Working Paper
Beyond Eliminating Waste: The Role of Incentives in Outcomes-Based Finance[1]
By Stuart Theobald[2]
[Download this paper as a PDF]
Abstract
Outcomes-based finance (OBF) eliminates waste by ensuring public payments only occur when verified outcomes are achieved. This feature is well established in the literature. This paper aims to extend the analysis by assessing the incentive effects on suppliers, which can reduce the unit cost of outcomes beyond waste elimination. Drawing on project finance principles for risk allocation, I demonstrate how properly designed OBF structures create incentives for innovation, efficiency, and superior risk management.
Using the South African Jobs Boost Outcomes Fund as a case study, this paper provides empirical evidence of these propositions. The programme’s competitive procurement process attracted over 100 applicants, with final cost-per-job ranging from R35,757 to R164,802 across sectors. Analysis of three implementation partner failures illustrates how risk allocation drives behavioural change and performance optimisation.
The findings suggest that an important value driver for OBF, in addition to selective efficiency in programme choice, is the creation of an incentive architecture that fundamentally alters how implementation partners approach service delivery. Optimal risk allocation, placing risks with parties best able to influence, manage, or absorb them, proves critical for minimising overall programme costs.
The paper contributes to OBF literature by shifting focus from payment mechanisms to incentive design, demonstrating how properly structured programmes can achieve both improved outcomes and reduced unit costs.
Introduction
Recent literature on the role of outcomes-based finance (OBF) vehicles emphasise their ability to enhance effectiveness and impact, ensure value for money, and strengthen accountability and transparency in delivering public services (OECD 2025). In this paper I aim to complement the analysis of OBF by assessing the impact on the incentives and behaviour of suppliers[3] from whom OBF schemes procure outcomes. My argument will show that these incentives are fundamental to the value-for-money, impact and effectiveness that OBF structures can provide. It is critical that the incentive effects of OBF designs are carefully calibrated to maximise value. I will refer to the Jobs Boost Outcomes Fund to illustrate the argument.
Cost effectiveness in OBF
Claims regarding the cost effectiveness of OBF are generally premised on eliminating wastage in public spending on social services. The early examples of OBF vehicles, social impact bonds, were promoted with the explicit objective of delivering tangible public financial savings (Mulgan et al. 2011). By tying spending to results rather than inputs and activities through traditional grant-based funding, money is only spent when verified outcomes are achieved. This means that public resources are not wasted on spending that fails to deliver results. These savings must be balanced against the transaction costs and complexity of OBF schemes which can involve numerous stakeholders and extensive negotiation, making them complex to design and costly to implement (GSG/EOF 2021). Over time, however, these costs have reduced as models have been standardised, and parties become more familiar with OBF structures. Therefore, there is general agreement that OBF, where suitable, achieves better value for money than equivalent activities-based funding structures that are intended to achieve the same outcomes.
Recent OECD analysis (OECD 2025) argues that OBF ensures resources get to interventions that work, thus stretching public funds. OBF appeals to those managing public funds because it can help maximise the impact of each unit of expenditure.
These arguments focus on the benefits that arise from selective efficiency, i.e. that in the spending of public money, selecting projects that use an OBF framework in place of an activities-based framework causes an improvement in the overall results of public spending. The argument is that OBF delivers greater efficiency and value-for-money than equivalent activities programmes, and therefore there will be better results for public spending if spending is to such schemes rather than alternatives.
The value-for-money that arises on this conception does so from eliminating wastage. This is a clearly an important consideration in weighting up OBF versus traditional spending approaches.
However, a considerable driver of value for money and effectiveness is the incentive effects on the suppliers that are funded through OBF schemes. This aspect has been relatively little analysed. The claim I will argue for is that the savings do not arise only from the fact that the state does not pay in the event of the failure of a supplier to deliver outcomes, but that savings also arise because the unit cost of the outcome is lower when OBF schemes are optimally designed.
Incentives and behaviour of implementation partners
OBF shifts the risk of implementation failure from the state to the supplier. Should the supplier partner fail to achieve the outcomes, it will not receive payment. Given that the partner is now carrying this risk, an important part of OBF is to give the partner autonomy to manage risk (World Bank 2020). Some authors also point to the role of increased competition and private sector terms and conditions that can reduce costs (GO Lab, n.d.).
