2 minute read
In an outcomes-based contract, either part or all of payment to the provider (and in some cases, an investor) is linked to the achievement of specified outcomes. As a result, the price that is paid for particular outcomes is important to ensuring the contract is successful. Outcome payers should engage with all parties throughout the price development process, to strike a balance between the value to the outcome payer, the cost of delivering the service to providers and the distribution of financial risk.
The parameters for price setting
Ideally, pricing outcomes involves first deriving the value to set the upper bound, then estimating the cost to set the lower bound, before finally setting the most efficient price within the range between value and cost.
Value represents the maximum amount of money the outcome payer is prepared to pay for a good/service.
The value of a particular outcome will vary based on who is paying for it, and depends on a range of financial, political, social and economic considerations. It includes the intrinsic value, comprising all of the long-term benefits of individuals achieving the outcome, and the prevented costs, the fiscal benefits of preventing costly social problems.
Cost is the cost of delivery, development, management and financing to the service provider (and where applicable, investor).
An outcome payer can attempt to estimate the cost in a number of ways. These include engaging with the market, examining the historical costs of the service if it has been delivered previously, or – if it has not - comparison to similar services which are likely to use similar resources.
Price is the amount to be paid for the good/service agreed between the provider and outcome payer.
If there are many potential providers, the price can be set by working downwards from the value, where providers bid at a discount against a maximum price. If the contract is developed with a single provider, the price can be set by working upwards from the cost, negotiating a price that reflects the distribution of risk between parties.
Factors which affect the price
Cohorts that are harder to help will require more support, resulting in higher costs for the provider, and therefore a higher price. Cohorts with diverse needs require more complex pricing structures, and may be susceptible to perverse incentives.
Level of improvement
The greater the level of improvement required for service users to reach desired outcomes, the more intensive support they will require, resulting in a higher price.
Likelihood of success
If there is more uncertainty about whether outcomes will be achieved, there is more risk to the provider/investor. To compensate for this, they may expect a higher price.
Timing of payment
Later payments for outcomes are more likely to reflect long-term policy goals, but increase financing costs and uncertainty for the provider, requiring a higher price. In contrast, earlier payments can reduce financing costs and lower uncertainty, leading to a reduced price, but potentially fail to deliver long-term goals.
Ideally, payment should only be made for outcomes over and above what would have happened anyway. If it is believed some outcomes would have happened without the provider’s intervention, the price should be reduced accordingly.
If an outcomes-based contract is particularly successful, the outcome payer may have to pay out more than expected. This can be mitigated by defining payment caps - either on the total payment to providers, or on the total payable outcome per individual.
5 minute read
An outcomes-based contract (OBC) involves the provider’s charges being linked, in whole or part, to the achievement of defined business outcomes for the customer, rather than being based on input costs, such as labour, or outputs, such as transaction volumes. For instance, a contract for the provision of training to support unemployed people into work may have payments linked to the proportion of trained individuals that find employment and maintain it for a minimum period.
An OBC can be underpinned by an impact bond (IB), where a third party investor provides up-front working capital to fund provision prior to outcome payments being made, and takes on the risk of non-payment if outcomes are not achieved. This relatively new type of financing method has been used in countries across the world. IBs originated in the UK which as of mid-2019 has the greatest number globally. They are in line with approach taken by the World Bank Group’s Maximising Finance for Development. They have been adopted in a number of countries to address a range of problems including supporting children at risk of being taken into care, tackling homelessness, improving health outcomes, reducing offending and a range of other social issues.
The idea is simple and powerful, but it requires a clear definition of who is affected by the social issue you aim to address (the cohort), what ‘good’ looks like for those people (the outcomes), and how much to pay if that is achieved (the price). These three considerations – cohort, outcomes and price – are all inter-related. Changing one will affect at least one of the others.
Approaches to defining the cohort and choosing which outcomes to pay for are examined in our guide to Setting and Measuring Outcomes. They are equally important in circumstances where it is desirable to measure outcomes, but the funder or buyer of the service does not want to move immediately to outcomes-based payment. However, in this guide we aim to create a framework to support an outcome payer who wants to pay, in part or in full, for the outcomes achieved.
The basic ingredients underpinning a payment mechanism may be summarised in the following sentence:
“If individuals in a clearly defined cohort achieve (or improve towards) a desired outcome, as measured by using an agreed process or tool, over a certain time span, compared to what we might expect to happen otherwise, then an outcome payer will pay an agreed amount of money.”
A theory of change, or logic model, can be a useful tool to describe the social issue you are considering and how people may benefit from the service or ‘intervention’ the provider(s) will deliver. A theory of change describes how ‘inputs’ (i.e. resources such as money and staff time) are used to deliver ‘activities’ (e.g. recruiting participants and setting up a mentoring scheme), which produce outputs (e.g. beneficiaries participating in the mentoring scheme), which will lead to desired outcomes (e.g. improved engagement at school) and impact (e.g. better education results and life prospects). The theory of change tool is useful in the processes of setting and measuring outcomes and evaluating outcomes-based contracts. While it relies on a much simplified version of the messy reality of improving outcomes for people, it remains a popular approach.
Figure 1.3 shows how the basic ingredients underpinning a payment mechanism map against a basic “theory of change”.
The payment mechanism is an important aspect of the contract that will influence how the outcome payer(s) manage the relationship with the delivery organisation and other stakeholders, and monitor progress.
It is worth noting that it is not the only tool that should be used in order to effectively manage your outcomes-based contract. It is important to distinguish between:
You may wish to consider a payment structure that combines elements of a ‘fee-for-service’ contract and elements of an ‘outcome based’ contract. For instance, you may agree initial payments for outputs or activities (such as engagement and assessment) and later incentivisation payments subjects to the achievement of desired, longer term outcomes.
For an outcomes-based contract to work effectively it needs to strike a balance between what is valuable to the outcome payer and what is possible from a service delivery perspective, and acceptable from a financial risk perspective. That is why it is important for outcome payers to engage with providers and (in the case of impact bonds) investors throughout the process, from start to end. A fundamental purpose of using an outcomes-based contract and SIBs is to generate a shared understanding between an outcome payer and providers of ‘what good looks like’, i.e. the desired outcomes, and to allow more autonomy for the provider to use their skills in bringing about the desired outcomes. Although defining the aspects of a payment mechanism will, by necessity, seem like ‘technical’ work, remember that the contract and the payment mechanism should strengthen and not weaken the relationship between outcome payer and provider.
For this reason, often outcome payers see a benefit in treating the contract more as a partnership than a transactional relationship. As this guide will show, there are many aspects of pricing the outcome where feedback from the market is helpful, or even essential. There will of course be limitations around how much providers and investors can share and some information might be commercially sensitive. An outcome payer should be clear upfront as to how the information shared will be used, and agree terms of sharing information with all those engaged. A partnership approach (especially if it is to work in the long-term) requires trust and transparency, and relies on parties conducting the negotiations in a spirit of openness and honesty.
An outcome payer should be prepared to allow room for revisions (prior to procurement) and negotiations (as part of the procurement process to appoint provider(s) and/or investor(s). It is important to allow sufficient time throughout the project development process for these conversations to take place. An outcome payer might find it helpful to set the right expectations within their own organisation around (i) the timeframes for completing the work, (ii) the need to have some flexibility in the negotiations with the provider(s) and investor(s) and (iii) the fact that numbers are indicative in the early stages and subject to change during procurement negotiations.
In the process of pricing outcomes, it can be very helpful to discuss with others who have faced similar questions. The GO Lab has a range of activity that supports peer learning, including regional workshops and events, as well as the impact bond knowledge club.
18 minute read
The fundamental economic concept of supply and demand is important to the pricing approach. The demand side contains the customers who have a willingness to pay up to a certain amount to acquire a good or service. The supply side is represented by the providers who are willing to sell their good or service for an amount greater than (or equal to) the cost of producing it. In an outcomes-based contract, the outcome payer (e.g. the funder or commissioner) represents the demand side, and the supply side is represented by the provider. In an impact bond model, the investor usually joins the supply side (see our guide to awarding an outcomes-based contract for more on how impact bonds are structured).
If the price is lower than the value, the outcomes payer benefits from the transaction. If the price is higher than the cost, then the provider benefits from the transaction. Therefore, ideally a transaction only takes place when the value is greater than (or equal) to the price and price greater than the cost. Figure 2.2. shows this graphically.
If it is not practical to both derive the value and estimate the cost, it is possible to set the price by either working downwards from the value, or upwards from the cost, as explained in 2.4 how to set the price.
Deriving the value allows the setting of an upper-bound to what can be paid. Value can be subjective: not only might value of an outcome vary according to who is paying for the outcomes, but there may also be different perspectives within a governing body or outcome payer, and also across time. In short, value is in the eye of the beholder, and is dependent on a range of financial, political, social and economic considerations, which vary according to the organisation or body concerned.
As an example of a national government’s finance ministry perspective, on defining the value in the UK public sector, the HM Treasury's Green Book addresses the ‘Scope of Benefits’ concept by grouping benefits with respect to their scope into three sections: (i) Direct public sector benefits (to originating organisation); (ii) Indirect public sector benefits (to other public sector organisations); (iii) Wider benefits to society (e.g. households, individuals, businesses).
This kind of national perspective can be very important when a central government is the outcomes payer, or is supporting others to pay for outcomes.
However, most outcomes payers have to look at the value of outcomes from a much narrower perspective. If there is a private outcome payer, local and regional government, or single central government department, they are more likely to consider:
It is usually hard, for example, for an education department to pay for reduced offending outcomes, or for a health department to pay for reduced homelessness, even though those benefits could well occur as a result of work those departments paid for.
This is one of the reasons why some countries have set up outcomes funds. Government as a whole can support projects which produce wider societal benefits, which an individual department or regional / local government cannot afford to fund.
Depending on the circumstances, defining the value may not always be possible or useful. Where this is the case, you might wish to refer to 2.3 estimating the costs.
However, it may be helpful to put a figure on the value for two reasons:
The intrinsic value of an outcome is an attempt to put a monetary figure on all the long-term benefits that might occur for society when an individual achieves particular outcomes. It goes beyond direct fiscal benefits to the outcomes payer. Often, this value is defined in moral and political terms – it is the reason why we educate children, rehabilitate offenders, help rough sleepers, and so on. But especially in the case of national government, it is possible to attempt to quantify the value using Cost-Benefit Analysis (CBA). CBA is widely used across governments to appraise spending options. It attempts to calculate the value to society of achieving the outcome. Most governments will have an approved approach to CBA so we do not cover in detail in here. For example, the UK has CBA guidance for local partnerships. This gives a word of caution, though: “CBA is not an exact science and its outputs are a guide to decision-making, not a substitute for thought’’.
