Why do national energy scenarios disagree?
Line up four scenarios for a fictional nation, trace each gap to its assumptions, and sort them as exploratory or target-seeking
The idea in one table: four scenarios for Prairieland, one accounting
| Scenario | Kind | Demand | Policy | CO₂ in 2050 | CO₂ 2026–2050 | Electricity made, 2050 | Oil and LNG imports, 2050 |
|---|---|---|---|---|---|---|---|
| Current policies | exploratory | Central demand | no carbon price; slow electrification | 222.9 Mt | 5,484 Mt | 339.5 TWh | 2,258 PJ |
| Announced plans | exploratory | Central demand | carbon price $40 a tonne from 2030; electrification as announced | 166.9 Mt | 4,801 Mt | 387.6 TWh | 1,369 PJ |
| Gas boom | exploratory | High growth demand | no carbon price; slow electrification | 290.1 Mt | 6,268 Mt | 410.9 TWh | 3,358 PJ |
| Budget pathway | target-seeking | Central demand | carbon price found: $84 a tonne; the fastest electrification allowed | 65.9 Mt | 2,586 Mt | 442.7 TWh | 197 PJ |
- Two kinds. Exploratory scenarios start from assumptions and ask what happens; a target-seeking one starts from a goal (the 2,600 Mt budget) and finds the policy that meets it.
- Gaps have causes. From Announced plans to Gas boom, CO₂ in 2050 moves +123.2 Mt; averaged over all 24 orders of swapping: Demand +39.0 Mt, Fuel +22.1 Mt, Policy +38.9 Mt, Costs +23.2 Mt.
- Order matters. Swapped one at a time, in a different order, the same groups get different shares of one total.
- Why M13 disagrees. It raises electrification in a straight line to 2050, not early as here, so it burns more fuel and cannot reach 2,600 Mt (about 2,700 Mt at best).
A screening accounting, not a forecast. With JavaScript on, you trace the gaps yourself.
Prairieland is fictional; its numbers are illustrative. Its balance, plants, demand cases, fuel and cost ranges (EIA's AEO2026 assumptions, 2025 dollars) and CO₂ factors (IPCC 2006) come from the course's shared national data.
Set for teaching. The four scenarios, the clean mix and the sharing rule.
About 25 guided minutes in eight steps. Helpful first: A1, cash flows and the cost of capital, and lesson 13, planning a power system.
What you just did, as diagrams
- Two kinds of question
- Exploratory scenarios ask what happens if their assumptions hold; a target-seeking one asks what it would take to meet a goal.
- The spread is information
- Prairieland's four end 2050 between 65.9 Mt and 290.1 Mt of CO₂; none of them is a forecast.
- Gaps have causes
- From Announced plans to Gas boom, the four groups alone add up to +170.3 Mt against a real gap of +123.2 Mt.
- Order matters
- Swapping technology costs adds 0.0 Mt to +56.3 Mt, depending on the order; its average over the 24 orders is +23.2 Mt.
- The policy is an output
- The Budget pathway's 2,600 Mt budget is fixed; its carbon price, $84 a tonne, is found.
Go deeper · AnalystThe accounting behind the four scenarios
- Demand. Each sector’s useful energy grows at its demand case’s rate; a share of it switches to electricity at the policy’s pace, up to Prairieland’s limits.
- Existing plants run as in 2025 until their retirement year; from 2030, a coal or combined-cycle plant closes early when its fuel and carbon cost per MWh is above the clean mix’s full cost.
- New power. The rest is shared between new gas plants and a clean mix (wind, solar, the generic firm clean plant, batteries) by their full costs per MWh:
\[s = \min\left(1,\ \max\left(0,\ \tfrac{1}{2} + \frac{C_{\text{gas}} - C_{\text{clean}}}{2W}\right)\right), \qquad C = \frac{K \cdot \mathrm{CRF}(r, n) + F}{\mathit{cf} \times 8.76} + V\]
- Symbols, in the table below; \(C\) is a plant’s full cost per MWh.
| Symbol | What it is | Unit |
|---|---|---|
| \(s\) | the clean share of new power | 0 to 1 |
| \(W\) | the rule’s half-width: the cost gap at which all or none of new power is clean | $ per MWh |
| \(K\) | capital cost | $ per kW |
| \(F\) | fixed upkeep | $ per kW a year |
| \(V\) | variable cost; for gas, also fuel and the carbon price | $ per MWh |
| \(\mathit{cf}\) | capacity factor | 0 to 1 |
| \(\mathrm{CRF}(r, n)\) | the capital recovery factor, \(r / (1 - (1 + r)^{-n})\): the share of \(K\) paid each year | per year |
| \(r\), \(n\) | the discount rate; the life | per year; years |
| 8.76 | the MWh one kW makes in a year at full output (8,760 hours ÷ 1,000) | MWh per kW |
- Fixed costs. Every capital cost is a fixed value inside Prairieland’s range, the same in 2030 and 2050; none changes with how much is built.
