Periodic Reporting for period 4 - NewMonEc (Monetary Economics and Communication: New Data, New Tools, New and Old Questions)
Période du rapport: 2023-12-01 au 2024-11-30
Central banks are charged with setting interest rates, and other related policies, to manage the business cycle. But the world in which central banks operate, and the way in which they operate in it, has changed dramatically in the last 25 years. Not that long ago, central banks would operate behind closed doors and not even announce their decisions on these important policies. Now, they are a part of the 24-hour news cycle, make their decision announcements with statements and press conferences, and they actively try to engage not only their traditional audiences, financial markets, but also a wider public.
This revolution of increasing transparency and an increasing role of communication brings challenges to central banks, and to the academics who study them (and their policies). The changing framework shifts the emphasis from interest rates to words. But most of economic analysis is not designed to study words. To do so, requires the use of tools from computer science and the field of natural language processing.
NewMonEc (New Monetary Economics) is my ground-breaking research agenda comprising of pioneering projects across three related themes addressing questions in monetary economics that emphasise the increasingly important role of central bank communication in policy. The projects rise to the challenge by combining new data, new approaches and innovative use of new statistical methodologies from outside economics.
There are a number of key important findings that come from the research.
In Haldane et al. (2021), we show that trust affects the extent to which agents will pay attention to the messages sent by the central bank. We show that simplifying communication can increase the engagement of the public with central bank messages making their expectations more precise. We also describe the key challenges in terms of communicating with the general public as relating to 3Es: Explanation, Engagement, and Education.
In Cieslak et al. (2023), we study how uncertainty impacts decision-making by central bankers. We find that inflation uncertainty, and particularly concerns about upper-tail inflation risks, prompt a more hawkish policy stance. This effect occurs when expected inflation nears or exceeds the target, and we can link this to narrative evidence that suggests a driving concern is a loss of credibility. We link this behaviour to the risk management approach to monetary policy.
Byrne et al. (2022) show the data covering the macroeconomy do not uniquely define the state of the economy. Instead, the central bank, like other people in the economy, need to form an assessment of the state of the economy. We show, using a newly developed methodology from Byrne et al. (2022), that, contrary to standard models, backward looking information reflecting this assessment function is important for the market reaction to central bank communication.
Cieslak and McMahon (2024) look at how alternative-policy scenarios, expressed by policymakers in speeches, impact risk premia in financial markets. We show how market perceptions of policy mistakes can raise risk premia against policy intentions. Communicating a more hawkish policy stance in what an alternative policy might look like predicts lower risk premium in the intermeeting period.
In McMahon and Naylor (2023), we distinguish between the kind of semantic complexity measured by Flesch-Kincaid, and what we term conceptual complexity which relates to the use of economics jargon. Our novel measure, termed the Conceptual Complexity Index (CCI), seeks to better reflect the true information-processing costs identified by theory. We designed and implemented an information provision experiment to distinguish between dimensions of semantic complexity and conceptual complexity and show that managing technical complexity has a bigger impact on understanding and trust than simplifying in a semantic way.
McMahon and Rholes (2023) offers a salutary lesson that precisely when inflation is high, it may be that the publics willingness to listen to what the central bank has to say is reduced. This result comes from an evaluation of how forecast performance, especially the timing of errors, affects the extent to which the public use the central banks forecast to inform their own outlook. Not only does forecast performance matter, we find a form of recency bias when subjects evaluate forecast accuracy. This bias, which applies to both short-term and medium-term forecasts, is especially strong after poor forecasting performance.
This revolution of increasing transparency and an increasing role of communication brings challenges to central banks, and to the academics who study them (and their policies). The changing framework shifts the emphasis from interest rates to words. But most of economic analysis is not designed to study words. To do so, requires the use of tools from computer science and the field of natural language processing.
NewMonEc (New Monetary Economics) is my ground-breaking research agenda comprising of pioneering projects across three related themes addressing questions in monetary economics that emphasise the increasingly important role of central bank communication in policy. The projects rise to the challenge by combining new data, new approaches and innovative use of new statistical methodologies from outside economics.
There are a number of key important findings that come from the research.
In Haldane et al. (2021), we show that trust affects the extent to which agents will pay attention to the messages sent by the central bank. We show that simplifying communication can increase the engagement of the public with central bank messages making their expectations more precise. We also describe the key challenges in terms of communicating with the general public as relating to 3Es: Explanation, Engagement, and Education.
In Cieslak et al. (2023), we study how uncertainty impacts decision-making by central bankers. We find that inflation uncertainty, and particularly concerns about upper-tail inflation risks, prompt a more hawkish policy stance. This effect occurs when expected inflation nears or exceeds the target, and we can link this to narrative evidence that suggests a driving concern is a loss of credibility. We link this behaviour to the risk management approach to monetary policy.
