By Shobhit Shukla
On November 22, 2022, the Ministry of Electronics and Information Technology released India’s draft data protection law, the Digital Personal Data Protection Bill, 2022 (‘Bill’).* The Bill sets out certain situations in which seeking an individual’s consent for processing of their personal data is “impracticable or inadvisable due to pressing concerns”. In such situations, the individual’s consent is assumed; further, they are not required to be notified of such processing. One such situation is for processing in ‘public interest’. The Bill also illustrates certain public-interest purposes and notably, includes ‘credit-scoring’ as a purpose, in Clause 8(8)(d). Put simply, the Bill allows an individual’s personal data to be processed non-consensually and without any notice to them, where such processing is for credit-scoring.
Evolution of credit-scoring in India
Credit-scoring is a process by which a lender (or its agent) assesses an individual’s creditworthiness i.e., their notional capacity to repay their prospective debt, as represented by a numerical credit score. Until recently, lenders in India relied largely on credit scores generated by credit information companies (‘CICs’), licensed by the Reserve Bank of India (‘RBI’) under the Credit Information Companies (Regulation) Act, 2005 (‘CIC Act’). CICs collect and process ‘credit information’, as defined under the CIC Act, to generate such scores. Such information, for an individual, comprises chiefly of the details of their outstanding loans and history of repayment/defaults. However, with the expansion of digital footprints and advancements in automated processing, the range of datasets deployed to generate credit scores has expanded significantly. Lenders are increasingly using credit scores generated algorithmically by third-party service-providers. Such agents aggregate and process a wide variety of alternative datasets relating to an individual, alongside credit information – these may include the individual’s employment history, social media activity, and web browsing history. This allows them to build a highly data-intensive credit profile of (and assign a more granular credit score to) the individual, to assist lenders in deciding whether to extend credit. Not only does this enable lenders to make notionally better-informed decisions, but also to assess and extend credit to individuals with meagre or no prior access to formal credit.
While neither the Bill nor its explanatory note explain why credit-scoring constitutes a public-interest ground for non-consensual processing, it may be viewed as an attempt to remove the procedural burden associated with notice-and-consent. In the context of credit-scoring, if lenders (or their agents) are required to provide notice and seek consent at each instance to process the numerous streams of an individual’s personal data, the procedural costs may disincentivise them from accessing certain data-streams. Consequently, with limited data to assess credit-risk, lenders may adopt a risk-averse approach and avoid extending credit to certain sections of individuals. Alternatively, they may decide to extend credit despite the supposed inadequacy of personal data, thereby exposing themselves to higher risk of repayment defaults. While the former approach would be inimical to financial inclusion, the latter could possibly result in accumulation of bad loans on lenders’ balance sheets. Thus, encouraging data-intensive credit-scoring (for better-informed credit-decisions and/or for widening access to credit) may conceivably be viewed as a legitimate public interest.
However, in this post, I contend that even if this were to be accepted, a complete exemption from notice-and-consent for credit-scoring, poses a disproportionate risk to individuals’ right to privacy and data protection. The efficacy of notice-and-consent in enhancing informational autonomy remains debatable; however, a complete exemption from the requirement, without any accompanying safeguards, ignores specific concerns associated with credit-scoring.
Deemed consent for credit-scoring: Understanding the risks
First, the provision allows non-consensual processing of all forms of personal data, regardless of any correlation of such data with creditworthiness. In effect, this would encourage lenders to leverage the widest possible range of personal datasets. As research has demonstrated, the deployment of disparate datasets increases incidences of inaccuracy as well as of spurious connections between the data-input and the output. In credit-scoring, historical data using which the underlying algorithm is trained may conclude, for instance, that borrowers from a certain social background are likelier to default in repayment. Credit-scores generated from such fallacious and/or unverifiable conclusions can embed systemic disadvantages into future credit-decisions and deepen the exclusion of vulnerable groups. The exemption from notice-and-consent would only increase the likelihood of such exclusion – this is since individuals would not have any knowledge of the data-inputs used, or the algorithm using which such data-inputs were processed and consequently, no recourse against any credit-decisions arrived at via such processing.
