Economics Letters 207 (2021) 110028

Contents lists available at ScienceDirect

Economics Letters
journal homepage: www.elsevier.com/locate/ecolet

Do green policies catalyze green investment? Evidence from ESG

investing developments in China
∗
Xiaoke Zhang a , Xuankai Zhao b,c , , Linshan Qu c
a
School of Economics, Peking University, Beijing, China
b
China Center for Internet Economy Research, Central University of Finance and Economics, Beijing, China
c
School of Economics, Central University of Finance and Economics, Beijing, China

article info a b s t r a c t

Article history: This study examines the heterogeneous performance of ESG investing in China before and after 2016,
Received 15 May 2021 when the ‘‘Guidelines for Establishing a Green Financial System’’ was announced. In the portfolio
Received in revised form 25 July 2021 analysis, high ESG portfolios earn significantly higher abnormal returns compared to low ESG portfolios
Accepted 28 July 2021
in the post-2016 context, but this is not the case pre-2016. This outperformance is not closely related
Available online 4 August 2021
to common firm characteristics. Next, the stock analysis also shows that good ESG profiles predict
JEL classification: higher future excess returns on average post-2016. The rising performance of good ESG stocks largely
G11 depends on equity cost advantage other than profitability improvement.
© 2021 Elsevier B.V. All rights reserved.
Keywords:
ESG investing
Portfolio analysis
Alpha
Chinese stock market
Green policy

1. Introduction of ESG investing by comparing the risk-adjusted returns (Al-

phas) of ESG portfolios relative to different factor models; at
In 2016, seven ministries and financial regulators of China the stock level, we conduct Fama–MacBeth regression to provide
jointly issued the Guidelines for Establishing a Green Financial supporting evidence about the relationship between ESG and
System (hereinafter, Guidelines), which is designed to support future excess returns. The results show that ESG investing clearly
becomes more profitable after implementation of the Guide-
sustainable development. Following the Guidelines, a sequence
lines, and high-ESG portfolios earn significantly higher abnormal
of important policies, initiatives, and rules were announced to
returns since June 2016, but not prior. Moreover, the outper-
promote environmental, social, and governance (ESG) investing.1
formance is independent of most common firm characteristics
However, the real performance of ESG investing is still unknown leading to abnormal returns.
in China. This study examines the heterogeneous performance Furthermore, we identify the drivers of rising ESG perfor-
of ESG investing before and after 2016 at both the portfolio mance by separating firm value into profitability and cost-of-
and stock levels using samples covering June 2010 –March 2019. equity capital under the discounted cash flow framework. The
This elucidates how ESG investing has evolved in the context of result shows that green policies convey a clear signal to the mar-
substantial policy change. ket, thereby increasing valuation of high ESG firms by reducing
Unlike developed markets, policies are key driving forces for cost of equity capital.
investment practices in China, as they convey clear signals from Our comparative analysis of heterogeneous ESG investing per-
top management, which are indispensable for catalyzing mar- formance highlights the impacts of green policies. In this regard,
ket activity. At the portfolio level, we examine the effectiveness we provide novel explanations for mixed evidence on the perfor-
mance of ESG investing, with previous studies having attributed
the country, industry or time intervals difference (Renneboog
∗ Correspondence to: School of Economics, China Center for Internet Economy
et al., 2008; Díaz et al., 2021; Lins et al., 2017).
Research, Central University of Finance and Economics, 39 South College Road,
Our findings also challenge the view that ESG is a risk factor
Haidian District, Beijing, China.
E-mail address: [email protected] (X. Zhao). for reward compensation (Becchetti et al., 2018), while strongly
1 In appendix, figures A1 and A2 provide some factual evidence of ESG support that ESG contains information about future fundamen-
investing. High-ESG portfolios slightly underperformed low-ESG portfolios before tals not fully priced (Edmans, 2011). This study helps to under-
2016, but outperformed them after. stand how green policies incentivize investments in green sectors

https://doi.org/10.1016/j.econlet.2021.110028
0165-1765/© 2021 Elsevier B.V. All rights reserved.
X. Zhang, X. Zhao and L. Qu Economics Letters 207 (2021) 110028

