chevron icon Twitter logo Facebook logo LinkedIn logo YouTube logo download icon link icon audio icon quote icon posted icon clock icon author icon arrow icon arrow icon plus icon Search icon location icon location icon document icon menu icon plus-alt

There has been widespread debate among the community of practitioners and researchers around the transaction costs of impact bonds. Some believe that the costs of impact bonds have been higher than alternative commissioning arrangements. Yet what lies beneath this statement is more murky. Which costs are higher, and what are they higher than? How high is ‘high’? What drives the costs, and might this reduce over time? And crucially, do the benefits of the partnership model outweigh the costs – whatever they might be?

We came together to respond to these issues at the latest INDIGO Hack and Learn event. For the 2023 Spring edition, hackteam#28 worked on transaction costs and social impact bonds. The goal was to understand whether the transaction costs in government outcomes-based contracts are typically more or less than in more traditional government inputs-based contracting. And we wanted to describe the features of projects that might lead to greater or lesser transaction costs.

We reviewed academic papers that focus on these issues and found the approach proposed by Ole Petersen and colleagues useful for our challenge. Petersen et al. (2019) focus on the government ‘purchaser’s’ costs, defining transaction costs as the resources that governments expend on finding and selecting vendors, negotiating contract terms, monitoring vendor performance, and ensuring that the delivered product meets specifications. Government incurs two broad types of transaction cost: ex ante costs, such as scanning the market for potential vendors, evaluating formal bids and negotiating contract terms, and ex post costs, such as monitoring the performance of services, executing contract options or resolving disputes.

Hackteam#28 decided to use this framework as a lens to explore the different types of transaction costs associated with social impact bonds. One sub-team looked at two similar homelessness projects launched in Greater Manchester between 2017 and 2022. One was an outcomes-based contract, where the delivery partners were free to design and evolve their own intervention and organise their budgets in whatever way they found worked best.  Payment from government was made in arrears, according to the verified progress achieved by each participant. The second project was a traditional inputs-based contract, with an agreed ‘service specification’ and a pre-defined set of delivery principles. Here, payment from government was made in advance, according to a forecast ‘inputs budget’.  The project then tracked data on the progress made by each participant.   Payment made was not contingent on these progress measures, but standard enforcement and termination options were available if results fell significantly behind plan.

Shuff Tariq, Homelessness Transformation and Performance Manager, Manchester City Council explains: “… we wanted to trial a range of ways to help people experiencing homelessness.  We launched one project where the delivery consortium was paid for their input costs, to deliver an evidence-based intervention “Housing First”. We also launched another project where the delivery consortium was not obliged to adhere to any specific ‘intervention’, but was paid only for the outcomes they achieved.  The desired outcomes were the same - both projects aimed to help people move into safe homes and remain stable there as well as having access to the support they need.”

Building on the experience of the practitioners and policy makers in the team, we identified the following list of transaction costs. For those who are not familiar with the process of public service contracting, The Sourcing Playbook offers helpful context and guidance. These are the activities that government teams undertake to prepare for and manage each type of contract. And this is the work that accrues transaction costs.

Tables of transaction cost categories across inputs-based and outcomes-based contracts
Search GM inputs-based contract GM outcomes-based contract
Scanning the market for potential vendors Same process used for both contracts Same process used for both contracts
Developing product requirements and specifications Detailed service specification developed with preferred delivery consortium No service specification. The outcomes are specified in a ‘Rate card’ adapted from previous MHCLG projects
Incentivising or training potential bidders Same process used for both contracts Same process used for both contracts
Negotiation GM inputs-based contract GM outcomes-based contract
Evaluating formal bids Same process used for both contracts Same process used for both contracts
Conducting reference checks of proposers Same process used for both contracts Same process used for both contracts
Negotiating contract terms (e.g., method of compensation) Payments made against pre-defined inputs budget, agreed with preferred delivery consortium Payments made against progress of each person helped by the project
Monitoring GM inputs-based contract GM outcomes-based contract
Developing performance metrics Projects measured number of people safely housed, and how long they successfully sustained this safe home Projects measured number of people safely housed, how long they successfully sustained this safe home and some wider metrics (health, employment etc)
Gathering information from product users and other stakeholders Referrals made onto the project by local councils, but not verified by GMCA Referrals made by local councils, but then verified by GMCA to ensure they met the eligibility criteria
Assessing deliverables GMCA verified in monthly meetings that the service specification was being followed by the delivery consortium No prescribed deliverables. The delivery consortium is free to develop a bespoke intervention for each person
Enforcement GM inputs-based contract GM outcomes-based contract
Executing contract options or termination GMCA met monthly with delivery consortium to assess progress against objectives GMCA met monthly with delivery consortium to assess progress against objectives
Implementing performance incentives GMCA collected housing data from the consortium, but had no formal verification criteria or process GMCA collected housing and wider data from the consortium, and verified data against a pre-defined set of criteria. This was then used to vary the payment levels
Resolving disputes (negotiation, arbitration, litigation) No formal disputes arose No formal disputes arose

What did we learn, and what next?

For the inputs-based contract, more time and hence cost arises through government’s effort to define the service specification and inputs budget, and then to ensure that these were delivered as specified.  For the outcomes-based contract, more effort (and cost) is spent on formally verifying the eligibility criteria for the participants, and verifying the data on progress of each participant against a clearly defined set of criteria.  Additionally, a broader set of data on the progress made by each participant was collected, which could be useful to help future partnerships better understand how participants achieve the desired outcomes across time.

Shuff Tariq elaborates: Although these two approaches required a different commissioning focus at times, both required about the same amount of time and effort from us.  It's not true that paying for outcomes automatically increases the ‘transaction costs’ of launching a service.   But neither is it true that paying for inputs reduces the obligation to contract manage a service.  Whichever payment method is chosen, good procurement and contract management practice are key, equally as important is having good and transparent relationships with delivery partners. This is necessary to achieve the best outcomes for people.”

Our collaborative 'hacking and learning’ has advanced our understanding of transaction costs and offers a preliminary set of data items and definitions to describe the varied activities that might contribute to transaction costs. But there is still more work to be done. To understand how these costs stack up we will need to work closely with practitioners to understand the time and effort aligned to these different activities. Future research could investigate the relationship between transaction costs and the evolution of the impact bond market. As governments become more experienced in commissioning impact bonds and working with outcome funds, do transaction costs decrease? Does the availability of contract templates reduce transaction costs? We will continue this conversation with the community in our next Hack and Learn event in September 2023. We hope more practitioners and policymakers join us in this exploratory work.