Neural networks versus multiple regression : can they explain strategic performance?

• Autor(a) principal: Gonçalves, Vítor da Conceição
• Data de Publicação: 2002
• Outros Autores: Reis, António Palma dos, Duque, João
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10400.5/23613
• topic: Neural Networks; Statistical Methods; Multiple Regression; Strategic Performance; Financial Services; Banking Industries; Insurance; Portugal

This paper evaluates how effective are neural networks versus multiple regression on identifying the economic and financial variables that determine the goodness of the corporate strategic choices and releases the results obtained on the evaluation of strategies. As research sample, we used the strategic decisions taken by the Portuguese financial services industry (banks and insurance companies). Based on the data collected by a questionnaire that was sent to all of the Portuguese financial firms, we were able to run several linear regressions, as well as neural networks, towards justifying firms performance based on their strategic choices, which were decomposed into four dimensions: technology adoption, strategic alliances, geographic based strategies and attention focused on competitive aspects. Even though the relevance of the four strategic dimensions was observable through linear regression analysis, neural networks always provided an R-square several points above the equivalent multiple linear regression.