2019第二届金融工程青年学者学术研讨会

苏州大学金融工程研究中心

为了加强金融工程领域青年学者间学术交流，促进金融工程学科的发展，由苏州大学金融工程研究中心承办的《第二届金融工程青年学者学术研讨会》，将于2019年4月25日-26日在苏州举行。本次研讨会将讨论金融工程与金融风险管理领域的前沿问题与最新研究进展，并开展CSIAM金融数学、金融工程与精算专业委员会金融工程青年委员会2019年-2020年度工作会议。

会议地点

江苏省苏州市十梓街1号苏州大学览秀楼105报告厅

住宿与报到

1. 住宿、报到地点

   苏州市姑苏区十全街滚绣坊20号南林饭店山水楼大厅

2. 报到时间：2019年4月24日12:00-21:00

3. 联系人：支康权 15152091178

钱晓松 18051296936

会议费用

本次研讨会不收取会务费，参会者交通住宿费用自理。

参会学者（按姓氏拼音顺序）

会议日程I

会议日程II

报告题目和摘要

基于大数据框架下的小微企业成长的动态演变机制评估体系建设的金融科技全息画像方法

袁先智（上海立信会计金融学院 & BBD公司）

本文的目的是全面介绍我们(BBD)公司如何在大数据框架下利用在金融科技中独立创建的“全息画像 (Hologram)”工具建立对中小型（包含小微企业）(SME)成长变化的动态演变机制进行评估体系。

本工作的核心思想是：通过使用“全息画像”（一种提供数据融合的新平台）作为工具来描述中小型企业（SMEs）在其业务动态行为方面的演变机制。在金融科技大数据环境下处理由给定实体的商业行为网络结构提供的结构化或非结构数据信息，并进行数据融合（Data Fusion），进一步从处理与商务生态系统相关的复杂金融网络结构中提取与风险相关风险基因（Risk Genes）, 从而建立对中小型企业（SME）成长“好与不好”的风险评估体系，这也是通过企业（商业）行为的网络结构信息研究中小企业发展机制的新方法。我们的研究表明，

本文介绍的针对中小企业发展演变的动力机制建立的框架可以让我们对中小企业（及其贷款申请）进行风险评估（在没有传统的会计，金融或者相关的资产等待消息情况下），这一新方法是支持普惠金融服务在金融创新方面的革命性突破。另外，以我们2016年推出的支持中小企业贷款的“Holo Credit Loan”纯信贷贷款作为案例，也解释了我们的全信画像方法为如何解决长期困扰人类对中小型企业（和个人）信贷进行信用风险评估提供了新的理论基础。　

从专业层面，我们的新方法以描述双状态的随机共振（SR）理论作为基础，用来解释“资本与产品”转换效应相关的周期性力量，以及与中小企业内部和外部资本以及风险因素环境的宽带带来的随机波动，相应的竞争和合作驱动响应力量作为驱动，使实体中小型企业（SME）根据最大化效用在两个局部稳定状态（即“资本”和“产品”状态）之间进行转换。在并存于最佳风险水平，与周期性力量协同一致的情况下，一个好的中小型企业存在周期性的发展变化，应该导致输出“光谱放大指数（SAI）”中的最大值，以及在固定单位下产生的“风险

收益率（URR）”来反映其系统风险水平。因此，中小企业商业行为的知识图谱揭示了驱动随机共振（SR）行为以及与中小企业发展成功相关的竞争与合作机制，有助于中小企业发展机制的调整和发展。这样，中小企业通过相关的投入资本和风险控制等基于关联方（Related-Parties）提供的信息，我们可以建立其成长“好与不好”的（信用）风险评估体系。这就解决了对中小型企业贷款需要的风险评估问题。

最后，我们也分享了支持中小企业贷款的“Holo Credit Loan”(好企贷)信贷贷款在实践和应用中的表现。

数据特征驱动的信用风险分类研究

余乐安（北京化工大学）

本报告主要从多维信用分类数据出发，首先回顾了信用分类的各类研究方法，然后对多维信用分类数据的特征进行系统梳理，总结出数量特征、质量特征和分布特征等三类主要特征。基于三类主要特征，提出面向分类数据的数据特征检测方案及对应特征解决方案。随后使用目前常见的三个信用数据集，对提出的数据特征检测和解决方案进行了具体实证分析。同时基于数据样本的非均衡特征、缺失性特征以及稀疏性特征，提出了一套基于数据特征驱动的机器学习模型，改进信用分类的预测结果。

