Algorithmic Pricing in Emerging E-Commerce Markets: The Roles of Price Sensitivity, Digital Literacy, and Consumer Trust
Algorithmic pricing – the data- and conditions-driven automation of retail prices via software is quickly evolving into common infrastructure in online commerce, yet its psychological aspects from the consumer side have not yet been thoroughly specified, especially in emerging markets where price competition is fierce, digital competencies are distributed unevenly, and institutional trust is relatively weak. Based on the theories of price-fairness and organisational trust, this study proposes and examines a moderated serial mediation model that shows how perceived exposure to algorithmic pricing negatively affects purchase intention through reducing the perceived price fairness and thus consumer trust. Price sensitivity has been theorised to be a factor that aggravates this fairness erosion, while digital literacy has been theorised to be a factor that acts as a buffer to this trust erosion. Path analysis with regression-based and 5,000 bootstrap resamples was performed on survey results from 612 online consumers in Bangladesh. Exposure had an indirect effect through fairness (β = −0.23) and trust (β = −0.24) that jointly mediated the effect to purchase intention (total indirect effect = −0.19, 95% CI [−0.24, −0.14]) while the direct effect was not different than zero. The exposure–fairness link became stronger (β = −0.10) and the exposure–trust link became weaker (β = +0.10) with the influence of price sensitivity, resulting in a conditional indirect effect that differs by a factor of 2.5 across the literacy distribution. The results shift the discussion on algorithmic pricing from one of competition policy to one of consumer capability, and suggest that interventions to increase transparency will under-serviced the consumers most vulnerable to algorithmic pricing.
Retail prices are more and more being produced than posted. In grocery stores and ride-hailing apps, prices are dynamically determined by software that will process the demand signal, inventory, feeds from competitor prices, and individual buyer histories and adjust the price displayed to each customer in a matter of seconds (Stucke &Ezrachi, 2016). It's no longer just for the high-tech incumbents: repricing tools in the software-as-a-service category have been spread among small third party sellers, like Amazon or Daraz, and a price may now be displayed on a product's page without any conscious decision-making by any human.
Studies of the impact of this change have made most progress on the supply side. Economists have recorded an increase in prices when algorithmic traders are introduced in retail gasoline markets (Assad et al., 2024a), that pricing algorithms that respond to the actions of rival firms can lead to higher margins than the competitive benchmark (Brown et al., 2023), and that reinforcement-learning agents can reach supra-competitive outcomes without being programmed to communicate with rival firms (Calvano et al., 2020). Algorithmic pricing is therefore, for the most part, a competition and collusion issue, and thus has been approached as such by legal scholarship (Ezrachi et al., n.d.). In the parallel marketing literature, two themes have focused on the personalisation and its privacy costs: conditioning offers on individual data shifts the distribution of surplus between firms and consumers (Acquisti& Varian, 2005); and perceived privacy invasion harms customer and firm performance (Martin et al., 2017). What's lagged behind is an understanding of what consumers make of an algorithmic price. This is a significant oversight. A visibly contingent price is not just a price that is at or near the bottom of a distribution; it is a statement about the way the seller thinks about her. The price-fairness research literature has long documented that consumers not only compare prices with costs, but with comparison standards and dual-entitlement norms as well (Kahneman et al., 1986); and that deviations from the comparison standards and dual-entitlement norms elicit emotional and behavioural responses that are disproportionate to the monetary cost or benefit at stake (Bolton et al., 2003). Trust research in turn, has demonstrated that e-commerce viability depends on consumers' belief in a vendor's ability, benevolence and integrity (Gefen et al., 2003). If algorithmic pricing is interpreted as a violation of fairness, it should therefore be interpreted as a violation of trust, and trust is the immediate cause of most online purchase decisions (Pavlou & Fygenson, 2006).There are two limitations of the consumer-side evidence that does exist. First, it is overwhelmingly experimental, pitting an ‘algorithmic’ label against a ‘human’ label in vignette settings; it has yielded a contested picture in which, sometimes, people discount algorithms after seeing errors (Dietvorst et al., 2015) and, sometimes, they prefer them outright (Logg et al., 2019), depending on the task and on whether the decision is perceived as affective or objective (Longoni et al., 2019). These designs work great for clearly marking the mechanism but are not great for everyday shopping when there is no label and the mechanism must be guessed at by the algorithm. Second, nearly all of this research was done in high-income economies in which there are well-established consumer-protection regimes and relatively high levels of consumer trust in platforms. However, there is disagreement on exactly what dimensions should drive the response: there is a wide disparity in the level of competition, overwhelmingly based on price and not service or brand (marketing & 2011, 2011); the dispersion of digital skills is high because a large proportion of shoppers were onboarded during the one-decade diffusion of the smartphone Hargittai, E. (2002); and there is little institutional trust in redress, as people routinely pay cash on delivery. In these circumstances the meaning of the same price algorithm can be quite different, and the "edge" behavior of consumer response can be more important than the average behavior. This study thus poses three questions. First, does perception of exposure to algorithmic pricing diminish consumer willingness to purchase, and does it do so by lowering the perceived price fairness and thus consumer trust? Secondly, to the extent that there is fairness erosion, is it magnified when a consumer is sensitive to the price she pays? Third, does a capacity to read the mechanism by which a platform produces what it displays—what has been termed digital literacy—cushion the blow of the trust loss? Using survey data from 612 online shoppers in Bangladesh, where e-commerce is rapidly expanding on the back of mobile internet diffusion, but consumer-protection infrastructure is underdeveloped and most online transactions are still completed by cash on delivery, we answer these questions. We propose a moderated serial mediation model in which exposure is passed on to purchase intention through price fairness and trust.
