杠杆率如何影响资产价格？——来自中国债券市场自然实验的证据

杠杆周期理论表明,杠杆周期会影响金融体系的稳定性,资产的质押率(相应地,杠杆率)上升会提高资产价格;由于杠杆率一般是内生的,所以在实证上一直很难确立杠杆率与资产价格之间的因果关系。本文首次运用2017年1月到8月中国证券交易所债券市场和银行间债券市场的债券发行数据,利用中国银行间债券市场和交易所债券市场对同类债券的不同质押率规定的自然实验,对杠杆率与资产价格之间的因果关系进行了实证检验。结果表明,债券的可质押属性可以提高债券的价值,这验证了杠杆周期理论的预测,即杠杆率上升会使得资产价格上升。因此,杠杆率对资产价格有重要影响,是宏观审慎监管的一种工具,也应该成为货币政策的一种工具。

How Leverage Affects Asset Prices: Evidence from a Natural Experiment in China's Bond Markets

Asset leverage has been the focus of academic research since the recent global financial crisis, and deleveraging has been at the heart of policy discussion. Leveraging and deleveraging are associated with the rise and fall of asset prices.
In the modern financial system, collateral-based financial markets and monetary markets have become increasingly important. Investors can use their assets as collateral to borrow money. For instance, if a haircut is 20%, they can borrow an amount equal to 80% of the asset value (here Loan-to-Value, or LTV is 80%). Generally, for an asset, the following holds: leverage=1/(1-LTV)=1/haircut. The haircut reflects the lender's fear that the value of the collateral will fall in the future. The greater the fear, the higher the haircut, hence the leverage will be lower.
Leverage theory a la Geanakoplos shows that an increase in leverage will increase the asset price, because higher leverage better reflects optimism about the price; while due to short-sale constraints, the pessimistic view cannot be incorporated in the price. Leverage-induced boom-bust cycles lead to a fragile financial system and a volatile economy. Leverage cycles are a recurring historical phenomenon. Margin finance in the stock and housing markets is related to the Great Depression of 1929–1933 and the Great Recession following the 2008 financial crisis, respectively. China's stock market crash of 2015 is also related to margin finance.
Although theory and history both have shown the importance of the leverage cycle, due to the endogeneity of leverage and asset prices in the financial system, it is notoriously hard to establish the causal effect of leverage on asset prices. This study is the first to exploit a natural experiment in China's bond markets in which the same type of bonds traded in both the inter-bank market and the exchange has differential regulatory margin requirements. It exploits this experiment to identify the causal effect of leverage by using data on bonds issued from January 2017 to August 2017 in these two markets.
The study makes the following contributions. First, it is among the first to use a natural experiment to identify the causal effects of leverage on asset prices. Second, it sheds light on the market microstructure in the bond market; third, it sheds light on macro-prudential regulation and monetary policy.
The natural experiment is as follows. The China Securities Depository and Clearing Corporation Limited (CSDC) is in charge of the registration, depository, and clearing of bonds listed on the exchange, while the Shanghai Clearing House is in charge of those listed in the inter-bank market. On April 7^th, 2017, the CSDC adjusted the pledging qualification for corporate bonds and removed from the collateral basket all corporate bonds listed on the exchange with a rating below AAA. This change only affected bonds issued after April 7^th, 2017.
In this context, this paper tries to show that the change in spreads of AA bonds listed on the exchange is significantly larger than those listed on the inter-bank bond market.
The main idea is to exploit this new regulation as a natural experiment, thus the AA bonds listed on the exchange are taken as the treatment group and those in the inter-bank bond market as the control group. Using a difference-in-differences methodology, we test whether the collateral requirement (hence leverage) caused a change in price. The sample is the AA bonds issued on the exchange (including both the Shanghai and Shenzhen Stock Exchanges) between January and August 2017.
The paper finds that leverage can increase the price of a bond by as much as 70 basis points. This finding is consistent with theoretical predictions, and is robust to tests with different time windows for the new regulation and to placebo tests. This paper also tests for parallel trends to see whether companies in the sample strategically choose the venue of issuance, and finds that potential strategic shopping does not affect the results.
The paper thus establishes the causal effect of leverage on asset prices. It has important implications for financial markets and monetary policy. For a long time, economists and central bankers have regarded interest rates as the most important variable in the economy. However, in the modern collateral-based financial system, leverage is sometimes more important. In terms of monetary policy, central banks have never regarded leverage as a tool, and this neglect has caused the failure of interest rate-based monetary policy and the recurrence of leverage cycles. An ideal monetary policy should take both leverage and interest rates into account.