Calculating prevented cost is an attempt to calculate the fiscal benefits that might occur when an individual or group achieves particular outcomes. This approach focuses on how the achievement of certain outcomes now might reduce costs to the outcome payer in the future. It uses a similar form of analysis to the ‘intrinsic value’ approach, but with a narrower focus.
Preventing costly social problems from arising in the first place, or reducing their incidence, is often likely to lead both to better social outcomes for people and lower public sector spending.
However, it is difficult to allocate resources to preventative activities when there are budget pressures, and the available funding often has to be prioritised towards people in need of urgent care to be supported first. Furthermore, the difficulty of knowing for sure whether the preventative activities will actually deliver their demand reduction promises, challenges the ability to invest in long-term preventative activities.
An outcome-based contract may help enable a focus on preventative activities that will lead to better future outcomes, and reduced demand (and therefore costs). It also has the potential to enable ‘double running’ of budgets, where preventative and remedial work is run in parallel, provided the demand reduction unlocked by the preventative work is able to generate genuine future budgetary savings.
Whether you are seeking to calculate intrinsic value or prevented costs, the approach is similar.
Most importantly, as mentioned at the start of this guide, it is important to have a clear understanding of the cohort or group of people for whom better outcomes are targeted. You can read more about this in our guide to Setting and Measuring Outcomes.
You should then start to think about all the benefits arising from the achievement of a certain outcome. If you are seeking to calculate the overall intrinsic value, you will need to consider them all. But if you are seeking to calculate prevented costs to a particular department or sub-national government, who will also be the outcome payer, you probably just need to focus on the first two: cohort member and outcome payer.
Your aim is to quantify, each of these benefits as well as being able to describe what the impact looks like. Sourcing this information can be difficult; ideally the scope and scale of these benefits will come from research done on the cohort you are aiming to target. It is advised to follow a conservative and evidence-based approach to pre-empt future criticism when discussing with other partners.
In some cases either an outcome payer or service providers might be able to provide data or feedback on the types and/or quantity of benefit. Alternatively, you may need to rely on published research for similar cohorts.
These benefits will fall into two broad categories: Fiscal benefits and economic/social benefits. Fiscal benefit sare impacts on the public sector budget which are either ‘cashable’(where expenditure released from the change in outcomes is freed up and can be reallocated elsewhere, e.g. no longer needing to spot-purchase residential children’s care) or ‘non-cashable’ (which represent a benefit to the public from freeing up resources even if public expenditure is not reduced, e.g.. reduced demand from frequent attenders in A&E allows staff to concentrate on more critical cases). Economic/social benefits not only consist of such benefits, but also wider gains to society that are often harder to identify and value in monetary terms.
There are some benefits for which it is either unfeasible, or undesirable, to derive monetary values for – such as greater wellbeing. These should still be considered alongside those in monetary terms, in order to decide if your actual ‘willingness to pay’ is higher than the final monetary figure.
Where a program is commissioned on the basis of prevented costs, then unlike in the intrinsic value approach where a degree of estimation of the benefits of outcomes being achieved is permissible, it is important to be as precise as possible about the fiscal benefits to the outcomes payer and when these will occur. The outcomes payer is agreeing to pay for outcomes directly as a result of the savings that they expect to make, which will often actually finance those outcomes payments. Therefore, the focus will be on the “cashable” savings.
It can sometimes be quite straightforward to determine the cashable savings of a programme. For example a programme might be intended to reduce the number of children who have to be placed in the care of the state. If that care provision is paid for on a spot purchase basis at a known cost, and by the same agency that intends to pay for outcomes in a proposed outcomes-based contract, then there is a known direct saving to the outcomes payer if the number of children who go into care is reduced.
In other cases, though, it is much more difficult to determine cashable savings.
This may be because for example:
Once the analysis of value is complete, you should reflect on these figures to decide the total value of the outcomes, and this total should represent the upper limit of public sector’s ‘willingness to pay’ for achieving that outcome (although in reality it is unlikely the analysis has been able to capture and monetise all the benefits).
If you are one of many public sector organisations identified in the analysis, you may wish to price an outcome based only on your own fiscal benefits. Alternatively, you may wish to use the analysis to engage other public sector outcome payers in order to price the outcome in accordance with realised benefits for them.
Proper analysis of the direct and wider impacts of outcomes can help to justify or reject the use of particular outcomes for payment, and provide a rigorous basis for a decision on expenditure by an outcomes payer. It can also help construct the case for co-commissioning or facilitating cross-agency working. For example, this approach can be useful for contracts which aim to meet the needs of individuals who move through different existing services which do not work well for them under the status quo. It can also provide a starting point if a service is completely new.
On the other hand, it can be difficult for teams with little or no analytical resource to make use of the tools of cost benefit analysis. Furthermore, if the analysis has not used local data, but relies on standard datasets, it may undermine trust in the resulting figures.
Depending on the issue being addressed, the time delay between the intervention and the achievement of medium to long term outcomes which are of most value to the outcome payer may make an outcome based contract unfeasible for a provider and/or social investor.
There can also be a problem in this approach with "prevalence”: the number of people in the target group who display (or might go on to display) the undesirable results which the intervention is intended to mitigate (or avoid).
There can also be a problem with “deadweight” which is the term often used for those who would have achieved a positive outcome even without the intervention.
The issues of timing of payments, prevalence and deadweight are all discussed further in Chapter 3.
Where a precise calculation of value cannot be carried out, it is possible to commission services through an outcome based contract using an estimated value derived from limited data.
It may then be possible to run an evaluation which makes a more robust assessment of the range of benefits which are delivered. For example, an IB designed to reduce rough sleeping might tie payment only to that – but the price paid may also have factored in expected reductions in crime and health service costs. A robust evaluation can show if these were achieved as expected, and therefore whether the price paid was justifiable, and further commissioning is warranted.
|Intrinsic value||Prevented Costs|
|Key concept||Government’s ‘willingness to pay’ for a social outcome||Potential future budgetary savings|
|When helpful?||When value of an outcome is not primarily financial. When the desired benefit is co-commissioning or cross-agency working.||When an intervention to prevent a social problem from developing for an at-risk group is likely to be effective|
|Type of analysis to do||
Monetary value of economic, social, and political benefits using economic theory.
Statistical analysis of how people ‘flow’ through the system, what are their needs on the journey and what is currently offered at what cost (to whom). Cost benefit analysis of successful prevention.
|Strengths||Can help make the case for central government funding, co-commissioning or collaborative working. Cashable savings not a priority. Can provide a starting point if the service is completely new.||Develops understanding of root causes of problems. Can enable flexibility in delivery as different means of prevention are attempted.|
|Limitations||Can require considerable analytical resource. A budget line to pay an amount based on this method of calculation must exist or be created.||
Can create a large financing gap for providers (if the avoided negative outcomes would be a long time in the future).
Risk of paying for the prevention of things that would never have happened anyway.
|Resources||UK Treasury green book Unit cost database that has been developed by New Economy.||
What Works Centre evidence synthesis.
Unit cost database.
Estimating the cost allows the setting of a lower-bound to what the provider should expect to receive from the contract, if they achieve a satisfactory level of performance in achieving outcomes. The lower bound of the price paid for outcomes should cover the costs the provider (and/or investor) have to endure in order to achieve the outcomes at the level expected. As a result, it is useful for the outcome payer to derive an understanding of what these costs might be. Of course, if they do not achieve the expected level of outcomes, their costs will not be covered.
It is important to distinguish between the cost to the provider for delivering the outcomes (the lower bound), and the cost to the outcome payer, which includes any surplus (or "profit") the provider may seek (which the outcome payer will want to minimise) and management and monitoring costs to the outcome payer. This is illustrated in Figure 2.4.
While outcome payers may seek to minimise provider surplus / profit, underpaying for outcomes can be as dangerous to the success of an outcomes-based contract as overpaying, because it may mean that no bids are received, may discourage high quality providers, and may encourage cherry picking whom they work with i.e. focus their efforts primarily on people for whom the defined outcomes can be achieved most cheaply, whilst ignoring those who need the most help. This is discussed further in Chapter 3.
The importance of estimating provider costs may be reduced if you are operating in a competitive environment where real competition on price is likely to occur, and the market can be relied upon to deliver value for money. However, where there may be very few providers or investors in a position to deliver the contract, there may not be a competitive market, and a greater degree of transparency may therefore be needed around provider costs than would be necessary where a functioning provider market exists.
An example of how costs might be broken down into a number of components is shown in Table 2.2. This example assumes that the impact bond is being led by an investor who sets up a special purpose vehicle (see our guide for more on IB structures). In other impact bonds, these costs may be split between the investor and provider, and perhaps also an intermediary and evaluator as well.
It is important to note that whilst the operational costs of delivering the contract are likely to be easily the largest component, there are sometimes significant costs associated with development and design, contract set up and negotiation and evaluation.
|Cost Category||Cost Sub-category|
DEVELOPMENT & DESIGN
Pre-contract signing search and information costs
|Due diligence of potential providers|
CONTRACT SETUP & NEGOTIATION
Bargaining and Contracting Costs
Legal advice on contract design (including tax advice & either external or internal)
Financial advice on contract design (either external or internal)
SPV Setup Costs
Reworking of business and financial case
Governance, monitoring and evaluation costs
SPV operational costs (including salaries)
Performance management (Staff)
Data management (Technical)
Outcome reporting &/or compliance with validation method cost
Service delivery costs (upfront payments)
Service delivery costs (later payments)
Governance costs (e.g. board meetings)
Outflow to investors*
In outcomes contracts, the finance costs borne by the investor and/or the provider are passed on to the outcomes payer if the program is successful. These may vary significantly depending on the source of the finance, the amount of investment required, and the perceived level of risk of achieving enough outcomes to break even.
There are three ways that an outcome payer can try to gain an overall understanding of costs: market engagement, historical costs, or comparable services.
It may often be possible and desirable for outcomes payers to gain an understanding of the likely investor and provider costs of a proposed program through market engagement and/or asking for the information as part of a tender process.
This approach applies if the intended outcome requires a service that has been used previously on a similar target population. In this case, you should have an idea of the price of such services which enables you to get an estimate of the likely total cost of outcomes, provided that a similar service is proposed for the outcomes contract.
