Textual Analysis-based Measurement of Fintech and tests of Enabling Effect for Commercial Banks
With the deep integration of digital technologies,such as big data and artificial intelligence,with financial business,the rapid development of financial technology(Fintech)is profoundly changing and reshap-ing the financial industry all over the world.In China,there are over 4600 licensed institutions in the banking industry and their total assets have accounted for more than 90％of that of all financial institutions as of 2019,and hence commercial banks are seen as the most important force for serving the real economy and driving the development of technology-enabled finance.In practice,most commercial banks have treated developing Fintech as an important strategy for realizing digital transformation and promoting financial inclusion.However,existing research pays little attention to the measuring of Fintech for each bank at the individual level,and thus there is a lack of the corresponding quantitative investigation on the enabling effect of Fintech for banks.It aims to measure the level of Fintech for commercial banks at the individual level and then investigate the enabling effects of Fintech on banks'operating performance as well as inclusive finance.Specifically,based on more than 170000 textual news publicly reported by an authoritative financial website,multiple natural language processing technologies including named entity recognition,pre-training word embedding model,and LDA topic model are used to construct two basic thesauruses about commercial bank names and banking Fintech,and then Fintech Development Index for 1566 banks from 2011 to 2019 is constructed.Using a panel data from 472 banks with complete financial data,the findings of the investigation about the enabling effects of Fintech show that Fintech can not only improve the banks'operating performance by significantly enhancing their operation ability,service ability,and risk control ability,but also is beneficial for expanding access to credit by stimulat-ing banks to increase loan supply.However,it is found that banks'Fintech fails to decrease loan rates,which indicates a"universal but not inclusive"problem in Fintech lending due to convenience premium or pricing discrimination.The contributions of the paper are threefold.Firstly,compared with the existing literature on measuring the development of Internet finance or digital finance at the national or regional level,the develop-ment of Fintech at the individual bank level is measured using publicly reported textual news and a variety of natural language processing techniques.The methodologies associated with automatedly constructing thesaurus and separation of compound Chinese words also provide an important reference for textual analysis-based research.Secondly,the mechanism of how Fintech can enable the banking industry is uncovered by a large-sample investigation on the effects of Fintech on banks'operation as well as financial inclusion.Thirdly,as researchers are beginning to discuss the negative consequences of Fintech development,supplementary evidence at the cross-bank level on the"universal but not inclusive"problem in Fintech lending is provided.
Fintechcommercial banksenabling effectnatural language processing
万方数据
维普
