Palgrave Handbook of Econometrics: Applied Econometrics

708 Panel Methods to Test for Unit Roots and Cointegration

Table 13.10 Panel cointegration estimation results for SO 2

FM-OLS D-OLS Two-step

Fixed effects

ln yit 5.26 1.35 4.80

(35.65) (6.93) (13.34)

(ln yit)^2 –0.26 −0.01 −0.22

(-28.22) (−0.60) (−10.18)

TP (in $) 24,720 − 54,671

Fixed effects and linear trends

ln yit 4.82 −12.63 4.12

(34.45) (-68.80) (11.74)

(ln yit)^2 −0.23 0.86 -0.18

(−26.35) (16.12) (−8.34)

TP (in $) 35,534 − 93,381

Notes: Estimation results obtained with the panel cointegration estima-

tion methods described in sections 13.3.1–4. The upper panel contains

the results with only fixed effects included in the regression and the

lower panel reports the results with both fixed effects and linear trends

included. Bold indicates significance of the estimated coefficients at the

5% level and the numbers in brackets are thet-values. TP denotes the

implied turning point.

Table 13.11 Estimation results with de-factored observations

DF-1 DF-2

One-way Two-way Incl. Ft One-way Two-way Incl. Ft

ln yit 0.86 0.79 0.78 0.92 0.66 0.69

(14.75) (13.90) (13.47) (14.86) (10.80) (11.30)

(ln yit)^2 0.48 0.44 0.38 0.55 0.10 0.22

(12.24) (11.30) (9.60) (6.34) (1.09) (2.52)

Notes: One-way refers to equations including fixed individual effects, Two-way refers

to equations including fixed individual and time effects and Incl.Ftrefers to equations

including fixed individual effects and the estimated common factors of both SO 2 emis-

sions and GDP.DF-1 indicates that the common factors are obtained from the model

without linear trends andDF-2 indicates that the common factors are obtained from

the model with linear trends. Bold indicates significance of the estimated coefficients

at the 5% level and the numbers in brackets are thet-values.

The estimation results just mentioned are robust in a number of ways. First, they
are qualitatively unchanged if the de-factorization method allows for structural
breaks, as found when using the Bai and Carrion-i-Silvestre (2007) methodology.
The results are also robust to using smaller sets of countries or shorter time periods
to rule out the potential structural changes. Also, when allowing for parameter