SPUR

What do they do? SPUR researches and advocates for improvements in housing, transportation, environmental health and public institutions across the Bay Area. More

Why this approach interests us

• Better public decisions can benefit many more people than an organization could reach through direct service alone.
• Housing, transit and environmental policy can affect health through more stable homes, safer travel and reduced exposure to hazards.
• Research, implementation support and coordination offer plausible ways to turn policy ideas into changes in daily life.

Our main reservations

• A policy win does not establish how much SPUR caused or what an additional donation would change.
• Our quantified health pathways cover only part of SPUR's work; their size and attribution remain highly uncertain.

What do you get for your dollar? Our conditional central scenario is about $2.19 million per 10 Bay Area QALYs. It is an illustration of additional health effects, not a probability-weighted expected return or a valuation of every benefit of better public institutions. More

What do they do?

The unit being assessed

The official financial-disclosure page identifies the charitable legal entity and explains its April–March reporting year. SPUR is not SPURR, the separate public procurement organization. The model does not assign benefits of unrelated organizations with similar names. Nor does it create separate charity entries for SPUR’s topic areas or offices.

An ordinary gift funds an institution. There is no public basis in the inspected sources for saying that every new dollar goes to clean heat, every SF donor dollar benefits only SF, or a gift buys a named number of permits. Research staff, policy analysis, education, coordination, management, fundraising and facilities all support delivery. Some expenditures may contribute to several topics; this is a reason to allocate shared costs once, not to assume those costs disappear or to penalize overhead twice.

The current annual report describes regional housing, transit, building electrification, governance and resilience work. It also describes education and convening. Its financial summary is explicitly unaudited as of March 31, 2026. These provider accounts identify current activity and claimed contributions; they are not a controlled evaluation of health or an audited topic-by-topic allocation.

Housing and planning

Housing work can involve analysis of permitting, zoning, development feasibility and the institutions that implement these rules. The Reinventing Downtown brief proposes a coordinating authority for capital projects, real estate and public financing. Its February 2026 description is a proposal and rationale, not a completed housing intervention.

For health, the relevant sequence is not report → law → QALYs. A rule or administrative change must alter a project’s feasibility, scale or timing; financing and construction must follow; homes must be occupied; a subset of people must experience a net change in a health-relevant housing condition. Other renters may benefit through supply, vacancies and market adjustment. Some tenants or businesses may incur disruption. These are real mechanisms worth examining, but they need a common counterfactual and cannot all receive the full project effect independently.

The SF Planning implementation page states the Family Zoning Plan was signed December 12, 2025 and took effect January 12, 2026. It also identifies state-law obligations and the parallel tenant-protection ordinance. The present donor therefore cannot be modeled as purchasing adoption from a world where none of this occurred. A prospective contribution might improve implementation, remove a residual bottleneck or accelerate a later project, but that is a narrower claim requiring new evidence.

Transportation

Taking Muni’s Vitals, dated March 26, 2026, compares agency performance and identifies possible efficiency work amid operating deficits. This is an identifiable research product, not a trial of what SPUR funding does to bus service. A comparison with peer agencies can expose useful questions without proving that any recommended change will preserve the same service at lower cost.

The health model considers additional useful transit service. A useful hour is more demanding than an hour scheduled: the service must run where and when it changes a relevant journey, and the person must not have made the same journey by an equally beneficial alternative. Health may arise through access to care or necessities, less harmful travel, or additional physical activity. These effects can coexist, but a single aggregate utility prior must not later be supplemented with the same physical-activity benefit again.

The Metropolitan Transportation Commission’s January 30, 2026 agreement identifies a $590 million bridge loan, negotiated with the state and operators, for the FY2026–27 operating gap. This is a major funded baseline. The agreement is evidence of public resources and multiple decision-makers, not evidence that a current SPUR gift causes the full loan or all riders’ benefits. Loan principal is also a financing flow, not automatically the economic resource cost of all transit service.

