Price dispersion and online markets maturity – case of Poland

Folia Oeconomica Acta Universitas Lodzensis ISSN 0208-6018 e-ISSN 2353-7663 www.czasopisma.uni.lodz.pl/foe/ 3(329) 2017 DOI: http://dx.doi.org/10.18778/0208-6018.329.01 Łukasz Konopielko Lazarski University, Faculty of Economics and Management, Department of Economics, [email protected] Diana Radkova Lazarski University, Faculty of Economics and Management, [email protected] Price dispersion and online markets maturity in Poland Abstract: This paper is dedicated to the issue of persistent price dispersion in real markets in the context of Polish online market development. The two channels of retail (online and oline) are observed and compared with regard to price variations determinants and their values comparison. Market eiciency comparison between both markets is also provided on the basis of theoretical background. The data collected from shops representing the two channels of retail (online and oline), for a number of commodities within the home appliance product category in Poland, has been used as a sample for the analysis. Generally, results for the Polish market are in line with similar tests performed for EU as a whole and conirm the maturity of Polish online market. Keywords: Price dispersion, Poland, market maturity, online markets, e-commerce, Internet. JEL: D40, L11, L81, C10 [7] 8 Łukasz Konopielko, Diana Radkova 1. Introduction Online markets in Eastern Europe are developing at a rapid rate. The average an‑ nual online sales growth since 2010 achieved 35.1% comparing to 18% average for the whole Europe (E‑commerce, 2015). Particularly in Poland, the percentage of individuals ordering or purchasing goods or services on the Internet for private use increased from 28.9% in 2010 to 34.2% in 2014 (GUS, 2014), while estimated share of online retail of goods in total retail of goods reached 5.4% in 2014. How‑ ever, the question remains: to what extent is this market already developed and what is its “quality”? This paper investigates the price variations in real market as a measure for market maturity, with respect to Poland. The issue of price dis‑ persion is relevant in both traditional and online markets. Price dispersion refers to deviation of prices from the average value. This problem concerns both demand and supply holders as well as the whole market equilibrium. As online markets become more mature, more and more empirical evidence appears of higher ei‑ ciency of online markets and lower price variation rate. In the case study presented in this paper, commodities from the home appliance category in Poland are con‑ sidered as the ground for the analysis: 468 price observations were obtained at the end of data gathering period. Both online and oline markets are compared with regard to price variations rates. The hypotheses are generated on the basis of theo‑ retical expectations and are classiied into two groups. The irst group concentrates on the measuring of price dispersion in both retail channels; the second group tests drivers of dispersion and their relationship with prices. The paper is built as follows: the irst part provides the theoretical framework of price dispersion and expectations regarding price dispersion presence in real markets. Empirical evidence of persistent price dispersion in both online and tra‑ ditional markets is presented based on literature review of previous studies; to sum up, a brief conclusion on price dispersion as a microeconomic phenomenon, which can serve as an estimator of market eiciency, is presented. The remaining part is dedicated to empirical analysis and evidence of price dispersion on the example of home appliance and consumer electronics in Poland. This section also compares achieved results to previous indings. A inal conclusion mentions limitations of the study and obstacles that have been encountered, together with recommendations for future research. 2. Price dispersion phenomenon The term “price dispersion” is relatively new in modern economy. Stigler was the irst to mention the phenomenon of price dispersion in his article “Economics of In‑ formation” (Stigler, 1961: 213). According to Stigler, the main sources of price var‑ FOE 3(329) 2017 www.czasopisma.uni.lodz.pl/foe/ Price dispersion and online markets maturity in Poland 9 iation appearing in real markets are the following: search costs by market agents, property of information to become outdated over time, changing identity of buy‑ ers and sellers, size of the market, and products’ heterogeneity. Price dispersion is thus deined as a variation of prices for homogeneous products across various sellers in a given market. Dispersion rate can serve as a determinant of purchase behaviour from the buyers’ perspective, since the variation of prices represents the presence or absence of alternative beneicial oferings. A low rate of dispersion in‑ dicates smaller probability of the less expensive goods’ purchase, since