Second, the provision allows any entity to non-consensually process personal data for credit-scoring. Notably, CICs are specifically licensed by the RBI to, inter alia, undertake credit-scoring. Additionally, in November 2021, the RBI amended the Credit Information Companies Regulations, 2006, to provide an avenue for entities (other than CICs) to register with any CIC, subject to the fulfilment of certain eligibility criteria, and to consequently access and process credit information for lenders. By allowing any entity to process personal data (including credit information) for credit-scoring, the Bill appears to undercut the RBI’s attempt to limit the processing of credit information to entities under its purview.
Third, the provision allows non-consensual processing of personal data for credit-scoring at any instance. A plain reading suggests that such processing may be undertaken even before the individual has expressed any intention to avail credit. Effectively, this would provide entities a free rein to pre-emptively mine troves of an individual’s personal data. Such data could then be processed for profiling the individual and behaviourally targeting them with customised advertisements for credit products. Clearly, such targeted advertising, without any intimation to the individual and without any opt-out, would militate against the individual’s right to informational self-determination. Further, as an RBI-constituted Working Group has noted, targeted advertising of credit products can promote irresponsible borrowing by individuals, leading them to debt entrapment. At scale, predatory lending enabled by targeted advertisements could perpetuate unsustainable credit and pose concerns to economic stability.
Alternatives for stronger privacy-protection in credit-scoring
The above arguments demonstrate that the complete exemption from notice-and-consent for processing of personal data for credit-scoring, threatens individual rights disproportionately. Moreover, the exemption may undermine precisely the same objectives that policymakers may be attempting to fulfil via the exemption. Thus, Clause 8(8)(d) of the Bill requires serious reconsideration.
First, I contend that Clause 8(8)(d) may be deleted before the Bill is enacted into law. In view of the CIC Act, CICs and other entities authorised by the RBI under the CIC Act shall, notwithstanding the deletion of the provision, continue to be able to access and process credit information relating to individual without their consent – such processing shall remain subject to the safeguards contained in the CIC Act, including the right of the individual to obtain a copy of such credit information from the lender.
Alternatively, the provision may be suitably modified to limit the exemption from notice-and-consent to certain forms of personal data. Such personal data may be limited to ‘credit information’ (as defined under the CIC Act) or ‘financial data’ (as may be defined in the Bill before its enactment) – resultantly, the processing of such data for credit-scoring would not require compliance with notice-and-consent. The non-consensual processing of such forms of data (as opposed to all personal data), which carry logically intuitive correlations with creditworthiness, shall arguably correspond more closely to the individual’s reasonable expectations in the context of credit-scoring. An appropriate delineation of this nature would provide transparency in processing and also minimise the scope of fallacious and/or discriminatory correlations between data-inputs and creditworthiness.
Finally, as a third alternative, Clause 8(8)(d) may be modified to empower a specialised regulatory authority to notify credit-scoring as a purpose for non-consensual processing of data, but within certain limitations. Such limitations could relate to the processing of certain forms of personal data (as suggested above) and/or to certain kinds of entities specifically authorised to undertake such processing. This position would resemble proposals under previous versions of India’s draft data protection law, i.e. the Personal Data Protection Bill, 2019 and the Personal Data Protection Bill, 2018 – both draft legislations required any exemption from notice-and-consent to be notified by regulations. Further, such notification was required to be preceded by a consideration of, inter alia, individuals’ reasonable expectations in the context of the processing. In addition to this balancing exercise, the Bill may be modified to require the regulatory authority to consult with the RBI, before notifying any exemption for credit-scoring. Such consultation would facilitate harmonisation between data protection law and sectoral regulation surrounding financial data.
*For our complete comments on the Digital Personal Data Protection Bill, 2022, please click here – https://bit.ly/3WBdzXg)