Clean heat and environment

SPUR’s August 6, 2026 heat-pump update discusses installation barriers. The earlier Greenlighting Clean Heat brief offers permitting reforms. These sources support a current implementation subject, not an observed number of additional conversions attributable to a donor.

There are at least three different outputs that should not be confused. First, a permit process can become less costly. Second, an installation that would have happened without a permit can become recorded or compliant. Third, a household can choose electric replacement earlier than it otherwise would. Only the third necessarily advances the combustion-exposure health pathway used here. The first may have welfare benefits without the same health effect; the second may improve safety without increasing electrification. A convincing monitoring system would distinguish all three.

Governance, procurement and economic security

Purchasing Power, April 17, 2025, identifies procurement delays and recommends administrative changes. A faster contract for an effective service could create additional person-time of access. Faster procurement for an ineffective service need not improve health. Reduced scrutiny can also create quality or accountability risks. The appropriate unit is a specific earlier effective service, not a percentage of the entire city budget multiplied by an unrelated charity’s QALY estimate.

Economic and downtown work can affect jobs, public space, tax revenue and service capacity. The connections are potentially broad but contain substitution and distributional effects. More revenue in one district can be partly displaced from another; higher land values may benefit some people while worsening affordability for others. Income and GDP are not QALYs. A future report may value those outcomes separately, but this health assessment does not hide a monetary-to-utility conversion inside a broad positive coefficient.

Resilience and other unquantified work

Shared Risk, Shared Resilience, January 7, 2026, concerns cross-jurisdictional wildfire governance. The delivery chain would be better coordination → actual additional mitigation or response → changed fire/smoke exposure → finite health outcomes. A governance recommendation is not acreage treated, an insured property is not a person protected, and property loss avoided is not a clinical utility measure.

Earthquake resilience, flooding, environmental justice, education and institutional knowledge can matter even where no number is assigned. Each should be revisited if a concrete implementation cohort and outcome bridge become available. For now the portfolio assessment is broader than the quantified denominator. This is a substantive limit on interpretation, not a footnote that can be removed when displaying a favorable ratio.

Spending breakdown

Three latest posted tax years

These are the latest three Form 990 years linked from the official disclosure page at review, not a claim that an unpublished newer filing cannot exist. Amounts below use the functional-expense columns of each original return; these should not be silently replaced with subsequently restated comparatives.

SPUR — functional expense shares

Fiscal years ended March 31. Denominator is original Form 990 functional expense, NOT audited/gross addback totals.

Shares are calculated as category expense divided by the stated annual denominator. Rounding may make displayed shares differ slightly from 100%.

The latest return adds $30,000 to audited expense in Schedule D; Part IX places $30,000 in investment management. The previous return explicitly describes its analogous reconciliation as investment-management fees. FY2024–25 additionally reports $236,161 event costs outside functional expenses. Thus the gross accounting benchmark is $6,313,985, rather than either $6,047,824 audited functional expense or $6,077,824 tax functional expense alone. This is a resource-boundary clarification, not a reason to multiply the marginal-gift price by annual spending.

The original 2022 filing itself contains a reconciliation discrepancy: Schedule D’s $8,438,592 differs from Part IX’s $8,437,499 by $1,093. Preserve both in the financial ledger and use the explicitly identified Part IX value in this table. The older gross-cost series is not fully reconciled here. The 2023 filing reports $185,934 netted event costs; those are additional to its functional expenses, not an extra program-health multiplier.

Audited versus current unaudited information

The FY2024–25 audit, issued February 4, 2026, reports $6,047,824 functional expenses. It includes depreciation and in-kind accounting. Its liquidity note reports $1,756,502 available for general expenditures within one year; designated reserves and property are not all current spendable cash. The prior-year restatement reduces net assets by $550,081. The Food & Agriculture sponsorship transferred, so historical sponsored activity cannot be treated as current delivery.

The FY2023–24 audit originally reports $8,744,075 expenses; the newer audit’s comparative is $8,807,575. The $63,500 difference has not been fully reconciled in this review. Do not build a clean like-for-like trend from these inconsistent vintages or attribute the decline wholly to efficiency. Original and restated statements answer different questions.

