Alberta argues about the price of electricity using a number that follows a third of its households

1978–2026 · the series

First, the number everybody is arguing about

The consumer price index for electricityCPI electricity index. Statistics Canada's price index for electricity in Alberta, part of the consumer price index and published monthly since September 1978. It is an index, not a price: it is set to 100 in 2002 and every other month is a ratio to that, so it can tell you how much something changed and never what it costs. in Alberta has been published every month since September 1978. It is set to 100 in 2002, and in July 2026 it stood at 216.2. It is an index and not a price: it says how far something moved, never what anything costs.

For most of its life it behaved like a utility price index anywhere — a slow climb with occasional steps. From 2021 it stopped behaving that way. It reached 372.9 in August 2023, its highest reading on record. Thirteen months before that it had fallen 45.6% in a single month, in January 2023 — the largest one-month decline in the whole series. Then, in January 2025, it went nearly flat and has stayed there.

How much has it risen? The honest answer is that it depends entirely on which measure you use, by more than a factor of two, so here are all three. Against its 2019 annual meanAnnual mean. The average of a year's twelve monthly readings. For a series as volatile as this one, a comparison between annual means and a comparison between individual months can differ by a factor of two, so this page states which it is using every time., the index in 2025 — the last complete calendar year — is ×1.45. At its highest annual mean, in 2023, it is ×1.69. Only at the single monthly peak of August 2023 does it reach ×2.53. The claim that Alberta electricity prices “roughly doubled” is true of one month and of no year, and this page uses annual means and says so every time.

The two marked months are the subject of two later sections: a price ceilingPrice ceiling. A legal maximum on what may be charged. Alberta capped the regulated rate's energy charge at 13.5 cents per kilowatt hour for January, February and March of 2023. By the terms of the announcement it applied to regulated-rate customers and to nobody else. that by its own terms applied to regulated-rate customers and nobody else, and the month the regulated rate stopped being reset monthly. Both are dates in policy, not in the weather or the market.

Nothing here is a wholesale price. Alberta's hourly wholesale pool priceWholesale pool price. The price generators are paid in Alberta's wholesale electricity market, set every hour. It is published only by the Alberta Electric System Operator, under a licence permitting internal and non-commercial use, so it is absent from this page entirely and nothing here may be read as a substitute for it. is published by the Alberta Electric System Operator under a licence permitting internal and non-commercial use only, so it is absent from this page by decision and has no substitute here. Every price on this page is retail.

The question

An index of electricity prices has to be an index of somebody's electricity price

Alberta deregulated electricity retailing, which means an Alberta household's electricity price depends on a decision it made, or didn't. Sign a fixed competitive contractCompetitive contract. An electricity supply agreement with a private retailer rather than the regulated default — usually a fixed rate held for a term of one to five years, sometimes a floating rate that tracks the market. A household on a fixed contract sees no price change at all between renewals. and your rate is the same this month as last month and will be until it renews. Sign nothing and you are on the regulated default rateRegulated Rate Option. Alberta's regulated default electricity rate — what a household paid if it never signed a contract with a competitive retailer. Until the end of 2024 it was reset every month, which is why it moved so much and so fast., which until the end of 2024 was recalculated every single month.

Those two populations do not have one price between them. They barely have the same kind of price: one is a step function that moves at renewal, the other moved twelve times a year. So a single index for “electricity, Alberta” is doing something specific, and which of them it is doing determines what the inflation figure means.

The population on the regulated rate has shrunk considerably over the period this page covers. This is the part of the argument where a page like this usually asserts a tidy percentage, and an earlier draft of this one could not: the figures in circulation were reported at scattered dates, and the most-quoted were a news organisation's paraphrase of the regulator's data. The regulator's own series has since been read directly. The Market Surveillance Administrator publishes its retail statistics monthly, back to January 2012, and one definition governs every share on this page: the MSA counts active residential sites, and its notes say plainly that a site corresponds to a meter. A site is not a household — an apartment building behind one meter is one site, a second property is a second one — though for a rough sense of scale the two are close. With that said, here is the series, read from the workbook itself:

