Model migration and rough edges: British actuaries and the ontologies of modelling

The existing literature on modelling provides two main ways of viewing model migration: a modular view, which seeks to decompose models in their constitutive elements, and thus provides a view on what it is that migrates; and a practice-based view, which focuses on modelling as an activity, and understands a model as intricately entangled with its context of use. This article brings together these two sensitivities by focusing on ontologies of modelling. The paper presents a case study of the appropriation of modern finance theory's 'no-arbitrage' models by British actuaries - a process that gradually unfolded at around the turn of the century and led to significant friction within the UK's insurance industry. We can distinguish two main modelling ontologies: a 'risk-neutral ontology', which underpins no-arbitrage models and holds that the value of financial instruments is determined by 'arbitrage'; and, a 'real-world ontology', which assumes that the economic world consists of real probabilities that may be approximated through a combination of archival-statistical methods and expert judgment. The appropriation of the risk-neutral modelling ontology was made possible by the declining legitimacy of actuarial expertise as 'financial stewards' of life insurance companies. The risk-neutral modelling ontology provided an 'objective' alternative to the traditional actuarial models, which explicitly required actuaries to make 'prudent' judgments. Despite the fact that the no-arbitrage modelling was considered an 'objective' affair, the valuation models that insurers use today are strongly shaped by political compromises, a result of the 'rough edges' of models.