Purchasing power parity: further evidence and implications

Abstract

This study tests the validity of the purchasing power parity (PPP) theory using a panel of ten Arab countries. It also measures the speed of convergence for the panel and for the Gulf Cooperation Council (GCC), including Bahrain, Kuwait, Qatar, Saudi Arabia and United Arab Emirates which follow a hard dollar peg. The test is conducted using nine base currencies. The main findings are first on the validity of PPP. (1) The null of unit root is rejected using each of the nine currencies as a numeraire—the result appears to be invariant to the choice of the base currency. (2) The result appears to be stronger when the estimation is based on heterogeneous rather than homogeneous serial correlations among panel units. Secondly, on persistence. (3) The estimated average half-life of a shock to the real exchange rate in the full panel is 2.19 years. (4) The study reveals that for the GCC economies deviation from PPP will be more persistent due to the hard peg, the dollar volatility and the weak goods market arbitrage. Clearly, the GCC countries will need to move to a more stable peg to be able to achieve the benefits they hope by engaging in a monetary union by 2010.

Keywords:

• purchasing power parity
• real exchange rate
• panel unit root test
• Arab world
• JEL Classification: F31

Acknowledgments

The author would like to thank the editor and two anonymous referees for comments and suggestions, which have improved the paper. Any remaining errors are the author’s.

Notes

1.	There have been some studies that found that shocks do not die quickly, in this case, we have what is called mean reverting with long memory. That is fractional integration in the real exchange rate see Cheung and Lai (2000).
2.	There have been some studies related to the Arab world using co-integration and univariate unit root tests, e.g. Bahmani-Oskooee (1998), Salehizadeh and Taylor (1999) and Soofi (1998).
3.	The choice of the countries is largely dictated by data availability. For example, CPI data are missing for the UAE, Oman and Lebanon. Data are also missing for Yemen and Libya.
4.	There is a debate in the literature on the choice of the appropriate price series to be used, see McNown and Wallace (1989) who wrote on the sensitivity of the choice of price index.
5.	A black market exchange rate occurs when governments try to restrict capital flow by imposing various types of restrictions on the purchase of foreign currencies.
6.	Start with the highest possible lag order and choose lag length based on the highest significant lag order or, if no lag is significant, choose lag = 0; highest lag tried is 4.
7.	All the real exchange rate series were demeaned first by subtracting the cross-sectional mean from the original series before applying the ADF test for each individual. The highest average off diagonal elements of the sample correlations matrix of the panel-demeaned real exchange rates series is 0.22.
8.	Coakley and Fuertes (2000) did not find empirical evidence for the base currency effect. O’Connell (1998) argued that the sets of real exchange rates generated by different choices of numeraire are linear combinations of one another. Thus, changing the numeraire does not change the information that is used in the estimator, only its configuration, i.e. its interdependence.
9.	Papell (1997) has shown that serial correlation has an impact on the size of panel unit root tests.
10.	The major trading partners of the Arab world are the European Union first (with, Britain and Germany having top ranks). The USA and Japan alternate in the second position. Interregional Arab trade, however, is low (see Arab Monetary Fund annual report, various issues).
11.	The fact that all exchange rates in the panel are directly related to the dollar may help to explain the evidence for PPP even with the dollar as a base currency (other studies found weaker evidence when the dollar is used as a base currency).
12.	The study by O’Connell did not find evidence for PPP for the world as a whole and sub-samples of Europe, Asia, South America and Africa using the LL test. As was indicated earlier, the LL test has been criticized for assuming the same speed of reversion across countries. That is, both lag length and serial correlation structure are assumed the same, the IPS test used in this study allows for heterogeneous serial correlations by conducting individual ADF regressions.
13.	This is believed to be caused by the large appreciation and the subsequent depreciation of the dollar in the 1980s; all exchange rates in the panel are directly linked to the dollar.
14.	Openness measured as the ratio of trade to GDP ranges from 47 to 194 for countries in the panel.
15.	The approximate half-life of a shock is computed as –ln(2)/ln(α j ).
16.	, where ψ T = − (1 + α)/(T − 1), ω T = 1 − (1/T) (1 − α T )/(1 − α), and ϕ T = 1 − 2α(1 − ω T )/((1 − α)(T − 1)).
17.	This result may give support to the view that PPP, as a long-run relationship has nothing to do with the exchange rate regime.