- Imports and CO₂. Domestic oil and gas output stay at 2025’s levels, so extra fuel is imported; CO₂ uses the IPCC’s default factors for fuel burned.
Go deeper · AnalystSplitting a gap fairly when causes interact
- One at a time. Each change alone, from the same start: simple, but the parts need not add up to the gap.
- In order. Each change after the ones before it: the parts add up exactly, but they depend on the order.
- The average over every order. Four groups have 24 orders; each group’s step averaged over all of them gives shares that add up to the gap (the Shapley value of game theory).
- Index decomposition. For identities such as CO₂ = population × income × energy intensity × carbon intensity, the logarithmic mean Divisia index splits a change with no residual (Ang 2015, Energy Policy).
- Report the method. Whichever you use, say so beside the bars.
Go deeper · AnalystWhere a national budget comes from
- A target is chosen. A target-seeking scenario needs a target first; a national carbon budget is one, and choosing it is a value judgement, not a model output.
- Sharing a global budget. Common rules give each country a share by population, by today’s emissions, by capability to pay or by historical responsibility; each rule gives a very different national budget.
- Prairieland’s reference budgets. Step 7 lists three, arithmetic on 2025’s emissions (no change, or a straight line to zero by 2050 or 2040), not a fair share under any rule.
- Report the rule. A study that cites a national budget should say which sharing rule, which global budget and which start year it used.
Go deeper · ModelerHow an optimization model seeks a target
- A budget is a rule. A planning model adds one constraint: the sum over the model years of weight × emissions stays at or below the budget.
- Its price. In a least-cost model the constraint’s shadow price is what one more tonne of budget would save; the model finds plan and price together.
- This page’s price. Its implied carbon price is the lowest whole-dollar price that meets the budget under the sharing rule, found by trying prices; it need not equal a model’s shadow price.
- Foresight. A model that sees every period at once plans differently from one that plans decade by decade; that choice is one more assumption to report.
- Next. Modeler lesson M13 builds Prairieland’s multi-region, multi-decade model; the Pathway Explorer (SB4) lets you run it.
Go deeperWhat this lesson leaves out
- Hourly balance, storage timing, transmission between regions and adequacy: the job of a planning model such as M13's.
- How prices respond to demand, and how people and firms respond to prices beyond the sharing rule.
- The cost of heat pumps, vehicles and other end-use equipment.
- Methane and other gases, land use and carbon removal.
- Probabilities: this lesson assigns none, so its results do not tell us which scenario is more likely.
Check yourself
Practice, about 30 minutes: four myths, then six exercises with worked answers.
Myth or fact? “An energy outlook is a forecast of what will happen.”
- Myth
- Fact
Answer
Myth. EIA says the results of its Annual Energy Outlook 2026 are not predictions; the IEA calls its scenarios exploratory or normative.
Myth or fact? “When two outlooks disagree, at least one of them is wrong.”
- Myth
- Fact
Answer
Myth. Both can be consistent: Current policies and the Budget pathway answer different questions with the same data.
Myth or fact? “A target-seeking scenario shows what will happen once a country adopts the target.”
- Myth
- Fact
Answer
Myth. It shows one way to meet the target under its assumptions: with Gas boom's cheap gas, the Budget pathway needs $134 a tonne instead of $84 a tonne.
Myth or fact? “A gap between two scenarios splits into parts that do not depend on how you split it.”
- Myth
- Fact
Answer
Myth. When assumptions interact, the parts depend on the order: here, swapping technology costs adds 0.0 Mt to +56.3 Mt.
Practice · about 30 minutes on your own
The clean mix's cost. What does the clean mix cost per MWh?
Wind $49.00 per MWh; weight 55% Solar $67.93 per MWh; weight 35% Firm clean plant $79.77 per MWh; weight 10% Batteries beside the solar $9.32 per MWh of the mix Answer
0.55 × 49.00 + 0.35 × 67.93 + 0.10 × 79.77 + 9.32 = $68.02 per MWh.
The sharing rule. What share of new power is clean?
