Levenmouth Rail Link — Economic Analysis
Why does this matter? The case for measuring what rail really does to an economy

The problem with how we evaluate transport

Standard cost-benefit analysis of transport projects focuses almost entirely on time savings for existing travellers and compares them against construction costs. It largely ignores what happens to the local economy: whether people find work they otherwise couldn't reach, whether businesses open because footfall improves, whether a town stops losing young people because it's no longer cut off. Levenmouth went 50 years without a railway. This is an attempt to measure what that actually cost.

What I measured

I use difference-in-differences, comparing Levenmouth's claimant rate to similar Fife towns that didn't get rail (Cowdenbeath, Lochgelly). If claimant rates fell everywhere by 1% but fell 1.28% in Levenmouth, the extra 0.28 pp is attributed to rail. The early signal is −0.28 pp after 19 months: modest, directionally consistent with what the evidence predicts, and expected to grow. More negative = more people in work. Methil (Scotland's most deprived ward in this area) shows the largest early effect. That is the equity case for rail in one number.

The simulation

The agent-based model is calibrated to reproduce that −0.28 pp early-period signal, then projects forward to Year 5. It models 3,700 residents and 60 businesses across five settlements, including the friction of getting to a station from Methil (2km) or Kennoway (5km). Move the sliders to explore: what if adoption is slower? What if the Edinburgh labour market connection is stronger? The dashed red line on the employment chart is the observed early-period benchmark; the blue projection should drift below it as adoption grows.

Running simulation…

Claimant rate

% of working-age adults on unemployment benefits, with and without the rail link

Rail's net effect on employment

More negative = bigger benefit. Dashed red line = observed early-period effect (aggregate view only). In zone views, individual towns show wide fluctuations; the key pattern is direction and the Year 3 peak, not the exact month-to-month values. Methil and Buckhaven drive almost all of the aggregate effect.

Residents using the railway

Share of working-age adults who have adopted rail. Ramp-up is gradual, with slower adoption for residents further from Leven or Cameron Bridge station.

Local businesses

The two lines overlap in years 1–2; this is expected, not a bug. Business effects are too slow to detect at short timescales; the gap opens in years 3–5. Counter-pressure: residents shopping in Edinburgh may reduce local spend.

Methodology: Difference-in-differences (TWFE, 2022+ pre-period) using NOMIS claimant count data, Jun 2024 – Dec 2025. Empirical estimate: −0.28 pp (p=0.441, 4 clusters; statistically suggestive, not yet significant). Agent-based model: Mesa 3.5 · 3,700 resident agents · 60 business agents · 5 zones · 8 seeds. Business entry rates from Audretsch & Fritsch (1994); SIMD-adjusted. ORR station usage: 277k annualised (Year 1), consistent with Borders Railway ramp-up. Control wards (Cowdenbeath, Lochgelly) have pre-existing rail access; estimates are conservative lower bounds.