Abstract The rapid advancement of artificial intelligence (AI) relies on a complex supply chain comprising five key layers: hardware, cloud infrastructure, training data, foundation models and AI applications. This paper examines the market structure of each layer and highlights the economic forces shaping them: rapid technological change, high fixed costs, economies of scale, network effects and in some cases, strategic behaviour by dominant firms. We also highlight the expanding influence of big tech companies across the AI supply chain. We discuss the challenges for consumer choice, innovation, operational resilience, cybersecurity and financial stability.

Abstract This study investigates quality choices when a product version is outsourced compared to full in-house production of all product versions. Our findings suggest that outsourcing the high-end version leads to improvements in the quality of the low-end version, as the firm seeks to strengthen its bargaining position with the contractor to secure a lower wholesale price for the outsourced high-end version. In situations involving asymmetric information, this effect can help alleviate the downward distortion of low-end quality, thereby enhancing overall welfare. However, outsourcing may also result in a level of low-end quality that exceeds the socially optimal level. The key drivers of this trade-off are the proportion of high-type versus low-type consumers, which determines who is served in the event of a negotiation failure, and the relative bargaining power of firms and contractors, which defines the relative weight of the agreement and disagreement payoffs in setting quality. On the other hand, outsourcing low-end versions is less likely to contribute to social welfare improvement, as it diminishes the quality of the in-house, high-end versions. The study also shows that while outsourcing affects quality choice, the outcomes are largely unaffected by the specifics of outsourcing arrangements and game timing.

Abstract We empirically assess the impact of ride-hailing platforms on the incidence of drunk-driving fatal crashes and fatalities in Chile. Using a difference-in-differences approach, we study heterogeneous effects in fatalities by gender and role in the crash (driver or passenger). Our results suggest that the introduction of ride-hailing platforms has significantly reduced fatal crashes and fatalities, especially the number of female passengers’ fatalities and the number of male drivers’ fatalities at night. The former result may evidence that ride-hailing platforms like Uber can contribute to the mitigation of the mobility bias against women in the traditional transport sector.

Abstract Railroads remain a critical transportation mode for the movement of U.S. agricultural freight. Within much of the northwest and mid-central U.S., rail is often the only viable mode to transport bulky agricultural commodities, including wheat. With the potential for exploitation of market power by railroads over such movements, regulations exist that are designed to mitigate the effects of monopoly railroad situations. But what of duopoly railroad markets? Economic theory offers reliable predictions of firm behavior when there are either many firms serving a market, or conversely when there is just a single firm serving the market. But behavioral predictions are not as straightforward when evaluating oligopolistic market structures. Relevant to this research, it is not clear a priori what kind of firm and market behavior might emerge under a duopoly. Clarifying what happens in such cases ultimately becomes an empirical issue. In this paper, we investigate a significant U.S. wheat transportation market currently served by a Class 1 railroad duopoly, but railroad behavior in this market may also be moderated by intermodal competition from water barge. While our findings about railroad behavior over time indicate a tendency towards Cournot duopoly behavior, latent variable analysis offers a more granular understanding of how both intra- and inter-modal competition affect the chosen transportation market. With only a very limited number of Class 1 railroads left serving the entire country, the future of U.S. freight transportation by rail will be comprised of numerous important products and regions served by only one or two railroads. While some of the methods we use in this analysis are novel to the industrial economics literature, we believe this effort will help better inform future regulatory policy design for rail, further strengthening market vigilance for freight shippers who are destined to transport goods in increasingly concentrated railroad markets.

Abstract This study examines the impact of price-cap regulation and subsidies for truthful reporting on media bias in a duopolistic market with both traditional and digital media. While subsidies reduce bias and promote balanced news coverage, price-cap regulation exacerbates online polarization and displaces traditional media. These findings provide valuable insights for policy discussions on fact-checking, reporting standards, and social media regulation.

Abstract In this paper, we consider a platform that sells both the first-party product and the third-party product. The product recommendation of such a platform is interpreted as cheap talk, because it is unbinding and costless. We show that if the consumer has the outside option to exit from the platform, there exists a partially revealing communicative equilibrium in which the platform makes a biased recommendation with some positive probability while the consumer follows the platform’s self-referencing recommendation with some probability and takes the outside option to exit from the platform with the remaining probability. In this equilibrium, self-preferencing occurs. Thus, the consumer’s exit option is essential to this self-preferencing equilibrium. We also show that both the platform and the consumer are made better off in this partially revealing self-preferencing equilibrium than in an uninformative equilibrium or without using the search engine. We also extend our arguments to the Hotelling model with consumers’ exit option and draw an interesting policy implication that if a platform’s commission fee is regulated, it can increase the platform’s self-preferencing bias.