• ·58.1% of residential sites on the regulated rate (March 2014) — Market Surveillance Administrator, MSA Retail Statistics workbook. source
• ·Calgary 42%, Edmonton 72%; “nearly 60%” province-wide and “persistently declining” (March 2014, published that November) — Market Surveillance Administrator, State of the Market 2014. source
• ·45.0% of sites; averaged across 2021, 43.6% (January 2021) — Market Surveillance Administrator, MSA Retail Statistics workbook. source
• ·37.4% of sites — and 34.9% of residential consumption, a different measure that travelled widely as “35%” (September 2022) — Market Surveillance Administrator, MSA Retail Statistics workbook. source
• ·31.5% of sites averaged across the year, falling from 35.7% in January to 26.4% by December (2023) — Market Surveillance Administrator, MSA Retail Statistics workbook. source
• ·21.6% of sites averaged across the year, now on the Rate of Last Resort (2025) — Market Surveillance Administrator, MSA Retail Statistics workbook. source

A clear majority of residential sites on the regulated rate in 2014, about a third averaged across 2023, about a fifth by 2025. A default that erodes. Two figures that circulate widely — a 2021 share of about 45% and a 2025 share of about 26% — were flagged in an earlier draft of this page as untraceable. Review traced one and killed the other, and it is worth recording both outcomes. The 45% is real: the regulator's workbook shows 45.0% of residential sites on the default rate in January 2021, and it now sits in the list above with its source. The 2025 figure remains unsupported — no month of 2025 comes near 26%, the annual mean is 21.6%, and the nearest 26.4% reading in the whole series is December 2023 — so it stays off this page. The widely-quoted “35% of households” is a third case: real number, wrong population. It was the September 2022 consumption share — a share of energy volume — relabelled a share of households as it travelled.

2018–2024 · the falsifiable test

So we went and got the rates, and checked

This is answerable rather than arguable. If the index tracks the regulated rate, then month-to-month changes in the two should agree; if it tracks a broad mix of households mostly on fixed contracts, they should not, because most of that population's price does not change in a given month at all.

So we assembled 130 monthly regulated energy chargesEnergy charge. The commodity part of an electricity bill: the electricity itself, in cents per kilowatt hour. It sits alongside distribution, transmission, riders, administration and tax, which are billed separately and by different rules. Every regulated-rate figure on this page is the energy charge alone, and none of them is a bill. from the utilities that published them: 75 months of Direct Energy Regulated Services in the ATCO Electric service area, continuous from October 2018, and 55 months of ENMAX in Calgary, continuous from December 2021 to November 2024 with older figures from individual news releases. Every figure is the energy charge alone — no distribution, no transmission, no riders, no administration, no tax — and every one was read from a document we fetched. 12 sources, all listed in the method section.

Because an index and a price in cents share no scale, the comparison is between month-over-month changesMonth-over-month change. The percentage change from one month to the month immediately before it. Comparing two series as rates of change rather than as levels is what makes it legitimate to put an index and a price in the same sentence — they have no common scale, but their movements do. in both. On the 62 clean months in the Direct Energy Regulated Services territory the correlationCorrelation coefficient. A number between −1 and 1 measuring how closely two series move together: 1 is perfect agreement, 0 is none. It measures co-movement and nothing else — two series can correlate because one drives the other, because something drives both, or by accident. It is evidence, not proof. is 0.938. On the 23 clean months in Calgary — a different utility, a different service area, an independently sourced series — it is 0.955. The index moved in the same direction as the rate in 51 of 62 months in one territory and 22 of 23 in the other.

