Publications-Theses
Article View/Open
Publication Export
-
題名 基於公開新聞資訊的違約預警模型
A default prediction model based on soft information from news作者 呂朋怡
Lu, Peng-I貢獻者 陳威光<br>江彌修
呂朋怡
Lu, Peng-I關鍵詞 情感分析
媒體情緒指標
Sentiment analysis
Media sentiment indicator日期 2018 上傳時間 7-Aug-2019 16:10:11 (UTC+8) 摘要 過去違約預警模型之相關研究大多以總體經濟變數、財務比率變數、市場變數等量化資訊進行分析，然而公開之新聞資訊中往往隱含許多有價值的訊息，因此，本研究嘗試爬取美國著名報社之新聞進行情感分析(sentiment analysis) ，剖析公開新聞資訊中與樣本公司相關之新聞內涵，建立基於情感傾向和新聞報導量的媒體情緒量化指標，並採用 logit 模型建構財務預警模型，將媒體情緒量化指標與過去文獻常用之指標進行效能驗證與比較，觀察公開新聞資訊是否能夠提升預測公司違約之準確率。\n實證結果顯示，納入基於公開新聞資訊之媒體情緒量化指標，有助於提升違約預警模型之效能，也就是說，能夠提升模型預測違約發生之準確度;本研究亦進行兩種違約臨界值之比較，Begley, Ming, and Watts (1996) 建議之最適臨界值在模型之效能上，較一般文獻使用之 0.5 臨界值能夠明顯降低型一誤差，因此擁有較好的分類效果。
As most of the relevant studies of default prediction models are based on quantitative information such as financial ratio variables and market variables, public news information actually contains lots of valuable information. This paper uses Sentiment Analysis technology to extract the information related to sample companies from open news, quantifying the media sentiment indicator, and use Logit model and Probit model to conduct default prediction models. The open news is from The New York Times website and the US Newsstream database which contains Los Angeles Times, Chicago Tribune and The Wall Street Journal, I use VADER (Hutto, 2014) as the main tool of sentiment analysis, and create media sentiment indicator according to the amount of news and sentiment tendency.\nThe empirical results show that the media sentiment indicator does improve the effectiveness of the prediction of the corporate default probability. I also compare the effectiveness of the models with two different cut-off points. The optimal cut-off point suggested by Begley, Ming, and Watts (1996) create significantly lower Type I error than the general 0.5 cut-off point, so it has a better classification effect.參考文獻 Altman, Edward I, (1968). Financial ratios, discriminant analysis and the prediction of corporate bankruptcy, Journal of Finance , 23, 589-609.\nAntweiler, W. and M. Z. Frank, (2004). Is all that talk just noise? The information\ncontent of internet stock message boards, Journal of Finance, 59, 1259-1294.\nBegley, J., J. Ming, and S. Watts, (1996). Bankruptcy classification errors in the\n1980s: An empirical analysis of Altman’s and Ohlson’s models, Review of Accounting Studies, 1(4), 267-284.\nBharath, S.T., and Shumway, T., (2008). Forecasting default with the Merton distance to default model, Review of Financial Studies, 21, 1339-1369.\nCrosbie, Peter J., (1999). Modeling default risk. mimeo, KMV Corporation, San\nFrancisco, CA.\nCampbell, J.Y., Hilscher, J., and Szilagyi, J., (2008). In search of distress risk, Journal of Finance, 63, 2899-2939.\nDuffie, D., Saita, L., and Wang, K., (2007). Multi-period corporate default prediction\nwith stochastic covariates, Journal of Financial Economics, 83, 635-665.\nDuan, J.C., Sun, J. and Wang, T., (2012). Multiperiod corporate default prediction-A\nforward intensity approach, Journal of Econometrics, 170, 191-209.\nEspahibodi, P., (1991). Identification of problem bank and binary choice models,\nJournal of Banking and Finance, 15, 53-71.\nForss, T. and Sarlin, P., (2017). News-sentiment networks as a company risk indicator, Under review at Journal of Network Theory in Finance.\nHutto, C.J. and Gilbert, E.E, (2014). VADER: A parsimonious rule-based model for\nsentiment analysis of social media text, Eighth International Conference on Weblogs and Social Media (ICWSM-14), Ann Arbor, MI, June 2014.\nLopatta, K., Gloger, M.A. and Jaeschke, R., (2017). Can language\npredict bankruptcy? The explanatory power of tone in 10‐K filings, Accounting Perspectives 16(4), 315-343.