| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 759.99 KB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
Esta dissertação surge com o objetivo de escolher um modelo de previsão de falências para as PME portuguesas, dada a importância destas empresas na economia portuguesa.
Achamos útil a existência de um modelo de previsão de falências dedicado às PME portuguesas, que possa auxiliar os investidores, os órgãos de gestão, etc., na sua relação com a empresa.
A análise que realizamos permitiu aferir que o modelo de Altman e Sabato (2007), com os coeficientes atualizados a partir da amostra emparelhada, permitiu classificar corretamente as PME falidas, com um ano de antecedência, apresentando uma capacidade preditiva global de 92,9 por cento, com erro tipo I (empresas falidas classificadas como ativas) de 21,2 por cento e erro tipo II (empresas ativas classificadas como falidas) de 6,8 por cento.
Para testarmos a capacidade de melhorar o modelo com a inclusão de variáveis macroeconómicas: Taxa de crescimento real do PIB; Taxa de Inflação; Taxa de Juro sobre novos empréstimos e a inclusão das variáveis qualitativas: Localização e CAE, selecionamos, da base de dados SABI, duas amostras com empresas PME da Indústria transformadora. Construímos a primeira amostra para o horizonte temporal entre 2004 e 2019 - respeitando a proporção de empresas falidas, por CAE e ano de falência - para desenvolvimento dos modelos e construímos a segunda amostra com dados de 2019 e 2020, para teste dos modelos desenvolvidos.
Por consideramos que os dados obtidos da base de dados, não refletiam a realidade, criamos novas amostras, reduzindo o horizonte temporal. Por fim criamos uma nova amostra, desta vez emparelhada (68 empresas falidas e 68 ativas, em 2019 e 2020), que nos permitiu atingir os resultados mencionados.
This dissertation comes with the objective of choosing a bankruptcy forecasting model for Portuguese SMEs, given the importance of these companies in the Portuguese economy. We find it useful to have a bankruptcy forecasting model dedicated to Portuguese SMEs, which can assist investors, management bodies, etc., in their relationship with the company. The analysis we performed allowed us to correctly assess that the Altman and Sabato (2007) model, with the coefficients updated from the paired sample, allowed us to correctly classify the failed SMEs, one year in advance, presenting an overall predictive capacity of 92.9 por cento, with type I error (bankrupt companies classified as active) of 21.2 por cento and type II error (active companies classified as bankrupt) of 6.8 por cento. To test the ability to improve the model by including macroeconomic variables: Real GDP growth rate; Inflation Rate; Interest rate on new loans and the inclusion of qualitative variables: Location and CAE, we selected, from the SABI database, two samples with SME companies in the Manufacturing Industry. We built the first sample for the time horizon between 2004 and 2019 - respecting the proportion of bankrupt companies, CAE and year of bankruptcy - for model development and we built the second sample with data from 2019 and 2020, to test the developed models. Because we consider that the data obtained from the database did not reflect reality, we created new samples, reducing the time horizon. Finally, we created a new sample, this time paired (68 bankrupt and 68 active companies in 2019 and 2020), which allowed us to achieve the results mentioned.
This dissertation comes with the objective of choosing a bankruptcy forecasting model for Portuguese SMEs, given the importance of these companies in the Portuguese economy. We find it useful to have a bankruptcy forecasting model dedicated to Portuguese SMEs, which can assist investors, management bodies, etc., in their relationship with the company. The analysis we performed allowed us to correctly assess that the Altman and Sabato (2007) model, with the coefficients updated from the paired sample, allowed us to correctly classify the failed SMEs, one year in advance, presenting an overall predictive capacity of 92.9 por cento, with type I error (bankrupt companies classified as active) of 21.2 por cento and type II error (active companies classified as bankrupt) of 6.8 por cento. To test the ability to improve the model by including macroeconomic variables: Real GDP growth rate; Inflation Rate; Interest rate on new loans and the inclusion of qualitative variables: Location and CAE, we selected, from the SABI database, two samples with SME companies in the Manufacturing Industry. We built the first sample for the time horizon between 2004 and 2019 - respecting the proportion of bankrupt companies, CAE and year of bankruptcy - for model development and we built the second sample with data from 2019 and 2020, to test the developed models. Because we consider that the data obtained from the database did not reflect reality, we created new samples, reducing the time horizon. Finally, we created a new sample, this time paired (68 bankrupt and 68 active companies in 2019 and 2020), which allowed us to achieve the results mentioned.
Description
Esta versão contém as criticas e sugestões dos elementos do júri
Keywords
Modelo preditivo Falência Logit PME Portugal Predictive model Bankruptcy Logit SMEs