Results and discussion
The aim of our research is to uncover the influence of different policy instruments on subsequent investments in RE by private institutional investors over time in a longitudinal research design. Policy makers interested in improving their country’s transition towards RE should implement measures for attracting private institutional investors, as the capital required for large-scale RE projects by far surpasses the available funds of utility companies as well as the public budgets. Institutional investors’ capital played an important role in the development of the RE sector, and establishing a favourable environment for them, including specific policies, should increase capacity additions in the future. With our analysis we provide an integral picture of RE policies and their influence on RE capacity investments by institutional investors. We intend to contribute to the literature surrounding investor behaviour regarding RE technologies, as investors provide funds for large scale deployment (Bergek et al., 2013a; Luthi & Prassler, 2011; Wustenhagen & Menichetti, 2012).
The analysis is conducted on a sectoral basis to allow differentiated policy recommendations. In the following discussion, we highlight significant effective and ineffective policy measures and relate our results to previous studies in this literature stream (Aguirre & Ibikunle, 2014; Marques & Fuinhas, 2012a; Marques et al., 2010).
Table 19 (in the appendix) shows the descriptive statistics of our analysis. The correlation among explanatory variables has been subject to analysis as well. The simultaneous use of several drivers leads to the hypothesis of collinearity among explanatory variables. Table 19 and Table 20 (in the appendix) show the summary statistics and the correlation coefficients for our analysis. The analysis suggests the absence of collinearity among the exogenous (independent) variables.
|
Independent variables
|
PCSE |
|||||||
|
No autocorrelation |
No autocorrelation No autocorrelation |
No autocorrelation |
||||||
|
Multiple RE |
Wind |
Solar |
Biomass |
|||||
|
(I) |
(II) |
_(Ш)_ |
_(IV)_ |
|||||
|
Coeff. |
S.E. |
Coeff. |
S.E. |
Coeff. |
S.E. |
Coeff. |
S.E. |
|
|
EI FI FIjk |
0 69*** |
0.28 |
0.70*** |
0.26 |
1.18*** |
0.26 |
0.09 |
0.24 |
|
EI FI GSik |
0.33 |
0.25 |
0.13 |
0.24 |
0.61*** |
0.23 |
0.61** |
0.24 |
|
EI FI Lik |
-0.36 |
0.26 |
0.30 |
0.27 |
-0.91*** |
0.30 |
-0.16 |
0.32 |
|
EI FI TRik |
0.40** |
0.18 |
0.20 |
0.20 |
-0.01 |
0.28 |
0.10 |
0.18 |
|
EI FI Tjk |
-0.90* |
0.48 |
-0.51 |
0.34 |
1.00** |
0.45 |
-0.20 |
0.29 |
|
EI MI GAik |
1.48** |
0.62 |
1 59*** |
0.41 |
-1.17** |
0.60 |
0.97* |
0.47 |
|
EI MI GCik |
-0.02 |
0.34 |
-0.03 |
0.35 |
-2.27*** |
0.41 |
0.06 |
0.34 |
|
EI DI FSGik |
-0.65* |
0.40 |
-0.38 |
0.43 |
-1.43*** |
0.45 |
1 50*** |
0.39 |
|
EI DI IIik |
0.15 |
0.26 |
0.37 |
0.24 |
0.21 |
0.40 |
-1.20** |
0.39 |
|
PS ICik |
-0.41 |
0.28 |
0.13 |
0.26 |
-1.51*** |
0.38 |
0.83** |
0.37 |
|
PS SPik |
0.70* |
0.37 |
0.16 |
0.43 |
2.35*** |
0.30 |
-0.44* |
0.24 |
|
RI CSik |
0.45*** |
0.17 |
0.63** |
0.19 |
0.54** |
0.27 |
-0.41 |
0.37 |
|
RI OSik |
0.28 |
0.26 |
-0.09 |
0.31 |
0.16 |
0.31 |
-0.01 |
0.27 |
|
RI MRik |
0.52 |
0.36 |
0.21 |
0.32 |
0.77** |
0.37 |
0.28 |
0.27 |
|
Control variables |
||||||||
|
c TECik |
0.64 |
0.22 |
0.75** |
0.36 |
0.95 |
0.74 |
-1.97*** |
0.65 |
|
c CIik |
-0.09 |
0.67 |
1.94** |
0.94 |
-6.01*** |
2.24 |
4.37*** |
1.56 |
|
c LIRik |
-0.67*** |
0.19 |
-0.66*** |
0.19 |
-0.54 |
0.75 |
-0.57** |
0.29 |
|
c SPik |
1.84*** |
0.56 |
1.58*** |
0.62 |
-0.24 |
0.59 |
0.61 |
0.42 |
|
c GDPik |
-0.02 |
0.08 |
-0.06 |
0.08 |
-0.82 |
0.70 |
2.16*** |
0.58 |
|
cons |
-7.12** |
3.29 |
-6.74** |
3.47 |
21.59 |
17.18 |
-53.24*** |
13.25 |
|
Observations |
330 |
319 |
176 |
220 |
||||
|
R2 |
0.38 |
0.40 |
0.49 |
0.38 |
||||
|
Wald |
258.81*** |
208.87*** |
161.69*** |
169.40*** |
||||
|
Notes: The Wald test has a Chi2 distribution and tests the null hypothesis of non-significance of all coefficients of independent variables; panel corrected standard errors are reported. ***, **, *, denote significance at 1, 5 and 10% significance levels, respectively; Estimates include country and time dummies (Marques and Fuinhas, 2012a). xtpcse command was used. |
||||||||
Table 10 — Panel-corrected Standard Errors (PCSE) Regression Results
We estimated all models separately using the PCSE and the OLS estimator as well as REE for robustness checks. We conducted the analysis for Multiple RE data and distinct sectors. The estimation results are displayed in order of the categories and different policy structures. We report the results with models based on a time lag of 1 year, thus, investments lagging behind the introduction of policy by 1 period. With this analysis we go beyond extant work (Aguirre & Ibikunle, 2014; Johnstone et al., 2010a; Marques & Fuinhas, 2012a), ruling out reverse causality (i.e. investments driving policies for example through lobbying) and providing a more realistic approach to renewable deployment, taking into account a lagging reaction of investors to policy measures. The results of our complete policy variable analysis are presented in Table 10 and in Table 11. An overview about our results can be drawn from Table 12.