The triangles are not an inconvenience, they are a check. They are the 24 months touching July 2022–May 2023, when a provincial rebateElectricity rebate. A credit applied to a bill. Alberta's 2022–23 rebates went to all consumers, on the regulated rate and on competitive contracts alike — which is what distinguishes them from the price ceiling, and why the two cannot be treated as one intervention., a price ceilingPrice ceiling. A legal maximum on what may be charged. Alberta capped the regulated rate's energy charge at 13.5 cents per kilowatt hour for January, February and March of 2023. By the terms of the announcement it applied to regulated-rate customers and to nobody else., or both were changing what consumers paid without changing the underlying rate. Across all months including those the correlation falls to 0.638 and 0.622. That it breaks down exactly where a known third variable was moving, and is near-perfect everywhere else, is what you would predict if the index prices the regulated rate as it reached the customer — and it is the reason those months are drawn rather than dropped.

An index based in 2002 and a price in cents per kilowatt hour cannot share an axis as published. Both series here are divided by their own value in October 2018 and multiplied by 100, which is the only operation that puts them on one scale without asserting anything about their units. The vertical distance between the lines is a consequence of that choice and means nothing; their shape is the data.

Nothing here is a wholesale price. Alberta's hourly wholesale pool priceWholesale pool price. The price generators are paid in Alberta's wholesale electricity market, set every hour. It is published only by the Alberta Electric System Operator, under a licence permitting internal and non-commercial use, so it is absent from this page entirely and nothing here may be read as a substitute for it. is published by the Alberta Electric System Operator under a licence permitting internal and non-commercial use only, so it is absent from this page by decision and has no substitute here. Every price on this page is retail.

Falsification

A correlation would not be enough. Neither, on its own, is the slope

Correlation alone would not settle this, and it is worth saying why: a household on a floating competitive contract also sees its price move with the market, so an index covering everybody would still show some co-movement with the regulated rate. The question splits in two — how much of the index's weight sits on prices that move at all in a given month, which the slope can answer, and whose prices those are, which it cannot. This section does both, and is explicit about which tool does which job.

There is a blunter way to show that, and it is worth showing against ourselves. Statistics Canada's selling price index for non-residential customers under 5000 kW has no regulated-rate population in it whatsoever. Fitted on the same 62 clean months, it tracks the household index at 0.958 — higher than the regulated rate itself manages. Every electricity series in Alberta reprices off the same monthly market, so a tight correlation establishes only that the index follows something that moves every month. It cannot establish whose price that is.

Neither can the slope, on its own — an earlier draft of this page claimed it could, so this is a correction as much as a caveat. The household index moves 0.70 for each point that non-residential index moves, against 0.50 for the regulated rate, and the gap looks like it separates the two. It does not. A fitted slope is co-movement times relative size, and this gap is what you get when one series simply swings harder than the other: the regulated rate is an energy charge with no wires or fees diluting it, so it moves further each month than an all-in selling price index does. Run that through: if the household index tracks the regulated rate, its fitted slope on the non-residential index is forced to about 0.72 by that difference in swing alone — and the observed value is 0.70. The placebo's slope lands where the regulated-rate reading itself predicts it. It rules nothing in and nothing out.

Here is what the slopeRegression slope. How far one series moves, on average, for a one-unit move in another. Here: the percentage the index moves for each one per cent the regulated rate moves. Unlike a correlation, a slope has a magnitude that can be checked against what is physically possible. does establish. Across both territories it is about 0.50: the index moves roughly half a per cent for every one per cent the regulated energy charge moves. That number is a product of two things — the share of the index's weight sitting on prices that reprice every month, and the energy charge's share of a bill, since the rest of a bill is wires and fees that did not move. Fix the slope at what was measured, and each assumption about the first implies a value for the second.

The middle row is the one the arithmetic kills. An index covering every residential customer, with the competitive majority's prices sitting still between renewals, could not move 0.50% for each 1% move in the regulated rate; it would move about 0.18%. So the slope forces roughly four-fifths of the index's weight onto prices that reprice monthly — a fixed-contract-dominated index is excluded. The bottom row is the one the arithmetic cannot kill. On the slope alone, an index following the regulated rate and an index following floating competitive plans are indistinguishable: both reprice every month, off the same market.

