供应链融资业务中钢材质押贷款动态质押率设定的VaR方法

异于债券、股票等质押融资业务,存货质押业务动态质押的核心在于预测其长期价格风险。从分析存货质押市场收益率的统计特征出发,以场外现货交易为主的钢材((HRB335)日数据为例,建立能刻画钢材收益率序列异方差性和尖峰厚尾特性的VaR-GARCH(1,1)-GED模型。同时,提出置于多风险窗口下运用样本外预测未来质押期内钢材价格风险水平,给出厚尾分布下长期风险VaR的计算解析式,得出与银行风险承受能力相一致的质押率。进而,基于失效率法则建立长期风险的碰撞序列函数,回测多风险窗口下长期VaR值。实证分析显示,模型得到的质押率在控制好风险的同时降低了效率损失,为商业银行提供一种动态质押率的风险管理模式和框架。

VaR Models for Setting Dynamic Impawn Rate of Steel in Inventory Financing of Supply Chain Finance

Inventory financing,as one of the main business models of supply chain finance in China,uses inventory as a pledge to mitigate credit risk of loans.However,in current banking practice,the impawn rate of inventory is still determined through experiences of banks,which might not be aligned with the risk tolerance level of banks.Therefore,setting impawn rate is crucial not only to the risk control of supply chain finance but also to the development of supply chain finance itself.The current literature primarily focuses on theoretical modeling and cases based on individual samples.However,the empirical analysis based on a large number of samples is scarce and most of the existing research set impawn rate in impawn periods statically. Taking into account of macroeconomic environment,credit level counterparty,liquidity of pledged inventory and the risk preference of banks,this paper sets impawn rate dynamically by dividing the impawn period into different risk windows so as to trade off the dilemma between risk holding period and impawn period.Besides,this paper proposes that different from pledging bonds and stocks,setting impawn rate in inventory financing is to forecast long-term risks caused by the insufficient liquidity of inventory.Moreover,in order to better depict the features of heteroscedasticity,leptokurtosis and fat-tails,this paper introduces the Generalized Error Distribution,and establishes the VaR-GARCH(1,1)-GED model based on normal assumption.This approach can help predict volatilities of different impawn periods,and establishes the formula of long-term VaR to forecast long-term risks with short-term data.Finally,parameter K is introduced to improve risk coverage and calculate the impawn rate according to the risk tolerance level of banks. Based on the dataset of spot steel(φHRB335),usually traded in over-the-counter markets,this paper shows that the time series of returns have significant characteristics of volatility clustering,leptokurtosis and fat-tails.Besides,it is important to note that the model may be able to predict the long-term risk in most cases;however,the failure rates of risk windows in 3 months and 4 months are far beyond the confidence level in back testing.This finding indicates that the model is not able to predict the long-term risk perfectly even after considering the characteristics of fat-tails and volatility clustering.To improve the model,the corrected parameter K is introduced into this paper.After adding the new parameter,the risk coverage level has been improved remarkably via K,which plays an important role as capital cushion.Also,the impawn rates obtained from model positively correlate to the lowest price in the future risk window.This model could reflect reasonably the risk expectation of banks about pledged steel. In summary,the VaR-GARCH(1,1)-GED model could control the risk better while reducing the efficiency loss compared with existing methods.This paper introduced a dynamic impawn rate mode and a new framework for banks.