What if the Islands Trust had never existed?

Read this before the numbers

We want the Islands Trust abolished, so you should assume we will publish whatever helps us. Each of the five studies below states in advance the result that would count against it. 5 findings have already gone the wrong way for this campaign and are listed by date below, 2 of them squarely against our case.

And the bias in this comparison runs toward us, not away. The Trust Area was drawn in 1974 around islands already under development pressure, so the obvious comparison islands face less of it. Finding less development inside the Trust Area would therefore prove nothing. That is the whole methodological problem, and everything below is built around it.

The question

What difference does Islands Trust land-use governance actually make, compared with the ordinary regional-district governance every comparable BC coastal community has?

Nobody has ever measured it. Searched through OpenAlex full text and Crossref: no matching study, no difference-in-differences, no synthetic control, no before-and-after with a comparison region. A boundary-discontinuity evaluation of a supra-municipal land-use commission of this type — the Adirondack Park Agency, the Pinelands Commission, the Cape Cod Commission, the Tahoe Regional Planning Agency — does not appear to exist anywhere.

The two attempts that exist, and how each failed

• Carbon and Biodiversity Mapping and Assessment for the Islands Trust Area — Schuster, UBC (Arcese lab), 23 March 2014 — commissioned and paid for by the Islands Trust Conservancy.

  39% higher bird diversity and 82% more standing carbon inside the Trust Area.

  It makes no causal claim, has no time dimension and no controls, and its “non-Trust” comparison area is dominated by Greater Victoria, Nanaimo, Duncan and Parksville. Comparing ferry-only islands to a metropolitan corridor and crediting the zoning is not an evaluation.

  Source

• Audit of Selected Polygons of the Sensitive Ecosystems Inventory of East Vancouver Island and Gulf Islands, 1999–2001 — Caskey & Henigman, BC Ministry of Water, Land and Air Protection, 2002.

  The Gulf Islands were a labelled stratum against four regional districts.

  The Province declined to draw the comparison, at p.12: “No attempt was made to ensure random sampling or uniformity of sample sizes. Hence comparisons between geographic sub-units… are not statistically valid.” So the Province had the data and would not compare; the only study that did compare was paid for by the body being evaluated.

  Source

The objection, published first

The Islands Trust does not include Quadra and Cortes, which are more distant from Vancouver, and less directly threatened by development.

It cuts against us and we publish it first. If the controls faced less development pressure, then finding less development inside the Trust Area proves nothing — the islands may simply have been quieter to begin with. A raw Trust-versus-Quadra comparison is biased toward the conclusion this campaign wants, and that is precisely what makes it worthless as evidence.

So development pressure has to be measured, not assumed away. The best measure available consistently across every island is the share of homes with nobody living in them.

Method. Share of private dwellings not occupied by usual residents, 2021 Census, the same Statistics Canada definition for every unit. Used here as the best amenity-pressure proxy available consistently across the whole set. Source. Statistics Canada, 2021 Census of Population (measured 2026-08-22).

The controls are not uniformly low-pressure, which is the specific thing the objection would imply. Cortes sits above Denman, Gabriola and Salt Spring; Quadra sits at the bottom of the entire set. Cortes is therefore our best-matched control and Quadra the weakest, and treating the three as one non-Trust bloc would repeat exactly the error that sank the 2014 study. We match island to island and never pool.

And it is not the ferry

Method. Read from BC Ferries’ own published seasonal timetables, weekday sailings each way. Hornby and Cortes each require a second crossing and cannot be compared on their own second leg alone. Lasqueti has no vehicle ferry; Gambier and Keats have no BC Ferries service at all and are not scored. Source. BC Ferries published schedules — seasonal, and to be re-captured before republishing (measured 2026-08-22).

Quadra — a control island — has the best ferry service in the entire set: ten minutes, 28 weekday sailings, better than any Islands Trust island including Denman. And it has the lowest seasonal-dwelling share in the set. So “the controls are simply harder to get to” is false as a general claim. What Quadra is far from is Vancouver, not a ferry. Metro proximity and ferry access are different variables and must enter the matching separately rather than being collapsed into one remoteness score.

