本文關(guān)鍵詞： 金融危機(jī) 預(yù)警指標(biāo) 因子分析 BP神經(jīng)網(wǎng)絡(luò)　出處：《浙江大學(xué)》2014年碩士論文　論文類型：學(xué)位論文

【摘要】：在世界經(jīng)濟(jì)發(fā)展史上,曾爆發(fā)了多次金融危機(jī),金融危機(jī)一旦爆發(fā)對世界范圍內(nèi)的經(jīng)濟(jì)市場都將產(chǎn)生不可估量的破壞,金融危機(jī)是世界經(jīng)濟(jì)市場發(fā)展史上極具破壞力的不定時(shí)炸彈。本文探尋預(yù)警金融危機(jī)的典型表現(xiàn)因子和數(shù)據(jù),以提前預(yù)測金融危機(jī)為目的,構(gòu)建了危機(jī)指標(biāo)體系,探索了因子分析法和BP神經(jīng)網(wǎng)絡(luò)預(yù)警方法兩種預(yù)警方法,并進(jìn)行了實(shí)證研究。 通過對國內(nèi)外的文獻(xiàn)研究對比分析,梳理了金融危機(jī)理論和金融危機(jī)預(yù)警方法及相關(guān)模型。對20世紀(jì)30年代的經(jīng)濟(jì)大蕭條、1997年的亞洲金融危機(jī)和2007年的美國次貸危機(jī)進(jìn)行對比研究,重點(diǎn)分析了危機(jī)爆發(fā)的原因,總結(jié)它們的表現(xiàn)形式,特別是相關(guān)經(jīng)濟(jì)指標(biāo)的變化特征,為提取和識別金融危機(jī)的預(yù)警指標(biāo)提供事實(shí)的基礎(chǔ)。 接著構(gòu)建了金融危機(jī)的預(yù)警指標(biāo)體系并進(jìn)行了初步的數(shù)據(jù)處理,即分析選取了15個(gè)預(yù)警指標(biāo),為各指標(biāo)確定了安全、基本安全、警惕、危險(xiǎn)四個(gè)等級并確定了相關(guān)指標(biāo)界限,查詢了歷年的統(tǒng)計(jì)數(shù)據(jù),獲取和計(jì)算了這些指標(biāo)的數(shù)值并進(jìn)行了歸一化處理。 之后闡述了因子分析方法的基本原理和操作流程,采用SPSS21.0軟件對所獲得的15個(gè)典型指標(biāo)的統(tǒng)計(jì)數(shù)據(jù)進(jìn)行了分析,采用主成分法提取了4個(gè)公共因子并按權(quán)系數(shù)計(jì)算了總因子,計(jì)算了因子得分系數(shù),根據(jù)該系數(shù)計(jì)算了歷年各因子的得分,并對各因子反映的經(jīng)濟(jì)狀況進(jìn)行了研究,結(jié)果表明,通過對因子得分的分析可以直觀解析當(dāng)前經(jīng)濟(jì)狀況或者預(yù)警情況,為后期的經(jīng)濟(jì)政策的制定提供參考。 最后在介紹BP神經(jīng)網(wǎng)絡(luò)的作用、模型特點(diǎn)和應(yīng)用流程之后,基于MATLAB工具箱函數(shù)進(jìn)行了BP神經(jīng)網(wǎng)絡(luò)構(gòu)建,并選擇了合適的傳遞函數(shù)和訓(xùn)練函數(shù)進(jìn)行網(wǎng)絡(luò)訓(xùn)練得出預(yù)測模型,驗(yàn)證了預(yù)測網(wǎng)絡(luò)的泛化能力有效性后,對下一年的經(jīng)濟(jì)預(yù)警程度進(jìn)行了預(yù)測,最后對預(yù)測結(jié)果進(jìn)行了評析。結(jié)果表明,本文構(gòu)建的預(yù)測方法,能有效預(yù)測下一年的經(jīng)濟(jì)情況。
[Abstract]:In the history of world economic development, there have been many financial crises. Once the financial crisis breaks out, it will cause incalculable damage to the world economic market. Financial crisis is a destructive time bomb in the history of world economic market. This paper explores the typical performance factors and data of early warning financial crisis in order to predict the financial crisis in advance. This paper constructs a crisis index system, explores two early warning methods, factor analysis and BP neural network, and makes an empirical study. Through the comparative analysis of domestic and foreign literature, this paper combs the theory of financial crisis, financial crisis warning methods and related models. On 1930s, the Great Depression. The Asian financial crisis in 1997 and the subprime mortgage crisis in the United States in 2007 were compared. The causes of the crisis were analyzed, and their forms of expression, especially the change characteristics of related economic indicators were summarized. For the extraction and identification of financial crisis warning indicators to provide a factual basis. Then it constructs the early warning index system of financial crisis and carries on the preliminary data processing, that is, analyzes and selects 15 early warning indicators, determines the security, the basic security, the vigilance for each index. The four grades of risk and the limits of related indexes are determined, the statistical data of the past years are queried, the values of these indexes are obtained and calculated, and the normalization processing is carried out. After that, the basic principle and operation flow of the factor analysis method are described, and the statistical data of the 15 typical indexes obtained are analyzed by SPSS21.0 software. Four common factors were extracted by principal component method, the total factors were calculated according to the weight coefficient, the factor score coefficients were calculated, and the scores of each factor over the years were calculated according to this coefficient. The results show that the analysis of factor scores can directly analyze the current economic situation or early warning situation, and provide a reference for the later formulation of economic policy. Finally, after introducing the function, model characteristics and application flow of BP neural network, the BP neural network is constructed based on MATLAB toolbox function. After selecting the appropriate transfer function and training function to train the network to get the prediction model, the validity of the generalization ability of the prediction network is verified, and the economic early warning degree of the next year is predicted. Finally, the prediction results are evaluated, and the results show that the prediction method can effectively predict the economic situation in the next year.
【學(xué)位授予單位】：浙江大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2014
【分類號】：TP18;F831.59

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 王春娟;馬克思的虛擬資本理論與虛擬經(jīng)濟(jì)[J];財(cái)經(jīng)問題研究;2004年11期

2 黃賾琳;;改革開放三十年中國經(jīng)濟(jì)周期與宏觀調(diào)控[J];財(cái)經(jīng)研究;2008年11期

3 崔軍;杜建偉;;全球經(jīng)濟(jì)危機(jī)背景下我國新一輪宏觀調(diào)控政策評析[J];成人高教學(xué)刊;2009年01期

4 黃力克;;馬克思經(jīng)濟(jì)危機(jī)理論與西方經(jīng)濟(jì)周期理論的比較[J];當(dāng)代經(jīng)濟(jì)研究;2006年05期

5 易培強(qiáng);;馬克思虛擬資本理論與國際金融危機(jī)[J];當(dāng)代經(jīng)濟(jì)研究;2009年01期

6 張作云;;試論金融危機(jī)的可能性和現(xiàn)實(shí)性[J];當(dāng)代經(jīng)濟(jì)研究;2010年05期

7 國紀(jì)平;;國際金融危機(jī)的成因[J];黨建;2008年12期

8 王力;美國經(jīng)濟(jì)周期的新因素[J];國際關(guān)系學(xué)院學(xué)報(bào);2005年01期

9 張立群;我國的經(jīng)濟(jì)周期問題研究[J];管理世界;1997年06期

10 范燁;趙宏潔;;虛擬資本與金融危機(jī)[J];經(jīng)營管理者;2010年18期

，

本文編號：1455914