A MULTI-MODEL APPROACH TO ASSESS THE RELATIVE WEIGHTS AND SENSITIVITIES OF THE FACTORS OF REGIONAL COMPETITIVENESS
A MULTI-MODEL APPROACH TO ASSESS THE RELATIVE WEIGHTS AND SENSITIVITIES OF THE FACTORS OF REGIONAL COMPETITIVENESS

Summary/Abstract: In order to prioritize the intervention to augment regional competitiveness, it is essential to assess the relative weights and sensitivities related to the factors of competitiveness. The improper assignment of relative weights is prominent in the case when multi-co-linearity exists among independent variables. The paper tests the suitability of multiple models for their capacity of assessing relative weights, and subsequently for forming a competitiveness index. The relative weights of critical components of economic infrastructure have been assessed with Zero-order correlation, Structure coefficient analysis, Beta coefficient analysis, Product measure analysis, Relative weight analysis, and Commonality analysis. Subsequently, regional competitiveness indices have been formed with relative weights as a linear combination. The most suitable technique to form an index has been identified through the Pearson correlation and Spearman rank correlation analyses. The multiple regression analysis assigns the relative weights and consecutively forms the regional competitiveness index, better than other applied techniques. Zero-order correlation and Structural coefficient analysis performed reasonably well. Commonality analysis is a very appropriate technique for the detailed investigation of unique and shared effects among variables. The result shows that the common effects of the critical components of the economic infrastructure are stronger than their unique effects. The sensitivity of competitiveness related to the variables has been assessed through Artificial Neural Network. Regional competitiveness is most sensitive to the variable of rural roads. The results indicate that better connectivity triggers capital and labor drain from the rural areas of the region.

• Details
• Contents