How and why do prices of identical products vary across countries?

(1)

How and why do prices of identical

products vary across countries?

Bachelor Thesis within Economics

Authors: Mercédesz Dani

Gabriele Noreikaite

Supervisors: Sara Johansson

(2)

Acknowledgements

We would like to express our gratitude foremost to Sara Johansson and Sofia Wixe for their outstanding tutoring, support and engagement throughout the process of this thesis. Valuable feedback and guidance have been very relevant and stimulating to accomplish this work. Additionally, we would like to thank all members of our thesis seminar group for their important inputs and contribution.

Finally, we are really grateful to our parents and friends who have been supporting us in every step of the thesis writing process.

(3)

Bachelor’s Thesis in economics

Title: How and why do prices of identical products vary across countries? Authors: Gabriele Noreikaite & Mercedesz Dani

Tutors: Sara Johansson Sofia Wixe Date: May 2015

Key terms: Price differentials, the law of one price, purchasing power parity, exchange rates, identical goods

Abstract

The validity of purchasing power parity and the law of one price have been constantly questioned as it is usually assumed that, at least in the long-run, commodity prices are perfectly arbitraged. The purpose of this paper is to investigate how and why do prices of identical goods vary across countries even today, in the globalized world, when trade costs are relatively low due to high mobility and the countless number of free-trade agreements. In order to fulfill the purpose of the study, prices of identical products from H&M are analyzed. The main factors taken into consideration when investigating the nature of price deviations are: tariffs, productivity, trade volume of clothing and how well the producing company is established within the sampled countries. The findings of the regression analysis indicate that price disparities exist and the law of one price, as well as purchasing power parity, fail to hold even when identical goods are compared.

(4)

1 Introduction

The world we currently live in is characterized by globalization; a process by which ‘national and regional economies, societies, and cultures have become integrated through the global network of trade, communication, immigration and transportation’ (Financial Times Lexicon). Thus, one would expect that the more globalized the world becomes, the vast extensity of market conditions and prices of goods would be equal-ized. However, reality demonstrates different patterns.

The law of one price (LOP) states that when expressed in the same currency, the prices of identical goods and services are equal across countries. The purchasing power parity (PPP) supports this theory, however, it reflects the prices of basket of goods instead of individual goods in order for the LOP to hold.

There is a lack of available research investigating the question of why do prices vary across countries for standardized products even today, in the globalized world, when the countries and markets are becoming increasingly integrated. The purpose of this paper is to test if the LOP holds for a basket of identical and standardized goods, as well as to analyze the determinants for the potential price differentials. Previous research has demonstrated that prices in Japan in the 1980s and 1990s were 40% higher on average, than in the other OECD member countries. In the Nordic countries and Switzerland, prices were 15–25% higher. In comparison, prices in the United States were 10% lower than the OECD averge, whereas in Portugal 20% lower (Lapse & Swedenborg, 2010).

In order to compare the prices of the same goods, exchange rates must be incorporated into the analysis in order to get the price of the same good translated into one currency. Another useful tool that is considered in the analysis is the widely used Big Mac Index, that is intended to represent price deviations from the PPP with the help of the current prices of a standardized Big Mac burger at McDonald’s. With the help of this index, currency under- and overvaluations can be detected. However, the use of the index alone is simply inefficient because of trading issues.

This paper is designed to contribute to the investigation of the nature of price differ-ences in the clothing sector, as developed countries are becoming more service-oriented, which implies that more emphasis is put on the retail industry. The reason why the clothing industry is chosen as an example is due to the fact that it has played a decisive

(6)

role in the past economic development. Palpacuer, Gibbon and Thomsen (2004) claim that the clothing sector has played a fourfold role in the development process: (1) it helped reduce unemployment by targeting unskilled labour, (2) it satisfied the basic needs of a large population by mass-production, (3) it created the capital for more tech-nologically demanding production in other sectors, (4) the large volume of clothing ex-ports financed imex-ports of more advanced technologies. Furthermore, price data for ap-parel are easily accessible and highly transparent; clothes are relatively easy and cheap to transport, and significantly important for employment and growth for the developing countries.

In order to test price deviations empirically, the well-established international com-pany, H&M is chosen as an example, whose prices are analyzed in this paper. The main focus relies on the price levels of four chosen identical products in 53 countries incorpo-rating four different regions: North and South America, Middle East and North Africa, Asia Pacific and Europe. Sweden is chosen as a geographical benchmark, as H&M was founded in Västerås and still has its parent company located there. When investigating the prices of identical H&M products in different markets, it is clear that even after con-verting them to a common currency, the prices of identical products strongly differ from each other across numerous countries. Thus the LOP does not hold and further research is necessary.

1.1 Limitations

The four products used in the analysis are strictly identical in terms of size, color, product ID and material from which it is produced. An identical shopping environment (physical and online stores respectively) is assumed across the 53 chosen countries, since H&M stores are standardized and all equipped with the ‘Basic’ products.

There are some limitations regarding the data collection, as there is no available in-formation regarding the prices of the same products from the previous years, collections have changed and catalogues are no longer available. Moreover, as H&M is only pre-sent in 53 markets, the sample size of this research is limited to these countries. It is also hard to predict or collect data on transportation costs associated with these products, since the place of origin and production costs are not stated in the catalogue or on the company’s website. Furthermore, data on tariffs, trade volume and wages are not avail-able for all sampled countries, which results in conclusions based on fewer observations

(7)

whilst incorporating all relevant information. Additionally, the results on the effect of tariffs on price changes are limited, as the European Union member countries posess the same tariff as Sweden.

All necessary measures for the analysis, such as: exchange rates, GDP per capita, wages, tariffs, number of stores, exports and imports of clothes, and total population are collected from various official databases, causing a differentiation throughout time hori-zons. For some variables the most recent information dates back to 2007, while others provide data for 2015. Nevertheless, all data is presented as recent as possible.

The remainder of the paper is structured as follows: in Section 2, the theories used for the analysis of differences in prices are presented. In Section 3, a description of the data set and model formation are presented, as well as the choice of dependent and inde-pendent variables are introduced. In section 4, the empirical findings are analyzed and interpreted. In section 5, the concluding thoughts are discussed and the possible further developments of the paper are addressed.

(8)

2 Theoretical Framework

In this chapter, the theories of the LOP and PPP are discussed and as an implication of the PPP, the real and nominal exchange rates are explained in order to see the pos-sible reasons why the LOP and PPP do not hold. The relationship between demand and pricing, the impact of total trade on price levels as well as previous studies are also in-cluded.

2.1 The Law of One Price

The law of one price is one of the most dated theories in economics. Krugman, Obstfeld and Melitz (2012) define it as a condition, where competitive markets are free of transportation costs and official barriers to trade (such as tariffs), and identical goods in different countries are sold for the same price when prices are expressed in the same currency.

If the law of one price does not hold, arbitrage opportunities exist. This means that a good can be bought for a lower price in one country and then resold for a relatively higher price in another country, making the transaction beneficial for both the supplier and the customer.

