The article is an epistemological examination of the concept of ‘border’ in economics through an analysis of the relationships of economics with other social sciences in the 21st century. Taking developing countries as an example, it focuses on two sub-fields of mainstream economics, i.e. development economics and the economics of institutions. Anthropology and political science are, for their part, among the social sciences that have taken developing countries as a major subject. The article is therefore centred on these social sciences, as their focuses exhibit substantial intersections with those of development and institutional economics. Firstly, it highlights the conceptual elements of a ‘golden age’: e.g., theories recognised both by economics and anthropology fostered seminal conceptual exchanges until the 1990s. It examines the epistemic weakening of social sciences from the end of the 20th century onwards, while economics extended its investigations to phenomena previously situated out of its scope (social norms, political phenomena, cognitive representations). Supported by a substitution of techniques for theoretical conceptualisations, mainstream economists have thus increasingly seen themselves as analysing the core concepts of other social sciences with more rigour than these social sciences, thus legitimating the ‘absorption’ of social sciences by economics. In a second step, the article demonstrates the epistemological flaws of this extension of economics to other social sciences ‘core’ concepts. Mainstream economics’ exclusive focus on causalities describable by quantifiable notions misses the crucial point that concepts are in essence composite and relational constructions built over time by a given (social) science. This extension of economics is also driven by epistemic confusion, because social sciences’ concepts cannot be reduced to their measurable attributes (‘variables’). The article takes as an empirical example the institutions that organise membership and shows that concepts such as ‘institution’ or ‘rule’ are not measurable variables. For the analysis of the economic dimensions of such concepts, holistic social sciences remain epistemologically superior.< Réduire