Secure Computing to Improve Data Access in PFS

Through our work with governments, we think critically about how social services are contracted. Data availability is often a limiting factor in the project design or in the evaluation of a project’s success. Many of our projects benefit from timely access to data in order to evaluate the unmet need, identify characteristics of the intended beneficiary population, and define outcomes. Additionally, lack of data availability results in high transaction costs for one-off data use agreements, and decreased data value due to lag times. Third Sector’s work is often dependent on sensitive data such as incarceration, health, or child data. Balancing data access and privacy is an ongoing issue.

In “Secure Computing Needs in Pay for Success Projects” we look at challenges that we face in using data and propose potential solutions.

In our work we’ve identified two main challenges that we face in using data. The first is to how to share data through approaches that structurally maintain privacy by providing access to only those fields approved to share. The second challenge is that the system restricts the use of data is to approved purposes, creates a log of data access, and is robust to unauthorized access.

Often, our use cases touch on both of these challenges. In one of our projects, we are working with the local government to achieve better outcomes through services designed to combat homelessness. In collaboration with the county and state data providers, we are using multi-level data to determine unmet needs, design impact goals, and monitor progress towards those goals. Given the sheer number of data sources and the intricacies in combining and matching data across datasets, we had to establish an elaborate and time-consuming framework to execute upon a multi-party data use agreement.

We have also found that as the data sensitivity increases, the longer it takes to get a data use agreement in place. Our projects are constructed to augment impact, which naturally leads us to gravitate towards areas of greatest need in a community – primarily underserved sensitive populations. Some examples of the data we are working to integrate are health care utilization, health care claims, reports of child abuse or neglect, homeless shelter stays, behavioral health utilization, wage and employment, and criminal justice involvement data.

Secure computing may allow the encrypted data to be analyzed, thereby removing the need to expose private information, reducing the risk of a data spill, maintaining privacy, and accelerating the project. The value of secure computing is maximized when multiple data sources are brought together to better understand those receiving benefits and the impact on their lives.

Potential Solutions

Sensitive data can be used securely through new technological approaches. Secure computing offers solutions by providing access to information while maintaining personal privacy on a secure system. Using secure computing or other approaches to make the data more available while protecting privacy rights is an important path to understanding which intervention models successfully break the intergenerational cycle of poverty we witness today.

We must keep in mind that tools are only part of the solution. A cultural adaptation to use the full power of the tools rather than maintain the same practices needs to occur. Significant efforts are necessary to allay concerns about privacy risks. Along with the development and deployment of secure computing tools, there is a clear need for leadership on establishing regulatory guidance and norms around secure computing, so that its value can be made tangible to government agencies. Only by incorporating the application of the data into the design can the system address these needs. Third Sector is ready to help our partners think through what is possible and evaluate options to realize that vision.

For a more in-depth look and to learn more about the challenges and potential solutions we propose, please read the full paper here.