all the pric‑ es are close to the average and there are no extremes. However, when dispersion is high, buyers are more likely to put more efort into the search for the cheaper product and beneit from the purchase. From the sellers’ perspective, dispersion relects competitors’ pricing strategies. For the whole market, the rate of price dis‑ persion can show if the market is close to or far from full eiciency; the lower the dispersion rate, the closer the market is to a perfectly competitive model and the higher its eiciency. Thus investigation of the price dispersion level can be used as a benchmark for the level of a given market development. Insuicient information about product prices forces consumers to search among various sellers, spend time and efort, which means facing considerable costs; meanwhile, cost is a discouraging feature of any activity. If we consider the discussion of price dispersion in terms of prospects theory (Kahneman, Tversky, 1979), in real markets, people underestimate the possible beneits of product search and overestimate the costs. Therefore, byers settle for the irst listed price they ind and forget about marginal savings, assuming that in such a way they avoid poten‑ tial costs in the future (such as: time, transport, etc.) necessary to perform further search. Regardless of time loss, there is always a low of new uninformed sellers and buyers. As they enter the market, information and prices that have already been investigated become outdated and irrelevant. Obsolete information chang‑ es conditions of search costs and the of the whole market equilibrium and makes sellers incapable of maintaining a perfect correlation of prices. In this case, the buyers are not aware of time, eforts and costs they should devote to the search ei‑ ther, or that they should adapt to the new equilibrium. Hence, market conditions and market stability afect price variations. As the market grows, dispersion be‑ comes smaller, as irms that specialize in information collection arrive. Moreover, there are no purely homogeneous products across various sellers in real markets. Every seller is diferent; hence, the services and the commodities they ofer are considered unique. Thus, information is the key element when considering price dispersion. Even in terms of theoretical framework, price dispersion deinitely ex‑ ists and is persistent. The emergence of e‑commerce has enriched the economic theory with a lot of new issues for discussion. As far as the price dispersion on the online mar‑ kets is concerned, the theoretical framework claims that it is lower in compar‑ www.czasopisma.uni.lodz.pl/foe/ FOE 3(329) 2017 10 Łukasz Konopielko, Diana Radkova ison with traditional markets. The simplest logic behind this expectation is that increased availability of information and decreased search costs give a prior advan‑ tage to buyers to choose whatever seller they consider most appropriate, whereas sellers in these conditions seek to be price competitive; hence, no extreme markups are predicted to occur. Nevertheless, the results of empirical studies suggest var‑ ious results and indings. Relying on theoretical background, Bacos (1997) expected that prices in of‑ line stores should be higher for the same items than in online shops; the author argues that the search costs difer within these two channels of retail with regard to changes in information low and consumers’ price sensitivity. The appearance of online markets is crucial for the information low, since now consumers have to make less efort to obtain price information and product characteristics infor‑ mation from diferent sellers, as well as alternative ofers in the market for a giv‑ en good. Lesser efort for buyers relates to smaller search costs; comparing two channels of purchase, consumer search costs in online market are lower. Sell‑ ers know that buyers can compare prices from several retailers so that it makes no sense for them to apply high markups in order to be price competitive against the competition. Bailey (1998) was probably the irst to investigate price dispersion with‑ in the online retail channel. He took books, CDs and software items as prod‑ uct categories for comparison of prices between online and traditional markets (all the chosen products were considered fully homogenous). The prices were corresponding to each other in both markets. The point of his analysis was that search costs and the level of competition in the market were the determinants of dispersion rate. During the period of data (prices) collection in 1996–1997, it was observed that online prices were higher than the prices of the same prod‑ ucts in traditional market. As Bailey explained, his irst indings may conlict with the hypothetical expectations and point to higher search costs on immature online markets with low numbers of sellers. Little competition among e‑retailers and consumers’ search intensity was expected to increase over the time, so