Study A — one electoral area, two land-use authorities

Comox Valley Regional District Electoral Area A — officially “Baynes Sound – Denman/Hornby Islands” — contains both the two Trust-planned islands and the CVRD-planned Baynes Sound mainland. One electoral area, one director, one set of regional services, one labour market, one assessment authority, two different land-use authorities. Both CVRD land-use bylaws carve the islands out in their own operative text, and the Islands Trust Act s.36(2) expressly contemplates partly-Trust electoral areas: the design is a structure the statute anticipates, not a quirk we found.

Almost every confounder we cannot measure is held constant by construction. The treatment changes at a jurisdictional line while geography, economy and services vary continuously across it.

Method. 2021 Census of Population. The two islands are read at designated-place level, the mainland as the eight dissemination areas that make up the remainder of the electoral area. Derived two ways — direct sum, and census subdivision minus the two designated places — which reconcile exactly. “Not occupied by usual residents” is Statistics Canada’s own standard definition, not one we invented. Source. Statistics Canada, 2021 Census of Population (measured 2026-08-22).

Across one regional-district boundary, inside a single electoral area with one director and one set of regional services, the share of dwellings with nobody living in them is 6.7% on the mainland against 21.1% and 41.5% on the two Trust-planned islands. Core housing need runs the same way, and so does the share of residents over 65.

Read that carefully before using it. It is a difference across the boundary, not yet an effect of the boundary. Island status is confounded with Trust status here — the line is also a water line — and the seasonal share is simultaneously an outcome and our pressure covariate. It is the strongest single number this project has produced and it is not a causal claim.

What would count against us. If parcelization, dwelling growth and approval times are indistinguishable across the boundary once island status is accounted for, the Trust has no measurable land-use effect in either direction, and we say so.

The Province already built this comparison — and refused to let the Trust out of it

Every unit in this study has its own legally required housing needs report on the same prescribed indicator list. The Local Government Act defines a regional district’s “applicable area” as its electoral areas other than an area within the trust area, and separately requires a local trust committee to report for its own local trust area. The disaggregation is mandatory at collection, not merely at presentation. Housing Needs Report Regulation, B.C. Reg. 90/2019.

This Housing Needs Report does not include the Denman and Hornby Island Trusts.

Inside one electoral area, the mainland is measured by the regional district and the two islands by their local trust committees — separately, by statute, on the same indicators, against the same regional denominators. Study A does not have to construct its comparison. The Province already built it.

In July 2025 the Islands Trust Council asked the housing minister to be released from the standardised method, saying it “found that the formula was inadequate to capture the true and diverse housing need in rural island communities”. On 2 October 2025 the minister refused:

for planning purposes it’s important to ensure a more robust, consistent, and comparable calculation of the overall amount of housing needed in each community. This is why the legislation requires use of the standardized ‘HNR Method’

The body being evaluated asked to be excused from the standardised comparative metric, and the minister with the authority to grant it said no, because comparison is the purpose.

What each authority does with the sea

Same question, same method, both sides of the comparison: how much water does the body that plans this island put a zone on?

Method. Marine and aquaculture zones are attributed to an island by the authority’s OWN layer — the Trust’s water classes for the Trust islands; the regional districts’ marine, marine-commercial, marine-industrial, marine-reserve, marine-institutional and aquaculture codes for the controls — then restricted to polygons lying within 5 km of that island’s shoreline, and measured in BC Albers. The denominator is the island’s own land area from the provincial Freshwater Atlas island polygons, the same source for all five. The two sides use each authority’s own classification, and those are differently structured: a real limit on precision, though not on the order of magnitude. Source. Islands Trust zoning layer; Strathcona Regional District zoning polygons; qathet Regional District OCP designation layer; WHSE_BASEMAPPING.FWA_ISLANDS_POLY for island land area (measured 2026-08-29).

Texada. Texada is scored on designation, not zoning. qathet publishes no zone polygons, and Texada’s zoning bylaw 99 (1982) is a ten-page PDF that exists in no digital layer. Its OCP designation layer is digitised — 817 polygons for Electoral Area D — and carries no marine designation at all. So the zero is a real measurement on the designation instrument and an absence of evidence on the zoning one.