Assumptions of the LOP:

- The goods are identical;

- No barriers to trade exist (costless and open trade);

- The commodities’ value is expressed in the same currency. The law of one price is modeled as follows:

PH=PF*E (1)

Where PH represents the price of a good in the home country, PF is the price of the

same product in a foreign country, and E is the nominal spot exchange rate (price of foreign currency in terms of domestic currency). Therefore, arbitrage opportunities exist if PH ≠ PF*E. For example, if PH<PF*E, then it is cheaper to buy the commodity at home

and to resell it on the foreign market for a price between PH and PF. On the other hand,

if this occurs in a larger volume, the demand will increase in the local market and de-crease in the foreign market. This will result in a new equilibrium condition in the short-run with reduced foreign prices and increased domestic prices. However, in the

(9)

long-run, one would expect prices to converge.

Several researchers, such as Engel and Rogers (2001) and Fenestra and Kendall (1997) find evidence on the fact that a significant portion of the observed deviations from the law of one price is attributable to an incomplete exchange rate pass-through (respon-siveness of prices to exchange rate changes) as a result of local currency pricing.

2.2 Purchasing Power Parity

The theory of PPP is one of the most fundamental economic concepts, firstly put into practice by the economist, Gustav Cassel (Officer, 1976) in the early 20th century. Fol-lowing, Krugman et al. (2012) define purchasing power parity as a condition where two countries’ price level ratio equals the exchange rate of their currencies. Therefore, an increase (decrease) in the domestic price level would result in a depreciation (apprecia-tion) of the domestic currency in the foreign exchange market, indicating a decrease (in-crease) in purchasing power.

According to the theory of absolute PPP, the home price of a domestic commodity basket of goods and services should equal the foreign price of a foreign basket in order to reach the equilibrium condition. For the absolute PPP to hold, the baskets must meet certain prerequisites:

1) The baskets must contain identical goods;

2) Prices of the goods have to be expressed in the same currency; 3) The LOP has to hold for all the identical goods.

Every year since 1986, The Economist issues the alleged Big Mac Index, a set of data that indicates the current prices of a Big Mac hamburger throughout different countries, making a comparison between the price levels with the help of the actual exchange rate. The reason why this index is used as a measure of PPP instead of the LOP is that the hamburger is considered as a “basket” of its ingredients. ‘Most of the ingredients that go into a Big Mac are individually traded on international markets, so we might expect that the law of one price would hold, at least approximately’ (Pakko & Pollard, 2003, p. 11). To determine if the Big Mac index can give an explanation related to tradable goods, the correlation between the relative Big Mac prices and the selected pieces of H&M clothes’ prices are incorporated into the analysis.

(10)

by the exchange rate), are the same. Since in reality this is not the case, the Big Mac In-dex is rather used as an indicator of how abscure price levels are set. Many economists have studied the applicability of the Big Mac Index, as it raises the question of why does a single product have different prices across the globe. Alessandria and Kaboski (2008) find that the average price of the burger is lower in relatively poor countries be-cause the labor cost is low. Almås (2012) finds that due to substitution and quality bias, relatively poorer countries’ incomes are overestimated, causing PPP bias within the food industry. On the other hand, since the production process of a hamburger is not traded, Balassa and Samuelson’s idea, developed in 1964, of how non-traded goods sys-tematically affect the deviation from PPP seems to be appropriate. The Balassa-Samuelson effect claims that an increase in the relative productivity of tradables in a country raises relative wage, resulting in the increase of relative average price (Mac-Donald & Ricci, 2001).

According to Sarno and Valente (2006), who study the deviations from PPP under dif-ferent exchange rate regimes, ‘regardless of the great interest in this area of research, manifested by the large number of papers on PPP published over the last few decades, and despite the increasing quality of data sets utilized and the econometric techniques employed, the validity of long-run PPP and the properties of PPP deviations remain the subject of ongoing controversies’ (Sarno & Valente, 2006, p. 3148).

2.3 Nominal Exchange Rate and Real Exchange Rate

Broadly speaking, an exchange rate is the price of one currency in terms of another. However, there is a difference between nominal and real exchange rates. As Krugman et al. (2012) state, because of their strong influence on the current account and other mac-roeconomic variables, exchange rates are among the most important prices in an open economy. The relationship between the exchange rates and the law of one price (thus PPP) is of key importance in order to understand the nature of price differences.

Following Krugman et al. (2012), the nominal exchange rate is defined as the relative price of two currencies:

E = (2)

where E is the nominal exchange rate and P* and P are the domestic and foreign price levels respectively.

(11)

One step to broaden the theory of PPP is to describe what real exchange rate, which is the relative price of basket of goods between two countries. Since the real exchange rate is defined in terms of both prices and the nominal exchange rate, the relationship is the following:

q = (3)

where q represents the real exchange rate.

The main prediction of PPP is that the real exchange rate must remain unchanged. As equation (3) shows, the real exchange rate can never change when absolute PPP holds. Thus, the real exchange rate plays a more important role in relation to this research as it can tell when PPP fails to hold and it also takes the relative price differences between countries into account.

Furthermore, the shifts in supply and demand in open economies cause nominal and real exchange rates to move, causing PPP not to hold anymore. From equation (3) it can be seen that with stable output prices, nominal depreciation (appreciation) implies real depreciation (appreciation).

Krugman et al. (2012) conclude that exchange rates play a central role in international trade because they allow us to compare the prices of goods and services produced in dif-ferent countries.

Several economists have analyzed the impact of the nominal exchange rate against the real exchange rate on various issues related to PPP. Sarno and Valente (2006) find that the relative importance of exchange rates and relative prices in restoring the long-run equilibrium level of the exchange rate varies over time and is affected by the nominal exchange rate arrangement in operation. Empirical studies in the literature typically find slow long-run convergence of the real exchange rate to PPP (Xu, 2003).

2.4 Why LOP (and thus PPP) does not hold?

Despite the fact that PPP and the LOP are well-established economic models, empiri-cal research has demonstrated that they have several limitations and do not hold under certain circumstances. One of the empirical analysis has shown that ‘a Big Mac devotee could buy one and a half of the sandwich in the USA for every one he could purchase in Switzerland’ (Pakko & Pollard, 2003, p. 16). According to the PPP, a Big Mac should

(12)

have a fixed price globally. It would be impractical to ship a burger from a cheaper country to a more expensive one, even though the components of the sandwich are traded on world markets and the prices should be equalized.

Ardeni (1989) states that where a distinction between traded and non-traded goods is made in other models, deviations from the LOP are existing for non-traded goods only. Other empirical research about the general price levels find PPP to hold in respect to the long-run. However, there are an increasing amount of studies showing significant devia-tions even in the long-run. Isard (1977) states that ‘in reality the law of one price is fla-grantly and systematically violated by empirical data’ (Isard, 1977, p. 942). Further-more, Ardeni’s (1989) research on commodity trade demonstrates that the prices of primary commodities are generally considered independent of the country of origin as such commodities are expected to be identical or nearly perfect substitutes and therefore perfectly arbitraged. However, some perversions may occur in these markets. Factors which potentially create permanent price differentials across countries are import quo-tas, tariffs and various means of trade control. There are also some other reasons such as institutional aspects, high cost of arbitraging and errors in data. All applicable when evaluating the failure of the LOP.

