Per-country overland departure rates from the E3 border-throughput evidence

Weekly outbound departure probability per person, assembled in the MOSAIC-OCV E3 investigation from border-throughput counts (Goma ~50k/day, Busia ~23k/day, Ressano Garcia ~7.1k/day), IOM DTM flow-monitoring, and the StatsSA P0351 overland share. Values are the four regional params_<region>.json margins.

Usage

MOSAIC_E3_TAU_WEEKLY

Format

Named numeric vector, weekly departure probability per person.

Source

MOSAIC-OCV output/E3_mobility_route2b_*/params_*.json; method in notes/E3-mobility-west-refit.md section 3.

Details

These are evidence-anchored approximations, not measurements. Every underlying count is a floor (recorded crossings only, movements not persons), so the band is deliberately wide and is carried into the prior as such – see est_overland_tau_prior.

Where two regions disagreed the larger value is kept (UGA: central 0.003 vs eastern 0.002 -> 0.003), on the grounds that each regional estimate counted only that region's corridors and so understates total outbound flow.

Ethiopia is revised up from E3's 3e-4 to 1e-3/week. E3's value implies ~5,400 outbound crossings/day for 127M people – fewer than Rwanda (13.8M) and fewer than a single Mozambican border post – and the Horn had the thinnest border-throughput evidence of the four regions. E3's own border-district decomposition (tau_national ~ f_border x tau_border, calibrated off MOZ) gives ~6.7e-4. 1e-3 keeps Ethiopia comfortably the lowest evidenced value, preserving the ordering, without asserting something arithmetically implausible. Any conclusion about Ethiopian cross-border spillover must be re-run across this band: the earlier “no resolvable spillover” finding was an arithmetic consequence of the least-evidenced number in the panel.