Asset Pricing Models in China
The CAPM model was a breakthrough for asset pricing, but the times where the market factor was most widely used are long gone. Nowadays, if we exaggerate a bit, we have as many factors as we want. Therefore, it might not be straightforward which factor model should be used.
Hanauer et al. (2021) provide several insights into factor models. The authors postulate that the factor models should be examined in the international samples since this can be understood as a test for asset pricing models. The domestic Chinese A-shares stock market seems to be an excellent “playground” for the factors models, given the size of the Chinese stock market, but mainly because of its uniqueness. The paper compares the models (and factors) based on various methods (performance, data-driven asset pricing framework, test assets, turnovers and even transaction costs). Apart from valuable insights into the several less-known factors, the key takeaway message could be that the “US classic” Fama-French factor models perform poorly in China. The modified Fama-French six-factor model or q-factor is better, but overall, it seems that factor models designed for China, such as the model of Liu, Stambaugh and Yuan (2019), are the best.
Authors: Matthias X. Hanauer, Maarten Jansen, Laurens Swinkels and Weili Zhou
Title: Factor models for Chinese A-shares
We compare the pricing ability of popular asset pricing models for the cross-section of U.S. equities on a large, liquid, but mostly segmented equity market of Chinese A-shares. The q-factor model performs well among factor models developed for the U.S. equity market, but is outperformed by a modified Fama-French six-factor model and by a four-factor asset pricing model adapted to the Chinese A-shares market. A data-driven method to detect the preferred asset pricing model results in the same four factors, plus three additional ones. However, these three additional factors do not reduce the pricing errors to a set of test assets. When taking transaction costs into account, the ranking of asset pricing models changes. The preferred model from both the direct and data-driven model comparison methods now consists of a three-factor model comprising the market, size, and an earnings-based value factor.
As always we present interesting tables:
Notable quotations from the academic research paper:
The apparent existence of a large number of anomalies has spurred research into new asset pricing models that better account for these seemingly anomalous returns relative to the Capital Asset Pricing Model or the popular Fama and French (1993) three-factor model. Competing factor models are, among others, the five- and six-factor model by Fama and French (2015, 2018), the q-factor model by Hou, Xue, and Zhang (2015), and the mispricing model by Stambaugh and Yuan (2017).
The Chinese economy is large. In 2000, its gross domestic product was already 36% of that of the U.S., but this has steadily increased and reached 116% in 2020. Furthermore, the Chinese equity market has grown to the second-largest in terms of market capitalization and trading volume, after only the United States. Despite its economic importance, Chinese A-shares have been mostly segmented from other equity markets in the world (see Hu, Pan, and Wang, 2021), and firms tend to have different corporate governance features, leading to different agency problems (see Jiang and Kim, 2020). This makes the A-share market an excellent out-of-sample environment to test for the existence of factors that have originally been documented in U.S. equity markets.
The contribution of this paper is threefold. First, we compare the pricing ability of popular asset pricing models for a large, liquid, but mostly segmented equity market of Chinese A-shares. This analysis sheds light on which factors are more likely to be universal drivers of risk and return, and which factors are more local or temporary in nature.
Second, we employ a data-driven method to come up with the preferred asset pricing model for the Chinese A-share market and compare these with the pricing ability of models developed and tested using U.S. stock markets. This contributes to our understanding in what sense the cross-section of returns of the Chinese A-share market is different from that of the rest of the world.
Third, we evaluate the relative performance of factor models for a more real-life situation in which we account for transaction costs. In a recent paper, Detzel, Novy-Marx, and Velikov (2021) show that by taking transaction costs into account, the ranking of competing asset pricing models may substantially change. Model selection studies that ignore transaction costs favor models that include more frequently updated factors with less persistent characteristics. However, Novy-Marx and Velikov (2016) show that such high turnover factors often do not generate significant net returns despite high gross returns. Our paper adds to this literature by studying the implications of including transaction costs for the order of preferred asset pricing models. Each of these three contributions also adds to the recent discussion on the existence of the value premium.
Our main findings are summarized as follows: First, we find that the conventional three- and fivefactor models from Fama and French (1993, 2015) are poor asset pricing models in the Chinese A-share market. Adding a momentum factor and replacing the original book-to-market and operating profitability factors with monthly updated book-to-market and return-on-equity factors improves the performance of the asset pricing model substantially.
Second, our findings suggest´ that a parsimonious four-factor model developed specifically for the China A-share market outperforms the existing models developed to price the cross-section of U.S. equity returns. A Bayesian asset pricing framework, as in Barillas and Shanken (2018) and Chib, Zeng, and Zhao (2020), to detect the preferred asset pricing model results in the same four factors, plus three additional ones. This data-driven asset pricing model is also superior in terms of the classical maximum squared Sharpe ratio metric. However, the three additional factors do not reduce the pricing errors to a set of non-model factors that are deemed to be relevant for China A-shares in the literature, a seemingly conflicting result that we can resolve by extending the set of test assets in the spirit of Barillas and Shanken (2017).
Third, we find that accounting for transaction costs alters the order of preferred asset pricing models for the China A-share market, similar to the findings of Detzel, Novy-Marx, and Velikov (2021) for the US stock market. The preferred model, including transaction costs, is a three-factor model with the market, size, and monthly updated earnings-based value factor. Finally, we find that value factors are part of the data-driven asset pricing model, no matter whether we include or exclude transaction costs. Our empirical evidence thus indicates that value is not a redundant asset pricing factor and that its dismissal by some based on its recent poor performance in the U.S. equity market may be premature.”
Are you looking for more strategies to read about? Sign up for our newsletter or visit our Blog or Screener.
Do you want to learn more about Quantpedia Premium service? Check how Quantpedia works, our mission and Premium pricing offer.
Do you want to learn more about Quantpedia Pro service? Check its description, watch videos, review reporting capabilities and visit our pricing offer.
Or follow us on:
Facebook Group, Facebook Page, Twitter, Linkedin, Medium or Youtube
Share onLinkedInTwitterFacebookRefer to a friend