Rättskraft och korrektiv : En förvaltningsrättslig studie

Sammanfattning: In Swedish administrative law, there are numerous opportunities for authorities and others who fulfill public tasks to correct decisions. As an important limitation of this power, there are general principles on the immovability of administrative decisions. As a general rule, a favorable decision may not be corrected. Some would say that this position is based on the principle of protection of legitimate expectations, a principle which is recognized in many legal systems. Individuals should be able to rely on the fact that decisions rendered by authorities will not suddenly be withdrawn.This thesis examines rules and principles on the immovability of administrative decisions (res judicata) and the conditions under which such decisions can be corrected by the authorities. The issue of res judicata in general administrative law has not been investigated deeply in jurisprudence over the past sixty years. However, Swedish administrative law has been influenced by the administrative codifications in the 1970s, the expanding public sector and legislation, the introduction of the mandatory two-party procedure and by Sweden’s EU-membership in the 1990s. An overarching purpose of the thesis is therefore to investigate the extent to which the concept of res judicata has been affected by these factors or events. Further, the thesis takes up the issue of concept formation, both in terms of res judicata and in terms of the correction rules, which have a variety of designations such as: amendment; review; modification; and withdrawal.It can be concluded that it is, and has been, difficult to agree on the content of the res judicata concept. This is also manifested in the absence of a general res judicata rule in the Administrative Procedure Act (1986:223). Based on the investigation undertaken and the results that emerged, a possible categorization of the res judicata concept is presented. The categorization made comprises three different models. Finally, an integration of the models is recommended.