From risk to reward: Learning ESG return dynamics in Chinese equities

Maurice Kyla Octaviano; Jin-Taek Seong; Maurice Kyla Octaviano; Jin-Taek Seong

doi:10.3934/math.2026675

AIMS Mathematics

2026, Volume 11, Issue 6: 16448-16478. doi: 10.3934/math.2026675

Previous Article Next Article

Research article

From risk to reward: Learning ESG return dynamics in Chinese equities

Maurice Kyla Octaviano ,
Jin-Taek Seong ^,

Graduate School of Data Science, Chonnam National University, Gwangju 61186, Republic of Korea

Received: 23 April 2026 Revised: 21 May 2026 Accepted: 29 May 2026 Published: 09 June 2026
MSC : 91G10, 68T05, 62P05

This study re-evaluated the role of environmental, social, and governance (ESG) characteristics in China's A-share market over the period 2012Q1 to 2021Q3. Contrary to the prevailing view of ESG as a defensive, risk-mitigating attribute, we showed that passive high-ESG strategies fail to generate excess returns. To address this inefficiency, we introduced a meta-learning framework that integrates three heterogeneous base learners—LightGBM for nonlinear cross-sectional effects, ridge regression as a linear benchmark, and a temporal convolutional network (TCN) for sequential earnings dynamics—whose out-of-fold predictions were stacked via a second-stage ridge meta-learner trained on a held-out validation set. This architecture predicts next-quarter earnings per share (EPS-TTM) and ranks firms by fundamental growth potential within a high-ESG universe screened to the top tercile of ESG scores. The framework was trained on 2012Q1–2017Q4, validated on 2018Q1–2019Q1, and evaluated out-of-sample on 2019Q2–2021Q3 using a strict chronological split that eliminates look-ahead bias. We documented a clear performance hierarchy. While the broad high-ESG universe exhibits negative alpha relative to the CSI 300 benchmark, a refined top 50% portfolio largely neutralizes this underperformance, and economically meaningful excess returns are concentrated in the top 10% portfolio, with annualized alpha exceeding 40% over the out-of-sample evaluation period (Sharpe ratio: 1.11; market beta: 1.16). Industry-neutralized robustness tests confirmed that this alpha is attributable to stock-level fundamental prediction rather than passive sector concentration. These results support a risk-signaling interpretation of ESG in emerging markets: ESG-related return premia in China reflect compensation for growth-related systematic risk under conditions of disclosure heterogeneity and information asymmetry, rather than downside protection.
- ESG investing,
- machine learning,
- meta-learning,
- China A-shares,
- return premia
Citation: Maurice Kyla Octaviano, Jin-Taek Seong. From risk to reward: Learning ESG return dynamics in Chinese equities[J]. AIMS Mathematics, 2026, 11(6): 16448-16478. doi: 10.3934/math.2026675

Related Papers:

Abstract

This study re-evaluated the role of environmental, social, and governance (ESG) characteristics in China's A-share market over the period 2012Q1 to 2021Q3. Contrary to the prevailing view of ESG as a defensive, risk-mitigating attribute, we showed that passive high-ESG strategies fail to generate excess returns. To address this inefficiency, we introduced a meta-learning framework that integrates three heterogeneous base learners—LightGBM for nonlinear cross-sectional effects, ridge regression as a linear benchmark, and a temporal convolutional network (TCN) for sequential earnings dynamics—whose out-of-fold predictions were stacked via a second-stage ridge meta-learner trained on a held-out validation set. This architecture predicts next-quarter earnings per share (EPS-TTM) and ranks firms by fundamental growth potential within a high-ESG universe screened to the top tercile of ESG scores. The framework was trained on 2012Q1–2017Q4, validated on 2018Q1–2019Q1, and evaluated out-of-sample on 2019Q2–2021Q3 using a strict chronological split that eliminates look-ahead bias. We documented a clear performance hierarchy. While the broad high-ESG universe exhibits negative alpha relative to the CSI 300 benchmark, a refined top 50% portfolio largely neutralizes this underperformance, and economically meaningful excess returns are concentrated in the top 10% portfolio, with annualized alpha exceeding 40% over the out-of-sample evaluation period (Sharpe ratio: 1.11; market beta: 1.16). Industry-neutralized robustness tests confirmed that this alpha is attributable to stock-level fundamental prediction rather than passive sector concentration. These results support a risk-signaling interpretation of ESG in emerging markets: ESG-related return premia in China reflect compensation for growth-related systematic risk under conditions of disclosure heterogeneity and information asymmetry, rather than downside protection.