In the following sections some of the reasons why LOP (and thus PPP) does not hold are presented in more detail such as trade barriers, transportation costs and productiv-ity.

2.4.1 Trade Barriers

There are many government restrictions, such as import/export licenses, import quotas and subsidies which function as trade barriers. Tariffs play a crucial role in international economics as a barrier to trade. This government intervention usually influences income distribution within the country. For the purpose of the research problem, tariffs are con-sidered as the most important means of trade barriers, as they can create a difference be-tween prices for the goods which are traded bebe-tween countries as well as the differences in prices within a country. ‘The direct effect of a tariff is to make imported goods more expensive inside a country than they are outside the country’ (Krugman et al., 2012, p. 124).

Tariffs are used in most countries as protective tools against competition originating from other countries. If the government imposes a tariff, it means that the imported

(13)

goods become relatively more expensive than the domestically produced ones. On the other hand, Ravn and Uribe (2007) argue that price differences among countries, thus deviations from the LOP, hold even when tariffs or quotas are not existing.

There are several academic papers that analyze the trade restrictions in relation to dif-ferences in commodity prices. Parsley and Wei (1996) state that due to barriers to flows of goods and services, minimal expectations exist that price differences will immedi-ately disappear as they do, for example, in financial markets. In an empirical paper, Pandit (2009) examines U.S.-Canada softwood lumber trade. One of the main findings of this research shows that ‘U.S. demand factors and trade barriers on the import price of Canadian softwood lumber suggests that these factors and barriers explain about 91% of the variation in the import price’ (Pandit, 2009, p. 417). As it can be seen, barriers to trade are relevant for price fluctuations and should be taken into account when analyz-ing the price differences of identical products.

2.4.2 Transportation Costs

The expenses encountered when changing the geographical locations of products are one of the main determinants of price differences. Transportation costs arise mostly on three occasions: (1) when raw materials are transported from the supplier to the cus-tomer, (2) when the products are transported to distribution centers or stores, and (3) when the products are delivered to the end-customer.

Ravn and Mazzenga (2004) have studied the quantitative role of transportation costs in the international business cycle. Their findings support the hypothesis according to which ‘costs of transportation can have large welfare costs while their effects on the pat-tern of trade depend crucially on the degree of substitutability between domestic and foreign goods’ (Ravn & Mazzenga, 2004, p. 668). Thus, transportation cost is a factor contributing to the failure of PPP and the LOP as it causes prices to diverge across countries.

2.4.4 Productivity

‘It’s a good bet that most of the clothing you are wearing as you read this came from a

country far poorer than the United States.’(Krugman et al., 2012, p. 279)

As it was already mentioned, the Balassa-Samuelson effect contributes to the explana-tion of how productivity affects price differences: an increase in the relative

(14)

productivi-ty of tradables increases the relative average price level in the country. Their notion is based on non-traded goods: variation in prices for non-tradables contributes to the dis-crepancies between rich and poor nations. Krugman et al. (2012) state that ‘rich coun-tries with higher labor productivity in the tradables sector will tend to have higher non-tradables prices and higher price levels’ (Krugman et al., 2012, p. 402). In their study, García-Solanes, Sancho-Portero and Torrejón-Flores (2008) give an empirical support to the Balassa-Samuelson differential productivity postulate. The conclusions are that the difference in relative prices of non-tradables leads to overall differences in price levels and wealthier countries are associated with relatively elevated price levels.

Moreover, productivity differences explain variations in wages and GDP per capita. Wages play an important role when examining price differences of goods as they take the marginal productivity of labour (MPL) into account. MPL is defined as a unit change in output due to one unit change in labour. Krugman et al. (2012) also define wages as the function of marginal productivity of labour and price. Thus, wages are not only related to productivity, but also to the actual price of a product:

MPL * P = w (4)

thus

P= w / MPL (5) Due to lack of available data on wages, one can use GDP per capita as a proxy for wage (hence productivity) differences. GDP is the most widely used measure of living standards as well as an indicator of the overall economic performance of a country. Compared to wages, GDP per capita is a boader measure of average income and stand-ards of living, implying that GDP per capita may also reflect differences in consumption patterns and preferences. Higher GDP per capita represents higher income and living standards, while at the same time, real income per capita is positively related to the price level of the country if expressed in a single currency. Causa, Araujo, Cavaciuti, Ruiz & Smidova (2014) find a positive correlation between the levels of GDP and household income. Thus, productivity and hence GDP per capita can partially explain price differentials in different countries and that is why it is relevant to include this measure as one of the explanatory variables in the empirical analysis.

(15)

GDP per capita variations across countries are demonstrated in Figure 1 where the da-ta for 2013 is obda-tained from the World Bank1 database.

0 200,000 400,000 600,000 800,000 1,000,000 L u xe n b o u rg h N o rw a y (N O K ) Q a ta r (Q A R ) S w itze rla n d ( C H F ) A u st ra lia ( A U D ) S w e d e n ( S E K ) D e n m a rk ( D K K ) S in g a p o re ( S G D ) U S A K u w a it (K W D ) C a n a d a ( C A D ) T h e N e th e rla n d s A u st ria Ir e la n d F in la n d B e lg iu m G e rm a n y U A E ( A E D ) F ra n ce U K ( G B P ) H o n g K o n g ( H K D ) Isr a e l ( IL S ) It a ly S p a in S a u d i A ra b ia ( S A R ) B a h ra in ( B H D ) G re e ce O m a n ( O M R ) P o rt u g a l C ze ch R e p u b lic (C Z K ) E st o n ia S lo va ki a C h ile ( C L P ) L a tvi a R u ss ia ( R U B ) P o la n d ( P L N ) C ro a tia ( H R K ) H u n g a ry ( H U F ) T u rke y (T R Y ) M a la ysi a ( M Y R ) M e xi co ( M X N ) Lebanon (LBP ) R o m a n ia ( R O N ) B u lg a ria ( B G N ) C h in a ( C N Y ) S e rb ia ( R S D ) T h a ila n d ( T H B ) Jo rd a n ( JO D ) In d o n e si a ( ID R ) E g yp t (E G P ) M o ro cc o ( M A D ) P h ill ip p in e s (P H P ) T a iw a n ( T W D )

GDP per Capita SEK

Figure 1- GDP per Capita (SEK) 2013

In this paper, GDP per capita is used as an estimator of wages as there is a strong cor-relation (~92%) between the two variables by using 41 pairs of observations (Figure 2).

0 4,000 8,000 12,000 16,000 20,000 0 200,000 400,000 600,000 800,000 1,000,000 GDP per Capita (SEK)

W a g e ( S E K )

Figure 2 - Correlation between GDP per Capita and Wage (SEK)

When analyzing productivity, it is crucial to take the volume of trade, which is exports plus imports, into consideration. Adam Smith and David Ricardo argue that overall openness is very important – nations grow more prosperous through imports and ex-ports (Ondrich & Richardson, 2004). Since changes in GDP also reflect price

1

(16)

ences across countries as it is discussed above, the impact of trade would affect changes in price level through changes in GDP.