that more mature market would let empirically reairm the theory and achieve low‑ er dispersion rates. Later, a similar study on books and CDs by Brynjolfsson and Smith (2000), also seeking to compare dispersion in the online and traditional mar‑ kets, emphasizes that e‑commerce brought any market, for whatever goods and services, a possibility to have lower barriers to entry, due to alleviated infrastructure, which is simply a website page in a browser. The advantag‑ es of an online market such as lower variety of choices within the menu and absence of infrastructure issues of traditional markets contribute to a lower rate of dispersion among e‑retailers. Moreover, the authors have documented a new tendency, according to which sellers in online markets tend to change FOE 3(329) 2017 www.czasopisma.uni.lodz.pl/foe/ Price dispersion and online markets maturity in Poland 11 prices more often with lesser difference than traditional retailers. This find‑ ing can also serve as a contributor to a more competitive market and expected lower price dispersion. The empirical part dedicated to the measure of posted prices by a proxy for market share accepts the hypothesis of lower dispersion favouring online markets. As for online electronics market, Baye, Morgan and Sholten (2001) found that the number of companies posting prices tends to vary with the rate of price dis‑ persion. An interesting case of price dispersion has been described through the analysis of online auctions by Sun and Hsu (2007), which detects another factor inluencing dispersion rate. The paper concludes that sellers’ reputation generally has the most signiicant impact on auction prices yet such auction features as: du‑ ration of bargain, the opening bid and Buy‑it‑now option (BIN) have a signiicant impact on the online auction prices. In later study regarding prices in online auc‑ tions, Pate (2006) gained the results that also conirm sellers’ reputation contrib‑ uting to changes in auction prices. Shankar, Pan and Ratchford (2002) attempt to determine the drivers of price dispersion over time as well as to discover if these drivers can change. Empirical analysis showed that the provision of product information, shipping, handling and reliability were the main contributors to dispersion among e‑retailers characteris‑ tics. As for market characteristics, the number of sellers was negatively correlated with the rate of dispersion; nevertheless, the authors emphasize that empirically it works only at a diminishing rate over time, and in time this efect is likely to be insigniicant. Even when considering approximately similar reasons for price dis‑ persion such as search costs, product characteristics, channel of retail, shopping convenience, brand loyalty, or level of competition, as long as e‑commerce market matures, the majority of studies suggest that price dispersion streams at diminish‑ ing rate favoring the online channel. Bailey (1998), Clemons, Hann and Hitt (2002), Erevelles, Rolland and Srinivasan (2001), Lee and Gosain (2002) observed higher prices on the internet and higher price dispersion in online markets in comparison with the traditional ones. All these studies use diferent approaches as well as dif‑ ferent subjects of analysis, but still their empirical results ind higher price varia‑ tions among e‑retailers. The only similarity observed is for the time period from 1998 to 2002. This was a very early period of e‑commerce development, therefore market maturity has to be considered here again. Nevertheless, early evidence of lower price variation in online markets also ex‑ ists. For instance, Brynjolfsson and Smith (2000), Morton, Zettelmeyer and Silva‑ ‑Risso (2001), Tang and Xing (2001) have documented lower dispersion rates in on‑ line markets. The study by Ancarani and Shankar (2004) discovers price dispersion for books and compact discs between three retail channels: Internet only, tradi‑ tional, and multichannel. Cooper (2006) has chosen an unusual subject of analy‑ sis when comparing dispersion rates in traditional and online markets. Consistent www.czasopisma.uni.lodz.pl/foe/ FOE 3(329) 2017 12 Łukasz Konopielko, Diana Radkova with previous papers, he inds evidence of lower dispersion rate in the market for disposable contact lenses. The idea behind it was that traditional shops set pric‑ es assuming that buyers are not aware of the opportunity of buying lenses online. As a result of facilitated search costs, online price dispersion was revealed to be lower than oline with nearly 11% diference. Basically, the results showed that buyers seemed to be unaware of their options. The paper by Rosello and Riera (2012) compares online and oline price levels and variation rates for tourism expenditures among emerging online tour agencies and existing traditional travel operators. Despite the fact that results have shown a persistent online price dispersion, it has been observed to be considerably lower among new