What drives green betas? Climate uncertainty or speculation

Full text

What drives green betas? Climate uncertainty or speculation
Onur Polata,b, Riza Demirerc,˙
Ibrahim Halil Ek¸sid
December 2023
Finance Research Letters, forthcoming
Abstract
Examining green equity sectors including geothermal, wind, solar, bioclean, and clean energy within a
DCC-MIDAS framework, we show that green betas are predominantly driven by speculative sentiment in
technology stocks rather than climate uncertainty. We argue that the lottery-like features of green assets,
whose values are highly tied to the success of new technologies, result in a negative risk-mispricing rela-
tionship driven by technology speculation. Further economic analysis shows that a forward-looking invest-
ment strategy conditional on technology sentiment yields improved risk-adjusted returns for passive equity
investors, particularly following the 2016 Paris Agreement. Our findings establish a new speculation-based
channel for characterizing the systematic risk of green assets.
Keywords: Beta, Climate risk, Speculative sentiment, Asymmetric DCC-GARCH, Reverse MIDAS.
JEL Classification: C22, D81, G15, Q42, Q54, Q56.
_______________________________________________________
aDepartment of Applied Statistics and Operations Research, Universitat Politècnica de València, 03801 Al-
coy, Spain. Email: opolat@upvnet.upv.es
bDepartment of Public Finance, Bilecik ¸Seyh Edebali University, Bilecik, Turkiye.
cCorresponding author. Department of Economics and Finance, Southern Illinois University Edwardsville,
School of Business, Edwardsville, IL 62026-1102, USA. Email: rdemire@siue.edu
dFaculty of Economics and Administrative Sciences, Gaziantep University, Turkiye. Email: eksihalil@gmail.com

1. Introduction
According to the Capital Asset Pricing Model (CAPM), exposure to aggregate market fluctuations mea-
sured by beta is the sole asset-specific determinant of expected returns. Beta estimates are often used by
fund managers in their strategic and tactical allocation models and are essential in risk management appli-
cations to hedge market exposures. However, the market forces that drive asset betas are often understudied,
although beta is a critical input in financial decisions. While prior works have established a significant re-
lationship between uncertainty and systematic risk in equity markets (e.g. Yu et al., 2017; Naeem et al.,
2020), the literature has not yet extended the analysis to green equities that have attracted much attention
in the burgeoning climate finance literature. The main contribution of this paper is to extend the emerging
literature on climate finance in a novel direction by exploring the role of climate uncertainty as a driver of
market betas using data from a wide range of green equity sectors that capture geothermal, wind, solar, bio-
clean, and clean energy stocks. This is an important consideration as asset betas capture market exposures,
which in turn serve as determinants of expected returns. As another novelty of our analysis, building on
the evidence that establishes a close link between technology stocks and renewable energy companies (e.g.
Kumar et al., 2012; Inchauspe et al., 2015), we explore whether sentiment in technology stocks plays a pre-
dictive role in green betas. Indeed, our results show that green betas are predominantly driven by speculative
sentiment in technology stocks rather than climate uncertainty, thereby establishing a new channel that links
the clean energy market to the technology industry. Subsequently, we propose a forward-looking investment
strategy in green industries that is conditional on technology sentiment and presents evidence of improved
risk-adjusted returns for passive equity investors.
The literature on climate finance is growing rapidly, and there is now a well-established literature on
the role of climate as a driver of stock returns (e.g. Bolton and Kacperczyk, 2021; Dutta et. al, 2023;
Faccini et al., 2023), while recent studies also show that the uncertainty associated with climate policy is
a dominant determinant of return and volatility in this growing asset class (Bouri et al., 2023). In a recent
study, Treepongkaruna et al.(2023) show that stocks with low exposure to climate policy uncertainty (CPU)
earn 5.5%-6.3% higher returns, suggesting that uncertainty-averse investors are willing to pay a premium
on low-CPU exposure stocks. The importance of climate risk on firm valuations is further highlighted by
the growing attention on "double materiality" that aims to capture the bidirectional interactions between
firm operations and climate-related impacts in the financial reporting regulations (Chiu, 2022; Pizzi et al.,
2023; Carvajal and Nadeem, 2023; Xie et al., 2023). Separately, a burgeoning literature establishes a close
link between technology stocks and renewable energy companies as the performance of alternative energy
companies is highly sensitive to the success of technological innovations in their efforts to transition into
1