What separates them is scope — two occasions when the law moved one population's price and not the other's, each treated at length below. In January 2023 a price ceiling that by statute covered only regulated-rate customers took effect. The household index fell 45.6% — its record — in the same month both zero-regulated-rate selling price indexes rose, 23.7% and 28.5%. An index weighted toward floating plans follows the market, and the market went up. Then in January 2025 the regulated rate was fixed for two years, and the household index's monthly swings collapsed 12-fold while the market-facing index settled only 2.8-fold. If floating residential bills moved even as much as that index did, floating plans can carry at most about a quarter of the household index's weight — and full residential coverage would then need the energy charge to be 0.80 of a bill, which no Alberta bill resembles. The slope narrows the suspects to prices that move monthly; the two events name which one.

Where this argument stops. The energy-share figure the arithmetic turns on is derived from our own regression, not read off a published tariff breakdown — a utility document giving the energy share of a residential bill would sharpen every bound above, and we have not found one. And bounds are what these are: the slope caps how much of the index can sit on prices that hold still, the 2025 break caps how much can sit on floating plans, and the ceiling month requires the regulated rate to carry most of what remains. Together they establish that the regulated rate dominates the index — not an exact percentage, and not that nothing else is in it.

2025-01 · the natural experiment

Twice the rate stopped moving. Twice the index stopped with it

The strongest evidence here is not statistical. It is two occasions when the regulated rate stopped moving for reasons that had nothing to do with the index, and the index stopped too.

The first is small, and weaker than we once made it. In 8 months of the Direct Energy Regulated Services series the energy charge printed exactly the same figure two months running — the rate did not move at all. In those months the index moved an average of 0.8%, against 8.2% in the other 66. A 9.8-fold damping. But honesty requires the context: the whole market was calm in those months — the non-residential selling price index moved an average of only 1.6% in the same months — so an index weighted toward floating plans would also have printed small changes. This test is consistent with the finding, not proof of it. The proof is the two dates that follow.

The second is the whole character of the series. On 1 January 2025 the monthly-reset Regulated Rate Option was replaced by a Rate of Last ResortRate of Last Resort. The regulated default rate that replaced the Regulated Rate Option on 1 January 2025. The instrument that created it requires it to be set for a two-year term rather than recalculated monthly, so it does not move within a term.. The instrument that did it is explicit about the change:

“An owner must set each Rate of Last Resort for a term of 2 years (a) for the first 2-year term, beginning on January 1, 2025 and ending on December 31, 2026, and (b) for each subsequent 2-year term, beginning on January 1 of the year following the end of the previous term.”

The same regulation describes what it replaced in its own words, as “the transition from a monthly regulated rate setting plan to a 2-year fixed regulated rate setting plan”. The published rate has been flat ever since: 12.01¢ per kilowatt hour in the Edmonton and Fortis territories and 12.02¢ in the ATCO Electric service area, every month through September 2026.

And the index: through 36 months of monthly resets from 2022, its monthly changes had a standard deviation of 16.46% and averaged 10.76% in absolute size. Over the 19 months since January 2025 those figures are 1.35% and 0.80% — a 12-fold collapse in volatility, beginning on that exact boundary. Some of that quiet is genuinely the market's: new generation came online through 2024 and 2025 and Alberta's wholesale market settled. The page's own control measures how much that can explain — the under-5000 kW selling price index, exposed to the same calmer market and to no regulated rate, settled 2.8-fold across the same two windows. A quieter market accounts for a part of the calm; it does not account for a household index that went quiet 12-fold on the statutory date while the market-facing series kept moving.

A price index for a population mostly on competitive contracts has no particular reason to change character on 1 January 2025. Competitive contracts did not all renew that month, and the market calm that did arrive came to every Alberta electricity series — only the household index went flat. What changed on that date was the rule for setting the default rate — and this page's reading of it was written down as a prediction, from the shape of the break alone, before the instrument that caused it had been found.

2023-01 · an intervention with a known scope

A measure that covered only one kind of customer moved the index by a record amount

In the winter of 2022–23 Alberta did two things to electricity bills, and the difference between them is the point. The rebate went to everybody:

“the $50 rebates will be on all eligible bills – protecting consumers on the variable regulated rate option as well as those on competitive contracts.”