The cost impact of risk allocation depends on the ability of the risk holder to manage that risk. The state has the biggest balance sheet and therefore has considerable risk-baring capacity. However, in managing the risk that activities and inputs will result in the desired outcomes, non-state implementing partners may be better able to reduce the probability of outcomes failure through innovation, agile programme delivery, and upfront contracting that ensures outcomes follow the activities that are implemented.
In designing an OBF scheme, the priority consideration should be to allocate risks to whichever entity is best able to manage them. That effectively reduces the cost of achieving the outcomes. Private sector implementing partners may be more agile in responding rapidly to an event that threatens successful attainment of outcomes. For example, a candidate for an employment OBF scheme may be a victim of crime and need support to return to work, which an agile implementing partner is incentivised to do to ensure the outcome is still obtained. However, an implementing partner has little control over other risks such as the outbreak of a pandemic leading to lockdowns, or varying exchange rates that affect the cost of key inputs.
The right approach to risk is that which leads to the lowest cost of producing the outcome. Literature from project finance sets out some approaches to allocating risk that can be adopted in the OBF context. The following principles are adapted from Irwin (2007). The lowest cost is likely to arise when risk is allocated to the party most able to:
Influence the likelihood of the risk occurring. An implementing partner can affect the risk of outcomes being achieved in many ways, starting with the choice of what inputs and activities to undertake in order to obtain the outcomes. Partners can study and assess particular contexts and build experience in managing programmes that maximise the probability of successful outcomes. They can also innovate and develop new delivery mechanisms that are lower cost. They can contract with partners upfront to ensure that the outcomes will be achieved following agreed inputs and activities. Furthermore, partners can also manage some kinds of unexpected risks during the implementation phase, such as a job candidate requiring an unexpected intervention, or a medical intervention recipient encountering an unanticipated contraindication. Partners can be agile in responding to such unexpected events and ensure that outcomes are still achieved.
Influence the cost to the project should a risk occur. No one can anticipate a pandemic, but an OBF programme can be designed to minimise the consequences should one occur, for example by ensuring there is a backup all-online delivery mode that can be adopted quickly if needed. Making implementing partners responsible to adapt in the face of a set of definable eventualities can lead to better risk management for those. However, the cost of preparing to manage a risk can be high, and it may be cheapest to remove such risks from implementing partners depending on the value of the outcomes to the state.
Absorb the risk. If the likelihood of risks cannot be controlled by either the implementing partner or state, then it should be allocated to the party best able to absorb the risk if it occurs. The relevant factors will be the size and features of the revenue of the risk-carrying party, including the correlations with other activities of that party. An implementation partner may be able to insure risks with third party insurers at lower cost than the state absorbing the risk. It may also have diverse non-correlating revenue streams. Some risks may damage the probability of delivering outcomes but benefit the implementing partner in other ways. For example, a pandemic may lead to a surge in demand for new vaccine distribution but disrupt an existing OBF vaccine programme. The implementation partner may suffer negligible net harm from failing to deliver the outcomes.
The state should be the risk carrier of last resort, where the implementing partner cannot influence or absorb the risk. Such risks may be catastrophes, civil unrest, terrorism, power grid failures, some categories of economic crises, and changes in laws that make the outcomes illegal or unobtainable. If the state does not absorb these risks, the implementing partner has little choice but to price for them, resulting in an outcomes price that is inefficient (given that the state could absorb the risks at lower cost). The OBF design can also lessen the risks by having several outcomes milestones, spreading cashflows.
Implementation partners will also be conscious not only of the operational risks that may lead to failure to deliver the outcomes, but also of the risks presented by the outcomes payer. Given that outcomes payments will become due at some point in the future, only after the implementing partner has incurred costs, the partner faces credit risk. Outcomes payments may not be made when due because of administrative failures or because of political or policy change. In our experience, implementation partners are much more conscious of this risk and it is a material contributor to pricing. The public sector can improve the pricing of outcomes by actively reducing this risk, for example by placing the full outcomes funding budget in trust with a third party and ensuring there is automatic payment on presentation of outcomes audit verification or other agreed evidence.
In practice, however, OBF contracting provides limited risk assessment and allocation detail. Apart from a few scenarios, implementing partners do take on risks that they could not absorb were they to materialise. In place of detailed risk plans, OBF programmes have explicit or implicit flexibility to adapt in the face of unexpected events and partners may well price for the implicit risk reallocation back to the state were catastrophic events to occur. On balance, however, costs are ultimately lower when risks are transferred to the implementing partner.
