Research

Working Papers:

How Urbanization Affects Regional Wealth Inequality: the Housing Channel (with Jiaofen Hu)

China went through rapid urbanization in the past decade. We document that cheap apartments appreciate much faster than expensive apartments in this period. We identify two channels for the pattern: (1) Demand for cheap apartments increases more relative to expensive apartments. (2) The location values in poor neighborhoods increase more due to city construction. Combined with the evidence that cheap apartments are mostly owned by poor people, we find that the housing market causes regional wealth inequality to decrease among homeowners. Our findings suggest new win-win scenarios where economic growth and inequality reduction can be achieved simultaneously.

How do ETFs Affect the Liquidity of the Underlying Corporate Bonds?

There is considerable concern over the impact of ETFs on illiquid markets. This paper shows that corporate bond ETFs improve the liquidity of the underlying bonds. ETF ownership increases turnover and reduces transaction costs. I identify an important channel for this positive impact: ETF arbitrage. The profits that dealers can earn from ETF arbitrage are considerable. Consequently, when ETF arbitrage opportunities arise, dealers are incentivized to both compete and to give price concessions. In the cross-section, the impact of ETFs is stronger on the less liquid bonds because these bonds are rarely traded by other types of investors.

Self-Feeding Housing Price Boom: Evidence from Cross-Generation Transfer (with Tao Shu and Sheridan Titman)

There is considerable concern over the impact of ETFs on illiquid markets. This paper shows that corporate bond ETFs improve the liquidity of the underlying bonds. ETF ownership increases turnover and reduces transaction costs. I identify an important channel for this positive impact: ETF arbitrage. The profits that dealers can earn from ETF arbitrage are considerable. Consequently, when ETF arbitrage opportunities arise, dealers are incentivized to both compete and to give price concessions. In the cross-section, the impact of ETFs is stronger on the less liquid bonds because these bonds are rarely traded by other types of investors.

Creating More Reliable Measures in the Bond Market with Matrix Pricing (with Jinfan Zhang and Delong Li)

The major difficulty in corporate bond research is dealing with missing data. This paper proposes a method to fill in the missing data using matrix price techniques. I then document several important findings. First, the monthly return data on WRDS Bond database which uses the last trade in TRACE could be unreliable. The correlation between returns reported on WRDS and the returns I calculate using adjusted month-end prices is only 70%. Second, many commonly used liquidity measures are unreliable. This is in stark contrast with Schestag et al. (2016) who claim that most measures can capture the cross-sectional variations of trading costs. I show that regression-based methods systematically underestimate the magnitude of trading costs, and they do not highly correlate with the effective spreads. Of all the liquidity measures that use only TRACE data, Roll's measure is the best at both estimating the magnitude of trading costs and identifying the cross-sectional variation in bond liquidity.

中国新基金发行之谜和投资者保护 (with Jinfan Zhang and Kaixuan Zheng)
截止2022年1月，中国公募基金管理资产25万亿，基金数量已超过9000支。虽然管理资产规模只有美国公募基金行业的1/9，但中国基金数量已经超过美国。中国公募基金行业表现出强烈的发行新基金偏好。我们发现造成这一现象的主要原因是银行在基金发行中的巨大渠道优势以及对于基金托管业务的重视。基金公司为了获得更大的银行客户群，在不同银行发行新基金并托管是一个重要手段。过度发行基金造成权益类基金经理平均管理2.6支基金。比较同一个基金经理管理的新老基金，我们发现新基金年化收益率平均优于老基金1.5\%-2\%。当一个基金经理发行了新基金后，其管理的老基金表现出现下降。由于新基金的资金流入更加依赖其近期表现，因此基金经理有更强动机提升新基金业绩。我们的发现对于规范中国公募基金行业发展，加强投资者保护提供了依据。

“北上广深杭”顶级学区溢价是如何分布的？ (with Jiaofen Hu)
本文针对中国房地产市场的特点，改进了传统文献中计算学区房溢价的方法，实现了对大部分学校的学区溢价的逐一计算。本文首次揭示了一个城市内学区房溢价的分布情况，并以此量化优质教育资源被垄断的程度。我们收集了北京、上海、广州、深圳和杭州2011-2019年的房产交易数据，发现北京、上海、深圳最顶尖的小学在2017年的学区溢价高达500万，即使学区内最便宜的房子也让全市约95\%的家庭负担不起。广州和杭州在2017年顶尖学区溢价在200万左右。该溢价有非常强的头部效应，排第十的学校往往溢价只有第一的五分之一。如果对学区房溢价计算基尼系数，在只考虑前10的学校时，北上深杭在0.5到0.6之间，广州只有0.26；如果考虑所有学校，那这些城市均超0.98。可见这些城市教育不平等问题非常严重。为了破除优质教育资源的垄断现象，国家需要一定的行政手段来打压天价学区房。我们通过准自然实验发现大学区政策确实是抑制学区溢价增长行之有效的方法，但不能完全消除学区溢价。要进一步促进教育机会平等，我们需要在教育制度上进行改革，如实行教师轮岗。

Work-in-progress:

Why do Retail Investors Underperform? (with Xiaomeng Lu)

According to the retail trading data from the Shenzhen Stock Exchange, the average alpha of small retail investors can be as low as -10%. We decompose this negative alpha into different components caused by various behavioral biases. The research aims to find the most important behavioral bias that contributes to investors' loss, and it has a strong implication on investor education.