# More horsepower. More miles per gallon.

Original Meplar analysis and design, prepared 17 October 2026. This is a new power/fuel-economy comparison using the existing frozen EPA release. The numeric inputs were not re-fetched.

Planned permanent insight: https://meplar.com/insights/us-vehicle-power-and-economy/

## Finding and interpretation

EPA production-weighted averages for new US light-duty vehicles rose from 137.3 to 258.0 horsepower and from 13.1 to 27.2 estimated real-world MPG between model years 1975 and 2024.

EPA’s reported fuel-economy average more than doubled over this interval, while average horsepower rose by 87.9%. The two gains describe the same new-vehicle production aggregate, measured in different units. This endpoint comparison does not imply that either measure improved in every intervening year; the CSV retains all 50 annual observations.

The mix of vehicles sold changes over time, so this is not a controlled comparison of the same car. The averages cannot isolate the effects of engines, electrification, vehicle size or other changes. EPA includes battery EVs and plug-in hybrids using gasoline-equivalent energy, so fleetwide MPG is not a gasoline-only average. These summaries do not establish literal fuel consumption or emissions.

## Scope and method

Production-weighted new US light-duty vehicles; final model years 1975–2024; EPA All / All aggregate.

Select the existing EPA All / All production-weighted summary for each final model year 1975–2024. Retain all 50 annual observations and both exact source metrics. Plot only 1975 and 2024, with separate zero-based horsepower and MPG scales and one-decimal display rounding. Do not sum overlapping aggregates, recalculate model-level weights, interpolate or infer emissions. All 100 annual MPG/horsepower values match the archived EPA qMatrix numeric cells.

| Metric | 1975 exact | 2024 exact | Percentage change |
|---|---:|---:|---:|
| EPA estimated real-world MPG | 13.0597 | 27.24521 | +108.6204889852% |
| Horsepower (HP) | 137.3346 | 258.0114 | +87.8706458533% |

Labels show 13.1 / 27.2 MPG and 137.3 / 258.0 HP, matching the endpoint display precision of the original EPA table. MPG more than doubles; horsepower nearly doubles. No source values have been manually altered.

EPA includes battery-electric and plug-in hybrid vehicles using gasoline-equivalent energy (MPGe) in fleetwide MPG unless otherwise noted. The 27.2 MPG value is therefore not a gasoline-only average. See the official report, printed page 6 footnote 1 (PDF page 12, zero-based index 11). This chart retains the reported metric and does not reproduce the full adjustment formula.

## Definitions and limits

- These are EPA production-weighted averages for new light-duty vehicles delivered for sale in the United States by model year. They are not calendar-year sales totals, an unweighted average over models or the existing road fleet.
- The All / All row is an existing EPA aggregate. All, All Car, All Truck and component vehicle-type rows overlap and must not be summed.
- The chart uses the final 1975 and 2024 endpoints; the download retains every final annual observation from 1975 through 2024. Preliminary 2025 is excluded.
- Horsepower and MPG use different native units and separate zero-based scales. Display labels round to one decimal; both CSVs preserve source precision.
- The mix of models, types and powertrains changes. This is not a matched-vehicle or causal estimate and does not establish uninterrupted growth.
- EPA’s official report, printed page 6 footnote 1 (PDF page 12), includes battery-electric and plug-in hybrid MPGe in fleetwide MPG unless otherwise noted. This chart does not reproduce the full adjustment formula. Do not convert the aggregate MPG to literal gasoline consumption, CO₂ emissions or mechanical efficiency.
- Source values can be revised. This analysis uses the fixed EPA summary retrieved on 13 October 2026, not a new data collection.

