我院王雅琦副教授与中国人民大学劳动人事学院周广肃副教授合作撰写的论文《住房价格、房屋购买与中国家庭杠杆率》发表在《金融研究》2019年第6期。
Mian et al.（2013）和陈斌开和杨汝岱（2017）等人刻画了房价变动对居民消费储蓄等微观行为的影响。另一方面，从家庭金融资产配置角度看，资产证券化、次级住房抵押贷款等金融创新刺激了家庭负债的增长和家庭信贷的可获得性（Casolaro et al.，2006），李凤等（2016）基于中国家庭金融调查数据发现在2015年家庭住房资产占据家庭总资产比例高达70.1%。在此基础上，本文重点研究了房价变动对于家庭杠杆率的影响，并进一步探索房价变动对于我国家庭杠杆率影响的主要特征和渠道，同时探讨对不同特征家庭的异质性影响，对现有文献进行补充。
本文利用中国家庭追踪调查（CFPS）2014和2016年数据衡量了家庭的杠杆率水平，定义为家庭为购买或建造、装修住房而进行的借贷总额除以家庭当年的总收入。研究结果发现，房价对家庭杠杆率的提高有显著的正向作用，房价增长1倍，家庭杠杆率会提高39.2%，家庭贷款数额将会增长288.1%，而此部分贷款的增加主要来源于银行贷款而非私人借贷，由房价导致的家庭借贷数额和杠杆的的上升约占购房家庭总体借贷数额和杠杆率上升的90%。
进一步地，本文探讨了房价推动家庭借贷比率上升的的影响机制。首先，本文发现住房价格上涨刺激了家庭首套刚需住房需求的实现，这表明房价上升会提高家庭未来的购房成本，促使那些没有住房的家庭通过增加借贷来购买首套刚需住房，即“直接效应”。其次，我们发现房价上升刺激了家庭多套住房购买需求的实现，说明房价上升会增强家庭对于未来房价上涨的预期，投资住房的预期回报将会进一步提高，最终使得家庭通过借贷购买多套住房，即“投机效应”。住房价格的快速上涨还提高了家庭的借贷意愿和风险偏好，而“抵押效应”则并不明显。此外，分样本的探讨发现，这一效应对于非农户口家户、有配偶和子女家户的影响更为明显。
Mian et al. (2013) and Chen andYang (2017) portrayed the impact of housing price changes on microeconomicbehaviors such as household consumption and savings. Asset securitization andsubprime mortgages have stimulated household debt growth and household creditavailability (Casolaro et al., 2006). Based on Chinese Household Finance Surveydata, Li et al (2016) found family housing assets account for as much as 70.1%of total household assets in 2015. On these studies, this paper focuses on theimpact of housing price changes on household leverage, further explores themain channels of housing price changes on China's household leverage, andexplores the heterogeneity impact on households with different characteristics.
Based on household level housingand debt information drawn from the Chinese Family Panel Studies(CFPS)for 2014 and 2016, we show that the appreciation in house prices indeedcontribute to the increase in household leverage. Quantitatively, we find thata 100% increase in the price of a house leads to 288. 1% and 39. 2% increasesin household debt and leverage, respectively. By further splitting our sampleinto several subsamples, we also find that the house price movements haveheterogeneous effects on household leverage. Specifically, we find that urbanhukou households with spouses and children have a greater increase in householdleverage than rural hukou households without spouses and children. Moreover, wefind that some of the data patterns are at odds with or cannot be well explainedby the “collateral channel”. Therefore, we propose two additional channels, the“financing channel” and the “speculation channel”, and confirm that these channelshave a strong influence on our sample. Furthermore, we find that most householdsfinance their housing purchases by bank loans instead of private lending.
