Skip to main content
. 2023 Feb 16:1–26. Online ahead of print. doi: 10.1007/s12197-023-09617-y

Table 5.

CS-ARDL regressions

(1) (2) (3) (4)
Short-run coefficients
   TGE / GDP –1.7157*** –1.7012*** –0.6801* –0.7291**
(0.5962) (0.6231) (0.3713) (0.3698)
   PGE / TGE 1.6403*** 1.2185
(0.6164) (1.0571)
   UGE / TGE –1.5142** –1.2502
(0.5854) (1.0289)
   GDP 0.4826 0.4730
(0.3194) (0.3550)
   INTEREST -0.0049** -0.0033*
(0.0023) (0.0021)
Long-run coefficients
   TGE / GDP –5.5561*** –5.6710*** –2.1134*** -2.3279***
(0.5962) (0.6231) (0.3713) (0.3698)
   PGE / TGE 5.3122*** 3.7863***
(0.6164) (1.0571)
   UGE / TGE –5.0477*** –3.9920***
(0.5854) (1.0289)
   GDP 1.4996*** 1.5103***
(0.3194) (0.3550)
   INTEREST -0.0031* –0.0046*
(0.0018) (0.0025)
   ECT(t-1) –0.3088** –0.3000* –0.3218** –0.3132*
(0.1399) (0.1529) (0.1454) (0.1572)
N*T 608 608 489 489
Groups 28 28 25 25
RMSE 0.0566 0.0570 0.0513 0.0517
CD p-value 0.0713 0.0513 0.108 0.109
Hausman p-value 0.632 0.115 0.454 0.090

Notes: PMG estimates. Dependent variable: Real private investments per capita (logarithm). Linear country-specific time trends are included in the models but not reported. The standard errors (reported in square brackets) are calculated through the delta method. The p-values of both cross-sectional dependence and Hausman tests are reported. The asterisks ***, **, * denote significance at 1%, 5%, 10%, respectively. The optimal lag length is selected according to the value that minimises AIC