Predictive power of social preferences over social norms over Pigou Tax Policies

Pigouvian taxation is one of the most efficient means of reducing carbon emissions from consumption. Typical considerations for supporting tax policy at the individual level include whether the consequences or means of the tax policy are fair, whether the tax policy is efficient, and whether the cost to the individual is acceptable. Climate change presents a unique challenge, as the consequences of today’s actions (at least partially) lie in the future. Considerations of fairness therefore expand from traditional intra-generational concerns to inter-generational ones. The development of a large-scale online experiment, designed to identify preferences for inter- and intra-generational fairness, is used to predict support for Pigouvian taxation aimed at correcting negative externalities associated with consumption of goods and services.

Work in Progress, Leading Author

Geographical variations in risk tolerance and the use of financial instruments: evidence from a multi jurisdictional survey

A substantial body of literature examines the determinants of risk preferences and their impact on risk-taking. However, few studies offer a nuanced understanding of behaviour situated in a cultural and geographical context. Using data from a representative sample survey of working individuals in jurisdictions, this paper addresses this gap by exploring how context influences individual-level determinants of risk preferences. The findings reveal that, despite commonalities, context significantly affects how individual characteristics shape risk preferences and behaviour. Incorporating institutional and cultural factors into our understanding of human decision-making provides a more accurate representation of the geographical diversity of individual behaviour when faced with similar issues.

Forthcoming, Co-Author, Source

Firm-level production networks: what do we (really) know?

Are standard production network properties similar across all available datasets, and if not, why? We provide benchmark results from two administrative datasets (Ecuador and Hungary), which are exceptional in that there is no reporting threshold. We compare these networks to a leading commercial dataset (FactSet) and published results on national firm-level production networks. Administrative datasets with no reporting thresholds have remarkably similar quantitative properties, while a number of important properties are biased in datasets with missing data.

Working Paper, Co-Author, Source

Supermarket Checkouts Online: Studying the Effects of Algorithmic Recommendations on Consumer Satisfaction

Automated product recommendations play an increasingly important role for consumer decision-making processes on e-commerce marketplaces. Nevertheless, little is known about the effects of recommendations consumer preference satisfaction. This paper’s central hypothesis is that automated product recommendations transform consumer search, and that this transformation moves consumers away from their ‘true’ preferences. Investigating this hypothesis involves (1) identifying naturally occurring variation in the frequency with which products are recommended to consumers; and (2) studying the difference in consumer preference satisfaction — as expressed via consumer-generated product ratings — between products bought via recommendations and direct search. The analysis has been applied to Amazon data web-crawled over a 32-day period. The results show (1) that consumers tend to be less satisfied with products bought via recommendations as opposed to without recommendations with varying effects for different product types (i.e., substitute and complement goods); (2) that consumers tend to take up recommendations for product types that yield relatively higher preference satisfaction; and (3) that Amazon, when choosing its recommendations, does not maximise for consumer preference satisfaction nor direct recommendation uptake. The results of this study call into question the notion that recommendations are effective decision aids and support consumers in identifying the right choices for them. Instead, consumers need to be wary about both that Amazon does not suggest recommendations in consumers’ best interest, and that even though consumers tend to choose FBT recommendations with relatively high preference satisfaction payoff, taking up recommendations still leaves consumers worse off than had they not taken up recommendations.

Master Thesis, Main Author, Resumé