| Year | {GOM} $ | donors | Beth Oloth $ | donors | Moral Arc $ | donors |
|---|
THE HEIR One revocation · {N['funders']} funders · one new home
The Channel ended on a question it could not answer from its own five years of data: after CRA revokes a charity, do the charities that funded it move to one particular new organisation more than they move anywhere else? Twenty-five years of CRA releases answer it. When Beth Oloth Charitable Organization was revoked in January 2019, {N['adopters']} of the {N['funders']} registered charities that had funded it began funding one foundation within two years — a foundation that had reported {d(H['dormant_assets'][0])} in assets for years. Scored against every comparable collapse in the country, nothing else comes close. The Act guards a revoked charity's assets and its directors. It has no rule about its donors.
{N['adopters']}/{N['funders']}
Funders who followed to one heir
1st
Of {T50['events']} collapses with 50+ funders · median {pct(T50['median_share'], 0)}
{mm(H['abroad_2019_24'])}
Sent abroad by the heir, 2019–2024
{d(bo_after['dollars'])}
Reported as gifts to Beth Oloth after it stopped being a charity
In 2022 Miles Howe and Paul Sylvestre, writing in the Journal of White Collar and Corporate Crime, tracked three Toronto charities across twenty years of CRA data and called the pattern a “burner charity”: a dormant registration activated, run at scale, revoked, and replaced by the next one, with the same funders moving each time.
CRA's own findings
Beth Oloth was revoked effective 12 January 2019 and Gates of Mercy effective 13 April 2019, each by notice in the Canada Gazette. CRA's case against Beth Oloth, released in a 94-page letter, cited a “lack of direction and control” over money sent to non-qualified donees, support for Israel's armed forces through pre-army academies, about $1.2 million to projects in the occupied territories, and inadequate books and records. Neither charity appealed, the study reports.
What the study could not show
That the funders' movement was unusual. It picked its “anchor foundations” because they gave across all three charities, then cited that consistency as the evidence. It never asked how often funders move together after any collapse. That is the test The Channel listed as its sixth unfinished join.
SOURCES · CANADA GAZETTE PART I, 12 JAN 2019 & 13 APR 2019 · THE CJN, 29 JAN 2019 · HOWE & SYLVESTRE (2022)
The study's numbers are sound. Its headline was not: the “$400 million” it called unlawful includes the third charity, which it also wrote it “cannot conclusively” call a conduit. This page publishes only what the returns show.
What each charity received from other registered charities, year by year, read from the donors' returns — so a charity that stopped filing still shows up. One scale across all three panels. Tap or hover a bar for the year.
The rule was fixed before the named charities were looked up, then run over all {M['generated_from'][-9:]} of CRA's gift files. A collapse is any charity that had been receiving at least $1 million a year from other charities and then lost at least half of it over the next two years. Its funders are the charities that gave it $5,000 or more in the two years before. A funder follows when it starts giving to a recipient it had not given to in the previous four years.
The named case
FIRST OF {T50['events']} · {T['heir_to_runner_ratio']}× THE RUNNER-UP
The control
WHAT A COLLAPSE NORMALLY LOOKS LIKE
The ranking holds at every size cut tested: first of {T['tiers']['30']['events']} collapses with 30+ funders, first of {T['tiers']['100']['events']} with 100+.
Coordination is not the only thing that sends many funders to one place. A shared outside event does it too, and the data shows exactly what that looks like.
The shape
It was the top new recipient for the funders of {CC['events']} unrelated collapses in the same window — a United Way, a university, a food bank network, a children's hospital foundation. Funders everywhere gave to Ukraine relief. The largest single move was {CC['largest_single_move']} funders.
The difference
A common cause pulls a few funders each out of many charities. The Beth Oloth move pulled {N['adopters']} funders out of one. Moral Arc was also the top new recipient for {len(T['heir_elsewhere'])} smaller collapses in the same years, at {min(x['adopters'] for x in T['heir_elsewhere'])} to {max(x['adopters'] for x in T['heir_elsewhere'])} funders each — the shape of a common cause, around one charity.
