Who pays, who receives — household incidence from StatCan distributional accounts

2026-07-18 · wave 3 · ESTIMATE lane · answers Floyd's constraint: "the goal shouldn't be that the vast majority pay out more — let the data guide us."

Basis note (w18): figures below are on the wave-3 gross basis. After the w17 property-tax netting, scale magnitudes by ≈0.91 — signs and orderings are unchanged. Canonical current figures: NUMBERS.md.

Headline

Combining StatCan's National Balance Sheet (who holds land, by sector) with StatCan's Distributions of Household Economic Accounts (who holds real estate, by household group), under the national per-resident design at r = 5.5%:

The data

Sector split of land (NBSA Table 36-10-0580-01, Q1 2026, market value):

households $4.669T · corporations $1.450T · governments $0.356T · total $6.493T. Resource wealth excl. land: $1.369T. Settlement pool base: $7.862T.

Real estate by household group (DHEA Table 36-10-0660-01, Q4 2025, Canada):

Group Real estate per household Implied land per household (52.7% share) Net/yr (avg-size household)
All households (17.24M implied) $514k $271k +$10,200
Lowest wealth quintile $59k $31k +$23,400
Second quintile $170k $90k +$20,100
Third quintile $391k $206k +$13,700
Fourth quintile $640k $337k +$6,500
Highest quintile $1,310k $690k −$12,900
Renter households $50k $26k +$23,600
Owner households $725k $382k +$4,100
One-person households (1 share) $283k $149k +$2,200

Workings: equal share $189,574/person (pool $7.862T ÷ 41.47M residents); average household size 2.41 (41.47M ÷ 17.24M implied households, N derived from DHEA value ÷ value-per-household); average household allowance $456k; land per group = real estate per group × 52.7% (the economy-wide NBSA-land ÷ DHEA-real-estate ratio).

Assumptions, stated

  1. Land distributed proportional to real estate within households. Conservative for progressivity: land fractions are highest in expensive metros, whose owners skew wealthy — the true top-quintile land holding is likely above 52.7% of its real estate, making the true result more progressive than this table.
  2. Within-group heterogeneity is real. Averages hide the politically loud cases: an average-size household in a $1.2M Vancouver home with a 70% land share holds ≈ $840k of land and pays ≈ $21k/yr despite possibly-median income. The regional incidence cut (BC Assessment land/improvement splits) remains the open task.
  3. Resource-stock annuitization (5.5% × $1.37T ≈ $75B/yr) is an approximation; actual annual resource rents are the better number (queued).
  4. Corporate charges land on shareholders, including pensions — concentrated at the top of the wealth distribution, which adds progressivity; the evidence that landlords cannot shift land charges onto tenants is documented on the wiki: https://www.progress.org/wiki/landlords-cannot-pass-lvt-to-tenants/
  5. Government-held land ($0.36T) paying into the pool is a design choice (it effectively transfers from general revenue to the dividend); excluding it shrinks the pool ≈ 4.5%.

What this settles and what it doesn't

Floyd's constraint is met by the data under the national per-resident design: the vast majority of households receive; payment concentrates in the top wealth quintile plus entity-held land. Not yet settled: the regional cut (metro homeowners), the per-adult vs per-resident choice (one-person households receive only +$2.2k — the "who counts" question has real stakes), and the price-reflexivity dynamics.

Sources

  1. Statistics Canada, Table 36-10-0580-01 (NBSA, Q1 2026) — sector holdings of land.
  2. Statistics Canada, Table 36-10-0660-01 (DHEA, Q4 2025) — real estate value and value-per-household by wealth quintile, tenure, and household size; full CSV via WDS, retrieved 2026-07-18.
  3. Statistics Canada, quarterly population estimate (41.47M, 2026-01-01).
  4. Wiki evidence pages: land-value-tax-can-be-progressive; landlords-cannot-pass-lvt-to-tenants.

Groundshare — a proposal in open development. Every number traces to a cited public source with its retrieval date; corrections are published, not erased. Rebuilt 2026-07-19 from the repo's research files.