Data Governance for the Data-Driven Government

In the fall semester of 2016-2017, three of my students, Anders Rahbek , Bo Kofoed Jensen and Djaco Khabat Ahmad did a research on how the local Danish municipalities work with the issue of data quality and data governance in their business intelligence applications.
The challenges in data quality and data governance have been prompted to the whole information management industry for over 10-15 years. When we look back at the solutions and practices in the past years, most of them came from sectors like finance, insurance, retail, telco, pharmaceutical and so on. There are plenty of successful implementations in the private sectors but we can rarely find good case stories in the public sector. In the current wave of digitalization at the Danish public sector, data-driven decision-making, BI and business analytics are dominating the IT implementation projects in local municipalities. In such situation, the never-ending issues of data quality and data governance become outstanding problems and in many cases the show-stopper.
Motivated by this discovery and the semester theme, the three young man chose to dive into this topic of data quality and data governance for the local municipalities.
After iterations of discussion with the Digitaliseringsstyrelsen (the Danish Agency for Digitilisation Organisation) and a few other contacts, the research scope was settled on the challenges of data quality in the FLIS system. In the year 2012, the Danish organization Kommuneres Landsforening (KL, or Local Governance Denmark) launched the BI system called FLIS (Fælleskommunalt LedelsesInformationsSystem). The FLIS system collects different types of data from 98 local municipalities every month. These data, covering areas from schools, economy, senior citizen, healthcare, etc., can be used for different reporting and analysis purposes. One of the main business drive to use FLIS is when a local municipality uses measures and KPIs from other municipalities as benchmarks to identify places where it can potentially become more effective according to the other municipalities.
The research starts with interviews of employees from the KL organization and consultants from Rehfeld (i.e., QuintilesIMS, a well-known Danish consulting firm in the public and healthcare sectors). Three subject areas, data quality, master data and data governance are kept in focus in the theoretical discussion and analysis process where the facts and information collected from the interviewees are thoroughly inspected.
Seeing from the data quality angle, classical root-causes and issues remains classical in the public organizations. Different IT system vendors deliver in different formats and standards, which naturally leads the complexity to the integration process. In certain areas at a local municipality, it is difficult to find motivations and applications that helps with the data registration and validation. Just to demonstrate this, I personally would not like to write down too much information just for sending my daughter to a weekly swim course.
Different municipalities have different practices on registering data. That is a natural reflection of how different group of people understand “good quality” for data. These challenges become more explicit when there comes the need to consolidate data in different systems. For example, the financial report made by the local municipalities are almost always different from the financial result made by Statistics Denmark (considered as the national bureau of statistics for Denmark). Actually a very good question to ask here is “which result should I trust?”
Master data can easily be identified in the public sector. For example, demographics about citizens, health data, and information on the unemployment benefit. However, the challenge in the matter of master data lies in the timely integration and validation of the master data at different IT systems. In many local municipalities, the task of integrating the senior and handicap citizen data to the healthcare systems is too expensive to be done through a carefully-designed SOA architecture with a well-founded infrastructure. In such circumstances, building a SOA-based master data hub is unrealistic.
The FLIS project has established a set of standard practices of data governance. There are training programs, well-maintained guidelines and steering organizations for covering debates and discussions. Two features make the data governance at the municipalities stand out from the classical examples in the private sectors.
  • First, in terms of organizing training sessions and ensuring a unified organizational practice on data registration and data processing, the data governance program needs to cover a majority of the employee at all Danish municipalities. Note that by the year 2014, there are 500,000 employees in all Danish municipalities. Most private organizations will not be able to see such challenges in their data governance programs.
  • Second, from the viewpoint of the local municipalities, the data governance organization defined through the FLIS project is in fact an “external” organization that facilitates and coordinates the relevant processes. In general, the definitions of “ownership” and “stewardship” at the local municipalities is even harder to clarify compared to the private sectors where the VPs and C level executives can take always the ultimate roles.
The idea of using measures and KPIs from other municipalities as benchmarks for performance improvement and process optimization is smart and can be potentially very beneficial. In the pursuit of being more intelligent, cognitive and effective in social services through innovative IT applications, quality of data is crucial. Finding out the right initiatives for data ownership and creating an agile data governance process become the critical path for the success of data quality and the success of well-functioning public services.
Denmark is among one of the first few countries to totally digitalize the public sector. While bringing the new power of technology, the role of IT applications and the understanding of data value chain require a total refreshment in the public domain. The research work by Anders, Bo and Djaco shows a great step in KL and the FLIS project. I believe there will be a great need to new economic models to support the development and management of the data analytics and data-driven applications.
From the classical process of data warehousing to the new trends such as big data, data science, NoSQL, etc., there is always the need to ensure data quality, verify and validate master entities, align definitions and transformation rules for business logics. When it comes to the matter of one-version-of-the-truth for data, there is no workaround.