References

[1]	D. C. Broadstock, K. Chan, L. T. W. Cheng, X. Wang, The role of ESG performance during times of financial crisis: Evidence from COVID-19 in china, Finance Res. Lett., 38 (2021), 101716. https://doi.org/10.1016/j.frl.2020.101716 doi: 10.1016/j.frl.2020.101716
[2]	S. Jin, R. Xiong, H. Peng, S. Tang, Esg performance and private enterprise resilience: Evidence from chinese financial markets, Int. Rev. Financ. Anal., 98 (2025), 103884. https://doi.org/10.1016/j.irfa.2024.103884 doi: 10.1016/j.irfa.2024.103884
[3]	A. S. Garcia, W. Mendes-Da-Silva, R. J. Orsato, Sensitive industries produce better esg performance: Evidence from emerging markets, J. Cleaner Product., 150 (2017), 135–147. https://doi.org/10.1016/j.jclepro.2017.02.180 doi: 10.1016/j.jclepro.2017.02.180
[4]	S. Chen, X. Han, Z. Zhang, X. Zhao, Esg investment in china: Doing well by doing good, Pacific-Basin Finance J., 77 (2023), 101907. https://doi.org/10.1016/j.pacfin.2022.101907 doi: 10.1016/j.pacfin.2022.101907
[5]	X. Gong, F. Xie, Z. Zhou, C. Zhang, The enhanced benefits of esg in portfolios: A multi-factor model perspective based on lightgbm, Pacific-Basin Finance J., 85 (2024), 102365. https://doi.org/10.1016/j.pacfin.2024.102365 doi: 10.1016/j.pacfin.2024.102365
[6]	$\check{\rm{L}}$. Pástor, R. F. Stambaugh, L. A. Taylor, Sustainable investing in equilibrium, J. Financ. Econ., 142 (2021), 550–571. https://doi.org/10.1016/j.jfineco.2020.12.011 doi: 10.1016/j.jfineco.2020.12.011
[7]	C. Shrestha, P. Andrikopoulos, J. S. Park, Esg rating and financial performance in the emerging markets: The moderating effects of cross-listing and industry type, Res. Int. Business Finance, 77 (2025), 102916. https://doi.org/10.1016/j.ribaf.2025.102916 doi: 10.1016/j.ribaf.2025.102916
[8]	Z. Shang, B. Yu, K. S. Lam, Does esg explain stock returns? evidence from chinese stock markets, Finance Res. Lett., 79 (2025), 107214. https://doi.org/10.1016/j.frl.2025.107214 doi: 10.1016/j.frl.2025.107214
[9]	L. Ruan, J. Li, S. Huang, News or noise? esg disclosure and stock price synchronicity, Int. Rev. Financ. Anal., 95 (2024), 103483. https://doi.org/10.1016/j.irfa.2024.103483 doi: 10.1016/j.irfa.2024.103483
[10]	R. Albuquerque, Y. Koskinen, C. Zhang, Corporate social responsibility and firm risk: Theory and empirical evidence, Manag. Sci., 65 (2019), 4451–4469. https://doi.org/10.1287/mnsc.2018.3043 doi: 10.1287/mnsc.2018.3043
[11]	S. Bai, J. Z. Kolter, V. Koltun, An empirical evaluation of generic convolutional and recurrent networks for sequence modeling, arXiv preprint arXiv: 1803.01271, 2018. https://doi.org/10.48550/arXiv.1803.01271
[12]	T. Afzal, M. A. Afridi, M. N. Jan, Integrating LSTM with Fama-French six factor model for predicting portfolio returns: Evidence from Shenzhen stock market China, Data Sci. Finance Econ., 5 (2025), 177–204. https://doi.org/10.3934/DSFE.2025009 doi: 10.3934/DSFE.2025009