The regional districts zone sites in the sea. The Islands Trust zones sites in the sea, and then zones the sea.

Median marine zone size is nearly the same on both sides of the comparison — between 1.9 and 6.2 hectares — so both kinds of authority zone specific places in the water: an aquaculture tenure, a marina, a dock. The difference is entirely in the tail. Denman’s single W1 Water Conservation polygon is 13,390 hectares, running to mid-channel, which is 2.6 times the land area of Denman Island itself. The largest marine polygon either regional district applies anywhere near its islands is 538 hectares.

Per hectare of land, Denman carries 54 times the marine zoning that Quadra does.

This cuts both ways and the study takes no position on which reading is right. A body created to preserve and protect a marine archipelago either is doing a job its neighbours leave to the Province and to Ottawa, or is reaching further than its neighbours think necessary.

It is not an argument that the Trust lacks the power. It plainly has it: the two leading British Columbia authorities for a local government zoning the foreshore and the waters over it are both Islands Trust cases, and both were upheld on appeal. This measures how far the power is used, not whether it exists.

The land itself

Parcelization is the outcome the Islands Trust exists to restrain, so it is the first thing anyone would want measured. Here it is, with the honest limit attached.

Method. Every parcel is assigned to an island by majority area against the provincial Freshwater Atlas island polygon — asserted, not inferred from a bounding box or from a legal description. Parcel areas are computed from geometry in BC Albers and clipped to the island, so a foreshore lot contributes only its land. Density uses the island’s own land area as the denominator, never the sum of parcel areas, which double-counts rights-of-way registered over the lots beneath them. Coverage is the de-overlapped union of the parcel fabric against that same island polygon. Source. ParcelMap BC via the BC Data Catalogue, scoped against WHSE_BASEMAPPING.FWA_ISLANDS_POLY (measured 2026-08-29).

Why the density column is not an answer yet. The parcel fabric observes 94.7% and 96.3% of the two Trust islands but only 77.0%, 54.4% and 46.9% of the three controls. The residue is land the fabric genuinely does not carry — unsurveyed Crown land and provincial parks — not a filter we can lift. Missing parcels make a control island look LESS subdivided than it is, and the Trust islands are the two best-observed units in the set, so the gap in the density column is an upper bound on the real gap and it runs in the direction this campaign would like. On top of that, Denman and Hornby are small islands within sight of the Comox Valley, and the controls are large islands two to five hours further from any metro. Read the column as a description of what is there, not as a measurement of what the Trust did.

How current the rules are, and how many bodies write them

Method. Each authority’s own bylaw. Vintage replaced page count as the regulatory-weight indicator after page count failed — see the retired findings below. Source. Islands Trust, Strathcona Regional District and qathet Regional District bylaw registers (measured 2026-08-22).

This is the indicator that replaced a broken one, and it does not run the way a reader might expect. Cortes rewrote its zoning last year; Texada is running a ten-page bylaw from 1982 with one amendment in forty-three years. Regional-district planning is not uniformly newer or older than Trust planning — it is simply different in each district, which is itself the answer to “the Trust brings professional planning where there would be none”.

Improvement districts counted from the CivicInfo BC directory. A control island carries a regional district, a school board and its improvement districts. A Trust island carries all of those, and a local trust committee, and the Trust Council above it. Cortes: none · Quadra: Whiskey Point · Texada: Van Anda and Gillies Bay.

Method. Improvement districts counted from the CivicInfo BC directory. A control island carries a regional district, a school board and its improvement districts. A Trust island carries all of those, and a local trust committee, and the Trust Council above it. Source. CivicInfo BC local government directory (measured 2026-08-20).

What we got wrong

Dated, in full, at the same prominence as everything above. 2 of these make our case weaker and we have left them standing.

• 2026-08-22 · against our case

  Retired: “Official community plan page count is a usable proxy for regulatory weight.”

