Working Papers

Liquidity Constraint, Transition Dynamics, and the Chinese Housing Return Premium [PDF new version coming soon] [Slides new]
SED 2017, NBER Chinese Economy 2017, Princeton, CityUHK, PKU GSM, SAIF, SHUFE, SJTU Antai, Tsinghua SEM, XJTU

Endogenous evolution in funding liquidity can drive large medium-to-long run asset price movements. In China, housing prices grew 170% during 2003–2012 in real terms. Returns on housing commanded a sustained 12% premium annually over the risk-free rate. Across Chinese cities, the relationship of the housing boom with income or population growth is weak using official or unofficial measures. However, increases in the value of housing are instead associated one-to-one with increases in household wealth, whether measured with or without housing. The same relationship is obtained instrumenting wealth growth using differences in savings rates driven by the "experience mechanism" via different exposures to the 1959-61 Great Famine across cities. This relationship motivates an explanation based on the upward transition dynamics in household wealth from a low initial condition, interacted with liquidity constraints.

Evaluated in a consumption-housing two-asset dynamic portfolio choice model with housing being priced in equilibrium, calibrated with realistic liquidity constraints and low initial wealth, this explanation explains the high housing return premium and generates the bulk of the observed aggregate increase in housing prices. Value of housing in the model is the lower of (1) the maximal sustainable level of housing value sensitive to the required rate of return (RRoR), and (2) endogenous evolution in funding liquidity driven by wealth dynamics, from a low initial condition. Housing in the model is a hedge to fundamental-driven recessions: A permanent slowdown in economic growth leads to a reduced intertemporal smoothing motive and a higher target wealth, hence housing price experience only a small temporary decrease which quickly overturns.

Global Supply Chain and Cross-Border Financing (with Jie Peng and Jing Wu) [SSRN]
INFORMS 2018, CityUHK, George Washington (scheduled), George Mason (scheduled), HKU, Renmin University, SHUFE, Tsinghua PBCSF

Does the formation of global supply chain relationships help firms access cross-border financing in the global capital market? Comparing two firms that are otherwise similar in covariates and in pre-formation outcome through a matching difference-in-differences strategy, we observe the firm that has formed global supply chain relationships to access more cross-border financing, after the global supply chain formation. This result is supported empirically by all major financing vehicles, including stock cross-listing, bond issuance, bank loans, and M&A deals, and is robust to controlling for strategic disclosure concerns. The estimated effect of global supply chain relationship formation on cross-border financing is stronger for firms in countries with stronger accounting and ethical standards and for small and growth firms, but weaker for firms in countries with tighter capital controls and more volatile exchange rates. This suggests global supply chain formation helps cross-border financing more through information production, not cashflow relocation or re-denomination.

Bubbly Bitcoin (with Feng Dong and Zhiwei Xu) [SSRN]
PHBS Workshop in Macro and Finance 2018, PKU GSM

What are the potential causes and macroeconomic consequences of Bitcoin-like cryptocurrencies? We construct a model for Bitcoin-like cryptocurrencies as a stochastic rational bubble that is costly to produce, in an infinite-horizon production economy with incomplete markets. Entrepreneurial firms choose to hold Bitcoins as liquid assets to buffer idiosyncratic investment distortions. The intrinsically worthless Bitcoins can emerge as rational bubbles only when market distortion is severe and public sentiment is optimistic enough. Bubbly Bitcoins provide market liquidity to facilitate investment in the real sector, but they also crowd out production through costly mining activities. Deterioration in market sentiment depresses Bitcoin prices but ambiguously affects aggregate output and investment. This model is consistent with the following facts: i) the surging Bitcoin market presents enormous volatility, and ii) its price dynamics are significantly sensitive to both investor sentiment and policy stances, and iii) the Bitcoin market exhibits diverse cyclical features for US and China.

Capital Leakage, House Prices, and Consumer Spending: Evidence from Asset Purchase Restriction Spillovers (with Yinglu Deng, Li Liao and Jiaheng Yu) [PDF coming soon]
CICF 2019 (scheduled), Asian Meeting of the Econometric Society 2019 (scheduled), SHUFE

This paper studies the consequences on house prices and consumer spending of capital leakage across housing markets. We exploit plausibly exogenous spillovers from the imposition of restrictions on housing asset purchases in certain large Chinese cities on nearby unregulated cities as a unique laboratory. We find that capital leakage from large regulated cities causes house prices in small, neighboring, non-regulated cities to rise significantly. These exogenous house price booms in non-regulated cities lead to a substantial increase in household spending on automobiles, explaining approximately 40% of the sample period annually average aggregate increase in car sales. The spending responses are highly heterogeneous across household types, suggesting redistributive effects of housing market capital leakage. We find the largest spending response in locals born in the cities they reside in, no significant average response among migrants, and a U-shaped spending profile across age groups, consistent with predictions from Favilukis and Van Nieuwerburgh (2017). The quasi-experimental estimate of the MPC from housing wealth is larger than the OLS estimate because saving propensities, which drives both spending and house prices, introduce a negative omitted variable bias in the OLS estimate, highlighting that the source of house price variations matters in understanding the effects of housing wealth.

Heterogeneity in Consumption Responses to Long-lasting Income Shocks [PDF]

Standard life-cycle incomplete-markets models predict large cross-sectional differences for age and wealth in agents’ consumption responses to long-lasting income shocks; whether such predictions hold in the data is an open question. Using household microdata for the US, I document economically important age and wealth cross-sectional heterogeneity in the the adjustment of nondurable consumption in the face of long-lasting income shocks. A calibrated standard life-cycle incomplete-markets model predicts heterogeneity in consumption responses that are quantitatively similar to empirical estimates.

Work in Progress

The Transmission of Income Inequality to Consumption Inequality: The Role of Firms (with Miao Jin, Yu-Jane Liu and Juanjuan Meng)