Methodology
How the index is calculated, classified and aggregated.
How the Transport Poverty Index is computed. TPI is a 0–100 composite score built from four pillars with different weights and aggregated from the settlement level up to municipality and province.
0.90×–1.00×, from the municipality's dependency ratio and its share of people with disabilities. It multiplies the final TPI, with the same factor for every settlement in the municipality. The four pillars measure the place, not its population.
Each 0–100 score is mapped to one of three classification bands. The bands provide a quick read of how a place stands relative to the rest of the country and drive the colour scale used across maps, charts and tables.
Transport service meets the population's needs across accessibility, availability, affordability, and adequacy
Transport service is fragile and could erode further; gaps already put some residents at risk of mobility exclusion
Severe transport poverty with substantial deficits across multiple pillars
Measures how easily residents reach essential services and the regional centre. Built from two sub-pillars of equal weight: Service Access (hospital, pharmacy, school, kindergarten) and Capital Reach. Both are measured as driving time and scored on the same scale.
Time to nearest hospital, pharmacy, school, and kindergarten from settlement centroid. Scored by car-travel-time bands (≤5 min = 100, >60 min = 0).
Driving time from each settlement to the nearest oblast capital. Oblast capitals concentrate universities, regional administrative offices (prefecture, traffic police, social services), courts and larger labour markets. Same scale as Service Access: ≤5 min = 100, 15 min = 75, 30 min = 50, 60 min = 25, >60 min = 0.
Service Access covers acute care (hospitals) and routine care (the dense pharmacy layer); GPs are intentionally excluded as near-ubiquitous, which would not differentiate transport-poor places. Service Access and Capital Reach use routed driving time — the realistic mode for reaching a hospital or oblast capital — and are scored on the same scale. Walking access to public transport is measured separately by Transit Reach in the Availability pillar. The GISCO dataset is used only to cross-validate coverage, not to score. For the regional-centre town itself the travel time to the regional centre is zero and the score is 100.
Measures transport supply through four sub-pillars: Transit Reach (30%), Service Frequency (40%), Motorway Proximity (15%) and Road Network Density (15%). For a settlement with no timetable within reach, Transit Reach carries 70% and the two road sub-pillars keep their 30%.
Time from settlement centroid to the nearest transit stop — minimum across BDZ rail, Ivkoni intercity bus, urban GTFS (8 cities), and SARS BDP transit stops. Scored by 5/15/30/60-minute walking-time bands. Where no timetable is within reach it also carries the frequency weight (70%).
Daily departures from rail stops within 5 km, intercity bus stops within 3 km and urban stops within 1.5 km of the settlement centroid, summed across all GTFS sources. ≥80 a day = 100, 40 = 75, 20 = 50, 10 = 25, below 10 falls proportionally to 0.
Driving time from each settlement to the nearest operational motorway interchange: ≤15 min = 100, 30 min = 75, 60 min = 50, 90 min = 25, >90 min = 0.
Total length of Class I, II and III national roads within the municipality, divided by municipal area (km/km²). Densities below 0.03 km/km² score 0; above 0.15 score 100.
Transit Reach is the walking-access measure — routed on foot for every settlement — and is stricter than the 2 km JRC/TSI standard. The radii used by Service Frequency (5 km for rail, 3 km for intercity bus, 1.5 km for urban transport) are departure-counting catchments, not walking thresholds: a railway station is also reached by car or bus, an urban stop on foot. Service Frequency carries the largest weight in the pillar: a nearby stop is only as useful as the departures from it. Motorway Proximity reflects private-vehicle and strategic-road connectivity, and Road Network Density inter-urban connectivity; the road-density bands are calibrated to the real Bulgarian distribution. Rail and intercity-bus schedules are sourced via the National Access Point.
Measures how affordable transport is through two sub-pillars: Fare Burden (65%) and Economic Vulnerability (35%). Computed at municipality level; all settlements within a municipality share the same score.
Monthly travel cost as a share of the net mean monthly wage. Where an urban pass is published: the price of the full monthly pass. Elsewhere: the fuel cost of 12 return car trips a month to the regional centre (8 l/100 km, €1.90/l), four times lower where a rail stop is within 5 km. ≤2% = 100, 5% = 75, 10% = 50, 15% = 25, ≥20% = 0.
