میزان پویایی قیمت گذاری آنلاین در خرده فروشی اینترنتی : مطالعه موردی بازار DVD
|کد مقاله||سال انتشار||تعداد صفحات مقاله انگلیسی||ترجمه فارسی|
|1854||2011||10 صفحه PDF||33 صفحه WORD|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Electronic Commerce Research and Applications, Volume 10, Issue 2, March–April 2011, Pages 227–236
جمع آوری داده ها و خلاصه آماری
جمع آوری داده ها
رقابت بین شرکتهای اینترنتی و خرده فروشان چند کانالی
رقابت در میان شرکتهای اینترنتی و در میان خرده فروشان چند کانالی
مدل های اقتصادسنجی
ویژگی های داده های جمع آوری شده
مدل ضریب رگرسیون تصادفی
ویژگی های مدل رگرسیون ضریب تصادفی
تجزیه و تحلیل سطوح قیمت
بحث و نتیجه گیری
The explosive growth of Internet retailing offers an excellent opportunity to collect online prices at a disaggregate (e.g., individual store and/or individual product) level over time and to investigate the evolution of Internet markets. In this paper, we generalize the results obtained in existing static analyses and develop two random coefficient regression models to capture the dynamics of prices in the US online DVD market. On the basis of the models, we test hypotheses to compare the rates of change in price levels and in price dispersion at both pure dotcoms and online branches of multichannel retailers in the DVD market. The results, based on the analysis of 6759 price quotes over a 12-month period, suggest that multichannel retailers effectively differentiated themselves from pure dotcoms on nonprice dimensions so that they charged higher prices and maintained the difference in price levels throughout the time period of the study. Head-to-head price competition within pure dotcoms tended to be more severe. Our results also suggest that there is a sign of maturity in the current US online DVD market.
Although in the early years of Internet retailing it was widely predicted that online marketing would lead to frictionless e-commerce (Alba et al., 1997 and Bakos, 1997), a considerable number of recent studies have overwhelmingly shown that this is not true (see e.g., Lal and Sarvary, 1999, Brynjolfsson and Smith, 2000 and Pan et al., 2004). This study shifts the research focus and by investigating pricing behavior aims to gain a better understanding of how different types of retailer compete with each other in online markets. The issue of online pricing is of particular importance in the online marketing research. This is because pure dotcoms tend to differentiate themselves from other types of retailer via flexibly pricing their products. In addition, the competition among dotcoms also tends to be on the price dimension. Such competition leads to substantial price dispersion in the Internet markets. It is thus crucial for researchers to understand the characteristics of online pricing behavior and how it evolves over time. The results on this research issue also have important managerial implications. Most of the earlier empirical studies performed a static analysis where price competition was measured in terms of price levels and price dispersion. From the perspective of marketing research, both price levels and price dispersion are summaries of the price distribution in a market that reflects how retailers interact with each other. In these empirical studies price levels and price dispersion were compared between bricks-and-mortar (traditional) and online retailers (Bailey, 1998, Brynjolfsson and Smith, 2000, Erevelles et al., 2001 and Clay et al., 2002). Further studies focused on comparisons of various retailing channels. They compared pure Internet retailers (hereafter dotcoms) and online branches of multichannel (hereafter multichannel) retailers (Tang and Xing 2001); or traditional retailers, dotcoms, and multichannel retailers (Ancarani and Shankar 2004). These studies resulted in some interesting findings that suggest substantial differences in pricing behavior among different retailing channels. It has also been recognized that the stage of development of Internet retailing has a substantial influence on the pricing behavior of retailers. In the early stage, for instance, online retailers priced products at a higher-level than traditional retailers (Bailey, 1998 and Erevelles et al., 2001). As Internet markets developed in the early years of this century, online retailers substantially lowered their prices. During the transition period there was a mixture of findings, some of which contradicted each other. For example, Clay et al. (2002) did not find any significant difference in prices between online retailers and traditional retailers, whereas Brynjolfsson and Smith (2000) compared prices of CDs and books and found that online retailers had a lower price level than traditional retailers. There were also conflicting results on price dispersion. See Pan et al. (2004) for a comprehensive review. Internet markets are now more mature. It is thus of interest to investigate which of the earlier findings on Internet retailing can be generalized to the current online markets. In addition, since the majority of the existing researches were carried out at a fixed time-point, it is of interest to investigate which of the findings in these static analyses can be generalized to a longer time period so that the evolution of online prices over time can be investigated. The emergence of Internet data sources offers an impetus to the development of dynamic models that capture price dynamics (Dekimpe and Hanssens, 2000 and Pauwels et al., 2004). Consequently, recent studies on online pricing have used more sophisticated dynamic approaches. As Pauwels et al. (2004) have pointed out, however, few existing studies in marketing research recognize that the neglect of heterogeneity across the entities over which the data are averaged is a serious issue in dynamic modeling. For instance, in the recent analysis in Xing et al. (2006), cross-sectional heterogeneity was absent and the correlation of the prices posted at different retailers for the same product (e.g., a particular DVD title) during the same time period was ignored. Statistically, when aggregation bias is not addressed properly, it may result in parameter estimates being inconsistent, inefficient, and/or biased (Pauwels et al. 2004). This paper incorporates a sophisticated statistical technique to address these econometric issues. On the basis of our models, we focus on the pricing dynamics in online market evolution and investigate how different types of retailer compete with each other in an online