Aerial view of the Oculus at the World Trade Center with anonymous pedestrians crossing its vast white interior.

How It Works

How We Measure Social Value

Individual clarity. Societal insights. Our methodology is the tagline made literal: we achieve individual clarity by calculating welfare for each person, then build societal insights by aggregating those individuals to understand the group.

Start With the Change to a Real Person

Social value is the difference an intervention makes to people's wellbeing compared with what would likely have happened anyway. SVT keeps that comparison at the person level before any total is reported.

1. Establish the counterfactualDefine what would likely have happened without the programme, policy, asset, or intervention.
2. Follow welfare pathsEstimate how wellbeing changes over time for individual people rather than applying one average value to everyone.
3. Aggregate after the peopleTranslate welfare changes into comparable monetary equivalents, report who gains or loses, then show the total.

Core Terms

Decision-grade
A model is decision-grade when its outputs would be defensible as the basis for a decision whose consequences depend on the represented situation being feasible. It is a higher bar than "plausible numbers" or "fits the data", and lower than "truth". ONEMODEL is designed to be decision-grade.
Counterfactual
The counterfactual is what would likely have happened anyway if the programme, policy, or asset did not exist. SVT compares actual outcomes against the counterfactual, so the reported social value is the change attributable to the intervention.
Welfare paths
Welfare paths describe how each person's wellbeing changes over time. Because ONEMODEL represents individual people, it can model how wellbeing evolves for each one - rather than assigning an average value per participant.
Agent-based modelling
Agent-based modelling represents individual people and how they interact with systems. It differs from population-average methods by keeping each person distinguishable throughout the analysis, so results describe who gains, who loses, and by how much.
Monetary equivalent
A monetary equivalent is a dollar translation of a welfare change. It expresses "how much wellbeing improved" in dollars, so social value can be compared with financial cost and reported alongside other decision inputs.

The Calculation Sequence

Our methodology follows a strict order. The sequence cannot be reversed or compressed without losing the transparency that makes the evidence credible.

  1. Define the Question

    What intervention is being evaluated? What population does it affect? What is the relevant time horizon? What counts as the counterfactual - what would have happened without this intervention?

  2. Generate Welfare Paths

    For each affected person (or representative agent), ONEMODEL generates two paths: a scenario path (with the intervention in place) and a counterfactual path (without). The difference between them, period by period, is the person's welfare change attributable to the intervention.

  3. Preserve Individual Paths

    Welfare changes are not immediately aggregated or averaged. Individual paths are preserved - showing who gains, who loses, through what channels (income, health, security, connection), and over what time horizon.

  4. Translate to Monetary Equivalents

    Each person's wellbeing change in each period is translated into dollars using a stated valuation rule (a valuation function). The rule accounts for the person's circumstances, income position, and local context. This is not a single universal exchange rate.

  5. Discount Monetary Equivalents

    Only monetary equivalents are discounted (using standard financial present-value methods). Raw welfare paths are never discounted - because discounting future wellbeing is a separate ethical question from discounting future money.

  6. Aggregate as a Reporting Operation

    Only after individual paths have been preserved, valued, and discounted does the methodology produce a total. The aggregate is a bottom-up sum of person-level present values - not a top-down estimate imposed on a population.

What Makes This Rigorous

Person-level, not population-level

Every result traces to identifiable persons (or representative agents). You can always ask "who benefits?" and get an answer from the model - not from an assumption.

Counterfactual-based

Social value is defined as the difference between two worlds: with the intervention and without it. This prevents the common error of counting outcomes that would have occurred anyway.

Mechanism-based

Every welfare change in the model arises from a represented mechanism - a causal process that is specified, calibrated, and testable. Nothing is inserted by assumption.

Assumption-transparent

Every valuation choice, calibration parameter, and boundary condition is declared and can be sensitivity-tested. The methodology does not hide its assumptions - it invites scrutiny of them.

Distribution-first

Distributional results (who gained, who lost, by how much) are a primary output - not an afterthought. The total is reported alongside the distribution, never instead of it.

The Single-System Principle

A critical feature of SVT's methodology: hundreds of social value topics are modelled together within one coherent system.

This means:

  • Every topic shares the same persons, definitions, and boundaries
  • A benefit counted under one topic cannot also be counted under another (no double-counting)
  • Interactions between topics are captured (e.g., housing stability affecting health and life satisfaction)
  • The conclusion should not change because the same situation is described differently (technically, frame-invariance)
hundreds of topics, one connected model, person by person. Most social value methods value each topic separately from a fixed table of averages, then add the totals together. That approach creates hidden errors. SVT's single-system design prevents them by construction.

What the Methodology Does NOT Do

Transparency requires stating limitations:

  • It does not claim to be true - no model makes this claim. It claims to be decision-grade: fit for the purpose of supporting real decisions.
  • It does not eliminate uncertainty - it measures uncertainty, reports it, and sensitivity-tests key assumptions.
  • It does not produce a single "right number" - it produces a range, a distribution, and a set of declared assumptions under which the central estimate holds.
  • It does not replace qualitative evidence - stories, testimony, and lived experience complement quantitative measurement. They are different kinds of evidence serving different purposes.

Documented for Independent Review

SVT's methodology has been refined through more than 20 years of external review and practical application. Its formal methodology is documented to journal-publication standard. The methodology paper:

  • States a formal estimand (what is being measured, precisely)
  • Specifies a reproducible calculation algorithm
  • Declares calibration standards and validation protocol
  • Compares with existing approaches (WELLBY, SROI, hedonic pricing, cost-benefit)
  • Reports limitations and minimum reporting requirements

This is not a rule of thumb. It is a documented methodology designed for independent scrutiny and replication.

Common Questions

What is a counterfactual in social value measurement?
A counterfactual is the scenario representing what would have happened to affected persons if the intervention had not occurred. Social value is the measured difference between the actual scenario and this counterfactual - ensuring only the intervention's contribution is counted.
What is "decision-grade" evidence?
Evidence is decision-grade when it would be defensible as the basis for a decision whose consequences depend on the described situation being feasible. It is higher than "plausible numbers" but lower than "certain truth".
How is this different from SROI?
SROI assigns monetary values to outcomes using proxy databases. SVT generates person-level welfare paths through simulation, translates them to monetary equivalents using a declared local valuation function, and reports distributions. The evidence is specific to the population studied - not transferred from another context.
How is this different from the WELLBY approach?
WELLBY defines one life-satisfaction point for one person for one year as a standard unit. SVT does not assume raw satisfaction scores are automatically comparable across people. Instead, it preserves individual paths and applies a local valuation function that accounts for heterogeneity before aggregating.
Can the methodology be applied outside Canada?
Yes. The methodology is implementation-independent. While ONEMODEL is calibrated primarily to Canadian data, the calculation sequence applies anywhere person-level welfare data and counterfactual construction are feasible.

Want to Understand How This Applies to Your Context?

Schedule a consultation. We will tell you what evidence is possible.

Start a Conversation
Externally reviewed methodology, developed over 20+ years400+ engagements (CANCEA, 2002-2026)Featured in the ICPM Social Infrastructure Blueprint (2026)