Simulated trial and error experiments on productivity

Karn Thamprasert, Ahmad Yahya Dawod, Nopasit Chakpitak

Abstract


Trial and error experiments in socioeconomics were proved to be beneficial by Nobel prize laureates. However, replication is challenging and costly in term of time and money. The approach required interventions on human society, and moral issues have to be carefully considered in research designs. This work tried to make the approach more feasible by developing virtual economic environment to allow simulated trial and error experiments to take place. This research demonstrated the framework using 19 macroeconomic indicators in 6 interested categories to study the effect on productivity if each indicator value grew by 5 percent for each of 65 countries. Seven predictive models including some machine learning (ML) models were compared. Neural network dominated in accurateness and was selected as the core of the simulator. Experimented results are in full of surprises, and the framework acted as expected to be a data-driven guide toward country-specific policy making.

Keywords


data-driven economic policy; machine learning; productivity; simulated trial and error; virtual economic environment;



DOI: http://doi.org/10.11591/ijai.v11.i4.pp%25p

Refbacks

  • There are currently no refbacks.


View IJAI Stats

Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.