Interest-Only Investment Loan

Objective — invest a lump sum borrowed at t=0, servicing only the interest, so the full principal stays invested for the entire holding period.

Risk level — 2 — Moderate risk. Risk justification — net market exposure increases and loss can exceed the amount of the investor’s own capital, which puts it at level 2 minimum under the assignment rule; it is not level 3 because a dedicated term investment loan is not callable on a market decline the way margin is. Benefit justification — unverified — needs Talbot. Jurisdiction — Canada. The model is Canadian-tax-specific: 13 provinces with Quebec’s limited-deductibility divergence. U.S. support is designed but not implemented (sd-math/docs/us-tax-design.md).

Prerequisites

• Capacity to service the interest from income alone, through a market decline — CIRO’s suitability guidance treats debt-service capacity as a required assessment, not a preference.
• A long horizon. The strategy’s benefit is a compounding spread, not a near-term gain.
• High-cost consumer debt cleared first — see eliminate-high-cost-consumer-debt-first.
• A non-registered (taxable) account. sd-math models account_type="taxable" only; registered and tax-free account types deliberately raise NotImplementedError rather than silently applying the wrong model.

Mechanism

A single loan is drawn at t=0 and invested. Only interest is paid, so the loan balance stays constant and the whole principal remains invested for the full period. At the end, the position is sold, the loan repaid, and tax on the accrued gain settled — the net leveraged balance is what remains.

The engine computes, per year: the leveraged balance, adjusted cost base, distributions before and after tax, interest cost and its deduction, and the net leveraged position against an unleveraged baseline. It also solves for the break-even return — the return at which leveraging matches not leveraging (IntOnlyBetterThan in the VB6 source).

Benefits

• The full amount is invested for the whole horizon; nothing is diverted to principal repayment.
• Interest is deductible under ITA §20(1)(c), so the after-tax cost is gross rate × (1 − marginal rate) — 7% becomes 3.25% at the Ontario top rate.
• The tax arbitrage is structural: the deduction is taken at the full marginal rate while dividends and capital gains are taxed at preferential rates. At the Ontario top bracket, $7,000 of interest paid against $7,000 of eligible dividends received nets +$2,217/year (canada-investment-taxation.md §9.4).

Risks

• The principal never amortizes. After a decline, the debt is unchanged while the asset is smaller.
• Rate risk: interest cost rises with rates while returns do not follow.
• Deductibility risk: if the investments stop paying income, or the position is sold, deductibility can be jeopardized.
• Behavioural risk: CIRO’s own guidance names emotional tolerance as separate from risk tolerance — an investor may accept risk intellectually and still sell at the worst time.
• In Canada there is no §163(d)-style cap, so losses flow fully through income. The source calls this “a double-edged sword.”

Failure modes

• Servicing interest from portfolio distributions rather than income, so a distribution cut forces a sale.
• Return-of-capital distributions spent personally — the Van Steenis trap; see interest-tracing-hygiene.
• A horizon shorter than the strategy needs, forcing a sale at a low.
• Sizing the loan to what the lender will approve rather than to what the investor can service.

Costs

Interest for the full term; setup costs; tax on the crystallized gain at exit.

Tax considerations

Interest deductible under ITA §20(1)(c) in a non-registered account. Quebec limits deductibility to taxable investment income, with carryforward — modelled as prov_deductibility_limited. Eligible-dividend gross-up of 1.25 applies to the Canadian deductibility-limit calculation.

Who it may suit

An investor with stable income, a 10+ year horizon, reserves, no high-cost debt, and — per the sourced booklet position — a conservative loan size; see conservative-leverage-ratio.

Who should avoid it

Most people. Under Rule 1 of Strategic-Plan, the honest output for an level 2 strategy is usually “most people in your position should not do this.” Specifically: unstable income, short horizon, no reserve, existing high debt ratios, or an investor who would not hold through a 50% decline.

Implementation outline

1. Establish suitability first — debt-service capacity, horizon, reserves, emotional tolerance.
2. Model the outcome across a range of returns including negative ones, and read the break-even return.
3. Size the loan to serviceable interest, not to approved credit.
4. Hold in a non-registered account with segregated tracing.
5. Document the rationale — CIRO requires the reason leverage is suitable, not merely the client’s consent.

Evidence status

implemented-and-verified — the highest confidence in the library. sd-math/strategies/interest_only.py is a verbatim port of LevPro’s CalcIntOnlyLevAnalysis, verified four ways: golden fixtures against LevPro’s own PDF reports (to the dollar), hand-derived first-principles tests (to the cent), a line-by-line VB6 source cross-check, and an independent from-scratch algebraic re-derivation sharing no code with the library.

Known limits carried from the audit: constant tax parameters across the horizon (today’s tax environment as the proxy for the future); DistribReinvestRatio fixed at 1.0; taxable accounts only; first-principles coverage is one scenario per strategy family.

Counterarguments

• FAIR Canada documented leverage being recommended to unsuitable retail investors, argued that the industry’s incentives (larger AUM through leverage) create a structural conflict, and recorded the real harm when 2008–09 margin calls arrived. That is the strongest available argument against this strategy and it is evidence, not opinion.
• The rebuttal SDC can honestly make is narrow: the objection is to unsuitable recommendation, which is why Rule 1 makes the product architecturally incapable of pushing anyone into leverage.

Variants

Implemented in sd-math: annual and monthly cash-flow variants, and historical-return variants (interest_only_monthly, interest_only_historical, interest_only_historical_monthly). Also implemented: solving for the loan amount.

Related strategies

term-investment-loan · interest-only-then-term-loan · conservative-leverage-ratio · better-rates-investment-loan · disappearing-source-continuation · margin-account-leverage

Sources

• ~/projects/monorepo/packages/sd-math/src/sd_math/strategies/interest_only.py — the implementation
• ~/projects/monorepo/packages/sd-math/docs/accuracy-audit.md — the four verification layers, assumptions and known limits
• ~/projects/monorepo/packages/sd-math/docs/design.md — algorithm derivation and the “No Math Inference” rule
• Core/_WorkingOn/Research/canada-investment-taxation.md §9.1–9.5 — hurdle rate, after-tax cost, tax arbitrage, Canada-specific risks
• Core/_WorkingOn/Research/LevPublications/Leverage-Publications-Summaries.md — CIRO suitability and supervision guidance; FAIR Canada, Regulators Need to Act on Leveraged Investing

Open questions

• U.S. implementation (designed, unbuilt) — gated behind the Canada restart.
• Whether the published record should show the break-even return by default. A product decision that Rule 1 arguably answers yes.