## Files

- chart_data_1975_2024.csv: All 50 annual aggregate observations, with exact source precision.
- plotted_endpoints.csv: The two exact endpoint rows used for the graphic.
- chart.png: Approved chart PNG, 1440 × 2585 pixels.
- chart.svg: Approved scalable SVG, native viewBox 390 × 700; uses installed system fonts.
- chart_mobile390.png: Actual 390 × 700 raster inspected at native size.
- reproduction.zip: Standalone offline reproduction package: this chart, frozen EPA inputs, code and notes.
- README.md: Methods, interpretation, source provenance and reuse.
- SOURCE_TERMS.txt: Retained EPA source attribution and reuse record.
- SOURCE_REFERENCE.json: Machine-readable frozen-source hashes and selection.
- VALIDATION.json: Numerical, archived-source, typography and render checks.
- build_chart.py: Relocatable chart builder; use the reproduction ZIP for its complete frozen inputs.
- SHA256SUMS.txt: SHA-256 hashes of all other public download files.

Both CSVs have the same six fields: model_year, regulatory_class, vehicle_type, production_share, real_world_mpg and horsepower. All values remain at their retained source precision. The annual file has 50 rows; the endpoint file has two.

## Reproduce completely offline

Download `reproduction.zip`, extract it, and run `python3 build_chart.py` inside the extracted `reproduction/` directory. Its default source is the bundled `sources/` folder. The package contains the unchanged frozen 400-row EPA CSV and original source ZIP, whose archived response is needed to independently verify all 100 annual MPG/horsepower values. The original ZIP also retains the earlier weight-chart provenance; it is an input archive, not the new graphic.

The builder needs Python 3 plus existing local `rsvg-convert` (librsvg) and `pdftotext` (Poppler) executables. It never fetches data or installs anything. SVG and data generation use only Python's standard library; PNG rendering and shaped-font text-bound QA use those installed local tools. The reviewed build used rsvg-convert 2.62.3. Fonts use the existing system stack `-apple-system, BlinkMacSystemFont, Segoe UI, Arial, sans-serif`; no font is downloaded or redistributed. Numeric outputs are independent of font substitution; raster pixels can vary on a different rendering/font platform. The package includes reviewed outputs for hash comparison.

Inputs are guarded by SHA-256. CSV bytes must match the original archived CSV; all 50 final All / All years must appear exactly once, all selected production shares must equal 1, and all 100 MPG/horsepower values must equal archived EPA qNum cells. The output has two native-unit, zero-based bar comparisons. Native 390-pixel rendering was visually inspected; font sizes are at least 14.25 px for essential labels, 20 px for values and 11 px for source-only credits. Automated native PDF checks reject clipped or overlapping text. Independent content QA verified the numbers and definitions.

## Source provenance and reuse

US Environmental Protection Agency, **2025 Automotive Trends Report**, public Qlik summary retrieved **13 October 2026**. Source table title: FE by Vehicle Type; original measure labels: Real-world MPG and Horsepower (HP). This analysis uses the final All / All aggregate from the already-frozen, production-weighted summary.

Frozen source CSV SHA-256: 05c531ccbfd1c274086e4ad42bdf8baddc17244df626b399b33e4a14839f8c94

Frozen archived raw-response SHA-256: 2e6b98e9048eb5e96451ebaa8bc04bcba304c7aa9d5a4a88a7edb1dd42742507

- US EPA, 2025 Automotive Trends Report data: https://www.epa.gov/automotive-trends/explore-automotive-trends-data
- EPA Automotive Trends definitions: https://www.epa.gov/automotive-trends/about-automotive-trends-data
- EPA Standard Open Data License: https://edg.epa.gov/EPA_Data_License.htm
- EPA report, printed page 6 footnote 1: MPG includes BEV/PHEV MPGe: https://www.epa.gov/system/files/documents/2026-02/420r26001.pdf#page=12

Retained source terms state that EPA-produced numerical data are generally US public-domain material unless otherwise specified; no contrary data-specific restriction was found in this archived source. Preserve EPA attribution and the method/transformation notes. No EPA logo or endorsement is implied. This does not assign a blanket source licence to other works or a new licence to Meplar's original chart/code.

## Tools and affiliation

Original Meplar analysis/design with AI assistance from ChatGPT/Codex. Tools: Python standard library, Decimal arithmetic, local librsvg and Poppler. I run Meplar. No paid assets, services or API calls were used.