The Bank for International Settlements (BIS) is launching Project Neo to explore how more frequent and granular data from novel sources combined with advanced analytic technologies can help central banks better fulfil their mandates.
A joint effort by the BIS Innovation Hub Swiss Centre and the Swiss National Bank (SNB), Project Neo aims to ensure that central bank policymaking decisions are grounded in timely and precise intelligence.
Central banks are actively examining how advanced data analytics (such as artificial intelligence ) can strengthen the effectiveness of their policy work. Project Neo is the first in a series of experiments that the BIS Innovation Hub plans to undertake to understand how technology can help central banks in their core functions. The Innovation Hub is therefore expanding the scope of one of its six core themes to include "monetary policy tech" together with "suptech and regtech". These are the use of technology for supervision and oversight and for meeting regulatory and compliance requirements, respectively.
Project scope
Official macroeconomic statistics, such as gross domestic product and inflation, often suffer from delays and a lack of detail, complicating effective decision-making. To address this, Project Neo aims to:
- explore novel data sources from both private companies and public institutions, such as air transport, road traffic, air pollution, electricity, cargo transport and retailers and payment services; and
- use advanced data science techniques to gain new economic insights by creating unique economic indicators and produce relevant forecasts of macroeconomic statistics.
Project Neo is innovative in that it aims to work with unique data from national private companies. It will combine this information with typical macro-financial data sets, targeting a broad range of official statistics. At the same time, Project Neo will quantify the relevance of using disaggregated micro data to understand and forecast macroeconomic indicators. The project will benefit from exploring the synergy between economists and data engineers and use machine learning and predictive modelling.
The project will initially work with data from Switzerland, but the prototype and learning experience will be applicable to any other country.
Frequently asked questions
Project Neo aims to assess the added value of granular and timely data from companies in monitoring economic activity. It evaluates how these data sources can generate new economic insights and enhance short-term forecasts of key macroeconomic statistics, such as inflation, GDP, and consumption.
Official macroeconomic statistics are essential for conducting monetary policy as they provide key insights into economic trends and developments. However, these statistics are often published with substantial delays and are periodically revised. Project Neo leverages detailed and near real-time data from companies to provide a more up-to-date view of current economic development. For instance, using flight data enables us to monitor the tourism sector on a weekly basis and gain insights into the origins of tourists in a timely manner. Another example is using health insurance data to gain an understanding of current health consumption at high frequency and across different types of sociodemographic groups. Official health consumption is annual and delayed by more than 2.5 years.
Project Neo examines how emerging technologies, such as artificial intelligence (AI), can support central banks in their core functions. It innovates by providing insights and tools to assist central bank analysts and decision-makers in conducting monetary policy.
The BIS Innovation Hub is well-positioned to lead Project Neo for three reasons. First, its insights can be shared with other central banks. Second, the Hub fosters cross-expertise collaboration by combining diverse IT, economics and data science skills within a single team. Third, its flexible IT environment enables rapid experimentation and innovation, offering capabilities that surpass those of many central banks.
Project Neo is a collaboration between the BIS Innovation Hub Swiss Centre and the Swiss National Bank (SNB).
The aim is to create a prototype that utilises granular and timely data from novel sources, combined with advanced analytical technologies, to provide central banks with new economic insights and enhance short-term forecasting of key macroeconomic statistics.
Yes, Project Neo will publicly share its key findings through a detailed report and an informative video.
For initial testing and prototyping, Project Neo will use data from Switzerland. However, the prototype and insights gained can be applied to other countries as well.
Project Neo utilises two main categories of data: public and from Swiss companies. Public data encompasses freely available information from statistical offices and commercially available datasets from data vendors that are accessible to anyone. Company data comprises granular, high-frequency information from bilateral agreements with Swiss companies.
To be valuable, data must be detailed, up-to-date and highly frequent (ideally weekly or daily). Data from companies fits this description better than most public data.
The company data used in Project Neo offers valuable insights into key sectors of the Swiss economy, including health, retail, construction, real estate, and tourism.
The project does not use personal data or personally identifiable information. The project aims to strike a balance between the need for granular, high-frequency data and strict privacy protections. While detailed data is essential for improving economic monitoring and forecasting, all information is aggregated to prevent the identification of individuals. All socio-demographic information is only available at a grouped level. For example, a dataset might include insights on women within age and income groups living in a certain district but does not contain individual records or personally identifiable details. Project Neo adheres strictly to Swiss data protection regulations and confidentiality agreements to ensure the highest levels of privacy and security.
All data used in Project Neo is stored securely in Switzerland and remains within the country. Furthermore, all data processing is conducted exclusively within Switzerland, ensuring compliance with local data protection regulations and maintaining strict confidentiality standards. The project's underlying data will remain strictly confidential per agreements with data providers. Project Neo is committed to neither sharing nor monetising this data with third parties.
Project Neo trains supervised machine learning models to nowcast and forecast relevant macroeconomic variables, such as GDP, consumption, and inflation. Algorithms such as linear regressions, tree-based approaches or dimensionality reduction strategies will be implemented.
Project Neo neither uses nor trains large language models (LLMs) or generative AI. The data we work with is already structured, eliminating the need for AI-based text processing. Since the datasets are well-organized and can be directly analysed or converted into time-series formats, our approach does not require generative AI.
Project Neo will quantify the value of company data, i.e., timely and granular data, to better understand and forecast macroeconomic indicators. It will serve as a complement to the toolset traditionally employed by macroeconomists at central banks.
The primary purpose of official statistical offices is to compile and publish the most accurate representation of various aspects of the economy. This crucial data is often available with a lag. Project Neo is focusing solely on data to provide more up-to-date information about the current state of the economy. This, however, comes at the cost of less precise statistics. Moreover, official statistics are the primary targets of central banks. Project Neo will thus not alter the relevance or work of official statistical offices but serve as a complementary tool for policymakers.