1 of 42

��CEO Environmental & Safety Compensation Goals Disconnect: Evidence from the Oil & Gas Industry

Sudheer Chava (Georgia Tech), Lubo Litov (OU),

Bing Xu (OU), Runzu Wang (OU)

2 of 42

Motivation

With the increasing public awareness of environmental & safety issues, how do companies improve their environmental & safety (i.e., sustainability) performance?

More specifically, we examine whether CEO’s Annual Incentive Plan (AIP) includes corporate sustainability goals and, if so, how do such goals impact corporate sustainability performance.

3 of 42

Research Questions

What is the structure of sustainable compensation in CEO annual incentive plans (e.g., targets, quantifiable targets, modifiers)?

Who chooses to adopt AIP sustainable compensation goals?

How common are AIPs with such goals? We find that mostly oil & gas firms (i.e., toxic firms) employ sustainable AIP goals.

4 of 42

Research Questions

Do AIP sustainable goals lead to reduction in carbon emissions, environmental violations, and toxic chemicals releases?

Do AIP sustainable goals lead to increase in green innovation (patenting)?

Do firms with AIP sustainable goals grow faster and remain profitable?

Are the effects above causal?

5 of 42

Contributions

We specifically focus on environmental & safety (i.e., sustainable) goals in the AIP of CEOs and characterize the structure of such compensation goals.

We show three elements of the compensation agreement related to sustainable goal setting: (i) sustainable targets; (ii) sustainable targets with quantitative weights; and (iii) sustainability modifiers (bonuses or penalties).

6 of 42

Contributions

Importantly, we show that only a small fraction of Fortune 250 firms are including sustainability goals in AIP.

Among those, nearly all are oil & gas firms. Hence, a disconnect.

Shifting attention to oil & gas firms exclusively, we show that environmental goals are effective in reducing carbon emissions, toxicity, and environmental violations only for high polluters.

7 of 42

There is Limited Prior Literature

Hong et al (2016) links corporate governance to the existence of executive compensation incentives for corporate social responsibility (CSR) and indicate that firms with more shareholder-friendly corporate governance are more likely to offer compensation to executives linked to firm’s social performance outcomes.

Flammer et al (2019) examines how CSR criteria is integrated in executive compensation and find that the adoption of CSR contracting can help increase social and environmental initiatives, and green innovations.

Cohen et al (2021) depicts an empirical pattern in green patents production and show that oil, gas, and energy producing firms are key innovators in the United States’ green patent landscape and that these firms produce more, and significantly higher quality, green innovation.

8 of 42

There is Limited Prior Literature

Bebchuk and Tallarita (2022) document the negative effects of ESG goals in CEO contracts on the shareholder wealth.

Cohen, Kadach, Ormazabal, & Reichelstein (2022) would be the closest to ours.

They show that a coarse definition of CEO ESG goals in an international sample improves overall CO2 emissions.
In their regional analysis, they document that the ESG goals has no impact on U.S. firms on average.
Our study are different from theirs that we look deeply into the U.S. firms, and test under what condition the sustainability related incentives would contribute.
We also look at a broader set of environmental outcomes beyond the carbon emissions.

9 of 42

The number of Fortune 250 firms with sustainable targets increase steadily in the sample period, with some small drops in 2003 and 2014.

10 of 42

Similar conclusion when we measure it as a percentage of total firms.

11 of 42

Red columns for Fortune 250 oil & gas firms with sustainability targets and blue columns for Fortune 250 non-oil & gas firms with sustainability targets.
In 2000, non-oil & gas firms with sustainability targets debuted in Fortune 250 but oil & gas with sustainability targets dominated in # till 2020.
No less, oil & gas firms with sustainability targets has seen an overall drop relative to Fortune 250 companies with sustainability targets.

Note the decrease in the % of oil & gas firms with sustainable goals – at the early part of the sample nearly 100%.

The line indicates the ratio of the number of petroleum firms with ESG targets in Fortune 100 over the number of total Fortune 100 firms with ESG targets, i.e., # petroleum firms with ESG targets/# total firms with ESG targets

We use red columns to represent Fortune 100 petroleum firms with ESG targets and blue columns to show Fortune 100 non-petroleum firms with ESG targets. From 2000, non-petroleum firms with ESG targets debuted in Fortune 100, but petroleum firms with ESG targets always dominated in number until 2020

Despite its dominance, petroleum firms with ESG targets has seen a flexural drop in its fraction relative to the entire Fortune 100 companies with ESG targets

12 of 42

From 2000, non-oil & gas firms with sustainability targets started to appear in Fortune 250.
In 2020, non-oil& gas firms with sustainability targets dominated oil & gas firms with sustainability targets for the first time.

13 of 42

Data on CEO Sustainable Compensation

Hand collected data on Fortune 250 firms, 1994-2021, data on the annual incentive plans (AIP) sustainability goals.

Next, hand collected data on AIP sustainability goals in 2,564 firm-years for 228 oil & gas firms.

Describe the structure of the sustainability compensation mechanism:

Sustainability Targets.
Sustainability modifiers of total comp (i.e., similar to bonus targets).
Quantifiable importance of sustainability targets in overall compensation.