Creation of surplus
From an implementation partner’s perspective, the risks faced must be justified by the potential returns, whether financial or the achievement of impact objectives (or social return). Whether the implementing partner is a for-profit service provider, or a non-profit charity, the OBF model must deliver outcomes payments (or intangible returns) that justify the risk taken. In line with expected utility maximisation, implementation partners participate because the returns, discounted for risk, justify the investment required to produce the outcomes. The returns can be financial or social returns. Indeed, one review of the first five years of social impact bonds found that most implementation partners and service providers are motivated by opportunity to scale up a successful intervention or deliver social returns while none cited financial returns or savings as a motivation (Gustafsson-Wright et al. 2015). However, as OBF matures we should expect that service providers will adapt to manage risk, and financial motives will become more common.
Given that an OBF programme fixes outcomes payments and leaves the method of delivering outcomes open, partners have scope to generate a surplus by reducing the cost of delivery. The pursuit of surplus is seldom explicitly considered in outcomes fund design. Even if the implementation partner is a charity aiming to maximise an impact outcome (or social return), the possibility of generating surplus funds from an OBF programme enhances their ability to cross subsidise other activities in maximising impact overall. Both for-profit and non-profit providers therefore have an incentive to generate a financial surplus, or profit, from OBF programmes. This incentive creates other risks that the public sector must manage, including the risk of achieving outcomes but skirting on implicit expectations such as the quality of programmes. To manage this risk, delivery standards must be clear and able to be monitored, without conflicting with the freedom given to implementing partners to innovate.
To our minds, the limited consideration of surplus reflects the approach common in the literature of seeing OBF as an alternative to activity-based grant funding, in which the funding quantum is determined essentially on a cost-plus basis. Grant applicants set out the costs they face, include a pre-determined margin if any, which determines the grant amount. This amount is constant irrespective of the outcomes achieved. In the transition to an OBF scheme, outcomes are priced “as if” the implementation partner was taking a grant-based approach with a “reasonable” expectation of outcomes derived from evaluations of previous programmes. A grant-based skills programme, for example, will have funded specific skills interventions and, upon evaluation, have created a certain number of jobs. This becomes the benchmark for an OBF programme, with the state benefitting from the reduced risk of outcomes failures. Early OBF programmes were often set up with the existing service provider being migrated from a grant contract to an OBF contract (for example, New Zealand took this approach, see Ramasamy and de Boer 2004).
Where a surplus arises, it does not harm the state which is benefitting from efficiency improvements and reduced risk. However, the distribution of surplus can bring additional value-for-money and efficiency. Furthermore, the OBF scheme can be calibrated to maximise other policy objectives including the specific type of outcomes that should be prioritised.
To maximise the value-for-money, impact and efficiency of the OBF programme, it is important that some level of competition is introduced in the procurement process. This allows potential implementation partners to offer prices for outcomes that reflect their risk-adjusted returns requirements without accumulating excessive surplus. However, any OBF procurement process must manage the risk to the state of outcomes failure. It is therefore important that procurement is not only driven by price but has strong tests for implementation partner capability of delivery as well as other non-financial features of the delivery approach. An approach that overly centres price risks appointing partners who fall victim to the “winners curse” having under-priced to win and then finding themselves financially unable to deliver.
The above discussion sets out the motivation for the approach adopted for the Jobs Boost Outcomes Fund, which provides a case study of these considerations in action.
Jobs Boost Outcomes Fund
The Jobs Boost Outcomes Fund was developed by Krutham in partnership with the South African Presidential Youth Employment Intervention. The PYEI had taken the view that OBF approaches could improve the efficiency of public spending on skills development to create jobs. South Africa has a deep unemployment crisis that particularly affects the youth (StatsSA 2025). An OBF programme would complement several other interventions overseen by the PYEI. Krutham was appointed to design and implement the scheme. South Africa’s National Skills Fund become the sole outcomes funder for the first phase, which was positioned as a pilot, with funding of $18m.
