Elections 2019: Who is really getting a tax cut?

It’s Canadian federal election season again, and that means party platforms and arguments about tax cuts, deficits, and how much each campaign proposal will cost. As usual, the Office of the Parliamentary Budget Officer (PBO) provides independent estimates of the cost of campaign proposals.

But the cost to government is only one half of the equation. To whom does this money go?

Why might this matter? Maybe you think the increasing income inequality in our society is a problem. Or maybe you think our society needs more income inequality to incentivize people and capital to be more productive. In any case, I think it’s interesting to see who these election tax policy proposals are really benefitting or costing. After all, almost everyone can be “middle class” if the definition is sufficiently vague.

In this article, I look at personal income tax proposals from four parties (Conservative, Liberal, NDP, and People’s Party), and how the benefit or cost is distributed among people at varying income levels. I try to emulate the methodology the PBO uses in its cost estimates, and then break down the financial impact by income group (a “distributional analysis”). Both the PBO and I use Statistics Canada’s Social Policy Simulation Database and Model (SPSD/M) to model tax system policy. And according to the SPSD/M license agreement, I must make the following statement:

This analysis is based on Statistics Canada’s Social Policy Simulation Database and Model. The assumptions and calculations underlying the simulation results were prepared by Henry Wong and the responsibility for the use and interpretation of these data is entirely that of the author.


Fed. Govt. Cost (2023, $M) Fed. Govt. Cost (PBO FY 2023-24, $M) 50% of impact shared by top x% individuals
Conservative Party 5748 5890 (link) 27%
Liberal Party 5390 5634 (link) 36%
New Democratic Party -1279 -964 (link) 0.1%
People’s Party 37899 n/a 6%
Green Party Too complicated (link)

Of the four proposals, three are tax cuts (positive cost to government), while the NDP proposes a tax increase (negative cost). The table above includes the PBO estimates as a sanity check. My estimate of the cost to the federal government should be close to PBO’s estimate. I chose to look only at year 2023, which is the year when the Conservative income tax rate reduction is fully phased-in. The People’s Party does not appear to have submitted any estimate requests to the PBO.

The Conservative and Liberal proposals are surprisingly similar. Both offer about $5.5B/yr of tax cuts, targeted at roughly the highest-income 40%–50% of individuals. The Conservative proposal gives more money to fewer people than the Liberal proposal ($340 to the top 30% of individuals vs. $250 to the top 50%), but neither proposal provides much benefit to the lower third of the income distribution.

The NDP proposes increasing the top tax bracket rate from 33% to 35%. Very few people have enough individual income to be affected by the top tax bracket, so half the burden of this tax increase falls on the highest-income 0.1% of individuals.

The People’s party proposes an increase to the Basic Personal Amount, lower tax rates, and elimination of personal capital gains tax. The net result is a high cost ($38 billion/year), with half of the money going to the top 6% of income earners.


I used Statistics Canada’s SPSD/M to estimate the static impact of personal income tax policy changes. I break up the population into income groups, then plot a graph showing the financial impact to each income group.

For each proposal, I plot four graphs. The first two are based on individual income, while the last two are based on per-person family income. In all graphs, the horizontal axis is a ranking of persons, sorted by (pre-tax) income: Low-income on the left, high income on the right. The first set of graphs uses individual income, with non-tax filers excluded (31.6M estimated tax filers in 2023). The second set sorts persons (including non-tax filers) by the average per-person income of their family, because families typically share incomme among all family members. In this model, whether a child with no income is a “low-income” or “high income” individual depends on family income, not the child’s individual income. A single person with $1000 income is ranked the same as both people in a family of two that has $2000 total family income.

Within each pair, the first graph shows the average change in “consumable” (post-tax) income per person in each income group. This graph shows how much money each person gets (or pays) as a result of the policy change. The second graph shows the cumulative cost to the federal government, with the same horizontal axis as the first graph. The vertical axis is scaled to the total cost of the proposal. This graph can be used to see how the pool of money is distributed between different income groups. The second graph is close, but not exactly the same as summing over the first graph, because changes in federal income tax policy affects the amount of other taxes (provincial and federal sales taxes) collected.

