Constructing an alternative dollar index to gauge the movements in currency markets

Abstract

The methodology of building up a dollar index capturing the appropriate weights of different currencies of trading partners has been challenged. This paper tries to suggest a methodological improvement. The authors employ multivariate statistical techniques to identify appropriate financial and economic weights based on variables that are more relevant and attempts to use those variables to suggest an alternative dollar index that would gauge more accurately the movements in the currency market. Cronbach's (1951) alpha is also used to estimate the internal consistency of the proposed index's components. Our proposed index is based on 12 indicators. We calculate the average inter-item correlation, and alpha reliability coefficient. By calculating the inter-item correlations among the 12 component variables, we study the variables in the expected direction, and statistically significant relationships at the 0.05 level. We then remove the variable which has statistically insignificant relationship. The process works as follows: first, we aggregate the 12 objective indicators to construct a dollar index with equal weight, then, for constructing the dollar index we choose the appropriate weights by allowing them to be determined by the statistical procedure of principal components to determine the weights for an index of the variables.