A machine learning approach of exploring nexus between geopolitical risk and Indian economic uncertainty: a comparative review

The purpose of the study is to explore the causal connection between geopolitical risk and the Indian Uncertainty Index, with a focus on comparing the old and new uncertainty index. The study aims to evaluate the impact of different elements of geopolitical risks on the uncertainty index by employing advanced and optimized regression models using MATLAB. The relevant data related to GPR indices on monthly basis are extracted from https://www.matteoiacoviello.com. So far as dependent variables are concerned, two kinds of Indian uncertainty index are considered and the data extracted from https://www.policyuncertainty.com/india_monthly.html. The first one is Old Indian EPU Index (Old) in which Consensus Economics forecaster dispersion data were used. The data available for Old Indian EPU Index on monthly basis, ranges from January 2003 to March 2014 i.e. 135 data points. The second one is India News-Based Policy Uncertainty Index (New). The data available for India News-Based Policy Uncertainty Index (New) on monthly basis, ranges from January 2003 to present (May 2024) i.e. 257 data points. By comparing the both formulated models for each kind of uncertainty index, it seems that although both the models got less r-squared (0.3407 in case of model with old uncertainty index as response variable and 0.16 in case of model with new uncertainty index as response variable), but the model formed for Old Indian Economic Policy Uncertainty Index (Old) seems minutely better. In other words, the GPR indices were more capable of explaining the old uncertainty index data as compared to new uncertainty index data.