Table 1
ESG score sorted portfolio regressions.
Quintile Jun. 2010–May. 2016 June. 2016–Mar. 2019
ESG αM α3 α5 α6 α CH3 α CH4 ESG αM α3 α5 α6 α CH3 α CH4
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
1.557*** 0.766*** 0.789*** 0.811*** 1.219*** 1.285*** −0.145 0.615** 0.713*** 0.674*** 0.695*** 0.767***
1 11.28 15.39
(3.464) (4.862) (4.624) (4.106) (7.002) (8.370) (−0.305) (2.541) (3.946) (3.311) (3.597) (3.731)
1.060*** 0.281 0.439** 0.296* 0.563*** 0.566*** −0.027 0.413** 0.575*** 0.586*** 0.599*** 0.644***
2 18.09 22.11
(2.816) (1.567) (2.466) (1.737) (4.116) (3.835) (−0.080) (2.180) (3.164) (2.973) (4.312) (4.265)
0.685** 0.090 0.251 0.187 0.349* 0.479** 0.608*** 0.778*** 0.925*** 0.875*** 0.821*** 0.893***
3 23.14 31.07
(2.505) (0.476) (1.612) (0.960) (1.845) (2.604) (3.521) (4.120) (3.482) (3.024) (4.649) (5.299)
0.365* 0.248* 0.248 0.280* 0.346** 0.400** 0.980** 0.811* 0.917* 0.939* 0.975** 1.275***
4 30.00 39.55
(1.876) (1.769) (1.635) (1.726) (2.050) (2.284) (2.201) (1.903) (1.839) (1.883) (2.597) (3.144)
0.827** 0.758** 0.888** 0.904** 0.882*** 0.862*** 0.933 1.626** 1.786** 1.668** 1.326* 0.756
5 37.53 50.15
(2.386) (2.122) (2.446) (2.464) (2.705) (2.830) (1.300) (2.071) (2.593) (2.224) (1.759) (0.890)
High–Low 26.25 −0.730 −0.008 0.099 0.093 −0.337 −0.424 31.66 1.078 1.010 1.074* 0.994 0.631 −0.012
(−1.383) (−0.023) (0.261) (0.236) (−0.911) (−1.242) (1.564) (1.550) (1.777) (1.405) (0.810) (−0.013)