The current unaudited annual summary reports $6,916,547 expenses: policy/programs/publications $4,910,748; development $1,008,507; administration $997,293. Components exceed its stated total by $1. This minor discrepancy is preserved. Revenue is $6,424,324 and the stated deficit $492,223. Neither a deficit nor the reported net-asset balance independently identifies the next funded task. Source: FY2025–26 annual report.

What a spending breakdown cannot tell us

The tax returns group policy areas into a program total. They do not provide audited separate expenditures for housing, transportation, clean heat, governance and economic security. The model’s 35%/30%/20%/10%/5% allocation is therefore an analyst picture of an additional gift, not a transcription of the return. Historical functional costs show the scale of the institution and necessary support functions. They do not identify the marginal production function.

It would be tempting to divide annual expense by the $100,000 hypothetical gift and multiply benefits by the same factor, presenting an annual organizational total. That would be algebra, not evidence. Policy work has fixed costs, shared knowledge, discrete opportunities and capacity limits. The present model is a local marginal-gift illustration, with whole-organization allocation, not a claim that every $100,000 block has identical impact or that a gift equal to the annual budget can be deployed linearly.

Similarly, gross accounting cost is not complete societal cost. Public staff, private capital, contractors, agency operations, households’ time and equipment can be necessary complements. Some would have been incurred under the comparator, so charging all gross capital as incremental would also be wrong. The correct quantity is the change in real resources under the same outcome counterfactual, with financing and transfers kept distinct. This review supplies a resource stress but not a measured comprehensive resource ratio.

Monitoring and information sharing

Research products are verifiable outputs

Named reports with dates, authors, methods and recommendations are observable. They provide more evidence than a generic mission statement: a reader can inspect whether the proposed mechanism is coherent and whether public agencies considered similar actions. That is an important organizational strength. It still does not establish whether the research was decisive, whether implementation succeeded, or whether a marginal gift expands the work.

For a direct-service provider, useful monitoring might follow referral, treatment and retention. For SPUR, the analogous ledger would follow a recommendation, a specific administrative decision, implemented physical output and resident exposure. Each transition should include dates, responsible actors, counterfactual work and negative cases. A list containing only adopted recommendations would be selected on success and could overstate the average contribution of the full portfolio.

What a housing ledger should contain

A useful project or reform entry would identify the legal baseline, the residual bottleneck, expected construction absent further intervention, net units after losses and when occupancy begins. It should distinguish additional production from acceleration. If a home would be occupied two years later anyway, the health difference stops when the comparator catches up; the donor does not receive the entire building lifetime.

The direct health denominator is not simply new tenants. It is person-time in an improved harmful state relative to the alternative. An occupant moving from an equally healthy Bay home might increase consumption or convenience without the direct health improvement modeled here. The person leaving a vacated unit may enable another move, but vacancy-chain effects and citywide rent effects cannot both be counted in full without a shared market ledger.

The nonoccupant spillover denominator needs equal care. The model currently uses a small number of nonoccupant adult-years per home-year, not all Bay renters. A broader citywide spillover is possible in principle. Expanding that denominator requires specifying the rent or availability change, who actually receives it, whether leases reset, and the share experiencing a health-relevant rather than purely financial change. That is the central distinction between acknowledging citywide leverage and inventing citywide causality.

What a transit ledger should contain

Transit monitoring should identify a service change relative to a published funded plan, actual hours operated, marginal trips rather than all boardings, and recurring rider equivalents. A person can board many times in a year; counting each boarding as a distinct health recipient would greatly inflate benefits. Conversely, one person can obtain distinct access and activity benefits, but the aggregate utility must make clear which are included.

A service preserved in one corridor can be offset by cuts elsewhere or by fare changes that alter access. Peer-agency comparisons do not resolve these distributional effects. The model’s useful-hour quantity is therefore an outcome-equivalent prior, not a claim that a publicly reported scheduled hour is uniformly useful. Publicly funded service in the baseline receives no new donor credit merely because SPUR supported the funding.