There are two possible explanations how trade could influence changes in income and productivity and thus the price level. Firstly, GDP is affected positively by an increase in net exports (exports minus imports). As Ondrich and Richardson’s (2004) research indicates, countries with large exports (and low import penetration) have high income per capita which would lead to an increase in the GDP and thus in the price level. Therefore, relatively more exports than imports increases GDP. Secondly, many re-searchers have analyzed how trade volume (imports plus exports) affect GDP and in-come level in the country. Brunner (2003) has found that trade has a significant and large effect on the level of income. Li, Chen and San’s (2010) result suggests that there exists a long and short-term causality between GDP and total trade. In general, it could be expected that the increase in the trade volume (exports plus imports) in a country in-creases the level of income and thus GDP leading to an increase in the price level.

2.5 Demand-Based Pricing

A widely used approach to generate profits when entering a new market is to charge a relatively higher price to both high-price and low-price products, whilst gradually ad-justing it to market-demand over time. Saturation is an essential factor that needs to be considered during the setup of a pricing-strategy, that is: a downward shift of the de-mand function due to increased market penetration that results from the loss of potential buyers (see Figure 3) (Raman & Chatterjee, 1995; Khouja & Robbins, 2005). The satu-ration effect will eventually result in the decrease of the price of a product. In other words, the less common the product is in the market (either because it has not been pre-sent for long or the products are not supplied in the amount that would meet the de-mand) the higher the price can be set and maintained by ocmpanies. Different demand conditions in different countries are reflected by pricing-to-market that is connected to the elasticity of demand to the country’s consumers. Empirical studies have shown the clear existence of this practice in manufacturing trade (Raman & Chatterjee, 1995).

(17)

Figure 3- The Saturation Effect

2.6 Previous Studies

Several previous researches have been conducted in order to examine the reasons be-hind the price differences across countries of the products of another Swedish giant, IKEA. Baxter and Landry (2012) have differentiated high-price and low-price products when examining the entire assortment of IKEA and found a remarkable difference tween the price-deviations between the two categories. Furthermore, the difference be-tween the deviations of new and continuing products have been studied. The results show that the mean price differences of continuing goods differ more than in the case of new goods.

The other study by Haskell and Wolf (2001) examines absolute prices for more than 100 identical goods sold in 25 countries by IKEA. They find significant common cur-rency price divergences across countries for a given product and across products for a given country pair.

Moreover, other studies have used the Big Mac Index to model price differences and the results unanimously show that the LOP does not hold for identical goods. Alessan-dria and Kaboski (2008) find that the prices of Big Macs across countries provide a clear example of the LOP failure. When converted into U.S. dollars, Big Macs sell for up to 65 % more than in the U.S. and down to 57 % less than in the U.S.

A project on price comparisons in Europe was led by The European Consumer Centre Network (ECC-Net) in 2009. 27 ECCs participated in the survey in Luxembourg, called ‘Price Research, Price Differences in Europe’. This survey was based on some of the most common textile products of Zara, C&A and cosmetic products of Body Shop. Prices of three items for women and two items for men from each company have been compared and the result has demonstrated that the price differences of identical goods

(18)

exist between countries and those differences are not easily explained. The main find-ings clearly show that:

 shopping in Portugal is cheaper than in other European countries;

 Scandinavian countries such as Norway, Finland, Denmark and Sweden appeared to be more expensive than the rest of Europe;

 prices within the non-Euro zone deviate more from the average prices in Europe than inside the Euro zone.

(19)

3 Methodology

3.1 Primary and Secondary Data

The research method chosen for the paper consists of the collection of both primary and secondary data. As previous researchers on the topic use the Big Mac Index as the main indicator of price differences, this paper makes a contribution to the literature, since the data collected is about true tradables and is obtained by the authors them-selves, providing a unique dataset.

3.2 Data Collection and Measures

Data are collected on exchange rates, tariffs, GDP per capita, wages, total population, exports and imports of clothes, the PPP conversion factor and Big Mac Index from vari-ous databases such as World Bank or World Trade Organization (WTO)2. It ensures that the research has enough information to be comparable and analyzable. The data re-trieved from the observations are transcribed and evaluated by the authors themselves. All data are converted to a relative form, taking the values for Sweden as a basis in or-der to see how the price difference is affected by the differences in the selected explana-tory variables.

Data are also collected on the prices of the chosen sample products and the number of stores in each country from H&M’s website for all sampled countries. The company´s products are divided into several departments and in order to make easier comparisons, only low-price and continuing - ‘Basic’ - products are selected and examined from women’s and men’s departments respectively. Overall, 4 products are compared whose short descriptions can be seen in Table 1.

(20)

Table 1- Description of Products

Name Department Description Details

Sweatpants Men - Basics - Bottoms

Sweatpants with an elasticated drawstring waist, side pockets, one back pocket and ribbed hems. Brushed

inside.

60% cotton, 40% polyester

Polo shirt Men - Basics - T-Shirts & Vests -

Short Sleeve

Polo shirt in stretch jersey with a col-lar, short sleeves and buttons at the

top.

95% cotton, 5% elastane

Jersey skirt Ladies - Basics - Dresses & Skirts Short skirt in jersey with an

elasti-cated waist.

95% cotton, 5% elastane Long jersey

top Ladies - Tops - Vests

Long top in cotton jersey with a racer back.

95% cotton, 5% elastane

Prices are collected from 53 countries and all include value added tax (VAT). After collecting prices in local currencies, the first step is to convert them into Swedish Krona with the help of exchange rates. All exchanges rates were obtained on the 15th March 2015 as the most recent data to work with. The current date is important in order to see the most recent price deviations possible. To avert from creating misleading calculations and conclusions, relative price deviations have been calculated for each country for all four chosen products. Sweden is chosen as a benchmark with the relative price of 1 (see Appendix Table 1). Visual deviations of the mean prices can be seen in Figure 4 where 100 corresponds to the value of 1. As presented in the bar chart, relative prices deviate from 1, indicating that further research is needed and it is appropriate to work with the data not making biased estimates. A value higher (lower) than 1 indicates that it is less (more) expensive to buy the selected products in Sweden. Moreover, it is clear that prices heavily deviate in countries outside the EU, while among the EU member states the prices are relatively constant.

