Economic research at central banks

Scientization in central banks

Goutsmedt et al. in this paper put forth the following characteristics of a scientized central bank, one that has the resources, internal organization, and network to produce scientific contributions: a) Leadership and staff with PhDs; b) Internal resources dedicated to research (e.g. working papers, visiting scholars); c) External network composed of academic researchers and other scientized organizations; d) Focus on publishing in peer-reviewed journals; e) Use of scientific concepts and references in speeches.

In 2015 the Bank of England (BoE) launched “One Bank Research Agenda,” a research strategy designed to “transform the way research is done at the Bank.” What has the BoE’s experience been?

• There was an increasing proportion of staff with PhDs hired, especially from the mid-1990s onward. Before the mid-1990s, around 25% of new hires were from academia. This share declined as more employees were hired from other central banks and international organizations.
• There was a high research output in late 1980s/early 1990s, especially in top journals. This drops in the mid-1990s as the focus shifts to policy analysis. Publications rose again in the early 2000s and the BoE returned as a leading publishing central bank, though still behind the ECB.
• There has been a growing use of scientific references and terminology in speeches over time. But language is adjusted based on audience. There is less scientific jargon used when addressing the business community vs academics. Governors exhibit a lower propensity for ‘science speak’ than other leadership roles like MPC members.

The authors warn:

Becoming a stronger contributor to science is not a foolproof strategy: more scientization can get in the way of maintaining the credibility of a central bank and can be perceived as diverting resources away from more pressing policy work.

The Different Paths of Central Bank Scientization: The Case of the Bank of England
Authors: Aurélien Goutsmedt, Francesco Sergi, François Claveau, Clément Fontan
From: UCLouvain, Université Paris-Est Créteil, Université de Sherbrooke

The evolution of forecasting models

Goutsmedt et al. analyze how the BoE changed its forecasting models during the last 50 years. The authors focused on four key moments:

• In 1979, Margaret Thatcher’s election did not lead to changes in the BoE’s “Keynesian” model, despite the growing popularity of monetarist ideas. The model persisted due to the defensive stance of BoE executives against the government’s monetarist policies, the autonomy given to modelers, and material constraints on expanding the modeling work.
• In 1992, after the UK exited the European Exchange Rate Mechanism, the BoE developed a new smaller forecasting model called MTMM. This shift was largely driven by the new BoE Chief Economist Mervyn King, who advocated for simpler models with clearer theoretical foundations.
• In 2003, the MTMM model was replaced by BEQM, marking a move towards DSGE modeling standards growing in academia and other central banks. This change was pushed by a new generation of modelers at the BoE who were more receptive to these standards.
• In 2011, BEQM was replaced by COMPASS. This change resulted from dissatisfaction with BEQM’s empirical strategy and inability to address questions raised by the Global Financial Crisis. Unlike BEQM, COMPASS was designed as part of a large suite of models aimed at improving forecasting.

Understanding why policymaking institutions change (or not) their models, requires a historical, contextual perspective taking into due account three factors (modellers’ own dispositions, policymakers’ visions, and the model function within the institutions) and their interplays.

To change or not to change The evolution of forecasting models at the Bank of England
Authors: Aurélien Goutsmedt,Francesco Sergi, Béatrice Cherrier, Juan Acosta, François Claveau, Clément Fontan
From: UCLouvain, Université Paris-Est Créteil, Ecole Polytechnique, Universidad de los Andes, Facultad, Université de Sherbrooke