The Bundesbank has created MILA (Monetary-Intelligent Language Agent), an innovative model that uses artificial intelligence (AI) to analyse central bank communication. Public statements by central banks influence market expectations regarding the future path of monetary policy, much as conventional monetary policy instruments exert impacts on aggregate demand and inflation. The pivotal factor here is whether the communication signals a restrictive “hawkish” policy or an accommodative “dovish” stance. Tone also plays a role.

Precision analysis thanks to AI

MILA uses state-of-the-art language models to assess monetary policy statements in a transparent, granular and consistent manner. The model analyses sentences individually, takes into account the economic context and provides a comprehensible justification for each classification. The sentence-by-sentence approach to analysis makes sure that the assessment AI reaches remains understandable for human users, allowing the Bundesbank’s experts to review and query the results directly.

Detailed assessment of monetary policy communication

MILA’s first step when presented with an ECB press conference is to derive the inflation context; this inflation context serves as background information for the actual classification. MILA then analyses what is said about monetary policy instruments and assesses whether those remarks are of a more hawkish or dovish nature. In a third step, MILA examines how hawkish or dovish the economic narrative is by classifying individual sentences from the press conference. MILA also conducts a sentence-level analysis of the tone of the Monetary Policy Statement as a whole and determines whether it is positive, neutral or negative. In addition, it can be used to analyse speeches given by members of the ECB Governing Council.

Opportunities and risks of AI analyses

From a central bank perspective, integrating AI into the analysis of monetary policy texts holds the potential of improving central banks’ communication. But this development also brings associated challenges: leaning too heavily on AI analyses could reduce diversity in market opinions, lead to unexpected market reactions and make it more complex for central banks to communicate, the Bundesbank’s experts explain. 

A scenario where both central banks and market participants increasingly rely on AI might well usher in a future where machines end up communicating with other machines – the repercussions of which could be incalculable. It is therefore imperative to engage critically with the technology so as to leverage the opportunities presented by AI without losing sight of the risks.