An estimation based on historical costs is often easier to pursue if an existing service is being re-let, as that gives a ready comparison. As there is already a service in existence to compare with, then a lot of the uncertainty about referral rates, level of need of the client group, and to some extent level of anticipated success, are also reduced. As we discuss later in "3.6. dealing with uncertainty”, in the case of IBs this may also imply that investor returns could be lower.
After finding an estimate of the intervention costs, it is worth considering the possible additional costs of an outcomes-based contract, as shown in Table 2.2. above. However, it is not necessarily the case that an outcomes-based contract is more expensive to deliver than a standard contract. The extra discipline of having to achieve outcomes in order to receive payment might drive increased efficiency, which is likely to bring down the costs of delivering a given number of outcomes.
Where there is no historical service to compare, it may be possible to refer to comparable services which are likely to use similar resources/inputs. This should provide an approximate of expenses required to deliver the outcomes. Once again, additional costs specific to outcomes-based contracts should be taken into consideration.
Payment mechanisms should include generous enough payments to provide a quality service to the target population, which accepts that the risk taken by providers/investors of not succeeding needs to be balanced by the potential to make a surplus if they do well. Prices should be fair for all parties, high enough to be commercially viable and low enough to avoid windfall profits for investors and unnecessary expenditure by outcomes payers.
Whichever of these is used, a number of assumptions will need be made, which will affect the price. These “parameters” are as follows, and are discussed in detail in Chapter 3.
The best way to analyse the effect of these parameters is to build a financial model, where different assumptions can be made, and the probable effect of these on the price explored.
An additional parameter might be currency risk, if costs are incurred in a different currency from that used to pay for outcomes (which can be the case in international development). This parameter is not discussed further.
Some of these parameters are illustrated in figure 2.4.
Often the analysis of these parameters is best done through active, structured engagement with market. We discuss different approaches to market engagement and procurement, and the pros and cons of these, in our guide to awarding outcomes-based contracts.
Whichever pricing approach is used, a full and detailed assessment of value-for-money can be done. There is a full discussion of how this might be approached in the Appendix. There are three criteria that can be used for this (this is based on the UK’s National Audit Office and Department for International Development).
17 minute read
As outlined at the beginning of this guidance, the price you set will interact with the cohort definition and outcomes you expect to achieve, which is why the approaches described in Chapter 2 only give you a sense of the upper and lower bound, rather than a final answer. You need to strike a balance between multiple considerations in order to create useful incentives to achieve outcomes. There are two risks if these considerations are not taken into account. The first is the risk of underpaying for outcomes – in which case you will end up with a contract that is impossibly expensive or risky for a provider to deliver (and most likely none will offer to). The second is the risk of overpaying for outcomes – in which case providers make unreasonable profits / surplus and you do not get value for money.
As a basic rule-of-thumb, cohorts that are harder to help or have further to ‘travel’ in order to achieve the desired outcomes will be more costly to work with, as they will need more intensive support. As we describe in the likelihood of success section below, harder-to-help cohorts will also tend to be less likely to achieve the outcomes than easier-to-help ones. This increases costs. These increased costs need to be compensated through the payment mechanism.
Cohort specification will enable you to (i) define a specific cohort of people with key characteristics and (ii) estimate variation in difficulty-to-help amongst beneficiaries. How easily can you describe the characteristics of those for whom you are commissioning the service? The more targeted and similar the cohort, the most straightforward the approach to price setting can be. On the other hand, the more diverse the needs are of a cohort of people who exhibit the specific problem which is being targeted, e.g. being homeless, or being unemployed, or having a drug problem, the more an outcomes-based contract may be beneficial, because of the difficulty in specifying a single service intervention applicable to the whole cohort.
The more comparable the levels of need in a cohort, the less the risk of ‘perverse incentives’ (for example, there is less risk that a provider may ‘cherry-pick’ and only work with ‘easier cases’). This makes knowing how much to pay easier: it can be discovered through market engagement or by examining the cost of in-house provision.
In many cases, however, it will be difficult, or incompatible with the purposes of the program, to identify a cohort of sufficiently similar people, especially for services aimed at supporting people with complex needs. Often a cohort will be large and diverse with broad a range of needs and potential intervention / support packages. In these cases, it is still possible to reduce the risk of ‘cherry-picking’, in three ways:
A key part of defining payable outcomes is setting a point of improvement from a baseline at which payment is made. This can be referred to as the ‘threshold’, ‘target’, ‘metric’, ‘milestone’, or ‘trigger’ at which the payment outcome is deemed to have been achieved. We will use the word ‘target’ here. Essentially, it means defining ‘what does good look like’ or ‘what is a meaningful improvement’? The basic rule of thumb here is that the greater the level of improvement desired within the cohort identified, the greater the value of the improvement and the more costly it is likely to be for providers to achieve the targets, as they will need to offer more intensive support, leading to a higher price paid for the outcomes. In the last section we discuss ways to tackle uncertainties around a provider’s likelihood of achieving the level of improvement set out.
The longstanding discussion in education about how to measure learners’ performance is perhaps a helpful analogy to think about the forms these targets can take. The question is whether student performance is best measured by attainment / proficiency (meaning a student’s performance against a universal benchmark at a given time), or progress / growth (meaning a student’s performance improvement or decline over time, relative to the average or their own starting point). Proponents of progress or growth scores say that using attainment or proficiency scores encourage teachers to focus less on those students who fall far below the attainment threshold, and unfairly stigmatises schools whose intake has more of these students. They argue that we should be using progress or growth measures if the goal is to assess schools on how well they serve students, not on which students they serve. To take an example from the UK, the long-standing but now abandoned ‘A*-C’ measure of GCSE grades is an example of how an attainment-based cut off can have these effects, as it encouraged schools to disproportionately focus resources on students on the C/D borderline, at the expense of those expected to get lower and higher grades.
In the world of defining outcomes, the same logic holds true. You can think about fixed, binary targets – like an ex-offender not re-offending during a set period, or someone who is homeless living continuously in accommodation over a long period – as though they are attainment scores. Indeed, evidence suggests that they risk focussing providers’ attention on beneficiaries who are around the cut-off point, often at the expense of individuals who have further to travel. For instance, a binary payment for ‘not re-offending’ would incentivise providers to work with offenders who only offend few times to bring this down to zero, rather than working to reduce reoffending rates for people who have a history of many offences (without being able to eliminate them entirely).
Targets that reflect the ‘distance travelled’ may be more accurately estimated and reduce these risks, but it can be more challenging to measure them – typically requiring more granular and ‘sensitive’ measures or measurement approaches. In taking this approach, you will need to either (a) measure, and take into account, an individual or cohort’s starting point, and determine an acceptable amount of progress to have made by the end; and/or (b) show degrees or ‘steps’ of improvement, and add extra payment targets to reward these. Either of these approaches adds complexity, but it is key to think about them and to create a solution that addresses these points.
On the other hand there are areas where distance travelled towards a goal may have little value unless the goal is achieved. For example, supporting a person who is a long way from being “work ready” to being closer to work ready may not be worth much if that person still doesn’t get a job, unless the work done has other benefits such as improved mental health.
You can read more about tackling these issues in our guide to setting and measuring outcomes.
All stakeholders in an outcomes-based contract need to understand how likely it is that the project will achieve the proposed outcomes. If outcomes seem harder to achieve, either because the cohort is difficult to help or because the desired level of improvement is high, then the probability of achieving outcomes will be lower, and the contract will be deemed riskier – and risk demands compensation. In the case of impact bonds, social investors backing bids to provide working capital to finance the contract will take this into account in the level of financial return they expect – higher risk calls for higher potential returns.
There are a number of ways to make future performance projections and you will want to use a combination of all of these. To have a goal in mind, it is helpful to identify a range of values. For instance, you can estimate a ‘minimum expected scenario’ (sometimes termed ‘base case’), a ‘best case scenario’ and a ‘worst case scenario’.
Using historical data – If you have well documented data and a good historical data record, these would be a good starting point for a first estimate of the likely success of the project and how many outcomes are likely to be achieved.
Using existing evidence / academic research – in some cases, there will be existing evidence or academic research indicating how successful a particular programme or approach will typically be. For example, the Ways to Wellness social prescribing programme used the results of a pilot programme carried out by Nesta to predict how successful the project would be.
Running a procurement process with dialogue – by running a procurement process that allows for dialogue with multiple providers, you can compare competing claims on likelihood of success, and whether the estimates you are being given are based on robust assumptions. While providers will naturally want to show they have the greatest chance of success, it is important to assess how realistic the prediction is – or whether it seems overly optimistic. Please refer to the guide on awarding outcomes-based contracts for more about the different procurement approaches that allow for this sort of dialogue to take place.
Using the expert judgment and data of a provider – in some cases, you may be procuring a completely innovative service, or working with a new cohort who have not been previously identified or worked with. In these cases, you may be using an outcome based contract because you are not able to determine the likelihood of success and need to rely almost entirely on the projections of a provider, which you will want to test the rationale for. The risk is higher in these experimental scenarios, though the contract enables some (or all) of the financial risk to be transferred.
Use a learning contract or pilot period – You could procure services anticipating an initial period where you closely monitor the implementation of the intervention and improve your understanding of what level of outcomes it is feasible to expect. This would also allow you to identify the key barriers and strengthen your performance management system and payment metrics. At the end of this initial phase you can firm up your payment mechanism in collaboration with other stakeholders. There are examples in the UK of approaches like this being used. You still need to have a well reasoned baseline scenario before starting the contract, and should use this flexibility as a genuine opportunity for learning in partnership with other stakeholders. Good stakeholder relationships and a level of trust, as well as a clearly defined process for future price adjustments, are required to avoid the danger of the provider exploiting the flexibility to ‘move the goalposts’ in, say, year 3 of a 7 years contract. Our guide to awarding outcomes-based contract has more detail on how contractual terms can be used to safeguard this approach.
The timing of payments is important. Generally, the later the payments can be, the more likely it is that they will align with the achievement of your long-term policy goals, as you are more likely to know if the ultimate outcome is both achieved and sustained. There is also a benefit to the outcome payer of later payments, because money can be used for other things in the interim. Furthermore, in economies where inflation is high, there may be an additional benefit to later payment (unless you choose to include inflation in the pricing). These timings benefits can be quantified by using discounting (as shown, for example, in Chapter A6 of the UK treasury’s Green Book).