What a clean-heat ledger should contain

The key comparison is a time path of appliance use and emissions with and without additional work. Record the replaced equipment, expected failure/replacement date, electric versus combustion choice, installations completed, and any interruption or safety outcome. Permit-processing time is useful process data but is not a sufficient endpoint. Rebate applications are likewise not completed conversion or reduced exposure.

The current Air District rule-development page is the relevant baseline record, including implementation work and proposed amendments. Draft flexibility is not automatically an adopted exemption; a tentative future decision is not an implemented result. A model based on an obsolete schedule could overstate or understate the acceleration window. This V2 does not infer a new exact regulatory timetable from old SPUR text.

A more concrete public baseline is now available. The March 18, 2026 C/CAG packet, item 5, identifies a BayREN Codes and Standards-funded San Mateo County permit-simplification pilot, scheduled to launch April 1 with seven jurisdictions. It does not report completed installations or a SPUR grant requirement. This supports the reality of implementation work while weakening any assumption that a new donor must fund the pilot from scratch. A later evaluation could distinguish incremental adoption from improved documentation; the March packet itself does not.

Missingness and reporting bias

No inspected source supplies a public organization-wide dataset linking marginal donor dollars to these completed outcomes. We did not conduct staff interviews, inspect internal time records, audit monitoring systems or independently interview beneficiaries. Lack of public data is not proof SPUR lacks internal data. It is a limit on what this report can claim.

Self-reported contributions are most useful as leads for verification. They should be checked against agency records and the accounts of other participants, especially where the event had substantial public funding or a statutory deadline. Successful examples may reveal institutional competence without identifying the average effect of all attempted work. The model must pay for failed and inconclusive efforts too.

For procurement, an October 7, 2025 Board President release independently acknowledges SPUR's report in describing Board passage of contracting reform, then pending mayoral signature. It also traces the process to a 2022 request and 2024 City Administrator response, before the April 2025 SPUR publication. That chronology supports contribution rather than sole initiation. The release reports a legislative step, not measured faster contracts, better services or additional health. An ordinary gift now would need a residual implementation role beyond that baseline.

Clinical evidence and policy transfer

Housing production: stronger scale evidence, weaker donor attribution

The Controller’s October 29, 2025 Family Zoning economic analysis, pp. 18–21, forecasts 8,504–14,646 additional produced units over 2026–45, with 463–1,031 additional losses: 8,041–13,615 net units. Modeled citywide price reductions reach 2.5%–4.2%. These are conditional forecasts, not observed outcomes, and the analysis predates some amendments. Its model uses historical parcels and market conditions, not randomized adoption.

This provides a useful scale discipline. Legal housing capacity should not be treated as completed housing, and gross production should not omit lost units. It does not supply SPUR’s contribution or the effect of a gift today. The current comparator already includes adoption. A future implementation model must identify what remains changeable, rather than taking a fraction of the entire pre-adoption forecast and calling it an observed marginal return. The forecast’s twenty-year horizon cannot be inserted into a shorter health model without a completion and comparator-catch-up schedule.

Housing supply and affordability

Asquith, Mast and Reed’s primary working-paper page describes evidence that large new apartment buildings slow nearby rent growth. This supports a supply mechanism against the claim that every new building necessarily raises nearby rents. It does not establish a SPUR effect, an identical effect for small infill, or a clinical utility gain per home.

There are several transfer problems. Locations receiving new construction may differ from a future Bay project. A local effect near a building is not automatically a citywide average. A percentage rent change can matter differently for a rent-burdened household, a protected tenant, a homeowner or a high-income newcomer. The study’s outcome is not the health state used by our calculator. These limitations do not eliminate the mechanism; they prevent a housing result from silently becoming a QALY coefficient.

