This paper aims at analysing the existence of neighbourhood effects in welfare transitions. We do so by analysing the effect of being surrounded by individuals transiting in and out of social assitance, but also other environment characteristics. Our identification strategy follows three steps. We first apply the approach developed by Bayer, Ross and Topa (2008) by focusing on an exogenous source of variation with respect to the neighbours, provided by very fine granularity of the data. We then extend this strategy to check whether interactions may take place at higher level, that is at city level or local labour market level. Finally, we use an IV strategy based on the shift-share approach to examine whether diversity at the level of local labour market mediate social assistance transitions. We apply these methods to Luxembourg administrative social security data available from 2001-2015 which provides precise information at a small geographical unit – the postal code. Preliminary main findings highlight a similar positive impact of the behaviours of the neighbours on both transitions. However, such similarity vanished when dealing with impact of the specificities of the environment.