Title: Who Really Makes the Decision? AI, Algorithms and Human Behaviour in Digital Markets

From personalised advertising and online platforms to AI-driven recommendations, algorithms increasingly mediate our everyday choices. But how do these technologies interact with human behaviour? Are they simply learning what we want, or are they also shaping what we come to want? And what are the implications for consumers, organisations and society?

Drawing on his interdisciplinary work spanning machine learning, behavioural economics, game theory, digital advertising and online platforms, Prabir will explore the increasingly blurred boundary between human and algorithmic decision-making, and the opportunities, risks and ethical questions that emerge as AI becomes embedded in markets and everyday life.

The seminar should be of interest well beyond business and marketing, particularly to colleagues and students interested in AI, economics, behavioural science, computer science, public policy, ethics and the changing relationship between technology and society.

Title: Can a University Go Global Without Losing Its Roots? Education, Language and Public Value

Universities increasingly aspire to be global: attracting international students and faculty, building international partnerships, and preparing graduates for a world without neat geographical boundaries. But can a university become more international while remaining meaningfully connected to the place, languages and communities in which it exists?

Drawing on her experience of Cardiff University’s internationalisation, Welsh-medium and bilingual education, Cardiff Business School’s Public Value agenda, and engagement with local communities, Professor Rosier will invite us to think about some deceptively simple questions: Who is a university ultimately for? What does it owe to its place? What happens to local language and identity as universities internationalise? And should preparing students for the world also mean preparing them to contribute to the communities around them?

Title: Self-Affirmation and Network Behavior: A Field Experiment of Women Entrepreneurs in Kazakhstan

Research on gender-based network disadvantage has largely attributed women’s less effective professional networks to structural barriers. Yet structural explanations alone cannot account for persistent gender differences in network behavior. Psychological constraints – specifically, stereotype threat triggered by agentic networking – can also prevent women from pursuing the high-status and bridging ties most beneficial for resource acquisition.

Drawing on self-affirmation theory, we propose that a brief values-affirmation intervention can reduce these psychological barriers and increase women’s agentic network behavior. We test this argument through a longitudinal field experiment with 427 women entrepreneurs (224 treated, 203 controls) in Kazakhstan, where conservative cultural norms make stereotype threat especially salient. Self-affirmation significantly increased entrepreneurs’ intentions to form high-status ties, a type of agentic network behavior, and these elevated intentions mediated the effect of self-affirmation on the initiation of high-status ties.

However, we did not find effects of self-affirmation on bridging tie intentions or initiation. Our findings contribute to research on gender and networks by highlighting pre-formation psychological barriers as a distinct source of network disadvantage, extend network agency theory by documenting psychological enablers of agentic behavior under societal constraints, and advance self-affirmation theory into the interpersonal domain.

Title: Exchange Rate Pass-through in Vertical Markets with Price Regulation

This paper studies how price regulation and the structure of vertical markets affect exchange-rate pass-through (EPRT) in the pharmaceutical industry. Across different market structures, we analyze ERPT for drugs with regulated and non-regulated prices, as well as for retail and wholesale transactions in Kazakhstan. First we show that regulated ERPT are around one-third of the non-regulated ERPT. For non-regulated drugs, we justify theoretically and show empirically two properties: (i) the wholesale ERPT is higher than the retail ERPT, (ii) drugs that experience stronger competitive pressure exhibit higher ERPT, while monopolistic markets demonstrate the lowest ERPT. When the price regulation is introduced, we show that the wholesale ERPT is more sensitive to the regulated price caps, especially for monopolistic markets, and the property (ii) may not hold. We contribute to the literature by showing that price regulation, position in the supply chain, and market structure jointly determine the transmission of exchange rates into prices.

Title: Consensus on the Lesser of Two Evils

We propose a new ordinal axiom for interpersonal comparisons, Consensus on the Lesser of Two Evils. It requires that whenever both agents agree that the position one allocation assigns to one of them is better than the position another allocation assigns to the other, society ranks the first above the second. Imposed on all bundles, the axiom is incompatible with Pareto efficiency as soon as the two agents’ preferences differ; restricted to the positions the efficient frontier makes available, it yields a sharp representation. We show that a social preference satisfies the axiom if and only if it is represented by a mixed-benchmark metric: the welfare of agent (i) is the sum (m_i^1 + m_i^2) of two continuous increasing representations of (\succsim_i), where for each (k \in {1,2}) the pair ((m_1^k,m_2^k)) agrees on agent (k)’s projection of the frontier, that is, (m_1^k(x)=m_2^k(x)) for every bundle (x) that agent (k) receives in some Pareto-efficient allocation. Because the axiom bears only on single-recipient allocations, the representation constrains the calibration of welfare alone and holds for any aggregation rule—utilitarian, prioritarian, or leximin. We relate the result to existing welfare indices built on a single reference path, and show that our criterion is logically independent of envy-freeness while always compatible with it, and that it symmetrizes Suppes’s grading principle across agents.

Title: Emotional Coherence and Venture Outcomes: Machine Learning Approaches in the Study of Startup Pitches

In this paper, I examine the role of emotional coherence, defined as the degree of similarity in the sentiments expressed between verbal, vocal, and visual communication, in startup pitches in predicting financing outcomes for startup companies. Using data, including videos, from a prominent accelerator, a premier startup pitch competition, and a pre-accelerator program, I use machine learning tools to quantify a measure of emotional coherence in startup pitches. The paper (1) codifies and empirically measures a dimension of “gut feel” in early-stage startup investing, (2) contributes to the literature on investor decision-making by showing how investors interpret behavioral cues and communication signals, and (3) contributes to the literature on managerial cognitive capabilities and entrepreneurial communication. I find that greater emotional coherence expressed in venture pitches is positively associated with venture outcomes.

Title: Beyond the Interview: Whose Stories Are We Really Telling?

The Art, Ethics and Politics of Qualitative Research

Whether you are a qualitative researcher, a quantitative researcher curious about what numbers cannot capture, or simply someone interested in how knowledge is constructed, I think you will find something worth questioning in this conversation.