【作者】 林茂；

【导师】 胡代光；

【作者基本信息】 西南财经大学 ， 政治经济学， 2010， 博士

【摘要】 2007年至2008年的次贷危机,不但引发美国的经济衰退,还殃及全球其他国家和地区。如何改革现有的金融监管框架,提高金融监管能力是各国政府和金融业监管当局近期讨论的焦点。利率风险是银行业面临的一大风险。如果通货膨胀降临,利率大幅上升,银行业将面临巨大的负面冲击,20世纪80年代至90年代美国的储蓄与贷款协会危机就是典型的案例。近年来,随着我国商业银行资产负债规模的不断扩大和利率市场化改革的逐步深入,银行账户利率风险已成为我国商业银行稳健经营所面临的实质性风险。利率是经济理论研究的重要变量,凯恩斯(John M. Keynes)和莫迪里阿尼(Franco Modigliani)都曾为利率期限结构理论做出过巨大贡献。鉴于银行风险管理对国家经济安全的重要性,近二十年来商业银行利率风险已成为经济和金融研究领域备受关注的课题之一。巴塞尔委员会2004年7月专门颁布了《利率风险管理与监管原则》,为商业银行管理银行账户利率风险提供指引。中国银监会于2009年12月25日颁布了《商业银行银行账户利率风险管理指引》,也明确要求商业银行建立银行账户利率风险管理框架体系。这些文件的颁布和实施,必将推动银行利率风险管理研究的进一步发展。本文根据我国商业银行的具体情况,前瞻性地探索商业银行的银行账户利率风险暴露的计量方法,在此基础上整合分析银行资产负债结构与净利息收入和经济价值的利率风险之间的关系。研究方法方面,比较分析和情境模拟是本文使用的两种主要方法。由于银行利率风险管理涉及净利息收入和经济价值两个目标,研究中又使用了多种计量技术和不同期限的样本数据,通过比较分析有利于发现不同模型的适用条件,以及利率波动对两个目标的不同影响。基于特定模型的情景模拟,有助于预测最坏情景下银行的潜在损失。技术路线方面,基于实证研究发现问题、理论模型解释问题的逻辑,本文分为层层递进的三步：收益率曲线动态的估计以及模型预测能力的检验；基于收益率曲线动态的银行利率风险暴露的计量、比较和验证；建立分析框架模型进一步揭示资产负债结构和银行利率风险的内在关系。本文研究的主要内容如下：首先,基于我国收益率曲线动态对短期利率模型进行实证研究。第一步,基于CKLS嵌套模型框架,对不同的单因素短期利率模型进行时间序列估计。第二步,针对Vasicek模型和CIR模型,进一步使用特定时刻的实际收益率曲线对模型进行横截面的校准,校准时保留了部分来自时间序列估计的信息。第三步,对校准后的模型进行样本外的检验。研究结果表明：模型拟合效果较好,校准得到的参数也比较稳定。但是前推预测的天数越长,预测的误差越大。因此,即使是样本内拟合度很高的模型,随时间的推移也需要再校准。其次,基于我国收益率曲线动态进行主成分分析的实证研究。第一步,对收益率曲线的每日变化进行主成分分析。第二步,对主成分变化序列进行核密度估计。第三步,对估计的模型进行样本外的包络性检验。研究结果表明：不同年限样本的主成分系数存在较大差异。主成分变化的分布呈现明显的厚尾特征,正态分布假设可能会低估利率风险。非参数主成分模型的预测范围总体上能包络正态假设模型,能更有效地反映极端情景下的利率风险。再次,研究银行净利息收入利率风险的计量。银行一年内净利息收入的利率风险,属于短期效应。第一步,以我国五家商业银行2007年年底的重新定价缺口数据为样本,使用标准冲击方法模拟净利息收入可能的下降值。第二步,使用Vasicek和CIR模型模拟净利息收入可能的下降值。第三步,根据净利息收入风险对银行排序。结果表明：如果不扣除活期存款和存款准备金这些利率变动频率很低、变动幅度很小的负债和资产,模拟的结果和实际的情况不相符。扣除以后,所有银行净利息收入面临的都是利率下降的风险。同时发现,Vasicek和CIR模型预测的净利息收入风险,要小于标准冲击方法预测的结果,这与2007底收益率曲线处于低位有关。基于净息差变化对银行净利息收入风险进行排序,发现不同模型的排序结果非常一致。接下来,研究银行经济价值利率风险的计量。经济价值分析法是针对利率变动的潜在长期影响的评价,更值得银行管理者和监管当局的关注。第一步,使用标准久期方法模拟经济价值可能的下降值。第二步,使用Vasicek模型和CIR模型,以及主成分分析模型和历史模拟方法,模拟经济价值可能的下降值。第三步,根据经济价值风险大小对银行进行排序。研究结果表明：五家银行的经济价值面临的是利率上升的风险。使用2年和3年历史数据估计的短期利率模型预测的经济价值风险,都小于标准久期方法的结果。主成分VaR和历史模拟VaR的研究发现,非正态模拟的经济价值风险都要大于正态模拟的结果,反映了主成分变化的厚尾特征。标准久期方法预测的经济价值风险,介于2年和3年历史样本的非正态主成分1年持有期VaR的预测结果之间。从银行经济价值风险排序结果看来,正态和非正态主成分VaR的排序结果与标准久期方法非常一致,与历史模拟也比较接近,但与Vasicek模型和CIR模型的排序结果差距较大。最后,是利率风险管理目标的一致性问题的研究。第一步,建立一个框架模型,分析银行资产负债的缺口结构与净利息收入和经济价值的利率风险之间的关系。第二步,进一步探讨模型的政策含义。研究结果表明：针对我国商业银行的典型情况,面对同样的利率冲击,短期净利息收入和经济价值这两个目标的表现不一致,资产负债管理策略的效果也不一致,监管当局的监管目的和银行管理者的经营目的容易出现背离,监管难度较大。本文的研究主要有以下创新：整合研究银行净利息收入和经济价值的利率风险。当利率变化时,银行的净利息收入和经济价值的变化方向是否一致,对银行的管理者和监管当局很重要。整合研究净利息收入和经济价值的风险,有助于银行管理者统筹管理利率风险,也有利于监管当局制定有效的监督检查方案。使用主成分VaR方法,逐日盯市管理经济价值的利率风险。本文使用正态主成分VaR和非正态主成分VaR的模拟方法,逐日基础上计量经济价值风险,并进行返回检验,有助于银行逐日盯市管理经济价值的利率风险。建立分析框架模型,直观地揭示银行资产负债结构与利率风险的关系。模型只使用一个变量,1年内缺口值占总缺口值的比率,就能很好地表征不同的资产负债缺口结构,有助于揭示银行资产负债结构与利率风险的内在联系。总之,商业银行的银行账户利率风险,具有风险来源多样性、风险影响多重性、风险计量复杂性等特点。由于知识和时间有限,本文的研究还存在一些局限性,有待进一步改进和提高。更多还原

【Abstract】 From 2007 to 2008, the outbreak of the sub-prime mortgage crisis, not only lead to economic recession in the United States, but also threaten the economic security of other countries and regions. How to reform the financial regulatory framework and improve the financial supervision is recently a hot topic among governments and the financial supervisory authorities. Interest rate risk (IIR) is a major risk in the banking sector. If inflation comes, interest rates will increase substantially, the banking sector will suffer a huge adverse impact. Savings and Loan Associations Crisis in the United States from 1980s to 1990s is a typical case. In recent years, as China’s commercial banks continue to expand the size of assets and liabilities, and the reform of the market-oriented interest rate gradually make progress, interest rate risk in the banking book has become the substantial risk.Interest rates are important variables in the economic theory. John M. Keynes and Franco Modigliani have made tremendous contributions to the theory of the term structure of interest rates. In view of the importance of bank risk management on national economic security, interest rate risk of the commercial banks has become one of the topics of concern in the economics and finance in last two decades. Basel Committee on Banking Supervision in July 2004 promulgated the "Principles for the Management and Supervision of Interest Rate Risk", in order to guide the management of interest rate risk in the banking book of the commercial banks. China Banking Regulatory Commission on December 25,2009 also promulgated the "Guidelines for the Management of Interest Rate Risk in the Banking Book of Commercial Banks", requesting