Often, though, there are practical implications for providers (and investors) to such long-term payments, meaning you should consider adding payments for ‘proxy’ or ‘lead’ outcomes (or even outputs as long as the final target in an outcome) in addition or as alternatives. You can read more about identifying these sort of outcomes in our guide on setting and measuring outcomes.
The cost implications of the timing of payments is all to do with the fact that an outcome based contract creates a financing need for a provider organisation. A provider provides a service upfront but is not paid until later, when outcomes are achieved – so they have to use their own money, or borrow it from someone else (like a social investor). The effect of different timings on this is best illustrated with a simple example:
The same basic principle applies whether providers self-finance, take out a loan, or receive backing from a social investor who pays them a traditional service fee and takes the financial risk on themselves. The longer the provider or investor has to wait to receive payment for the delivering of outcomes, the higher the strain on their finances, and the greater the cost. On the other hand, later payments help with the outcome payer cashflow and enables them to use their money on other projects, as well as reducing the risk that they are wasting their money on outcomes of limited value e.g. a homeless person who sustains accommodation for a short time but then becomes homeless again. The outcome payer might therefore attempt to ensure the sustainability of results by setting both early and late payments in a way to keep the providers (and/or investors) incentivised.
Lower risk for provider (& investor)
Guarantees some early indications of progress
Greater risk to outcome payer
Weaker incentives for provider (& investor) to achieve long-term outcome
Lower risk for outcome payer
Defers outgoings for outcome payer, so cash can be used elsewhere
Greater risk for provider (& investor)
Can be complicated to budget for uncertain future payments
You could add an additional earlier payment, for example at the end of 6 months, for any participant who has enteredaccommodation. This is not really the goal you are looking for – as they may soon return to sleeping rough – but, it means your provider would expect to get some income sooner, so will go less into the red, need to borrow less money to run the contract, pay less interest, and require a lower amount to be built into the contract cost.
In addition you may consider payments for activities and outputs as well as for outcomes. This will have much the same effect in enabling the provider to get some payment sooner, but at the possible expense of reducing the focus on desired outcomes, or alignment to the policy objectives.
In short: outcomes which are paid early on in the delivery phase can lower the cost of financing for the providers (and/or investors), but may lessen the focus on longer term outcomes that are usually more aligned with the overarching policy aims of the project.
Note that a greater focus on longer term outcomes, as well as increasing the financing amount, can also increase the provider’s (and/or investor’s) perception of risk. This can result in an increase in interest payments or financial return expectations. This is discussed further in the next section, ‘likelihood of success’.
In projects which feature both early and later-term outcome payments, you need to be aware that your payment structure may create incentives for providers to treat these outcomes ‘interchangeably’ and focus solely on delivering short-term outcomes for a higher number of beneficiaries. There are ways to mitigate this: having a limit on the total number of participants whom the provider is allowed to work with within an overall payment envelope (or “contract cap” – see section 3.6 below), and/or higher payments for longer term outcomes.
Although later payments help with the cashflow and cost of capital for outcome payers, it is important to plan their payment in the budget in advance. Public sector commissioners can often only pay money in the year they have the money budgeted for. So it is important to profile the expected level of outcomes payments in each financial year, and if possible build in some flexibility to move money between financial years in case outcomes are achieved later than initially expected.
“Additionality” is one of several related concepts – the others are attribution, counterfactual, and deadweight. While these terms can seem daunting if you are not familiar with them, we use them because they describe useful concepts and some are increasingly widely used.
“Additionality” refers to an impact that is “over and above what would have happened anyway”. You could describe it as “over and above business as usual” or “what we currently expect to happen”. A description of what would have happened anyway is known as the “counterfactual”. Determining the level of “additionality” in a robust way helps to show that any positive effect was indeed caused by the work that was done – this is the concept of “attribution” (i.e. it shows that the outcome is “attributable” to the intervention).
The answer to “what would have happened anyway” is very rarely “nothing”. For instance, in projects aiming to support people back into work, an obvious outcome payment is sustained employment. However, some of the participants might have found an employment over time even without the intervention. The amount of this natural improvement that takes place is called the “deadweight”.It is worth pointing out that deadweight can also work in the other direction e.g. if something happens to other participants to make them less likely to get work than they were at the start of the project.
Unfortunately, it is never possible to observe “would have happened anyway” unless you know how to create a parallel universe, and so the best you can do is to estimate it.
If you are running a quantitative evaluation alongside (or as part of) the contract, then it is feasible to think about using statistical methods to estimate “additionality”, such as experimental (e.g. randomised-controlled-trials) or similar quasi-experimental techniques (e.g. pre test-post test). These tend to use some sort of comparison group who share similar features to the cohort being worked with, but who do not receive the same service. We explain these techniques in our introduction to evaluation. For instance, the Peterborough SIB and the Ways to Wellness SIB each estimate additionality by using a comparison group technique.
If you are using one of these techniques as the basis for measuring the outcomes you are paying for, you will have a good degree of confidence that those outcomes are attributable to the intervention. The risk of paying for things that would have happened anyway will be reduced, and there is no need to adjust the price set at the start to account for these considerations.
Often, it is not practical or affordable to use a comparison group to estimate additionality as part of the outcome measurement approach. When this is the case, you should think about factoring a prediction of what the deadweight might be into the price you offer. If the contract is using a “proven” intervention, there may be previous research and/or evaluations that will give you a reasonable indication of the likely level of additionality for that intervention. Alternatively, you can estimate additionality by articulating a “business as usual” scenario, and comparing it to various success scenarios for the new service. If you can access the right data, you might do this through an analysis of historical trends for the particular cohort that is eligible for the intervention, and project those trends in the future. There will be a degree of uncertainty and you might include a range of scenarios which account for high- and low-estimates.
Be aware that this extrapolation into the future may be inaccurate, especially for longer-term outcomes, as other external factors could influence the trends positively or negatively – such as a change in the economic outlook or in other areas of policy. In defining additionality using past trends, you could end up paying for outcomes that would have occurred anyway, because changes outside the control of the provider are making the outcomes easier to achieve. The reverse is also true: you may end up not paying for outcomes that the provider has legitimately achieved, because changes outside the control of the provider are making the outcomes more difficult to achieve than they would have been in the past. As we explain in our guide to Setting and Measuring Outcomes, you can mitigate this by aiming to set outcome measures and targets that are less susceptible to such external factors, but it is difficult to eliminate this risk entirely.
Although such projections carry a level of uncertainty depending on the quality of analysis and the eco-political stability of the study group, they provide us with useful information of the expected outcomes to help communicating better with other stakeholders (see more on uncertainty, optimism bias, and risk in Annex A5 of the Green Book and international equivalents.)As a general rule, you should adjust either the price or total payment for outcomes downwards if you believe some outcomes would happen anyway (i.e. there is “deadweight”) but you don’t have confidence that you will be able to accurately determine how many (which you might do by measuring a comparison group). This is a legitimate measure to avoid paying for things that would have happened anyway, and to ensure good value-for-money to taxpayers. The possibility of price adjustment should be covered in the contract documentation. See more on this in our guide on awarding outcomes-based contracts.
1. Setting a cap on the total payment to providers or on the number of beneficiaries
You can consider setting a cap linked to the available budget for outcome payment. Consider the following relationship:
Total Payment = Payment per outcome x Number of people receiving the service) x Likelihood of success in achieving each outcome
In this approach, you will already have established a price per outcome, so you are setting a payment cap based on a maximum number of beneficiaries for which you will make payments under the ‘best case’ likelihood of success; i.e. Maximum pay is expected when ‘Successful Outcomes’ is equal to ‘Best Case Scenario’. You need to consider that this makes it unlikely the provider would continue investing effort and resources in providing the interventions to more people, as this would be at their own cost. Some providers (or the investors providing the finance) might continue to deliver the service anyway as they value the achievement of outcomes for their own sake and can access the required extra funding. You could discuss in advance with the provider what they expect to do if the cap is reached.
In some cases the cap follows from identifying a specific cohort. For instance, the “Street Impact” London Rough Sleepers SIBs identified the target cohort as consisting of 415 named individuals identified as sleeping rough in a particular dataset.
2. Setting a cap on the total payable outcome per individual
You may have defined a number of different outcomes that you will pay for as and when an individual in a cohort achieves them. As described earlier, this could be because your cohort is diverse, or because you want to include an earlier payment and/or reward progression towards an end outcome. If every individual in the cohort achieves every one of the available outcomes, you will overshoot your budget. However, rather than limiting the overall payment across a cohort, you may want to consider a cap to limit the total payment on each individual. This type of cap signals to providers the need to balance doing intensive work with a single individual with their ability to engage with a higher number of individuals overall (an ‘equity’ consideration). It also helps to protect from the possibility that providers focus on individuals who are ‘easier’ to work with, and who can progress through multiple outcomes more quickly. This is especially helpful when considering that these individuals were more likely to achieve some of the outcomes even without the intervention.
Can a separate cap be set on the provider’s surplus / investor returns?
Setting a payment cap as described above allows you to limit the return that investors can expect to earn from the project. If there are no investors, the same principle applies, but the ‘return’ would be reflected as provider profit or surplus. Providers and investors can use a number of methods to determine the level of surplus or return they anticipate, such as carrying out a financial sensitivity analysis. This analysis plays out a number of scenarios based on the expected outcome success levels.
Investors may be able to take risks on a particular project where they can take a portfolio approach. Across a portfolio of projects that they support, they expect that some projects will be more successful and lead to a higher return, whilst others may be less successful and lead to a loss. If you set a payment cap that limits investor returns, the investor may also wish to discuss a ‘floor’, i.e. a minimum payment that limits the total loss they could incur.It is worth mentioning that there is not necessarily an inherent need to cap surplus or returns, as in a well-designed payment mechanism, these will be higher when you are getting more outcomes, which is the aim.
8 minute read
The West London Zone (WLZ) social impact bond (SIB) launched in 2016 and is ongoing. It is the first SIB to launch using the “collective impact bond” model.
Core project stakeholders (and their project roles) include: WLZ (service provider), the London Borough of Hammersmith and Fulham (LBHF) (co-commissioner), the Royal Borough of Kensington and Chelsea (RBKC) (co-commissioner), The National Lottery Community Fund (co-commissioner), local schools (co-commissioner), private philanthropy (co-commissioner), and Bridges Fund Management (BFM) (investor).
WLZ is an organisation that partners link workers, charities, schools and other local organisations to support children and families.