a less carbon-intensive economy (Bouri et al., 2023). Thus, studies including Kumar et al. (2012) and
Inchauspe et al. (2015) argue that investors’ perceptions of clean energy and technology stocks are often
similar.Interestingly, however, the literature has not yet explored the potential impact of these dynamics on
industry betas, particularly on the clean energy industry. Our key finding is that green betas are predomi-
nantly driven by speculative sentiment in technology stocks rather than climate uncertainty, thus establishing
a new investor sentiment channel that links clean energy and technology stocks. We argue that the lottery-
like features of green assets, whose values are highly tied to the success of new technologies, result in a
negative risk-mispricing relationship driven by speculation in technology stocks. Further economic analysis
shows that a forward-looking investment strategy conditional on technology sentiment yields improved risk-
adjusted returns for passive equity investors who supplement their positions with green stocks, particularly
following the 2016 Paris Agreement.
It is worth noting that financial regulators encourage market actors to transition their investments from
high-carbon to low-carbon assets, while also encouraging efforts to diminish the carbon footprint of exist-
ing high-carbon operations (Gunningham, 2020), fostering investments in green assets, as highlighted by
D’Orazio and Popoyan (2019). Considering the evidence that investors already demand compensation for
their exposure to climate-related risks, our findings are important in characterizing the systematic risk of
green assets with significant investment implications. This is especially important considering that asset
betas are not directly observable and the growing emphasis on double materiality in financial reporting will
only make the assessment of risk exposures in green assets more challenging. In that regard, our findings
offer novel insight to the assessment of systematic risks in this growing asset class by focusing on the role
of speculative sentiment, rather than the dominant focus on climate risk. This reveals a novel connection
between the clean energy market and the technology sector, creating a new pathway for understanding their
interplay. Moreover, our findings offer a forward-looking investment strategy that relies on technology
sentiment and provides substantiation of enhanced risk-adjusted returns for passive equity investors.
The remainder of the paper is organized as follows. Sections 2 and 3 describe the sample data and
methodology. Section 4 presents the findings and Section 5 concludes with several remarks.
2. Data
We use daily data for five green equity indexes, namely the NASDAQ OMX Bio/Clean Fuels (GRN-
BIOX), Wind (GRNWIND), Solar (GRNSOLAR), and Geothermal (GRNGEO) total return indexes and the
iShares Global Clean Energy ETF (ICLN), over the period October 21, 2010–March 2, 2023, obtained from
Datastream. The market index is represented by the MSCI World Index, which captures large- and mid-cap
2

representation across 23 developed markets. The Climate Policy Uncertainty (CPU) index of Gavriilidis
(2021) is obtained from the Economic Policy Uncertainty database.1To capture investor sentiment in tech-
nology stocks, we use the speculative ratio originally proposed by Garcia et al. (1986) as a model-free
measure of the speculative tendencies of investors based on trading volume and open interest information
obtained from futures market transactions. For this purpose, we collect data for NASDAQ 100 futures
contracts from Commodity Systems Inc. that track the performance of the NASDAQ 100 index, which in-
corporates non-financial firms from industries that drive global innovation and growth. Following Balcilar
et al. (2017), the technology speculative ratio (TSR) on a given day tis computed as Vt/OItwhere Vtis the
trading volume during period t and OItis the value of the open interest at the end of the same period. Con-
sidering that speculators tend to engage in short-term positions compared with hedgers, Lucia et al. (2015)
argue that speculators’ trading activity of taking opposite positions during a given period results in faster
growth in trading volume relative to open interest, thus leading to a higher value for TSR, which indicates
the relative importance of speculative activity in the market with respect to hedging.
The time series plots in Figure A1 in the Appendix display a notable spike in the technology speculation
series following the official announcement of the COVID-19 pandemic by the World Health Organization
(WHO) on March 11, 2020, which also coincides with a positive shift in the green index values during that
period. Likewise, we observe a notable spike in the CPU index in mid-2016 following the initiation of the
Paris Climate Agreement and later in 2021 following the presidential elections. Although not reported due
to space considerations, the descriptive statistics show that Solar records the highest mean and volatility in
returns among the green equity indices. With the exception of Geothermal, all green equity indexes exhibit
a left-skewed distribution, indicating large negative outliers in daily returns.
3. Methodology
3.1. Estimating green betas
We estimate the betas for green equity indexes via the Asymmetric Dynamic Conditional Correlation
(ADCC) Model of Cappiello et al. (2006) which allows us to jointly estimate the time-varying covariance
between the green industries and the market index. Given the n ×1 vector of returns, rt, the mean equation
on the information set It−1is specified as
rt=µ+ψrt−1+εt(1)
1https://www.policyuncertainty.com/climate_uncertainty.html
3

