Making it real: Bringing research models into central bank projections

This paper aims to bridge the gap between models in research and models used to support policy decisions in central banks. Models used in central bank projection environments overlap with research models and benefit from lessons learned in research, but they differ from research models in important ways. For example, to deal with real-world macroeconomic projection issues, central bank models may have a broader scope. To inform policy decision-making, models generally need both a theoretical basis and an ability to 'fit' the data. For repeated projection exercises, forecasters need models that can be adapted to deal with data flows, including historical revisions. And, to provide valuable advice, forecasters must incorporate judgement into their projections to address issues outside the scope of the model. If all these challenges are met, then central bank models and projections will also inform the economic narrative that helps the public understand the policy decisions. In this context, this paper is organized around four main themes: 1) model requirements for central bank purposes; 2) overview of the Bank of Canada's main policy models-ToTEM and LENS; 3) challenges in meeting those modelling requirements; and 4) practical approaches to addressing some challenges under time constraints. The paper concludes with a description of how lessons learned from research and practice set the stage for the Bank's future modelling agenda, as discussed in Coletti (2023).