e‑retailers. Table 1. Summary of studies that compare price dispersion rates between online and traditional markets Authors Bailey Clemons, Hann and Hitt Brynjolfsson and Smith Erevelles, Rolland and Srinivasan Morton, Zettelmeyer and Risso Tang and Xing Clay, Krishnan, Wolf and Fernandes Lee and Gosain Brown and Goolsbee Ancarani and Shankar Cooper Rosello and Riera Sengupta and Wiggins Duch‑Brown and Martens Year Subject investigated 1998 1998 2000 2001 2001 2001 2002 2002 2002 2004 2006 2012 2012 2014 Books, CDs, Software Airline tickets Books and CDs Vitamins Cars DVDS Books CDs Insurance sevices Books and CDs Contact lenses Travel agency services Airline tickets Consumer goods Price dispersion is higher Online Online Oline Online Oline Oline Online Online Oline Oline Oline Oline Oline Oline Source: own research based on the literature review With regard to studies on the issue of price dispersion on Eastern European markets, only one paper exists, by Szopiński and Nowacki (2014), which investi‑ gates price variation rates between domestic and foreign light carriers in Poland, on the basis of airline tickets’ prices, posted on the internet for the most popular connections. The paper reported persistent price dispersion in the market for airline tickets varying from 5.63% to 8.07% between Polish carriers and 7.70% to 20.3% between carriers from destination countries. The highest diferential between ab‑ solute value of standard deviation among the two diferent carriers has accounted for 13.885%. The conclusion suggests that dispersion is much lower among Pol‑ ish carriers than among the corresponding foreign carriers for the same routes. FOE 3(329) 2017 www.czasopisma.uni.lodz.pl/foe/ Price dispersion and online markets maturity in Poland 13 The authors do not mention any reasons that would account for such a big gap be‑ tween prices. When reviewing the literature on price dispersion chronologically, more re‑ cent papers contradict the indings of research performed at the time when e‑com‑ merce was still in its infancy. For instance, earlier papers describe higher online dispersion; whereas the results of more recent papers are in accordance with the initial theoretical expectations regarding lower online dispersion. In more mature online markets, market players adapt to the conditions of perfect information and very low search costs, which makes the suppliers aim at being price‑competitive and still survive on the market, so that sellers keep prices on approximately equal level without considerable markups. This notion can be treated as one of the online markets’ maturity characteristics: mature markets experience a decrease in price variations rate. 3. Empirical evidence from Poland The main objective of this paper is to investigate price dispersion in real mar‑ kets as well as to apply basic theoretical expectations to practical situations, and to compare the results to previous studies. Speciically, the market of household appliance electronics in Poland is observed. Household appliance goods are di‑ vided into three subgroups: major home appliances, or white goods; small home appliances; consumer electronics, or brown goods. The point is to use a dataset consisting of prices for chosen types of electronic goods as the main subject of the analysis, in order to detect the presence of dispersions and highlight external and internal factors that may inluence prices. By the term “external”, market factors are basically implied, whereas internal determinant may underlie in the speciica‑ tion of the commodities. Table 2. Hypotheses classiication Price dispersion measurements and comparison H0 H1 Sources and inluence Price dispersion is present in both online Sellers’ reputation impacts price varia‑ H2 and traditional markets. tion rate. The rate of price dispersion is lower in on‑ line shops than in corresponding oline H3 Product reputation afects price variation. ones. Number of sellers is negatively correlated H4 to price dispersion rate. Source: own research www.czasopisma.uni.lodz.pl/foe/ FOE 3(329) 2017 14 Łukasz Konopielko, Diana Radkova Considering the theoretical background provided in the previous section, it is most obvious and relevant to achieve empirical evidence of lower dispersion rate in online markets and to deine its main determinants. The set of hypotheses is divid‑ ed into dispersion rate measurements and its comparison (H0, H1) and checking for sources of dispersion (H2, H3, H4). Table 2 represents hypothesis classiication. As long as hypotheses are classiied with regard to the aims of the analysis, two diferent methodological approaches will be applied for each group of hy‑ potheses. Methodological approaches in studies for price dispersion vary, however the coeicient of variation has been used by Ghose and Yao (2011), Sengupta and Wiggins (2012), Duch‑Brown and Martins (2014), Szopiński and Nowacki (2014), therefore, probably, it is the best known estimation method. In