with the residuals represented by εt=H1/2
tztwhere Htis the conditional covariance matrix of rtand ztis a
n×1 vector of i.i.d errors. Formulating Htas Ht=D1/2
tRtD1/2
twhere Dt=diag hi,t,...,hn,tcaptures the
diagonal conditional variances, the conditional correlation matrix Rtis defined as
Rt=diag q−1/2
1,t, . . . q−1/2
n,tQtdiag q−1/2
1,t, . . . q−1/2
n,t(2)
where Qtis a symmetric positive definite matrix with Qt=(1−θ1−θ2)Q+θ1zt−1z′
t−1+θ2Qt−1. Here, Q
denotes the n ×n unconditional matrix of the standardized residuals zi,twhere θ1and θ2are non-negative
satisfying the condition θ1+θ2<1. The beta estimator for green industry iis then formulated as βi,t=qi,m,t
qm,m,t
where mdenotes the market index. Cappiello et al. (2006) propose an asymmetric version of the DCC
model of Engle (2002) in which the conditional volatility of the GARCH(1,1) model is defined as
hi,t=ωi+αiε2
i,t−1+τihi,t−1+γiε2
i,t−1Iεi,t−1(3)
where the indicator function It−1=1 if εi,t−1<0 otherwise It−1=0.In this setting, the "asymmetric" effect
is captured by a positive value for d which implies that positive residuals tend to increase variance less than
negative residuals. Thus, Qtis represented by
Qt=¯
Q−A′¯
QA −B′¯
QB −G′¯
Q−G+A′zt−1z′
t−1A+B′Qt−1B+G′z−
tz′−
tG(4)
where A,Band Gare n ×n parameter matrices and z−
tare zero-threshold standardized errors with an uncon-
ditional matrix ¯
Q−.
3.2. Reverse MIDAS Model
Given that climate policy uncertainty is available on a monthly basis while the beta estimates are ob-
tained from daily returns, we follow the reverse-MIDAS (R-MIDAS) model, derived from the mixed data
sampling framework of Ghysels et al. (2006), to predict high-frequency beta series based on the low-
frequency CPU series. Assuming a dependent variable xgenerated by an AR(p) process and variable yas
the exogenous predictor, where xis observed at high frequency every t=1
kperiods, while ycan only be
observed at low frequency every kperiods, the R-MIDAS follows the form
xt=λi(Lk+i)yt+σ1iAi(L, θi)xt−1/k+ϵt(5)
t=0+i
k,1+i
k+1,2+i
k+2(6)
i=0, ..., k−1 (7)
4