(21)

0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6 Jo rd an (J O D ) Le ba no n (L B P ) C hi na (C N Y ) S w itz er la nd (C H F) E gy pt (E G P ) Is ra el (I LS ) Q at ar (Q A R ) U A E (A E D ) S in ga po re (S G D ) S au di A ra bi a (S A R ) M or oc co (MA D ) Th ai la nd (T H B ) B ah ra in (B H D ) O m an (O MR ) K uw ai t ( K W D ) P hi llip pi ne s (P H P ) N or w ay (N O K ) M al ay si a (M Y R ) Ta iw an (T W D ) H on g K on g (H K D ) A us tra lia (A U D ) U S A U K (G B P ) D en m ar k (D K K ) C hi le (C LP ) S w ed en (S E K ) In do ne si a (ID R ) C an ad a (C A D ) M ex ic o (M X N ) A us tri a B el gi um E st on ia Fi nl an d Fr an ce G er m an y G re ec e Ire la nd Ita ly La tv ia Lu xe nb ou rg h S lo va ki a Th e N et he rla nd s C ro at ia (H R K ) R us si a (R U B ) S pa in P ol an d (P LN ) B ul ga ria (B G N ) S er bi a (R S D ) C ze ch R ep ub lic (C ZK ) R om an ia (R O N ) P or tu ga l H un ga ry (H U F) Tu rk ey (T R Y ) Mean Price

Figure 4- Mean Price Distribution (SEK)

Furthermore, prices are also compared with the Big Mac Index. The index was ob-tained for the selected countries as of January 2015. Figure 5 indicates the relationship between the relative prices of Big Mac and the relative prices of the selected H&M products. It shows that the two variables are not strongly correlated, since there is no observable pattern between the dots, as shown by the regression line. The reason for this may be that clothes are tradable goods while the production process of the Big Mac is non-tradable. It could also be argued that the raw ingredients of a burger are traded, al-though the whole value of the Big Mac is not comparable to the meat or cheese used in preparation. Also, it is more reasonable to treat the product as a non-traded good. There-fore, for the further research of this paper the Big MacIndex is not considered as one of the explanatory variables.

Figure 5- Correlation between relative Big Mac and H&M products price deviations (SEK) 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

Relative price BigMac

R e la ti ve P ri ce H & M

(22)

The data on tariffs levied on clothes are collected from the websites of WTO and ITA3 and are expressed as the average tariff for HS chapter 62: ‘Articles of apparel and cloth-ing accessories, not knitted or crocheted’ from the followcloth-ing subchapters:

 6203 – Men's or boys' suits, ensembles, jackets, blazers, trousers, bib and brace overalls, breeches and shorts (other than swimwear);

 6204 – Women's or girls' suits, ensembles, jackets, blazers, dresses, skirts, di-vided skirts, trousers, bib and brace overalls, breeches and shorts (other than swimwear);

 6205 – Men’s or boy’s shirt;

 6206 – Woman’s or girl’s blouses, shirts and shirt-blouses.

The rate of ad valorem tax is collected for these four categories and their average is used in the regression model. As it can be seen from Appendix Table 1, all European Union members hold the same tariff, which is equal to 10%. It implies that all member countries can circulate their goods freely. The European Commission defines the 'Common Customs Tariff' (CCT) which applies to the import of goods across the exter-nal borders of the EU. Thus, tariffs are not explaining sufficiently the price differences between the European Union members. Only tariffs associated with non-EU countries can have a measurable influence on the price differences.

Sweden’s trade with EU countries is very important, because according to Statistics Sweden4 it accounts for a large share of exports and imports respectively. In 2014, the EU countries accounted for 73.7% of Sweden's exports and 84.6% of Sweden's imports. Having free trade within the EU countries is sufficient, since the world demand outside the EU is rapidly growing. This is a direct result of why opening markets with other countries is increasing market opportunities for negotiating new businesses and other benefits which can be obtained through free trade agreements. Currently, the European Free Trade Association (EFTA) States hold 25 free trade agreements (covering 35 coun-tries) with the partners shown in Appendix Table 2.

3_{Iternational Trade Administration - http://www.trade.gov/}

4

(23)

3.3 Choice of the Dependent Variable and the Company

The relative average price of selected H&M clothes is chosen as the dependent vari-able. According to the most recent data, the company is represented in 53 markets with more than 3,300 stores worldwide. Since the analysis requires identical products to be compared, the company’s production scale complements the research well. During the past decade, H&M has become one of the most famous clothing companies due to af-fordable price of its identical products. The company is growing at a fast pace (their tar-geted growth is 10-15 % per annum) and outsources to approximately 700 facilities, lo-cated mainly throughout Asia and Europe. These include particularly low wage coun-tries such as Turkey, Bangladesh, Indonesia and India (H&M Corporate Website). As one of the the company’s primary aims is to ensure the best price to its customers through cost-consciousness and efficient logistics it is interesting to investigate how the prices of the company’s products vary across countries worldwwide.

Lack of research regarding price differences of this company’s products combined with Sweden being the parent country, are two of the underlying principles when choos-ing this company and its production as an example. Furthermore, deviations in prices of identical products have interested many researchers that earlier mainly have previously worked with comparable Swedish giants, such as IKEA, which is presented in the pa-per.

3.4 Choice of Independent Variables

In order to explain the average relative price in foreign countries compared to Sweden, the following independent variables are chosen:

1) Relative GDP per capita: as GDP per capita is highly correlated with the level of minimum wages, it is presumed as a viable estimator of wages. Also, GDP per capita can tell a lot about the tastes and preferences related to the living standards of the residents of the country. Thus, relative GDP per capita can indi-cate if a country is wealthier or poorer than Sweden. A positive relationship is expected between the relative average price and the relative GDP per capita. As it increases, so does the wealthiness of the country, therefore the general price level is expected to be higher;

(24)

2) Relative PPP conversion factor5: In order to measure the purchasing power conditions for the selected countries, the PPP conversion factor is collect-ed. This analyzes how many units of the country´s currency have the same buy-ing-power for a given basket of goods and services to the US dollar. To make the measurement relevant to the study, the values are converted in relation to the Swedish value. One would expect a positive relationship between the relative average price and the relative PPP conversion factor, as a higher PPP conversion rate reflects a higher general price level.

3) Relative tariff (in %): In order to reflect how difficult it is to access a country with the product, the ad valorem tariff is an appropriate measure. Rela-tive tariff shows if it is more or less difficult to access a foreign market than to access the Swedish market. The change in price is expected to be influenced positively by the tariff, as the higher the tax imposed on the imported good is, the more difficult it is to enter the market, thus more expensive the domestic price of the good becomes.

4) Relative trade volume: The relative volume of trade (exports plus im-ports) can be used as a proxy for domestic productivity and competitiveness in the clothing industry. The volume of exports reflect how productive a country is in the given sector, while the volume of imports indicate the consumption patters of a nation in relation to Sweden. One would expect that domestic prices on clothes are relatively low in countries with large clothing industries due to in-creased competition.

5) Relative total population per store: The population per store indicates how unique H&M is within a country. The value would tell if the H&M store is more or less accessible in the sampled country than in Sweden. If the value is greater (smaller) than 1, it means that H&M stores are less (more) available for the inhabitants of the country, showing lower (higher) market-penetration, creat-ing a higher (lower) status and thus higher (lower) prices for the selected prod-ucts. The relative average price is expected to be affected positively by the total population per store, as the higher the share of population per store, the less the saturation is apparent, resulting in potentially higher prices of the company´s products.

5_{The PPP conversion factor was planned to be included as an independent variable, but since it shows}

(25)

Table 2 - Expected Signs

Variable Expected Sign

Relative GDP per capita +

Relative PPP conversion rate +

Relative tariff (in %) +

Relative trade volume +

Relative total population per store +

3.5 Model Formation

Firstly, the econometric model is tested by the use of OLS estimation as the expected relationship between the dependent variable and each of the independent variables is linear. The dependent variable; the relative average price of the products is tested against the independent variables, such as: relative GDP per capita, relative tariff, rela-tive total trade and relarela-tive total population per store.

