Identification of exchange rate shocks with compositional data and written press

Abstract

The evolution of exchange rates results from events that affect different countries differently, that is, asymmetric shocks. Being value ratios, exchange rates constitute what is known as compositional data. The compositional data analysis methodology transforms exchange rates in a way that ensures the validity of their subsequent statistical analysis. In this paper we submit the daily exchange rates of the US dollar, the yen, the pound sterling, the euro, the Brazilian real, and the yuan to transformation by centred log-ratios. We then use the residuals from a Box-Jenkins analysis to estimate shocks and represent them in statistical control charts for a simple visual identification of both significant revaluations and devaluations, and periods of greatest volatility, which we then relate to news stories in the written press.

Keywords

• compositional data
• control charts
• exchange rates
• asymmetric shocks
• written press