14 of 42

Example of AIP Elements of Sustainable Compensation (Andeavor Inc. in 2012)

Goal	Weighting	Performance Levels			Actual�Performance	% Achieved
Goal	Weighting	Threshold	Target	Maximum	Actual�Performance	% Achieved
EBITDA (measured on a margin neutral basis)	50%	$1.219 billion	$1.406 billion	$1.531 billion	$1.521 billion	192%
Personal Safety (measured by improvement in # of incidents)	5%	20% improvement over 3-year average (≤ 0.63)	5% improvement over 2011 (≤ 0.50)	15% improvement over 2011 (≤0.45)	0.50	100%
Process Safety Management (measured by improvement in # of incidents)	5%	30% improvement over 3-year average (≤ .13)	5% improvement over 2011 (≤ 0.076)	15% improvement over 2011 (≤ 0.068)	0.064	200%
Environmental (measured by improvement in # of incidents)	5%	20% improvement over 3-year average (≤ 25)	5% improvement over 2011 (≤ 23)	15% improvement over 2011 (≤ 20)	21	167%
Cost Management*	35%	$2.261 billion	$2.153 billion	$2.045 billion	$2.118 billion	132%
Overall Performance Achieved						166%

15 of 42

Data on Pollution

Carbon emissions: S&P TrueCost.

Environmental related penalties: Corporate Research Project’s Violation Tracker.

Toxic Chemical releases: Toxic Release Inventory (TRI) from EPA.

Provide knowledge of toxic chemical releases reported by individual facilities.
We create the facility-chemical-year level toxic chemical emissions data following King and Lenox (2000).

Manually matched to hand collected AIP data for oil & gas industry.

16 of 42

Defining Green Patenting

Air pollution abatement:

IPC class: B01D53/34-72, F23G7/06, F23J15, F27B1/18, C21B7/22, C21C5/38, F23B80, F23C9, F23C10, B01D53/92, B01D53/94, B01D53/96, B01J23/38-46, F01M13/02-04, F02B47/08-10, F02D21/06-10, F02M25/07, G01M15/10, F02B47/06, F02D41, F02D43, F02D45, F02M3/02-055, F02M23, F02M25, F02M27, F02M31/02-18, F02M39-71, F02P5, B01D46, B01D47, B01D49, B01D50, B01D51, B03C3, F01N3, F01N5, F01N7, F01N13, F01N9, F01N11, C10L10/02, C10L10/06

Water pollution abatement:

IPC class: B63J4, C02F, C09K3/32, E03C1/12, E03F, C05F7, E02B15/04-10, B63B35/32, C09K3/32

Waste management:

IPC class: E01H15, B65F, A23K1/06-10, A43B1/12, A43B21/14, B03B9/06, B22F8, B29B7/66, B29B17, B30B9/32, B62D67, B65H73, B65D65/46, C03B1/02, C03C6/02, C03C6/08, C04B7/24-30, C04B11/26, C04B18/04-10, C04B33/132, C08J11, C09K11/01, C10M175, C22B7, C22B19/28-30, C22B25/06, D01G11, D21B1/08-10, D21B1/32, D21C5/02, D21H17/01, H01B15/00, H01J9/52, H01M6/52, H01M10/54, C05F1, C05F5, C05F7, C05F9, C05F17, C10L5/46-48, F23G5, F23G7, B09B, C10G1/10, A61L11

Soil remediation:

IPC class: B09C

Environmental monitoring:

IPC class: F01N11, G08B21/12-14

Water-related adaptation technologies:

IPC or CPC class: F16K21/06-12, F16K21/12-20, F16L55/07, E03C1/084, E03D3/12, E03D1/14, A47K11/12, A47K11/02, E03D13/007, E03D5/016, E03B1/041, Y02B40/46, Y02B40/56, A01G25/02, A01G25/06, A01G25/16, C12N15/8273, F01K23/08-10, F01D11, F17D6/02, E03, F16L55/16, G01M3/08, G01M3/14, G01M3/18, G01M3/22, G01M3/28, E03B5, E03B3/06-26, E03B9, E03B3/04, E03B3/28-38, E03B3/02, E03B3/03, E03B3/00, E03B3/40, E03B11

17 of 42

Instrumental Variables

The inclusion in the Board Accountability Project (BAP).

BAP was intended to enable pensioners to speak out in oversight related to climate change, board diversity, and excess CEO pay.
Since one of its most important mandates was climate change risk, this campaign targeted firms based on their underlying carbon footprint.

The state adoption of director duties statutes.

Following Karpoff and Wittry (2018).
Board members are anticipated to take into consideration the interests of all corporate stakeholders including non-investor stakeholders.

BAP was initiated by the New York City Pension System (“NYCPS”) and intended to enable pensioners to speak out in oversight related to climate change, board diversity, and excess CEO pay.
Since one of its most important mandates was climate change risk, this campaign targeted firms based on their underlying carbon reserves.
Due to its target of carbon-intensive companies unable to fully tackle their climate change concerns, BAP is an almost ideal instrument for our study considering that it can absolutely make a difference to companies with heavy pollution while these companies did not anticipate the campaign before.