Housing and health: retain the negative evidence

Denary and colleagues studied a cohort of 400 low-income adults in New Haven. The primary abstract, independently retrieved through Europe PMC after the full-page browser challenge, reports lower distress among assistance recipients in the cross-sectional comparison but statistically nonsignificant decreases following within-person entry. This is not proof of no benefit. It does weaken a simple assumption that reduced housing costs reliably produce a large, measured health gain.

Rental assistance is not new market housing; the magnitude and recipients of relief differ. The relevant lesson is methodological: selection can make cross-sectional associations look more favorable, and a mental-health scale is not a preference-based utility. The existing model’s direct utility and spillover utility remain judgments. We do not claim that this study supplies .01 or .001, or that it justifies setting all housing health to zero.

Other evidence retained includes a benefit-cut natural experiment and the Moving to Opportunity randomized study. They concern different doses, populations and mechanisms. The V2 packet preserves their prior ledger rather than pretending to have newly re-audited every paper. None is used as a direct calibration of an ordinary SPUR gift. Symmetry also matters: worsening an already low-income household’s benefits does not automatically identify the equal-and-opposite effect of modest market rent relief.

A fresh abstract check of Bentley et al. identifies 10,047 Australian adults and a fixed-effects comparison of transitions into unaffordable housing. The lower-income group’s SF-36 mental-component change was −1.19 points (95% CI −1.97 to −.41); the higher-income group did not show the same association. This supports heterogeneity rather than a universal occupant benefit. SF-36 mental-component points are not SF-6D utilities. No transformation from that score to our .01 or .001 coefficients was established, and none is applied.

Transit health

The inherited transit bridge combines useful service, marginal trips, recurrent-rider equivalents, a health-relevant fraction and a net QALY per changed rider-year. This is deliberately not a count of all riders. Regional active-transport health modeling can illustrate mechanisms, but a scenario that shifts population travel patterns is not the effect of another hour of Muni service. The full health package includes injury and pollution tradeoffs, not only exercise.

The current V2 does not find a study matching SPUR-funded research to changed service and then to a measured utility outcome. It therefore retains the old parameters, identifies the required monitoring and exposes the pathway separately. A new empirical utility number would be less defensible than keeping a clearly marked prior until actual travel behavior is known.

The original Maizlish et al. paper, newly read through its methods and limitations, is particularly informative about transfer. Its ambitious regional active-travel scenario increases median walking/cycling from 4 to 22 minutes daily. It models substantial activity benefits alongside 5,907 additional injury DALYs, rather than assuming exercise is harmless. The authors explicitly exclude changes in public-transit use and apply a future steady-state accounting year. Therefore it is not an empirical coefficient for our incremental Muni service hours, and its DALYs should not be silently relabeled as QALYs. It supports a mechanism and an injury boundary, not the local donation effect.

The 2025 countrywide relocation study reports within-person activity changes across 7,447 relocations; moving between cities at the 25th and 75th walkability percentiles was associated with about 1,100 additional daily steps, sustained over three months. The publisher-indexed primary abstract was accessible, while the full-page open encountered an authentication redirect. The result is a useful counterpoint to purely cross-sectional selection concerns, but it does not establish years of sustained utility after a Bay policy change. Relocation is not randomized, walkability change is not a transit funding increment, and any later activity credit would need to be reconciled with housing-health assumptions.

Pollution and finite mortality health

The inherited Air District appliance assessment models 37–85 annual premature deaths avoided under the full regional emissions change. This is a population atmospheric/health model, not a randomized heat-pump trial. The former source audit visually inspected the table; this refresh preserves that provenance rather than claiming a fresh visual inspection.

The calculator uses only a small exposure fraction of that full modeled change, then explicit implementation, gift contribution, effective years and finite QALYs per mortality case. A premature mortality case is not a full healthy lifetime. The 3/8/12 QALYs-per-case scenarios are assumptions about remaining health, not source-provided life expectancy. Likewise, effective mortality years are not appliance lifetime: they represent exposure ramp, lag, discounting and comparator catch-up. Indoor health, climate and other morbidity are not added on top without an overlap model.

