Hours Are Not Dollars
Almost none of ALTO’s $49.5 billion is money. It is time — and a saved hour cannot service a loan. Here is what that figure actually is, how it was built, and why it says nothing about who pays for the railway.
In August 2026 ALTO published Canada’s Moment: The Economic Opportunity of High-Speed Rail, reporting $49.5 billion in benefits against a construction cost of $60 to $90 billion. Those benefits are not money in a bank account. They are mostly hours — time that travellers would have spent on the road or at an airport — stretched over sixty years and converted into today’s dollars.1
The tool that does the converting is called a discount rate. ALTO uses 3.5 per cent a year. Change that one number and the headline changes by tens of billions, without a single train or passenger changing.
The $49.5 billion is a measure of worth, not of funds. The tool that produces it, a discount rate, answers the question is this worth doing? It does not answer the question who pays, and how? Those are separate ledgers, and ALTO’s report is detailed on the first and thin on the second. This explainer sets out what the rate does, shows the arithmetic openly, and then follows the money to the place the appraisal never goes: the difference between what it costs the government to borrow and what a private partner needs to earn.
One finding runs against the grain and is stated here first. Two of the adjustments ALTO leaves out would have made its benefit figure larger, not smaller. The problem is not that the number is tilted. The problem is that a reader is given one number, no range, and no way to know that any of these choices were made.
What a discount rate is, in ordinary words
Ask yourself a simple question. Would you rather have $100 today, or $100 in forty years? Almost everyone takes it today. The money is useful now, the future is uncertain, and by 2066 we will probably all be somewhat better off anyway, so $100 will matter a little less to us then than it does now.
Economists turn that instinct into a percentage. A discount rate shrinks future amounts back to what they are worth to us today, by a fixed amount each year. At 3.5 per cent, a benefit arriving sixty years from now counts for about 13 cents on the dollar. At 8 per cent, the same benefit counts for about one cent.
That is the whole mechanism. It sounds technical and it is arithmetically simple. But it matters enormously for a railway, because of when the money and the benefits arrive.
The timing is what makes this decisive. Construction money is spent early — from 2029 through the early 2040s — so it is barely shrunk at all. The benefits arrive later and keep arriving for sixty years, so they are shrunk heavily. Anything that changes the rate therefore hits the benefit side hard and the cost side hardly at all. A project’s whole case can move from comfortable to marginal without anything physical changing.
How much the answer moves
The table below is the Initiative’s own arithmetic, not a re-run of ALTO’s model. It takes a steady stream of benefits running for sixty years, beginning fifteen years from now, and asks what that stream is worth in today’s dollars at different rates. The last column simply scales ALTO’s published $49.5 billion by the same proportion, to show the size of the swing.
| Discount rate used | Value of the stream | Relative to 3.5% | $49.5B scaled |
|---|---|---|---|
| 2.5 per cent | 21.34 | 1.43× | $71.0B |
| 3.0 per cent | 17.76 | 1.19× | $59.1B |
| 3.5 per cent — ALTO’s rate | 14.89 | 1.00× | $49.5B |
| UK declining schedule | 15.61 | 1.05× | $51.9B |
| 5 per cent | 9.11 | 0.61× | $30.3B |
| 7 per cent | 5.09 | 0.34× | $16.9B |
| 8 per cent | 3.90 | 0.26× | $13.0B |
Assumptions, stated so the arithmetic can be checked: a level benefit stream of one dollar per year, sixty years of operation beginning in year 16, discounted back to a year-zero base. ALTO’s real benefit stream ramps up rather than running level, so the exact figures would differ; the proportions are what matter here. The scaled column is illustrative and is not ALTO’s number at those rates.
Read the middle rows first. At 8 per cent, the same railway carrying the same passengers saving the same hours produces a benefit figure roughly a quarter the size. At 2.5 per cent it produces one roughly forty per cent larger. Nothing about the trains changed. Only the parameter changed.
This is why appraisal manuals require the calculation to be repeated at more than one rate and the results published as a range. It is not a bureaucratic formality. It is the only way a reader can tell whether a case is robust or whether it depends on a parameter choice.
A number with a family tree
ALTO’s report attributes its rate to one source: the Business Case Manual Volume 2: Guidance, published by Metrolinx, the Government of Ontario’s transit agency for the Toronto and Hamilton region.2 That manual sets a social discount rate of 3.5 per cent, alongside an evaluation period of five to sixty years.