It is also important to test for normality, model misspecification, multicollinearity and heteroscedasticity. After computing all estimations, the conclusion is drawn and the relevance of the chosen variables is determined.

3.6 Empirical Model

The data is processed with the statistical software, EViews. During the procedure, the linear regression model is used in order to analyze the relation between the dependent and independent variables. The regression model is developed for quantitative analysis based on the theories discussed previously and set up accordingly:

Yi=β0+β1ADVi+β2GDPCi+β3TTi +β4PSi+ε (6)

where Y denotes the price relative to Sweden, ADV denotes the rate of ad valorem tar-iff compared to Sweden, GDPC denotes the GDP per capita relative to Sweden, TT de-notes the total volume of trade of clothes relative to Sweden and PS dede-notes the popula-tion per store relative to Sweden. With the help of OLS estimapopula-tion, the β-values are es-timated in order to see how much each explanatory variable influence the extent of the price differences.

(26)

3.7 Diagnostic Tests

Tests for normality, heteroscedasticity, misspecification and multicollinearity are used in order to check the consistency and stability of the regression model.

Firstly, the Jarque-Bera test is used to check if residuals are normally distributed. Since the probability (p-value) is lower than the chosen level of significance (0.0011<0.05), the normality between the residuals is not present in the model (Appen-dix Table 3). The presence of non-normality can be due to the fact that the dataset con-tains too many outliers, resulting in a skewed distribution of each of the variables as well as the residuals (Table 3). However, as the assumption of normality is not neces-sary in order for the OLS estimation to be unbiased. Secondly, White’s test is applied to see if heteroscedasticity exists among the residuals. Since the probability associated with the ‘Observations*R2_{’ is lower than the level of significance (0.008<0.05), it can}

be concluded that there is heteroscedasticity in the model (Appendix Table 4). As a remedy, White heteroskedasticity-consistent standard errors and covariance have been applied. Thirdly, in order to test if the model is correctly specified, Ramsey’s RESET test is made. The probability of the likelihood ratio, which is the RESET test statistic, is higher than the significance level (0.59>0.05) meaning that the model is correctly speci-fied (Appendix Table 5). Finally, variance inflation factor values are less than 10, mean-ing that there is no multicollinearity in the model. The rule of thumbs is that if VIF<10, then there are no severe multicollinearity problems (Appendix Table 6). In addition to this, the correlations between the variables have been checked and it can be concluded that none of the variable pairs are reaching 50% correlation (Appendix Table 7).

Table 3 – Descriptive Statistics

Variable Mean Max Min Std. dev.

Price 0.9678 1.3530 0.7652 0.1268

Population per store 47.7184 601.2703 0.7893 22.1999

Total trade 2.3145 16.3344 0.0530 3.8661

Tariff 1.0037 2.1800 0.0000 0.4115

(27)

4 Empirical Results

The regression is performed on the regressand, the relative average price of selected products. The predictor variables are as follow: ad valorem tariff, GDP per capita, trade volume and finally, the relation between total population and the number of H&M stores in the country. As shown before, two of the selected explanatory variables, GDP per capita and the PPP conversion rate are showing strong correlation. Thus, one of the variables has to be excluded from the model. As GDP per capita captures the entire eco-nomic activity of a country, it is included in the regression. The results are displayed in Table 4.

Table 4 - Regression Output

Dependent variable: Relative price of H&M clothes

Variables Regression (1) Regression (2) Regression (3) Constant 1.0454*** 0.9027*** 0.9999*** (15.7086) (21.2722) (19.7111) Relative Ad valorem -0.0266 0.0117 -0.1499*** (-0.6038) (0.4375) (-3.5216)

Relative GDP per capita 0.0035 0.0299 0.1022**

(0.056) (0.7939) (2.728)

Relative trade volume 0.0002 0.0001 0.0144***

(0.0309) (0.0212) (3.3877)

Relative population per store 0.0002 -0.0001 0.0005***

(0.9853) (-0.7572) (3.7217)

Dummy Asia Pacific 0.2308***

(4.2148)

Dummy Middle East and North Africa 0.3478***

(8.9254)

Dummy North and South America 0.0616

(0.9709)

R2 0.0232 0.6778 0.4893

Adjusted R2 -0.0618 0.6253 0.4274

Number of observations 51 51 38

***p-value<0.01 **p-value<0.05 *p-value<0.1 The figures in parentheses indicate the t-statistics

When analyzing Regression (1), the intercept, β0, represents the relative price of H&M

for selected products in Sweden in comparison to the sampled 51 countries6.

(28)

ly, this figure equaled 1.0454, which means that the spread of relative prices between Sweden and the average is about 1.0454. Other than the intercept, no other variables are significant, meaning that their effect cannot be differentiated from zero.

The estimated R2 of the model is 0.0232; therefore, 2.32% of the change in relative price is explained by the change in independent variables. Since none of the selected in-dependent variables are significant in the first model, and the R2 is extremely low, it is concluded that this model is not correctly specified or that among the observed coun-tries, the chosen variables do not have any significant effect on the price differences.

A possible explanation for the weakness of Regression (1) is that due to the fact that H&M has recently endeavored to expand throughout Asian and Middle Eastern coun-tries, it is assumed that the prices in these areas differ greatly from those areas, where H&M has been present for several years. In addition to this, in most of these countries H&M still exists in a form of franchise, so prices are assumed to be higher. For exam-ple, the highest relative prices are in Lebanon (1.4) and Morocco (1.5), whose markets H&M has entered in 2012 via franchise, so it is assumed that the company has not pene-trated the market entirely.7

In order to account for these general price differences, dummy variables are intro-duced in Regression (2). Dummy variables reflect the region to which each country be-longs to. After collecting all the relevant data, Asian and Middle Eastern countries are reasonably excluded from the scope of the study. However, the use of dummy variables is a suitable remedy for this problem, making it possible to keep the original sample size. As a result, a second regression has been set up:

Yi=β0+β1ADVi+β2GDPCi+β3TTi +β4PSi+DUMA+DUMM+DUMN+ε (7)

Where DUMA represents the dummy variable for Asian and Pacific countries, DUMM represents the dummy variable for Middle Eastern and North African countries and DUMN represents the dummy variable for North American and South American countries.

When comparing the results of Regression (2) to those of Regression (1), it is clear that the inclusion of dummies have greatly improved the model. Less of the independent

7_{Included countries: Bahrain, Egypt, Hong Kong, Indonesia, Jordan, Kuwait, Lebanon, Morocco, Oman,}

(29)

variables are considered insignificant and a great effect of geographical location on price differences is visible, thus the assumption made earlier (that Middle Eastern and Asian markets have not been penetrated entirely) is supported by the empirical results.