Revision record and acceptance status

V2 retains the accepted v3 health coefficients and all ten scenario outputs. New work adds three-filing financial comparison, gross-cost reconciliation, current public-baseline checks, an official housing scale analysis, monitoring requirements and validation of finite schedules and price outputs. It does not claim new measured SPUR health outcomes, a new restricted offer or a complete portfolio expected return.

The result remains approximately $2.19M/10 Bay QALYs in the central ordinary-gift illustration. It is not the earlier clean-heat-only SF estimate and not a full annual-expenditure ratio. The executive summary and full narrative must travel with that scope when reused.

This assessment relies on public sources and does not include organizational interviews or private monitoring records. Its conclusions remain limited by the absence of measured SPUR-specific health outcomes and a verified additional funding opportunity.

Qualitative assessment

Transparency and correction

Publishing multiple financial years, named research and underlying policy arguments allows meaningful scrutiny. The restatement also matters: a correction is evidence that a discrepancy was recognized, while the discrepancy limits unqualified trend claims. An unmodified audit opinion is not an audit of policy effectiveness or proof that all public-facing totals are internally consistent.

This review found both minor and material accounting-boundary issues in the older assessment. The $1 current-summary difference is minor. The netted-event boundary and original-versus-restated historical comparisons matter more for financial interpretation. None directly measures the health return of a new gift. We have not observed how SPUR would respond to our questions because no outreach occurred; responsiveness cannot be inferred from the existence of a contact page.

Why the optimistic case is still plausible

Policy implementation can change large populations’ exposure. A useful administrative improvement may require relatively little nonprofit expenditure compared with the resources it unlocks. Housing and transport also affect nonparticipants through markets and networks. These are reasons not to restrict the analysis to workshop attendees or newsletter subscribers.

The strongest positive case is therefore a small but genuinely pivotal implementation contribution producing a durable, bounded difference beyond the funded baseline. It does not require SPUR to be the sole cause. It does require more than a retrospective statement that SPUR participated. Such an opportunity might be found in a project-specific implementation record, but it is not established by the current public evidence.

What most weakens the case

The most serious problem is not that some benefits are difficult to monetize. It is that the marginal physical-output and health assumptions are not jointly anchored. The citywide housing source improves scale understanding but describes a policy already adopted; the health study concerns a different housing intervention; the gift share remains judgment. Multiplying those individually plausible pieces can create a precise-looking number without making the joint claim reliable.

The favorable scenario also bundles optimism about several links. Even if one optimistic assumption is reasonable in isolation, their conjunction may be uncommon. We decline to invent a precise probability of that conjunction. The report consequently cannot claim a calibrated probability of outperforming a threshold, nor can it infer expected value by averaging the endpoints.

Finally, the organization’s wider benefits remain incompletely quantified. This limits a negative conclusion as well as a positive one. The correct finding is that this health model does not establish a compelling ordinary-gift opportunity, not that all unmodeled activity is worthless. Further evidence could raise or lower the assessment; it should not be collected only to rescue a favorable rank.

What do you get for your dollar?

What is retained and why

The central and named joint scenarios are inherited unchanged from the accepted unrestricted portfolio v3. The new evidence improves accounting and interpretation but does not identify replacement values for marginal output, utility or causal contribution. An unchanged headline is an honest outcome of additional research. We do not move coefficients merely because a deeper report is expected to produce a new number.

The portable V2 model validates finite inputs, calendar schedules, geographic sums and ratios. It preserves all inherited saved outputs exactly. It additionally reports a small-gift linear illustration, with an explicit warning that this is not proof of scalable capacity. It does not infer a probability distribution from the labels central, favorable and pessimistic. Weighted expectation and favorable-tail share of expectation are therefore unavailable, not zero.

One cost allocation

The illustrative $100,000 gift allocates $35,000 to housing/planning, $30,000 to transportation, $20,000 to heat/environment, $10,000 to governance/procurement and $5,000 to economic security. Those allocations are already embedded in the assumed physical outputs. They should not be multiplied into the health calculation a second time. Nor should a separate administration penalty be applied after the full gift has been charged.