The price ceiling did not:

“Consumers who purchase power through the regulated rate option (RRO) will benefit from temporary price relief through a 13.5 cent per kilowatt hour ceiling in January, February and March 2023.”

The 13.5¢ ceiling applied to regulated-rate customers and, by the government's own description, to nobody else. Into January 2023 the Direct Energy Regulated Services energy charge fell 38.0% to meet it. The index fell 45.6% — the largest monthly decline in its recorded history. And in that same month, Statistics Canada's two selling price indexes for non-residential customers — series with no regulated-rate population in them at all — rose, 23.7% and 28.5%. The one population the ceiling covered saw its index fall by nearly half; the populations it did not cover saw prices climb by a quarter. A floating competitive bill follows the market, and the market went the other way — which is why this month, more than any statistic, is what identifies whose price the index carries. It then printed 119.2 in January and 119.2 again in February: bit-identical, in the middle of the most volatile stretch the series has ever had, while the ceiling held the regulated rate flat.

This month is not a clean experiment, and we would rather say so. The rebate rose from $50 to $75 in the same month and applied to everyone, so two things moved at once. The cleaner signal is the pair of identical prints and the flat rate behind them. March is imperfect too: the index jumps while the ceiling nominally still held, and the obvious candidate is the credit stepping down from $75 to $25 — but we have not decomposed it and are not going to claim we have. What survives all of that is the scope distinction: an intervention restricted to regulated-rate customers coincides with the record move, against the market's direction. Worked through ordinary bill scenarios — even letting the rebate's step from $50 to $75 do as much of the work as it can — a fall of that size in a month the rest of the market rose a quarter requires somewhere between 57% and all of the index's weight on the regulated rate, and around 85% under middling assumptions. Those two bounds are scenario arithmetic rather than a measurement — the method section labels them as the hand-carried figures they are — but no scenario we could construct gets a one-third weight anywhere near the observed fall.

An audited null result

Statistics Canada does not say, in either direction

The obvious way to settle this would be to read the methodology. We did. All eleven chapters of the Consumer Price Index Reference Paper were swept mechanically, in both the 2023 and the 2019 edition, along with the survey record and the collection methodology pages. There are three occurrences of “electric” in the entire Reference Paper and none of them is responsive. Searches for “regulated rate”, “rate option”, “electricity retailer” and “supply arrangement” return nothing at all. Nothing Alberta-specific exists.

The nearest thing in existence is an endnote observing that examples of product classes with one seller in a geographical stratum “include some public utilities such as water or electricity companies”. That says electricity may be handled that way. It does not say it always is, or which seller, or anything about Alberta, or anything about default rates against competitive contracts. The hedge is theirs.

So this is a real, audited null result, and it is load-bearing for how this page describes itself. Nothing here is a Statistics Canada admission. We are not reporting that they told us what the index tracks; we are reporting that they have not said, and that we tested it ourselves and got an answer. Anyone who wants to disagree with the answer has the rates, the method and the arithmetic above to disagree with.

We are not the first to notice the gap. A report commissioned by the Alberta Federation of Labour — a union, publishing advocacy, and it should be read as that — made the same inference in October 2024, about the energy components of the Alberta electricity CPI and the Hydro-Québec price series together:

“…the respective energy components for residential users appear to only track changes in the RRO over time but excludes the majority of residential customers that are on contracts. This is an analytical challenge because it does not allow for an accurate assessment of the change in effective prices paid by households over time. This is a data gap that should be closed by the Government of Alberta.”

Note what that quotation is and is not. The hedge “appear to”, and the author going on to call it an analytical challenge and a data gap, mark it as that author's own methodological inference — not something Statistics Canada told him, and not a finding he tested against published rates. It is a second party reaching the same conclusion independently, which is worth something, and it is an advocacy document, which is worth knowing.

Show the gap

The number this page cannot show you at all

Anyone who follows Alberta electricity closely has been waiting for the pool price — the hourly wholesale price that spikes in a cold snap and collapses on a windy afternoon. It is the number the trade press reports and the one most arguments start from. It is not here, and it is not here for a reason worth stating plainly rather than routing around.

