Private Beta Phase
What is the purpose of the Private Beta?
The Private Beta Phase of the Project involved amalgamating data across five master trusts (L&G, Nest, now:pensions, The People’s Partnership and Smart Pension) with an overall objective of gaining insight into people’s savings patterns in multiple pots across these providers and specifically to:
identify and prove that the technology exists to merge data from across different providers legally, safely and securely using real data;
show the willingness of industry stakeholders to collaborate for the greater good of the UK
population and their savings for retirement;
highlight the importance of data-based evidence and its role in informing policy.
Alongside these overall high-level objectives, the Private Beta stage aimed to:
understand error rates in data matching;
understand the distribution of small pots and the implications for policy.
Initial funding for the Private Beta stage was provided by the Association of British Insurers (ABI), the Department for Work and Pensions (DWP), the Pensions UK and The Pensions Regulator (TPR).
Understanding error rates in data matching
Understanding error rates in data matching will be of particular importance to both the Pensions Dashboards Programme and the small pots solution. In particular, learnings about the quality of data, levels of matching across providers, and which data items can be seen to provide the most reliable match, will all be relevant when trying to match members across providers and trying to consolidate pots.
How the Private Beta stage was carried out
The scheme administrative data are by their nature commercially sensitive, and it would have been unsuitable for the providers to have any access to each other’s data, even in anonymised form. It was therefore decided to work with a third party that had the necessary capability and infrastructure to match and analyse the data. In this case, the third party was PricewaterhouseCoopers (PwC).
Prior to transferring the data to PwC the following steps were undertaken:
Step 1: Defining the data
In order to ensure that all providers supplied their data in the same format, a data scope was developed that outlined how the data should be formatted and what constituted a valid or non-valid data item. The three key personal identifiable data items were tightly defined, as due to the data being anonymous, it would be impossible to check why data had not matched. For example, due to different providers having different validations or system capabilities (e.g. some, but not all, allow for diacritics on surnames).
It was agreed that data would be provided for four financial years to both provide longitudinal findings. The data provided by the five master trusts encompassed the following years:
1st April 2019 to 31st March 2020
1st April 2020 to 31st March 2021
1st April 2021 to 31st March 2022
1st April 2022 to 31st March 2023
Step 2: Data protection, information security and transfers
The protection of individuals’ data rights was a central driver of the data approach. Not all providers had a suitable lawful basis for processing personal data for research purposes. Thus, to proceed with the project, the data needed to be anonymised, thus moving the research data outside the scope of the GDPR. This also needed to be done in way that would still enable PwC to match the data at the individual level. Based on guidance from the Information Commissioner’s Office (ICO), the following needed to be considered when anonymising the data:
The “motivated intruder” test
The grounds for legal access and processing of data by recipient