The uneven recovery of UK rail

archived · 2026-07-27 · github.com

What the national recovery of post-COVID rail demand hides at the regional and station level (IJC437, University of Sheffield)

Background

In 2025 the national passenger index for GB rail hit 100.8 (2019 = 100). Read that number alone and the story ends with “back to pre-COVID levels”. But did it really come back? The question was whether one national aggregate was hiding very different recoveries across 11 regions and 2,554 stations.

Design

Official ORR statistics only: regional passenger journeys (1996–2025), the national series, and station-level usage estimates. On top of a 2019 = 100 index, I measured each region against a counterfactual, its own 2010–2019 linear trend extended forward, because “back to 2019” and “back on trajectory” are entirely different questions. Regions were grouped with k-means (k = 2, chosen by silhouette width), all in R (tidyverse, readODS, cluster). No forecasting models, deliberately: the question was “how uneven is it now”, not “what happens next”. A companion coursework (IJC445) turned the same material into a composite visualisation.

Outcome

The national index says recovered (100.8), but only 3 of 11 regions exceed their 2019 level (North East +17.4, South West +6.7, London +5.1). Against pre-pandemic trends the whole country sits 17.5% short, and only 43.1% of stations have recovered. Even London pairs an above-baseline aggregate with a median station at 88.5, recovery concentrated in a few big hubs (the Elizabeth line effect). The national average says recovery; the distribution tells another story.