[13]	A. D. Kayit, M. T. Ismail, Leveraging hybrid ensemble models in stock market prediction: A data-driven approach, Data Sci. Finance Econ., 5 (2025), 355–386. https://doi.org/10.3934/DSFE.2025015 doi: 10.3934/DSFE.2025015
[14]	Z. Li, Y. Xu, Z. Du, Valuing financial data: The case of analyst forecasts, Finance Res. Lett., 75 (2025), 106847. https://doi.org/10.1016/j.frl.2025.106847 doi: 10.1016/j.frl.2025.106847
[15]	L. K. David, ESG scoring and forecasting in china, data set, 2023.
[16]	L. K. David, J. Wang, V. Angel, N. Amjad, Esg scoring and forecasting in china: advancing sustainable business with multidimensional modeling, J. Asia Pacific Econ., 30 (2025), 128–157. https://doi.org/10.1080/13547860.2023.2273017 doi: 10.1080/13547860.2023.2273017
[17]	E. F. Fama, K. R. French, Common risk factors in the returns on stocks and bonds, J. Financ. Econ., 33 (1993), 3–56. https://doi.org/10.1016/0304-405X(93)90023-5 doi: 10.1016/0304-405X(93)90023-5
[18]	J. Liu, R. F. Stambaugh, Y. Yuan, Size and value in china, J. Financ. Econ., 134 (2019), 48–69. https://doi.org/10.1016/j.jfineco.2019.03.008 doi: 10.1016/j.jfineco.2019.03.008
[19]	China 1-year government bond yield, Investing.com, 2026. Available from: https://www.investing.com/rates-bonds/china-1-year-bond-yield.
[20]	R. G. Sloan, Do stock prices fully reflect information in accruals and cash flows about future earnings? Account. Rev., 71 (1996), 289–315.
[21]	S. H. Penman, X. J. Zhang, Accounting conservatism, the quality of earnings, and stock returns, Account. Rev., 77 (2002), 237–264.
[22]	E. F. Fama, K. R. French, The cross-section of expected stock returns, J. Finance, 47 (1992), 427–465. https://doi.org/10.1111/j.1540-6261.1992.tb04398.x doi: 10.1111/j.1540-6261.1992.tb04398.x
[23]	S. Gu, B. Kelly, D. Xiu, Empirical asset pricing via machine learning, Rev. Financ. Stud., 33 (2020), 2223–2273. https://doi.org/10.1093/rfs/hhaa009 doi: 10.1093/rfs/hhaa009
[24]	S. P. Kothari, A. J. Leone, C. E. Wasley, Performance matched discretionary accrual measures, J. Account. Econ., 39 (2005), 163–197. https://doi.org/10.1016/j.jacceco.2004.11.002 doi: 10.1016/j.jacceco.2004.11.002
[25]	T. A. Gormley, D. A. Matsa, Common errors: How to (and not to) control for unobserved heterogeneity, Rev. Financ. Stud., 27 (2014), 617–661. https://doi.org/10.1093/rfs/hht047 doi: 10.1093/rfs/hht047
[26]	J. M. Wooldridge, Econometric Analysis of Cross Section and Panel Data, 2 Eds, Cambridge: MIT Press, 2010.
[27]	D. N. Gujarati, D. C. Porter, Basic Econometrics, 5 Eds, New York: McGraw-Hill, 2009.
[28]	B. T. Kelly, S. Pruitt, Y. Su, Characteristics are covariances: A unified model of risk and return, J. Financ. Econ., 134 (2019), 501–524. https://doi.org/10.1016/j.jfineco.2019.05.001 doi: 10.1016/j.jfineco.2019.05.001
[29]	N. Jegadeesh, S. Titman, Returns to buying winners and selling losers: Implications for stock market efficiency, J. Finance, 48 (1993), 65–91.