where Ldenotes the lag operator operating in high frequency and Ai(L, θi) can be an exponential Almon lag
polynomial
Ai(L, θi)=
Q
X
j=0
ai(j, θi)Lj,ai(j, θi)=ex p(θi1j+θi2j2
PK
j=0ex p(θi1j+θi2j2.(8)
In this setting, the daily beta series, xt, is always related to the latest available value of the low-frequency
monthly CPU series, y, and its monthly lags, and to the latest and additional daily lags of xitself. In our
application, our predictors include monthly CPU series along with the daily technology speculation series.
4. Empirical Results
4.1. Drivers of green betas
The ADCC model results reported in Table 1indicate positive and statistically significant short-term
persistence (α) for most green equity indexes, which are less than the long-term persistence parameters (τ),
suggesting the presence of volatility clustering in green returns. The significant γvalues reveal asymmetric
effects, indicating that positive and negative residuals tend to impact conditional volatility in different mag-
nitudes. The time-varying betas, reported in Figure 1, take on values above unity for most green equities,
notably Solar, highlighting the risky nature of these assets whose fluctuations are highly tied to the success
of new technologies (Bouri et al., 2023). The estimated betas display notable spikes in 2020, likely triggered
by the COVID-19 pandemic, as the beta values drop back to unity subsequently, with the exception of Solar
and Wind displaying a rising pattern in betas during the latter part of the sample.
Although the theoretical framework that one can use to relate betas to uncertainty is rather limited,
recent studies offer some theoretical insight to the role of speculation as a determinant of the risk-return
tradeoffs in equities. In the first approach, Hong and Sraer (2006) argue that high beta assets are more
sensitive to disagreement among investors which makes them prone to speculative overpricing due to short-
sale constraints. This, in turn, obscures the risk-return relationship for high beta stocks – which green
equities certainly qualify as, leading to a negative expected return-beta relationship when disagreement is
high. In the second approach, Ghazi and Schneider (2022) propose a framework in which asset returns are
decomposed into speculation and non-speculation components and show that the non-speculative component
captures a positive risk premium that the standard models hypothesize, while the speculative component is
associated with a negative speculative premium. The authors show that these two components of returns can
help explain the risk and mispricing related factors that relate to CAPM anomalies including the well-known
beta anomaly.
5

In line with the arguments stated above, examining the R-MIDAS estimates reported in Table 2, we
find that technology speculation (TSR) has a robust impact on green betas consistently across all five green
industries. While we obtain mixed results for climate uncertainty with insignificant effects observed for
Wind and Geothermal, we find that technology speculation has a consistently negative effect on green betas
in all industries at the highest level of statistical significance.2This result could be explained by the lottery-
like features in these high-beta assets, which make them more subject to speculative demand. Bali et al.
(2017) note that lottery investors generate demand for stocks with high upside probabilities in the stock
price, which is more likely during periods of high speculative sentiment. The disproportionately high price
pressure based on lottery demand then pushes the prices of these stocks up, thereby decreasing future returns.
This, in turn, generates an intercept greater than the risk-free rate (positive alpha zero-beta stocks) and a
slope less than the market risk premium, leading to a negative alpha on these stocks. Our results show that
this manifests itself in the form of a negative relationship between speculation and betas, which is consistent
with the recent evidence in Ghazi and Schneider (2022) that lottery-like stocks earn a negative speculative
premium. Indeed, examining the green betas and average returns during high and low market states reported
in Table 3, we observe in Panel A that green industries experience consistently lower betas during periods
of high technology speculation, indicated by negative mean differences in betas (High-Low), while we also
observe in Panel B that these higher green betas during high speculation are coupled with negative mean
returns in all green industries, supporting a negative risk premium conditional on speculation. These findings
are also in line with the evidence in Scheinkman and Xiong (2003) that stocks are overpriced when the level
of speculative trading is high as lottery-seeking investors drive up the valuations of these stocks, leading to
their underperformance.
4.2. Economic Analysis
To explore the economic implications of our findings, given that the return-beta trade-offfor green
equities is driven by speculative sentiment in technology stocks, we examine a forward-looking investment
strategy that is conditional on the level of speculation. To that end, we consider a passive investor currently
invested in the MSCI World index and form a diversified portfolio by supplementing the passive index with
each green industry index separately. Using the conditional volatility and covariance estimates obtained
from the ADCC model, we estimate each month the optimal portfolio allocation for the passive market
2Our findings are robust to the inclusion of the global VIX and EPU as control variables in the model as presented in Table A1
in the Appendix.
6

