Testing hedonic theories using online data

Sun, Weiming (2024) Testing hedonic theories using online data. Doctoral thesis, University of East Anglia.

Preview

PDF Download (2MB) | Preview

Abstract

The theme of this thesis is the use of online data to test hedonic theories. “Hedonic theories” are theories which embody the idea that consumer behaviour is motivated by the pursuit of satisfaction. Online market data is very useful in the testing of such theories, because it often includes large amounts of information in the form of customer reviews, which represent a natural measure of customer satisfaction. In some hedonic theories, customer satisfaction data is used to explain outcomes such as product price, producer’s sales volume, or customer purchase intentions. In other hedonic theories, customer satisfaction takes the role of the the dependent variable, and the focus is on the features of the product (including price) which influence satisfaction.

The thesis comprises three main chapters.

In the first main chapter, a meta-analysis is conducted to investigate and explain between-study differences in the estimated effects of electronic word of mouth (eWOM) on consumer behavior outcomes, such as purchase intentions and actual sales, in peer-to-peer online markets. Drawing from 155 studies and analyzing 836 reported effects, the research explores the factors that influence the significance of eWOM-related reputation effects. Using meta-regression analysis, we find that larger sample sizes are consistently associated with more statistically significant results, in line with expectations from statistical theory. The study also reveals important geographic variations, with studies from Asian countries showing higher significance levels compared to those from other regions. Additionally, the analysis highlights a noticeable shift in findings for studies published after 2020, where lower significance levels are observed, potentially reflecting a fall-off in consumer trust in online reviews following the COVID-19 pandemic. Furthermore, our results show that experimental studies are more likely to yield significant outcomes than observational ones, possibly due to their controlled environments that allow for clearer causal inferences. These findings contribute to the broader understanding of how eWOM impacts consumer behavior across different contexts and offer valuable insights for future research on online reputation systems.

In the second and third main chapters, the focus narrows to a particular online market: wine. The second main chapter presents a Hedonic Pricing Model for wine, focusing on the impact of weather conditions prevailing during the year in which the wine was produced on the current prices of red and white wines across different regions of the world. This study expands upon the work of Ashenfelter et al. (1995), who used weather variables to explain the prices of Bordeaux wines. Our extension is that we incorporate a more diverse global dataset. The research aims to identify how variations in temperature and rainfall during the grape growing, maturation, and harvest seasons influence wine prices. By including regional dummies, we are also allowing for regional differences in “terroir” (that is, the environment in which the wine is produced, including factors such as the soil, topography, and method of production). We find that an array of weather variables observed around the time of production of the wine, have a significant effect on the current prices of the wines. These sorts of results are clearly very useful for wine producers, investors, and policymakers, since they provide a means of predicting the future quality of a wine at (or even before) the time at which the wine is produced. More importantly for our own purposes, these results provide a valid set of instrumental variables which allow for the possible endogeneity of the price variable in the analysis conducted in the final main chapter of the thesis.

In the third and final main chapter, Online customer review data is used, together with price data, to test the hypothesis that wine is a “Veblen Good” (Veblen, 1899). We define a Veblen good as a good which becomes more desirable to consumers as the price increases, even after controlling for true quality. The Veblen phenomenon is very interesting to economists because it can be perceived as a violation of the almost universal “Law of Demand”. The test is performed in the context of a weighted regression model with average customer rating as the dependent variable, and the log of price as an independent variable. The key to the empirical strategy is that the “true quality” of the wine is controlled for, by including a measure of “expert rating” as an independent variable in the regression. As mentioned above, we also allow for possible endogeneity of price using weather variables in the year of production as instruments. We find strong evidence of the Veblen effect for red wine, but no such effect for white wine.

Item Type:	Thesis (Doctoral)
Faculty \ School:	Faculty of Social Sciences > School of Economics
Depositing User:	Chris White
Date Deposited:	04 Nov 2025 11:25
Last Modified:	04 Nov 2025 11:25
URI:	https://ueaeprints.uea.ac.uk/id/eprint/100886
DOI:

Downloads

Actions (login required)

View Item