the commercial banks to establish the framework of interest rate risk management in the banking book. The issue and implementation of these documents will further promote the study of interest rate risk management of the commercial banks.The dissertation is studying the measurement of interest rate risk in the banking book based on the specific situation of China’s commercial banks, and the the relationship between the structure of asset-liability and interest rate risk exposure exposure to banks’net interest income (NII) and economic value of equity (EVE). Comparative analysis and scenario simulations are the two major methodologies. Because banks’interest rate risk includes net interest income risk and economic value risk, and a variety of measurement techniques and different sample data are applied, comparative analysis will be an effective method to find out which measurement techniques are more applicable, and how the interest rate changes influence net interest income and economic value in different way. Scenarios simulations based on the specific models will help measure the banks’ potential loss in the worst-case. Based on the logic of finding problems through empirical studies and analyzing problems through theoretical models, the study has three steps:the estimation of the yield curve dynamics and the test of the predictability of the models; Measurement, comparison, validation of the banks’ exposure to interest rate risk in terms of the different yield curve dynamics models; Establishment of an analytical framework model to further reveal the relationship between the structure of asset-liability and interest rate risk.The main contents of the dissertation are as follows:Firstly, the short-term interest rate models are empirically estimated using China’s yield curve dynamics. The first step, the different single-factor short-term interest rate models are estimated time-seriously under the CKLS nested model framework. The second step, Vasicek model and CIR model are further calibrated cross-sectionally with the actual yield curve on specific dates, but retain some of the information from the time-series estimates. The third step, the models after calibration are tested with out-of-sample data. The results showed that:the models fit well, and the parameters obtained from calibration are also relatively stable. However, the longer is forecasted, the greater the prediction error. Thus, even the models fit well with in-sample data, it is also necessary to recalibrate the models with the passage of time.Secondly, China’s yield curve dynamics are estimated using Principal Component Analyses (PCA). The first step, the daily yield curve changes are estimated using PCA. The second step, the PCA series distribution is estimated with kernel density estimation. The third step, the envelope abilities of the estimated PCA models are tested using out-of-sample data. The results showed that: There exists a huge difference among the PCA coefficients estimated with the sample of the different periods. The changes in PCA series showed significantly the distribution of heavy-tailed characteristics. Therefore, the normal distribution assumption may underestimate the interest rate risk. The scope of the prediction using non-parametric PCA models, to better reflect the extreme scenarios of interest rate risk, can overall envelop the scope of the prediction using normality assumption models.Thirdly, a study of the measurement of the interest rate risk exposure to banks’net interest income (NII). Changes in banks’net interest income to interest rate fluctuation in one year are only short-term effects. The first step, compute the potential decline in net interest income following Basel standardized interest rate shock, using China’s five commercial banks’re-pricing gap data in the end of 2007. The second step, compute the potential decline in net interest income through the simulation using Vasicek model and CIR model. The third step, rank five banks in terms of their interest rate risk exposure to net interest income. The results show that:If don not remove demand deposits and legal deposit reserve out of the re-pricing gaps, whose interest rate doesn’t adjust often and adjustment is very small, simulation results are very different from the actual situation. After deduction, all banks’net interest income is subject to the risk of the fall in interest rates. Also found that, the net interest income risk