index and each green industry via
wx/y
t=hy
t−hx/y
t
hx
t−2hx/y
t+hy
t
,wx/y
t=

















0,if wx/y
t<0
wx/y
t,if 0 ≤wx/y
t≤1
1,if wx/y
t>1
(9)
where hx/y
tis the conditional covariance and the weight of the passive index xin a one-dollar portfolio of
the two assets is given by wx/y
t. Next, following Cepni et al. (2022), we supplement the passive market
index with a position in each green industry when speculative sentiment is low and remain passive during
the high speculation state, characterized by the (TSR) values above the 90th percentile of the trailing 6-
month empirical distribution. We then compare the performance metrics of the subsequent returns for the
passive market index against the active portfolio strategy supplemented with each green industry one at a
time. Furthermore, motivated by the finding by Monasterolo and Angelis (2020) that the Paris Climate
Agreement (PCA) has had a prominent impact on the systematic risk for green equities, we further split the
sample into pre- and post-PCA to examine the efficiency of our active portfolio strategy.
The portfolio metrics reported in Table 4indicate that supplementing the passive market index with
green equity indexes conditional on speculative sentiment provides remarkable benefits, particularly during
the post-Paris Agreement period. While the results for the pre-Paris period are dismal, we observe in Panel
B that all portfolios supplemented with green industries, except wind yields, improved returns compared
with the passive market index during the post-Paris agreement period. Bioclean and Solar stand out with the
most significant improvement in portfolio returns of 0.0719% and 0.0967%, respectively. In contrast, the
active portfolio supplemented with positions in Solar offers significantly improves risk-adjusted returns with
Sharpe ratio and Treynor index value of 4.117 and 0.066 percent, respectively, compared to 2.161 and 0.025
percent for the passive market index. In short, our results show that the effect of technology sentiment on
systematic risk and return dynamics in green equity sectors can indeed be exploited within a forward-looking
investment strategy to supplement passive investments with green equities.
5. Conclusion
Beta is a key factor in assessing the market exposure of assets and beta estimates are often used by fund
managers in their strategic and tactical allocation models. We extend the emerging literature on climate
finance in a novel direction by exploring the drivers of betas in a variety of green equity sectors, including
geothermal, wind, solar, bioclean, and clean energy stocks. Our findings show that green betas are pre-
dominantly driven by speculative sentiment in technology stocks rather than climate uncertainty. We argue
7

that the lottery-like features of green assets whose values are highly tied to the success of new technologies
result in a negative risk-mispricing relationship driven by speculation in technology stocks. Subsequently,
we show that a forward-looking investment strategy conditional on technology sentiment can offer improved
risk-adjusted returns for passive equity investors. Our findings establish a new speculation-based channel
for characterizing the systematic risk of green assets. This is an important consideration as betas are not di-
rectly observable although beta estimates are widely used by fund managers in investment decisions. Thus,
our findings offer an alternative setting to estimate betas to be utilized in tactical asset allocation models
for this growing asset class. Furthermore, our findings open a new channel of interaction between green
and technology stocks from a new perspective, highlighting the role of technology sentiment on green asset
performance. Finally, our findings underscore the lottery-like nature of green equities, tied to the success of
emerging technologies, and suggest that mispricing in these assets could be driven by investor sentiment in
technology stocks, which in turn, can be used in betting against beta strategies that involve green equities.
8

