A married couple in Ontario, both in their sixties and two years apart in age, moving from employment into full retirement across the period modelled. Their income in retirement comes from four sources that behave very differently for tax purposes: two defined benefit pensions, Canada Pension Plan and Old Age Security starting at different times for each of them, dividends from a private corporation they continue to own, and registered and non-registered savings.

The trigger for the engagement was a single event. One partner was leaving employment and was owed a retirement bonus of roughly $130,000, payable at their election in either of two calendar years or split between them.

What makes this harder than it sounds is that almost every other source of income was also in motion at the same time. One pension included a bridge benefit that would step down. The second pension was starting partway through a year. Canada Pension Plan was beginning for one partner several years before the other. And the couple owned a corporation with a finite pool of retained earnings available to distribute, which meant that any year they took less salary or bonus, they would need to draw more from the company to cover their living costs.

The clients asked: “I can take my retirement bonus this year, next year, or split it. Which costs us the least tax? And do we have enough coming in either way to actually live on without touching things we don’t want to touch?”

Restated technically: for a two-taxpayer household transitioning into retirement with defined benefit pension income, staggered Canada Pension Plan and Old Age Security commencement, private corporation dividend capacity and available registered contribution room, what allocation of a six-figure retirement allowance between two consecutive taxation years minimises combined household tax across a four-year projection, while meeting a target net spending requirement each year, preserving corporate distribution capacity, and avoiding Old Age Security recovery tax?

The spending requirement, built from the bottom up

The model does not start with income. It starts with what the household actually spends, itemised across fixed costs, discretionary costs and annual lump sums, indexed for inflation across the projection, with a deliberate annual cushion added on top. Only once a required net figure exists does the model work backwards to the gross income needed to produce it. Running the analysis the other way round, from income down, would have answered a question nobody asked.

Four separate marginal rate columns, not one

Ontario combined marginal rates were built out across eleven brackets, and each bracket carries four different rates depending on what kind of income fills it: ordinary income, capital gains, eligible dividends and non-eligible dividends. A single blended rate would have produced the wrong answer here, because this household receives ordinary income and dividends simultaneously and the interaction between them is precisely what the bonus timing decision turns on.

Pension income splitting, optimised year by year

Eligible pension income can be split between spouses up to fifty percent, and the optimal split is not the same every year. In some years of the projection the transfer runs one way and in other years it reverses, because the underlying income mix changes as pensions start, bridge benefits end and the bonus lands. The model recalculates the split each year rather than fixing it once.

Corporate distribution capacity as a running constraint

The couple’s private corporation holds a finite pool available for distribution. Every scenario tracks how much of that pool has been consumed and how much remains at the end of the projection. This is what stops the analysis from being gamed: a scenario can always look better on tax by quietly drawing more from the company, and unless the remaining balance is tracked, nobody notices.

Withholding versus the balance due in April

Tax withheld at source and tax actually owed are separated throughout. Payroll withholding, pension withholding and withholding on registered withdrawals are each modelled, and the residual balance payable by April 30 is shown per person per year. A household can be perfectly optimised on total tax and still be caught out by a five-figure cheque they were not expecting.

Old Age Security recovery tax

An explicit clawback line is carried in every year of every scenario. Pushing a large bonus into a year where Old Age Security has already commenced can trigger recovery tax that a headline marginal rate comparison will not reveal.

• A summary comparison of all four scenarios showing total household income, total tax, the balance payable by each spouse at April 30 in each year, and the corporate distribution capacity remaining at the end of the projection.
• Four full scenario models, one per tab, built identically so they can be read side by side: spending requirement, gross income target, income build by person and by source, splitting, withholding, balance owing, and a cash sufficiency check.
• A live Ontario marginal rate table across eleven brackets and four income types, driving the tax calculation in every scenario rather than sitting beside it as reference.
• A household expenditure schedule broken into fixed, discretionary and annual categories, which is the input the whole model is built on.
• Six numbered assumption notes covering the pension bridge benefit, expected corporate income and expenses, investment yield treatment, Canada Pension Plan commencement, Old Age Security commencement, and the treatment of the retirement bonus itself.
• A written recommendation naming one scenario, with the reasoning for choosing it over the one that scored marginally better.

The workbook is seven tabs: a summary, four scenario models, a rate table and an expenditure schedule. Every figure below is reproduced from the sample, so the structure and the depth can be assessed without downloading anything.

Tab 1. Summary

This first tab is the only page the clients need to read, and it is built to be read in about ninety seconds. Each of the four scenarios gets an identical block: total household income by year, total tax by year, the balance each spouse owes at April 30 in each year, and one line that most personal tax plans leave out entirely, which is how much of the corporation’s distributable pool is still there at the end.

Underneath the four blocks sits a two-year comparison, because two years is where the entire decision lives. Everything from the third year onward is identical across all four scenarios, since the bonus is long gone by then. Isolating the two years that actually differ is what makes the comparison legible.

Figures rounded. Scenarios 2, 3 and 4 produce identical household income and identical remaining corporate capacity, which is what makes them directly comparable on tax alone. Scenario 1 does not, for the reason set out in section 6.

