The collective model of labour supply opened the household “black box” and allowed for individual treatment of partners in couples. However, the literature on labour supply has so far largely ignored a broader issue with special relevance to transition and developing countries – the distinction of single versus multi-family (“complex”) households. We propose a method to account for multi-family household structure by borrowing from recent applications of the collective model and combining estimation and calibration to identify the degree of resource sharing. We assume that each household is characterised by a between-family sharing parameter, which is calibrated on estimated preferences, the observed labour market status and other characteristics. The key identifying assumption is that preferences over income and leisure of specific family types living in single and multi-family households are the same conditional on observable characteristics. We apply the method to Polish labour market data.