\nOhlson, J.A., (1980). Financial ratios and the probabilistic prediction of bankruptcy,\nJournal of Accounting Research, 18, 109-131.\nOdom, M.D. and R. Sharda, (1990). A neural network model for bankruptcy\nprediction, IEEE/INNS, 2, 163-168.\nSingh, L., Scheuermann, P. and Chen, B., (1997). Generating association rules from\nsemi-structured documents using an extended concept hierarchy, ACM IKM, 193-200.\nShumway, T., (2001). Forecasting bankruptcy more accurately: a simple hazard model, Journal of Business, 74, 101-124.\n王懷德，(2003)。KMV模型於國內未上市、未上櫃之公開發行公司之研究。未出版之碩士論文，東吳大學，會計學系研究所。\n古永嘉、陳達新、陳維寧與楊延福，(2007)。以會計資訊及盈餘品質指標預測企業信用風險之研究-區別分析與類神經網路模型之比較。管理科學研究，4(1)，43-60。\n李沃牆，(2010)。考量總體變數下的公司違約預測—遺傳規畫決策樹與Logit模型之比較。台灣銀行季刊，61(1)，168-186。\n金雪軍、祝宇與楊曉蘭，(2013)。網絡媒體對股票市場的影響——以東方財富網股吧為例的實證研究。新聞與傳播研究，12，36-51。\n林長瑞，(2010)。預測財務危機公司樣本配置之研究。管理與資訊學報，15，1-30。\n唐麗英與林麗甄，(2009)。應用驗證因素分析與Logit模型建構台灣中小企業之財務預警模型。中華管理評論國際學報，12(1)，1-23。\n游和正、黃挺豪與陳信希，(2012)。領域相關詞彙極性分析及文件情緒分類之研究。中文計算語言學期刊，17(4)，33-47。\n蔡璧徽與黃鈺萍，(2010)。財務危機預測模型之比較分析。當代會計，11(1)，51-78。\n盧陽正、魏裕珍、張倉耀與廖婉茹，(2012)。公開新聞之資訊內涵能否增進臺灣企業信用評級慣用指標的預測能力？。台灣金融財務季刊，13(4)，27-53。 描述 碩士
國立政治大學
金融學系
105352017資料來源 http://thesis.lib.nccu.edu.tw/record/#G0105352017 資料類型 thesis dc.contributor.advisor 陳威光<br>江彌修 zh_TW dc.contributor.author (Authors) 呂朋怡 zh_TW dc.contributor.author (Authors) Lu, Peng-I en_US dc.creator (作者) 呂朋怡 zh_TW dc.creator (作者) Lu, Peng-I en_US dc.date (日期) 2018 en_US dc.date.accessioned 7-Aug-2019 16:10:11 (UTC+8) - dc.date.available 7-Aug-2019 16:10:11 (UTC+8) - dc.date.issued (上傳時間) 7-Aug-2019 16:10:11 (UTC+8) - dc.identifier (Other Identifiers) G0105352017 en_US dc.identifier.uri (URI) https://ah.lib.nccu.edu.tw/item?item_id=143189 - dc.description (描述) 碩士 zh_TW dc.description (描述) 國立政治大學 zh_TW dc.description (描述) 金融學系 zh_TW dc.description (描述) 105352017 zh_TW dc.description.abstract (摘要) 過去違約預警模型之相關研究大多以總體經濟變數、財務比率變數、市場變數等量化資訊進行分析，然而公開之新聞資訊中往往隱含許多有價值的訊息，因此，本研究嘗試爬取美國著名報社之新聞進行情感分析(sentiment analysis) ，剖析公開新聞資訊中與樣本公司相關之新聞內涵，建立基於情感傾向和新聞報導量的媒體情緒量化指標，並採用 logit 模型建構財務預警模型，將媒體情緒量化指標與過去文獻常用之指標進行效能驗證與比較，觀察公開新聞資訊是否能夠提升預測公司違約之準確率。\n實證結果顯示，納入基於公開新聞資訊之媒體情緒量化指標，有助於提升違約預警模型之效能，也就是說，能夠提升模型預測違約發生之準確度;本研究亦進行兩種違約臨界值之比較，Begley, Ming, and Watts (1996) 建議之最適臨界值在模型之效能上，較一般文獻使用之 0.5 臨界值能夠明顯降低型一誤差，因此擁有較好的分類效果。 zh_TW dc.description.abstract (摘要) As most of the relevant studies of default prediction models are based on quantitative information such as financial ratio variables and market variables, public news information actually contains lots of valuable information. This paper uses Sentiment Analysis technology to extract the information related to sample companies from open news, quantifying the media sentiment indicator, and use Logit model and Probit model to conduct default prediction models. The open news is from The New York Times website and the US Newsstream database which contains Los Angeles Times, Chicago Tribune and The Wall Street Journal, I use VADER (Hutto, 2014) as the main tool of sentiment analysis, and create media sentiment indicator according to the amount of news and sentiment tendency.