The 3.5 per cent figure is not original to Metrolinx. It is the rate used by HM Treasury in the United Kingdom, and the Treasury publishes exactly how it was assembled. Three judgements are added together:
0.5 per cent for simple impatience. People prefer good things sooner. Half a percentage point is the allowance for that.
1.0 per cent for the risk that the future does not arrive as expected. Wars, pandemics, collapses. A benefit promised in 2080 might never materialise, so it is discounted a little further.
2.0 per cent because people in the future will be richer. If incomes rise about 2 per cent a year, our grandchildren will be considerably better off than we are, and an extra dollar will matter less to them than it does to us. This is the largest of the three, and the most contestable.
Those three add to 3.5.3
Notice what the rate is not. It is not a market price, an interest rate, or anything anyone can look up. It is a set of judgements about how much weight to give people who are not yet born — and every one of the three is disputed by serious people. That is not a criticism of the figure. It is the reason a serious appraisal shows what happens when the figure moves.
HM Treasury does not apply 3.5 per cent forever. The rate steps down to 3.0 per cent for years 31 to 75, and 2.5 per cent thereafter,3 and the Treasury’s supplementary guidance instructs practitioners in the same terms: the standard 3.5 per cent for years 1 to 30, and 3.0 per cent for years 31 to 75.4 The reason is uncertainty: the further out you look, the less confident anyone can be in the parameters, and the lower the rate should be.
ALTO discounts a sixty-year stream at a flat 3.5 per cent throughout. Applying the stepped-down schedule instead would have made ALTO’s benefit total about five per cent larger, as the fourth row of the table above shows. This is a conservatism in ALTO’s favour, and it should be credited as one. It is recorded here because a reader assessing where a federal appraisal input came from deserves the whole picture, including the parts that do not fit a critical narrative.
What the federal government uses, and the gap where a manual should be
Canada has approached the same question from the opposite end, and it is worth understanding the difference, because it produces a far higher number.
The British method asks a question about values: how much should we care about the future? The Canadian method asks a question about alternatives: what else could this money have done? If public money invested elsewhere in the economy would have earned, say, 8 per cent, then a project has to clear that bar to be worth funding — otherwise the country was better off doing the other thing. That is what economists mean by the opportunity cost of capital.
Neither question is wrong. They are simply different questions, and the second one produces a much tougher test than the first.
The Treasury Board’s 2007 guidance identified 8 per cent as the appropriate rate, with sensitivity tests at 3 and 10 per cent, on that opportunity-cost basis.5 The current federal Policy on Cost-Benefit Analysis still directs departments to use the opportunity cost of capital as the discount rate, permitting a social rate only in defined cases — including where impacts run fifty years or more — and requiring that even when a social rate is used, results using the opportunity cost of capital must also be reported.6
Two honest qualifications belong here, and neither is small. First, that federal policy governs regulations, not capital projects, so it does not bind ALTO. Second, the current edition of the Treasury Board guide is no longer published on canada.ca and is available only through an internal government wiki page,7 so the Initiative has not been able to verify the figure it now specifies.
And the federal manual for transport projects specifically? Transport Canada’s guide to benefit-cost analysis dates from 1994.8 Thirty-two years later, there is no current, public federal appraisal manual for a project of this kind. That absence is very likely why a national railway is being appraised using a provincial transit agency’s parameters — and it is a finding about the machinery of government rather than about ALTO.
The most cited Canadian argument for a 3.5 per cent rate comes from the economists Boardman, Moore and Vining, who reject the 8 per cent approach. So there is a respectable Canadian case for ALTO’s rate. But it is a conditional case, and the conditions are specific.
Condition one: the project runs under fifty years. Beyond that, they recommend a rate that steps down over time, for the same reason the UK Treasury does — nobody can see that far ahead with confidence.
Condition two: the project must not pull money away from private investment. The money for a public project comes from taxes or borrowing, and it would otherwise have been used by someone else. Some of it would have been spent, and some would have been invested — a business expansion, new equipment, a factory. Those two are not equivalent. A dollar diverted from someone’s spending costs the economy that one dollar. A dollar diverted from investment costs more, because that investment would have gone on producing returns for years afterwards.
And if the project does pull money from investment, there is a fix. Rather than argue about the rate all over again, you take the portion of the cost that displaced private investment and mark it up by 26 per cent before putting it in the calculation — because that is roughly what the lost investment was worth to the economy over time. Economists call the 1.26 multiplier a shadow price of capital. It is simply a way of using a generous discount rate honestly, instead of using it to pretend the money was free.5
ALTO’s appraisal period is sixty years, which fails the first condition outright. Whether a $60 to $90 billion draw on Canadian capital displaces private investment is a real question, not a technicality — and the mark-up would apply only to the share that does, not to the whole sum. Neither condition is mentioned in the report.