The project targets children aged 5-16 in disadvantaged communities in West London who are at risk of negative outcomes in life due to being ‘off-track’ in school and in their wellbeing. The project covers the northern parts of two Local Authorities in West London: the London Borough of Hammersmith and Fulham (LBHF) and the Royal Borough of Kensington and Chelsea (RBKC).
Each authority has a separate SIB contract, though the contracts share the same features.
The project targets improvements in social and educational outcomes across several areas, and the contracting authority pays out based on children showing a measurable improvement. These milestone payments were negotiated among stakeholders.
The project funds WLZ to offer a 2-year tailored programme for each child that addresses a range of needs and builds a variety of strengths and skills empowered by their link workers, who are based in their schools and work with children and families alongside multiple local partner charities for specialist support. WLZ contracts its charity partners and operates a practical shared delivery relationship between the WLZ link workers and the delivery partner session leaders on the ground. Most, but not all, of West London Zone’s work is funded through a social impact bond (often referred to as a ‘collective impact bond’).
Once the cohort is identified, WLZ link workers approach at-risk children and their parents/carers in partnership with the child’s school. The link worker builds a trusted adult relationship with these parties while co-designing the child’s individual support plan. The child’s individual support plan is developed in a co-design phase which uncovers information about the child’s strengths, interests and skills – which informs design of a phased support plan. Developmental support is conducted by WLZ link workers and WLZ delivery partners (32 as of Autumn 2018) provide specialist support to participating children.
The project was inspired by the Harlem Children’s Zone, a charitable enterprise in New York initiated to support children from ‘cradle to college’. The premise of the West London Zone intervention lies in the idea that issues relating to children living in deprived neighbourhoods are complex and cannot be solved using a single agency or intervention.
WLZ values outcomes based on expected costs and success rates for the 2-year programme. In 2015-2016, a pilot implementation study was undertaken to inform these estimates. WLZ was the first time this set of stakeholders had worked together in this way.
The pilot was philanthropically funded and run in parallel with the development of the social impact bond financing model. The set up, delivery, and evaluation of the pilot together with the SIB development cost £580,000.
The WLZ pilot implementation study offered insights into what performance could be expected when otherwise individual services were combined in a new delivery model.
Project milestone payments in the WLZ SIB are based on:
Delivery costs per participating child incorporated the assumption that different children would require different support. This was converted into an average figure for payment value estimation.
Iterative discussions among WLZ and BFM took into account: average delivery costs per participating child, payment proportion assigned to different milestones, the likelihood of success among the cohort of achieving milestones.
The project identified eligible children in disadvantaged communities in LBHF and RBKC using risk factor analysis. Analysis steps are outlined in Table 2. WLZ expects to work with at least 700 children over the course of the SIB-funded project. Cohort identification data provides the baseline for measuring progress of children participating in WLZ services. Step 5 in the risk factor analysis (target cohort agreement) ensures that both schools and councils verify children for inclusion, which reduces the risk of perverse incentives leading to WLZ “cherry picking” children with a higher likelihood of reaching improvement milestones (and associated payments).
The risk factor analysis combines school level administrative and demographic data with interviews with school staff, and is verified using the WLZ My Voice survey. This collects data via multiple self-reported measures. This helps determine children’s emotional wellbeing, trusted adult networks, engagement with school, peer relationships, and parental relationships.
|1) Prioritisation and participation||
WLZ service is school-based and only includes children at participating schools
WLZ prioritises schools for participation based on combining publicly available data on demographics, progress, attainment gaps and absence rates.
Prioritised schools then have a choice on whether to participate in WLZ SIB
|2) Data generation (school)||
Following school participation, WLZ compiles school-level data to assess individual children's economic disadvantage (using the proxies of receipt of free school meals or 'pupil premium' payments), school attendance, and attainment levels of English and Maths.
Additional qualitative input is collected through discussion with staff to gather their insights into individual children's school engagement, wellbeing, and family context (incl. parental engagement with the school and presence of trusted adult network).
School staff also highlight if an individual child has: an Education, Health and Care Plan, a Child Protection Plan, involvement with Children and Adult Mental Health Services (CAMHS) or any other social services.
|3) Data generation (survey)||
WLZ collect additional data
Children 9-16 years old are eligible to complete the ‘My Voice’ survey, which is a composite of multiple self-reported measures.
Children 5-8 years old are eligible to compete the Strengths and Difficulties questionnaire, either themselves or via a parent or teacher on their behalf
|4) ‘At-risk’ identification||
Each piece of collected data has a ‘risk threshold’. Thresholds are determined via national datasets, academic paper, or government policy.
Individuals marker as above a risk threshold is deemed ‘at risk’ on given characteristics and assigned ‘1’. Anyone below the risk threshold is deemed ‘not at risk’ and assigned ‘0’ for that characteristic.
Children with the largest ‘at risk’ levels, summed across all characteristics are shortlisted for inclusion in the target cohort.
|5) Target cohort agreement||WLZ and school staff confirm those to target from the list|
The impact of services provided – improvement in social and education outcomes – is measured via a “rate card” of progress outcomes. This approach does not involve a control group, but rather achievement of defined milestones. Payments are allocated across six “milestones”, which are listed in Table 1.
Outcome funding is divided equally across the six possible outcomes payments (Table 1). Half is allocated towards service engagement (#1, #2, #3) and half for outcome payments (#4, #5, #6).
There was a different payment milestone framework at project launch, but this was revised after a year as WLZ and commissioners determined that the original mechanism was too complex, and not all of the data could be collected in the way required to conduct the measurement.
The WLZ SIB uses historical baselines for the outcome payments #4, #5, and #6. Historical baselines use the data collected during the risk analysis / identification process described above.
|Payment #||Milestone(s)||Timing||Payment %|
|No payment||Identification - Child is identified as eligible for early intervention support.||-||-|
|1||Sign up - Child/family gives consent to participate.||FY 1 - Q4||17%|
|2||Engagement - Sufficient interactions with link worker and attendance at partner support payable at the end of year. Applicable if child attends at least six formal engagement meetings with their link worker and attends at least 75% of the support sessions scheduled with partner charities and link workers.||FY 2 - Q1||17%|
|3||Engagement - Same interactions and attendance as required in Payment #2, but for following year.||FY 3 - Q1||17%|
|4, 5 & 6||
Achievement - Applicable at the end of 2 years of engagement, if one engagement payment (2 or 3) has been met. 3 final payments in a possible "rate card" of 7 at end of 2-year programme. Payments must include at least one outcome from "academic"/"attendance" outcomes.
Rate cards payments include:
|FY 3 - Q2||49%|
Likelihood of success is based on estimates of how many children might be expected to improve by participating in WLZ’s service, based on the pilot implementation study. WLZ and BFM agreed a ‘likelihood of success’ for each outcome, which incorporated expectations that most service recipients would engage with services (milestones #1, #2, #3), but fewer would achieve end outcomes (milestones #4. #5, #6).
WLZ SIB stakeholders used a sensitivity analysis to account for different project scenarios. Table 3 outlines the results of this analysis for base, high, and low likelihoods of success rates of children participating in the WLZ SIB. Figures used are examples, rather than than the actual ones used.
|Payment #||Milestone||High||Base case||Low|
|2||1st annual engagement payment||90%||80%||70%|
|3||2nd annual engagement payment||80%||70%||60%|
|4||1 out of 7 outcomes achieved||70%||60%||50%|
|5||2 out of 7 outcomes achieved||60%||50%||40%|
|6||3 out of 7 outcomes achieved||50%||40%||40%|
In Table 4.3, ‘sign-up’ is 100% for each scenario because commissioners agreed to outcome payments for each eligible child signed up. Total numbers of children worked with is generally limited by school budgetary constraints, rather than risk analysis or programme interest. WLZ had a target number of children to sign up and expected to be able to reach this target no matter what, as demand is sufficiently high.
This sensitivity analysis also informed the financial viability of the WLZ SIB. Different scenarios were explored to understand:
Financial modelling estimated:
Total payments from all commissioners together (Local Authorities, schools, philanthropists, and the National Lottery Community Fund) were estimated to be between £3.5m-£4m across both LBHF and RBKC. West London Zone received £550,000 loan from Bridges Fund Management as working capital to finance the up-front work required under the contract. The re-payment of this was linked to WLZ’s success in achieving outcomes, such that WLZ was partly protected if outcome payments were lower than targeted.
West London Zone received £150,000 from the “Stepping Stones Fund”, a collaboration between UBS Optimus Foundation and the City Bridge Trust. This was to be used as partial first-loss payment for the investors, if the intervention was unsuccessful against its targets. As explained in the Commissioning Better Outcomes Fund in-depth review:
This safety net within the model meant that the investors could essentially commit to a model with a number of innovative, and untested, elements. Importantly, though, if the SIB model was successful in the first year, then WLZ could use the grant from City Bridge Trust and UBS as additional money in their service. This therefore meant WLZ was still motivated to ensure the intervention was a success and only use the money as first-loss payment if necessary.
Primary and secondary sources have been used for this case study. The secondary sources are highlighted in the text, and the primary sources are listed below.
7 minute read
The Ways to Wellness (WtW) social impact bond (SIB) launched in 2015 and is ongoing. It was the 1st SIB funded in the United Kingdom (UK) targeting health outcomes. The project will run for 7 years and end in 2022.
Core project stakeholders (and their project roles) include: Newcastle Gateshead Clinical Commissioning Group (CCG) (commissioner), Commissioning Better Outcomes Fund (commissioner), Cabinet Office Social Outcomes Fund (commissioner), Ways to Wellness (provider), Bridges Fund Management (investor) and Social Finance (who provided advice during development). The SIB was initially developed by Newcastle West CCG, which merged with Gateshead CCG and Newcastle North and East CCG into Newcastle Gateshead CCG.
Ways to Wellness (WtW), a separate legal entity, was created to coordinate implementation of the SIB.
The project targets 8,500 patients aged 40 to 75 living with long term health conditions in areas of high socio-economic deprivation in West Newcastle Upon Tyne, UK.
Long term health conditions (like diabetes or some types of mental illness) disproportionately affect those facing socioeconomic difficulties. West Newcastle upon Tyne is among the 40th most deprived areas in England, with a higher-than-average receipt of sickness and disability-related benefits and 18% of residents recorded as living with a long term condition (LTC).
The project targets improvements in sense of wellbeing and reductions in use of secondary healthcare services through self-management of long-term conditions. WtW earns payments from commissioners through improved health outcomes and associated reductions in care costs for Newcastle Gateshead CCG.