  It does not distinguish Trust from non-Trust islands, and on the available evidence it points the other way. Quadra — governed by a regional district — has the longest OCP in the measured set at 124 pages, against 82 for Cortes and 82 for Texada. The three Trust page counts are still unread, and we do not get to hold an indicator open until it agrees with us. Page count is dropped, and must not be asserted in either direction.

• 2026-08-28 · against our case

  Retired: “The regional districts do not zone water at all.”

  They do. Ingesting the districts’ own zoning layers found Cortes running Aquaculture One and Two, Marine Commercial and Marine Industrial, and Quadra running three tiers of aquaculture, 39 marine commercial polygons and a marine reserve. The earlier statement had been flagged at the time as an absence in parcel data rather than as a finding, with an instruction to read the bylaws before publishing it. The caveat did its job. Nobody should repeat the earlier sentence.

• 2026-08-28 · our own arithmetic

  Retired: “The comparison parcel sets described the islands they were filed under.”

  They did not. 77% of “Quadra” parcels were Campbell River and 69% of “Texada” parcels were the Powell River mainland, because the ingest identified an island by taking the largest cluster of parcels in a bounding box, and in those boxes the largest cluster is a town. Quadra’s median parcel size moved by a factor of five once the city lots left the sample. Every per-area figure computed before 28 August 2026 is void.

• 2026-08-29 · our own arithmetic

  Retired: “Texada carries 1,410 hectares of marine designation.”

  It carries none. The 1,410-hectare “Marine Seasonal Recreation” polygon is in qathet Electoral Area C, 10.1 km away across Malaspina Strait; Texada is Electoral Area D. The figure had been picked up by a text match on the qathet layer with no distance check. Correcting it widens the gap this study is measuring, which is exactly why it had to be found before publication.

• 2026-08-29 · our own arithmetic

  Retired: “Quadra carries 2,436 hectares of marine zoning against 17,450 hectares of island land.”

  Both halves were wrong, in opposite directions. The numerator included Thurston Bay Marine Provincial Park and other Discovery Islands polygons up to 30 km away, because Strathcona Electoral Area C is a 1,064,539-hectare region and not an island. The denominator is not Quadra’s land area at all — the Province’s own island polygon gives 27,248 hectares. Restricting the zones to within 5 km of the shoreline and taking every denominator from the provincial island polygons gives 1,473 hectares against 27,248, by one method applied to both sides.

What we cannot compute yet

Listed so nobody has to ask, and so a gap in the evidence cannot be mistaken for a finding.

Assessed value, and any tax indicator, for most Trust islands

Assessed values are missing on ten of the twelve Trust islands — Galiano holds 16 of 1,477 parcels, Gabriola 980 of 4,047, and eight islands have none at all. Study B’s value indicators cannot run outside Denman, Hornby and the control group.

Agricultural Land Reserve share, outside four areas

ALR flags exist only for Denman, Hornby, Baynes Sound and the Nanaimo regional district. Everywhere else reads zero, which is a missing ingest and not an absence of ALR land. “Zero ALR parcels” must never be published for those islands.

Per-capita local-government cost for any unincorporated island

Confirmed absent at source, not merely unfound. BC Local Government Statistics carries one row per regional district with no electoral-area breakout, so island-level finances do not exist in it for any Trust island or for Cortes, Quadra or Texada. Bowen is the sole exception, because it is a municipality. A regional-district figure must never be presented as an island’s.

Any island-wide median for Quadra

Quadra has no matching census subdivision or designated place, so it has to be aggregated from four dissemination areas — and medians do not aggregate. Its four DA medians span $61,200 to $64,000 in income, $500,000 to $650,000 in dwelling value, and 46.8 to 61.6 in age. Report per dissemination area, or report nothing.

Texada on any zoning-stringency indicator

Its zoning bylaw is a 1982 PDF that has never been digitised. Texada is a control for land-use governance but not for zoning stringency, and must be labelled that way wherever it appears.