This differs from pricing only a restricted program and presenting that price as ordinary giving. Every dollar of the gift is included, including the allocations with unquantified health. It also differs from a measured annual whole-expense return. The yearly financial record informs plausibility and resource coverage but does not determine the number of additional homes, service hours or exposure change caused by that gift.

Housing equation and worked example

For each geographic cell:

Housing Q = integrated occupied-home-years × net gift contribution × [(people per direct home-year × improved-state fraction × conditional direct utility) + (nonoccupant household-years per home-year × adults per household × average spillover utility)].

The central Bay exposure is 4,000 integrated home-years, split 1,000 SF and 3,000 rest of Bay. The net contribution is .02. Direct parameters are 2 people, .15 improved-state fraction and .01 conditional utility. Spillovers are .5 household-years, 1.5 adults and .001 average utility. Thus 4000 × .02 × [(2×.15×.01)+(.5×1.5×.001)] = .3000 QALY.

The .001 spillover utility is already an unconditional average across counted nonoccupant adult-years. Adding a new responder probability would change its definition, not simply make it more cautious. The direct .01 is conditional and therefore uses the separate .15 affected fraction. That distinction prevents a superficially symmetric but erroneous double discount.

The annual exposure schedule covers years 3–8 after the gift at 3% calendar discounting. The integrated input is used once. It must not be multiplied by six again. Beneficiary geography is a prior, not a deduction from where a building stands. Migration and regional vacancy chains can separate project location from where health changes accrue.

Transit equation and worked example

Transit Q = integrated useful service-hours × net contribution × marginal trips/hour ÷ trips/recurrent-rider-year × health-relevant fraction × net QALY/changed rider-year.

Central Bay service exposure is 20,000 hours. With .02 contribution, 20 marginal trips/hour, 400 trips/recurrent rider-year, .10 health relevance and .005 QALY/changed rider-year, the result is .0100 QALY. The finite schedule is years 1–5, discounted once. The conversion to recurrent-rider equivalents is an accounting device, not evidence that those trips involve a verified number of unique individuals.

The .02 contribution already covers implementation, replacement funding and collaborators. Adding another general success factor would count some uncertainty twice. It would be legitimate to replace this bundled judgment with independently defined conditional terms if evidence supports them, but not to append overlapping discounts while keeping the original definition.

Heat equation and worked example

Heat Bay Q = regional exposure fraction × implementation probability × gift attribution × effective mortality years × regional annual mortality change × finite QALYs/mortality case.

Central: .002 × .5 × .1 × 3 × 61 × 8 = .1464 QALY. The .002 is a conditional equivalent exposure fraction associated with the smaller portfolio allocation, not the .01 in an older clean-heat-only hypothetical grant. Using the old narrow value in the whole portfolio would silently change scope. The SF share is .10, leaving .13176 QALY in the rest of the Bay. SF and rest-of-Bay are mutually exclusive accounting cells.

Results and sensitivities

The favorable outcome combines favorable values rather than varying one input. It is a possibility test, not a confidence interval or evidence that a $19,365 purchase exists. Removing the favorable scenario does not change the central result, because no mixture weights exist. The central housing pathway contributes .30/.4564, about 66%; heat about 32%; transit about 2%. Those are contributions to modeled central health, not shares of actual organizational impact.

The historical spillover sensitivity reproduces an earlier .005 utility average. It is not another current headline. The zero-spillover comparison shows that the central result remains unfavorable even if that contested channel is removed; it does not prove spillovers are absent. Keeping both comparisons helps a reader distinguish model fragility from a claim about the world.

Signed harms, overlap and uncertainty

Negative health can arise from disruption, unsafe implementation, loss of access or displacement of effective activity. Shared intervention harms belong in the net pathway parameters. The outside donor-specific harm term is reserved for an incremental adverse effect that can persist when the positive contribution is zero. It is not a second subtraction of all construction harms and not a monetary opportunity-cost penalty.

