This paper explores the politics of welfare retrenchment, but differs from much of the current literature in this area by focusing not on the decisions of politicians but those of private sector employers. In countries with a large private welfare sector, employers are major social policy players with a significant influence on the generosity of welfare provision, but the rationale behind their actions is not well understood. To explore these issues, a case study is used of the recent fundamental change in UK occupational pension provision, involving a rapid shift from defined-benefit to defined-contribution pensions. The paper shows by means of a micro-simulation of the relative performance of defined-benefit, defined-contribution and state pensions that this shift represents a significant retrenchment. It suggests, using historical material, interview data and insights from behavioural economics, that existing explanations for this change, while valuable, have important gaps because they are based on too narrow a conceptualization of business motives. In this regard, the paper highlights the importance of herd behaviour. . For an overview, see Grabher and Powell (). . In the summer of we conducted expert interviews with all the major members of the British pension policy network, including the Minister for Pensions, the social policy spokesperson for the Conservative Party, major trade unions, the insurance industry and the CBI. These interviews lasted one hour. Our main aim was to discuss policies that would make the British pension system more socially inclusive. . All five workers could be either men or women, even though the full-time career on average wages is more common for men. We chose a gender for the sake of style. . Citing Ostaszewski (), they describe this change as New Economy Theory. . Between and the number of UK VAT-registered companies with a turnover of more than £ million rose by percentage points (statistics.gov.uk, own calculations). . An interlocking directorship involves one person sitting on the boards of two companies. These can be directional interlocks, where the person has strong connections with at least one of the companies, or non-directional, where the person is affiliated to a third institution.