The question comes up once a client has run an Immediate Financing Arrangement for a decade and starts thinking about retirement income: can the same policy be borrowed against a second time, and can the interest simply be left to capitalize until then?
Most people expect the answer to turn on tax law. It usually does not. The tax rules permit most of what is being described. What fails first is the collateral arithmetic — and the second thing to fail is the deduction, for a reason almost nobody anticipates.
This page walks the constraints in the order they actually bind. It states what each provision says and links the primary source. It does not interpret the law for your facts — that belongs to your accountant and your tax lawyer.
An IFA pairs a permanent policy with a loan from a lender that takes the policy as collateral. A shareholder borrowing against a corporate-owned policy does the same thing: a lender advances money and takes the policy as collateral. The moving parts are identical. Only three things differ.
Who borrows The corporation in one case, the shareholder personally in the other. This single difference drives the entire tax analysis below.
What the money does Money used to earn income from a business or property supports an interest deduction under paragraph 20(1)(c). Money used to fund a lifestyle does not.
Whether interest is paid Paid interest and capitalized interest are treated differently, and not in the way most illustrations imply.
The consequence people miss
Lenders advance against the policy's cash surrender value, and published lender material describes arrangements reaching up to 100% of CSV for well-qualified borrowers, sometimes with additional security pledged alongside. Whatever the specific limit, the point is structural: an IFA that has been drawing for ten years has already encumbered a large share of the value the second loan would rely on.
A second borrowing is therefore possible only to the extent cash value has grown beyond what the first loan encumbers — and only if the lender will rank two borrowers against one policy at all. Many will not without a subordination agreement, and where the two borrowers are a corporation and its shareholder, some will decline outright.
Ask this before the structure is designed
If the corporation owns the policy and the shareholder borrows personally against it, the corporation is providing security for someone else's private debt. CRA has addressed this repeatedly under subsection 15(1), the shareholder benefit rule: where a private corporation guarantees or secures a shareholder's personal borrowing, a benefit may have been conferred.
The customary response is a guarantee fee paid annually by the shareholder to the corporation. Industry material generally puts a reasonable fee somewhere in the range of 1% to 4% of the outstanding balance, with 1.5% to 2% the most commonly cited. Three things matter more than the percentage: the fee has to be reasonable for the security actually provided, it has to be genuinely paid rather than recorded, and it has to be documented at the time.
Two loans, two sets of books
This is the provision that quietly destroys the “stop paying at 50, capitalize until 65” plan, and it is the one least often shown in an illustration.
Simple interest may be deducted under paragraph 20(1)(c) on either a paid or payable basis. Compound interest — interest charged on unpaid interest — is governed instead by paragraph 20(1)(d), and it is deductible only in the year it is actually paid, regardless of the taxpayer's accounting method.
| Treatment of interest | When deductible | Provision |
|---|---|---|
| Simple interest, paid | Year paid | 20(1)(c) |
| Simple interest, payable but unpaid | Year payable, on an accrual basis | 20(1)(c) |
| Compound interest, capitalized into the loan | Not deductible while it accrues | 20(1)(d) |
| Compound interest, later actually paid | Year of payment | 20(1)(d) |
So a client who stops paying interest and lets it roll into the loan balance is trading an annual deduction for cash flow. The deduction is not lost forever — it arrives in the year the compounded interest is finally paid, if it ever is — but it is unavailable for every year the balance is left to compound.
Where a policy is pledged for a loan, paragraph 20(1)(e.2) may permit a deduction for part of the premium. The conditions are cumulative:
Assigned to a restricted financial institution The policy must be assigned to the lender in the course of a borrowing from that institution.
Required by the lender The assignment has to be required as collateral — and the loan documentation should say so.
The interest must otherwise be deductible This is the condition that couples 20(1)(e.2) to everything above. If the interest ceases to be deductible, this deduction is exposed along with it.
Capped and apportioned The deduction is the lesser of the premiums payable and the net cost of pure insurance for the year, and only the portion reasonably related to the amount owing.