The pattern is the one the companion audit It Left the Rules Behind describes: a number travels, and the conditions attached to it stay behind.
The last time anyone published these numbers for this corridor
ALTO’s stated reason for publishing no benefit-cost ratio is that the cost estimate is not yet mature enough to support one. It is worth knowing that a predecessor project on the same corridor did publish one, at a comparable stage, and published the funding ledger alongside it.
In December 2021 the Joint Project Office — a body formed by VIA Rail and the Canada Infrastructure Bank — completed a Business Case Update for High Frequency Rail, the slower, cheaper predecessor to ALTO between Toronto and Québec City. It was released through access to information by the Canada Infrastructure Bank in November 2025.16
| What the 2021 business case published | Figure |
|---|---|
| Capital cost, with electrification (2020 prices) | $27.71B |
| Projected revenue over 30 years (2019 prices) | $33.7B |
| Operations, maintenance and rehabilitation, 30 years | $32.5B |
| Net present value over 30 years | −$21.1B |
| Benefit-cost ratio | ~0.13 |
| Expanded benefit-cost ratio | ~0.4 |
| Public subsidy over 30 years, by delivery model | $37.1B to $42.2B |
Source: Joint Project Office, High Frequency Rail Business Case Update V.002, 10 December 2021. Ratios at Table 14, page 43; capital, revenue, lifecycle and net present value figures in the executive summary, pages 7 and 8; subsidy comparison at Table 4, page 8.
A benefit-cost ratio of 0.13 means about thirteen cents of measured benefit for every dollar of cost. The wider figure of 0.4 is what the same table calls an expanded ratio, and the difference between the two is worth understanding, because it is the larger of the two numbers.
The expansion adds two items. One is agglomeration — the economic gain from businesses being better connected — worth $0.3 to $0.9 billion. The other, worth $5.6 to $7.6 billion, is a resource correction: the fares new passengers would pay, counted as a benefit because they arrive as revenue for the operator. That single item is larger than the journey time savings and all the external benefits put together. The business case itself notes that both are relatively new to Canadian economic appraisal, which is why it reports the ratio with and without them.
The 2021 economic case states where its parameters came from: the social discount rate, the value of time and the value of external impacts were taken from a combination of Metrolinx and Ministère des Transports du Québec guidance. Those are the same two sources ALTO cites in 2026.
So the identical parameter lineage, applied to a $27.71 billion version of this corridor, produced a published ratio of 0.13. Five years later, on a project costing two to three times as much, the same two sources are cited and no ratio is published at all.
The 2021 document also shows how the distinction this page has been drawing is meant to work in practice. Its net present value calculation used a discount rate of 2.5 per cent, sourced explicitly to the ten-year average of the 30-year Government of Canada benchmark bond — a financing rate, taken from what the government actually pays to borrow. Its economic case used the social parameters from Metrolinx and MTQ. Two questions, two rates, both disclosed, in a single document.
Three cautions, stated plainly. High Frequency Rail is not ALTO: different technology, different speed, a $27.71 billion cost rather than $60 to $90 billion, and a thirty-year evaluation rather than sixty. The JPO described its own results as preliminary. And none of these figures transfer to ALTO by arithmetic. What the document establishes is narrower and harder to set aside: a benefit-cost ratio can be produced for a project on this corridor at this stage of development, because one was.
One further point belongs on the record. The identical document was also released under a separate access request, and in that version the whole net present value section, the capital cost figure, the revenue figure and both ratios were blacked out — along with the subsection titles of the Economic Case within the table of contents.17 Same document, same date, two institutions, opposite outcomes.
Three different rates, and only one of them is in the report
Here is the heart of it. People use the phrase “the discount rate” for three quite different things, and conflating them is how an appraisal result gets mistaken for a financing plan.
1. The appraisal rate — 3.5 per cent
Used to decide whether a project is worth doing. No money moves because of it. It turns hours saved and collisions avoided into a single present-day figure so they can be compared with the cost. Nobody charges it, nobody pays it, and no bank uses it.
2. What it costs the government to borrow
Real money, actually paid. When the federal government borrows for thirty years it has been paying in the region of 3.7 to 3.9 per cent during 2026. Take off inflation, which the Bank of Canada aims to hold at 2 per cent, and the true cost of the money is roughly two per cent a year.9 If the state simply builds the railway and holds it, this is what the borrowing actually costs, and it is lower than the appraisal rate.