A PATTERN IS ONLY A FINDING ONCE YOU KNOW WHAT THE BASELINE LOOKS LIKE
The study's first succession is real in the returns. It cannot be put through the same test, for two reasons the page will not paper over.
It is not a collapse. Gates of Mercy kept operating, and kept receiving millions, until 2019; its funders moved while it was still open. And the move runs straight through 2011, the one year CRA's open release has no gift file for, so “new in 2012” cannot be told apart from “started in 2011”. Reported as a description, not ranked.
Every row below is a gift one of the three charities reported giving to another, on its own qualified-donee schedule.
| Year | Giver | Receiver | Amount |
|---|
Since {AL['first_year_asked']}, when the return began asking, {AL['flagged_Y']} of {AL['director_years']} director-years across the three boards are flagged “at arm's length” — including every year one person sat on all three. The exception is {AL['flagged_N'][0]['charity']}'s {AL['flagged_N'][0]['year']} return, which marks all {AL['flagged_N'][0]['count']} of its directors as not. The boards separated before the funders arrived. {F2['overlap']} of Moral Arc's {F2['ma_new_funders_2019_20']} new funders in 2019–2020 had funded Beth Oloth in 2017–2018, and they brought {pct(F2['overlap_dollars'] / F2['ma_new_dollars'], 0)} of its new money. {M['funded_all_three']} charities funded all three registrations.
A revoked charity is no longer a qualified donee. Gifts to it kept appearing on other charities' returns anyway.
Placeable
Fiscal years that began after 12 January 2019: {bo_after['returns']} returns from {bo_after['donors']} charities report {d(bo_after['dollars'])} to Beth Oloth, the last for a year ending {bo_after['last_fpe']}. Two charities account for nearly all of it, year after year. Gates of Mercy: {d(gom_after['dollars'])}, from {'one charity' if gom_after['donors'] == 1 else str(gom_after['donors']) + ' charities'}.
Not placeable
Fiscal years that straddle the revocation: {AF[BO]['straddling']['returns']} returns, {mm(AF[BO]['straddling']['dollars'])} to Beth Oloth; {AF[GOM]['straddling']['returns']} returns, {mm(AF[GOM]['straddling']['dollars'])} to Gates of Mercy. A gift on 2 January and a gift on 2 December are the same row.
THE FORM KNOWS THE RECIPIENT'S NUMBER · IT IS NEVER CHECKED AGAINST THE REGISTER
The donee's registration number sits on the row. The register of revoked charities is public. Matching one against the other at filing time would have flagged every placeable row. The file also carries negative “gifts” to Beth Oloth, which no form should accept.
When a charity is revoked, the Income Tax Act follows three things out of it. Read the provisions in a row and the gap is plain.
A revoked charity owes a revocation tax equal to what it holds, plus what it takes in while winding up, less debts, its own charitable spending, and transfers to an “eligible donee”. An eligible donee must have more than half its board at arm's length from the revoked charity's board (s.188(1.3)). The Act foresaw a related successor — for the property.
At its last return: Beth Oloth {d(LR[BO]['assets'])} in assets ({LR[BO]['fpe']}) · Gates of Mercy {d(LR[GOM]['assets'])} against {d(LR[GOM]['liabilities'])} in liabilities ({LR[GOM]['fpe']})
A director who served while a charity “engaged in conduct that can reasonably be considered to have constituted a serious breach”, and was revoked for it, is an “ineligible individual” for five years. A charity with one on its board, or controlled or managed by one, can be revoked.
Moral Arc's board, from the 2018 release on: no name in common with either revoked charity
A gift between charities whose purpose is “to avoid or unduly delay the expenditure of amounts on charitable activities” is a ground for revocation, for both sides. It is written against parking money, not against moving it.
Aimed at delay, not at destination
No provision addresses where a revoked charity's donors give next. A pass-through charity holds almost nothing, so the asset rule bites on almost nothing. Its directors can step off a board years before the revocation. What it actually has is its funder list, and the funder list walks.