In developing Jobs Boost, extensive consideration was given to using competitive dynamics to maximise value-for-money and impact. The programme began activities in 2023, with a first step intended to maximise market interest and develop understanding of the OBF approach, through series of public workshops. These workshops detailed the OBF approach, the evaluative approach that Jobs Boost would apply in selecting partners, and guidance on appropriate management of an OBF project. Various guides and Q&A documents were provided to participants to maximise understanding and enable the widest participation possible. To further maximise participation, a two-phase procurement process was designed with a low-cost (to the potential supplier) initial expression of interest phase launched on 20 November 2023 (Jobs Boost 2023), followed by a closed request for more detailed proposals from qualifying partners. No restrictions were placed on the type of organisation that may bid to join the programme, except that service providers could not themselves be the employer – they strictly had to be an intermediary. This provision aimed to ensure sustainability of the jobs created because the employer must be independent with a clear economic motive to hire the jobseeker (rather than be subsidised in providing a job). The market building workshops provided for two-way feedback, and some programme parameters were adjusted based on market feedback particularly related to how potential implementation partners were perceiving the risks. Through this process we developed clear insight into how risks were being managed, particularly the risks of recidivism during the programme making it difficult to achieve the final outcomes milestone (which was sustained employment for six months).
This approach led to extensive interest from the market. Over 100 organisations responded to the submissions of interest. Of these 48 were deemed able to deliver on the programme in terms of the evaluator’s assessment of the feasibility of the proposed approach. The evaluators selected 34 organisations for the closed request for proposals. Selections were based on the historic exposure to the sector, organisational capacity, alignment of jobseeker selection mechanism to the parameters of Jobs Boost, articulation of the market demand for jobseekers emerging from the programme and the probability of successfully placing jobs seekers into jobs that meet the parameters of the programme. Of these, 33 submitted proposals. There was also consideration of the proposed size of the intake, with all bidders capped at a maximum of 2,000 candidates and a minimum of 100. This cap was expected to ensure there were at least five implementation partners, based on estimates of the total number of outcomes, while the lower cap ensured some scale effects in performance management to make it cost effective. The cap was a design feature to enable risk management during programme implementation in that it made it possible to reallocate budget if partners fell short of outcomes targets (which proved valuable when the programme was implemented).
More detailed proposals were requested from the 34 invited to submit by the deadline of end February 2024. Bidders had to complete 163 questions in the online RFP submission process. A detailed RFP guide was provided to those invited to bid with significant detail, including an explanation of the economics of a OBF with examples. The examples anchored bidders around certain price expectations, but no price caps (e.g. maximum or minimum price) were given to bidders.
The RFP was structured as a “reverse auction” with a set total budget advertised and bidders then submitting prices per outcome. However, all bids were also assessed on non-financial factors using a scorecard with the following key elements:
| Dimension | Maximum score | Description |
| Additionality evidence
a) Additionality to job seeker b) Additionality to job market
Jobseeker flow |
15
(8 for additionality, 7 for flow) |
The additionality test had two aspects – additionality in that the jobseekers targeted would not be able to obtain a job without the intervention and that the intervention creates jobs that would not otherwise be created. Bidders could focus on either (or both).
The flow test assessed the realism of dropout assumptions. |
| Delivery
|
30 | This assessed the proposal’s strategy for selection and screening of candidates, enrolment process, sourcing of employment, inclusivity and transition of candidates through the programme. |
| Non attendance and complaints | 10 | The proposal needed a clear strategy for limiting attrition and for managing complaints. |
| Sustaining employment | 20 | Approaches to ensuring jobseekers stay in their jobs once placed including in-work support and relationship management |
| Job quality | 10 | A minimum requirement was that jobs were full time, formal sector paying above minimum wage. However, additional points were given if jobs significantly exceeded the minimum for example in expected salaries. |
| Target demographics | 10 | Additional points were given if the younger end of the 18-34 youth range was targeted, a higher ratio of women to men, more candidates were rural, and candidates had disabilities. |
| Staffing to deliver | 10 | Applicants had to demonstrate that they had a resourcing plan to match the needs of the programme with a performance management plan. |
| Premises | 10 | There needed to be a clear plan on how to put physical infrastructure in place to deliver the proposed programme. |
| Quality assurance and audit | 15 | There needed to be a clear data collection and assurance plan in place. |
| Governance | 10 | There needed to be a clearly defined governance structure to ensure integrity and performance. |
| Data authenticity | 20 | There were systems in place to ensure that data is maintained securely and made available for audit. |
| Risk management | 10 | Bidders needed to set out their risk management plan detailing the risks they face and their plan to manage these. |
| Implementation plan | 10 | There needed to be a detailed and pragmatic implementation plan across the milestones. |
| Funding plan | 20 | The organisation has a clear plan for how it will manage the cashflow needs of an OBF programme. |
| Total | 200 |
Several scorecard dimensions had minimum requirements and bidders were excluded if they did not meet them, irrespective of the rest of the dimensions. This included a strict requirement that the eligible job seekers meet the programme’s definitions which included deprivation indicators of having attended a fee-free school (which are the most under resourced schools in South Africa) or having been a recipient of a means-tested child support welfare grant.