The project funds a consortium of service providers to provide social prescribing for eligible patients via link workers. Social prescribing enables GPs to refer people to a range of local, non-clinical services. At the outset, project service providers included First Contact Clinical, Mental Health Concern, HealthWORKS Newcastle and Changing Lives. Health WORKS withdrew from service provision August 2017 and Changing Lives March 2018. GP practices and patients were redistributed to the remaining two providers.
WtW link workers offer support to patients by helping them to identify meaningful health and wellness goals, and providing support to help them access community and voluntary groups and resources. Social prescribing recognises that people’s health is influenced by social factors as well as clinical ones. Social prescribing aims to provide people with a variety of social activities, typically through voluntary and community sector organisations, such as volunteering, arts activities, group learning, gardening, befriending, cookery, healthy eating advice and a range of sports.
Stakeholders negotiated outcomes prices based on estimates of fiscal cost savings. Cost savings were calculated from primary and secondary health care costs, social care costs and health related benefits.
The WtW SIB was developed when there was a paucity of costing and implementation information on social prescribing. The original business case was informed by the “People Powered Health” programme, an earlier study under Diabetes Year of Care and consultations with prospective service providers.
The “People Powered Health” programme included social prescribing as an alternative approach to service provision, as well as a cohort from West Newcastle. North East Quality Observatory System (NEQOS) and Social Finance undertook further desk research to validate these findings.
Evidence informing the WtW SIB included:
WtW based outcome payments on (potential) net cost savings for patients with long term conditions, if the provision of social prescribing improved patient outcomes. Additionally, WtW conducted consultations with prospective services providers and presented expected costs of delivering a social prescribing service in West Newcastle upon Tyne. This informed refinements to cost estimates in the SIB’s business case. (This market engagement activity was conducted prior to a formal procurement process taking place – while some of the same service providers were involved, the two processes were independent and both conducted openly and fairly).
This work was used to create a business case for the SIB and social prescribing interventions predicated on cost savings.
This estimation approach does not account for the wider economic benefit, or ‘intrinsic value’ of benefits to participating people with long term conditions.
The project identified eligible patients in West Newcastle Upon Tyne, UK using eligibility criteria including:
WtW aims to offer services to over 80% of the patients who meet referral criteria. At the outset of the project, the project was expected to reach 8,500 patients in the area. The target population over 7 years (allowing for population growth and new entrants) was circa 15,000. The original population estimate was higher but was refined and reduced when it was discovered that patients with more than one condition were being double-counted.
The project measures improvement two outcome measures:
Triangle Consulting’s “Wellbeing Star” measures improvements in self-reported wellbeing across several categories (see also). This involves patient and link worker joint-assessments every six months and pre-post evaluation of the average change for the whole cohort between the initial and most recent Wellbeing Star measurements.
For Outcome 1 (improved sense of wellbeing), The National Lottery Community Fund and Cabinet Office agreed to pay more towards this lead indicator after considering: if Ways to Wellness would deliver wider patient benefits over and above the measurable secondary care savings; the level of risk to the social investor and whether this was acceptable; the level of buy-in of the commissioner and proposed providers (whether they were on board with the contract, and how confident they were they about delivering the outcomes in the contract).
Reduced expenditure on hospital services was calculated based on average costs within the WtW cohort compared to a matched cohort of patients in Newcastle North and East.
Payments for improved sense of wellbeing began in 2015. For healthcare expenditure reductions, payments are made per patient per year. Payments for reduced healthcare expenditures began in August 2017.
The wellbeing star (4.2.1) and reduced expenditure (4.2.2) outcomes were chosen to ensure that the project was viable over the short term. Expected savings were only expected to accrue, based on improved health and wellbeing for social prescribing recipients, 1-5 years after patient admission. To make the contract viable for providers, improved sense of wellbeing was incorporated as an outcome to create earlier “lead” payments. This reduced the level of risk to the social investor and therefore the cost of working capital.
WtW receives outcome payments from commissioners, and then sub-contracts with service providers via activity- and output-based payments. Payments to service providers involved variable activity payments for the first two years. These payments combined with a fixed retainer, which was redistributed toward variable activity payments at contract renewal after 2 years. Variable payments included:
WtW being the SPV in the SIB shielded service providers from downside risk (financial losses) of not achieving the overall payment outcomes. Upside (financial benefit) was shared between the social investors, WtW and the providers. The proportions of this sharing are not disclosed.
The WtW project uses both historical baselines and matched controls to evaluate the impact of provision of social prescribing services. The historical baseline involves joint completion of the Wellbeing Star, which reduces self-reported bias risks which would negatively impact attribution. The matched controls involve differences in average health expenditures within the WtW cohort compared to a matched cohort of patients in Newcastle North and East. It may be possible to isolate a causal effect for WtW on differences in average health expenditures.
Independently of the payment mechanism, the National Institute for Health Research began conducting research which began in July 2018 and includes economic analysis. Research is expected to conclude in October 2020.
Based on publicly disclosed information, it is not clear how likelihood of success was incorporatedinto the SIB payment mechanism.
Analysis was done to develop three basic scenarios:
Based on publicly disclosed information, it is not possible to verify the exact link between the different scenarios for outcomes achievement and the likely levels of investor losses or returns, beyond the basic principles stated above.
The project’s investor, Bridges Fund Management, committed to provide WtW with a £1.65m investment, with £1.1m drawn down by WtW. The facility availability period has since expired and it is not expected that any further SIB investment will be needed / received.
The investment from Bridges Fund Management is 100% at-risk. Repayments and returns are linked to achievement of project outcomes. The National Lottery Community Fund / Ecorys ‘Deep Dive’ review published in 2015 makes the following statement in regards to returns:
If, and only if, base case success targets are achieved the estimated money multiple over 7 years will be c.1.38 times the initial investment. If outcomes achieved are lower than base case the multiple could be much lower and conceivably all investment could be lost.
The figure given in this statement does not account for the costs to Bridges Fund Management of their involvement in project development and their ongoing role in project management. These costs mean the eventual financial return to the capital providers will be lower than stated here.
Primary and secondary sources have been used for this case study. The secondary sources are highlighted in the text, and the primary sources are listed below.
6 minute read
The Essex Multi-Systemic Therapy Social Impact Bond (SIB) launched in 2012, running to 2019. It was the first SIB to launch that was led by a local authority.
Core project stakeholders (and their project roles) included: Essex County Council (commissioner), Action for Children (provider), Social Finance (intermediary), and a consortium of investors. The consortium of investors included: Big Society Capital, Bridges Fund Management, Social Ventures Fund, Charities Aid Foundation, The King Badouin Foundation, Tudor Trust, Barrow Cadbury Trust, and the Esmee Fairbairn Foundation.
If a young person in Essex, UK who is at risk of entering state care (based on a combination of risk characteristics)…
The project targeted young people aged 11-16 in Essex, England who were at risk of entering care due to behavioural problems or family breakdown.
The project targeted improvements in social outcomes based on the reduction in days the cohort spends in care, and generated outcome-based payments by Essex County Council paying out a share of their cost savings resulting from reduced care placements for at risk young people. Care placements can cost over £200,000 per annum and research suggests it is substantially harder to address behaviours after entering care.
The Essex Edge of Care SIB operated via a special purpose vehicle- Children’s Support Services Ltd (“CSSL”). CSSL was responsible for managing the outcomes contract, the performance of the service and paying the service provider. The project intermediary, Social Finance, noted the aim of the intervention was to:
improve parenting skills of parents and carers which in turn impacts the behaviour of the adolescents so that they do not become looked after or the amount of time they spend in care is reduced.
CSSL funded Action for Children, a national children’s charity, to deliver multi-systemic therapy (MST), an intensive evidence-based family therapy. MST targets specific problems and breaks negative cycles of behaviour via the promotion of positive social behaviours. An Action for Children service delivery manager oversaw project implementation and the project was delivered by two teams of four therapists, each overseen by a supervisor with assistance from a business support officer.
Essex County Council valued outcomes based on their projected cost savings from young people diverted from care due to service provided.The outcome payment did not account for the wider benefit or ‘intrinsic value’ of improved outcomes, or improved efficiencies. This aligned with the “Prevention” approach outlined in the GO Lab “Pricing Outcomes” guide.
The impact of services provided - reduction in care placement days – was measured as the difference in aggregate days spent in care between a historical comparator group who did not receive MSTand those receiving MSTthrough the SIB. The reasons for using reduction in aggregate care placement days as measure of success were that:
Secondary outcomes that were measured, but not linked to SIB payments, included: educational engagement, offending, and personal wellbeing.
Projected cost savings in the Essex Edge of Care SIB were the difference between the comparator and intervention groups. This amounted to £120 per care day avoided.
Value calculations involved two UK government datasets about children in care: SSDA903 returns and Section 251 budget data. SSDA903 returns collate annual local authority returns and information based into categories about “looked after children” (LAC) - children who are in care. Care placement categories include residential, foster, unknown, and “other”. Section 251 budget data collates local authority statements on planned and actual expenditure for education and children’s social care.
Projected costs for intervention and comparator groups were the number of days spent in care multiplied by the unit costs of care. The number of days spent in care is published in SSDA903 returns. Gross costs were a multiple of days spent on each type of care multiplied by unit costs for care, which were estimated in three steps:
Once estimated, the unit costs of care placement were sense checked using outputs from the Local Authority Interactive Tool (LAIT). The LAIT provides a single central evidence base for data related to children and young people sourced from various UK government departments.
The reasons for using this payment structure, on top of measures of success, were:
This estimation approachdid not account for intrinsic value of children being out of care, or benefits associated with not entering care (e.g. positive socioeconomic spillovers). It is also not clear if the payment mechanism accounted for additional administrative costs. A 2016 review of the initial 3 years of implementation outlines some of these administrative activities, incurred due to using a SIB compared normal delivery of a MST programme. These included governance, performance management, and payment by results processes. The SIB required additional effort on data management and payments, by all parties, due to the complex and intensive nature of the focus on results.
The project identified service recipients via a single source referral process – referrals were taken from Essex County Council Children’s Social Care ‘quadrant resource panel’ and assumed the young person was already a child in need or subject to a child protection plan.
Young people were tracked for 30 months after the start of the MST course (which lasts 4-5 months), and their outcomes were measured quarterly.
Outcomes were measured quarterly and paid by Essex County Council to CSSL (the special purpose vehicle). This regular payment allowed investors to recycle their capital and reinvest in ongoing MST provision.
