Collection operations of a national census require careful planning in order to control its costs and maximize response rates. For the 2016 Canadian Census, Statistics Canada developed a microsimulation model designed to produce forecasts of the response rates. An important part of this microsimulation model is the self-responses process which was modelled via a nonparametric survival model. A major improvement of the microsimulation model could be the use of a parametric survival model for the self-responses process. It could allow for a better understanding of the efficiency of the waves methodology used for the 2016 Census and could be adapted more easily to potential changes in the methodology for the next 2021 Canadian Census. In this talk, I will discuss the 2016 Census methodology and present a parametric discrete survival model that could be used to model the self-response process.