\nThe empirical results show that the media sentiment indicator does improve the effectiveness of the prediction of the corporate default probability. I also compare the effectiveness of the models with two different cut-off points. The optimal cut-off point suggested by Begley, Ming, and Watts (1996) create significantly lower Type I error than the general 0.5 cut-off point, so it has a better classification effect. en_US dc.description.tableofcontents 第一章 緒論 1\n第一節 研究動機與背景 1\n第二節 研究目的 2\n第二章 文獻探討 4\n第一節 信用違約模型之相關研究 4\n第二節 文字探勘與情感分析技術 8\n第三章 研究方法 10\n第一節 樣本前測 10\n第二節 違約預警模型 11\n第三節 模型績效 14\n第四章 資料來源與處理過程 17\n第一節 Duan變數 17\n第二節 媒體情緒指標 21\n第五章 實證分析 29\n第一節 樣本前測 29\n第二節 建構違約預警模型 33\n第三節 評估模型績效 36\n第六章 結論與建議 42\n參考文獻 44 zh_TW dc.format.extent 2333963 bytes - dc.format.mimetype application/pdf - dc.source.uri (資料來源) http://thesis.lib.nccu.edu.tw/record/#G0105352017 en_US dc.subject (關鍵詞) 情感分析 zh_TW dc.subject (關鍵詞) 媒體情緒指標 zh_TW dc.subject (關鍵詞) Sentiment analysis en_US dc.subject (關鍵詞) Media sentiment indicator en_US dc.title (題名) 基於公開新聞資訊的違約預警模型 zh_TW dc.title (題名) A default prediction model based on soft information from news en_US dc.type (資料類型) thesis en_US dc.relation.reference (參考文獻) Altman, Edward I, (1968). Financial ratios, discriminant analysis and the prediction of corporate bankruptcy, Journal of Finance , 23, 589-609.\nAntweiler, W. and M. Z. Frank, (2004). Is all that talk just noise? The information\ncontent of internet stock message boards, Journal of Finance, 59, 1259-1294.\nBegley, J., J. Ming, and S. Watts, (1996). Bankruptcy classification errors in the\n1980s: An empirical analysis of Altman’s and Ohlson’s models, Review of Accounting Studies, 1(4), 267-284.\nBharath, S.T., and Shumway, T., (2008). Forecasting default with the Merton distance to default model, Review of Financial Studies, 21, 1339-1369.\nCrosbie, Peter J., (1999). Modeling default risk. mimeo, KMV Corporation, San\nFrancisco, CA.\nCampbell, J.Y., Hilscher, J., and Szilagyi, J., (2008). In search of distress risk, Journal of Finance, 63, 2899-2939.\nDuffie, D., Saita, L., and Wang, K., (2007). Multi-period corporate default prediction\nwith stochastic covariates, Journal of Financial Economics, 83, 635-665.\nDuan, J.C., Sun, J. and Wang, T., (2012). Multiperiod corporate default prediction-A\nforward intensity approach, Journal of Econometrics, 170, 191-209.\nEspahibodi, P., (1991). Identification of problem bank and binary choice models,\nJournal of Banking and Finance, 15, 53-71.\nForss, T. and Sarlin, P., (2017). News-sentiment networks as a company risk indicator, Under review at Journal of Network Theory in Finance.\nHutto, C.J. and Gilbert, E.E, (2014). VADER: A parsimonious rule-based model for\nsentiment analysis of social media text, Eighth International Conference on Weblogs and Social Media (ICWSM-14), Ann Arbor, MI, June 2014.\nLopatta, K., Gloger, M.A. and Jaeschke, R., (2017). Can language\npredict bankruptcy? The explanatory power of tone in 10‐K filings, Accounting Perspectives 16(4), 315-343.\nOhlson, J.A., (1980). Financial ratios and the probabilistic prediction of bankruptcy,\nJournal of Accounting Research, 18, 109-131.\nOdom, M.D. and R. Sharda, (1990). A neural network model for bankruptcy\nprediction, IEEE/INNS, 2, 163-168.\nSingh, L., Scheuermann, P. and Chen, B., (1997). Generating association rules from\nsemi-structured documents using an extended concept hierarchy, ACM IKM, 193-200.\nShumway, T., (2001). Forecasting bankruptcy more accurately: a simple hazard model, Journal of Business, 74, 101-124.\n王懷德，(2003)。KMV模型於國內未上市、未上櫃之公開發行公司之研究。未出版之碩士論文，東吳大學，會計學系研究所。\n古永嘉、陳達新、陳維寧與楊延福，(2007)。以會計資訊及盈餘品質指標預測企業信用風險之研究-區別分析與類神經網路模型之比較。管理科學研究，4(1)，43-60。\n李沃牆，(2010)。考量總體變數下的公司違約預測—遺傳規畫決策樹與Logit模型之比較。台灣銀行季刊，61(1)，168-186。\n金雪軍、祝宇與楊曉蘭，(2013)。網絡媒體對股票市場的影響——以東方財富網股吧為例的實證研究。新聞與傳播研究，12，36-51。\n林長瑞，(2010)。預測財務危機公司樣本配置之研究。管理與資訊學報，15，1-30。\n唐麗英與林麗甄，(2009)。應用驗證因素分析與Logit模型建構台灣中小企業之財務預警模型。中華管理評論國際學報，12(1)，1-23。\n游和正、黃挺豪與陳信希，(2012)。領域相關詞彙極性分析及文件情緒分類之研究。中文計算語言學期刊，17(4)，33-47。\n蔡璧徽與黃鈺萍，(2010)。財務危機預測模型之比較分析。當代會計，11(1)，51-78。\n盧陽正、魏裕珍、張倉耀與廖婉茹，(2012)。公開新聞之資訊內涵能否增進臺灣企業信用評級慣用指標的預測能力？。台灣金融財務季刊，13(4)，27-53。 zh_TW dc.identifier.doi (DOI) 10.6814/NCCU201900279 en_US
