BLOG | Rethinking Data Governance at UNDP

This blog was originally published on Medium by Andreas Pawelke, Gaia Bellavista and Shumin Liu

"Over the past months, however, we have seen not only how vulnerable our social, economic and health systems are, but how grossly inadequate the mechanisms, institutions and systems are that we rely upon to handle our data."

At UNDP, we’ve been experimenting with ways to harness new sources of data. The outcomes of many of these data innovation experiments are encouraging. Yet, these tactical innovations have largely operated within the existing data governance structures and processes — instead of challenging them.

Over the past months, however, we have seen not only how vulnerable our social, economic and health systems are, but how grossly inadequate the mechanisms, institutions and systems are that we rely upon to handle our data. The Covid-19 pandemic raises the big questions of the data age we live in — like who gets to have a say in how data is collected, used and shared for what type of decision-making and for what purposes. It’s time to find answers if we want to uncover and utilize the true value of data.

We need to rethink data governance

To that end we organized a series of webinars and internal workshops with six inspiring speakers (webinar 1, webinar 2) to discuss how to explore how “alternative” forms of data governance could be used by the UNDP, its partners and the development space at large.

Here’s what we learned:

1. Much of today’s data economy is of extractive nature, with data collection being more pervasive than ever before and large amounts of data being controlled by and benefiting the few, not the many. The economic and political power by those who control the data has created a lot of mistrust in current data ecosystems.
2. There’s an interesting paradox that has resulted from the way (personal) data is being handled at the moment: we observe too much and too little use of data at the same time, or more precisely, there is too much data misuse and mismanagement and underuse of data to address social, environmental and political challenges, with data being siloed and difficult to access for such purposes.
3. Data is relational: other people’s data can impact us as much as the data we generate ourselves. Using modern machine learning technologies, the data we generate individually can, when combined with data from others, be used to infer characteristics not just about ourselves but people like us. This calls for a shift from an individual to a collective responsibility burden and a need to rethink individual consent as the dominant mechanism we currently use to govern the use of the data we produce.
4. Data governance frameworks need to take into account three dimensions to adequately address questions around data control, access and sharing: 1) regulatory e.g. appropriate data privacy laws; 2) institutional e.g. new forms of data sharing mechanisms like data trusts; and 3) technological e.g. anonymization techniques to protect peoples’ privacy when data is shared.
5. Data stewardship models fulfill different functions and can be helpful for at least three things: 1) to empower individuals by providing effective control mechanisms over who gets to use different types of data and for what purposes; 2) to provide collective bargaining power to data subjects vis-a-vis data collectors; and 3) to bring about societal benefits by unlocking the potential of data for purposes other than just economic ones. A bottom-up, participatory approach to data stewardship is critical in unlocking the value of data while at the same time ensuring that individuals and communities are not just bystanders to their data use and sharing.
6. While not a one-size-fits-all solution to the challenges around data governance and data sharing, data trusts could be an effective mechanism to facilitate greater and improved data sharing, while at the same time empowering individuals and communities to address the data privacy vulnerability, notably by reversing the direction of consent. For a data trust to work, a certain level of data rights need to be in place. While everyone seems to love the idea of a data trust, there are many different understandings of how a data trust should operate and a range of open questions on its functioning such as the role and tasks of trustees.
7. Unequal power relations are inherent in many data ecosystems that often result in indigenous people not having control over their data. What’s needed is to ensure the right of indigenous people and nations to govern the collection, ownership and application of their own data. Such indigenous data sovereignty requires states to respect traditional values of local communities in the collection and use of data.
8. Various concepts, models and mechanisms for data governance exist, including personal data stores, data cooperatives or data trusts, and everyone seems to agree that we need to facilitate greater and improved data sharing. While no single model or mechanism will provide all the answers, we need to move from the theoretical frames into the real world and start building the capacity for practical testing.

The way forward

The perception of data has shifted from a means to generate value to a means to perpetuate unequal power structures, privileging private and for-profit entities over individuals. New mechanisms of data governance are emerging to reverse this trend and move towards greater empowerment, advocating for more decentralized systems to influence decision making.

There is no shortage of ideas when it comes to reinventing data governance, but practical experiences are few and far between.

At UNDP, we need to make use of our mandate, capabilities and networks to help the governments we work with to change the ways data governance is viewed, moving away from outdated, ineffective and unfair models to more inclusive and just ways of data collection, use and sharing.

Listen to our webinar recordings on reimagining data governance for inclusive and sustainable development and exploring emerging data governance models.

We would like to thank all the speakers for their inspiring inputs and the Rockefeller Foundation and the Government of Slovakia for their continued support.