基于CPU+GPU架构的期权定价算法及其库函数开发

邓明森（贵州财经大学）

随着大数据、大型模型、大型计算网格以及与之相关的大型监管规则的涌现，量化金融领域越来越意识到更大规模的计算变得越来越必要。自20世纪90年代末以来，“大金融”时代逐渐在银行业兴起，特别是在2008年金融危机时所暴露出来的金融市场中的各种深层复杂性所导致的更复杂的模型和更丰富的市场数据，它们呈现了爆炸性的增加，也需要更复杂的管理、定价及风险管理。即使时最简单的期权定价定位也因为规模的变化变得对市场数据的关键风险因素敏感度降低。以并行计算为基础的计算金融因此变得越来越重要，我们搭建了专门针对计算金融领域的高性能计算集群，并引入了适用于蒙特卡洛大规模模拟的Tesla P100 GPU卡，组建了CPU+GPU的混合架构计算系统。由于GPU的并行化处理往往需要采用CUDA等非标准的C++库，算法风格特殊，开发效率低下，同时也有不可避免的硬件开销，通常金融界认为GPU并不适用于金融计算。基于此，我们开发了适用于数据密集型及计算密集型通用的，基于GPU的随机数发生器，降低了数据传递的开销，建立以不同的Monte Carlo算法为基础的用于期权定价的算法库函数，在单块GPU卡上的期权定价数量计算规模可以达到目前难以管理的1014的尺度，并将开销降低到以秒为单位来定义。

Predictable Dynamics in the Implied Volatility Surface based on Weighted Least Squares: Evidence from Soybean Meal Futures Options in China

隋聪（大连海事大学）

This paper examines the dynamics and predictability of the implied volatility surface derived from weighed trading volume. We study the Chinese soybean meal futures options market. By assigning larger weights to options with higher trading volume, we find more precise fittings on the implied volatility surface. Our estimation method outperforms traditional methods in terms of the dynamics and predictability of the implied volatility surface, both in the sample and out of the sample. We also document that soybean meal futures options exhibit implied volatility smirk that is different from those in the stock index options. In addition, we find that the implied volatility term structure shows an inverted U-shape during the period of high implied volatility.

央行干预与市场预期：基于中国外汇衍生品市场的检验

陈蓉（厦门大学）

本文构建了独特的央行干预指标和外汇衍生品隐含预期指标，采用2014年6月5日至2017年9月30日的人民币/美元1市场的日度数据，对中国外汇市场上央行干预与市场预期之间的相互影响关系，以及在央行干预下在岸与离岸市场预期之间的相互引导关系进行了实证研究。结果表明：我国央行的干预目标既有“逆风向干预市场预期”，也有“熨平市场预期波动率”；前者是央行干预的主要目标，但随着外汇期权市场的发展，央行也开始越来越多地实行“熨平市场预期波动率”干预；央行“逆风向干预市场预期”对在岸市场是有效的，对离岸市场无效，而“熨平市场预期波动率”在两市场上均只能发挥微弱的效果；随着市场发展，在岸和离岸市场的互动关系逐渐提升，其中在岸市场对预期的主导作用逐渐增强。

Time-varying comovement and changes of comovement structure in the Chinese stock market: A causal network method

部慧（北京航空航天大学）

The driving force for the comovement in stock returns is a long-standing debate between classical asset pricing theory and behavioral finance theory. It has become critically important recently for understanding systemic risk and risk contagion in the market. In this study, we propose complex networks enabled new methods to measure the causal comovement of individual stocks and the comovement structure of the market, which facilitate the examination of all kinds of hypotheses of comovement theories in a unified framework. Using a sample of the Chinese stock market from Jan. 1, 2006 to Dec. 31, 2016, we find that the degree of comovement generally intensifies over time, with a drastic increase from 2011 to 2015, while the comovement structure of the market changes with different market situations. Most importantly, our study reveals the driving force of causal comovement among individual stocks; that is, sentiment-based factors related to the market index indeed induce excess causal comovement in returns beyond that can be justified by fundamental factors including beta coefficient, book-to-market ratio, liquidity, profitability and volatility. Our study also reveals the determinants of comovement structure, which are attributable to the change of investors’ behaviors in different periods. It turns out that investors in the Chinese stock market care about risk-return relationship in normal periods, while they seem to care only about risk in crisis periods.

