Classifying Currencies Using Common Factors

This study proposes a novel approach for classifying currencies on the basis of common driving forces that determine their behavioral features. In contrast with the conventional approach, which focuses on exchange rate systems that regulate variations in currencies, and classifies them according to the anchor currency and system flexibility of the system, we focus on the common factors that govern the dynamic behavior of currencies. We employed a multivariate factor stochastic volatility model to estimate the latent factors and applied K-means and hierarchical clustering to the estimated factor loadings. We observed three distinct clusters that can be labeled the US dollar, euro, and commodity currency blocks. This approach will prove particularly useful for investors and policymakers to evaluate the currency risks that are common within a currency block, but vary between different blocks.