References
Balcılar, M., R. Demirer, T. Ulussever (2017). Does speculation in the oil market drive investor herding
in emerging stock markets? Energy Economics 65, 50–63.
Bali, T., Brown, S., Murray, S. and Tang, Y. (2017). A lottery-demand-based explanation of the beta
anomaly. Journal of Financial and Quantitative Analysis 52, 2369–2397.
Bolton, P. and Kacperczyk, M. (2021). Do investors care about carbon risk? Journal of Financial
Economics 142, 517–549.
Bouri, E., Iqbal, N., Klein, T. (2022). Climate policy uncertainty and the price dynamics of green and
brown energy stocks. Finance Research Letters 47, 102740.
Bouri, E., Dudda, T. L., Rognone, L. & Walther, T. (2023). Climate risk and the nexus of clean energy
and technology stocks. Annals of Operations Research. https://doi.org/10.1007/s10479-023-05487-z
Cappiello, L., Engle, R. F., and Sheppard, K. (2006). Asymmetric dynamics in the correlations of global
equity and bond returns. Journal of Financial Econometrics, 4(4), 537-572.
Carvajal, M., Nadeem, M. (2023). Financially material sustainability reporting and firm performance in
New Zealand. Meditari Accountancy Research, 31(4), 938-969.
Cepni, O., Demirer, R., Rognone, L. (2022). Hedging Climate Risks with Green Assets. Economics
Letters 212, 110312.
Chiu, I. H. (2022). The EU sustainable finance agenda: developing governance for double materiality
in sustainability metrics. European Business Organization Law Review, 23(1), 87-123.
D’Orazio, P., Popoyan, L. (2019). Fostering green investments and tackling climate-related financial
risks: Which role for macroprudential policies?. Ecological Economics, 160, 25-37.
Dutta, A., Bouri, E., Rothovius, T., Uddin, G. S. (2023). Climate risk and green investments: New
evidence. Energy, 265, 126376.
Engle, R. (2002). Dynamic conditional correlation: A simple class of multivariate generalized autore-
gressive conditional heteroskedasticity models. Journal of Business & Economic Statistics, 20(3), 339-350.
Engle, R., and Colacito, R. (2006). Testing and valuing dynamic correlations for asset allocation. Journal
of Business & Economic Statistics, 24(2), 238-253.
Faccini, R., Matin, R. and Skiadopoulos, G. (2023). Dissecting climate risks: Are they reflected in stock
prices? Journal of Banking and Finance 155, 106948.
Garcia, P., Leuthold, R. M., Zapata, H. (1986). Lead-lag relationships between trading volume and price
variability: New evidence. The Journal of Futures Markets (1986-1998), 6(1), 1.
Gavriilidis, K. (2021). Measuring climate policy uncertainty. Available at SSRN 3847388.
9

Ghazi, S. and Schneider, M. (2022). Market Risk and Speculation Factors (Sep. 2, 2022). Available at
SSRN: https://ssrn.com/abstract=4187398
Ghysels, E., Santa-Clara, P., Valkanov, R. (2006). Predicting volatility: getting the most out of return
data sampled at different frequencies. Journal of Econometrics, 131(1-2), 59-95.
Gunningham, N. (2020). A quiet revolution: Central banks, financial regulators, and climate finance.
Sustainability, 12(22), 9596.
Hong, H., Sraer, D. A. (2016). Speculative Betas. The Journal of Finance 71 (5), 2095-2144.
Inchauspe, J., Ripple, R. D., & Trück, S. (2015). The dynamics of returns on renewable energy compa-
nies: A state-space approach. Energy Economics 48, 325–335.
Kumar, S., Managi, S., & Matsuda, A. (2012). Stock prices of clean energy firms, oil and carbon
markets: A vector autoregressive analysis. Energy Economics, 34(1), 215–226.
Lucia, J. J., Mansanet-Bataller, M., Pardo, Á. (2015). Speculative and hedging activities in the European
carbon market. Energy Policy, 82, 342-351.
Markowitz, H. (1952). Portfolio selection, Journal of Finance, 7:1, 77–91.
Monasterolo, I., and De Angelis, L. (2020). Blind to carbon risk? An analysis of stock market reaction
to the Paris Agreement. Ecological Economics, 170, 106571.
Naeem, M. A., Balli, F., Shahzad, S. J. H., de Bruin, A. (2020). Energy commodity uncertainties and
the systematic risk of US industries. Energy Economics, 85, 104589.
Pizzi, S., Principale, S. and De Nuccio, E. (2023). Material sustainability information and reporting
standards. Exploring the differences between GRI and SASB. Meditari Accountancy Research 31 (6), 1654-
1674.
Scheinkman, J. A., Xiong, W. (2003). Overconfidence and Speculative Bubbles. Journal of Political
Economy 111, 1183-1219.
Treepongkaruna S., Chan K. F., Malik, I. (2023). Climate policy uncertainty and the cross-section of
stock returns. Finance Research Letters 55, 103837.
Yousaf, I., Suleman, T., Demirer, R. (2022). Green investments: A luxury good or a financial necessity?
Energy Economics 105, 105745.
Xie, J., Tanaka, Y., Keeley, A. R., Fujii, H., Managi, S. (2023). Do investors incorporate financial
materiality? Remapping the environmental information in corporate sustainability reporting. Corporate
Social Responsibility and Environmental Management 30 (6), 2924-2952.
Yu, H., Fang, L., Du, D., Yan, P. (2017). How EPU drives long-term industry beta. Finance Research
Letters 22, 249–258.
10

