On the Joint Test of the Uncovered Interest Parity and the Ex-ante Purchasing Power Parity

Authors


  • The author is grateful to Fabio Bagliano and Claudio Morana for constructive comments. He is also thankful to Andrea Beltratti, Carlo Favero, Agostino Consolo, Giulio Nicoletti and Philippe de Rougemont for further discussion. The paper also benefited from comments provided by participants at the Rimini Time Series Workshop, together with participants at the GdRe Symposium on Money, Banking and Finance and participants at an internal seminar organized by the DG Economics of the European Central Bank. The author also acknowledges comments from two anonymous referees.

Abstract

This study revisits the relation between the uncovered interest parity (UIP), the ex-ante purchasing power parity (EXPPP) and the real interest parity (RIP) for the UK and Japanese vs US data. The original contribution is on developing some joint coefficient-based tests, obtained by rewriting the UIP, the EXPPP and the RIP as a set of cross-equation restrictions in a vector autoregression (VAR) framework. Test results point to a “forward premium” bias in both the UIP and the EXPPP. The latter result is novel in the literature and stems from testing the PPP in expectational terms. Moreover, the results suggest a currency-dependent pattern for the UIP, contrarily to the EXPPP equation. Finally, it is shown that conditioning the VAR on M3 growth differential has important explanatory power in resolving the aforementioned biases in both the UIP and EXPPP equations for the UK vs US data. At the same time, variables having a strong forward-looking component (i.e. share prices) help recover a unitary coefficient in the UIP equation.

Ancillary