The one asset of a conduit is the one thing the Act does not follow
Moral Arc Foundation is a registered public foundation, listed in CRA's 2024 release. CRA has published no finding against it. These are its own returns since the funders arrived.
| Fiscal year ending 31 May | Revenue (4700) | Spent outside Canada (200) | Management & admin share |
|---|
Nearly twice what Beth Oloth moved in its whole life, in six fiscal years, with management and administration at a fraction of one percent. Whether that spending is under the “direction and control” the Act requires is a question only an audit answers. What the public file shows is that the audience inherited the stage.
The obvious second test runs the logic backwards. It failed, and why it failed is the most likely mistake for anyone repeating this.
“For every charity whose funding surged from nothing to millions, find the one charity whose funder list supplied most of its new funders.”
It ran over {M['mirror_test']['surges']} surges. The median surge scored {pct(M['mirror_test']['median_share'], 0)}: more than half of almost any new charity's funders already gave to some large, popular charity — a university, the Salvation Army, a hospital foundation — so every surge appeared to have a “source”. A test that convicts the median is not a test.
It also could not score the named charities at all: their surges sit next to the missing 2011 file, or follow years of small receipts. Dropped. The collapse test above survives because it starts from one charity's own funders and asks where they went, which a popular bystander cannot fake.
Every field below already exists in a file CRA publishes.
When a revoked charity's funders move, in numbers, to one recipient, review the recipient. The test on this page ran in under three minutes over every gift file CRA has published.
Needs a join CRA can already run — the code is linked below
The qualified-donee schedule carries the recipient's registration number. Refuse a gift row whose number belongs to a revoked charity, and refuse a negative gift outright.
Needs a validation rule at submission
A schedule with no date cannot say whether a gift was made before or after a revocation. One field ends the straddle.
Needs one column on T1236
The 2011 release has identification, financial and director files, and no gift file. It is the year a succession in this record passed through.
Needs a file CRA holds
s.188(1.3) already asks whether the receiving charity's board is at arm's length from the revoked one. The same question can be asked of the receiving charity's funders.
Needs a policy decision — the only one of the five that does
Each is within reach of the data above. None follows from it.
Supported
That it received the largest coordinated funder move in CRA's record, from the funders of a charity revoked for lack of direction and control.
Not supported
Anything about how it directs its own spending. That requires an audit. CRA has published none.
INHERITING THE AUDIENCE IS NOT INHERITING THE FINDING
Supported
That it happened together, at a scale no other collapse approaches, to a registration with a shared history of boards and gifts.
Not supported
Who decided it. Intent needs a record of the decision. The public files hold the result.
A FUNCTIONAL READING · NOT A STATEMENT OF PURPOSE
What is published
Counts of funders, and dollars, exactly as their own returns report them.
What is not
The funders' names. A gift to a registered charity is lawful; naming {N['adopters']} charities on a page about a revocation would imply what the data does not show.
THE SUBJECT HERE IS THE FORM, NOT THE FILER
Every figure on this page is computed by measure.py from CRA's annual List of charities releases, {M['generated_from'][-9:]}, and written to measure.json. The page itself is rendered from that file by render.py, so it cannot state a number the build did not produce.
A gift counts at $5,000 or more in a year, read from the giver's qualified-donee schedule and dated by the release year of the giver's return. A collapse needs at least $1 million a year from charities before, and half of it gone over the next two years; only the first year of a collapse is scored, and only if the charity had at least ten funders. A follower is a funder that starts giving to a recipient it had not given to in the previous four years. Thresholds were set before the named charities were scored.
CRA's 2011 release contains no qualified-donee file. A missing year looks like a collapse, and a gift made in 2011 looks like a new one in 2012, so every event whose window touches 2011 is dropped rather than filled. That is why the first succession is described, not ranked.
Matched the Gazette notices, read the Act, and checked the study's tables. Director names are compared as written on the returns, which is the only identity the public file carries.
measure.json holds every figure. collapse-events.csv is all {T['events']} scored collapses by size and share, with only the named case labelled. three-charities.csv is the three registrations' filed lines, year by year. build/README.md is the recipe.