How does ppp hold between australia and its trading partners.pdf

How does ppp hold between australia and its trading partners

THE AUSTRALIAN NATIONAL UNIVERSITYSchool of Finance and Applied Statistics Research Project in International Finance

_

DANG H PHAM U4278723

HOW DOES PPP HOLD BETWEEN AUSTRALIA AND ITS TRADING

October the 26th, 2007

I INTRODUCTION

The interrelations between three major components of the international market (international trade, foreign exchange and capital market) are explained by the economic theories called the international parity conditions It is considered to be unique to the field of International Finance (Eiteman, Stonehill and Moffett 2006:102) Deriving from the Law of one price, the absolute form of purchasing power parity (PPP) states that if the market is frictionless the price of identical products and services should be the same in different markets For instance, if the price of the product in Australian dollar is PAUD, then the price of the identical product in U.S dollar (PUSD) should be equal to PAUD adjusted for exchange rate (SUSD/AUD), or PAUD x SUSD/AUD = PUSD Thus, the nominal exchange rate can be simply expressed as the ratio of the two prices (SUSD/AUD = PUSD / PAUD) If this is not the case, the arbitrageurs in an efficient market could make riskless profit by shipping the goods from locations where the price is low to locations where the price is high (Taylor and Taylor, 2004) They will trade the opportunity away and bring the conditions back to “parity” If the PPP holds perfectly the nominal exchange rate adjusts to maintain a constant real exchange rate equal to 1 The relative form of PPP involves only that the change in nominal exchange rate offsets the differential between home and foreign price level

While international trade plays a vital role in a PPP , obviously the market frictions such as transportation cost, taxes, tariffs and duties, and non-tariff barriers such as quotas and import licenses are the main causes of some deviations from PPP In recent years we have seen the emergence of multinational and bilateral free trade arrangements where governments set the objectives to bust up trades by removing trade barriers Australia has joined WTO and signed free trade agreements with the United States and Thailand in 2005, with Singapore in 2003 and with New Zealand from 1993 It is now in negotiating Free Trade Agreements with other countries and areas such as Japan, ASEAN, Malaysia, China, Chile and Gulf countries This research project tests the PPP between Australia and its trading partners over the period from 1984 to 2006 and provides some insights on particular effects that the level of bilateral trade and investment have on their PPP Using quarterly of foreign exchange rates and proxy for country price level (both consumer price index and producer/whole sales price index) evidence from my sample shows that PPP holds for

some countries while it does not hold for others Further, the higher level of bilateral trade significantly and positively associates with the higher level of PPP Foreign investment level is found to have positive impacts on PPP but not significantly This research is outlined as follows: Part II reviews PPP literature, in which the subsection 1 on PPP theoretical background highlights the law of one price and the role of international trade as crucial conditions for the existence of the parity condition An overall picture about extant empirical literatures on PPP including Australian evidences is provided in subsection 2 and 3 From these reviews we develop our arguments about the impacts of trade and investment on the PPP s between Australia and its trading partners Part III presents the hypothesis our research and the methodologies, models we use to test the hypothesis in details Part IV describes and discuss about the data used in our research The results of this research are reported and discussed in Part V Conclusion is provided in Part VI; and Part VII proposes some direction for future research

II LITERATURE REVIEW 1 Theoretical background

Economists widely believe that the PPP theory hold at least approximately because of the possibility of the international goods market arbitrage In the formal way, the PPP theory states that the percentage change in the exchange rates over a given period simply offset the difference in inflation rates in the countries concerned over the same

period Therefore, in the relative form of PPP, according to Eiteman et al (2006:105),

“if a country experiences a higher rate of inflation, and its exchange rate does not

change, then its goods prices will be relatively more expensive” Eiteman et al

(2006:105) also states that “if the spot exchange rate between two countries starts in equilibrium, any change in differential rate of inflation between them tends to be offset over the long run by an equal but opposite change in the exchange rate” Hence, the relative PPP is useful for forecasting the spot exchange (long run) rates from the expected inflation rates in the two countries The model is described as follows:

= (1.1) or ΔSh/f =πh −πf (1.2)

Where: is the expected spot exchange rate at time t, is the current spot exchange rate,

)(Sth/ f

While the traditional PPP theory holds that the real exchange rate should be constant and equal to 1, there is another assumption of long-run deviations from PPP Harrod (1933), Balassa (1964) and Samuelson (1964) build their model of the equilibrium exchange rate on the observation that rich countries tend to have higher price levels than poor countries This is because rich countries are relatively more productive in the traded high-tech goods and they grow richer as a result Income rises as a result of

the higher productivity in tradable goods will cause price level to rise also This effect modeled by Harrod, Balassa and Samuelson seem to be stronger over time where data show a strong positive relationship between income and price level If this is true, then the real exchange rate is not constant over time