Why this compounds the previous section
There is a reason this area attracts scrutiny. An earlier generation of leveraged insurance plans, known as 10/8 arrangements, paired a policy crediting roughly 8% with a loan charging roughly 10%, producing a large annual interest deduction with little economic risk. The 2013 federal budget ended them.
A 10/8 policy is now defined in subsection 248(1). Broadly, it captures arrangements where the borrowing is from the insurer, or where the return credited to the policy's investment account is determined by reference to the interest rate on the borrowing, and the policy or that investment account is assigned as security. For periods after 2013 the rules deny the interest deduction and the 20(1)(e.2) collateral insurance deduction, and reduce the capital dividend account credit arising on the policy.
What this means for a conventional IFA
Deductibility follows the current use of the borrowed money, not the use at the time it was advanced. Borrowed funds that were invested to earn income and are later redirected to personal spending cease to support a deduction from that point. CRA sets out its interpretation in Income Tax Folio S3-F6-C1.
This is the point at which retirement plans commonly collide with the structure. A second borrowing taken at 65 to fund lifestyle is not an income-earning use, so its interest is not deductible — and if the two borrowings share an account, the tracing problem infects the first one as well. Separate accounts from day one is not administrative fussiness. It is the evidence.
The burden of proof sits with the taxpayer, so the records have to be good enough to defend years later, when the person reconstructing them may not be you.
Taking the question as it is usually asked — a client who starts an IFA at 40, pays interest for ten years, stops at 50 without repaying principal, and wants a second borrowing at 65 with interest capitalized in the meantime:
| Period | What happens | Where it stands |
|---|---|---|
| 40 – 50, interest paid | Corporate borrowing, funds deployed to earn income, interest paid annually | Deduction available under 20(1)(c); 20(1)(e.2) conditions capable of being met |
| 50, principal not repaid | Loan left outstanding against the policy | Ordinary. IFAs are commonly designed to remain outstanding and be settled from the eventual policy proceeds |
| 50 – 65, interest capitalized | Unpaid interest compounds into the balance | No deduction for the compounding while it accrues (20(1)(d)); 20(1)(e.2) exposed; balance grows geometrically |
| 65, second borrowing | Shareholder borrows personally against the same policy | Possible only within remaining collateral capacity; engages 15(1) and a guarantee fee; deductibility depends entirely on use of funds |
Where this actually breaks
There is no percentage in the Income Tax Act, and CRA has said plainly that it has no firm guidelines on when it will assess a benefit — it is a question of fact. What exists instead is a range that recurs in industry material, a valuation principle from the courts, and one decided case that priced a guarantee at arm's length.
| Source | What it supports | Weight |
|---|---|---|
| Industry and insurer material | 1% to 4% of the outstanding balance, with 1.5% to 2% most commonly cited | Convention, not authority |
| GE Capital Canada, 2010 FCA 344 | A 1% annual fee on guaranteed debt was held to be within an arm's length range | Decided, but a transfer pricing case under section 247 — not a 15(1) case |
| Youngman, 90 DTC 6322 (FCA) | A shareholder benefit is measured by what the shareholder would have had to pay to obtain the same advantage from a corporation of which they were not a shareholder | The governing valuation principle |
| CRA 2006-0174011C6 | Factors CRA weighs: arm's length dealing, the shareholder's ability to repay, and whether a reasonable fee was paid | Administrative position |
Read together, these point the same way. The fee is not a ritual payment that buys immunity at any size. It is an attempt to price what the security is actually worth — which is what Youngman says the benefit is measured by, and what the court in GE Capital did by asking what the borrowing would have cost without the guarantee. A borrower who could not have obtained the loan at all without the pledge has received something worth considerably more than 1%.
The uncomfortable part
The case worth reading before doing any of this is Golini v. The Queen, 2016 TCC 174, because the facts are closer to this question than anything else in the reported cases: a shareholder borrowed personally against a life insurance policy and annuity owned by his corporation, and a guarantee fee was payable.
