Exploring socio-spatial inequalities in bike-sharing systems: Case studies from Brazil and Spain ESTHER ANAYA-BOIG CENTRE FOR ENVIRONMENTAL POLICY - IMPERIAL COLLEGE LONDON 28 th September 2017, Manchester
Contents vwho is involved? vgeneral background, incl. decolonising aspects. vstage 1: The Brazilian Study vstage 2: The Barcelona Study vconclusions
Who is involved? Brazil study: Ana Clara Duran and Thiago Hérick de Sá, Centre for Epidemiological Studies in Nutrition and Health, University of Sao Paulo, Sao Paulo, Brazil Esther Anaya-Boig, Centre for Environmental Policy, Imperial College London, London, UK Joshua Daniel Shake, Centre for Metropolitan Studies, University of Sao Paulo/Brazilian Centre of Analysis and Planning, Sao Paulo, Brazil Leandro Martin Totaro Garcia, National School of Public Health, Oswaldo Cruz Foundation, Rio de Janeiro, Brazil Leandro Fórnias Machado de Rezende, Department of Preventive Medicine, School of Medicine, University of Sao Paulo, Sao Paulo, Brazil Barcelona study: Esther Anaya-Boig, Centre for Environmental Policy, Imperial College London, London, UK Àngel Cebollada i Frontera, Department of Geography, Autonomous University of Barcelona, Barcelona, Spain
Transferability remarks I Of this study: Interestingly, less data was available in Barcelona (less transparency!) In Brazilian cities the inequalities are bigger than in Barcelona: GINI coefficient (2016) Recife (BR) 0.689 Sao Paulo (BR) 0.645 Salvador (BR) 0,645 Rio de Janeiro (BR) 0.639 Porto Alegre (BR) 0.614 Barcelona (ES) 0.325 Mobility alternatives, urban features, modal split, cultural aspects quite different in one place and in the other. Population density is a relevant indicators that seem counter-intuitive: Population density (inhabitants/km 2 ) (2014 & 2016) Recife (BR) 7,133 Sao Paulo (BR) 7,913 Salvador (BR) 4,187 Rio de Janeiro (BR) 5,286 Porto Alegre (BR) 3,030 Barcelona (ES) 15,881!!
Transferability remarks II Of this study: The element of the study where built environment and its perception and use has impacted the most was in the definition of the catchment areas. 1. We first tried to find willingness to walk to the nearest bike-sharing station. Of course, this was unavailable (wishlistindicator). 2. The perception of the acceptable walking distance to a station could be determined by the perception of the distances in the whole network (assuming the user had that knowledge). So we decided to calculate the average distance between the nearest stations, using the Nearest Neighbour Analysis: Brazilian cities: 500 m Barcelona: 150 m!!
Equity Bike-sharing is a public service, but for whom? HEALTH and other BENEFITS OF CYCLING. Based on mode shift, health impact assessment attributes health benefits to bike-sharing (if the shift is not from walking). ACCESS TO A PUBLIC SERVICE. Any population groups left aside?
The Brazilian Study Paper submitted to the Journal for Transport and Health, accepted with reviews. METHODS Spatial analysis: catchment areas (buffers around the stations) of the residents and geo-referenced indicators for sociodemographics (head of the household income and ethnicity). ØSource: Brazilian census 2010 Comparison between bike-sharing users and general population for sex, age, education level and household income. ØSource: Intercept survey by the Brazilian Centre of Analysis and Planning - CEBRAP, 2014. Public procurement analysis on case study (São Paulo): ØSources: Call for tenders and contractual agreements, media coverage.
Bicycle-sharing system catchment areas in the 5 selected Brazilian state capitals and neighborhood-level mean income. 1a. Recife 1b. Salvador 1c. Porto Alegre 1d. Sao Paulo 1e. Rio de Janeiro
The Brazilian Study FINDINGS Spatial analysis: areas covered by the systems: Encompass 8 to 25% of the cities areas and 6 to 18% of the cities population. Favour wealthier and centrally located neighbourhoods. Mean income of the head of the household was 1.6 to 2.3 times the cities mean! Have 13 to 36% higher proportion of white residents than citywide average. Comparison between bike-sharing users and general population: bikeshare users: Were overwhelmingly higher educated than the general population. Who studied less than high school were less than 3%, whereas this group represents 30 to 40% of the cities population! Are poorer than citywide averages, but not as much as found in the spatial analysis. This seems to indicate that a sub-group of users don t live but probably work in the catchment areas. (limitation) Public procurement analysis on case study (São Paulo): Contractual arrangements have a strong impact shaping the location and coverage of systems. Equity is absent of these documents!
The Barcelona study On-going study, no publications yet. We focused in deepening in the spatial analysis: We included access to other cycling infrastructure: cycle paths (generally segregated or semi-segregated). We defined a Spatial Equity Index inspired in the Available Household Income Index (RFD 2016, Barcelona City Council) for the census units gathering 4 selected available variables: Variables Weight Higher education proportion 0,175 Property value 0,35 SUV proportion 0,15 Non-commercial vehicle ownership 0,15 Introduction of the hilliness variable for validation. ØSources: Census, National Statistics Institute, Barcelona City Council.
The Barcelona study Results Buffer calculation for bike-sharing stations (Nearest Neighbour Analysis): 150m, vs. 500m for Brazilian cities. Temporarily the same for the bicycle lanes. Catchment areas: Bicycle lanes: 30% Bike-sharing stations: 23% Population serviced: Bicycle lanes: 48% Bike-sharing stations: 44% No differences found for sex ratio nor non-nationals ratio.
Spatial Equity Index Census units 175-250 250-300 300-350 350-400 400-550
Spatial Equity Index Under-serviced census units are poorer. Census units without bike-sharing service, have 7% lower index than those within the catchment area. Census units without bicycle lanes, have 6% lower index than those within the catchment area.
Limitations and further research Variable wishlist (transparency/open data issues): Refine catchment area calculations. Income. Work location, with associated sociodemographics (currently only residents and at the level of census unit). Including Home location for workers within catchment areas intermodality assessment. Bike-share users survey. Any other variable for accessibility (not only distance?) Further analysis: Integration with public transport networks intermodality assessment (tariff integration support). Other barriers to cycling and bike-sharing that might explain a lower uptake of the system by the closer residents - surveys/qualitative methods. Test the Index with more case-studies, ideally with variables from wishlist if available, in order to validate the model.
Spatial equity index - adjustments Controlling for hilliness proved irrelevant (>5% slope in half of the census unit area) Census units without hilliness had 3% higher index than the hilly ones. Validated for the case of Barcelona, but should be tested for other cases.
Conclusions Spatial bike-sharing equity still needs to be defined. We propose an index that mainly depends on available data and selected relevant indicators for bike-sharing use. There are local variables that need to be calculated for every case, e.g. buffer distance: We propose the Nearest Neighbour Distance analysis, but other behavioural aspects might apply (such as availability to walk to nearest bike-share station, or to cycle to nearest segregated cycle infrastructure). Relevance of bike-sharing for equity could be different depending on the proportion of cycle trips made by with public bicycles. Procurement processes and regulations might have a big impact for spatial distribution of bike-sharing stations. Political aspects also under-studied. The confounding effect of technical limitations (such as slope) for the location of stations was irrelevant in this case, but should be taken into account in every case.
Thank you! Contact: Esther Anaya-Boig Centre for Environmental Policy Imperial College London e.anaya14@imperial.ac.uk Observatorio de la bicicleta pública en España (Observatory of bike-sharing in Spain) esther.anaya@bicicletapublica.es