Payments were entirely based on a single outcome and reflect the reduction in costs of care to Essex County Council.
The SIB used a historical baseline to evaluate the effect of the provision of MST, within the SIB, on reduced care admissions. The historical baseline for each year used information from cases not receiving MST from the previous 3 years regarding: project eligibility, numbers entering care, numbers not entering care, and individual’s average length of stay in care. Estimations for the outcomes for the counterfactual of participating in the MST include:
There was no concurrent comparable control group and it is not possible to isolate the causal effect of the intervention. It is difficult to compare the performance of the Essex MST service to others due to different project scales and contexts.
Likelihood of success was based on the expected number of children that could benefit from MST each year. This data was collected through a six-month feasibility study and historical government datasets, Social Finance led data cleaning and analysis.
Based on an overall referral target over five years of 380 families, “medium level performance” was set at 110 young people being diverted from care.
Edge of care volume calculations assumed that 70% of individual children per year could be eligible for interventions. This was due to ineligible cases related to autism and a lack of parental engagement. Of this group, there was a 65% assumed likelihood of entering care in the following 12 months based on experience and case file analysis.
The Essex MST SIB tied outcome payments to an average of averted unit costs, and did not use future performance projections to account for the likelihood of different outcome scenarios.
Outcomes payments were capped at £7.2 million. Outcome payments from the fund varied based on cost savings achieved.
Primary and secondary sources have been used for this case study. The secondary sources are highlighted in the text, and the primary sources are listed below.
8 minute read
The Educate Girls development impact bond (DIB) launched in 2015. It was the 2ndDIB (worldwide) and focused on improving education outcomes for children in Rajasthan, India. The project ended in 2018. Project outcomes were met and core project stakeholders deemed it a successful pilot.
Core project stakeholders (and their project roles) included: Children’s Investment Fund Foundation (CIFF) (commissioner), Educate Girls (provider), UBS Optimus Foundation (UBS OF) (investor), Instiglio (intermediary), and IDinsight (evaluator).
If girls and boys in Rajasthan, India at risk of not achieving potential levels of enrolment and education outcomes…
The project targeted out-of-school girls aged 7-14 and girls and boys in grades 3-5 in the state of Rajasthan in India.
The project focused on children living in the Bijoliya, Mandalgarh and Jahajpur blocks in the Bhilwara district of Rajasthan. The DIB funded interventions for children attending 166 government schools across 141 villages. These schools were randomly selected from a sample of 332 schools in 282 villages.
The project targeted improved education outcomes via out-of-school girls (re)enrolled in school and education quality, as measured by test scores for girls and boys. It generated outcome-based payments by paying out based on cohort level payments assigned by commissioners to evaluated outcome metrics.
The project funded interventions by Educate Girls, an NGO operating in Rajasthan, India. Educate Girls uses an integrated community-based approach to education that involves increasing access to education for primary school-aged children in rural areas, especially girls. In rural parts of Rajasthan, girls are out of school at twice the rate of boys and only 50% of women can read or write.
Educate Girls has history of success with education enrolment and quality improvement via community and cultural engagement. In 2015 the DIB contract launched to increase the scale of their work. Educate Girls’ strong community ties allow them to positively communicate the value of education within rural communities. Educate Girls account for cultural context and accordingly adapts its education provision approach. They create tailored teaching programmes to adapt user needs and improve the quality of their education.
A DIB-funded approach allowed the flexibility and support required to address difficult education challenges in Rajasthan. These challenges included one in ten girls aged 11-14 not being enrolled in school and less than a quarter of rural children in Grade 3 being able to read at a Grade 2 level or solve a subtraction problem.
The Educate Girls DIB involved negotiated outcome prices based on estimates of expected project costs and project performance. Expectations were based on data from a randomised controlled trial (RCT) and additional baseline collection. RCTs involve one group of individuals receiving an intervention/service and their outcomes are compared to a control group of individuals, who share the same characteristics as the intervention group, but did not receive the additional intervention/service. This aligned with the “Efficiency” approach outlined in the “Pricing Outcomes ” guide.
Price per unit outcome did not inform the overall outcome payments – rather, outcome payment was determined by improvement across the whole cohort. Individual price per outcome can be calculated from overall outcome payments and allocations.
Projected performance was based on data from a RCT undertaken by Educate Girls in the Jalore district of Rajasthan. The RCT was designed pro-bono by University of Michigan faculty. This information outlined the effects of Educate Girls’ activities on enrolment, retention, and learning outcomes (see also). Instiglio, the intermediary, combined performance data from Educate Girls and target area characteristics to estimate enrolment and learning outcomes. The project gathered baseline data from a census-like door-to-door survey.
The value of project payments was decided via CIFF and UBS OF negotiating based on expected service costs and UBS OF receiving an internal rate of return (IRR) of 10%. An IRR reflects the rate of return to investors that accounts for the risk which they absorb with their investment and how much they are paid for holding this risk. It is possible to calculate unit values for the outcomes, however this is via working backwards from overall outcomes payments and costs.
UBS OF provided US$270,000 as investment based on being the equivalent value of expected service provision costs (17,332,967 Indian Rupees). An IRR of 10% on an investment of US$270,000 corresponded to an expected outcome payment was US$367,000, 87% of CIFF’s total available outcome funding, if the targets were met (79% enrolment and +5,592 points on ASER scores). If performance exceeded these targets, CIFF could draw from a total outcome funding of US$422,000 available, which would have enabled UBS OF to earn an IRR of 15%.
According to Alison Bukhari of Educate Girls, the estimated service delivery cost took into account some of Educate Girls’ standard administrative costs to operate the programme, but did not include any administrative or transaction costs for designing or managing the DIB. A 2016 Devex piece references that if the DIB had been bigger, the overhead cost would have been a lower proportion of the project’s total costs. However, further information on what was incorporated in overhead costs is not available.
The Educate Girls DIB was set up by stakeholders as a pilot development impact bond. Its payment approach has limitations - the payment mechanism does not set a clear financial value for individual education outcomes and is based on the cost of delivering the intervention. While a value can be extrapolated, it does not necessarily reflect an outcome preference or valuation.
The project focused on children living in the Bijoliya, Mandalgarh and Jahajpur blocks in the Bhilwara district of Rajasthan. The project targeted two different, but overlapping, outcome populations. One outcome (enrolment) focused on out-of-school girls aged 7-14 and the other (learning) focused on girls and boys in grades 3-5.
The DIB funded interventions for children attending 166 government schools across 141 villages, matched to a cohort of equal number. These schools were randomly selected from a sample of 332 schools in 282 villages, from a random sample of 332 schools from an eligible population of 396 schools.
IDinsight, the independent evaluator, divided the 332 sampled schools into treatment and control groups. IDinsight used pairwise matching to balance the characteristics of treatment and control groups. Pairwise matching involves assigning villages to “pairs” based on their characteristics and randomly assigning one to the treatment group and one to the control group. Pairwise matching characteristics included:
Outcome payments were allocated to “buckets” for education enrolment and quality. Pre-allocated payments only included those for minimum target outcomes, i.e. US$367,000 of the US$422,000 outcomes budget. 20% was allocated for enrolment rates and 80% for education quality. CIFF pushed for a focus on improved learning due to an identified disparity between the education outcomes of boys and girls in Rajasthan. The project measured success by improvement in enrolment and education quality based on:
Enrolment was measured by identifying the percentage of out-of-school girls in target villages based on door-to-door surveys. This involved a pre-post evaluation of enrolment rates. IDinsight, the independent evaluator, verified enrolments by visiting schools and cross-checking school registers against interview data from principals, teachers, and parents.
Education quality was measured using standardised testing for literary and maths. The project used the Annual Status of Education Report (ASER) test, which measures proficiencies in Hindi, English and mathematics. The ASER test is a widely used and accept method to provide rigorous assessment of the education outcomes of social sector programmes.
Measurement involved annually assessing a panel of students using ASER tests over the three-year evaluation. The impact was calculated by aggregating the differences between cohorts’ baseline and final learning levels for the intervention group and comparing this to the aggregate change in test scores in the matched control cohort. Using aggregate, rather than average, scores linked the quality education payments to improved enrolment in effective schooling.
Impacts were measured over three years. Data collection began in August 2015 and ended in February 2018. Final results were announced in July 2018. Project targets and final measures achieved were:
The figure below displays outcome progress over the course of the project (2015-2018).
The payment was made in one lump sum in July 2018, following the verification of successful project results.
Outcome funding was divided between lump sum payment for investors and an additional payment to Educate Girls to incentivise reaching programme milestones. UBS OF negotiated with Educate Girls to pay the incentive payment at a rate 32% of interest payments which the Foundation received if outcomes were met, up to an IRR of 15%.
The payment structure rationale was to pass 'upside' financial benefit risk to providers and incentivise performance without passing on downside risk (financial loss). This upside benefit was in addition to non-financial risks, such as reputational risks, that Educate Girls was exposed to in the DIB. Payments were set up as as a single pay out to the investor at the end of the project.
The Educate Girls DIB used matched controls and a historical baselines to evaluate the effect of the service provision on education outcomes.
Assessments for learning outcomes compared the intervention and matched control groups of schools. Evaluations were made at baseline in September 2015 and in February of each subsequent year (2016, 2017, 2018). In total, IDinsight conducted over 25,000 assessments across more than 11,000 students. For the intervention group, which was compared to a historical baseline, IDinsight enumerators followed up with students who were absent from school, such that the attrition rate over the three-year evaluation was below 4%. Each August, IDinsight validated additions to the out-of-school census.
Due to census costs, IDinsight did not estimate enrolment in control villages. The Instiglio DIB design memo notes several potential factors influencing the enrolment of out of school girls. A causal effect cannot be measured for the effect of Educate Girls’ program on enrolments, due to the a lack of comparator group.
IDinsight incorporated several measures to mitigate bias and ensure results were robust. To mitigate bias IDinsight operated independently from the Educate Girls program and its field staff. IDinsight project enumerators were sent to a variety of schools and were not informed of the village’s assignment to treatment or control. Their data were also collected digitally and reviewed daily to evaluate if there were any missteps in collection. The risk of skew, due to gaming or cheating, in the ASER tests was assumed to be low. This is due to the difficulty in “faking” ability in the language and mathematical reasoning tests. The size of the study (roughly 12,000 participants) also supported the statistical robustness of estimates.
It is not clear how likelihood of success was incorporated into planning of the DIB payment mechanism.