Blend of three measures: 50% Eurostat At-Risk-Of-Poverty rate at NUTS-2 level (the EU's standard relative income-poverty indicator, ilc_li41 2024), 25% income tier per municipality (net mean monthly wage), and 25% employment rate. The AROP component captures regional poverty risk; the income tier adds within-region municipal differentiation; the employment rate adds an independent economic-capacity signal. Measures the economic context that determines a household's capacity to absorb transport costs.
Fare Burden is the monthly cost of getting around relative to the net wage. In municipalities with a published urban pass it is the price of the full pass. In the others it is the fuel cost of 12 return car trips a month to the regional centre, computed for every settlement and averaged by population for the municipality; where a rail stop lies within 5 km the cost is four times lower, because the trip can be made by train. One scale applies to the pass and to the car. The pillar is one value for the whole municipality. Economic Vulnerability combines relative-poverty risk, municipal net-wage income and the employment rate. Income is the net mean monthly wage (employed persons), which can overstate typical income in pensioner-heavy municipalities.
Measures the safety and quality of the transport environment. Combines Vulnerable User Safety (30%), Local Mobility Quality (30%), Network Risk (25%) and Stop Quality (15%).
Per-settlement weighted average of pedestrian-crossing (50%), intersection (35%), and grade-separated-crossing (15%) SARS BDP inspection scores. Generalized-N redistribution when sub-components are missing.
Sidewalk coverage, pavement quality, and pedestrian crossing density from SARS road inspections. All indicators on a 0–100 scale.
Length-weighted share of road segments classified HIGH or VERY HIGH risk by SARS BDP inspections, plus republican-road segments flagged POOR ROAD CONDITION. Below 3% high-risk share scores 100; above 20% scores 0.
Mean SARS inspection score across transit stops and bus/train stations within the settlement polygon. 5 boolean fields per stop (signage, fencing, road surface, parking, lighting).
Vulnerable User Safety reflects how safe the transport environment is for vulnerable road users. Local Mobility Quality is comprehensive at municipal level and coarser for individual small settlements. A settlement with no inspected objects takes the average of its own group in the municipality — the municipal centre together with the towns, or the remaining settlements — and the municipal average when the group has no data either. Pedestrian-crossing density is one value per municipality. Stop Quality captures the condition of stops; accessible-stop (PRM) inclusion and ease of access to travel information are not available as national data.
The TPI is computed at the settlement level and then aggregated upward to municipality and province. Larger places contribute proportionally more to each aggregate because the rollup is weighted by population.
Every Bulgarian settlement (identified by its EKATTE code) gets its own pillar scores, computed from infrastructure data within or routed from that settlement.
Settlement scores combine into a municipality score as a population-weighted average. Population is taken from the GRAO 2026 register — current-address count where available, permanent-address otherwise — so a settlement with 10 000 residents has roughly 100× the weight of a settlement with 100 residents.
Province scores are the population-weighted average of the municipalities in that province. The same step, one level up.
Where a measurement is missing for a place, the pillar is scored on the data that exists for it. A settlement with no timetable within reach has no frequency score, and that weight moves to Transit Reach. A municipality with no published urban pass gets its Fare Burden from the cost of travelling by car. A settlement with no inspected objects takes the average of its own group in the municipality.
Availability carries 40%, Accessibility 25%, Affordability 20% and Adequacy 15%. Availability and accessibility are preconditions for everything else: the price and quality of a service matter only once the service exists and can be reached. The two pillars therefore carry 65% of the score together. Adequacy describes the quality and safety of the environment and carries the smallest weight.
Every indicator is converted to a 0–100 score against fixed, policy-anchored targets — a distance-to-target (goalpost) method — rather than by stretching each indicator across the current sample (min–max normalisation). This gives every score an absolute meaning and keeps results comparable over time and between places: a municipality's score reflects its own conditions, not its rank in this year's data.
Two inputs — published urban fares (Affordability) and timetable-based service frequency (Availability) — exist mainly for larger cities, because they are not published nationally. The index is defined for every place without them. Where no pass is published, Fare Burden is computed from the cost of travelling by car to the regional centre. Where no timetable is within reach, the frequency weight moves to Transit Reach, which is computed for every settlement. Weight does not move between pillars.
The three-way category (well-served / at-risk / transport-poor) is most decisive at the extremes — the best and worst places stay stable regardless of weighting. For everywhere in between, read the per-pillar breakdown rather than the exact rank: a place may score low on affordability but well on access, which points to different remedies (fare support vs. new services vs. infrastructure). The map's colour-by-pillar toggle is built for exactly this.