive


Distributed Lag (ARDL) model, a methodology introduced by Pesaran and
Shin (1998) and Pesaran, Shin, and Smith (2001). This modeling framework
addresses limitations observed in prior literature by discerning between
short- and long-term impacts. Given the constraints posed by a small sample
time period, opting for a separate empirical equation for each country within
the sample is deemed suboptimal. Consequently, a panel estimation
approach is deemed more appropriate, particularly in scenarios where data
availability is constrained.
Panel estimation leverages both time and cross-sectional dimensions,
thereby augmenting the total number of observations and their variability.
Furthermore, this method mitigates noise inherent in individual time-series
estimations, contributing to a more reliable inference. The adoption of the
ARDL model in a panel setting enhances the robustness of the analysis by
accommodating both short- and long-term dynamics in the relationship
under investigation.

Panel second-generation unit root tests


In panel data analysis, the assessment of stationarity issues is crucial,
requiring the conduct of unit root tests. These tests, advocated by Levin-
LinChu (2002) (LLC) and Im-Pesaran-Shin (2003) (IPS), are fundamental for
evaluating stationarity in panel data. They ascertain whether variables
exhibit stationarity at the level and must be stationary at the first difference.
Moreover, Levin et al. (2002) demonstrated the superior efficiency of panel
unit root tests compared to time series unit root tests. In this study, we
employ two panel unit root test methods: Levin et al. (2002), Breitung
(2000), and Im et al. (1997). However, the IPS unit root test holds greater
significance than the LLC unit root test due to its suitability for regression of
heterogeneity unit root.
Panel Autoregressive Distributed Lag Model
After conducting unit root and cointegration analyses, we proceed to estimate the Autoregressive
Distributed Lag (ARDL) model. The ARDL model is designed to differentiate between short-
term and long-term coefficients, making it particularly suitable for reliable application in
situations characterized by limited sample periods. Notably, Pesaran and Shin (1998)
demonstrate that, even in cases of small sample sizes, the long-term parameters exhibit super-
consistency, while the short-term parameters display √T consistency. Consequently, the
formulation of the equation transitions into a panel ARDL (p, q1, q2, q3, q4, q5) equation, where
p denotes the lags of the dependent variable, and q represents the lags of the independent
variables. The panel ARDL equation is expressed as follows:
p q1 q2 q3 q4 q5
()lnCDEit =αi+ ∑ α 1 , ij lnCDE i ,t − j + ∑ α 2 , ij GDPG i ,t − j + ∑ α 3 ,ij lnREC i ,t − j+ ∑ α 4 ,ij lnEii , t − j + ∑ α 5 ,ij APi ,t − j + ∑ α
j=1 j=0 j=0 j=0 j=0 j=0

where i = 1,2,3,...N and t = 1,2,3,...T, αi represents the fixed effects, α1 -α5 is


the lagged coefficients of the independent variables and the regressors and
εit is the error term which is assumed to be white noise and varies across
countries and time. In a panel error correction (ECM) representation equation
() is formulated as follows:
p q1 q2 q3 q4
lnΔ CDE¿ =α i+ ∑ α 1 , ij lnΔCDEi , t − j+ ∑ α 2, ij lnΔGDPG i ,t − j + ∑ α 3 ,ij lnΔREC i ,t − j + ∑ α 4 , ij lnΔEii , t − j +∑ α 5 ,ij lnΔAP
j=1 j=0 j=0 j=0 j =0

In this context, the symbol ∆ represents the first difference of variables. The
coefficients α1 to α5 are associated with short-term effects, while β1 to β5
pertain to the long-term effects of GDPG, debt, labor, human capital, and
physical capital, respectively. To assess the short-term relationship, the
approach suggested by Hendry (1995) is employed. The short-term impact of
debt on growth is computed using ∑qj=1 1 α2,i j (1-∑pj=0 α1,i j) for
significant coefficients.
Upon establishing a long-term relationship between the dependent variables
and regressors, the panel Error Correction Model (ECM) (as expressed in
Equation ()) can be articulated as follows:
p q1 q2 q3 q4
lnΔ CDE¿ =α i+ ∑ α 1 , ij lnΔCDEi , t − j+ ∑ α 2, ij lnΔGDPi ,t − j+ ∑ α 3 , ij lnΔREC i ,t − j + ∑ α 4 ,ij lnΔEi i ,t − j + ∑ α 5 ,ij lnΔAPi ,
j=1 j=0 j=0 j=0 j=0

Where θi signifies the coefficient of the Error Correction Model (ECM), serving as a measure of the
adjustment speed towards long-run equilibrium that occurs each year. The selection of the optimal lag
length for the ECM model is conducted using Akaike's lag selection criteria. Given the constraints posed
by the limited number of annual observations, a maximum lag length of three is chosen to determine the
most suitable specification.

You might also like