Proposals were then mapped against the overall targets of the programme, including its need to achieve geographic diversity across South Africa and meet certain demographic targets (e.g. that 5% of the jobseekers have disabilities). This meant that proposals might be accepted because they supported the overall programme objectives, while not necessarily scoring highest in the evaluation.
The scorecard result was considered alongside the value-for-money review that assessed the price per outcome implied by the grant request. There was no hard approach for balancing these, which enabled the adjudicators to optimise for the $18m budget. In practice, if the bids submitted added up to vastly more than the budget, then price would be more of a relevant factor in the adjudication, but if bids were more limited and prices tightly clustered then the scorecard would have greater weighting. In the event, budget requests from the 33 submissions totalled $48.5m, significantly in excess of available budget, so scorecard elements and relative pricing played an important role in adjudication. The cost-per-job among the bids received ranged from R17,612 ($1,057) to R365,000 ($21,908).
The uncertainty over how price would be balanced against the scorecard meant that bidders were incentivised to maximise for both. In our view, bidders did respond to price pressure but also focussed on meeting the non-financial assessment indicators. Short-listed bidders were also subject to a due diligence process. The scorecard process also made bidders take a more deliberate approach to assessing risk given they were required to set out a risk mitigation plan. In our analysis the ex-ante margins of implementation partners were relatively low and we did not detect any examples of excessive surplus being generated. We attribute this in part to the market building and guidance that emphasised that cost would be an important consideration in adjudicating competing proposals.
The submissions provided for a level of “price discovery” for different programmes, roles and demographics. We believe this provides valuable policy information, another benefit of OBF approaches. Jobs Boost was deliberately agnostic on the sector for employment, although the scorecard biased responses in favour of rural, women and people with disabilities where the unemployment challenge is highest. This design feature embedded the demand-led principle for skilling activities, allowing a bottom-up discovery of the sectors where jobs could be created.
Following the adjudication, 12 implementing partners were selected. In some cases, targets and budgets were negotiated with partners to achieve the overall programme geographic and demographic objectives. The table below outlines the key features and pricing of the cohort of implementation partners selected:
| Partner number | Final milestone target number* | Average grant per fourth milestone* | Programme description |
| 1 | 1000 | R47,148 | Various sectors including agriculture, waste management, renewable energy, retail and sales, hospitality and tourism, education and care work, vocational, and logistics, manufacturing, and administration. |
| 2 | 102 | R61,380 | Young girls for roles in beauty kiosks in department stores. |
| 3 | 227 | R112,331 | Specialised artisans in various trades (electrician, solar installation, plumbing). |
| 4 | 1,635 | R43,353 | Training and placement in call centre roles. |
| 5 | 54 | R57,856 | Training and placement as cannabis consultants in dispensaries. |
| 6 | 100 | R164,802 | Information technology training and placement, largely as coders. |
| 7 | 500 | R49,000 | Training and placement of young people into retail, manufacturing and other roles. |
| 8 | 324 | R60,000 | Training of lay mental health counsellors and placement into private sector and public sector roles. |
| 9 | 840 | R59,000 | Training and financing of capital equipment for last mile delivery riders using green mobility. |
| 10 | 57 | R63,579 | Trade and manufacturing using a work integrated learning model. |
| 11 | 470 | R35,757 | Hospitality roles in hotels and restaurants. |
| 12 | 106 | R41,457 | Retail and restaurants. |
| 5,415 | R53,134 |
*Outcomes were spread over four milestones: Enrolment (20%), placement (40%), three months sustained employment (20%), six months sustained employment (20%). The final milestone was used for targets, but with expected drop out rates, actual job placements were expected to be close to 7,000 and approximately 8,000 enrolled into training. The programme had a target of 4,500 candidates reaching the final milestone.
Partners were notified of acceptance in May 2024 and contracting concluded in July 2024 (with one outlier concluded in August). Programmes began as soon as contracting concluded.
The timelines for partners ran to final milestones being obtained in December 2025. This meant that training could be of a maximum duration of 12 months, before the start of the six months sustained employment period kicked in.