The five studies, and what would sink each one

• Study A — The boundary inside one electoral area (measuring now)

  Comox Valley Regional District Electoral Area A — officially “Baynes Sound – Denman/Hornby Islands” — contains both the two Trust-planned islands and the CVRD-planned Baynes Sound mainland. One electoral area, one director, one set of regional services, one labour market, one assessment authority, two different land-use authorities. Both CVRD land-use bylaws carve the islands out in their own operative text, and the Islands Trust Act s.36(2) expressly contemplates partly-Trust electoral areas: the design is a structure the statute anticipates, not a quirk we found.

  Its weakness. The line is also a water line. Island versus mainland is itself a treatment — ferry dependency, servicing cost, buildable land. Study A therefore cannot stand alone; Study B exists to correct it.

  What would count against us. If parcelization, dwelling growth and approval times are indistinguishable across the boundary once island status is accounted for, the Trust has no measurable land-use effect in either direction, and we say so.

• Study B — Matched island pairs, Trust against non-Trust (measuring now)

  Treated: the twelve Trust islands. Control: Quadra, Cortes and Texada — ferry-dependent BC coastal islands, unincorporated, inside an ordinary regional district, outside Trust land-use authority. Treatment assignment is machine-provable rather than our judgement: by point-in-polygon against the provincial boundary layers, Denman and Hornby return both an electoral area and a local trust area, while Cortes, Quadra and Texada return an electoral area and an empty trust-area result.

  Its weakness. The controls must be matched island by island on development pressure and never pooled. Pooling them is the specific error that sank Schuster (2014).

  What would count against us. If matched Trust islands show the same parcelization and dwelling-growth trajectories as matched controls, the Trust’s land-use regime is not doing what either side claims. A published “we could not construct a valid comparison” is also an acceptable output of this study.

• Study C — Bowen Island, before and after incorporation (designed, not yet run)

  Bowen was inside the Trust Area and incorporated as a municipality in 1999 after a referendum carried by 60%. It remains in the Trust Area and runs its own land use. It is the only unit in BC that has actually done the thing this campaign proposes — an interrupted time series with a real, dated intervention and a known counterfactual set.

  What would count against us. If Bowen’s post-1999 development, tax and housing trajectories are indistinguishable from comparable Trust islands that did not incorporate, then incorporation changes less than we claim, and the honest conclusion is that the case for it rests on democratic representation rather than outcomes. We would publish that.

• Study D — Governance cost, workload and accessibility (measuring now)

  Not about land cover at all: about how the apparatus behaves. Elected officials per 1,000 residents; how many separate governing bodies one community actually has; application processing time; zoning bylaw vintage and amendment count; enforcement backlog; document accessibility.

  What would count against us. If the Trust’s processing times, backlog trend and document accessibility are no worse than the comparison regional districts’, the process critique fails and only the representation argument survives.

• Study E — Small rural municipalities as the fiscal benchmark (designed, not yet run)

  Provincial local-government financial data has no electoral-area breakout at all, so per-capita cost cannot be computed for any unincorporated island, treated or control. It does break out municipalities. Bowen Island, the Village of Gold River and the Village of Tahsis carry their own rows 2005–2024 and bound the plausible range for a hypothetical island municipality using audited figures rather than a consultant’s projection.

  What would count against us. If per-capita cost in these municipalities is materially higher than the current per-capita cost of governing a Trust island, incorporation is more expensive than we imply, and the honest framing becomes “it costs more and buys representation” rather than “it costs less”.

What this study will not do

• Score forest cover against the Trust. Broad logging regulation on private managed forest land is outside its power by statute, so forest cover measures something the Trust is not empowered to control, and it stays out of the causal analysis for that reason.
• Score health outcomes against the Trust. It does not deliver health care, fund it, or site a clinic. Travel time to an emergency department belongs on this page as description, never as an effect.
• Pool Quadra, Cortes and Texada into one non-Trust bloc. That is the error that sank the only prior study, and it would be worse here, because the controls differ from each other more than some of them differ from the Trust islands.
• Present a difference across a boundary as an effect of the boundary. Where a number is not yet causal, the page says so beside the number, and not in a footnote.
• Bury a result that favours the Islands Trust. Adverse findings are published in the same place, at the same prominence, in the same language. That commitment is the only thing that makes the favourable results worth anything.