In general, as PPP theory is derived from the law of one price, its basic condition is the perfect international trade of goods and services If there is no trade, obviously no PPP exists The perfect trading market (without friction and transaction costs) is the key assumption in PPP theory In this research, an examination on PPP relation between Australia and its different trading partners to find out the consistency and inconsistency with the above-mentioned assumption is therefore necessary

2 Empirical evidences about PPP

There has been a heightened level of extant literature on the validity of PPP The results of the PPP tests varies according to according to whether (i) price and exchange rate levels (absolute PPP) or changes in prices and exchange rates (relative PPP) are studied; (ii) individual commodity prices or nominal price levels are employed; (iii) purely traded goods’ prices or non-traded as well as traded goods prices are considered; (iv) bilateral or multilateral approach is adopted; and (v) the short run or the long run is investigated (Rush and Husted, 1985) In the respect of long-run or short-run validity of PPP, for example, evidence from Edler and Lehmann (1983), Enders (1988) and Ardeni and Lubian (1991) reject long-run PPP However, Diebold, Husted and Rush (1991), Cheung and Lai (1993), Edison (1987), Frenkel (1981), Branson (1981), Desai (1981), and Miller (1986) strongly support that the PPP holds in the long run and poorly in the short-run Cooper (1994) claims that the main reasons for the inconclusive results are due to the difference in particular currencies under consideration, the price indices used to measure price levels of inflation, and the method of analysis employed

Early empirical tests (mainly in the 1970s) find support for the long-run and continuous PPP (Frankel, 1976 and1981) This might be partly due to the data of a period of stable exchange rates (few years post-float after 1971- collapse of the Bretton Woods Agreement) On the other hand, PPP deviates significantly from equilibrium in the short-run PPP due to the sticky nominal prices (Dornbusch, 1976) However, there is no evidence of how far do exchange rates deviate from the mean and how fast they revert to the mean, knowing that mean reversion of exchange rates

toward its mean is the key characteristic and necessary condition for long-run PPP to hold (Taylor and Taylor, 2004) Studies employing unit root test of real exchange rate (the null hypothesis is that real exchange rate follow a random walk) could not reject the random walk (see, for example, Roll 1979);

Contemporary empirical studies of PPP are dominated by unit root tests with more advanced techniques Frankel ( 1986) tests the first order autocorrelation model for the real exchange rates: (e1 − e~) = δ(et−1 − e~) +εt where e~ is the assumed

constant equilibrium real exchange rate and δ is the autocorrelation coefficient If

δ >1 it is an explosive process; if δ =1, then exchange rates follow a random walk; if

δ <1, then there is mean reversion The evidence from this test rejects the hypothesis of random walk at 5% significant level Abuaf and Jorion (1990) estimate a system of univariate autoregressions and find in general that PPP hold in the long-run Others studies using longer data sets and more sophisticated models also reject the null hypothesis of random walk (see, for example, Glenn, 1992; Dielbold, Husted &Rush,1991; and Cheung and Lai, 1994) As evidence from many studies have shown that exchange rates have mean reversion characteristics, it is desirable to know how fast the exchange rates revert to the mean or in other words to what extent of time horizon PPP holds? Reviews of Rogoff (1996) reveal that studies seem to agree that the deviation from PPP real exchange rate has half-life of 3 to 5 years

While some recent studies use non-linear dynamic models that allow for the autoregressive coefficients to vary find supportive evidence of non-linear long-run PPP For example, Taylor, Peel and Sarno (2001) find support for non-linear mean-reverting real exchange rates They also find that ‘modest’ deviation of less than 5% have half-life of less than 3 years while deviations greater than 5% have even shorter half-life Sarno and Valente (2006) test long-run PPP using complex non-linear model Their results indicate that long-run PPP does appear to hold Further, PPP deviations revert more quickly under floating exchange regimes In contrast, other latest studies on the validity of PPP even using advanced techniques still have different results For example, Alba and Papell (2007) examines the long-run PPP using panel data methods to test for unit roots in the USD real exchange rates of 84 countries finds that PPP holds for European and Latin American countries but not for Asian and African countries Arghyrou and Gregoriou (2007) tests for long-run PPP

using more sophisticated unit root test on G7 countries over the last 30 years They find that PPP does not hold in the long-run