through the lens of ESG investing, which is crucial for encouraging show the estimates/statistical significance of alphas from six dif-
sustainable development. ferent factor models; the last row presents long–short strategy
alphas. α M is the alpha relative to RMRF ; α 3 is the alpha relative
2. Data and methodology to RMRF, SMB, and HML; α 5 is the alpha to RMRF, SMB, HML,
CMA, and RMW ; α 6 is the alpha relative to RMRF, SMB, HML, CMA,
We use Bloomberg’s data from 2010–2018 for ESG perfor- CMW and UMD; α CH3 is the alpha relative to RMRF, SMB_CH, and
mance because of its long-term coverage.2 In portfolio analysis,
VMG; α CH4 is the alpha relative to RMRF, SMB_CH, VMG and PMO.
we follow Fama and French (1993) to group individual stocks into
Newey–West t-statistics with lag 6 are reported.
five portfolios according to their ESG performance at the end of
Obviously, both the lowest and highest ESG portfolio earn
last year. Then, the risk-adjusted return of each portfolio is cal-
substantially higher risk-adjusted returns, around 0.9% per month
culated relative to risk factor models using the data from Chinese
from June 2010–May 2016. The differences in alphas between
stock market. Together, nine common risk factors are controlled.
Six factors, including market (RMRF ), size (SMB), book-to-market the highest and lowest ESG portfolios are trivial and insignificant
(HML), investment (CMA), profitability (RMW ) and momentum with an average of −0.21%. Thus, investors could not expect to
(UMD), are calculated following Fama and French (1993, 2018). profit via the simple long–short strategy of buying the high-rated
We also construct three China-specific factors proposed in Liu portfolio and selling low-rated one
et al. (2019), such as adjusted size factor (SMB_CH), EP-based The magnitude of risk-adjusted returns for the highest ESG
value factor (VMG), and turnover based the sentiment factor quintile nearly doubles during the second period (except for CH4
(PMO). The model is as follows: model), while decreases for the lowest ESG quintile. The average
high–low alpha spread is 0.65%. Although most alpha spreads
Rt − rft = α + β1 × RMRFt + β2 × SMBt + β3 × HMLt
remain statistically insignificant, the t-values are obviously larger
+ β4 × CMAt + β5 × RMWt + β6 × UMDt (1) in the later subperiod. That inspiringly implies the possibility of
+ β7 × SMB_CHt + β8 × VMGt + β9 × PMOt + ϵt using a long–short trading strategy to harvest from good ESG
performance after 2016.
Rt denotes the quintile portfolio return in percentage at month t,
Furthermore, we perform bivariate portfolio sorts to verify
which is the equally-weighted return of stocks. rft is the risk-free
that it is the ESG performance rather than other characteristics
return. α denotes the risk-adjusted return.
that drives these documented differences. Table A5 and Table
However, the portfolio method suffers from loss of substantial
A6 show observed alphas are mostly independent of common
information for stock aggregation in cross-section. We further use
firm characteristics. Most of alpha spreads of High–Low portfolios
Fama and MacBeth (1973) method to show the average predictive
are insignificantly negative before 2016, and become significantly
power of ESG on future excess returns. With time-series averages
of the slope coefficients from the regressions of stock excess positive later.
returns on ESG, the test can determine whether ESG investing has
non-zero premiums. Monthly cross-sectional regressions are run 3.2. Fama–Macbeth regression
as follows:
Table 2 presents the average slope coefficients of ESG us-
Ri,t − rf ,t = β0 + β1 × ln(ESGi,t −1 ) + β2 × X ′ i,t −1 + γind + εi,t (2) ing Fama–Macbeth regressions. Newey–West adjusted t-statistics
i represents firm identity, t represents monthly time. The depen- with lag 6 are reported. The results indicate robust profitability
dent variable is excess return. ln(ESGi,t −1 ) represents the natural for ESG investing in China after the 2016 Guidelines. While ESG
log of ESG. γind represents industry dummies. X ′ i,t −1 are control has weak predictive power for excess returns before 2016 (Panel
variables. Summary statistics are in Table A3. A), this dramatically changes after, when good ESG performance
predicts higher performance (Panel B). The average slope of lnESG
3. Empirical evidence is 0.0035 after 2016, meaning that if ESG score is enhanced by
1%, the monthly return would increase 0.35%. The coefficients of
3.1. Portfolio regression ESG remain statistically significant over six months, thus showing
long-term prediction power.
Table 1 shows abnormal return of ESG portfolios.3 Columns 1
and 8 present average ESG scores in each portfolio. Other columns 3.3. Further analysis

2 Detailed explanation of ESG data in the Appendix A2. We check whether ESG relates with stock future returns by
3 TableA4 shows results for value-weighted portfolios. including information about firm fundamentals that has not been
2
X. Zhang, X. Zhao and L. Qu Economics Letters 207 (2021) 110028

Table 2
Fama–MacBeth cross-sectional regressions.
Panel A: Jun. 2010–May. 2016
(1) (2) (3) (4) (5) (6) (7)
Rt +1 Rt +2 Rt +4 Rt +6 Rt +8 Rt +10 Rt +12
lnESG −0.021% −0.174% −0.663% −0.746% −0.964% −1.170% −2.178%
(−0.157) (−0.653) (−1.267) (−1.057) (−1.013) (−1.010) (−1.348)
Industry control Yes Yes Yes Yes Yes Yes Yes
Panel B: Jun. 2016–Mar. 2019
lnESG 0.346%** 0.726%* 1.482%* 1.894%* 1.818% 1.212% 1.398%
(2.516) (2.065) (1.869) (1.854) (1.547) (0.825) (0.714)
Industry control Yes Yes Yes Yes Yes Yes Yes