3. What a private partner needs to earn
Considerably more. If a pension fund or infrastructure investor builds the railway, it is putting its own money at risk — the risk that construction costs more than planned, or that too few people ride. It requires a return for carrying that risk, and that return is paid out every year for decades. This is the rate that decides what the public actually hands over, and it appears nowhere in ALTO’s economic report.
The gap between the second and the third is the entire public-private question. If the government borrows at 2 per cent and builds the railway itself, that is what the money costs. If a private partner builds it instead and needs 8 per cent, someone has to make up the difference — every year, for as long as the arrangement lasts. That someone is the public.
So a project can pass the 3.5 per cent test comfortably and still require very large annual public payments to get built. The appraisal will go on saying “worth doing.” It will never say who writes the cheque, for how long, or at what return.
This is the point most easily missed, and it is not a technicality. Nearly all of ALTO’s benefit figure is not cash. It is hours of travel time, collisions that did not happen, tonnes of emissions avoided. These are real and they matter. But a saved hour cannot service a loan, meet a payroll, or renew a worn rail.
The money that actually funds a railway comes from two places only: fares, and government payments. ALTO’s report handles that second ledger in a few pages, supported chiefly by the operating margins of three foreign railways, and it publishes no fare, no revenue figure and no farebox recovery ratio. So the document is expansive about whether the project is worth doing and close to silent about how it would be paid for.
How the Montréal REM is actually funded
The Réseau express métropolitain is a 67-kilometre automated light metro in Greater Montréal, built, owned and operated by CDPQ Infra, a subsidiary of the Québec pension fund manager. It is the clearest Canadian illustration of what the third rate looks like once it becomes money, and its terms are public.10
That pair of numbers is the whole point of this section, made concrete. The same railway, the same track, the same passengers — and two participants requiring returns that differ by more than double. The difference is not a rounding error in an appraisal. It is paid out, in cash, on every passenger-kilometre, for as long as the agreement runs.
Why this case and not another. The REM is not an analogy picked at random. CDPQ Infra leads Cadence, the consortium selected in February 2025 as ALTO’s private development partner. AtkinsRéalis — formerly SNC-Lavalin, a member of the group that built the REM and, with Alstom, of the group that supplies and operates its trains — is also a Cadence member. The other Cadence members are SYSTRA Canada, Keolis Canada, SNCF Voyageurs and Air Canada.18 The REM is the lead sponsor’s own model, which CDPQ Infra presents publicly as an innovative approach to delivering public infrastructure. That is what makes it the most informative available guide to how a private partner’s return might be priced here.
An important caution. ALTO is nonetheless not the REM, and this is not a prediction. Canada is to retain permanent ownership of the ALTO network, which was never the REM arrangement; the project is in a co-development phase running to 2029; and no payment mechanism has been disclosed. Cadence is a different group with different members and a different contract. The REM is offered as the one Canadian case where the arithmetic of a private partner’s return has been made public — which is exactly what has not yet happened for a project several times its size.
Notice what a payment mechanism does with risk. Because CDPQ Infra is paid per passenger-kilometre, a shortfall in riders is a shortfall in its own revenue — the investor carries the demand risk, which is why it negotiated a clause barring competing services. Under a different structure, where the public pays for the railway simply being available, a shortfall in riders changes nothing the partner receives and everything the public pays.
Same railway, same disappointing ridership, opposite consequences. Which of those applies to ALTO has not been published.
What this does not claim
On the rate
On the comparisons
Two questions, answerable without releasing a model
1. Who absorbs it if the passengers do not come?
Not a forecasting question but a contract question. If a partner is paid per passenger, a shortfall reduces its return. If it is paid for availability, a shortfall costs the partner nothing and the public a great deal. Identical ridership, opposite outcomes — and ALTO has published neither the mechanism nor the cost of capital behind it.
2. What fare, and what revenue?
No fare level, average yield or farebox recovery ratio appears in eighty-three pages. Without one, the funding question cannot be examined by anyone outside the project.
Neither requires access to ALTO’s models, cooperation from its staff, or agreement about what the correct discount rate for a national railway ought to be. Both are answerable from work already done.
A third question — whether the calculation was ever run at any rate other than 3.5 per cent — belongs to the companion audit It Left the Rules Behind, which sets out the full list of tests the cited manual requires at this project’s scale and which of them appear in the report.