Dr Thomas Wainwright
School of Business and Management, Royal Holloway, University of London
Tags: Proptech, letting agencies, private rental sector, data ethics
Output type: Think piece
Target stakeholders: Proptech entrepreneurs, letting agencies, landlords, tenant groups, government
"Openness and Collaboration" by psd is licensed under CC BY 2.0
What’s the issue?
Data use and ethical implications for tenants. As highlighted in an earlier post [here], one set of issues facing rental proptech businesses concerns how tenant data is used by start-ups and the resulting potential for unintended effects. While data can be viewed as neutral and objective, the reality is that the web of social relations which govern how it is collected and the context in which it is viewed and analysed makes the resulting information more subjective. The goes for algorithms and artificial intelligence (AI) too, where biases can be unintentionally encoded into decision-making tools, or where AI reproduces biases within the data that we provide it with.
Data-related problems and misuse have been highlighted widely by the media, from Facebook, though to charities. While businesses may follow GDPR as a statutory requirement, ethical issues and wider laws on discrimination can be potentially breached through data use, creating wider risks, too. Housing is a biological necessity – it is not a luxury add-on and accidental discrimination or unethical tech design can potentially have profound and wide-scale negative impacts on prospective and current tenants. As will be explored later, these also create substantial business risks. This is not to say that rental proptech is designed to be intentionally unethical - problems will be unintended. However, having an awareness of the risks and deciding to not consider potential mitigations is perhaps less forgivable.
So far, the rental proptech sector appears to have escaped a crisis or scandal, but then the sector is still relatively new and it will not be immune to these issues. The technologies beneath the platforms are established and commonplace, but remain problematic. For example, it is well-established that AI can reinforce and reproduce racism, sexism and ableism, through biases detected in natural language processing - beware of your chatbot. As rental proptech grows and the technology is adopted more widely by high-street letting agents, the scale of proptech risks can grow, too. Building on my earlier briefing, I seek to consider how some of the risks associated with new digital technologies could be mitigated, and to sketch out what a Tenant Data Charter for the sector could look like, and how it could be developed. This is not in any way to be considered as exhaustive, but as a potential starting point.
What are the risks and issues if no action is taken?
As highlighted above, AI and algorithm technologies have the potential to discriminate against prospective tenants. Similarly, data asymmetries and ‘patchiness’ can also discriminate against minority groups. For example, a high reliance on open banking data could place recent migrants, those on zero hour contracts, or tenants in the cash economy, at a disadvantage owing to there being less data available. In theory, anyone can use rental proptech platforms, but they are often designed around a particular rental demographic: framed around young, urban professionals, for example. If a product and the underlying tech is designed around one key demographic profile, this may disadvantage ‘other’ renters who may have to use the platform or app to access certain properties in their local area. This becomes more likely as proptech is adopted by high-street letting agents more widely.
There are a series of risks that could directly emerge. These include discrimination against minority groups and breaches of legislation, but also reputational risk. If a rental proptech business is seen to discriminate, reputational risk could result in a reduction of tenants using it, voting with their feet to use other services, or may see landlords list their properties elsewhere, for example. This would reduce user numbers, triggering revenue risk with reduced opportunities for cross-selling other services. If partners and clients, for example, high-street letting agents are seen to be using technology associated with discrimination, the reputational and revenue risks could spread to those stakeholders, triggering calls for recourse or cancellation of contracts. For many proptech businesses, especially those backed by venture capital funding, scaling is key to enhance returns and raise more capital. These risks would spill over to investors too, but would also likely draw broader attention from regulators, tenants and the media to practices in the sector as a whole.
On the flip side, greater transparency and promotion of activities to mitigate these risks could drive new opportunities and revenue. Tech-savvy renters could be reassured with clarity and reassurance over how their data is used and shared, and what steps are taken to avoid discrimination. This in turn could build greater trust and attract more tenants over competitors who are less transparent.
Recommendations for action and change?
There are several steps that can be taken to mitigate some of the highlighted risks. Actively checking and monitoring AI and algorithms and examining where bias may emerge and affect different groups needs to be regularly undertaken. Where there is limited understanding about this issue within the business, awareness and training needs to be enacted. Rather than focus on the business and risk in general, it may be useful to develop an internal data risk group, perhaps with non-executive director and tenant input, to examine data risks, to support this work and to look for discrimination introduced to the business by tech.