The R2 value of 0.68 means that 68% of the price changes are explained by the second model. However, this increase compared to the value in Regression (1) is partly due to the increase of explanatory variables from 4 to 7. Gayawan and Ipinyomi (2009) study which criteria shows more reliable results when it comes to model selection. In their pa-per, ‘A Comparison of Akaike, Schwarz and R Square Criteria for Model Selection Us-ing Some Fertility Models’, they find that eventually, R2

would indicate more complex models to be more appropriate, but this can lead to over parameterization (the use of too many parameters in the model).

Other than the intercept term, only two dummy variables are shown to be statistically significant. The dummy ‘Asia Pacific’ implies that the countries in this region have generally higher prices (0.23 unit increase) than European countries. Moreover, the countries belonging to the ‘Middle East and North Africa’ region have even higher pric-es (0.38 unit increase).

As most of the variables are still insignificant in the Regression (2), it is found that the chosen variables do not have a measurable effect on the prices even when the countries are controlled for the geographical region which they belong to.

A new model, Regression (3) is set up with the exclusion of the countries whose mar-kets have not been penetrated entirely by H&M, thus reducing the sample size from 51 to 38. However, this reduction results in the fact that the model can only be applied to describe the prices in the European and American regions.

Regression (3) shows a relatively high R2 (0.49), which means that 49% of the price variations are explained by the model.

In contrast to the expectations, the relative ad valorem tax shows a negative effect on the relative price. This could be due to the fact that relative GDP per capita and relative ad valorem tariff shows a negative relationship (the higher the GDP per capita, the low-er the ad valorem is) (see Table 8 in Appendix). This implies that wealthilow-er countries levy lower taxes on imported clothes. It can also be noted that the main influence comes from the volume of ad valorem tax and GDP per capita, which is supported by the

(30)

theo-ry, as these two components reflect the economy and trade between countries more ac-curately than the other variables.

In accordance with theory, the relative GDP per capita shows a positive affect on the relative price. This is due to the fact that higher GDP per capita reflects higher produc-tivity and standard of living that is resulting in higher price levels.

The relative trade volume (exports plus imports) also show a positive relationship to prices according to the regression output, meaning that the more a country is involved when concerning trade, the higher the relative prices become. An underlying assump-tion in this case is that exports exceed imports, resulting in a less competitive environ-ment where higher prices can be set. According to the dataset obtained for exports and imports, only 2 out of the 38 countries have more imports than exports confirming the idea that trade volume positively affects the dependend variable.

Lastly, the total relative population per store has a positive impact on the dependent variable. In other words, the less available the store is in a country, the higher the aver-age price level is. This result is also in accordance with the underlying theoretical as-sumptions of the saturation effect: the higher population per store indicates less satura-tion, resulting in potentially higher prices of the company’s products.

To sum up, despite the fact that Regression (1) includes a full set of observations, the insignificance of the independent variables and the extremely low R2 suggest that the model is not performing well with the dataset. On the other hand, Regression (2) cap-tures the effect of the geographical location and demonstrates significant results. Fur-thermore, the R2 is considerably high, meaning that the model performs well, however, most of the independent variables are insignificant. Lastly, Regression (3) shows signif-icant relationships between the dependent and the independent variables, although, this result is obtained at the cost of excluding the countries that have demonstrated strong deviations.

4.1 Sensitivity Analysis

As the PPP conversion rate is considered as a pure price reflecting value, a new model is set up that includes this variable. In order to test the sensitivity of the regression model, relative GDP per capita is replaced by the relative PPP conversion rate (as they are 84% correlated) and the following model, Regression (4), is set up:

(31)

YI=β0+β1ADVi+β2PPPi+β3TTi +β4PSi+ε (8) Table 5 – Modified Model

Dependent variable: Relative price of H&M clothes

Variables Regression (4) Constant 0.9696*** (15.2508) Relative Ad valorem -0.1639*** (-3.7712) Relative PPP 0.1583** (2.3056)

Relative trade volume 0.0142***

(3.2493)

Relative population per store 0.0005***

(3.5571)

R2 0.4610

Adjusted R2 0.3956

Number of observations 38

***p-value<0.01 **p-value<0.05 *p-value<0.1 The figures in parentheses indicate the t-statistics

The results in Table 5 show that the inclusion of PPP instead of GDP results in a model with a slightly lower R2 (0.46<0.49). Even though all the explanatory variables are significant, they have slightly different coefficients in comparison to Regression (3). Therefore, relying on the model-selection criteria, the R2 value, Regression (3) is pre-ferred to Regression (4).

(32)

5 Conclusion

This paper assesses the possible impact of many legitimate reasons for price differ-ences of identical goods across countries. Based on the economic theories of the law of one price and purchasing power parity, variation in prices of selected H&M products were examined. The analysis was based on data collection, previous research and find-ings; finally, the empirical model was constructed and results were analyzed.

The main findings for price differences can be attributed to the variations in tariffs, GDP per capita, PPP, trade and total population per store within the European and American countries. Firstly, the empirical findings questioned the relationship between the dependent and the various independent variables, as none of the variables were sta-tistically significant. Later, with the reduction of the sample size based on market pene-tration, new results showed significant relationships for all the variables. GDP per capi-ta, PPP, trade volume and population per store were found to be directly proportional to the relative price. On the other hand, contradicting the expectations, tariffs were shown to have a negative effect on the relative average price of clothes, that can be explained by the negative relationship between GDP per capita and those tariffs. Furthermore, when introducing dummy variables for regions such as Asian Pacific, Middle Eastern and North African countries, it was found that they have significantly higher price for H&M clothes than European countries.

Since the products investigated are perfectly (and relatively cheaply) tradable goods, and the prices are easily accessed online, one would suggest that the prices around the world are set to be equal in order to eliminate arbitrage opportunities. On the other hand, it seems that even a company offering ‘fashion and quality at the best price’ with cost-consciousness in mind takes advantage of the fact that in some parts of the world their brand is still considered as unique.

5.1 Suggestions for Further Studies

Even though this study presents an indication of the determinants of price differences and their impact across countries, the research was highly limited due to the lack of available data for specific countries and previous case studies on this matter. In addi-tion, price data was collected only on one company for year, and the time horizons of the other variables differ, so the data on the same time horizon would produce more concrete and reliable results. Moreover, a study incorporating numerous countries into

(33)

consideration, and also applying truncated regression models would solve the problem of missing values. Furthermore, the level of penetration of a company in the selected markets could be further investigated within the sampled companies in addition to mark-up pricing, as it would give a good estimate of the demand and pricing patterns.

As Sweden is a high-income country, further studies can be conducted with a different benchmark, more preferably a middle-income country and a low-income country in or-der to reflect the diffences more accurately.

This is an important topic to be analyzed further, as nowadays, with the help of the developed consumer legislation and the increasing importance of electronic commerce, it is easy to compare the prices across countries and take the advantage of bargain with confidence.