market, the US DVD market. The US DVD market is chosen for several reasons. First, it is generally considered that DVDs are relatively homogeneous goods and thus likely to experience strong price competition given the characteristics of Internet channels (see e.g., Bakos, 1997, Lal and Sarvary, 1999, Brynjolfsson and Smith, 2000, Harrington, 2001, Tang and Xing, 2001, Iyer and Pazgal, 2003 and Xing et al., 2006). Secondly, there is a rich literature on the US online DVD market so it is easy to compare and contrast the findings of this study with other results, and in particular to compare the current price dynamics with those presented in Xing et al. (2006). In addition, it is more straightforward to compare DVDs because they are relatively homogeneous. For instance, prices of identical DVDs at different retailers can be compared directly. This is not the case for goods such as clothes, shoes, and electronics where there are many styles and/or models, and similar products may differ from each other to a considerable extent. Finally, the US online DVD market has a long history of Internet retailing and is likely to be more mature than other markets. The existing static analyses have revealed some interesting results on online marketing. Tang and Xing (2001) found that prices at dotcoms were significantly lower than prices at multichannel retailers. In addition, the corresponding price dispersion was much lower among dotcoms than among multichannel retailers. Contrary to Tang and Xing, 2001 and Pan et al., 2003 found that multichannel retailers generally had smaller price dispersion than did dotcoms. Ancarani and Shankar (2004) argued that multichannel retailers can combine the benefits of online shopping with physical inspection, pickup, and return of merchandise via support from their offline stores. In their static analysis they suggested that multichannel retailers may effectively differentiate themselves from dotcoms on nonprice dimensions and charge higher prices. Recently Xing et al. (2006) have investigated the dynamics of online prices in the US DVD market. On the basis of the online price data in the US DVD market collected during years 2000–2001, they have found that multichannel retailers charge higher prices than dotcoms and prices go up with time for both multichannel and dotcoms retailers. In addition, prices of dotcoms go up faster than those of multichannel retailers. In this paper we shall investigate which of these earlier findings can be generalized to the current online DVD market and can be generalized from a given time-point to a longer time period. More importantly, if there exists a difference in price levels between different types of retailer at a given time-point, we shall investigate whether the difference is maintained across the time period. To reveal the competitive pricing behavior of retailers, two dynamic models will be built at an individual product level, one model for price levels and the other for price dispersion. The nature of the data collected in this study raises several challenging issues for dynamic modeling, including extremely high dimensionality, and cross-sectional heterogeneity and the associated random effects. As indicated in Dekimpe and Hanssens, 2000 and Pauwels et al., 2004, it is difficult to address these issues in the framework of the widely used VARX approach. Hence, in this paper we shall consider an alternative approach, random coefficient regression models, to analyze pricing dynamics at an individual product level where the issues of time correlation and cross-sectional heterogeneity can be easily dealt with. We can also link marketing characteristics directly to the rate of change in price levels and in price dispersion so that the research issues of interest can be investigated. The next section is devoted to data collection and summary statistics. We then develop our main research questions. Then we build econometric models and test the formulated hypotheses. Finally we summarize the main results and discuss the managerial implications.
نتیجه گیری انگلیسی
In this study, we have investigated online pricing dynamics in the US online DVD market. Using the approach of two-level random coefficient regression models, we have tested hypotheses to compare the pricing behavior of dotcoms and multichannel retailers. The data collected in this study raised several challenging modeling issues. At first glance it seems that the well-known VARX model would be an obvious choice. Because of the nature of the data, however, the research carried out at the disaggregate level would cause aggregation bias in the VARX model. In addition, the large number of commodities included in the study would have the problem of over-parameterization. To address these issues, we used an alternative approach to analyze the comprehensive dataset of online DVD prices, the two-level random coefficient regression models. The results of the empirical analysis have provided support for our hypothesis that multichannel retailers charge higher prices than dotcoms. This is true not only at a given time-point as shown in earlier static analyses (e.g., Tang and Xing 2001, Pan et al., 2002, Ancarani and Shankar 2004) but also is the case over the whole time period of this study. This result is consistent with the findings in Xing et al. (2006) but the conclusion was drawn on a more solid basis because the more advanced econometric model was used in the analysis so that the estimates were more efficient and potential biases were avoided. Our results on the trend of the difference in price levels extend the existing research on the online DVD market. Based on the price data collected from 2000 and 2001, Xing et al. (2006) show that (a) price levels went up over time at both dotcoms and multichannel retailers; and (b) the prices at dotcoms increased faster than those at multichannel retailers. Xing et al. (2006) suggest that the difference in price levels tends to converge in the long term, and further predict that the two retailer types will have similar pricing behavior. The results of this study, based on the more recent price data, suggest some interesting findings which have reflected the evolution from the early stage to a more mature US DVD market. First, contrary to the dynamics revealed in Xing et al. (2006) where overall prices increased over time, this study has exhibited declining trends in price levels. This is in line with our observation that demand is