In general, PPP test appear to be a puzzle The extremely high short-term volatility of exchange rates hardly reconcile with the long-run PPP and the half-life of PPP reversion of 3 to 5 years is too long to be caused by short-run sticky nominal prices (Rogoff, 1996) states, Perhaps the international goods market is not as integrated as we think, with large trading frictions creating a ‘band of inaction’ that permits wild fluctuation in nominal exchange rate with no effect on price levels (Rogoff, 1996; Taylor and Taylor, 2004) Alba and Papell (2007) share the same view when they find evidence that PPP is stronger in countries that have higher levels of trade, closer to the US (physical distance), and countries characteristics affect PPP adherence Based on these finding, when testing PPP between Australia and its trading partners, this

paper hypothesizes that PPP hold better between Australia and its major trading

partners

3 Australian evidence

In the Australia context, empirical evidence of PPP is also inclusive For example, Corbae and Oularis (1991) claims that PPP does not hold in long run for the Australian dollar, as the data follow a random walk Conversely, Olekalns and Wilkins (1998), when re-examining the data used by Corbae and Oularis (1991), find different results that the long-run PPP holds for Australian dollar The research conducted by Bhati and McCrae (2004) supports the long-run PPP when the Australian dollar is used as the based currency in relation with the Asia-Pacific trading partners of Australia The exceptions include Singapore, Indonesia and New

Zealand with poor evidence of PPP Koedijk, Schotman and Vandijk (1998: p.60)

point out that “there is substantive evidence that PPP holds for many currencies, although not for every currency to the same extent”

The only one thing that is clear from the evidences for Australia is that PPP varies across its trading partners It is our conjecture that the level of PPP depends on the degree of trade relation between the host country and the foreign country Hence, it is deserved a comprehensive investigation in this research This research project tests PPP using the Australian dollar as a base currency with other 10 currencies selected from countries classification by the Reserve Bank of Australia (RBA) based on trade weights (described in Appendix 1) The test aims to find evidence which shows PPP

holding better with higher trade weighted countries Some abnormal cases rejected by the test will be further discussed for appropriate reasons

III METHODOLOGY 1 Test of PPP

The main objective of this study is to provide an overall understanding of how PPP holds between Australia and its trading partners We select the representing countries categorize them into groups which have different trading levels with Australia Test of PPP is conducted for each country/currency and compare to each others and between different groups

We test the PPP using relative form Based on the equation (1.2) we construct the regression model as follows:

H0: PPP holds or α= 0 and β = 1

HA: PPP does not hold or αis different from 0 and β is different from 1

We use F-statistics to test the null that α= 0 and β = 1 jointly (Wooldridge 2006:150-160) F-statistics are calculated using the error terms of restricted model and error terms of unrestricted model as follows:

The unrestricted model is the model (2.1)

Under the null hypothesis model of (2.1) becomes: this is called the restricted model

επ

q is the number of variables restricted

If F-statistic > critical value then we reject the null hypothesis to accept the alternative Otherwise, we cannot reject the null hypothesis and conclude PPP holds For the purpose of ranking the level of PPP between different trading partners, we also look at p value and R-squared as supplementing measures

2 Effects of trade and investment on PPP

Since the actual relationship between exchange rates and price levels often deviate from the parity and tend to revert from time to tine Reasonably, the more exchange rates deviate from the theoretical PPP the less PPP holds for that currency We measure the deviation from the equation of relative form of PPP as follows:

S −/1 Sth/f

Effects of trade and investments on PPP is estimated from the regression of PPP deviations against annual trade weights and foreign direct investment level in the following model:

PPPdev = β0 + β1 + β2 +ε (2.3)

The sign and the significance of β1and β2estimated from (2.3) indicate the effect of trade and foreign investment on PPP

IV DATA

Among Australia trading partners, we select 12 countries/currencies that their trading weights range from the high to low across the sample period from 1984 – 2005 (See Table 1) for our sample

Table 1 List of currencies of Australia’s trading partner countries selected for PPP test The trade weights represent the level of trading between that country and Australia

(According to trade weights table announced annually by RBA)

The quarterly nominal exchange rates of Australian dollar to the 12 currencies for the period from 1984 to 2006 are collected from the Website of the Reserve Bank of Australia We obtain quarterly Consumer Price Index (CPI) and Producer Price Index (PPI)/Wholesale Price Index (WPI)1 (from hereon PPI and CPI are referred to as PPI for convenience) of the 12 countries for the same period from Datastream-economics database

S /

Δ is calculated from the quarterly nominal exchange rates between the Australian dollar and the foreign currency The selection of quarterly data deriving from the needs that time interval should be long enough for the differences in countries’ inflation rates to take effect on the corresponding exchange rates Changes in nominal exchange rates are estimated from the following formula:

Inflation rates

1

WPI is used instead of PPI for countries that PPI is not available (for example Australia)