In checking websites and apps, few rental proptech businesses provide clearly accessible transparency information on tenant data. Some of this information is hidden amongst T&Cs, but the development of a clearer portal with information and explainers would be useful here. Particularly regarding what sources of data are used and what purposes the data is used for in different products and services across the platform. For example, who data is shared with externally and how can renters opt out of supplying some forms of data?
Access to individual data profiles would be useful, enabling tenants to review and to challenge incorrect data, or provide further data to address missing gaps to complete their profile, if they so wish, and if it increases the likelihood of securing a desired property. This feature is already commonplace with credit referencing agencies, which affects decisions on consumer credit and mobile phone contracts. Accessing information on what data is held and is used in decision-making on where you can rent, comparably seems like something that could be reasonably made available, yet is currently not (although perhaps through manual requests, but perhaps without context). While credit referencing agency products also give hints and guidance on how to improve the likelihood of being accepted for credit, could a similar feature not be made available on proptech firms for renting, too?
With the availability of new data sources that are increasingly more personal and detailed, it is unclear as to how some of the data is used fairly. For example, open banking data can be used in making affordability judgements, but it could also be used to profile a tenant’s lifestyle based on their activities, political and religious views. This level of detail moves far beyond what was previously available. However, does more data create more robust decisions, or new opportunities for misuse and discrimination? Explanations and justifications on what data is used, why and with potential opt outs to throttle unnecessary detail should be considered. After all, it is unlikely that many tenants will be aware of all the data collected about them, or that they would be comfortable with it either.
As the rental proptech sector matures, it would be useful to develop and share common principles and commitments to ethical data use and attempts to prevent discrimination and the associated risks. While there are already examples of ‘good practice’ in managing tenant data by rental proptech businesses, this would benefit the whole sector in sharing these practices, more widely, developing trust and legitimacy with tenants, and reassurance against risk for wider stakeholders.
Who could be involved in shaping a Tenants Data Charter?
Developing a Tenant Data Charter would require input from a genuinely interdisciplinary stakeholder group. It would need to be independent, inclusive and transparent to foster mutual trust. It would require support from representatives in the proptech sector and a diverse group of rental proptech businesses/early signatories, in addition to tenant groups, with perhaps independent academic expertise on how AI functions, legal knowledge and researchers who understand discrimination. This would provide a balanced range of stakeholder views and pool of expertise that could be used to develop objectives for the group, and later, core principles on tenant data, while developing and addressing questions on how data should be used/limited and transparency created around decision-making, for example.
It would be particularly important to determine, what transparency means to tenants, and how can it be best enhanced, for instance. From there on, early signatories can adopt the principles, adapt their businesses and promote the charter. The project would then develop further through a commitment to meeting code principles and the promotion of good practice examples, or perhaps through a graded approach, with minimum standards and higher grades for exceptional practice, which could be independently audited. Of course, there may be capacity and demand for different sets of principles and codes to accommodate the needs of different groups. However, where there is certainty, is that action is needed to manage these risks.
If this is of interest to you – do get in touch.
Resources: RED report - The growth of rental proptech: hybridisation and data risk
Funding: British Academy - Tackling the UK’s International Challenges
Acknowledgements: Comments from reviewers are much appreciated in clarifying and augmenting the key points raised in earlier drafts
 https://www.forbes.com/sites/forbestechcouncil/2021/02/04/the-role-of-bias-in-artificial-intelligence/?sh=4008afed579d  https://www.ftc.gov/news-events/press-releases/2019/07/ftc-imposes-5-billion-penalty-sweeping-new-privacy-restrictions  https://www.thirdsector.co.uk/charities-got-head-start-gdpr-ico-fines-says-dma-executive/fundraising/article/1465446  https://www.brookings.edu/research/detecting-and-mitigating-bias-in-natural-language-processing/#:~:text=Unsupervised%20artificial%20intelligence%20(AI)%20models,racism%2C%20sexism%2C%20and%20ableism.&text=Billions%20of%20people%20using%20the,exposed%20to%20biased%20word%20embeddings.  https://www.vice.com/en/article/akd4g5/ai-chatbot-shut-down-after-learning-to-talk-like-a-racist-asshole