The pool price is published by the Alberta Electric System Operator, whose terms of use permit internal and non-commercial use only. Tamrack is a commercial product. So it cannot appear — not as a level, not as a monthly average, not folded inside a composite of our own construction, and not redrawn as a chart of ours. There is no open, commercially reusable substitute, and deriving it from something else would be the same act with more steps. It was not consulted in assembling this page.

That has a consequence the reader should carry into everything above. This page can tell you what households were charged and what the index did; it cannot tell you what electricity cost to produce in any hour, so it cannot separate a rate that rose because the market rose from a rate that rose for any other reason. It does not try to.

2005–2024 · the only figure in money

One number on this page is in dollars, and it is a quotient rather than a price

Everything so far is an index or a rate. There is one series here denominated in money: the implied average residential priceImplied average residential price. Total residential electricity revenue divided by total residential volume — dollars collected over megawatt hours sold. It blends energy, transmission, distribution, riders and administration into a single number, and it shifts when the mix of customers shifts even if no tariff moves. No household pays it., which is all residential electricity revenue divided by all residential volume. It blends energy, transmission, distribution, riders and administration into one figure and it shifts when the customer mix shifts even if no tariff moves. No household pays it. Its one virtue is that it is defined identically for twenty years and is indifferent to which retail plan is fashionable — which makes it the only series here that is not a party to the argument above.

Over the full span, what this series says depends on the method more than on the market. On the form this page quotes everywhere else — the first five years of the record against the last five — it fell 32.2% in constant 2024 dollarsReal and nominal. Nominal is the number as it was billed. Real restates it in one year's dollars so amounts from different years can be compared. Over this page's twenty-year span the residential average moves in opposite directions depending which one is used, and both are correct., and fell 8.8% even in the dollars of the day. The figure that circulates, a rise of 5.8% from 2005 to 2024, is the point-to-point form — two single years of a noisy quotient, the comparison this page declines for the real series and so cannot lean on for the nominal one. It is kept here, labelled, because readers will have met it. Whether “electricity got more expensive in dollars” is true flips with that choice of window, and the flip is the finding. What no method reverses is the relative statement: against a general price level that rose 56.2%, electricity got cheaper.

Why that change is quoted as five years against five years. Because it is a quotient of two reported quantities, a point-to-point figure is a hostage to which single year it starts in. Starting in 2008 the same real decline reads −42.5%; starting in 2010 it reads −4.0% — the same series, the same endpoint, a tenfold difference in the answer. The five-year-mean form survives dropping any individual year, so it is the only form quoted here.

And why the wiggles are not read. Year to year the real series moves a median of 8.8% and as much as 27.8%. The two marked years are reporting artifacts rather than household behaviour: in 2010 the revenue numerator falls 23.4% while the volume it is divided by rises 5.1%, and in 2014 the volume denominator falls 16.7% while revenue falls only 2.3%. The denominator alone travels −21.8% between 2011 and 2014. A series shaped like that should have its level compared across windows and its individual steps left alone.

One thing this series is explicitly not used for here. It is tempting to explain the distance between the household index and this average by who is on the regulated rate. That explanation does not survive contact with the other two series on the page. Between 2005 and 2023 — on annual means — the two non-residential selling price indexes rose 469.7% and 342.8% against the household index's 175.0%. Those indexes have no regulated-rate population in them whatsoever, and they diverge from this average two to three times harder. The odd series is this one, for the reasons above. Its noise is not evidence for the argument in the earlier sections and is not used as any.

Three series, two base years, so all three are rebased to one month and the caption says which. The two selling price indexes cover non-residential customers only — there is no residential member in that table — so this is not three versions of one question. The gap in the larger-customer line across September 2012–December 2013 is 16 months Statistics Canada did not publish, left open rather than bridged.

