By Prof Albert Sanchez-Graells, Professor of Economic Law (University of Bristol Law School).
Blockchain or, more generally, decentralised ledger technologies (DLTs), are capturing the attention of policymakers. ‘Blockchain’ has become shorthand to refer to technology usually identified with the properties of a decentralised, trustless and immutable (or at least, tamper-proof) mechanism for information verification and recording that can enable self-executing digital transactions between anonymous parties (‘smart contracts’).
Blockchain’s touted tamper-proofness and potential to enable smart contracts are driving initiatives that seek to create automated ‘trust in trustless environments’ for public sector use cases, in particular concerning activities highly-exposed to corruption risks and/or the automation of administrative procedures devoid of discretion.
There are high economic stakes at play in public procurement—which represents around 12% of GDP and over a third of public expenditure in OECD countries, and even higher proportions in other economies. Coupled with the growing (over)reliance of policymakers on business consultants, the hype around blockchain—and, more generally, about public procurement 4.0—is perhaps particularly intense in this field of GovTech and RegTech.
Some legal scholars are rather optimistically jumping on the ‘disruptive technologies bandwagon’ and identifying blockchain as a main tool to increase the probity and efficiency of procurement governance at a national level.1 Some officially-backed ‘visions for the future’ go as far as promising blockchain-supported global e-procurement platforms capable of covering the entirety of procurement transactions carried out worldwide.
This is creating a set of expectations about how blockchain will revolutionise public procurement governance that does not translate into real action. Even further, I submit, blockchain is and will remain structurally inapt to generate such a governance revolution, for several reasons.
Blockchains for the public sector
First, it should be stressed that the types of DLT that create tamper-proofness are designed as public, permissionless and economically-incentivised ‘blockchains’ (based on proof-of-work, PoW consensus mechanisms). The underpinning characteristics are thus those of transparency, decentralised open participation, and economic incentives to curb self-interest. However, this model does not appeal to the public sector.
It is difficult to envisage the adoption of a fully decentralised blockchain solution, not solely due to technical issues of scalability and latency, but also because the public sector can hardly be expected to surrender control over the procurement system.
Emerging evidence shows that:
[M]any ongoing experiments are built on permissioned chains, which may seem most applicable for the public sector as they allow control to be retained over who can record new transactions, all while enhancing accountability through identity management and transactional transparency’
(Ubaldi et al, 2019: 16).
Along the same lines, even if solely due to environmental considerations and the need to ensure the sustainability of the solution, the consensus mechanism would need to be different (at most, proof of stake, but most likely proof of authority). This would also erode the potential of the technology to avoid control and make the system prone to manipulation through centralisation of decision-making power.
However, the more control of the system by the public administration, the more space for distrust from civil society and the more scope for corruption. As a result, added value in the implementation of a blockchain-based information management system would be limited.
Second, the additional benefits expected from smart contracts running atop the blockchain are mostly illusory. These proposals largely amount to a ‘salami-slicing’ of traditional contracts in order to change their structure and make them meet a (very large number of) ‘if/then’ clauses that can then be automated and executed in a deterministic manner by an algorithm. However, this fails to acknowledge important limitations.
One such constraint is that this approach can exponentially increase transaction costs – in particular, the negotiation and drafting costs of the (smart) contract – and is doomed to fail because, crucially, ‘the obvious problem is that blockchains only work on complete contracts, whereas most in-the-world firms … are largely (entirely?) made of incomplete contracts’; ‘a blockchain is an economic world of complete contracts’
Another limitation is that blockchain technology is not guaranteed to be able seamlessly to connect to the ‘real world’. This is because blockchain technology can only enable the automated execution of smart contracts to the extent that it does not require the generation of off-chain effects, as this would require a further integration of an oracle. Oracles are data interfaces that connect a blockchain to a database or a source of data. This makes them potentially unreliable as the oracle can only be as good as the data it relies on and the oracle can itself be manipulated.
Blockchains and smart contracts also raise significant data protection compliance difficulties, which exacerbate the issue of transaction costs.2
The case for blockchain implementations in procurement remains largely dubious – as further developed in my recent paper. The advantages of the type of DLTs that are likely to suit the public sector (both in terms of governance, control and compliance requirements) seem quite limited compared with alternative sophisticated databases; and the potential for smart contract implementations does not significantly change the analysis. There is thus a risk that blockchain-based procurement solutions turn out to be new white elephants sucking up public resources that could be put to better use elsewhere.
This blogpost is based on A Sanchez-Graells, ‘Data-driven procurement governance: two well-known elephant tales’ (2019) 24(4) Communications Law 157-170. The paper is available for download on SSRN: https://ssrn.com/abstract=3440552.
1. See eg (Williams-Elegbe, 2018) and (Carvalho, 2019).
2. See eg (Finck, 2019a) and (Finck, 2019b) and, for extended discussion, (Finck, 2018).