The structure A $6,000,000 loan to the shareholder, effectively non-recourse, with recourse limited to the policy and annuity pledged by the corporation.
The fee was paid Guarantee fees of roughly $40,000 a year over fifteen years — around $600,000 in total — were paid or planned.
It did not work The Tax Court found a shareholder benefit of approximately $5.4 million: the loan amount less the guarantee fees. The fee reduced the benefit. It did not prevent it.
Why On the court's reading of the arrangement there was no sensible reason the shareholder would ever repay — everyone understood the policy and annuity were how the obligation would be met. A loan nobody expects to be repaid is not really a loan.
The surrounding facts This was not a plain-vanilla arrangement. The policy and annuity came from non-resident insurers at rates that did not appear commercial, and money moved in a circle. The court found several elements to be a sham, and held in the alternative that the general anti-avoidance rule would apply. Those findings supported the shareholder benefit conclusion.
What survived The court did allow a deduction for interest the shareholder actually paid out of pocket, which is consistent with everything above about paid versus capitalized interest.
A conventional arrangement — a Canadian insurer, a Canadian bank, commercial rates, a borrower who is genuinely on the hook — is a long way from these facts, and it would be wrong to read Golini as condemning every shareholder borrowing against a corporate policy. But the reasoning the court used is general, and CRA has since cited the case at the 2018 CALU roundtable when describing the arrangements it considers aggressive. The three features below are what put daylight between a real financing and the pattern the court rejected.
The lesson, stated plainly
Recourse is the question of what a lender can pursue if the loan is not repaid. It sounds like fine print. In this structure it is the single term that most determines whether the borrowing is a loan at all.
| Type | If the borrower defaults, the lender can pursue | What it signals |
|---|---|---|
| Full recourse | The pledged policy, and then the borrower personally — their accounts, their property, their other assets — for any shortfall | The borrower genuinely owes the money |
| Limited recourse | Only specified security, typically the pledged policy and nothing else | The borrower's personal exposure is capped at an asset they do not own |
| Non-recourse | Only the collateral; the borrower walks away from any shortfall | Economically closer to a distribution than a loan |
Golini's loan limited the lender to realising on the corporation's policy and annuity. That is what let the court conclude the shareholder “would not have to repay” — if the only thing the lender can ever touch belongs to the corporation, then the corporation has in substance paid the shareholder's debt, and a payment by a corporation for the benefit of its shareholder is what subsection 15(1) taxes.
Under full recourse the order reverses. The shareholder is the primary obligor and their own assets are exposed first in practical terms; the corporation's policy is additional security the lender holds in case the shareholder cannot pay. The corporation is backing a debt, not discharging one.
How to tell from the documents
Full recourse has a second half that is easy to miss: the corporation's own right to be repaid. A guarantor that pays a debt ordinarily has a right to be reimbursed by the person whose debt it paid. If the lender ever realises on the corporation's policy, the corporation should have — and should keep — an enforceable claim against the shareholder, or against the estate, for what it lost. An agreement that waives that right turns a guarantee back into a gift.
Full recourse on paper is not enough if everyone involved knows the loan can only ever be settled one way. The second question is the one the court was really asking: is there a sensible reason this borrower would repay, and a realistic source to repay from, other than the corporation's policy? CRA lists the shareholder's ability to repay among the factors it weighs, alongside arm's length dealing and the fee.
A repayment path is real when it can be shown, not asserted:
Capacity at inception A personal balance sheet and income at the time the loan is taken that could plausibly service and retire it — personal non-registered assets, real estate equity, pension income, the value of other holdings. Not the corporation's.
Interest paid from personal funds Interest and the guarantee fee leaving the shareholder's own account on schedule. A borrower who never pays anything, where the whole cost capitalizes against a corporate asset, looks like someone not expecting to repay — and gets no deduction while it compounds.
A stated source and horizon A written answer to “repaid from what, by when” — a planned sale of a personal asset, a drawdown of personal investments, a schedule of partial repayments — revisited as circumstances change.