Various observations can be made about the RFP outcomes and grants per job. Generally, as expected, prices were lower for less skilled jobs, including last-mile delivery, retail and cleaning roles. Programmes in these areas focused on work readiness and on-the-job support, given that most candidates had never worked before and often had no concept of basic expectations such as punctuality. The six-month milestone meant that implementation partners had strong incentives to ensure candidates could not only obtain the job but keep it. The highest prices were for roles in IT and artisanal trades, which also had the longest training period. The lowest priced role was R35,757 and the highest R164,802. Generally, rural jobs were more expensive and focused on care jobs, agri-processing and hospitality, with the cost of delivering training the main driver given that partners were usually based in urban centres.
At the time of writing, data was emerging regarding the placements. The programme has run on target for overall placements and was set to deliver 5,400 candidates reaching the final milestone, in excess of the initial 4,500 target. The graphs in the data panel show overall features of 6,217 placements that had been recorded as of early August.
The risk management procedures designed into Jobs Boost have been extensively used. The variety of implementation partners provided wide scope for knowledge sharing. After the competitive procurement process, most implementation partners were quite comfortable sharing knowledge with others. The performance management team arranged regular knowledge sharing sessions where partners could air challenges they were facing and hear ideas from others. For example, several partners struggled with obtaining data and documentary evidence from jobseekers after they’d been placed into jobs. Some had developed innovative incentive reward programmes to encourage document delivery, innovations which other service providers could adopt.
When it came to risk of outcomes failure, several of the 12 service providers experienced challenges. Three in particular were unable to meet target outcomes. These failures, however, were managed through the programme by reallocating budget to other service providers who had exceeded targets. In this way, overall programme performance was maintained.
Those service providers who failed to achieve targets did carry the costs as a result, insulating the state. The examples are instructive.
USAID withdrawal. One service provider had intended to place trained youth into community care roles at predominantly USAID-funded organisations. When USAID froze funding world-wide in January 2025, the service provider could no longer achieve the outcomes intended as all of the USAID funded organisations immediately ceased hiring. The funding freeze was an unpredictable shock and the service provider had to minimise the impact. While the service provider did receive 20% of the funding for the enrolment milestone, they did not receive the balance of 80%.
Staff turnover. One service provider experienced the exit of several senior leaders. Much of the institutional knowledge of Jobs Boost was held by those who left. The team remaining notified Jobs Boost that they would be unable to deliver the outcomes that had been budgeted. The team made an appeal for funding despite the outcomes failure on the grounds that the organisation would experience significant financial distress were it not paid. Following discussion, a reduced outcomes target was set, but funding remained strictly tied to outcomes. This galvanised the organisation to focus effort on placing jobseekers, and ultimately it was able to exceed the reduced target and restore financial sustainability.
Unanticipated market demand dynamics. One service provider struggled to place jobseekers into qualifying artisanal roles, not because of a lower demand but because of the informality of the market. Artisanal roles in South Africa are initially filled for short term jobs with the jobseeker working on a temporary basis. This results in many of the placement contracts not meeting the quality criteria of Jobs Boost, which required qualifying jobs to be fulltime with at least a 12-month contract. A reduced target was agreed with the service provider. The service provider accepted the reduced budget but also committed to continue placing the jobseekers on a longer timeframe outside the Jobs Boost programme.
As of writing, the other nine service providers are on track to achieve their outcomes targets.
The examples of outcomes failures illustrate different risks faced by service providers.
The USAID withdrawal was somewhat of a black swan risk that could not be predicted in advance. It is arguable whether this risk should be left with the service provider, or if it should be “insured” against such unpredictable risks by the programme. In this case, the service provider acted swiftly to minimise the expenses it faced by ceasing recruiting and training for those components it could no longer place into jobs. Its model had minimised this risk because it only began training once it had a firm commitment to hire. While this did not insulate it from the USAID risk, the timing of the stop work orders meant it had only incurred limited expense so far. So, despite not having control of the risk, it was able to some extent to control the consequences of the risk when it emerged. In terms of the discussion of risk allocation above, the risk was correctly allocated to the implementing partner because it was able to minimise the cost of the risk when it emerged, though arguably this was down to luck of timing.
The staff turnover example illustrates weak management in the service provider. There was not sufficient planning for continuity to ensure success in delivering the outcomes. The service provider had historically operated solely on a traditional grant-based inputs and activities model. Its first instinct in its situation was to appeal for the OBF model to effectively revert to a grant model. This would likely have led to wastage as there was no guarantee that jobseekers would find jobs – indeed it was precisely this risk that the service provider was attempting to avoid by restructuring the agreement. When it became clear that restructuring the agreement was not an option, the service provider redoubled efforts to achieve a reduced placement target. This led a fundamental change to its whole approach and it worked hard to reach out to networks to place candidates in qualifying jobs. This illustrated an evolution in the risk understanding of the service provider, recognising that outcomes failure risk sat on its shoulders creating incentives for different behaviour.