Outcome calculations focused on the aggregate difference between treatment and control groups. IDinsight calculated the minimum detectable effect size base on a sample of 332 schools in 282 villages. This showed that if true treatment effect on ASER scores was 0.47 points, then the evaluation would have a 20% chance of failing to distinguish the treatment effect this – giving a false negative. The observed difference in learning gains was 1.08 ASER learning levels, a difference that statistically significant at the 1% level. This implies that the probability of observing this difference, if there is actually no treatment effect, is less than 1%.
Statistical uncertainty was not incorporated into outcome payment calculations. Example potential payment scenarios are outlined in Instiglio’s design memo, but these were updated following implementation of the EG DIB.
The maximum total outcome payment was US$422,000. The minimum outcome payment of US$0 was tied to the project having no impact at all.
Final evaluation reports from Educate Girls, IDinsight, and the UBS OF highlight limitations and lessons learned from stakeholder perspectives. External case studies offer further analysis, conclusions, and recommendations about the Educate Girls DIB.
The use of an outcomes-based payment mechanism created several positive spillovers. The mechanism created incentives for Educate Girls to more rigorously evaluate the data collected and analysed by IDinsight. Subgroup analyses, available due to the breadth of information collected about study participants, allowed Educate Girls to understand how their services were impacting individuals differently within the intervention group. This supported Educate Girls’ substantial adaptation and expansion of activities in project year 3.
Learnings from the Educate Girls DIB are directly feeding directly into the Quality Education India DIB, an ambitious US$11 million project coordinating service provision across three NGOs.
8 minute read
In this appendix, we discuss a new approach to assessing value for money, for those with analytical capacity and a strong need to provide clear VfM justification of the price set. We would like to invite discussion on this approach and its practical utility, and how it might be further developed.
After deriving the value and estimating the costs, you will have a better understanding of a range of possible prices (the range between P3 and P1 in the figure below). Depending on how you set the price, there might be per outcome net financial benefits (P2 less P1), non-financial benefits (P3 less P2), or the combination of both.
Another case scenario that could happen is where financial benefits are not greater than costs, but non-financial benefits are (P2 < P1 < P3). This is where decision making could become harder, depending on how much the society values the non-financial benefits at the time. See more on this in our blog.
Considering that resources are constrained, the goal should be to use resources optimally so to achieve the intended outcomes. In the UK, for example, the HM Treasury Green Book defines “potential Value for Money” as: “optimising social value (social, economic and environmental), in terms of the potential costs, benefits and risks”.
Value for money has become a much-buzzed about concept in international development in recent years. Primarily used in developed or middle-income countries—where there is more money and more options—a central question is this: should we also use this tool for projects in low-income countries?
The simplicity of the term ‘value for money’ belies the diversity of practical interpretations of the concept. At its best, it is a fully integrated, value creating, impact enhancing practice that is informed by various sources and stakeholders, which supports ongoing organisational and programmatic improvements. It is integral to organisational performance. However, there are also fears that it can be interpreted narrowly to focus on cost-minimisation, or on short-term rather than long-term value, or to justify doing what matters to those providing aid rather than those receiving it.
Based on guidelines published by the government of the UK, there are three criteria to assess the VfM, namely:
The three ‘E’s framework shows that the Value for Money agenda is not just about cutting costs. Maximising actual outcomes and impacts that are bought with (taxpayer) money is the key part of the VfM agenda.
There is a fourth criteria as well that is recommended to have taken into consideration when applicable: equity. This is relevant in all stages of impact bond models, and represents ‘Spending Fairly’, i.e. the extent to which services are available to and reach all people that they are intended to.
Are inputs of appropriate quality bought at a minimised price?
This criterion is about the costs borne throughout the programme. Inputs (e.g. time, staff, consultants, raw materials capital, and etc.) should have been procured at the least cost for the relevant level of quality.
In order to satisfy this condition, the price paid for outcomes should be the least possible and there should be no other way to achieve them cheaper. This requires some level of market study and then further negotiations with different providers (and/or investors). A comparison of costs is essential for price setting regardless of the intervention being original or based on a legacy programme. We have already discussed how to estimate costs earlier in this guide.
Keep in mind that the outcome price paid to the provider (and/or investor) is not the outcome payer’s only cost, as additional resource may be needed, such as for monitoring and evaluations, which count as inputs. The overall price which needs to be minimised is the sum of all the costs borne by the outcome payer. This price should be comparable to compatible services to satisfy the economy criterion of VfM.
Some of the costs of the outcome payer (i.e. demand side) in impact bond models are described in the table below. This should not be confused with the costs of the ‘supplier’ side discussed earlier in the guide.
DEVELOPMENT & DESIGN
Pre-contract signing search and information costs
CONTRACT SETUP & NEGOTIATION
Governance, monitoring and evaluation costs
|1. Development-staff time||1. Legal advice on contract design (internal &/or external)||
1. Monitoring Costs
a. Outcomes verification costs
b. Contract management staff costs
|2. Feasibility studies (internal &/or external)||2. Financial advice on contract design (internal &/or external)||2. Evaluation Cost|
|3. Business Case||3. Procurement Costs||3. Governance Costs|
|4. Early phase legal costs||4. Other Costs||4. Other Costs|
|5. Market engagement costs|
|6. Other Costs|
The outcome payer should have (or develop) a good understanding of how much will end up being paid for the outcome using template such as the above or based on comparable contracts already running. If you are considering an OBC model, you might feel that while the service is being delivered is in line with the service specification (or equivalent), it is not producing as many good outcomes for individuals as you think it could.Current costs may be difficult to assess in full as they include costs that may be difficult to apportion (e.g. managerial and other staff cost, including ‘overheads’). In an outcomes contract or SIB, there will be costs built in for investor returns if the service achieves its aims, and sometimes for increased performance management and data collection. But these costs might worth paying if performance is improved enough compared to the full cost of existing service provision, especially when considering managerial / overhead costs which are often left out. This is where the other criteria of VfM come into place.
How well are inputs converted into outputs?
Efficiency is generally defined as the value of ‘outputs’ in relation to the total cost of ‘inputs’ (at the relevant level of quality). This criterion also assesses anything affecting the conversion of inputs into outputs, including setting and measuring outcomes, monitoring, evaluation, project management, and so on. The foremost role of price in satisfying this condition is undeniable, as cost per output needs to be compared to benefit per output as part of the assessment.
Normally it should be possible to distinguish between output and outcome, e.g. in most of the preventative and early intervention programmes the target is of an ‘output’ nature which is positively associated with some ‘outcome’ in the future. Occasionally however, there might be cases where you cannot distinguish between output and outcome in an OBC due to the type of indicators you use for outcome payment. This is where the difference between ‘efficiency’ and ‘effectiveness’ becomes less apparent. So, you may be constrained by a lack of reliable information to indicate both full costs and output achievement for existing services.
Another issue affecting efficiency of IBs compared to either OBCs or conventional fee-for service might be lack of competition in the supply market due to the limited number of investors and/or providers. It is highly recommended however to encourage competition between potential providers as this will tend to promote efficiency. This will be important in an era of ‘more for less’ across the public sector.A comparison of expected costs and outputs could provide some understanding of how well the services are priced and being delivered, both in their own right and in comparison to the conventional methods. The ratio of output to costs could be a measure for comparison.
How well do those outputs achieve outcomes?
Unlike fee-for services, OBCs use outcomes as the basis for payments, and therefore the total cost of inputs should ensure that the outputs deliver the desired outcome. Price per outcome is probably the most influential factor in your total costs, along with the size of the cohort, so careful consideration is once again advised when setting the price to satisfy this criterion.
General opinion is that outcomes are more expensive than outputs. To illustrate, imagine you are an outcome payer and in a ‘block contract’ or in a ‘fee-for-service’ arrangement, you may be paying £10,000 to engage 100 participants in the training. This is £100 per participant (£10,000 divided by 100 participants). You pay this regardless of the success of the programme in achieving the set outcome. However, if only 20 of them achieve the desired outcome, the price per outcomeis in fact £500 – that’s £10,000 divided by 20 participants. So in reality, you paid £500 per outcome. In an OBC version of the similar programme, you might agree to pay £500 for each outcome to start with. Or, if the programme seeks to improve the “effectiveness” of the spend, you might seek to pay, say, £400 per outcome, such that you can get more outcomes for the same total payment, or the same outcomes for less total payment.
A potential complication with assessing effectiveness in OBCs is that sometimes outcomes are realised only in the long-run and are therefore hard to monitor. For instance, preventative and early intervention programmes are aimed to improve future chances. In such circumstances, the best way is using data analysis is to find (i) the rate of output to outcome conversion, and then (ii) use the information to predict future benefits and costs. Pay attention, discounting and measuring present values are essential when dealing with calculations happening at different times in the future (find more about these concepts in A6 of the Green Book).
The table below gives a simple comparison of the three criteria of Value for Money. You could learn more about VfM analysis in the Economic Evaluation chapter of our Evaluation Guide. For an overview of VfM analysis across the globe see this World Bank document.
|DESCRIPTION||EXAMPLES OF INDICATORS|
|ECONOMY||Evaluating whether the inputs of the appropriate quality are bought at the right price||
Cost drivers; e.g. outcome payment, staff costs, capital, cashflow, & etc.;
|EFFICIENCY||Evaluating how well inputs are converted into specific outputs with expected quantity and quality||
% of targeted outputs achieved;
% of eligible persons achieving target;
Per person cost and benefit comparison
Cost-efficiency (output input ratio).
|EFFECTIVENESS||Evaluating how well outputs from an intervention are converted into sustained actual outcomes. OBCs offer more control over outcomes than conventional methods.||
% of outputs translated into actual outcomes.
% of ‘expected outcomes’ translated into actual outcomes;
In the GO Lab team, Nigel Ball, Executive Director and Mehdi Shiva, Economist coordinated the development of this guide. This would not have been possible without the contributions made by our associates. These are: former GO Lab research interns Vaby Endrojono-Ellis and Lorcan Clarke; Fellows of Practice - Tim Gray, Neil Stanworth, Tara Case and Alison Bukhari; Former Fellows of Practice - Mila Lukic, Tanya Gillett; our partners at Social Finance - Jane Newman, and Marie-Alphie Dallest, Cat Remfry from the Centre for Social Impact Bonds at DCMS as well as Louisa Mitchell, CEO of West London Zone.
We warmly welcome any comments or suggestions. We will regularly review and update our material in response. Please email your feedback to email@example.com.