An estate that can settle it Where the intention is that the loan remains outstanding for life, an estate plan showing personal assets sufficient to repay it, with the policy as the lender's backstop rather than the plan. If the policy is the only realistic source, the loan is being repaid by the corporation.
Loan-to-value kept in range Monitoring that keeps the balance well inside the lender's limit. A balance deliberately run up to the policy's full value signals that the policy, not the borrower, was always going to pay.
The test in one sentence
Contemporaneous means created at the time the thing happened, not reconstructed when CRA asks. The difference matters because a review may come years later, the burden of proof sits with the taxpayer, and a document dated after the question was raised proves only that someone was worried. The file that defends this structure is built on the day it is set up and added to every year after.
| Document | Signed by / when | What it proves |
|---|---|---|
| Loan agreement | Lender and shareholder, at inception | The shareholder is the borrower, on full recourse, at commercial terms |
| Collateral assignment | Corporation and lender, at inception | The policy is assigned as security, and that the lender required it |
| Guarantee and fee agreement | Corporation and shareholder, at inception | The fee rate, the balance it is charged on, the payment dates, and the corporation's right to be reimbursed if the security is ever realised |
| Directors' resolution | The corporation's board, before the assignment | The corporation made a deliberate, authorised decision to provide security for consideration |
| Fee pricing memo | The tax advisor, before the rate is set | How the percentage was arrived at — ideally what the shareholder would have paid to borrow without the pledge, following the yield reasoning in GE Capital |
| Personal financial statement | The shareholder, at inception and on renewal | Capacity to repay from personal resources |
| Proof of payments | Bank records, every year | Fees and interest actually moved from the shareholder's account to the corporation and the lender — not accrued, not journal entries |
| Corporate books | The corporation's accountant, every year | The fee recorded and reported as income by the corporation |
| Use-of-funds tracing | A dedicated account, continuously | Where the borrowed money went, which is what decides whether any interest is deductible |
| Annual review | Advisor and shareholder, every year | Loan-to-value, whether the fee is still reasonable, and whether the repayment plan still holds |
| Tax opinion | The tax advisor, before implementation | The 15(1), 20(1)(d), 20(1)(e.2) and 10/8 analysis for these specific facts |
Three failure patterns show up again and again. The fee agreement exists but the fee was never actually paid, or was paid once and then forgotten. The resolution was written after the assignment, dated to look earlier. And the loan terms quietly limited recourse while every other document described a conventional borrowing. Each of them is worse than having no document at all, because it shows the right answer was known and not followed.
What the three safeguards have in common
This is the right question to ask, and the arithmetic is unkind. A 3% guarantee fee on top of a 5% lending rate is an 8% all-in cost on a balance that compounds — and the fee itself leaks, because it is income to the corporation and taxed again on the way back out to the shareholder.
| Guarantee fee | All-in annual cost | Balance after 20 years of $100,000 draws | Cost of the fee |
|---|---|---|---|
| 0% | 5% | $3,306,595 | — |
| 1% | 6% | $3,678,559 | +$371,964 |
| 2% | 7% | $4,099,549 | +$792,954 |
| 3% | 8% | $4,576,196 | +$1,269,601 |
| 4% | 9% | $5,116,012 | +$1,809,417 |
The fee is not lost outright, but less of it comes back than people assume. The corporation receives it as investment income. Because it pays the money straight back out as a dividend, the refundable portion of that tax is returned, leaving roughly 16% to 20% of corporate tax depending on the province. The shareholder then pays dividend tax on what arrives. Depending on the shareholder's bracket, between 40% and 60% of the fee leaks — about 0.8% to 1.2% of the balance a year at a 2% fee, and 1.2% to 1.7% at 3%. An earlier version of this page overstated the leakage by leaving out the dividend refund; the figures here correct it.
