Abstract: 【Purpose/significance】Entity relation extraction in financial field is the basis of constructing financial knowledge base and plays an important role in the utilization of text information in financial field. This paper proposes a joint extraction model of entity relations in the financial field, which increases the recognition of complex overlapping relationships of financial texts, and can effec⁃ tively avoid the erroneous delivery between different tasks in the traditional pipeline model.【Method/process】This paper constructs a high-quality financial text corpus, proposes a new tagging scheme and entity relation matching rule, constructs an end-to-end se⁃ quence annotation model based on BERT and combines BiGRU and CRF, and realizes the joint extraction of entity relation.【Result/ conclusion】Experiment with text data in the financial field, the experimental results show that the F1-score of the relationship extrac⁃ tion and overlapping relationship extraction of the model proposed in this paper reach 0.627 and 0.543 respectively, which preliminari⁃ ly verifies the validity of the joint extraction model proposed in this paper for entity relationship extraction in the Chinese financial field.【Innovation/limitation】Combining with the characteristics of financial text, a new sequential annotation pattern is proposed and a joint extraction model of entity relations in financial domain based on BERT is constructed, which realizes the recognition of overlap⁃ ping relations between entities in financial text.