this analysis of price dispersion, relative standard deviation is considered a primary method of meas‑ urement. All the hypotheses from the group one will be tested with descriptive statistics calculations. When calculating relative standard deviation, actual pric‑ es will be put in a dataset and analysed with software. With respect to the hy‑ potheses from group two a common tool of investigating sources of price varia‑ tion – econometric modelling – has been observed on the basis of literature review. Shankar, Pan, and Ratchford (2002), Cooper (2006), Leong (2013), Sengupta and Wiggins (2012), Duch‑Brown and Martens (2014) used linear regression models as a methodology approach of detecting factors inluencing price variation. Fol‑ lowing the study by Shankar, Pan and Ratchford (2002), cluster analysis of the re‑ gression will be used as a helping feature. The following log‑linear equation has been generated: Model 1 ln(Pj, k) = α0 + α1 · shr + α2 · prr + α3 · noshj + α4 · γON,OFF + α5 · γovens + α6 · γreig + k j α7 · γwash + α8 · γdish + α9 · γsmart. mach wash Price is used as a dependent variable of the model, since the primary interest lies in determining which factors afect it and to what degree. Price has been put into logarithmic terms in order to observe percentage change in dependent variable with each unit change in independent (explanatory) variables. Sellers’ reputation, product reputation, level of competition, dummies for product characteristics and channel of retail have been chosen as explanatory variables. Dummies for product characteristics and channel of retail are plugged in on purpose: for the irst group of hypotheses (in order to use “sort by” functions), and for the second group of hypotheses (with a view to check if product category may impact price variation). FOE 3(329) 2017 www.czasopisma.uni.lodz.pl/foe/ Price dispersion and online markets maturity in Poland 15 Table 3. Variables description Variable name Price for a good j in the shop k Pj, k sh_r(k) prr Explanation Seller’s reputation measured as a rate of shop k Product reputation measured as a rate of product j j noshj Number of sellers that ofer a good j as a measure of competition level γON,OFF Dummy that indicates whether the shop operates on the online market or the tradi‑ tional one. 1 – indicates that the shop is online 0 – indicates that the shop is in‑store Dummy that indicates whether commodities refer to a speciic group of product cat‑ egories (7) such as refrigerators, washing machines, dish washers, TV sets, smart‑ phones, small home appliances (indicated as sma), and ovens. 1 – indicates that the good is in its category 0 – otherwise Regression coeicients Error term that measures unobservable factors and model bias γn αn ε Source: own research A primary source of data has been used in this study in order to gain empirical results for the set of hypotheses; all the data has been collected manually. All in all, 396 price observations have been collected for the analysis. These are the observa‑ tions for 33 product categories from 8 online and 4 physical shops in Poland. Avail‑ ability of the commodity from all the stores has been a mandatory requirement. Initial list of commodities comprised 72 goods; subsequently, the list was iltered out in order to gain cross section data of 12 shops. All the price observations have been collected in domestic currency, i.e. Polish Zloty at the end of the irst quar‑ ter of 2015. Data gathering process consisted of several steps. The irst one was the choice of shops and creation of the list of commodities. Strictly, only retail shops and commodities referring to home appliances match the ield of study. Initially, the number of units was the same in each of the three groups of commodities. Howev‑ er, as soon as the ilter sorted the results, among the leftover 33 commodities suiting to cross‑section data distribution, 11 were form a group of consumer electronics, or brown goods, 4 refered to small home appliances, and the remaining 18 were ma‑ jor home appliances, or white goods. The prices from in‑stores like Komputronik, Saturn, Media Markt, and RTV Euro AGD were collected manually with the help of shop assistants, shop catalogue or a direct phone call to the shop. All the prices have been added into inal list of observations. In case of the online shops, things were much easier due to the existence of price comparison websites; in this case, Ce‑ neo.pl has been used as a source of data (Ceneo.pl, 2015). Price collection from price www.czasopisma.uni.lodz.pl/foe/ FOE 3(329) 2017 16 Łukasz Konopielko, Diana Radkova comparison websites is a common practice when investigating price dispersion rates. For instance, Shankar, Pan and Ratchford (2002) used Bizrate.com, Baye, Morgan and Shulton (2001) referred to Shopping.com in their studies. Following their exam‑ ple, Polish equivalent of such websites has been found and used as data source. Sell‑ ers reputation (shop rate), product reputation (product rate) and product reputation (product rate) values have also been collected from ceneo.pl platform. 