Where it stops making sense
Two things qualify that comparison, in opposite directions. In the borrowing's favour: the dividend permanently drains the corporation, while the borrowing leaves the corporate assets intact and compounding, and the loan is settled from the policy proceeds with a capital dividend account credit arising on the difference between the benefit and the policy's adjusted cost basis. Against it: the balance is still compounding, and modelled to age 85 a 3% fee on $100,000 of annual draws against $2,000,000 of starting cash value reaches a 90% loan-to-value around year 21 — at which point the policy can be collapsed to satisfy the lender, realising the accrued gain as income in a single year.
So the answer
From the lender Whether a second collateral assignment against the same policy, for a different borrower, will be accepted at all — and on what subordination terms and loan-to-value limit.
From the tax advisor The 15(1) analysis and a defensible guarantee fee; the treatment of 20(1)(e.2) during any period interest is not paid; and confirmation that the arrangement falls outside the 248(1) definition of a 10/8 policy.
From the illustration Loan balance against cash surrender value to age 85, stress-tested well above current rates, showing the year the structure would fail if it fails.
From the client Whether the interest can actually be paid during the years the plan proposes to capitalize it. Where it can, capitalizing is usually the more expensive choice.
None of this makes the combination impossible. It makes it a structure that only works when four separate parties agree in advance that it works — and one where the attractive-sounding step, letting interest ride, is the step that costs the most.
Primary legislation and CRA material first, then the industry and professional commentary used for context. Everything above is stated from these; where they conflict with a particular set of facts, the facts and your own advisors govern.
Income Tax Act, section 20 Deductions in computing income from business or property — paragraphs 20(1)(c), 20(1)(d) and 20(1)(e.2). Justice Laws Website.
Income Tax Act, subsection 248(1) Definitions, including “10/8 policy”. Justice Laws Website.
CRA, Income Tax Folio S3-F6-C1 Interest Deductibility — the current-use test, the treatment of compound interest, and the exclusions. Replaced Interpretation Bulletin IT-533.
CRA, IT-309R2 (archived) Premiums on Life Insurance Used as Collateral — the conditions for the collateral insurance deduction.
CRA, 2008-0270441C6 Roundtable: 20(1)(e.2) and net cost of pure insurance.
CRA, 2006-0174011C6 Roundtable: guarantees and shareholder benefits.
Department of Finance, 2013 Explanatory Notes Relating to the Income Tax Act — the 10/8 measures. See also BMO Insurance, Budget 2013: impact on 10/8 plans and leveraged annuities.
Golini v. The Queen, 2016 TCC 174 Shareholder benefit where a corporate policy and annuity secured a shareholder's non-recourse borrowing. Case summary, Tax Interpretations; Canada Life analysis; Advisor.ca commentary; CRA's response, 2018 CALU Roundtable Q.7.
Canada v. General Electric Capital Canada Inc., 2010 FCA 344 A 1% guarantee fee upheld as arm's length under the yield approach. Case report; Gowling WLG commentary.
Youngman v. The Queen, 90 DTC 6322 (FCA) Measuring a subsection 15(1) benefit — what the shareholder would have paid for the same advantage as a non-shareholder.
CRA, Income Tax Folio S3-F1-C1 Shareholder Loans and Debts.
Canada Life, Tax and Estate Planning Group Collateral life insurance deduction and Loan to shareholder vs. loan to corporation.
Manulife, 2018 What insurance structures are keeping the Canada Revenue Agency up at night? — CRA's 2018 CALU roundtable response and the arrangements it flagged.
Sun Life Corporate Ownership of Life Insurance — shareholder benefit exposure where a corporate policy secures personal borrowing.
Miller Thomson Fundamentals of interest deductibility in Canada.
What an IFA is, who it suits, and the corporate mechanics including the capital dividend account.
Paragraph 20(1)(c), the current-use test, and the record keeping the deduction depends on.
Where permanent insurance earns its place in a plan, and where term is the honest answer.
Model a corporate policy loan with the guarantee fee and capitalizing interest, including the year it collapses.
This page summarises publicly available Canada Revenue Agency material and general mechanics. It is not tax advice, and deductibility depends on individual circumstances and current legislation. Wallace Wang Financial Services does not prepare tax returns or provide tax opinions.