Tabs 2 to 5. The four scenario models

These four tabs are structurally identical, which is deliberate and is more important than it sounds. Because every row sits in the same place on every tab, the four scenarios can be compared line by line rather than only at the bottom. If pension splitting behaves differently in scenario 3 than in scenario 2, that difference is visible on the same row of both tabs.

Each tab reads top to bottom in five movements.

First, what the household needs. Monthly expenses, indexed forward each year. Then one-time items, which in this case included two significant gifts to family offset by an expected inheritance, netting to nil in the years they fall. Then a deliberate annual cushion. The result is a net annual income requirement, and this is the number everything else has to satisfy.

Second, the gross income target. The net requirement is converted to a gross figure and split between the two spouses. It is rounded upward on purpose, and the model says so, because the alternative is a plan that works to the dollar and fails the moment anything moves.

Third, the income build. For each spouse in each of the four years, every source is listed on its own line: employment income, Canada Pension Plan, Old Age Security, the retirement bonus, defined benefit pension, dividends from the corporation, the pension split transfer, registered plan minimum withdrawals, and non-registered investment income. Then the registered contribution is deducted to reach net income for tax purposes. Laying every source out separately is what allows the bonus to be moved between years without rebuilding anything else.

Fourth, withholding and the balance due. Payroll withholding, pension withholding and withholding on registered withdrawals are each subtracted to give net income after withholding. Below that sits the additional amount owing to Canada Revenue Agency at April 30, per spouse, per year, and then the Old Age Security recovery tax line. This is the block clients actually respond to, because it is the difference between knowing your tax bill and knowing what lands in your bank account and when.

Fifth, the sufficiency check. Total available cash for the year, a deficiency check that must return nil, and the expected leftover after all spending is covered. Every scenario carries a short note explaining why any leftover exists, whether from rounding the gross income target up or from the bonus itself. A leftover with no explanation is indistinguishable from a modelling error.

Below all of that sit six numbered assumption notes. They state, among other things, that the pension bridge benefit is assumed to equal the gap between a known prior-year figure and an expected future one, that expected corporate income sits in a stated range with an allowance for annual expenses net of depreciation and taxes, and that investment yield is assumed reinvested rather than drawn, per the client’s separate wealth management report. Naming the source of an assumption is what allows someone else to challenge it.

Tab 6. The Ontario rate table

This tab is a working component, not an appendix. Eleven combined federal and provincial brackets, and beside each one, four separate rates: ordinary income, capital gains, eligible dividends and non-eligible dividends. The scenario tabs pull from this table by cell reference, so a rate change is made once and flows through all four scenarios.

Two things on this tab are worth pointing at. The first is that the eligible dividend rate is negative in the lowest two brackets. That is not an error. It is the effect of the dividend gross-up and the dividend tax credit at low income levels, and it means eligible dividends received by a low-income spouse can reduce tax on other income. In a household where one spouse can receive corporate dividends and the other cannot, that is a real planning lever rather than a curiosity.

The second is that the tax calculation is built bracket by bracket rather than by applying a single top rate. Each spouse’s income is layered through the brackets in sequence. For a household whose income moves across five or six brackets between the highest and lowest scenario, a top-rate shortcut would materially misstate the comparison.

Tab 7. The expenditure schedule

This is the foundation the entire model rests on, and it is the tab most personal tax plans do not have at all. Household spending is broken into three groups: fixed monthly commitments such as housing, insurance and utilities; discretionary monthly spending; and annual or seasonal costs that would be invisible in a monthly average, such as maintenance, professional services and travel. Alongside sits the current income side of the household.

It matters that this is an input rather than an assumption. The whole analysis is anchored to what this household actually spends, so when the clients said a number looked wrong, there was one cell to change and the four scenarios updated together.

Timing the bonus was worth roughly $12,400. Across the two years that matter, combined household tax ranged from about $129,100 under the best scenario to about $141,500 under the worst. Nothing about the couple’s circumstances changed between those two outcomes. The same bonus, the same pensions, the same spending. Only the calendar year the money landed in.

Concentrating the bonus in either single year was the wrong answer, in both directions. Taking it all in the first year pushed both spouses into higher brackets simultaneously and produced the second-worst result. Taking it all in the second year produced the worst result, for a different and less obvious reason set out below. Splitting it was better than either extreme, which is the intuitive answer, but the intuitive answer was right for reasons the model had to be built to see.

The worst scenario looked like it had the most money, and that comparison was false. Scenario 1 shows the highest total household income across the projection and the highest after-tax income. It is still the worst option. Deferring the entire bonus to the second year left a shortfall in the first, and the only way to cover the household’s spending in that year was to draw roughly $13,100 in dividends out of the corporation. That is why its income looks higher: it is not extra income, it is money moved forward out of a finite pool. The corporate capacity line proves it, falling by almost exactly the amount of those dividends. Scenarios 2, 3 and 4 all finish with identical household income and identical remaining corporate capacity, so they can be compared on tax alone. Scenario 1 cannot, and comparing after-tax income across all four without adjusting for that would have pointed the clients at the worst option.