Table 3 costs among high-ESG firms and improving their market values.
Panel regressions of future profitability and cost of equity capital on ESG. Importantly, as the growing concern about climate changes, fu-
(1) (2) ture studies should put more attention on how the environmental
Profitability Cost-of-Equity
pillar of ESG materializes in global stock markets, such as Yang
ln ESGi,t −1 −0.0345*** −0.0031* (2021).
(−5.740) (−1.864)
Post × ln ESGi,t −1 0.0163 −0.0131***
Declaration of competing interest
(0.779) (−3.105)
Industry and year control Yes Yes
The authors declare that they have no known competing finan-
cial interests or personal relationships that could have appeared
to influence the work reported in this paper.
fully priced. Using the discounted cash flow framework, yearly
operating profitability and cost-of-equity capital are taken as Acknowledgments
proxies for fundamentals, to examine how ESG affects their move-
ment. The equation is as follows: We gratefully acknowledges financial support from Beijing
( ) Municipal Social Science Foundation (17JDYJB019) and Beijing
yi,t = a0 + a1 × ln ESGi,t −1 + a2 × Post × ln ESGi,t −1
(3) Outstanding Young Scientist Program (BJJWZYJH012019100-
+ β × X ′ i,t −1 + γind + δt + εi,t 34034).
yi,t denotes firm fundamentals. Post is the indicator for time pe-
riod after 2016. X ′ i,t are control variables including size, leverage, Appendix A. Supplementary data
and book-to-market ratio. γind and δt control for industry and year
fixed effects. Supplementary material related to this article can be found
Good ESG performance significantly predicts lower profitabil- online at https://doi.org/10.1016/j.econlet.2021.110028.
ity in the coming year (Table 3). The Post × ln ESGi,t −1 coefficient
is positive but insignificant, which implies that good ESG perfor- References
mance may have trivially improved profitability after 2016. This
Becchetti, L., Ciciretti, R., Dalò, A., 2018. Fishing the corporate social responsibility
shows that responsible behavior is detrimental to firm profitabil- risk factors. J. Financ. Stab. 37 (August), 25–48.
ity, which supports Friedman’s (1970) argument. ESG is slightly Díaz, V., Ibrushi, D., Zhao, J., 2021. Reconsidering systematic factors during the
negatively related to future cost-of-equity capital before 2016 Covid-19 pandemic – The rising importance of ESG. Finance Res. Lett. 38
(Table 3). Interestingly, this negative relation intensifies after (January), 101870.
Edmans, A., 2011. Does the stock market fully value intangibles? Employee
2016, which implies that firms with good ESG performance enjoy
satisfaction and equity prices. J. Financ. Econ. 101 (3), 621–640.
lower financial costs and in turn higher firm valuation. Fama, E.F., French, K.R., 1993. Common risk factors in the returns on stocks and
bonds. J. Financ. Econ. 33 (1), 3–56.
4. Conclusions Fama, E.F., French, K.R., 2018. Choosing factors. J. Financ. Econ. 128 (2), 234–252.
Fama, E.F., MacBeth, J.D., 1973. Risk, return, and equilibrium: Empirical tests. J.
Polit. Econ. 81 (3), 607–636.
Taking 2016 as a watershed year for green investment de- Friedman, M., 1970. The social responsibility of business is to increase its profits.
velopment, we elucidate the evolution of ESG investing in China In: Zimmerli, Walther Ch., Holzinger, Markus, Richter, Klaus (Eds.), Corporate
over the last decade. Our portfolio and stock-level analyses show Ethics and Corporate Governance. Springer, Berlin, Heidelberg, pp. 173–178.
that ESG was inconspicuous in the investment process before Lins, K.V., Servaes, H., Tamayo, A., 2017. Social capital, trust, and firm perfor-
mance: The value of corporate social responsibility during the financial crisis:
2016, however become incrementally important for enhancing
Social capital, trust, and firm performance. J. Finance 72 (4), 1785–1824.
portfolio performance after, with high-ESG profile stocks earning Liu, J., Stambaugh, R.F., Yuan, Y., 2019. Size and value in China. J. Financ. Econ.
substantially higher abnormal returns. After the 2016 Guide- 134 (1), 48–69.
lines, a sequence of measures and initiatives were announced to Renneboog, L., Ter Horst, J., Zhang, C., 2008. Socially responsible investments:
promote green investment practices, thereby reducing financial Institutional aspects, performance, and investor behavior. J. Bank. Financ. 32
(9), 1723–1742.
Yang, B., 2021. Explaining greenium in a macro-finance integrated assessment
model. SSRN Electron. J..