Detailed Results

Conservative Party of Canada: Personal Income Tax Rate Reduction – 32644536

This policy lowers the tax rate of the first income tax bracket from 15% to 13.75%, along lowering the value of non-refundable tax credits from 15% to 13.75%. (In SPSD/M, these are the FTX and FNTCR parameters)

Individual Income

Although “lowest tax bracket” sounds like it would benefit most people, in fact, nearly all of the benefit goes to the highest-income half of the tax-filing population. There are many tax filers with low income, and only taxpayers with taxable income beyond the upper end of the first tax bracket ($51667 in 2023) will see the full benefit of the tax cut. High-income individuals still see a benefit on the portion of their income that falls within the lowest tax bracket, so the per-person benefit flattens out at around $340/year at high incomes.

The cumulative chart shows that the lowest 50% of the tax-filing population only share about 12% of the new government “spending”.

Family Income

Families often consist of both high-income and low-income members (e.g., working parents and zero-income children), and the wealth of each member depends more on family income rather than the individual’s own income. For example, a tax policy that targets “low income people” doesn’t necessarily need to target low-income individuals (such as a child in a wealthy family), but should target individuals in low-income families. When sorting by per-person family income, the graph is more spread out horizontally, reflecting the diversity of families: There are single-person families, and multi-person families where there can be zero, one, or more income earners.

When plotted by family income, the trend is the same, but less extreme. The tax cut still mainly helps families with high income. The lowest-income half of people share about 20% of the new government spending.

Liberal Party of Canada: Basic Personal Amount Increase – 33121046

This policy increases the Basic Personal Amount (non-refundable tax credit) from $13092 to $15000 for those with taxable income below $160137 ($150605 in 2020 scaled by CPI inflation until 2023), and to $14046 for taxable income below $228136 (scaled from $214557 in 2020). In SPSD/M, the BXM parameter was changed. Although the PBO used SPSD/M in “glass box” (custom code) mode, I achieved the same result by running the simulation in “black box” mode three times (with different BXM values for taxable incomes falling into each of the three ranges) and summing the results.

Individual Income

Increasing the Basic Personal Amount benefits more people than decreasing the tax rate of the lowest tax bracket because a taxpayer requires less income before getting the full benefit. One difference in the Liberal proposal is that the increased tax credit does not apply to those with high incomes: the top 5% of income earners see reduced or no benefit from this proposal.

Family Income

About 2/3 of the benefits go to the highest-income half of families, though the top 4% of families see a reduced benefit.

This illustrates the difficulty of targeting the lowest-income third of the population using tax policy. This income group pays no or little tax, so they are unaffected by tax rate or tax credit changes.

New Democratic Party: Change in federal tax rate for high income earners from 33% to 35% – 32630416

The NDP proposed increasing the tax rate of the top tax bracket from 33% to 35%. (In SPSD/M, this is the FTX parameter.) Since the top tax bracket affects so few people, I plotted graphs with the horizontal axis starting at 0.9 (top 10% income) to keep the graphs readable.

Individual Income

There is essentially zero impact to about 98.5% of tax filers because they do not have enough income to reach the top tax bracket.

The PBO analysis of the cost to government includes an effect for an Elasticity of Taxable Income for high-income earners of 0.38. The elasticity is an estimate of how much the reported taxable income changes when the marginal tax rate changes, which tends to decrease the revenue generated when the tax rate is increased. For example, increasing the marginal tax rate might encourage people to work less (leisure is worth relatively more), or increase tax avoidance efforts or tax evasion, all of which reduce the reported taxable income.