Market demand and market dynamics are a significant risk to all service providers. This risk was a priority to manage and ensured that the skills provided were demand-led. Most service providers entered into agreements upfront with employers before they enrolled candidates to minimise this risk. While the service provider believed it had a level of undertaking from employers that gave it confidence to enrol candidates, market conditions were not as expected, and employers were not fully aligned to the programme requirements in that they would not enter long-term contracts. This risk was appropriately allocated to the service provider who has the most control over ensuring candidates would be placed into employment on the completion of training. But it also illustrates the need for better contingency planning, if employers are unable to follow through on commitments and meet contracting requirements.
These examples illustrate how the risk allocation changed behaviour in the programme. With the exception of the USAID example, which was contingent on the timing of stop work orders, the allocation of risk drove behaviour that prevented wastage and maximised outcomes.
Conclusion
I have argued that the value-for-money and effectiveness of outcomes-based finance extends beyond the commonly cited benefit of eliminating waste from failed programmes. While selective efficiency – choosing OBF over traditional grant funding – does provide important cost savings by ensuring payment only occurs when outcomes are achieved, a significant driver of value lies in the incentive effects on implementation partners and the resulting reduction in unit costs of outcomes.
The Jobs Boost Outcomes Fund demonstrates how thoughtful OBF design can harness these incentive effects to maximise public value. The competitive procurement process revealed substantial price variation across different sectors and skill levels. This price discovery alone provides valuable policy intelligence about the relative costs of job creation across different contexts and demographics. More importantly, the programme’s emphasis on both financial and non-financial evaluation criteria created incentives for implementation partners to optimise their delivery models while maintaining quality standards and targeting priority demographics.
The experience of Jobs Boost reinforces several key design principles for maximising OBF effectiveness. First, risk allocation must be carefully calibrated to place risks with the party best able to influence, manage, or absorb them. The programme’s experience with three implementation partners who failed to meet targets illustrates the nuanced nature of risk allocation in practice. The staff turnover case demonstrated how properly allocated operational risks drove behaviour change, forcing the service provider to evolve from a traditional grant mindset to an outcomes-focused approach. The market demand failure showed how implementation partners can and should manage employment placement risks through better contingency planning and employer relationship management. However, the USAiID withdrawal case highlighted the complexity of allocating unpredictable systemic risks, where timing and service provider agility determined the ultimate cost impact.
Second, the pursuit of surplus by implementation partners, whether for-profit or non-profit, creates a source of value that can be shared with the state through the right design. As the OBF market matures, financial incentives become increasingly important drivers of participation and performance. Rather than viewing surplus as problematic, OBF design should recognise it as a legitimate return on risk-taking that drives continuous improvement in delivery models. The key is ensuring that surplus generation occurs through genuine efficiency gains rather than quality compromises, which requires clear performance standards and robust monitoring systems as well as competitive procurement processes.
Third, competitive dynamics are essential for translating these incentive effects into public value. The Jobs Boost procurement process, with over 100 initial expressions of interest narrowed to 12 implementation partners, created genuine price pressure while maintaining focus on delivery capability and programme quality. This competition not only drove down unit costs but also encouraged innovation in programme design and service delivery models.
Fourth, the programme’s risk management mechanisms proved essential for maintaining overall performance despite individual partner failures. The ability to reallocate budget from underperforming to overperforming partners ensured that programme-level targets were exceeded even when some implementation partners struggled. The knowledge sharing sessions fostered continuous improvement across the partner network, demonstrating how competitive procurement can evolve into collaborative implementation without undermining performance incentives.
The broader implications for OBF policy and practice are significant. Policymakers should move beyond viewing OBF simply as a risk transfer mechanism and instead focus on how incentive structures can be optimised to drive systemic improvements in service delivery. This requires greater attention to procurement design, risk allocation frameworks, and performance management systems that balance autonomy with accountability.
For implementation partners, the OBF model offers opportunities to demonstrate superior delivery capability and capture returns from innovation. However, success requires sophisticated risk management capabilities and the financial capacity to manage extended cash flow cycles. The experience of Jobs Boost suggests that market development activities, including capacity building workshops and clear guidance materials, are essential for building a capable supply market that can effectively price and manage risks.
