Research

About | Presentations | Research | Blogs | 中文

Recent

Delayed and Distorted Price Discovery: Post-IPO Stocks in China

Master's thesis, draft available upon request.

Abstract: I find that two rules introduced to the Chinese stock market in late 2013 delay and distort price discovery among post-IPO stocks. The rules specify the maximum funds a firm could raise relative to its earnings and restrict the daily price change after listing. Undervalued new stocks hit the daily upper limit for two weeks with little volume. I document a 15% to 30% percent higher valuation among new stocks after the rule change, which may result from retail investors’ extrapolation of daily returns or institutional investors profiting with their liquidity advantage. The result implies more than 240 billion CNY higher valuation among all post-IPO stocks after 2014, though it reverses to the market level over the first year after listing. The result is consistent with a shift of IPO industry composition from high-PE to low-PE after the rule change.

Fig. 1 Overpricing rate of post-IPO stocks, before and after 2013-14 rule changes. The overpricing rate is the average of the ratios of initial trading P/E to the median market P/E for all IPO stocks in each month.

Past

with Kun Yuan, Bicheng Ying, and Ali H. Sayed, IEEE Transactions on Signal Processing 67 (2), 351-366. doi: 10.1109/TSP.2018.2872003.

In decentralized algorithms, computing nodes only communicate with their neighbors (no central servers involved), which eliminates server failure problems, relieves communication bottlenecks, and protects data privacy. Applications: supercomputers with millions of cores, networked self-driving cars. However, decentralized algorithms often (1) fail to converge, (2) require more computation time, and/or (3) cost excessive communication compared to single-machine algorithms. Our algorithm, Diffusion AVRG, solves (1) and outperforms state-of-the-art algorithms on both (2) and (3).

Fig. 1 Our algorithm, under the "best" setting shown in the figure, reduces the time cost of a standard machine learning task from 21.3 to 4.4 units of time.

About | Presentations | Research | Blogs | 中文