I attempted to use the same methodology to account for ETI, but could not get a similar number as the PBO (I got $1279M in calendar year 2023 including ETI, while the PBO analysis expects $964M in fiscal year 2023-24). If anyone knows how the PBO arrived at their estimate, I’d like to know.

I plotted the increase in federal government revenue including ETI in red (the rest of the plots assume the tax base doesn’t change in response to tax rate, or ETI=0). Because the ETI is assumed to be fairly high for high-income earners (the PBO assumes 0.38 for high-income, but 0.10 for low income), and the large amounts of per-person income involved, the ETI effect was not negligible for this policy. Paradoxically, there are some taxpayers (those near the lower bound of the highest tax bracket) where increasing the tax rate actually causes a decrease in tax revenue collected, because the behavioural effect (reducing taxable income) exceeds the amount generated by the higher tax rate (+2% on income above $228,200).

Family Income

The trends are the same: The burden of the tax increase is almost entirely carried by the top few percent of families, with almost no effect on everyone else.

People’s Party of Canada: Cutting Income Taxes and Abolish Personal Capital Gains Tax

The People’s Party proposes several major changes to the income tax system:

  • Increase the basic personal amount to $15000 (SPSD/M parameter BXM)
  • Decrease tax rates to 15% for the first $100,000 of income, and 25% for the rest (SPSD/M parameter FTX)
  • Phasing out the personal capital gains tax (SPSD/M parameter CAPGIR=0)

This tax policy can be seen as approximately the opposite of the NDP proposal. It proposes a large government expense that mostly benefits the highest income few percent of the population.

Individual Income

The total cost of the proposal (neglecting ETI) is about $38 billion in 2023, or about 7× bigger than the Liberal or Conservative tax cuts. Unlike all of the other proposals, the benefits overwhelmingly go to the highest-income individuals: Half the benefit goes to the top 6% of individuals.

Most of the cost ($31B) is due to the income tax cut. Eliminating the capital gains tax costs another $7B on top of that. Eliminating the capital gains tax on its own would have costed $13.1B without the income tax cut, because the income tax cut reduces the value of also removing the capital gains tax.

As one might expect, capital gains tax is mainly paid by those with a large amount of capital to invest. About 40% of the benefit of eliminating the capital gains tax goes to the top 2% of high-income individuals.

Family Income

The story is the same when looking at family income per person instead of individual income. About half the benefit of the tax cuts go to the top 7% highest-income families.

Green Party: Raising the Inclusion Rate of Capital Gains for taxation – 32631069

The Green Party’s proposals that affect personal income tax are too complicated for me to model. See the PBO’s estimate for a description of the policy proposal. The PBO’s model isn’t simple, contains some questionable assumptions, and the PBO thinks their estimate has a “high uncertainty” as a result.

Income Distribution

All of the above graphs used persons sorted by income on the horizontal axis, rather than income. How does this map to income? The following two graphs show the pre-tax income distribution (in 2023), with the first one sorted by individual income (excluding non-tax filers) and the second sorted by family income per person. (The SPSD/M variable is imictot.)

The median tax-filing individual has about $46,000 of pre-tax income, while the median family has about $36,000 of pre-tax income per person.


I think a moderate decrease in income inequality would be a good thing. Thus, I prefer tax policy changes that don’t predominantly benefit high-income individuals.

Both the Conservative and Liberal proposals provide a moderate tax cut for the high-income 40%–50% of the population. While these may not reduce income inequality, I think it is partly forgivable. It’s hard to design a tax cut (which attracts votes) that also targets low-income individuals (who already pay little tax and aren’t affected by income tax changes).

In contrast, the People’s Party’s large tax cuts are clearly aimed at increasing the income of a small fraction of high-income individuals.

The NDP’s proposal to increase the tax rate of the highest tax bracket does move in the direction of reducing income inequality. But the increased tax burden so narrowly targets the top 0.1%–0.5% of income earners that it risks being seen as punitive, rather than useful.

Leave a Reply

You can use these HTML tags

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>