VaR Models for Setting Dynamic Impawn Rate of Steel in Inventory Financing of Supply Chain Finance
HE Juan,JIANG Xiang-Lin,ZHU Dao-Li,WANG Jian,CHEN Lei
DOI: https://doi.org/10.3969/j.issn.1004-6062.2012.03.019
2012-01-01
Abstract: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.