Nothing here is a wholesale price. Alberta's hourly wholesale pool priceWholesale pool price. The price generators are paid in Alberta's wholesale electricity market, set every hour. It is published only by the Alberta Electric System Operator, under a licence permitting internal and non-commercial use, so it is absent from this page entirely and nothing here may be read as a substitute for it. is published by the Alberta Electric System Operator under a licence permitting internal and non-commercial use only, so it is absent from this page by decision and has no substitute here. Every price on this page is retail.

2005–2026 · the physical system

What Alberta actually generates, and who generates it

Underneath the prices is a fleet. Alberta generated 90.2 TWh in 2024, 41.5% more than in 2005 — and most of that growth did not come from utilities. Utility generation rose 13.6% over those twenty years, supplying 29.4% of the province's added output. Industrial self-generationClass of producer. Who generated the electricity: electric utilities, or industry generating for itself. Alberta's industrial self-generation is mostly oil sands cogeneration and is a large share of provincial output, so a count of utility generation alone materially understates the province., mostly oil sands cogeneration, rose 291%, from 10.1% of provincial output to 27.8%. Any count of Alberta electricity that looks only at utilities now misses more than a quarter of the province.

What that fleet burns is a harder question to answer honestly, because the classification changed underneath it. Statistics Canada recut the type-of-generationType of generation. What the electricity was generated from. Statistics Canada recut this classification in January 2016, from the technology of the machine to the fuel it burns, and the two cuts share no month — so no published series carries coal, or natural gas, continuously across that date. dimension in January 2016, from the technology of the turbine to the fuel it burns. The old members publish nothing after December 2015 and the new ones nothing before January 2016. No plant closed at that boundary and none opened.

So there are two charts here and not one. Drawing a single stacked area across that date would show Alberta's entire thermal fleet vanishing in one month, which is not something that happened, and no published series carries coal, or natural gas, continuously across it. Within each era the published members sum to the published total exactly.

The two panels are two different questions and must never be spliced into one series. Within the fuel era, combustible fuels fell from 91.5% of output in 2016 to 76.0% in 2025, while wind went from 5.8% to 17.3% and solar, which did not exist as a published category before 2016, reached 4.1%. Only complete twelve-month years are drawn, so 2026 is absent until it finishes. 2 breaks in this series are shipped as structured metadata in the source file, which is how this page knows where they are.

These two charts come from two different surveys and do not agree with each other. The annual by-producer table and the monthly by-type table are separate collections, and summing the monthly one to a year disagrees with the annual one by as much as 25%, in 2015. Each is internally consistent and neither is corrected against the other, so a producer-class share and a fuel-type share never appear on the same chart or in the same sentence on this page.

Reading it back

What we don't know

Show the work

Method, sources, and what this page can't do

1. 1.The electricity balance's disposition side (does not reconcile) — available-for-use should equal producers' own use plus sales plus unallocatedUnallocated. In the electricity balance, what is left after own-use and sales are subtracted from what was available: transmission and distribution losses plus unbilled volume. In Alberta's published figures it goes negative in five of twenty years, which a genuine loss term cannot do.. It does not, in 8 of 20 years, by as much as 18.5 TWh in 2014. The unallocated term is described in the source as losses of around 7 TWh a year, but it is negative in 5 of those years — 2013, 2014, 2018, 2019, 2020 — which a loss term cannot be. Nothing on this page is charted from the disposition split. The supply side of the same table does reconcile, to within 11.1 GWh in its worst year, and the residential share of total sales taken from it — 15.3% in 2005 rising to 21.2% in 2024 — is quoted here and nowhere else.
2. 2.Regulated rate levels against the implied average (does not reconcile) — the energy charge alone runs at close to double the implied all-in average in some years, which would be a striking figure if it were interpretable. It is not: an independent tracker's figure for the average fixed contract in the same year already exceeds the all-in average, which is arithmetically impossible if that average really includes wires on top of energy. Something in that comparison is wrong and we have not established what, so it appears nowhere.
3. 3.EPCOR, Fortis, and a fifth significant figure (sourced, unused) — 2024 monthly rates for two further territories were sourced and verified. One year of two extra utilities adds no independent test beyond six years of two others, and both publish a separate deferral-rate column that is not part of the energy charge. The data file carries the exclusion and its reason rather than the rates themselves, so the 130 figures behind this page are Direct Energy Regulated Services and ENMAX only.
4. 4.Figures that circulate widely (two untraceable, one traced on review) — a Reference Paper passage about electricity being collected at intervals longer than a month, a commonly cited share of a 2024 inflation gap, and a regulated-rate share of about 26% for 2025. Each was searched for directly at its supposed source. The first does not exist — the real passage gives tuition fees and property tax as its examples, not electricity. The second could not be traced. The third is contradicted by the regulator's own series, which averages 21.6% across 2025 and never approaches 26% in any month of it. None appears on this page, and a search summary is not a source. A fourth figure, a 2021 share of about 45%, stood in this list in an earlier draft as untraceable — it turned out to be the Market Surveillance Administrator's own January 2021 reading, and it now appears above with its source. This page made a sourcing claim that was wrong; the correction stays visible because the method is the point.