Several areas warrant further research and policy attention. OBF requires ongoing investment in supplier development and standardisation of contracting approaches to reduce transaction costs. The role of intermediary organisations in providing technical assistance and risk management support may become increasingly important as OBF scales. Additionally, more sophisticated approaches to measuring and rewarding quality alongside quantity outcomes could further enhance the value proposition of OBF structures.
The experience with unpredictable external shocks, such as the USAID withdrawal, highlights the need for clearer frameworks around force majeure and systemic risk allocation. Future OBF design should consider how to balance protecting implementation partners from truly uncontrollable risks while maintaining the performance incentives that drive value creation. The COVID-19 pandemic and other global disruptions provide natural experiments for understanding how OBF programmes adapt to external shocks and what this reveals about optimal risk allocation and programme design.
Ultimately, the success of outcomes-based finance depends not just on payment mechanisms but on creating an ecosystem of incentives that align the interests of all stakeholders around outcomes achievement. When properly designed, OBF can transform public service delivery by harnessing market forces in service of public goals. The Jobs Boost experience demonstrates that this transformation is achievable but requires careful attention to the incentive architecture that makes such programmes work. As governments worldwide face increasing pressure to demonstrate value-for-money in public spending, understanding and optimising these incentive effects will be critical to realising the full potential of outcomes-based finance.
References
GO Lab. n.d. ‘Outcomes-Based Contracting’. The Government Outcomes Lab. Accessed 20 August 2025. https://golab.bsg.ox.ac.uk/the-basics/outcomes-based-contracting/.
GSG/EOF. 2021. ‘Tying Funding to Results: A Primer in Results-Based Finance to Support a Just Covid Recovery and Foster Impact-Driven Economies’. GSG Impact; Education Outcomes Fund (EOF). https://www.gsgimpact.org/resources/gsg-impact-publications-and-reports/tying-funding-to-results/.
Gustafsson-Wright, Emily, Sophie Gardiner, and Vidya Putcha. 2015. The Potential and Limitations of Impact Bonds: Lessons from the First Five Years of Experience Worldwide.
Irwin, Timothy. 2007. Government Guarantees: Allocating and Valuing Risk in Privately Financed Infrastructure Projects. World Bank Publications.
Jobs Boost. 2023. ‘Jobs Boost Outcomes Fund Launched: R300 Million Allocated to Youth Training That Results in Jobs’. Jobs Boost, November 20. https://jobsboost.org.za/2023/11/20/jobs-boost-outcomes-fund-launched-r300-million-allocated-to-youth-training-that-results-in-jobs/.
Mulgan, Geoff, Neil Reeder, Mhairi Aylott, and Luke Bo’sher. 2011. Social Impact Investment: The Challenge and Opportunity of Social Impact Bonds.
OECD. 2025. Outcomes-Based Financing in the New Financing for Development Architecture: Lessons and Opportunities for Governments, Development Partners, and Multilateral Organisations. OECD. https://doi.org/10.1787/dcr-2002-en.
Ramasamy, Sankar, and Marc de Boer. 2004. ‘Outcomes Based Funding (OBF) Pilots Final Evaluation Report’. New Zealand: Centre for Social Research and Evaluation–Ministry of Social Development.
StatsSA. 2025. South Africa’s Youth in the Labour Market: A Decade in Review | Statistics South Africa. May 13. https://www.statssa.gov.za/?p=18398.
World Bank. 2020. ‘An Introduction to Outcome-Based Financing: GPRBA’s Outcomes Fund MDTF’. Global Partnership for Results-Based Approaches, World Bank Group. https://www.gprba.org/knowledge/resources/introduction-outcome-based-financing.
[1] This paper was written for a presentation to the Social Outcomes Conference 2025 at the Blavatnik School of Government, Oxford. It has benefitted from comments by Lameez Alexander and Riyaadh Ebrahim from Krutham.
[2] Stuart Theobald led the design and implementation of Jobs Boost and is executive chairman of Krutham which now manages it. Theobald is a visiting researcher at the Southern Centre for Inequality Studies at University of the Witwatersrand and at the Centre for the Philosophy of Natural and Social Science at the London School of Economics.
[3] I use “suppliers” and “implementing partners” interchangeably to refer to the organisations that are contracted to deliver the specified outcomes in return for payment.