当下，我国信用债市场里违约越发常见。违约通过影响情绪与风险偏好、增加信息不对称、促发“羊群效应”等方式，干扰债市投研人员的预期和行为，进而使得信用风险可由预期渠道对具有相似性的企业造成传染，且经由该渠道的传染往往更为迅速。因此，本文对其展开研究，并在诸多相似性中选取地区相似性作为传染的主要方向。本文利用2015-2021年全国30省的月度数据，以区域“违约次数”为自变量，以各省二级产业债与城投债的利差为因变量，构造面板回归。主效应结果显示： （1）当省内债券违约时，省内产业债受到传染，但城投债却展现“避风港效应”；（2）当经济大区内其他省债券违约时，省内产业债同时受到传染，跨省传染弱于省内传染；但城投债受该影响不显著；（3）当接壤省债券违约时，省内产业债受该影响不显著；但城投债却受到传染，可能源于片区系统性风险发生的可能增强。本文后续单独针对产业债所受的传染展开异质性分析，结果如下：（1）资产收益率较高的省份抵御传染的能力更强；（2）财务负担较重的省份抵御传染的能力更弱；（3）国企占比较高的省份抵御传染的能力更弱，可能源于违约使政府隐性担保变得更不可信；（4）评级背书较强省份受到的传染不显著；（5）金融机构实力较强的省份抵御传染的能力更强；（6）政府财力较强的省份抵御传染的能力更强。为了解决内生性问题，本文利用“省委书记是否更替”构造工具变量，考虑新官上任后引发的地区政治生态和政企关系变化，及其可能对违约造成的影响。2SLS结果显著，本文的模型具有稳健性。地区信用通常“一荣俱荣，一损俱损”，新上任的省政府领导会格外重视避免违约，以保障任期内债券融资渠道畅通。本文随后利用解释变量滞后1期再次构造工具变量，并替换解释变量与样本时间区间后再回归，结果均显示模型具有稳健性。考虑到线性模型无法描述非线性的缺陷，本文最后引入Copula-GARCH模型，对各省利差两两间依次建模，计算上尾相关系数，进一步补充描述跨省传染发生的倾向性。结果显示：（1）东部各省之间的尾部相关性较强；（2）东部与中西部部分省份间也存在较强的尾部相关性；（3）中部、西部内各省之间的尾部相关性并未明显偏强。

At present, default is more and more common in domestic credit bond market. Default interferes with the expectations and behaviors of bond market researchers by affecting sentiment and risk appetite, increasing information asymmetry, and promoting ＂herd effect＂, so that credit risk can be transmitted to similar enterprises through the expected channel, and the transmission through this channel is often more rapid. Therefore, this paper studies it, and selects regional similarity as the main direction of infection.This paper uses the monthly data of 30 provinces in China from 2015 to 2021, takes the regional ＂times of defaults＂ as the independent variable, and takes the spreads of secondary industrial bonds and City Construction investment bonds of each province as the dependent variable to construct panel regression. The main results show that: （1） when the provincial bonds default, the provincial industrial bonds will be infected in the month, but the provincial City Construction investment bonds show a ＂safe haven effect＂; （2） When the bonds of other provinces in the economic region are in default, the industrial bonds in the province are infected at the same time in the same month, and the degree of cross provincial infection is weaker than that in the province; However, the impact of default in the economic region on City Construction investment bonds is not significant; （3） When the bonds of neighboring provinces defaulted, the industrial bonds in the province were not infected at the same time; However, City Construction investment bonds are not only immune to infection in the province, but also infected by the default of the bordering Province, which may be because the increase of default in the bordering province indicates that the possibility of systemic risk rises.Furthermore, this paper analyzes the heterogeneity of the infection of industrial bonds, and the results are as follows: （1） provinces with higher overall asset return have stronger ability to resist intra provincial and inter provincial infection; （2） Provinces with heavy financial burden as a whole have weaker ability to resist intra provincial and inter provincial infection; （3） Provinces with a high proportion of state-owned enterprises have a weaker ability to resist intra provincial and inter provincial infections, which may be because the regional default makes the implicit government guarantee more unreliable; （4） The provinces with strong rating endorsement are not significantly affected by intra provincial and inter provincial infection; （5） Provinces with strong rescue capacity of local financial institutions have stronger ability to resist intra provincial and inter provincial infections; （6） Provinces with strong local government assistance capacity have stronger ability to resist intra provincial and inter provincial infections.In order to solve the endogenous problem, this paper uses ＂whether the provincial Party committee secretary is replaced＂ to construct instrumental variables, and considers the development goals of new officials after taking office and the possible impact of changes in the relationship between government and enterprises. 2SLS results are significant, and the model in this paper is robust. The newly appointed leaders of the provincial government will pay special attention to avoiding default, so as to ensure smooth bond financing channels during their tenure. This paper then uses the explanatory variable lag 1 period to construct the instrumental variable again, and replaces the explanatory variable and the sample time interval. The results show that the model is robust.Considering the defect that the linear model can’t describe the nonlinearity, this paper finally introduces the Copula-GARCH model to model the interest spread between two provinces in turn, calculate the upper tail correlation coefficient, and further supplement the description of cross provincial infection. The results show that: （1） the tail correlation between the eastern provinces is strong; （2） There is also a strong tail correlation between the eastern and central or western provinces; （3） The tail correlation between the central and western provinces is not significantly stronger.