3.1. Empirical evidence of the hypotheses of the irst group Table 4 represents inal results of price dispersion rates in both markets. Table 4. Descriptive statistics of price in online and traditional markets Descriptive statistics Mean Medium Minimum Maximum Range Standard Deviation Relative standard deviation, % Online (264) 1449.82 1299.495 61 4022.22 3961.22 872.468 60.1775 Oline (132) 1450.6 1309 67.9 4049 3981.1 865.468 59.6628 Lower value is: Online Online Online Online Online Oline Oline Source: own calculations This step allows us to conclude that zero hypothesis is accepted since relative standard deviation (proxy to measure dispersion rate) does not equal zero in ei‑ ther of the retail channels: 60.2% for online markets and 59.7% for the traditional ones (Table 4). Generally speaking, dispersion rate is really high if compared, for instance with 5.63% and 20.3% of price variations in airline tickets sales in the study by Szopiński and Nowacki (2014). When discussing hypothesis which picks out the market with a lower rate of dispersion, the last column in table 4, shows that price variation is higher in the online market. However, it is worth mention‑ ing that the irst run of descriptive statistics does not account for diferent product categories and their heterogeneity; so, in order to gain more reliable and accurate results, the problem of product classiication should be solved. The second run is going to distinguish prices by product category with the use of dummies indicating retail channel and commodity characteristics. Table 5 shows the rates achieved from the second run of descriptive statistics. After commodities characteristics were held ixed, ive out of seven product categories indicated a lower relative deviation rate than in corresponding oline market; except for goods from the smartphones and washing machines category (Table 5). Yet, the lowest rate of price variation was found at 18.10% for washing FOE 3(329) 2017 www.czasopisma.uni.lodz.pl/foe/ Price dispersion and online markets maturity in Poland 17 machines and the highest rate of dispersion was at 77.5% for small home applianc‑ es category (both in physical stores). As long as majority of product categories fa‑ vour lower rate of dispersion in online markets, the hypothesis one is accepted. Table 5. Descriptive statistics of price in online and traditional markets by product category Descriptive statistics Mean Medium Range Relative standard deviation, % Smartphones Mean Medium Range Relative standard deviation, % Mean Medium Range Relative standard deviation, % Mean Medium Range Relative standard deviation, % Mean Medium Range Relative standard deviation, % Mean Medium Range Relative standard deviation, % Mean Medium Range Relative standard deviation, % Online (264) Oline (132) Small home appliances 488.3797 530.7975 424 487.7 1138 1131.1 76.20433 77.50612 1278.128 1231.25 2343.98 58.38473 Tv‑sets 2267.161 1661.4 2793.23 50.75589 Refrigirators 2130.8 2187.4 2333.09 37.09722 Dish washers 1299.669 1322 660.5601 21.809 Washing machines 1180.493 1259.96 927 19.81226 Ovens 1515.514 1399.05 2123.05 52.26113 Lower value is: Online Online Oline Online 1247.25 1198.5 1960 56.82995 Oline Oline Oline Oline 2278.74 1691.89 2810 51.23243 Online Online Online Online 2139.736 2177.99 2408.02 38.15863 Online Oline Online Online 1299.291 1319 694 22.11982 Oline Oline Online Online 1189.498 1240.65 706.16 18.0838 Online Oline Oline Oline 1509.58 1359.05 2070 52.91494 Oline Oline Oline Online Note: the number in brackets in columns` names indicating the channel of retail shows the number of price observations Source: own calculations www.czasopisma.uni.lodz.pl/foe/ FOE 3(329) 2017 18 Łukasz Konopielko, Diana Radkova 3.2. Empirical evidence of the hypotheses of the second group The second group of hypotheses aims to determine drivers of price dispersion in both markets. In order to accept or reject all the hypotheses from the second group, it would be enough to run OLS regression and analyze general model reli‑ ability and statistical signiicance of explanatory variables and check coeicient signs in case of hypothesis four. The inal equation includes price in logarith‑ mic terms as a dependent variable, whereas product and sellers` reputation, lev‑ el of competition, dummy indicating channel of retail and small home appliance commodities serve as independent variables: ln(Pj, k) = α0 + α1 · shr + α2 · prr + α3 · noshj + α4 · γON,OFF + α10 · γsma + ε. k j Table 6. Presentation of OLS regression output Number of observations = 396 F(5, 32) = 6.20 Prob > F = 0.0004 R‑squared = 0.4330 lnprice Coeicient pr_r –0.4681863 nosh –0.0017269 sh_r –0.1728411 onof –0.0188441 Sma –1.33511 Cons 9.475434 t‑statistic –2.17 –0.15 –4.16 –2.86 –2.83 8.33 P>|t| 0.037 0.881 0.000 0.007 0.008 0.000 Source: own calculations