In this quote, David Dullaway describes the appropriation of marketconsistent quantification techniques as a paradigm shift, techniques that belong to what Chiapello and Walter (2016) refer to as the third age of financialization. At least two other interviewees also referred to the transition from traditional actuarial modelling to market-consistent modelling as a paradigm shift, and the term was also used to make sense of the events at the time the paradigm shift allegedly happened (for example Clarkson, 1997). Kuhn's notion of the paradigm shift has become a convenient shorthand in public discourse for describing a period of rapid cultural change in knowledge practices both within and outside of scientific fields. The notion of paradigm shift tends to be accompanied in popular usage by an understanding of cultural change that is somewhat non-sociological. It assumes that paradigm shifts follow an internal logic, where new paradigms emerge as unsolvable anomalies proliferate. A more sociological reading of Kuhn's work, however, proposes that paradigms can be understood as exemplary problem solutions that are central in the organization of research fields (for example Barnes, 1982). What Kuhn refers to as ‘normal science’ then involves the analogical extension of already existing problem solutions to new problems, a creative endeavour that already implies change or shifting practices. In this view, paradigms co-exist even in highly specialized fields of research, and paradigm shifts occur with the diffusion, institutionalization and analogical extension of exemplary problem solutions and may be either gradual or abrupt, in which case they may be experienced as significant discontinuities.

The paradigm shift my interviewees referred to, from traditional actuarial modelling to market-consistent modelling, had already started in the 1990s, if only in a very limited way. Especially in the context of mergers and acquisitions, consulting actuaries increasingly drew on key exemplars from modern finance theory to estimate the economic value of insurers’ liabilities, even if the calculations were somewhat indicative and the models rather crude.