Dear Kaan Kessler ,

I think that the question should be better framed.

I think you want to look at this:

Productivity_it = intercept + b1*Public_R&D_it + b2*Private_R&D_it + b3*Public_R&D_it*Private_R&D_it + ...

b3 gives you how a change in private R&D changes with the level of public R&D. A positive b3 indicates that private R&D has a larger effect on productivity if public R&D is higher.

If you want to just purely look at the effects of public R&D to the amount of private R&D, I would estimate:

Private_R&D_it = intercept + b1*Public_R&D_it + ...

I am almost sure you want to look at the first case.

I hope it helps.

José-Ignacio

Dear José-Ignacio,

Thank you very much for your help! This helps me a lot and makes everything clearer for me. Since I am not very familiar with this topic I apologize for the framework of my question. Regarding the first case you presented to me I would have couple of last questions: Would you recommend me to use cross-sectional data or rather panel data? Right now I am looking and I find only percentages of GDP as data for private and public R&D.

And in the next step I dont really know what tests or operations I should make once I created the regression. If my variables are significant that's good but would you test for causality or other things?

I am sorry for all these "naiv" questions. I just started my research and I am usually not from this field but it seems very interesting to me.

Once again, thank you very much!!

Kaan Kessler

If possible, panel data with fixed effects.

Causality depends on research design. Using fixed effects you control for time-invariante heterogeneity, but there is no silver bullet.

In the first specification, although R&D (private or public) might be endogenous, it is unlikely you might find instruments, so forget about it. Nevertheless, check the literature on the topic on the sort of model used for it (e.g., lags, etc.). In principle, you might proceed with share of GDP, but obviously your approach will be limited to countries...

I hope it helps.

Thank you so much for your insights on this topic. You really helped me a lot! Have a great week.

Dear Kaan,

You can examine the R&D effect on economic growth. You should note that economic growth and development are different concepts (You can see the book chapter in my profile for detail. "Seçilmiş OPEC ve MENA Ülkelerinde Doğal Kaynak Bağımlılığı ve Kurumsal Gelişmenin Ekonomik Kalkınma Üzerine Etkisi: Panel Kantil Regresyon Yaklaşımı") In this case, the ARGE will be an independent variable. GDP or GDP per capita can be taken for growth.