Dear all,
I have a question of how to interpret my results:
I am estimating the effect of a set of determinants on the share of renewables in a countries energy supply (dependent variable).
I have a set of explanatory variable (GDP, commodity prices, energy dependence, public policies supporting renewables etc.). I use the natural logarithm of both dependent and independent variables. I do not take logs of dummy variables (for example Kyoto Protocol). However, the interpretation of the public policies supporting renewables variable is difficult because I constructed the variable following two different approaches:
First approach: Dummy variable (taking on the value 1 if a country has 10 or more policies in the given year)
Second approach: Equal-weighted policy index. If, for instance a policy comes into force in a respective country in 1993, then the policy index will equal to one in 1993, if an additional policy is introduced in 1995, the policy index variable will equal to two in 1995 and so on.
How are results interpreted for the second approach?
Best, Teresa
Hi Teresa,
I'm not sure I'm fully understanding your question.
Have you got any visual element you might be able to share to help you interpret the results? I haven't completely grasped whether you are after support in terms of methodological / mathematical / numerical interpretation of your results.
Best wishes,
Francesco
Hi Franceso,
thanks for your reply and your time.
Here is a simplified version of my regression equation.I have a panel data set and I am runnings OLS, FE and RE models.
CRES=log(GDP)+log(oilprice)+dummy_Kyoto+(energydep)+(publicpolicies)
CRES is the share of renewables in a countries energy supply (in %)
As I mentioned above, I constructed the public policy variable according to the two approaches.
1) Dummy: 1 if country > or = 10 policies
2) Equal weighted policy index: (Continous variable, each policy is attributed the value 1, e.g.: Sweden has in total 39 renewable economic instruments) . Hence receives the value 1 in the year the first policy is introduced, the value 2 when the second policy is implemented and so on
I obtain statistically significant effects for the public policy variable on CRES for both approaches (dummy and equal-weighted policy index). However, I have difficulties to interpret the results for the second approach.
I hope I clarified my question.
Thanks again, Teresa
Thanks for clarifying Teresa.
I would suggest a sensitivity analysis to understand the effect of policies. Every time a policy comes into force you increase the index by one. That will change the results and you can assess which policies or how much time or which set of policies produced a statistically significant impact on the output. What I'm not sure about in your modelling is that as you say all policies are equally accounted for and I'm not particularly sure this is a realistic assumption. Perhaps you need a positive factor (up to you to define intervals, can be 0 to 1 if 1 means the policy has achieved 100% of what set out to do, or 0.5 to 1.5 where 1 would be the expected/average performance of a policy and you can in this way monitor under/overperforming of policies) to bias (in a good sense) that equal value of 1 that you factor in when policies come into force. Hope it makes sense.
Francesco
Hi Teresa,
I couldn't help reading into the discussion between yourself and Francesco. But frankly, I am still unclear about what exactly you imply by "difficulties to interpret the results" - what exactly is your doubt ?
However, I do agree that a fractional factor would be a more realistic way to handle the policies. It may or may not entirely depend on past performance, as Francesco suggests. But it would be better than what you are doing right now.
BTW, any specific reason as to why you chose to weigh the policy indices equally ? Would be interesting to know about your thinking regarding that option, to start with.
With best wishes.
-Sanjay
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Energy
- Energy-Generating Resources
- Renewable Energy