APPG on the Rule of Law: The EU Settlement Scheme

Background

On Tuesday 16 July 2019, the APPG for the Rule of Law met to discuss the Rule of Law implications of the EU Settlement Scheme. The meeting was convened to discuss a new Public Law Project report, authored by Dr Joe Tomlinson, titled 'Quick and Uneasy Justice: An Administrative Justice Analysis of the EU Settlement Scheme'  which offers an end-to-end administrative justice analysis of the design, and thus the underpinning values, of the Scheme.

Dominic Grieve QC MP chaired a panel of expert speakers, and a wide audience from a variety of sectors: Parliamentarians, legal professionals, NGO and charity representatives, among others.

Speakers

Dr Joe Tomlinson began the discussion with an overview of his important and timely report. The report focuses primarily on processes, specifically how decisions are made and how individuals can challenge these decisions. The scheme represents a new way of handling immigration with automation at the heart of the model, with multiple implications for the surrounding system. This is the first time that automation in the public sector is mainstreamed. One of the most significant considerations from a Rule of Law perspective is that this modern scheme is being put into a 'traditional' system. There are numerous outstanding issues: the position regarding tribunal appeal rights is still unclear, with the speculative line being "no deal = no appeal", raising serious questions around administrative justice. Further, how judicial review will work in this context is still unknown. For example, how do traditional judicial review principles of law apply to this new type of decision-making? Dr Tomlinson ended his talk by noting the appeal of automated decision-making systems, particularly when the Government has to process large amounts of information, however we need to be aware of the dangers of the safeguards shifting under this pressure.

The second speaker, Bijan Hoshi, is a lead lawyer from the Public Law Project, heading a project on the Settled Status Scheme. Mr Hoshi discussed what the Public Law Project hopes to do over the next 12 months regarding the Scheme. Substantive work will be centred on four activities: capacity building and support work, research and monitoring, policy, and casework. As the Scheme takes advantage of automated decision-making processes, it is likely that there will be greater divergence in individual experiences of administrative justices. There is therefore a need to collect robust, empirical evidence on the operation of this new model and its effect on vulnerable people.

The third speaker, Jessica Simor QC of Matrix Chambers, discussed the risks of the system. She began her talk by noting that the Settlement Scheme arises out of an international treaty commitment that the United Kingdom has not yet ratified. An estimated three million people will be registering with the Scheme, therefore speed is essential, but the consequences of the risks are not fully known yet. However, what is certain is that those particularly affected by the scheme are already vulnerable groups: risk arises out of an inability to produce evidence and documents, possession of a smartphone, and difficulties with children registering. Ms Simor QC pointed to paragraph 38 in the report for an overview of these issues.

She further pointed out that data use in the Scheme may itself give rise to discrimination. Firstly in relation to poorer members of society, highlighted by the fact that the data relied upon comes from the Department of Work and Pensions and HMRC. The poorer in society are likely to rely on DWP data, which isn't as high quality as HMRC data. Secondly, women are likely to be affected by the recorded data as working tax credits and child tax credits are held by HMRC, and this data won't serve the Scheme. Discrimination is therefore a real possibility, and these groups are also less able to access mechanisms of redress.

Ms Simor QC's final point was on the difficulties of redress in a situation with an automated decision. In order to exercise rights of redress against a state decision, you need transparent decision-making, access to an independent court or tribunal, and access to representation. As to the first necessity, transparency will be difficult to secure if the decision-making process is based on an algorithm. The second point may be dependent upon ratification of the Withdrawal Agreement, without which there may be no right of appeal. For access to representation, there is already a limited availability of people and organisations to provide this, hampered by a lack of money and resources.

The final speaker was Dr Jennifer Cobbe, the Coordinator of the Trust & Technology Initiative at the University of Cambridge. Dr Cobbe observed that the EU Settlement Scheme is part of the ongoing trend of 'digitalisation' in the public sector. 'Increasing efficiency' is the commonly given justification for public sector automation. Many systems are, by their nature, opaque and unaccountable. They use complex statistical models, which are not usually audited. They are also known to have problems in reinforcing biases, which is not a consequence of the 'maths', but rather the consequences of the designers. Much of machine learning works by grouping people into characteristics.

Dr Cobbe continued that public authorities must be prepared to open themselves up to scrutiny, even in this rush to automate. Officials seem to be clueless as to whether or not systems are congruent with administrative law principles. Dr Cobbe argued that court processes and judicial review is the best way of handling complex systems. Instead, a comprehensive statutory framework for decision-making is a solid way forward, perhaps one that establishes higher standards for automated systems than we already do for humans.

Discussion

Following the speakers' contributions, discussions took place which highlighted various concerns about the advent of digital decision-making and the Rule of Law.

The first question concerned Dr Cobbe's contention that a case-based system may not be the best way to achieve justice: if not, what should we do? She responded that having judges getting to groups with complex algorithmic systems may not be the best way to do things. This is because the issues that are likely to be raised do not run so deeply with regards to individuals, but rather there are issues with the datasets and training data which entrench biases. If there is a problem, it is not so much one concerning the individual, but rather the whole system.

The use of data in the Settlement Scheme also raised questions. In particular, what would happen whereby data acquired for one purpose is used for another purpose? This is particularly pertinent in relation to data not covered by the Data Protection Act, such as immigration data. The consensus from the speakers and members of the audience also doing work on the Scheme is that at present, there is little to no information on how the Home Office plans to use the data, and we know very little about the way that data will be transferred. This draws into the broader concern of lack of transparency, as we also know nothing about how the algorithm reaches a decision.

Alberto Costa MP raised the question of discrimination: at what point may we see litigation, and under what point(s) of law? Bijan Hoshi answered that litigation is likely to be imminent. One example of a discrimination claim could be the fact that there is no real provision for people lacking in capacity to access the Scheme. Dr Joe Tomlinson also pointed out that there are practical questions concerning accessing judicial review: how does one build a case? If there is a failure in an automated system, at what point does the failure occur? If there's an issue with the system, is there an issue with the whole system?

A further point was raised concerning capacity to access the Scheme: what rights do prisoners have, and how could they sign up to the Scheme? According to Mr Hoshi, in theory, they can. But there doesn't appear to be any attempts to explain this to prisoners. The Home Office carried out several testing phases of the Scheme, and during one of these phases, organisations were urged to work with vulnerable people. Prisoners were not included in this. Mr Hoshi said that there was a risk that prisoners may be subjected to 'discrete' deportations at the end of their sentences. Dr Tomlinson said this possibility stressed the real need for an analysis of particular demographics.

Overall, the meeting was a success and provided a meaningful forum for debate and discussion on a live Rule of Law issue, yet another consequence of the UK's decision to withdraw from the European Union.