The paper has three contributions. In principle, it recasts algorithmic pricing as a property of markets that consumers interpret and sets out the interpretive chain – fairness, then trust – between an opaque pricing practice and a behavioural outcome. This is alongside competition-focused narratives which view consumer harm as an "inevitable by-product of higher prices. Second, it finds two boundary conditions that pull in opposite directions and that represent different theory-based logics, namely stake, which increases the accessibility of fairness norms, and capability, which shifts the attribution a shopper makes for a price difference. The practical implication is that the platform managers are most worried about the loss of trust among those shoppers that are least able to rationalise the price that they see. Third, the study provides a validated measure of perceived algorithmic pricing exposure and conditional effect estimates from a market underrepresented in the literature, enabling the moderators to be analysed over a realistic range, rather than being limited to a convenience sample of students.
2.1 Research Context -
The study was conducted in Bangladesh, an emerging e-commerce market characterized by the fast spread of platforms and weak consumer-protection infrastructure. By 2025, Internet subscriptions totaled about 136 million, with the majority of these accessed via mobile networks (Bangladesh Telecommunication Regulatory Commission [BTRC] 2025), and the retailers selling on the marketplaces and social-commerce platforms compete for a first-generation online population. The context has 3 aspects that make it appropriate to the research question. Given that the differentiation of platforms with regard to service quality, returns and delivery reliability is limited, price is the most important competitive variable. Digital skills are ubiquitous and unevenly distributed with the majority of shoppers having their first digital device capable of online access in the last 10 years. Trust is expressed transactionally: cash on delivery is the predominant mode of payment, which means that consumers can go for the goods only after the items are physically examined. It is clear from this marketplace that algorithmic pricing is indeed there, but the results of this algorithmic pricing, such as marketplace repricing tools or app-only discounts, delivery-fee personalisation, are not often clearly disclosed, meaning that consumers must rely on indirect cues to infer it.
2.2 Sample and procedure
A structured, self-administered questionnaire was used for data collection via online and intercepts in Dhaka and two divisional cities. The respondents were adults (18 years of age and older) who made at least one online purchase in the past six months. The sample was evened out using quotas for age band and gender to ensure that it was not dominated by the student population that is more likely to respond to an online survey. Sixty-one percent of the people who answered the questionnaire (612 of 659) did not straight-line, did not complete in less than one third of the median time, or were unable to pass two embedded attention checks and were included in the analysis. These were not coerced; there was no personally identifying information collected and the respondents were debriefed about the purpose of the study.
The composition of the samples is shown in table 1. The sample is gender balanced (53.6% male, 45.6% female), younger adults (69.0% aged 18-34), well-educated with 46.6% holding a bachelor's degree and 33.0% a master's degree and spread across income bands. Marketplace shopping (28.6%) and social commerce (29.2%) are the two dominant channels, fashion (30.2%) and grocery (20.1%) are the two dominant categories, and cash on delivery (56.2%) is the dominant payment mode, which is in line with the above-described transactional nature of the market
-
Ahnaf, M. A., Tahira, K. T., Sani, M. A. I., Islam, F., Bonik, A., Mannan, S. I., Sifat, M. O., Mukta, M. I., Sunbeam, I. F., Tasneem, A. T., & Antmonitoail, R. (2026). Algorithmic pricing in emerging e-commerce markets: The roles of price sensitivity, digital literacy, and consumer trust. Contin. Acad. J. Integr. Res. Stud., 1(1), 1–16. https://doi.org/10.5281/zenodo.24213191
https://doi.org/10.5281/zenodo.24213191