(34)

Appendix

Table 1. Ad valorem tax (%) and relative average price deviations (SEK)

Country Average price deviations Ad valorem (%)

Australia 1.03 10,0 Austria 0.92 12,0 Bahrain 1.18 5,0 Belgium 0.92 10,0 Bulgaria 0.88 10,0 Canada 0.94 17,5 Chile 1.00 6,0 China 1.35 16,5 Croatia 0.92 10,0 Czech Republic 0.86 10,0 Denmark 1.02 10,0 Egypt 1.30 30,0 Estonia 0.92 10,0 Finland 0.92 10,0 France 0.92 10,0 Germany 0.92 10,0 Greece 0.92 10,0 Hong Kong 1.04 0,0 Hungary 0.85 10,0 Indonesia 1.00 14,1 Ireland 0.92 10,0 Israel 1.30 6,0 Italia 0.92 10,0 Jordan 1.51 5,0 Kuwait 1.15 5,0 Latvia 0.92 10,0 Lebanon 1.40 5,0 Luxenbourgh 0.92 10,0 Malaysia 1.09 0,0 Mexico 0.93 21,8 Morocco 1.22 25,0 Norway 1.13 10,7 Oman 3.66 5,0 Phillippines 1.13 15,0 Poland 0.89 10,0 Portugal 0.85 10,0 Qatar 1.27 5,0 Romania 0.86 10,0 Russia 0.91 5,0 Saudi Arabia 1.24 5,0 Serbia 0.86 13,5 Singapore 1.25 0,0 Slovakia 0.92 10,0 Spain 0.90 10,0 Sweden 1.00 10,0 Switzerland 1.33 - Taiwan 1.07 11,3 Thailand 1.19 30,6 The Netherlands 0.92 10,0 Turkey 0.77 12,0 UAE 1.26 5,0 UK 1.02 10,0 USA 1.02 11,4

(35)

Table 2. EFTA free trade agreements

Albania

Bosnia and Herzegovina Canada

Central American States (Costa Rica and Panama) Chile

Colombia Egypt

Gulf Cooperation Council (GCC) Hong Kong, China

Israel Jordan Korea, Republic of Lebanon Macedonia Mexico Montenegro Morocco Palestinian Authority Peru Serbia Singapore

Southern African Customs Union (SACU) Tunisia

Turkey Ukraine

Table 3. Jarque-Bera Test

Normality Test:

Jarque-Bera 13.5839

Probability 0.0011

Table 4. White’s Test

Heteroskedasticity Test:White

F-statistic 46.3336 Prob. F(14,23) 0.0000

(36)

Table 5. Ramsey’s RESET Test

Breusch-Godfrey Serial Correlation LM Test:

F-statistic 0.2887 Prob. F(1,32) 0.5948

Obs*R-squared 0.3398 Prob.Chi Square(1) 0.5600

Table 6. Variance Inflation Factors

Variance Inflation Factors: Coefficient Uncentered Centered

Variable Variance VIF VIF

Constant 0.0026 10.6169 NA

Relative Ad valorem 0.0018 8.7630 1.2323

Relative GDP per capita 0.0014 3.0855 1.1148

Relative Trade Volume 1.80E-05 1.4792 1.0812

Relative Population per Store 2.05E-08 1.4223 1.2297

Table 7. Correlation Matrix between variables

Correlation Matrix Relative Trade Volume Relative Ad valorem Relative GDP per capita Relative price Relative population per store

Relative Trade Volume 1.000000

Relative Ad valorem _0.231989 _1.000000

Relative GDP per capita 0.008245 -0.203143 1.000000

Relative price _0.298802 _-0.283868 _0.305857 _1.000000

Relative population per

(37)

References

Alessandria, G., & Kaboski, J. (2008). Why Are Goods So Cheap in Some Coun-tries? Business Review, Q2, 1-12. Retrieved February 28, 2015, from http://www.philadelphiafed.org/

Almås, I. (2012). International Income Inequality: Measuring PPP Bias by Esti-mating Engel Curves for Food. American Economic Review, 1093-1117.

Ardeni, P. (1989). Does the Law of One Price Really Hold for Commodity Prices?

American Journal of Agricultural Economics, 71(3), 661-669.

Batista, J., & Filho, G. (2010). Trade Costs and Deviations from the Law of One Price. American Journal of Agricultural Economics, 1011-1023.

Brunner, A. (2003). The Long-run effects of trade on income and income growth.

Working Papers, 1-38.

Causa, O., Araujo, S., Cavaciuti, A., Ruiz, N., & Smidova, Z. (2014). Economic Growth from the Household Perspective: GDP and Income Distribution Develop-ments Across OECD Countries. OECD Economics Department Working Papers,

1111, 1-83. Retrieved February 28, 2015, from

http://www.oecd- ilibrary.org/economics/economic-growth-from-the-household-perspective_5jz5m89dh0nt-en

Engel, C., & Rogers, J. (2001). Deviations from purchasing power parity: Causes and welfare costs. Journal of International Economics, 29-57.

European Commission | ECC Reports. (n.d.). Retrieved April 28, 2015, from

http://ec.europa.eu/consumers/solving_consumer_disputes/non-judicial_redress/ecc-net/reports/index_en.htm

European Commission | What is the Common Customs Tariff? (n.d.). Retrieved

April 28, 2015, from

http://ec.europa.eu/taxation_customs/customs/customs_duties/tariff_aspects/index_ en.htm

European Free Trade Association | Free Trade Agreements. (n.d.). Retrieved April 28, 2015, from http://www.efta.int/free-trade/free-trade-agreements

(38)

Eurostat - Statistics Explained. (n.d.). Retrieved April 1, 2015, from http://ec.europa.eu/eurostat/statistics-explained/index.php/Main_Page

Feenstra, R., & Kendall, J. (1997). Pass-through of exchange rates and purchasing power parity. Journal of International Economics, 237-261.

Financial Times Lexicon - The definitive dictionary of ... (n.d.). Retrieved March 28, 2015, from http://lexicon.ft.com/

García-Solanes, J., Sancho-Portero, F., & Torrejón-Flores, F. (2008). Beyond the Balassa–Samuelson effect in some new member states of the European Union.

Economic Systems, 17-32.

Gayawan, E., & Ipinyomi, R. (2009). A Comparison of Akaike, Schwarz and R Square Criteria for Model Selection Using Some Fertility Models. Australian

Jour-nal of Basic and Applied Sciences, 3(4), 3524-3530.

H&M. (n.d.). Retrieved February 28, 2015, from http://www.hm.com/

H&M | Annual reports. (n.d.). Retrieved April 12, 2015, from

http://about.hm.com/en/About/Investor-Relations/Financial-Reports/Annual-Reports.html#cm-menu

H&M | History. (n.d.). Retrieved April 15, 2015, from http://about.hm.com/en/About/facts-about-hm/people-and-history/history.html#cm-menu

International Trade Administration. (n.d.). Retrieved March 20, 2015, from http://otexa.trade.gov/msrpoint.htm

Isard, P. (1977). How Far Can We Push the ''Law of One Price”? The American

Economic Review, 67(5), 942-948.

Khouja, M., & Robbins, S. (2005). Optimal pricing and quantity of products with two offerings. European Journal of Operational Research, 530-544.

Knoema | World GDP Ranking 2015 | Data and Charts. (n.d.). Retrieved March 28, 2015, from http://knoema.com/nwnfkne/world-gdp-ranking-2015-data-and-charts

How and why do prices of identical products vary across countries?