Deferring the bonus also cost a registered contribution. With no bonus received in the first year, there was no room to make a meaningful contribution that year, so a deduction available in the split scenarios simply disappeared in scenario 1. Two separate effects, the forced corporate draw and the lost contribution, both flowed from the same decision to defer.

The optimiser found a better answer and the advice was not to take it. Scenario 3 is the mathematically optimal split. Scenario 2 is the split the clients had already worked out for themselves based on when they wanted the money. Scenario 3 beats scenario 2 by $85 across two years, on a combined tax bill of roughly $129,000. That is a difference of about six one-hundredths of one percent, well inside the error bars of every assumption the model rests on. The recommendation was scenario 2. Restructuring a household’s cash flow around $85 of theoretical benefit, and asking clients to change a plan they were comfortable with, is not advice. It is arithmetic being mistaken for judgment.

Nobody was going to be surprised in April. The balance payable to Canada Revenue Agency was modelled per spouse per year in every scenario, and in the bonus year it ranged from a few thousand dollars to over $35,000 depending on which scenario was chosen. Knowing which of those was coming, twelve months ahead, was worth as much to the clients as the tax saving itself.

Scope decisions are part of the work, and each of these was made for a reason.

• The optimal scenario was not recommended. Scenario 3 wins by $85 across two years. Recommending it would have meant asking the clients to rearrange their plans for a benefit smaller than the rounding in several of the model’s own assumptions. The scenario the clients had chosen was recommended instead, and the $85 gap was disclosed rather than hidden, so the choice was theirs.
• Investment returns and portfolio drawdown were not modelled. The household has a separate wealth management report covering asset allocation and yield. This model takes that report’s outputs as an input, states that it is doing so on the assumptions tab, and does not second-guess it. Two advisors independently modelling the same portfolio produce two answers and no clarity.
• No investment advice was given. Whether to hold, sell or reallocate anything is outside the scope of a tax plan and outside what a CPA firm should be opining on.
• The corporate side was not optimised. Corporate distribution capacity is tracked as a constraint so that no scenario can quietly borrow from it, but the corporation’s own remuneration strategy, its investment income and its long-term wind-down were not modelled here. That is a separate engagement with a separate question.
• The projection stops at four years. Far enough to capture the bonus decision and the years immediately after it settle, not far enough to require assumptions about legislation, indexation and longevity that would make the output look more precise than it could possibly be.
• Estate, probate and succession were not addressed. Relevant to this household and deliberately not folded into a document whose job was to answer one timing question well.

The engagement involved a four-year projection across two taxpayers in a single household; eleven combined Ontario and federal marginal brackets applied separately to ordinary income, capital gains, eligible dividends and non-eligible dividends, with the eligible dividend rate turning negative in the lowest brackets through the interaction of the dividend gross-up and the dividend tax credit; eligible pension income splitting of up to fifty percent recalculated annually and reversing direction between years as the underlying income mix changed; a defined benefit pension containing a bridge benefit that steps down at a known point; a second defined benefit pension commencing partway through a year; Canada Pension Plan commencing in different years for each spouse; Old Age Security commencement and the recovery tax tested in every year of every scenario; a retirement allowance electable across two taxation years in any proportion, with part of it directed to a registered plan and the registered contribution room depending on which year the allowance was received; registered retirement income fund minimum withdrawal rates; dividends from a private corporation drawn against a finite distributable pool tracked as a running balance across all four scenarios; withholding at source on employment, pension and registered withdrawals reconciled against the balance payable by April 30 for each spouse in each year; a tax-free savings account modelled as a non-taxable top-up where a scenario ran short; a household expenditure schedule split into fixed, discretionary and annual components and indexed for inflation to derive the required net income; and intergenerational gifts offset against an expected inheritance in the years they fell.

Relevant services

• Personal and corporate tax planning
• Advisory and fractional CFO services
• Accounting and tax for owner-managed businesses
• Owner-manager remuneration, salary versus dividend calculator

Related case studies

• Canada-US expansion: branch, subsidiary or sister corporation
• Not-for-profit board reporting and variance analysis

Supporting work sample

The full model is available below as an illustrative sample. It reproduces the structure, the calculations and the analytical approach of a real engagement, with the household, the individuals and all figures replaced. It is provided so that the depth of the modelling described above can be verified. It is not tax advice, and it is not a template for use without professional consultation, since personal tax outcomes depend entirely on individual circumstances and on legislation in force at the time.

Download: personal tax plan and bonus timing model (illustrative sample)

Seven tabs: summary comparison, four full scenario models, an Ontario marginal rate table and a household expenditure schedule.

Includes:

✓ Four-scenario comparison of retirement bonus timing

✓ Four-year household income and tax projection for two spouses

✓ Ontario marginal rates across eleven brackets and four income types

✓ Annual pension income splitting optimisation

✓ Corporate distribution capacity tracked across every scenario

✓ Withholding reconciled to the balance owing at April 30

✓ Old Age Security recovery tax tested in every year

✓ Household expenditure schedule driving the required net income

✓ Six numbered assumption notes with stated sources