[30]	K. Hou, C. Xue, L. Zhang, Digesting anomalies: An investment approach, Rev. Financ. Stud., 28 (2015), 650–705. https://doi.org/10.1093/rfs/hhu068 doi: 10.1093/rfs/hhu068
[31]	N. Kim-Duc, P. K. Nam, Earnings growth rates in business valuation models: The impossible quaternity, Global Finance J., 60 (2024), 100930. https://doi.org/10.1016/j.gfj.2024.100930 doi: 10.1016/j.gfj.2024.100930
[32]	G. Ke, Q. Meng, T. Finley, T. Wang, W. Chen, W. Ma, et al., LightGBM: A highly efficient gradient boosting decision tree, Adv. Neural Inform. Proc. Syst., 30 (2017), 3146–3154.
[33]	A. E. Hoerl, R. W. Kennard, Ridge regression: Biased estimation for nonorthogonal problems, Technometrics, 12 (1970), 55–67. https://doi.org/10.1080/00401706.1970.10488634 doi: 10.1080/00401706.1970.10488634
[34]	T. Hastie, R. Tibshirani, J. Friedman, The Elements of Statistical Learning, 2 Eds, Berlin: Springer, 2009.
[35]	D. H. Wolpert, Stacked generalization, Neural Netw., 5 (1992), 241–259. https://doi.org/10.1016/S0893-6080(05)80023-1 doi: 10.1016/S0893-6080(05)80023-1
[36]	J. M. Bates, C. W. J. Granger, The combination of forecasts, J. Oper. Res. Soc., 20 (1969), 451–468. https://doi.org/10.1057/jors.1969.103 doi: 10.1057/jors.1969.103
[37]	F. X. Diebold, R. S. Mariano, Comparing predictive accuracy, J. Busin. Econ. Stat., 13 (1995), 253–263. https://doi.org/10.1198/073500102753410444 doi: 10.1198/073500102753410444
[38]	A. Kempf, P. Osthoff, The effect of socially responsible investing on portfolio performance, Eur. Financ. Manag., 13 (2007), 908–922. https://doi.org/10.1111/j.1468-036X.2007.00402.x doi: 10.1111/j.1468-036X.2007.00402.x
[39]	J. Cohen, Statistical Power Analysis for the Behavioral Sciences, 2 Eds, Mahwah: Lawrence Erlbaum Associates, 1988.
[40]	E. A. H. Elamir, Modeling and predicting earnings per share via regression tree approaches, Invest. Manage. Financ. Innov., 17 (2020), 51–68. https://orcid.org/0000-0002-9430-072X
[41]	J. Y. Campbell, A. W. Lo, A. C. MacKinlay, The Econometrics of Financial Markets, Princeton: Princeton University Press, 1997.
[42]	W. F. Sharpe, Mutual fund performance, J. Bus., 39 (1966), 119–138.
[43]	A. Chekhlov, S. Uryasev, M. Zabarankin, Drawdown measure in portfolio optimization, Int. J. Theor. Appl. Finance, 8 (2005), 13–58. https://doi.org/10.1142/S0219024905002767 doi: 10.1142/S0219024905002767
[44]	K. V. Lins, H. Servaes, A. Tamayo, Social capital, trust, and firm performance: The value of corporate social responsibility during the financial crisis, J. Finance, 72 (2017), 1785–1824. https://doi.org/10.1111/jofi.12505 doi: 10.1111/jofi.12505
[45]	R. F. Stambaugh, Y. Yuan, China factor data, Wharton School, University of Pennsylvania, 2021. Available from: https://finance.wharton.upenn.edu/stambaug/.

Reader Comments

Your name:*

Email:*
© 2026 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)