About 10 years ago the finance ministers and central bank governors of the G20 set up the Data Gaps Initiative (DGI) in response to data gaps that became evident in the wake of the global financial and economic crisis, which emerged in 2008. One key recommendation aims at improving access to data (and data sharing), particularly with regard to granular data and microdata. Better accessibility and sharing of granular or microdata open up new possibilities for analysis by providing new insights into the effects of policies. For instance, the effects of monetary and foreign exchange policy differ depending on the bank, enterprise or household that is exposed to the policy. Only analysing granular data allows for taking into account this heterogeneity and thereby to understand the channels through which monetary policy has an impact on prices. Granular data are likewise required to assess the effects of regulatory measures.

Central Banks, amongst other institutions like statistical offices, are not only interested in using the microdata collected at their separate divisions but also need to share this data internally between divisions. One key issue is to harmonise und combine data in central banks, but also across central banks. Under secured conditions, the institutions may even want to allow for data access by external researchers given their scientific independence and personal integrity. Nonetheless, a greater disaggregation of data implies an increased need for data protection. The leading principle when working with microdata is compliance with the respective statutory secrecy and data protection requirements, and thus maintaining the confidentiality of the information submitted by the reporting agents. Thus, it is of interest to learn how some of the data collections could be shared and thereby be operationalised for an improved evidence-based policy making approach.

To establish an exchange on the topic, in 2018 the Bundesbank’s ICBC course “Data sharing” was launched and experts from 17 countries visited the Bundesbank’s Central Office to learn about different aspects with regard to the sharing of sensitive granular data and microdata. The Bundesbank’s research data and service centre (RDSC) in cooperation with different units of the Directorate General Statistics contributed to an agenda for the course. Experts from these units presented their knowledge accumulated and their experiences made with the practical implementation over the past years when data sharing was promoted at Deutsche Bundesbank. In this context, the Bundesbank’s course on “Data sharing” aimed at providing an overview on how microdata sources can be shared internally as well as with independent external researchers and in both ways support a central bank in fulfilling its tasks.

At the beginning of the week-long course, presentations on the Bundesbank’s microdata and the shift towards an increased importance of these data for policy analysis were given. Next, major topics like “collect-data only once“, “the value chain of microdata”, “data protection for microdata” and “data quality” were dealt with in detail. Thereafter, participants gave presentations on the different (micro-)data collections their sending institutions share or want to share - either internally or externally. They also presented on the role data sharing plays in their respective institutions, on how this data sharing is organised and they described the country-specific legal framework data sharing is regulated by. Moreover, in their presentations course participants’ explained the current challenges their institutions face with regard to data sharing and how these are dealt with. The participants’ presentations and the corresponding discussions raised awareness that many institutions dealing with microdata face the same challenges.

In the middle of the week a presentation was given on the Bundesbank’s strategical initiative called the “Integrated Microdata-based Information and Analysis System (IMIDIAS)”. Two further presentations dealt with topics that are particularly important within the IMIDIAS initiative: on the one hand the classification of microdata by SDMX, and on the other hand the metadata system currently established by the statistical directorate and used all over the Bundesbank for harmonized and digital access to the metadata information. Day four of the week began with a presentation on how to give access to sensitive microdata at the Bundesbank’s RDSC and how to establish its internal service for microdata analysis (IRMA). As an alternative approach to share data and metadata within one tool the so-called “Administrative Data Research Facility (ADRF)” was presented. Turing to techniques for sharing and integrating shared data, an introduction to record linkage presented how to integrate microdata from different source. It was followed by a presentation on machine learning techniques in the daily data work. A third presentation dealt with techniques of how to collect data in a survey that is harmonised across several countries. This was explained in a presentation on data production at a glance using the example of the Panel on Household Finance, a major Euro Area cross-country survey.

At the end of the week, an international perspective on data sharing was offered in a presentation on the “G20 Data Gaps Initiative” that shed light on key recommendations on better data sharing and data access recommended by the G20. How these recommendations can be put into practice was explained in the following presentation on accessing and harmonising microdata across central banks within the cross-country initiative called the International Network for Exchanging Experience on Statistical Handling of Granular Data (INEXDA). Though INEXDA does not deal with data exchange itself the undertaking is a fundamental basis for closer collaboration and understanding each other’s information, and thereby a good entry point to also get involved into the ongoing data sharing discussion.

The development at central banks and other institutions towards analysing more granular and microdata, preferably already well-integrated, will continue and microdata collections will further grow in volume, in number, or in both. Accordingly, various stakeholders at central banks and statistical offices, but also in academia, will be confronted with issues related to data sharing and data integration while at the same time obeying statutory data secrecy and data protection requirements. Promoting further discussion and an exchange of views as well as best practices on these issues is necessary and recommended. In every sense the Bundesbank’s course on data sharing contributes to the field. It was highly welcomed by the participants and according to their feedback the course content was viewed as helpful. The course on data sharing will be offered again in 2019.

Text: Stefan Bender and Judith Flory