Accessibility indicators take land use, the transport system, and their interdependencies into account in a holistic fashion. In many areas, however, spatial data to perform accessibility calculations are hard to obtain or not available at all. Freely available volunteered geographic information (VGI) like from OpenStreetMap (OSM), which is increasingly becoming a world-wide standard for geo-spatial data, may be a solution to this problem. In fact, some accessibility studies use data from OSM to create representations of the transport network and to perform network-based computations. In this paper, two approaches for accessibility assessment for Nelson Mandela Bay Municipality in South Africa are presented. The approaches possess different levels of utilization of OSM data which both exceed the use of OSM data for network creation. In the first approach, the transport network as well as locations and types of activity facilities are taken from OSM. Additionally, a synthetic population is created based on a census. The corresponding travel demand is generated based on a travel survey. Local expert knowledge is applied to design a household-specific accessibility indicator that takes into account various characteristics of travel and land use, such as travel time to work and/or education, travel time to the nearest health/shopping facility, availability of different transport options, and availability of various facilities within walking distance. Weights are used to combine the respective values of aforementioned characteristics into a composite, household-based accessibility score. This approach appears particularly suitable in the South African context where housing locations and travel characteristics are highly diverse among the population. The second approach relies exclusively on OSM data, which is -- as before -- used to create the network and activity facilities in the model. The approach applies an econometric accessibility indicator, which calculates the accessibility of a given measuring point as the weighted sum over the utilities of all opportunities including the costs of reaching them. In contrast to the first approach, no synthetic population, but only household locations -- collected from OSM in the same way as (other) activity facilities -- are used for the calculation. The approach is highly portable since no input data other than those from OSM are used. It is found that the second approach, though being much more lightweight in terms of data requirements, yields the same quality of insights concerning accessibilities of different areas of the region. Both approaches detect areas where levels of accessibility deprivation are high and where interventions in the transport-land-use system are advisable. Consequently, the paper is a contribution to accessibility analyses based on easily obtainable and ubiquitously available OSM data to obtain similar results as with respective approaches based on traditional spatial data.