G-正态分布和G-布朗运动在金融中的应用

李欣鹏（山东大学）

我们首先简要回顾了彭实戈院士提出的非线性期望理论，并着重介绍了G-正态分布这一重要的非线性分布以及由其生成的G-布朗运动，随后通过一个单边不完全信息重复博弈模型来说明股票价格由G-布朗运动的合理性，并给出了G-正态分布和G-布朗运动在金融中的一些应用。

Optimal Retirement in a General Market Environment

杨舟（华南师范大学）

We study optimal retirement, consumption/portfolio selection problem of an agent in a non-Markovian environment. We show that under a suitable condition the optimal retirement decision is to retire when the individual's wealth reaches a threshold level. We express the value and the optimal strategy by using the strong solution of the backward stochastic partial differential variational inequality (BSPDVI) associated with the dual problem. We derive properties of the value function and the optimal strategy by analyzing the strong solution and the free boundary of the BSPDVI. We also make a methodological contribution by proposing an approach to investigate properties of the strong solution and the stochastic free boundary of BSPDVI by combining a probabilistic method and the theory of backward stochastic partial differential equations (BSPDEs).

Expelling Retail Investors and Market Efficiency

吕大永（上海立信会计金融学院）

Using a unique event in China, after which retail investors were not allowed to invest in structured fund market, this paper investigates the effect of expelling retail investors on market efficiency. The Regression Discontinuity Design shows that after this exogenous event, tranche-B funds, which were most preferred by retail investors, experience a significant decrease in market efficiency. In addition, we find that transaction costs and illiquidity increases after the event, but non-informational selling activities rarely increase. We also show that retail investors are not informed traders in Chinese structured funds market. Overall, our results suggest that retail investors provide liquidity and lower transaction costs, thus improve market efficiency.

Optimal asset allocation for participating contracts under position and VaR-PI constraint

董迎辉（苏州科技大学）

Participating contracts provide a maturity guarantee for the policyholder. However, the terminal payoff to the policyholder should be related to financial risks of participating insurance contracts. We investigate an optimal investment problem under a joint Value-at-Risk and portfolio insurance constraint and position constraints faced by the insurer who offers participating contracts.The insurer aims to maximize the expected utility of the terminal payoff to the insurer. We adopt a concavification technique and a Lagrange dual method to solve the problem and derive the representations of the optimal wealth process and trading strategies.

大数据风控应用研究分享

李华（郑州大学）

大数据已上升为国家战略，以大数据为代表的信息技术正在推动着金融机构传统风控和信贷机制的转型升级，大数据风控正在为实体经济发展注入新的动力。报告将以河南省为例，分享大数据风控在金融行业的发展动态，以及需要解决的主要科学问题。

A semi-analytical formula for European options under a hybrid Heston-CIR model with regime switching

陈文婷（江南大学）

In this paper, we consider the pricing of European options under a regime switching Heston-CIR hybrid model, where the mean-reversion levels of both the stochastic volatility and interest rate are assumed to change among different states. Albeit difficult, we have still managed to derive an semi-analytical pricing formula for European options after the generalized moment generating function of this particular model is worked out. Numerical experiments are carried out to demonstrate the accuracy of the newly derived formula as well as the influence of the introduction of regime switching mechanics on option prices. Finally, through a preliminary empirical study, our model is shown to be superior to the Heston-CIR model, which demonstrates the importance of introducing the regime switching mechanics.

A Dynamic Mean-Variance Analysis

徐玉红（苏州大学）

We propose a dynamic portfolio choice model with the mean-variance criterion for portfolio log-returns. The model yields time-consistent portfolio policies and is analytically tractable even under some incomplete market settings. The portfolio policies conform with conventional investment wisdom (e.g. richer people should invest more absolute amount of money in risky assets; the longer investment time horizon, the more proportional amount of money may be invested in risky assets; and for long-term investment people should not short sell major stock indices whose returns are higher than the risk-free rate). Moreover, the model is able to elicit investors' risk prole by asking the investors to input their target annual returns in a fictional complete market.

乘车路线

（外地赴苏州可选择乘高铁至苏州高铁北站或苏州站，或乘飞机至上海虹桥机场。）