Figure 1: Time-varying green betas.
Note: This figure plots the time-varying betas for green equity indices obtained via the Asymmetric DCC-GARCH model.
11

Table 1: Asymmetric DCC-GARCH parameter estimates
Bio/Clean Wind Solar Geothermal GCI
µ0.00038* 0.000649*** 0.000636*** 0.000518** 0.000369
ψ0.047823*** 0.062645*** 0.084555*** 0.013424 0.020862
ω0.000002*** 0.000007 0.000006** 0.000007 0.000003
α0.044996* 0.049039*** 0.036509*** 0.045533 0.052973
τ0.920305*** 0.900163*** 0.926681*** 0.890472*** 0.912033
γ0.063691** 0.057728** 0.049886** 0.084367** 0.051225
DCC Estimation
θ1θ2Akaike Bayes Shibata Hannan-Quinn
0.0166*** 0.9658** -37.238 -37.123 -37.239 -37.197
Note: This table reports the ADCC model estimates for green equity sectors. *, **, *** denote significance at 10, 5, 1% levels,
respectively.
12

Table 2: The effect of speculation and climate uncertainty on green betas
c CPU TSR R2
Bioclean -0.136 0.268*** 0.073
(0.434) (0.085)
1.171*** -0.299*** 0.087
(0.065) (0.086)
-0.065 0.254*** -0.286*** 0.152
(0.486) (0.093) (0.086)
Wind 2.507 -0.293*** 0.143
(0.285) (0.057)
1.086*** -0.161*** 0.041
(0.049) (0.061)
2.551*** -0.301*** -0.176*** 0.194
(0.290) (0.058) (0.056)
Geothermal 0.444 0.095 0.012
(0.364) (0.076)
0.910*** -0.207*** 0.054
(0.048) (0.073)
0.495 0.085 -0.203*** 0.064
(0.358) (0.076) (0.075)
Solar 2.265*** -0.137 0.013
(0.667) (0.141)
1.602*** -0.275*** 0.053
(0.072) (0.072)
2.336*** -0.151 -0.282*** 0.069
(0.681) (0.144) (0.071)
GCI 2.099*** -0.163*** 0.045
(0.271) (0.057)
1.307*** -0.234*** 0.089
(0.044) (0.064)
2.160*** -0.175*** -0.243*** 0.14
(0.286) (0.062) (0.062)
Note: This table reports the MIDAS regression estimates for green betas regressed against climate policy uncertainty (CPU) and
technology speculative ratio (TSR). *, **, *** denote significance at 10, 5, 1% levels, respectively (standard errors reported in
parentheses).
13

Table 3: Green betas, returns and market states
Panel A: Green betas
Technology Speculation Climate Uncertainty
High Low High–Low High Low High–Low
Bioclean 1.1785 1.1771 0.0013 1.3741 1.0927 0.2814***
(0.053) (2.771)
GCI 1.1893 1.3259 -0.1365*** 1.2643 1.2960 -0.031
(-4.707) (-0.407)
Geothermal 0.8554 0.8722 -0.016 0.9842 0.8689 0.1152
(-0.637) (1.3613)
Solar 1.3995 1.5899 -0.1904*** -0.00134 0.01346 -0.0148
(-5.205) (-0.1301)
Wind 0.7286 1.1049 -0.3762*** 0.00322 0.00418 -0.00096
(-23.29) (-0.0086)
Panel B: Green returns
Technology Speculation Climate Uncertainty
High Low High–Low High Low High–Low
Bioclean -0.52% 0.079% -0.60%*** 0.014% 0.41% -0.39%
(-4.363) (-0.9093)
GCI -0.44% 0.060% -0.50%*** -0.017% 0.093% -0.11%
(-3.936) (-0.3302)
Geothermal -0.32% 0.064% -0.39%*** 0.071% -0.21% 0.29%
(-2.762) ( 0.7969)
Solar -0.39% 0.10% -0.49%*** -0.49% 0.19% -0.69%*
(-3.238) (-1.4324)
Wind -0.30% 0.063% -0.36%*** -0.12% 0.13% -0.25%
(-3.017) (-0.7685)
Note: Panel A (B) reports the average green betas (green equity returns) during high and low market states associated with each
column. A High (Low) market state is characterized by the index value above (below) the 90% percentile. The column ‘High-Low’
reports the mean difference between the high and low market states. *, **, *** denote significance at 10, 5, 1% levels, respectively
(t-statistics reported in parentheses).
14