high when titles are launched and will diminish when their rankings drop and new titles become more popular. Consequently, prices reduce gradually over time. We note that the analysis in Xing et al. (2006) was based on the data collected during an earlier stage of Internet retailing. It is likely that their results reflected the fact that during the initial years of Internet retailing, online retailers charged lower prices at first to attract customers, and then gradually raised prices. When the DVD market became more mature, however, such a high price level was not sustainable due to competition. This is likely the reason that we observe a pattern of declining DVD prices in this study. Next, in contrast to the prediction by Xing et al. (2006) that the two retailer types will have similar pricing behavior, this study shows that the difference in price levels is maintained over time. This seems logical for a mature market since there are strong theoretical reasons for different pricing behavior between the two retailer types: (a) multichannel retailers can combine the benefits of online shopping with support from offline stores; (b) multichannel retailers can translate market power from offline to online mode; and (c) dotcoms have lower operating costs. With all these fundamental retailing characteristics being maintained, it is unlikely that the difference in price levels will decrease in the long term. From the stability in pricing behavior revealed in this study, there is a sign of maturity in the current US online DVD market. The results obtained in this study thus should not be generalized to immature online markets since it is unlikely that a market in a transition period will exhibit such stable pricing behavior. Our results also suggest that on average the price levels of popular titles at both dotcoms and multichannel retailers are higher than those of random titles. This is consistent with our observation that the demand for popular titles is normally higher. Our results are consistent with Lee and Gosain (2002) in that the degree of price dispersion depends on the product type. Contrary to Tang and Xing (2001) and Xing et al. (2006), however, our results show that for dotcoms the standard deviation was lower for popular titles, whereas for multichannel retailers there was no statistical difference in the price dispersion of the two product types. Our results show that price dispersion among dotcoms increases over time. Contrary to the prediction in Xing et al. (2006) that price dispersion levels among dotcoms and multichannel retailers will converge, our results show that among dotcoms the price dispersion continues the increasing trend but among multichannel retailers the price dispersion seems to have become stable after several years of competition. Hence, there is no sign that the price dispersion levels for the two retailer types are similar. We conjecture that the different patterns in price dispersion between dotcoms and multichannel retailers are fundamental because of the following reasons. First, some dotcoms have established their reputation and have a considerable market share, whereas new dotcoms are being launched every year because online markets have easier entry than offline markets. Consequently, dotcoms tend to be more heterogeneous than multichannel retailers. Secondly, dotcoms tend to have head-to-head price competition, whereas multichannel retailers can differentiate themselves from dotcoms on nonprice dimensions. The major findings of this paper include: (a) multichannel retailers charge higher prices than dotcoms throughout the whole time period of this study; (b) contrary to the prediction in the previous studies that the two retailer types will have similar pricing behavior, this study shows that the difference in price levels is maintained over time; and (c) contrary to their prediction in the previous studies that price dispersion levels among dotcoms and multichannel retailers will converge, our results show that the price dispersion among dotcoms continues the increasing trend but the price dispersion among multichannel retailers seems to have become stable. All these findings show that there exist some fundamental differences in pricing behavior between dotcoms and multichannel retailers. Managers are interested in a better understanding of the pricing behavior in different channels. Our results have some important managerial implications. First, it is suggested that the two retailer types compete with each other in quite different ways. Multichannel retailers tend to differentiate themselves from dotcoms on nonprice dimensions by combining online shopping with support from their offline stores. They may also translate their market power and brand names from offline to online modes. They can charge higher prices without losing all their customers and this avoids severe price competition among them. It is thus essential for multichannel retailers to fully utilize their offline stores in retailing and to ensure that their brand names are successfully translated into the corresponding online markets. Dotcoms tend to differentiate themselves from multichannel retailers through pricing, and competition among dotcoms also tends to be on the price dimension. Since online markets have easier entry than offline markets, direct price competition among dotcoms tends to be more intense. It is thus vital for dotcoms to use effective pricing strategies such as applying random pricing, adopting different shipping costs, and offering product bundles. To differentiate themselves from each other and to avoid direct price competition, dotcoms should also provide better service and communicate with customers effectively. The findings of the present study must be considered in the light of the limitations in the data collection process. Because the sampling frame (i.e. the full list of the DVD titles in the market) was not available during the study period, probability-based sampling methods (such as simple random sampling and stratified sampling) could not be used in this study. Instead, in this study we followed the method used in the existing studies (e.g. Brynjolfsson and Smith, 2000, Tang and Xing, 2001, Ancarani and Shankar, 2004 and Xing et al., 2006) when selecting retailers and DVD titles. Although the market share of these online retailers was substantial, the data was not randomly selected and potentially it may lead to bias in the data collection process.