|
Article Excerpt
Introduction
According to the law of one price (LOOP), identical goods sold at different geographical locations should sell for the same price when prices are expressed in the same currency. The law is based on the assumption that price differences would provide the opportunity for arbitrage and, in the absence of transaction costs, the activities of arbitrageurs would cause prices to converge. Over the past two decades, a large body of literature has been dedicated to analysing the empirical evidence for the LOOP as well as for purchasing power parity (PPP), which generalizes the LOOP to aggregate price levels. (1) Due to the lack of appropriate econometric techniques and insufficient or unsuitable data, early empirical studies were mostly unable to find evidence for either of the two hypotheses. (2) Later studies which extend the data set by including more observations on cross country data are generally able to find support in favour of price convergence. While these studies typically find half lives of deviations from PPP or the LOOP of 3-5 years, (3) more recent studies that account for non-linear dynamics in the adjustment process report much lower half lives. (4) Regarding the LOOP, intra-country and intra-regional studies also find half lives well below 3 years. (5)
Our analysis contributes to the empirical literature on the LOOP in an intra-regional framework by analysing the development of relative prices on a disaggregated level for the entire European Union (EU) over the past 10 years. (6) The data used for the empirical assessment of the LOOP consist of harmonized consumer price indices for 90 different product groups from 25 EU countries. In addition to analysing the validity of the LOOP for the entire EU, our broad data set also allows us to investigate differences between the 15 old EU countries and the 10 countries of Central and Eastern Europe and the Mediterranean that joined the European Union on May 1, 2004. (7) To our knowledge, this has not yet been dealt with in the literature.
Rogoff (1996) shows that trade barriers, transportation costs and exchange rate volatility can explain why the LOOP might not hold in a large international framework. Concentrating our analysis on an integrated market with restricted geographical expansion allows limiting the impact of these distorting factors. First, the EU is a free trade area, so arbitrage opportunities are not affected by trade barriers. Second, due to the proximity of the EU countries, transportation costs play a less important role on an EU level than on an intercontinental level. Third, the exchange rate mechanism as well as individual exchange rate arrangements of new EU member states lead to stabilization among the nominal exchange rates of almost all EU currencies and, together with the implementation of the single currency in a major part of the EU increased price transparency, reduce information and search costs for potential arbitrageurs. (8)
Additionally, the break down into product groups provides the possibility for a differentiation between tradable and non-tradable products. Since cross-regional arbitrage is more likely to equalize prices of tradable goods than prices of nontradable goods, evidence for the LOOP is more likely to be found in the former case. While we do not pre-categorize the product groups into tradable and non-tradable, we show that the degree of tradability has an impact on price convergence for consumer goods and services among EU countries.
Our analysis is most closely related to the work by Jenkins and Snaith (2005); Jenkins (2004), and Chen (2004). Jenkins and Snaith (2005) employ panel cointegration tests to data on 27 consumer price indices and subindices from six EU countries and the United States. While the authors find support for the LOOP for those goods and services that are highly traded between countries, the evidence is fairly weak for goods and services that are not traded. Jenkins (2004) investigates price convergence for eight consumer price indices across 12 EU countries in a country by country analysis. He finds strong support for mean reversion and reports median half lives between 5 months and 4 years across countries and products. Chen (2004) looks at 17 price indices for Belgium, France, Germany, Italy, Spain, and the Netherlands. While Belgium and the Netherlands display some evidence in favour of the LOOP towards all other countries, the results are less clear-cut for the remaining countries. Regarding half lives, Chen reports values ranging between 4 months and 2 years.
The rest of the paper is structured as follows: Methodology section gives a brief overview of the econometric methodologies used in the empirical assessment of the LOOP. Data Set and Empirical Evidence section first presents the data set employed in our analysis and then reports and discusses the estimation results. Conclusions section concludes with a summary of the main findings.
Methodology
The LOOP in its absolute version states that the price of any good or service should be the same across countries and the relative price should be equal to unity. In its relative version, the LOOP allows transaction costs to drive a constant wedge between prices at different locations and, hence, does not require the relative price to be equal to one but to be mean-reverting.
Mean reversion of relative prices can be tested by considering the following formula:
[q.sub.jt.sup.i] = [s.sub.jt] - [p.sub.ref,t.sup.i] + [p.sub.jt.sup.i], (2)
where [s.sub.jt] is the log of the nominal bilateral exchange rate between country j and the reference country at time t and [p.sub.ref,t.sup.i] and [p.sub.jt.sup.i] are the logs of the prices of good i in the reference country and in country j at time t, respectively. In this setting [q.sub.jt.sup.i] could be seen as a product specific real exchange rate. If [q.sub.jt.sup.i] is stationary this can be interpreted as evidence in favour of the LOOP in its relative version. (9)
In this paper, we employ the panel unit root tests suggested by Levin et al. (2002), Im et al. (2003) and Maddala and Wu (1999) to assess the unit root properties of [q.sub.jt.sup.i]. Panel unit root tests offer the advantage that including a limited amount of cross-sectional information significantly improves the power of the test so that they can be employed to relatively shorter time series. All three tests are based on estimating some version of a standard AR(1) model for a panel, such as:
[q.sub.jt.sup.i] = [[rho].sub.j][q.sub.jt-1.sup.i] + [[epsilon].sub.jt.sup.i] (2)
and testing whether the parameters [[rho].sub.j] are equal to unity. The subscript j (= 1, 2,..., N) again distinguishes the N countries included in the panel and the subscript t (= 1, 2,..., T) denotes the...
|