predicted by Vasicek model and CIR model is less than that by standardized interest rate shock approach, because yield curve remains very low at the end of 2007. Ranked five banks in terms of the declines in net interest margin (NIM), and found the ranking results with different models are very consistent.Fourthly, a study of the measurement of the interest rate risk exposure to banks’economic value of equity (EVE). Changes in economic value may stand for the potential long-term impact of interest rate shock to banks, to which bank managers and banking supervisory authorities should pay more attention. The first step, compute the potential decline in economic value following Basel standard duration approach. The second step, compute the potential decline in economic value through the simulation using Vasicek model and CIR model, as well as the PCA model and the historical simulation method. The third step, rank five banks in terms of their interest rate risk exposure to economic value. The results showed that:All five banks’economic value is subject to the risk of the rise in interest rates. The economic value risk predicted by Vasicek model and CIR model, using 2-year or 3-year historical data sample, is less than that by standard duration approach. The studies of the principal component VaR and historical simulation VaR showed that economic value risk predicted by non-normal simulation to be greater than by normal simulation, reflecting the heavy-tailed distribution of PCA series. The economic value risk predicted by standard duration approach is between results predicted by non-normal principal component 1-year holding period VaR using 2-year and 3-year historical data sample. Ranked five banks in terms of the economic value risk, and found the ranking results predicted by historical simulation, normal and non-normal VaR models, as well as standard duration approach, are very consistent. But the ranking results predicted by Vasicek model and CIR model are very different.Finally, a study of the consistency of the objectives of interest rate risk management. The first step, establish a framework model to analyze the relationship between re-pricing gap structure of bank’s assets and liabilities, and interest rate risk exposure to net interest income and economic value. The second step, further explore the policy implications of the framework model. The results show that:For the typical situation of China’s commercial banks, when interest rate fluctuates, the changes in the direction of short-term net interest income and economic value are inconsistent; the effect of asset-liability management strategy is also inconsistent. In this case, the purposes of supervisory authorities and bank managers prone to departure, and then increase the difficult in banking supervision.The innovations of the dissertation are as follows:Studying integratedly the interest rate risk exposure to banks’net interest income and economic value. When the interest rate changes, it is very important for bank managers and banking supervisory authorities whether the changes in the direction of net interest income and economic value are consistent. Studying the impact of net interest income and economic value simultaneously will be helpful for bank managers’integrated management of interest rate risk, and for banking supervisory authorities’effective supervision programs. It’s helpful to manage the interest rate risk exposure to banks’economic value through marking to market, after introducing the principal component VaR method. It’s helpful to manage the interest rate risk exposure to banks’economic value through day to day marking to market, using principal component normal or non-normal simulation VaR with back-testing.An analytical framework model is established to reveal the relationship between the structure of asset-liability and interest rate risk. Only one variable, the ratio of one-year cumulative gaps to total cumulative gaps, could effectively stand for different gaps structure of assets and liabilities. This model could intuitively reveal the relationship between the structure of banks’asset-liability and interest rate risk.In conclusion, the interest rate risk exposure to commercial banks’banking book has a diversity of sources and multiple impacts. The measurement of the risk is also very hard. As knowledge and time constraints, there are some limitations. A lot of work should be done in the future study.更多还原