# `compression` [Back to L8](../index.md) | [Browse all axes](../../browse_by_axis.md) | [Browse all options](../../browse_by_option.md) > Axis ``compression`` on sub-layer ``L8_A_export_format`` (layer ``l8``). ## Sub-layer **L8_A_export_format** ## Axis metadata - Default: `'none'` - Sweepable: False - Status: operational ## Operational status summary - Operational: 3 option(s) - Future: 0 option(s) ## Options ### `gzip` -- operational Gzip-compress every output file individually. Each ``.json`` / ``.csv`` becomes ``.json.gz`` / ``.csv.gz``. Reduces artifact size by 60-80% for typical macro panels with marginal write-time overhead. Read-side: pandas / pyarrow auto-detect the gzip extension. **When to use** Reducing artifact size for archival; production sweeps. **References** * macroforecast design Part 3, L8: 'reproducibility = manifest + provenance + bit-exact replicate.' **Related options**: [`none`](#none), [`zip`](#zip) _Last reviewed 2026-05-05 by macroforecast author._ ### `none` -- operational No compression (default). Default. Files are written uncompressed -- cheapest at write time and most convenient for direct browsing / spot-checking. Recommended for development. **When to use** Default; cheapest write-time option. **References** * macroforecast design Part 3, L8: 'reproducibility = manifest + provenance + bit-exact replicate.' **Related options**: [`gzip`](#gzip), [`zip`](#zip) _Last reviewed 2026-05-05 by macroforecast author._ ### `zip` -- operational Zip-archive the entire run output directory. Wraps the run output directory in a single ``.zip`` archive after writing. Convenient for transferring an entire run via email / web upload as a single file. Slightly less efficient than per-file gzip but shipping a single archive matters for some workflows. **When to use** Packaging the run for transfer over email / file-sharing services. **References** * macroforecast design Part 3, L8: 'reproducibility = manifest + provenance + bit-exact replicate.' **Related options**: [`none`](#none), [`gzip`](#gzip) _Last reviewed 2026-05-05 by macroforecast author._