|
Independent variables
|
OLS |
|||||||
|
Standard errors |
Standard errors Standard errors |
Standard errors |
||||||
|
Multiple RE |
Wind |
Solar |
Biomass |
|||||
|
(V) |
(VI) |
(VII) |
(VIII) |
|||||
|
Coeff. |
S.E. |
Coeff. |
S.E. |
Coeff. |
S.E. |
Coeff. |
S.E. |
|
|
EI FI FIik |
0.69** |
0.30 |
0 70*** |
0.26 |
1.18*** |
0.31 |
0.09 |
0.26 |
|
EI FI GSik |
0.33 |
0.26 |
0.13 |
0.25 |
0.61** |
0.25 |
0.61** |
0.25 |
|
EI FI Ljk |
-0.36 |
0.44 |
0.30 |
0.40 |
-0.91** |
0.43 |
-0.15 |
0.39 |
|
EI FI TRjt |
0.40 |
0.33 |
0.20 |
0.29 |
-0.01 |
0.34 |
0.10 |
0.27 |
|
EI FI Tjt |
-0.90* |
0.49 |
-0.51 |
0.44 |
1.00* |
0.58 |
-0.20 |
0.39 |
|
EI MI GAit |
1.48** |
0.67 |
1.59*** |
0.61 |
-1.17 |
0.80 |
0.97 |
0.53 |
|
EI MI GCit |
-0.02 |
0.44 |
-0.03 |
0.41 |
-2.27*** |
0.48 |
0.06 |
0.37 |
|
EI DI FSGit |
-0.65 |
0.52 |
-0.38 |
0.47 |
-1.43*** |
0.47 |
1.51*** |
0.52 |
|
EI DI IIit |
0.15 |
0.44 |
0.37 |
0.37 |
0.21 |
0.56 |
-1.20*** |
0.46 |
|
PS ICjt |
-0.41 |
0.41 |
0.13 |
0.34 |
-1.51 |
0.43 |
0.83** |
0.36 |
|
PS SPjt |
0.70** |
0.35 |
0.16 |
0.36 |
2.35*** |
0.34 |
-0.44 |
0.30 |
|
RI CSjt |
0.45 |
0.33 |
0.63** |
0.32 |
0.54 |
0.36 |
-0.41 |
0.33 |
|
RI OSjt |
0.28 |
0.39 |
-0.09 |
0.35 |
0.16 |
0.41 |
-0.01 |
0.34 |
|
RI MRjt |
0.52 |
0.37 |
0.21 |
0.34 |
0.77* |
0.47 |
0.28 |
0.31 |
|
Control variables |
||||||||
|
c TECjt |
0.64*** |
0.19 |
0.77*** |
0.20 |
0.95 |
0.72 |
-1.97*** |
0.52 |
|
c CIjt |
-0.09 |
0.96 |
0.96 |
1.02 |
-6.01** |
2.58 |
4.37*** |
1.60 |
|
c LIRjt |
-0.67*** |
0.24 |
-0 74*** |
0.24 |
-0.54 |
0.67 |
-0.57* |
0.30 |
|
c SPjt |
1.84*** |
0.33 |
2.10*** |
0.35 |
-0.24 |
0.46 |
0.61* |
0.37 |
|
c GDPjt |
-0.02 |
0.10 |
-0.05 |
0.09 |
-0.82 |
0.70 |
2.16*** |
0.50 |
|
cons |
-7.12*** |
2.65 |
-8.33*** |
2.59 |
21.59 |
17.01 |
-53.24*** |
11.51 |
|
Observations |
330 |
319 |
176 |
220 |
||||
|
R2 |
0.38 |
0.40 |
0.49 |
0.38 |
||||
|
F |
10.05*** |
10.93*** |
8.01*** |
6.52*** |
||||
|
Mean VIF |
2.03 |
1.94 |
9.17 |
5.15 |
||||
|
Notes: The F-test is normally distributed N(0,1) and tests the null hypothesis of non-significance of the coefficient estimates taken together. ***, **, *, denote significance at 1, 5 and 10% significance levels, Estimations include both country and time dummies (Marques et al., 2010). regress command was used. |
||||||||
Table 11 — Ordinary Least Square (OLS) Regression Results
