Taxa real de câmbio e poder de paridade de compra: uma perspectiva via regressão quantílica com dados seculares para a relação libra esterlina-dólar americano

Abstract

This paper analyzes the validity of the Purchasing Power Parity (PPP) theory using annual data, from 1800 to 2018, for the real exchange rate (RER) between the pound sterling and the US dollar. The filtered series, to take into account the deterministic trend as predicted by the Balassa-Samuelson effect, is stationary, indicating PPP validity. Significant cubic fit of this series indicates behavior in line with international trade models with transaction costs (“iceberg costs”) and econometric analyzes that incorporate thresholds and non-linearities, such as those of the TAR family. Analysis via quantile regression reveals asymmetry in the temporal adjustment of the filtered series. For example, the empirical distribution conditioned on a highly valued pound is bimodal. Another example is the fact that the probability of the dollar devaluing, conditioned on a highly appreciated dollar, is different from the probability of the pound devaluing conditioned on a highly appreciated pound. From the point of view of adjustment dynamics, the perception of convergence speed for the unconditional distribution turns out to be more optimistic (in relation to PPP) and faster in relation to the perception based on the half-life focused adjustment on the conditional mean, standard adopted in the literature.