This article is also available in
In a time when it is difficult to collect data from the field, the use of administrative registers for statistical purposes can be an important alternative source of information to continue making data and statistics available to the public. While many countries already have systems in place for the use of administrative data and would only need to adjust the approach to include more sources, other countries may need to start with the basics.
There are many different administrative registers available which can be used for different purposes, and if interoperable, also can be linked to other sources of data. A health register can say something about number of health facilities and health workers in different regions. Combined with population data, this could be used to assess capacities to handle COVID-19 patients in those regions. A business register can be used in light of COVID-19 to help inform how many businesses there are in different sectors, as well as provide information on their size in terms of number of employees which again could say something about the potential impact on the economy.
As is the case with other data sources, administrative records may suffer from various limitations with respect of completeness, coverage, representativity of the target population, and other data quality issues. This will depend on the collection approach (online or in person), how often information needs to be updated and the type of respondents (individuals vs businesses for example), and the existence of data validation controls at different stages of their life cycle, among other factors.
During the time of crisis, national statistical offices (NSOs) should prioritize the use of data sources that are already available to them, before focusing on collecting new data. As a first step, it is crucial to ensure that data reporting flows that were already in place form entities owning administrative data systems to the NSO continue operating without interruption. This may be challenging, particularly in the current situation where the staff responsible for running and overseeing such data flows may be working from home without adequate access to the necessary IT infrastructure.
It is thus very important for NSO staff to team up with the owners of key administrative systems to ensure the continued sharing of data. They should also work together to explore how best to overcome access restrictions and other disruptions in existing data flows. For instance, during the COVID-19 crisis, the owners of administrative data systems may also experience disruptions in the inflow of information, particularly if the source data is not provided to them online or by other automated means. In these cases, it is crucial that NSO officials collaborate with the administrative data owner to establish new approaches for data collection, at least on a temporary basis. NSOs may even contact the disaster risk management agency and humanitarian actors, who can help on the ground in order to keep data flows going. This is particularly relevant in the context of COVID-19, given that most health and vital statistics originate from administrative records. So do other statistics that are important to monitor and assess the impact of the pandemic, such as statistics on social services and education statistics.
There could be disruptions in the collection of administrative data even if most information is provided online. This may particularly be the case for businesses which are severely affected by the shutdown that is taking place in many countries. In these situations, the statistical office can support the register owner in reaching out to key stakeholders and the media to highlight the importance of continued reporting to help ensure continued quality statistics which reflect the changes in the situation, and which may be helpful for policy makers.
To be able to produce the information needed by policy and decision markers responding to the COVID-19 pandemic, it is crucial so secure access to all relevant administrative data sources, particularly those linked to the production of health, education and business information. Below a few steps that could be relevant in this context.
Establishing contact: It may be challenging to establish contact with a register owner at these times, but there may also be more willingness to collaborate than under normal circumstances. At the end of this text is a list of key potential administrative data sources which could be considered. How to reach out to them will depend on local context. Generally, informal personal contacts may be helpful, but a formal request at the highest level, specifying the objectives and the urgency of accessing the data, may be necessary. One could, for instance, highlight the importance of statistics for policy making in times where so many aspects of life are changing,
Formal agreements: In many countries there is generally a need for a formal agreement in order to ensure data exchange. This may take time and may also be more difficult in a situation where meetings are discouraged. In light of this, one could agree to test out data exchange informally given the situation and start working in practice, saving the formal agreement for a later stage. In other countries, the Statistical Act does clearly specify that the statistics office should be given access to relevant data sources. This should then be sufficient to start sharing data.
Specifying what to share: It will be important to agree on the specific scope of the that to be shared. That means specifying which data or variables, for what reference period, at what level of detail (microdata vs aggregated data), in which format (type of file), and how often. A template for how data should be structured should be quickly developed and agreed in collaboration with the administrative data owner.
Specifying how to share: Electronic exchange will be a prerequisite, but there are many different options. Automated exchange from the administrative data owner´s database to that of the statistical office would be an ideal option but may not be feasible at present. In those cases, data could be exchanged using in open, machine-readable formats such as .csv or excel. Unstructured or non-machine-readable file formats, such as word or .pdf, should be avoided. And if micro data are exchanged, this should be done using secure transfer options, for instance by encrypting the files or similar.
Data quality assessment and feedback mechanism: Many administrative data sources face various data quality challenges. They can be incomplete, just cover a certain sub-set of a population, or there can be many errors in the data collected. Data quality assessments and the application of consistent standards is therefore important before the data are used for statistical purposes. A statistics office often has expertise on many of these matters and this can be offered to the register owner. The UN National Quality Assurance Framework can be a useful resource in this work.
Establishing a statistical database or register: If data are shared on a micro data level, it will be important to also establish a statistical database or register based on the administrative data. Key elements of a statistical database or register is that it mirrors the administrative data source but allows for corrections to the data for statistical purposes. While it for instance is illegal to remove a business from an administrative data source in many countries if the cease of the business has not been reported, the same company can be removed from a statistical register following specific rules such as inactivity for a certain amount of time with the aim of improving the overall quality of statistics. It will be key in this regard, to create clear rules on how new information from the administrative source is added to the statistical register when updates are received.
Ensuring interoperability: Data sharing, linking and processing will be substantially easier if the data are interoperable. Where possible, the use of common classifications, data models, and data formats across various providers of administrative data should be encouraged and supported as much as possible.
Linking data sources: One register may not be sufficient to satisfy specific data needs from policy and decision makers. If data are interoperable, it may be possible consider linking different sources in order to ensure coverage of certain population groups and to be able to obtain a more detailed picture of linkages between different populations and phenomena.
Confidentiality and security: When sharing and when linking micro data, it will be key to ensure that data are kept confidential to external actors. No data should be published that lead to breaking confidentiality of respondents, and it is also important to ensure that only key people are given access to the data and that they also agree to keep information confidential. Linked to this, it is also key to ensure that the data systems are secure so that externals can’t access the systems.
It may be that full scale data exchange is not possible at present due to access, legal issues or other aspects. In those cases, NSO staff can consider collaborating with the owners of administrative data systems, offering advice and support on how to conduct data quality assessments, how to codify and clean the data, and how to tabulate it to produce useful statistics and information. This could be done through video conferences and screen sharing. That way, more data could potentially be published under the current circumstances and a closer relationship would be established. International partners could also be involved in this work if relevant.