Table 4: Economic implications
Panel A: Pre-Paris Agreement Period (10/22/2010-12/31/2015)
Bioclean Wind Solar Geothermal GCI MSCI World
Average -0.0032% 0.0265% -0.0238% -0.0098% -0.0455% 0.0209%
Std. Dev. 1.514% 0.936% 1.796% 1.265% 1.624% 0.9731%
SR -0.319% 1.620% -1.337% -0.792% -2.814% 2.381%
Beta 0.946 1.107 1.344 0.694 1.344
TR -0.003% 0.020% -0.017% -0.014% -0.034% 0.020%
Panel B: Post-Paris Agreement Period (1/4/2016-3/2/2023)
Bioclean Wind Solar Geothermal GCI MSCI World
Average 0.0719% 0.0142% 0.0967% 0.0394% 0.0362% 0.0255%
Std. Dev. 1.282% 1.733% 2.270% 1.927% 1.754% 1.624%
SR 5.019% 0.631% 4.117% 1.880% 1.879% 2.161%
Beta 0.999 0.732 1.399 0.810 1.220
TR 0.022% 0.014% 0.066% 0.044% 0.027% 0.025%
Note: Panel A(B) reports the performance metrics for the dynamic investment strategy that involves the passive market index and
each green equity index before (after) the 2016 Paris Climate Agreement. Beta, SR, and TR are the beta, Sharpe, and Treynor ratios
for the dynamic portfolio strategy, respectively.
15

Appendix
Figure A1: Green Equity Indices, CPU, and Technology Speculation
16

Table A1: The effect of speculation, climate and economic uncertainty on green betas
c CPU TSR GEPU VIX R2
Bioclean -0.118 0.287*** 0.287 - 0.263 0.073
(0.838) (0.001) (0.202) (0.205)
1.024*** -0.293*** -0.122 0.050 0.089
(0.451) (0.082) (0.207) (0.156)
-0.017 0.270*** -0.277*** -0.179 -0.041 0.159
(0.641) (0.098) (0.081) (0.186) (0.142)
Wind 2.995*** -0.242*** -0.219 0.250*** 0.207
(0.302) (0.053) (0.158) (0.069)
2.083*** -0.152*** -0.222 -0.339 0.152
(0.049) (0.061) (0.154) (0.078)
3.056*** -0.252*** -0.167*** -0.168 -0.253 0.252
(0.309) (0.053) (0.055) (0.161) (0.070)
Geothermal 0.505 0.130 -0.463 -0.076 0.061
(0.648) (0.071) (0.211) (0.187)
1.030* -0.191*** -0.382 -0.040 0.086
(0.561) (0.069) (0.207) (0.190)
0.571 0.119 -0.184** -0.408* -0.0180 0.103
(0.659) (0.072) (0.072) (0.205) (0.186)
Solar 2.230*** -0.109 -0.467 -0.032 0.038
(0.805) (0.137) (0.201) (0.197)
1.842*** -0.257*** -0.414* -0.080 0.076
(0.531) (0.068) (0.190) (0.186)
2.326*** -0.125 -0.265** -0.387* -0.037 0.087
(0.797) (0.140) (0.069) (0.191) (0.194)
GCI 1.836*** -0.163*** -0.303* 0.090 0.068
(0.515) (0.053) (0.167) (0.141)
1.234*** -0.223*** -0.0271 0.025 0.106
(0.393) (0.060) (0.146) (0.143)
1.920*** -0.178*** -0.233*** -0.233 0.086 0.155
(0.536) (0.055) (0.061) (0.159) (0.142)
Note: This table reports the MIDAS regression estimates for green betas regressed against climate policy uncertainty (CPU) and
technology speculative ratio (TSR) along with control variables including the CBOE Volatility Index (VIX), and global economic
policy uncertainty index (GEPU). *, **, *** denote significance at 10, 5, 1% levels, respectively (standard errors reported in
parentheses).
17