The Impact of Monetary Policy and Housing-Purchase Restrictions on Housing Prices in China

ABSTRACT

This paper examines the level and volatility effect of monetary policy on housing prices in China utilizing a novel set of housing price indices constructed by (Fang, H., QuanlinGu, W. X., & Zhou, L.-A. (2015). Demystifying the Chinese housing boom. NBER Macroeconomics Annual 2015, Volume 30. University of Chicago Press.). We find that in the long-run, average housing prices react positively to inflation, money supply and bank lending growth, and negatively to the reserve requirement ratio and benchmark lending rate. Housing prices in Tier 1 cities respond more sensitively to monetary shocks relative to Tier 2 and 3 cities, possibly due to surging demand and limited supply under housing-purchase restrictions (HPR). We further study the volatility effect of monetary shocks using the GARCH model and find that the benchmark lending rate, reserve requirement ratio and money supply growth have strong negative impact on the volatility of housing price growth. Our benchmark results remain robust after incorporating the HPR policy variable in the estimation, with a significant negative effect of HPR on housing price growth in Tier 1 and Tier 2 cities. Lastly, we conclude with recommendations on future monetary policy design and implementation, with a specific focus on the heterogeneous characteristics of China’s housing market.

KEYWORDS:

• Monetary policy
• housing price
• housing-purchase restrictions
• volatility

JEL CODES:

• E2
• E4
• E5
• G1
• R31
• R38

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Sherry Yu http://orcid.org/0000-0002-5067-4455

Notes

1 Please see Du and Zhang (2015), Wang and Huang (2013), Liu (2013), etc.

2 Loans made to homebuyer and real estate developers have been growing at an increasing speed in China since the 2008 financial crisis. China's top five banks had mortgages and loans to the sector of 12.4 trillion yuan at end of 2015 and representing 28 percent of total loans.

3 Multicollinearity may exist between M2 and bank lending. According to Farrar and Glauber (1967), this multicollinearity will not affect the accuracy of our estimation with sufficient sample size.

4 The estimates obtained from an ARDL model of cointegration with appropriate modification as suggested by Section 3 of Pesaran and Shin (1999) are unbiased and efficient since they avoid issues that may arise in the presence of endogeneity and serial correlation. Thus, the ARDL model itself can effectively deal with endogeneity and serial correlation in the model.

5 As mentioned earlier, real consumption is a proxy for real disposable income in our estimation.

Additional information

Notes on contributors

Sherry Yu

Sherry Yu is an Assistant Professor of Economics at New College of Florida. She obtained her PhD in Economics from Boston University in 2015.

Lini Zhang

Lini Zhang is a quant at Barclays PLC. She obtained her PhD in Economics from The Ohio State University in 2014. This work was done when she was an Assistant Professor of Finance at the School of Finance, Central University of Finance and Economics.