# Denman housing affordability evaluation protocol

## Primary question

Compared with enabling additional dwellings while retaining an environmentally screened subdivision route, does enabling additional dwellings while removing ordinary subdivision permission increase the number of completed, occupied, year-round homes accessible to lower- and middle-income households?

This protocol is recorded before estimating a Denman policy effect. The initial evidence assessment is not a pre-registered randomized experiment. Adoption date and final rules must be verified before any post-policy evaluation.

## Hypotheses and outcomes

H1: Reusing suitable existing building space reduces incremental construction expenditure relative to constructing an equivalent new detached dwelling. Measure total project expenditure and floor area; separate conversion from detached construction.

H2: Additional dwellings have lower total housing payments than otherwise comparable homes on separate titles. Compare rental with rental; evaluate equity ownership and rental-cooperative access separately. Match bedrooms, floor area, condition, location, utility inclusion, lease duration and year of construction for this direct title-cost comparison. Report family discounts separately.

H3: Removing ordinary subdivision permission improves affordable housing delivery beyond the effect of enabling additional dwellings. Compare actual policy packages, not permission counts. An outcome is not attributable to the subdivision restriction merely because it occurs after the combined reform.

H4: Retaining a bounded subdivision route improves first-time ownership access without unacceptable servicing or environmental costs. Record deposit, qualifying income, finance availability, completed transactions, resale restrictions and actual site effects. The existence of a separate title is not sufficient evidence of affordability.

H5: Enforceable affordability mechanisms preserve lower housing costs through turnover more reliably than voluntary initial discounts. Record agreement terms, duration, eligibility, compliance and subsequent occupancy, with no assumption that a covenant alone produces construction.

Primary outcomes: net additional year-round occupied homes; homes affordable at specified gross-income bands; unrelated renter access; first-time equity ownership; rental-cooperative access; shelter cost including utilities and mandatory fees; tenure duration and security. Separately count improved legal status, safety and security for existing occupied dwellings, even where no net home is added. Record member capital, refund and exit terms rather than equating cooperative membership with equity ownership. Secondary outcomes: completion time, financing failures, abandonment, replacement/demolition/displacement, seasonal conversion, land consumed and servicing costs per occupied household.

Pre-specify gross household income bands of $30,000, $45,000, $60,000 and $90,000 in 2026 CAD, with household sizes one, two and four. The corresponding 30% monthly shelter budgets are $750, $1,125, $1,500 and $2,250. These are analytical bands, not verified local quantiles or program eligibility rules. Report outcomes across all bands and household sizes, including adverse distributional effects. Index future comparisons consistently to 2026 dollars with a documented price series.

Report household-size suitability alongside income. A one-bedroom unit does not automatically meet a larger family's need. Report both the conventional 30% shelter-cost threshold and remaining income after shelter; neither is a complete welfare measure.

## Sample and measurement

Construct a census of identifiable Denman subdivision, additional-dwelling and shared-lot projects from 2015 onward, retaining earlier examples as historical cases. Link archived applications and decisions to completion and occupancy evidence where available. Deduplicate repeated reports, application amendments and renamed projects. Record withdrawn and refused projects and eligible owners who did not build, not just successes.

Record each source's document date, event date, retrieval date, version, exact page/section or timestamp, publisher, original URL, archive ID and evidence class. Keep the status chain separate: proposed, applied, permitted, financed, started, completed, occupied, withdrawn. Never infer a later status from an earlier one.

Record each parcel's ALR status, land-use designation, agricultural use, permitted residential uses and any required Agricultural Land Commission approval. Do not assume a housing permission or the removal of a municipal restriction makes an ALR project eligible. Compare feasible alternatives only, with unresolved legal eligibility marked explicitly. Measure farmland removed from production, farm-operating disruption, farm-worker housing access and agricultural continuity alongside general land consumption; parcel fragmentation alone is not a complete farming outcome.

Stratify household access by available liquid savings, land ownership, existing debt, usable collateral, donated equity and family support as well as income. An income-affordable monthly payment may still require inaccessible entry capital. Record whether finance is a preliminary conversation, indicative quote, approved commitment or closed loan; include required deposit, qualifying income, collateral, guarantees, rate, term, insurance, draw conditions and exit/refinance risks. Model unsuccessful finance applications and owners unable to fund construction rather than treating them as lack of housing demand.

For asking rents and sale listings, record listing date, dwelling size, bedroom count, utility inclusion, tenure conditions, seasonal limitations and duplicate advertisements. Asking prices are not completed transactions; old tenancy rents are not new-entry rents. Do not treat missing records as zero activity. Published household-level data should be de-identified; do not publish private residents' names, precise financial histories or occupancy details.

Phase-two requests, if authorized, seek voluntary actual project budgets, signed finance offers, occupancy evidence and anonymized rents. Include tenants and unsuccessful would-be builders, and record nonresponse. Interviews must identify whether a statement is a firsthand outcome, forecast, opinion or recalled event.

## Comparison design

First compare the same feasible parcel, the same total homes, bedroom capacity, tenure, building specification and environmental constraints under two title arrangements. Keep land valuation, subsidies and gifts consistent. This estimates direct title-cost differences only. Then permit realistic uptake, dwelling size, density, family discounts and financing differences in a second model: these may be policy mechanisms and must not be controlled away when evaluating the total package.

Evaluate at least four policy packages: existing permissions; additional dwellings with ordinary subdivision retained; the proposed combined package; and additional dwellings plus a bounded subdivision route tied to suitable sites and explicit affordability/community-benefit mechanisms. Track non-market projects separately within every package.

Where sample size and overlap allow, use matched comparisons with pre-specified controls. A simple raw average cannot identify a title effect. Do not fit a many-variable model to a handful of observations. If cases are too few, publish audited case histories and ranges instead of a causal coefficient.

Any comparison-island design must document pre-policy trends, baseline rules, ferry access, servicing, land constraints, tourism demand, income, financing conditions and concurrent policy changes. Islands Trust reforms on other islands may contaminate controls. A before/after decline in rents, by itself, is not evidence of the bylaw's effect.

Use annual snapshots and rolling multi-year outcomes; reassess at one, three and five years after any adopted change. Report time from permission to completion, and treat unfinished projects as censored rather than failed solely because observation ends. Publish null and adverse results alongside favorable ones.

## Decision rules

Evidence supports an affordability benefit from the added subdivision restriction only if it improves the pre-specified household outcomes against an ADU-enabled comparison while accounting for lost ownership routes, servicing costs, project selection and other interventions. Cheaper ADUs compared with large detached homes do not pass this test. Environmental or cultural objectives can justify a separate policy choice without proving an affordability benefit; report those tradeoffs explicitly.

Evidence favors retaining subdivision only if feasible projects deliver relevant housing or ownership benefits at acceptable environmental and servicing costs. Paper capacity, owner windfalls or an urban analogy do not pass this test.

Revise conclusions if observed financing and year-round occupancy undermine the predicted uptake, if affordability disappears at turnover, or if the preferred option imposes materially higher costs on excluded households. List unresolved evidence explicitly. No fixed sample-size target or significance threshold is justified until an inventory establishes the attainable sample and measurement quality.