Excerpts from opening remarks by Fabio Panetta, Deputy Governor of the Bank of Italy, at the workshop on “Harnessing Big Data & Machine Learning Technologies for Central Banks”, Bank of Italy, Rome, 26 March 2018.
Big data and machine learning are the products of digital technology, whose widespread adoption has important implications for how communication occurs, education is delivered, and knowledge is spread. As we become accustomed to the new digital environment, we are seeing significant shifts in society. Around 80% of Europe’s citizens own a smartphone, personal computer or tablet. The internet is widely used to gather information, communicate and carry out activities that directly or indirectly affect people’s behavior.
Central banks could, and definitely should, play an active role in exploiting digital technologies and the enormous amount of data they generate. The flow of high-volume, high-velocity and high-variety information on agents’ preferences and choices can be put to good purpose by policy makers to make better decisions.
Along with the traditional data stored in well-defined records, today mobile networks and social media generate a ‘data rainforest’ with diverse sources of semi-structured (such as XML or JSON) and unstructured data (audio/video and free text). The traditional and the new data sources can be used to construct better and more timely measures of economic activity.
There are prominent examples of big data being used for policy analysis. Big data is used to estimate unemployment rates or the inflation rates, to improve the forecasts of policy-relevant variables, to compute measures of sentiment of consumers and firms. The number of potential applications for central banks is enormous, but there are challenges ahead. Before using the relationships estimated from the internet or social media for policy, we have to first ascertain that they are sufficiently robust, representative, and reliable. To do this we have to invest in research.