All the p‑values of coeicients are treated as statistically signiicant except for the level of competition variable which has not been excluded earlier, since the hypothesis four focuses on its relationship with dependent variable. Product and sellers’ reputation variables are treated as statistically signiicant. Hence, the hy‑ potheses two and three are accepted and both product and shop rates inluence price variations. Negative signs of product and sellers’ reputation and the level of com‑ petition variables show that the more popular the product and the retailer, and the higher the number of sellers who ofer a commodity, the lower the price dispersion, since there is a negative, linear relationship between these explanatory variables and price. Hypothesis four is accepted which can be paraphrased as follows: as the number of sellers, or the rate of competition grows, price variation diminishes. To sum up, all the hypotheses from the second group have been accepted in conformity with expectations and empirical studies dedicated to similar issues that have been mentioned earlier. Econometric analysis reairmed anticipation FOE 3(329) 2017 www.czasopisma.uni.lodz.pl/foe/ Price dispersion and online markets maturity in Poland 19 in regard to external factors that may inluence price variations, such as: product, seller’s reputations, retail channel and rate of competition as well as internal fac‑ tors which underlie commodity speciication assignment in case of small home appliance goods. When comparing exact values of price dispersion rate, Duch‑Brown and Mar‑ tens’ (2014) paper can be taken as a reference for collation, since their analysis has been undertaken? In the EU and home appliance goods were the subject of inves‑ tigation. Not all the product categories from this study can be compared to theirs, but for TVs, refrigerators, ovens and washing machines categories coeicients of variations can be collated. Table 7. Relative standard deviation values comparison for chosen product categories, % Product category TVs Refrigerators Washing machines Ovens This paper 51.23* 50.76 (0.48) 38.16* 37.10 (1.06) 18.08* 19.81 (1.73) 52.91* 52.26 (0.65) Duch‑Brown and Martens (2014) 60.80* 59.80 (1.00) 46.40* 47.90 (1.50) 34.00* 35.10 (1.10) 83.00* 82.70 (0.30) Diferentials in studies 9.57* 9.04 8.24* 10.80 15.92* 15.29 30.09* 30.44 Note: A star next to some values indicates that this value refers to physical stores. The number in brackets is a module diference between the two values in one cell. Source: own calculation and calculation from the report by Duch-Brown and Martens (2014) In summary, the results for TVs, washing machines and ovens categories col‑ late identically and prefer similar channels of retail. However, in case of refriger‑ ators, there is a discrepancy. Moreover, all price diferentials between the two pa‑ pers are very small: the highest diferential accounts for 0.63% diference in case of washing machines and 0.35% is the lowest for ovens. 4. Conclusion In conclusion, empirical price dispersion rate exists in both markets. Empirical ev‑ idence on the example of home appliance and consumer electronics commodities in Poland proves that price dispersion is lower in online markets. Results gained www.czasopisma.uni.lodz.pl/foe/ FOE 3(329) 2017 20 Łukasz Konopielko, Diana Radkova in this study are consistent with previous studies on the issue of price variation. Nevertheless, relatively small sample size leaves space for improvement and fur‑ ther investigation could be suggested with increased amount of information. If considering probable drivers of price dispersion in both physical and on‑ line markets, the results of econometric analysis suggest that product and sellers’ reputation, commodity’s category and a channel of retail are perceived as drivers of price variations, as it has been expected in the section about hypotheses settle‑ ment. As it has been mentioned in theoretical framework review, online markets’ maturity also has an impact on the level of price variations. In line with theoret‑ ical expectations and previous studies, this research of online markets in Poland brings successful evidence of a mature online market. References Ancarani F., Shankar V . (2004), Price Levels and Price Dispersion Within and Across Multiple Retailer Types: Further Evidence and Extension, “Journal of the Academy of Marketing Sci‑ ence”, no. 32, pp. 176–187. Bailey J.P. (1998), Electronic commerce: Prices and consumer issues for three products: Books, Compact Discs, and Software, “Organisation for Economic Co‑Operation and Development”, vol. 4, Paris. Bakos J.Y. (1997), Reducing buyer search costs: Implications for electronic market places, “Man‑ agement Science”, no. 43, pp. 1676–1692. Baye M.R., Morgan J., Scholten P. (2001), Price Dispersion in the Small and in the Large: Evidence from an Internet Comparison Site, “The Journal of Industrial Economics”, no. 52, pp. 463–496. Brown