Staff Papers
Published Date: 01 May 1998

What Lies Behind Singapore's Real Exchange Rate? An Empirical Analysis of the Purchasing Power Parity Hypothesis

MAS Occasional Paper No. 6 May 1998 - By Domestic Economy Division, Economics Department 

Executive Summary

1          This paper provides an empirical examination of the behaviour of exchange rates in Singapore. The analysis is in two main parts.

2          First, Section 2 of the paper reviews the alternative measures of the real effective exchange rate (REER). Three measures of the REER are evaluated, based on different deflators, namely relative consumer price indices (CPI's), wholesale price indices (WPI's) and unit labour costs (ULC's).

3          Conceptually, it can be shown that the REERCPI is the most general measure of the three. However, it is this general nature of the CPI which makes the REERCPI unsuitable for making inferences about international competitiveness. Among the three indicators, the REERULC is a better measure of competitiveness because, as a cost-based measure, it is a purer proxy of long-run prices as it is less distorted by changes in the exchange rate and profits. The REERULC is especially relevant for Singapore because remuneration costs comprise about 42% of production costs (excluding materials cost) in the manufacturing sector . However, we recognise that there is no single, comprehensive indicator of international competitiveness. The alternative versions of the REER, as well as other micro data that describe industry-specific cost conditions, should be studied as well.

4          The second part of the paper provides a careful empirical study of the time series properties of the alternative variants of the REERs. In Section 3, we subject the series to various non-stationary tests using data over the period 1980-1996. Our results establish that the REER series deflated by the CPI and ULC show persistent volatility and do not return to a constant time path within a finite horizon. On the other hand, there is evidence of mean reversion behaviour when the WPI series is used as the deflator over the period 1988 to 1996.

5          It has been difficult to provide an economic or structural interpretation of the persistent deviations of REERULC and REERCPI from a constant time path, because the relationships estimated in this study have been primarily of a reduced-form nature. The attempt to relate the movements of the real exchange rate to its underlying economic determinants will be the focus of further research.