The current harm coordinates are hypothetical, not allegations of observed SPUR harm. Central zero donor harm means unquantified, not proven absent. A null output does not entitle the model to erase an independent adverse pathway. Ratios are reported only for positive net health; a negative denominator is not a cheap intervention.

Housing and transportation can benefit the same person. That alone is not duplicate health, but counting the same physical activity or mental-health improvement in both pathways would be. The current aggregate utility priors require the health states to be distinct and net. A future person-level implementation ledger should replace this abstract safeguard with explicit accounting. Governance that enables those same outcomes should receive no additional full benefit without separating an independent effect.

Resource and funding sensitivity

Adding a hypothetical $1 million of net complementary resources once to the $100,000 gift raises the central ratio to approximately $24.10 million per 10 Bay QALYs. This is not an estimate of actual construction or transit cost. It illustrates why a donor-only favorable tail cannot establish complete resource cost-effectiveness.

For a small ordinary gift, the code linearly scales the reference health result only as an illustration. It makes no claim of an available marginal tranche. Cash could support ongoing staff, replenish reserves, replace other gifts, or permit new work. Those possibilities affect the contribution terms and must be defined once. General fundraising, a deficit, or a famous policy success cannot independently identify the marginal response.

Model input appendix: inherited coordinates, not observations

All entries below remain analyst assumptions except the regional mortality source range. The exposure schedules are stored in the companion JSON and checked by the executable model. These are joint coordinates, not independent probability distributions.

The names retain the executable identifiers so each value can be checked without guessing an abbreviation. conditionalSfOccupiedHomeYears and conditionalRestBayOccupiedHomeYears are already discounted exposure integrals. netGiftContribution includes funding replacement, implementation and collaborator attribution for housing/transit. Heat separates conditional exposure, implementation and gift attribution. donorSpecificHarm is additional to already-net shared intervention effects and is nonzero only in separately named harm stresses. No clinical coefficient is sourced merely because an input has a decimal value.

Funding and previous grants

What can a donor reasonably do today?

SPUR has an official general donation route. A donor can support the institution because they value its work and accept uncertainty about the resulting portfolio. This report does not verify a heat-pump restriction, a particular SF project, or a grant agreement purchasing technical services. It should not be phrased as though the donor selects a measured number of health outcomes at checkout.

For a donor whose primary goal is near-term measurable Bay health, the present evidence supports further diligence rather than a leading recommendation. A useful request would concern the next realistic unrestricted workplan and what changes if funding is lower or higher—not a promise to achieve our selected QALY threshold. The donor should also ask which work is already underwritten and how a new grant changes timing or scale, rather than merely renaming existing expenditure.

Which information would change the decision most?

First, identify one or more actual prospective implementation cohorts across the portfolio. For each, record the remaining public baseline, dates, attempted and failed work, physical outcome and funding response. Second, distinguish institutional contribution from the additional gift’s share: work accumulated over years is not purchased anew. Third, match the health bridge to the actual residents and exposure dose.

A verified offer without a credible outcome mechanism would not be sufficient. A credible mechanism with no priced offer can still justify exploratory research. The decision improves most when the funding and causal ledgers meet: an actual additional action, a reason it changes implementation, and a bounded health consequence. No single generic percentage can replace all three.

Alternatives and portfolio concentration

Other policy organizations may work on housing or transit, and public agencies already have responsibilities and budgets. A comparison should examine which bottleneck each organization is positioned to change rather than assigning every coalition member the same project total. The additional donor can be important even if many partners exist, but evidence must identify a missing contribution or plausible incremental acceleration.

Direct-service health organizations offer a different evidence profile, often with a clearer delivery denominator but less citywide leverage. This does not make them automatically better or worse. It means a donor should not compare a measured completed-service cohort and SPUR’s favorable joint policy scenario as if they carried equal evidential weight. The central price is useful as a transparent opinion under assumptions, not as a precisely comparable market price.