Update natural gas regression parameters with AEO 2026 data

[Yunzhi-Chen]

Summary

Update aeo_pipeline_config.json for AEO 2026 data. AEO 2026 (released April 8, 2026) renamed and reorganized its scenario cases compared to AEO 2025, so the config has been updated to reflect the new scenario IDs.

Technical details

Implementation notes

• aeo_year: 2025 → 2026
• beta_regression.start_year: 2024 → 2025
• Reference case renamed from ref2025 to cb2026 ("Counterfactual Baseline")
• lm2025 (Low Macroeconomic Growth) → lm2026 (Low Economic Growth)
• hm2025 (High Macroeconomic Growth) → hm2026 (High Economic Growth)
• Removed nocaa111 (no longer an AEO 2026 case)
• Added new AEO 2026 cases: higheldmd (High Electricity Demand), altelec (Alternative Electricity), electrnp (Alternative Electricity and Transportation)
• Retained unchanged scenario IDs: highZTC, lowZTC, alttrnp
• Alpha regression reference changed from ref{aeo_year} to cb{aeo_year} to match the new baseline ID
• AEO dollar year updated from 2024 to 2025; price_deflator_to_2004 updated accordingly: 0.602410 → 0.586750

Relevant sources or documentation

• AEO 2026 Case Descriptions: https://www.eia.gov/outlooks/aeo/assumptions/case_descriptions.php
• AEO 2026 Table Browser: https://www.eia.gov/outlooks/aeo/data/browser/

The full list of AEO 2026 scenario IDs (from the EIA API /aeo/2026/facet/scenario endpoint):

Case	Scenario ID
Counterfactual Baseline	cb2026
High Economic Growth	hm2026
Low Economic Growth	lm2026
High Oil and Gas Supply	highogs
Low Oil and Gas Supply	lowogs
High Zero-Carbon Technology Cost	highZTC
Low Zero-Carbon Technology Cost	lowZTC
High Electricity Demand	higheldmd
Alternative Electricity	altelec
Alternative Transportation	alttrnp
Alternative Electricity and Transportation	electrnp

Beta regression results: AEO 2025 vs AEO 2026

National beta:

	AEO 2025	AEO 2026	Change
beta	0.080017	0.078702	-1.6%

Regional betas (cd_beta0.csv):

Region	AEO 2025	AEO 2026	Change
New_England	3.875323	2.359082	-39.1%
Mid_Atlantic	0.151497	0.185420	+22.4%
East_North_Central	0.033492	0.013440	-59.9%
West_North_Central	0.856915	0.513493	-40.1%
South_Atlantic	0.175498	0.029744	-83.1%
East_South_Central	0.139571	-0.026099	sign flip
West_South_Central	-0.054900	-0.126897	+131.1% (mag)
Mountain	0.670353	0.049527	-92.6%
Pacific	1.341567	0.704760	-47.5%

Beta regression fit diagnostics: AEO 2025 vs AEO 2026

Model: demeaned price ~ beta_reg × Q_reg + beta_nat × Q_nat

r2 = R² on demeaned (within region×year) variation only — measures beta's explanatory power across scenarios.

Scope	Region	AEO 2025 R²	AEO 2026 R²	AEO 2025 N	AEO 2026 N
national	ALL	0.728	0.620	1701	2106
regional	New_England	0.588	0.209	189	234
regional	Mid_Atlantic	0.192	0.229	189	234
regional	East_North_Central	0.038	0.003	189	234
regional	West_North_Central	0.453	0.216	189	234
regional	South_Atlantic	0.408	0.009	189	234
regional	East_South_Central	0.135	0.003	189	234
regional	West_South_Central	0.049	0.194	189	234
regional	Mountain	0.575	0.008	189	234
regional	Pacific	0.670	0.449	189	234

Notes:

• N increased from 1701 to 2106 (189→234 per region) due to additional AEO 2026 scenarios (9 vs 7 in beta regression).
• National demeaned R² dropped from 0.73 to 0.62. The new AEO 2026 policy-off scenarios (altelec, electrnp, higheldmd) introduce price variation not fully explained by NG demand alone, which is expected.
• Regional R² dropped substantially in several regions (e.g., South_Atlantic 0.41→0.01, Mountain 0.58→0.01), suggesting the new scenarios have price-demand relationships that differ structurally from the prior scenario set.

Alpha regression results: AEO 2025 vs AEO 2026

Reference / counterfactual case:

High oil and gas case:

Low oil and gas case:

Alpha difference (AEO 2026 - AEO 2025):

Additional details

• Output verification: Confirmed all AEO 2026 scenario IDs via the EIA API endpoint (/v2/aeo/2026/facet/scenario). cb2026 is the formal NEMS run prefix for the AEO 2026 baseline (Counterfactual Baseline).