This dissertation consists in three essays with complementary approaches on the economics of social integration in an urban setting. The first essay analyzes the emergence of ethno-cultural hierarchies in a multi-cultural context, typical of nowadays large metropolises. This emergence is studied using an evolutionary game theory model according to which, in a society, a common hierarchy view emerges from a multitude of independent interactions between members of the different ethno-cultural groups. The originality of the model lies in the featuring of several minorities and hierarchical views (i.e. multi-group and multi-strategy model) and in the reciprocal effects that minorities may have on each others' social statuses. These effects allow to explain the non-linear relationship between a minority's size and its status suggested by the empirical literature, as well as the complex impacts of a new minority's arrival on the other minorities. The evolutionary process implies that the adopted ethno-cultural hierarchy is, in most cases, too inegalitarian and thus economically inefficient. The second essay presents an urban economics model adapted to the sub-Saharan African city context where land ownership is often informal and uncertain and where land transactions are often hampered by important information asymmetries between buyers and sellers. The model allows to theoretically study the impact of two institutions aimed at reducing transaction uncertainty. The first one consists in a formal land registration system administered by the government, the second is a traditional social trust norm that links specific social groups. This model is, to the best of our knowledge, the first one to study the effects of a social norm on the functioning of an urban housing market and the urban structure. It shows that the land registration system is more efficient than the traditional trust norm if registration costs are limited, but also that the two institutions are partly substitutable. The model predicts that, with the gradual decrease of registration costs, land registration will progressively replace social trust norms in the future.Eventually, the third essay consists in an econometric analysis of a large urban renewal program launched in France in 2003 for the renovation of 600 deprived neighborhoods (i.e. the « Programme National de Rénovation Urbaine », PNRU). In order to avoid possible biases linked with heterogeneities in the program's effects across neighborhoods and across time periods, we rely on the very novel DID_M estimator developed by De Chaisemartin and D'Haultfoeuille (forthcoming) and complement its results with a more traditional difference-in-differences estimation. Our results suggest that the program had non-significant and, in any case, very limited effects (i.e. smaller than 3.5%) on housing prices in renovated neighborhoods. The program's effects on transaction volumes are also non-significant. However, the program led to a sizable upward evolution in the socio-professional status of housing buyers as compared to sellers, suggesting some improvement in the attractivity of renovated neighborhoods.