Clean mix $68.02 per MWh New gas, with Announced plans' $40 carbon price $71.48 per MWh Half-width, the cost gap at which all or none is clean $20 per MWh Answer
0.5 + (71.48 − 68.02) ÷ (2 × 20) = 0.5 + 3.46 ÷ 40 = 0.59, about 59%.
A carbon price per MWh. What does a $40 carbon price add to each MWh from a new combined cycle?
Gas burned 5.912 GJ per MWh CO₂ from gas 56.1 kg per GJ Answer
5.912 × 56.1 ÷ 1,000 = 0.332 t per MWh; × $40 = $13.3 per MWh.
Cumulative CO₂. What is Current policies' total CO₂ for 2026–2050?
Weight how many years of 2026–2050 a model year counts for, with straight lines between model years 2025 214.9 Mt; weight 2 2030 219.9 Mt; weight 5 2035 220.4 Mt; weight 5 2040 217.9 Mt; weight 5 2045 218.9 Mt; weight 5 2050 222.9 Mt; weight 3 Answer
2 × 214.9 + 5 × 219.9 + 5 × 220.4 + 5 × 217.9 + 5 × 218.9 + 3 × 222.9 = 5,484 Mt (from unrounded values).
Two orders, two answers. What does swapping in Gas boom's fuel prices add in each order?
Current policies, CO₂ in 2050 222.9 Mt With Gas boom's demand swapped in 262.1 Mt With Gas boom's fuel prices swapped in 244.6 Mt With both swapped in 290.1 Mt Answer
Demand first, fuel adds 290.1 − 262.1 = +28.0 Mt; fuel first, 244.6 − 222.9 = +21.7 Mt. Averaged over the two orders: (28.0 + 21.7) ÷ 2 = 24.85 Mt.
Sort a description. An outlook “assumes laws and regulations in force on 31 December and reports what follows to 2050.” Exploratory or target-seeking?
Answer
Exploratory: it fixes its assumptions and asks what follows; nothing forces a result. EIA describes its Counterfactual Baseline case this way.
Prairieland is fictional; its numbers are illustrative. Every number here comes from the page's screening accounting.
Numbers, sources and what’s next
Numbers to remember
| Number | What it means | Source |
|---|---|---|
| 2 + 2 | The IEA's World Energy Outlook 2025 has two exploratory scenarios (Current Policies, Stated Policies) and two normative ones (Net Zero Emissions by 2050, ACCESS). | IEA commentary · read 2 Nov 2026 |
| Dec 2025 | The laws and regulations in force that EIA's Annual Energy Outlook 2026 Counterfactual Baseline case assumes; EIA says its results are not predictions. | EIA, AEO2026 narrative · 8 Apr 2026 |
| 1990–2014 | The years of the UK's electricity transition that Trutnevyte (2016) compared with cost-optimal scenarios: the least-cost answer departed from what happened. | Energy 106 · 2016 |
Keep going
In the real world
The IEA's World Energy Outlook 2025 (November 2025) reports exploratory and normative scenarios side by side, and EIA's Annual Energy Outlook 2026 (8 April 2026) adds side cases on growth, oil and gas supply and technology costs.
Comparing one outlook's exploratory scenario with another's normative one compares answers to different questions. This lesson makes no claim about which outlook is right.
Sources and further reading
- International Energy Agency. Scenarios in the World Energy Outlook 2025 (commentary): which scenarios are exploratory and which normative (read 2 November 2026).
- U.S. Energy Information Administration. Annual Energy Outlook 2026, narrative: foreword and introduction, the Counterfactual Baseline case and the side cases. Released 8 April 2026.
- U.S. Energy Information Administration. Assumptions to the Annual Energy Outlook 2026: Electricity Market Module, Tables 3 and 4: the capital costs behind Prairieland's ranges (2025 dollars).
- IPCC (2006). 2006 IPCC Guidelines for National Greenhouse Gas Inventories, Volume 2, Energy: default CO₂ factors.
- Trutnevyte, E. (2016). Does cost optimization approximate the real-world energy transition? Energy 106, 182–193.
- DeCarolis, J. F., et al. (2017). Formalizing best practice for energy system optimization modelling. Applied Energy 194, 184–198.
- Ang, B. W. (2015). LMDI decomposition approach: A guide for implementation. Energy Policy 86, 233–238.
- Prairieland: the course's shared national data (prairieland.json), with the sources listed there. The four scenarios and the sharing rule: set for teaching.
Spotted an error, or stuck? Write to hlli@ksu.edu. Corrections are made in place and dated here.
Text licensed CC BY 4.0, code MIT: teachers and students may reuse and adapt it, with credit to Honglin Li.