J.R., Goolsbee A. (2002), Does the Internet Make Markets More Competitive? Evidence from the Life Insurance Industry, “Journal of Political Economy”, vol. 110, no. 3, pp. 481–507. Brynjolfsson E., Smith M.D. (2000), Frictionless Commerce? A Comparison of Internet and Con‑ ventional Retailers, “Management Science”, no. 46, pp. 563–585. Clay K., Krisham R., Wolf E., Fernandes D. (2002), Retail strategies on the web: Price and non‑price competition in the online book industry, “The Journal of Industrial Economics”, no. 50(3), pp. 351–367. Clemons E., Hann I., Hitt L. (2002), Price Dispersion and Diferentiation in Online Travel: An Em‑ pirical Investigation, “Management Science”, no. 48, pp. 543–549. Cooper J. (2006), Prices and Price Dispersion in Online and Oline Markets for Contact Lenses, “Federal Trade Commission Bureau of Economics Working Paper”, no. 283. Duch‑Brown N., Martens B. (2014), Consumer Beneits from the EU Digital Single Market: Evi‑ dence from Household Appliances Markets, “Digital Economy Working Paper” no. 3. E‑commerce (2015), Eastern European B2C E‑commerce Report 2015, Ecommerce Foundation, Brussels. Erevelles S., Rolland E., Srinivasan S. (2001), Are Prices Really Lower on The Internet?: An Ana‑ lysis of The Vitamin Industry, “Working Paper, University of California”. Ghose A., Yao Y. (2011), Using transaction prices to re‑examine price dispersion in electronic markets, “Information Systems Research”, no. 22, pp. 269–288. GUS (2014), Społeczeństwo informacyjne w Polsce. Wyniki badań statystycznych z lat 2010– 2014 [Information Society in Poland. Results of statistical research over 2010–2014 peri‑ od], Szczecin. FOE 3(329) 2017 www.czasopisma.uni.lodz.pl/foe/ Price dispersion and online markets maturity in Poland 21 Kahneman D., Tversky A. (1979), Prospect Theory: An Analysis of Decision Under Risk, “Econo‑ metrica”, no. 47, pp. 263–291. Lee Z., Gosain S. (2002), A Longitudinal Price Comparison for Music CDs in Electronics and Brick‑and‑Mortar Markets: Pricing Strategies in Emergent Electronic Commerce, “Journal of Business Strategies”, no. 19, pp. 55–71. Leong J. (2013), Price Dispersion in Online Grocery Stores: An Analysis of the UK Market, “At‑ lantic Economic Journal”, no. 41(2), pp. 185–187. Morton F.S., Zettelmeyer F., Silva‑Risso J. (2001), Internet Car Retailing, “Journal of Industrial Economics”, no. 49, pp. 501–519. Nellis J., Parker D. (2002), Principles of business economics, Pearson Education Ltd, New York. Pan X., Rachford B.T., Shankar V . (2004), Price dispersion on the internet: A review and directions for future research, “Journal of Interactive Marketing”, no. 18, pp. 116–135. Pate J. (2006), Seller reputation as a determinant of price in online auction: Theory and evidence from gift card sales, “Working Paper”, Loyola Marymount University, Los Angeles. Rosello J., Riera A. (2012), Pricing European Package Tours: The Impact of New Distribution Channels and Low‑Cost Airlines, “Tourism Economics”, no. 18, pp. 265–279. Sengupta A., Wiggins S. (2012), Comparing Price Dispersion on and of the Internet Using Air‑ line Transaction Data, „Review of Network Economics”, no. 11, doi:10.1515/1446–9022.1244. Shankar V., Pan X., Ratchford B.T. (2002), Do Drivers of Online Price Dispersion Change as On‑ line Markets Grow?, “Working Paper”, University of Maryland, College Park. Stigler G. (1961), The Economics of Information, “Journal of Political Economy”, no. 69, pp. 213–225. Sun C.H., Hsu M.F. (2007), The Determinants of Price in Online Auctions: More Evidence from Quantile Regression, “Working Paper 07–18 Department of Economics”. Szopiński T., Nowacki R. (2014), Plane Ticket Price Dispersion in the Online Selling System in Po‑ land, „Contemporary Economics”, no. 8, pp. 207–218. Tang F., Xing X . (2001), Will the growth of multi‑channel retailing diminish the pricing eiciency of the web?, “Journal of Retailing”, no. 77, pp. 319–333. Dyspersja cenowa i dojrzałość rynków online w Polsce Streszczenie: Artykuł dotyczy zjawiska trwałej dyspersji cenowej w kontekście rozwoju rynków online w Polsce. Obserwacja dwóch kanałów sprzedaży (online i oline) ma na celu dokonanie porównania determinantów zmienności cenowej i ich kompozycji. Bazując na przesłankach teoretycznych, dokonano również porównania efektywności obu rynków. Dane do analizy zostały zabrane w sieciach dystrybucji dóbr AGD zarówno online, jak i oline. Rezultaty są zbliżone do podobnych, uzyskiwanych na innych rynkach Unii Europejskiej i wskazują na dojrzałość polskiego rynku sprzedaży online. Słowa kluczowe: dyspersja cenowa, Polska, dojrzałość rynku, rynki online, e-handel, Internet JEL: D40, L11, L81, C10 © by the author, licensee Łódź University – Łódź University Press, Łódź, Poland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution license CC‑BY (http://creativecommons.org/licenses/by/3.0/) Received: 2016‑06‑09; veriied: 2016‑08‑30. Accepted: 2017‑07‑31 www.czasopisma.uni.lodz.pl/foe/ FOE 3(329) 2017