• ·AESO — Alberta Electric System Operator, wholesale pool price and hourly load — AESO's terms of use permit internal and non-commercial use only. Tamrack is a commercial product, so the pool price cannot be shipped, charted, or derived from — not as a level, not as a monthly average, not inside a composite. This is the reason this page has retail price indexes and one derived retail average and no wholesale price at all. terms
• ·Roughly thirty archived ENMAX “DT Rate Schedule” / “DT Tariff Rate Schedule” PDFs, 2007–2021 — Those are distribution tariff (wires) schedules — service and facilities charge, system usage charge, transmission variable charge — not the RRO energy rate. Mixing them in would have blended bundled charges into an energy-only series and wrecked the comparison.
• ·EPCOR/EDTI and FortisAlberta monthly rates for 2024 — Sourced and verified, but they cover one year only and both carry a separate GOA/RRO deferral column that is not part of the energy rate. One year of two extra territories adds no independent test beyond what DERS and ENMAX already provide over six years.
• ·Alberta Utilities Commission and Utilities Consumer Advocate — reference-only licensing. Their material was read for facts and is cited by URL where it bears on the argument; no AUC or UCA table is reproduced.
• ·Pembina Institute — all rights reserved. Not quoted and not charted from.

1. 1.Wholesale (pool) price (unavailable) — The Alberta wholesale pool price is published only by AESO, under a non-commercial licence. There is no open, commercially reusable substitute. Every price series on this page is retail: an index of what households pay (CPI electricity), two indexes of what nonresidential customers pay, and one derived implied residential average in dollars. None of them tracks the pool.
2. 2.Generation by fuel across 2016 (partial) — No published series carries coal, or natural gas, continuously from 2008 to now. Before 2016-01 the type dimension cuts by turbine technology; after it, by fuel. The two cuts overlap in no month. Only the all-types total, hydraulic and wind run unbroken, and generation-mix.json ships the breaks as structured metadata so a chart can show the discontinuity rather than imply a fleet turnover that did not happen.
3. 3.Non-residential sales by customer class (partial) — Table 25-10-0021 publishes agriculture, mining and manufacturing, and other industries sales for Alberta. Those three classes are not collected, so supply-disposition.json has total sales and residential sales but nothing between them. A residential share of total sales is computable; a full customer-class breakdown is not.
4. 4.Real (inflation-adjusted) prices (partial) — The one dollar series on this page — the implied average residential price — ships nominal and real side by side, deflated into 2024 dollars by the Alberta all-items CPI, with the annual averaging checked against Statistics Canada's own published annual figures. What is NOT deflated is the three index series in prices.json: they are published indexes and are shipped as published. A relative price — electricity against all-items — is computable from what is here but is not shipped as a series, and the base-year mismatch between the two index families makes it a per-chart decision rather than a column in a file.
5. 5.Monthly generation before 2008 (unavailable) — Table 25-10-0015 begins in January 2008 for every type member. The annual by-producer and supply-disposition tables reach back to 2005; nothing here reaches further, and the monthly mix cannot be extended backwards from the annual figures because they come from a different survey.