Byrne et al. (2022) show the data covering the macroeconomy do not uniquely define the state of the economy. Instead, the central bank, like other people in the economy, need to form an assessment of the state of the economy. We show, using a newly developed methodology from Byrne et al. (2022), that, contrary to standard models, backward looking information reflecting this assessment function is important for the market reaction to central bank communication.
Cieslak and McMahon (2024) look at how alternative-policy scenarios, expressed by policymakers in speeches, impact risk premia in financial markets. We show how market perceptions of policy mistakes can raise risk premia against policy intentions. Communicating a more hawkish policy stance in what an alternative policy might look like predicts lower risk premium in the intermeeting period.
In McMahon and Naylor (2023), we distinguish between the kind of semantic complexity measured by Flesch-Kincaid, and what we term conceptual complexity which relates to the use of economics jargon. Our novel measure, termed the Conceptual Complexity Index (CCI), seeks to better reflect the true information-processing costs identified by theory. We designed and implemented an information provision experiment to distinguish between dimensions of semantic complexity and conceptual complexity and show that managing technical complexity has a bigger impact on understanding and trust than simplifying in a semantic way.
McMahon and Rholes (2023) offers a salutary lesson that precisely when inflation is high, it may be that the publics willingness to listen to what the central bank has to say is reduced. This result comes from an evaluation of how forecast performance, especially the timing of errors, affects the extent to which the public use the central banks forecast to inform their own outlook. Not only does forecast performance matter, we find a form of recency bias when subjects evaluate forecast accuracy. This bias, which applies to both short-term and medium-term forecasts, is especially strong after poor forecasting performance.
The project has produced 24 papers, including work-in-progress projects that continue, as well as an edited book.
In the early years of the project, the work focused on the task of measuring central bank communication through the use of tools from data science and natural language processing. Using these measures generated from the language and communication of central banks, I have been able to examine the drivers of policy decision making, and to study the reaction of financial markets. This work involved both the collection and cleaning of large amounts of textual and other contextual meta-data, as well as the development of the tools of NLP to do this analysis. These approaches have given rise to the analysis of the the risk premium reaction to speeches, highlighted the importance of how central banks analyse the data for their effects on market expectations, and shows the importance of conjunctural assessment.
Starting a little bit later in the project life, I started a series of papers that made use of macroeconomic information provision experiments. These are key to understand what language actually helps the general public, as opposed to experts, to understand and how to teach them to understand the rationale for decision making.
The final stage of work was the realisation that the common theme of my research extends beyond the narrow study of central banks. As the research evolved, I also started to develop collaborations with data scientists and linguist to develop techniques to study NLP issues beyond monetary economics and focusing on the issue of measuring narrative.
One of the highlights of the project was the invitation to summarise the lessons from this research as the opening speaker at the Reserve Bank of Australia’s Annual Research Conference; I presented a paper entitled "Lessons for monetary policy communication".
In the early years of the project, the work focused on the task of measuring central bank communication through the use of tools from data science and natural language processing. Using these measures generated from the language and communication of central banks, I have been able to examine the drivers of policy decision making, and to study the reaction of financial markets. This work involved both the collection and cleaning of large amounts of textual and other contextual meta-data, as well as the development of the tools of NLP to do this analysis. These approaches have given rise to the analysis of the the risk premium reaction to speeches, highlighted the importance of how central banks analyse the data for their effects on market expectations, and shows the importance of conjunctural assessment.
Starting a little bit later in the project life, I started a series of papers that made use of macroeconomic information provision experiments. These are key to understand what language actually helps the general public, as opposed to experts, to understand and how to teach them to understand the rationale for decision making.
The final stage of work was the realisation that the common theme of my research extends beyond the narrow study of central banks. As the research evolved, I also started to develop collaborations with data scientists and linguist to develop techniques to study NLP issues beyond monetary economics and focusing on the issue of measuring narrative.
One of the highlights of the project was the invitation to summarise the lessons from this research as the opening speaker at the Reserve Bank of Australia’s Annual Research Conference; I presented a paper entitled "Lessons for monetary policy communication".
These projects contain both methodological advances, as well as novel answers to important questions that are facilitated using those developed approaches (see above).
A great deal of my research has been based on developing new methodologies for economists to study communication, in general, and central bank communication specifically. This includes developing supervised multimodal natural language processing methods to study economic communication. I have also development new approaches to measure the complexity of language as well as shifting focus to conceptual complexity of policy communication.
I have also advanced the use of macroeconomic experiments in the study of central bank communication. This remains a relatively new approach and allows us to explore mechanisms driving expectation formation that are impossible to disentangle even with the best survey data.
A great deal of my research has been based on developing new methodologies for economists to study communication, in general, and central bank communication specifically. This includes developing supervised multimodal natural language processing methods to study economic communication. I have also development new approaches to measure the complexity of language as well as shifting focus to conceptual complexity of policy communication.
I have also advanced the use of macroeconomic experiments in the study of central bank communication. This remains a relatively new approach and allows us to explore mechanisms driving expectation formation that are impossible to disentangle even with the best survey data.