A comment on ‘Testing the validity of quasi-PPP hypothesis: evidence from a recent panel unit-root test with structural breaks’

Abstract

In a recent study, Guloglu et al. (2011) took 18 Turkish real exchange rate series and used a panel unit-root test, which allows for structural breaks in the level and the trend of the real exchange rate. When the practitioner allows for structural breaks, the rejection of the null hypothesis of unit root in the real exchange rate does not imply that Purchasing Power Parity (PPP) holds. The authors claim that they found evidence of Quasi-PPP (Q-PPP). Yet, the latter does not include either linear trend or trend breaks. Their results cannot be considered evidence of either Trend-PPP (T-PPP) or Trend-Qualified PPP (TQ-PPP), since both exclude trend shifts.

Keywords:

• PPP
• Q-PPP
• T-PPP
• TQ-PPP

JEL Classification:

• C22
• F40

Notes

¹ Abuaf and Jorion (1990) used 100 years of dollar–franc–sterling data. Lothian (1990) used slightly over 100 years of yen–dollar–sterling–franc data. Lothian and Taylor (1996) used 200 years of dollar–franc–sterling data. Frankel and Rose (1996) used a panel of 150 countries and 45 annual observations. Taylor (2002) used real exchange rate data for over 100 years for 20 countries. All of these studies found evidence in favour of PPP. That said, and quoting Taylor and Sarno (1998, p. 308), ‘panel unit root tests may be quite powerful’. Therefore, ‘the rejection of the null hypothesis of joint non-stationarity of a group of real exchange rates may be due to as few as one of the exchange rates series under investigation being generated by a stationarity process’. They suggested a complementary test where the null hypothesis is ‘that at least one of the series is generated by a non-stationary process. This null hypothesis will only be violated if all of the series in question are realizations of stationary processes’ (Taylor and Sarno, 1998, p. 308). In doing so, evidence of PPP is mixed; their results indicate that while Consumer Price Index (CPI)-adjusted real exchange rates among the G5 countries (the United States, the United Kingdom, Germany, France and Japan) are found to be mean reverting over the floating rate period, only a subsample of real exchange rate series, constructed using Gross Domestic Product (GDP) deflators or producer prices indices, are found to be mean reverting.

² Low power implies that the procedure may fail to reject the null hypothesis when it is false.

³ Taylor and Taylor (2004) provide an excellent literature review on the subject.

⁴ Among others, Papell and Prodan (2006) have referred to these by the acronyms used in Table 1.