Why do policymakers and economists within a policymaking institution choose to throw away a model and to develop an alternative one? Why do they choose to stick to an existing model? This article contributes to the literature on the history and philosophy of modelling by answering these questions. It delves into the dynamics of persistence, change, and building practices of macroeconomic modelling, using the case of forecasting models at the Bank of England (1974–2014). Based on archives and interviews, we document the multiple factors at play in model building and model change. We identify three sets of factors: the agency of modellers, institutional factors, and the material factor. Our investigation shows the diversity of explanations behind the decision to change a model: each time, model replacement resulted from a different combination of the three types of factors.