Furthermore, enlightened by Karpoff and Wittry (2018), we use constituency statutes as our first instrument variable. Constituency statutes, also known as directors’ duties (“DD”) law, force board members to take into consideration the interests of all corporate stakeholders including non-investor stakeholders. This law is evidently not influenced by any specific company but indeed cause some effect on companies in states that have it.

18 of 42

Table 3. Descriptive Statistics for Main Variables

Variable	Mean	St. Dev.	Median
Sustainability Target	0.35	0.48	0
Sustainability Target with Weight	0.20	0.4	0
Sustainability Modifier	0.03	0.18	0

Carbon Emissions, t+1 (S&P’s TrueCost)	1.7	2.75	0
Environmental violations penalty, t+1 (Violations Tracker)	2.94	5.38	0
Toxicity, t+1 (TRI data)	0.98	2.86	0
Green Patent Applications, t+1 (U.S. PTO)	0.3	0.3	0

Carbon Emissions, t-1	1.67	2.7	0
Environmental violations penalty, t-1	2.83	5.31	0
Toxicity, t-1 (TRI data)	1.06	2.96	0

Boardroom Accountability Project Company Indicator	0.04	0.2	0
Director Duties State Statutes	0.56	0.5	1
Log (Total Assets)	7.97	1.64	7.85
Leverage	0.33	0.19	0.33
Log(Cash/Assets)	3.73	2.31	3.64

35% of the firm-year observations have a sustainability target, and 20% has a sustainability target with weight.

Only a handful of firms choose to adopt sustainability modifier, only 3% in our sample.

19 of 42

Table 4. Adoption of Sust. Targets in AIP

The inclusion of the BAP significantly increases the probability of observing a sustainability target with weight.

The past carbon emissions seem to have no impact on the propensity to set up a sustainability target, no matter with or without weight.

Moreover, the past carbon emission leads to a decrease in the propensity that a firm introduces a sustainability modifier in their AIP as an incentive to improve their environmental performance.

	Sust. Target	Sust. Target	Sust. Target	Sust. Target w/ Weight	Sust. Target w/ Weight	Sust. Target w/ Weight	Sust. Modifier	Sust. Modifier	Sust. Modifier
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
VARIABLES	t	t+1	t+2	t	t+1	t+2	t	t+1	t+2

DD Statutes Adoption	0.053	0.029	0.027	-0.028	-0.046	-0.051	-0.025	-0.008	0.008
	(0.720)	(0.364)	(0.341)	(-0.517)	(-0.775)	(-0.762)	(-0.523)	(-0.194)	(0.231)
BAP Focus Firm	0.079	0.048	0.080	0.208^**	0.191^*	0.219^**	-0.055	-0.047	-0.050^*
	(0.832)	(0.486)	(0.834)	(2.060)	(1.853)	(2.658)	(-1.486)	(-1.333)	(-1.800)
Lag CO₂ Emissions	0.004	0.001	-0.003	0.008	0.006	0.011	-0.004^*	-0.006^*	-0.004
	(0.321)	(0.049)	(-0.234)	(0.700)	(0.600)	(1.106)	(-1.879)	(-1.771)	(-0.957)
Log(Assets)	0.045	0.046	0.063^*	0.012	0.014	0.028	-0.001	-0.003	-0.005
	(1.437)	(1.296)	(1.739)	(0.463)	(0.498)	(1.006)	(-0.059)	(-0.197)	(-0.335)
Book Leverage	-0.143	-0.229^**	-0.227^***	-0.024	-0.145	-0.182^*	0.025	0.031	0.049
	(-1.300)	(-2.135)	(-2.870)	(-0.251)	(-1.358)	(-1.992)	(0.421)	(0.568)	(1.192)
Log(Cash/Assets)	0.014^*	0.020^**	0.021^**	0.008	0.012	0.019^**	-0.003	-0.006	-0.008
	(1.769)	(2.709)	(2.469)	(0.938)	(1.407)	(2.681)	(-0.369)	(-1.027)	(-1.636)

Observations	1,548	1,451	1,332	1,548	1,451	1,332	1,547	1,450	1,332
R-squared	0.682	0.696	0.702	0.632	0.638	0.642	0.492	0.520	0.558

20 of 42

Table 5A. CO₂ Emissions and Sustainability Targets in Annual Incentive Plans

The overall effect of the sustainability related incentives.

Among all three sustainability related incentives, only sustainability target with weight has an impact on the carbon emission.

On average, the inclusion of sustainability target with weight in the AIP will lead to a 52.5% decrease on carbon emissions in the following year, an 56.9% decrease one year after that, and a 60.3% decrease in the third year.

Since the three sustainability related incentive measures are correlated, to address the concern of multi-collinearity, we also test the three sustainability related measures separately. The results are consistent.

	(1)	(2)	(3)
VARIABLES	t+1	t+2	t+3

Sust. Target	0.23	0.243	0.167
	(1.32)	(1.09)	(0.65)
Sust. Target with Weight	-0.745^***	-0.842^***	-0.923^***
	(-3.621)	(-3.599)	(-3.429)
Sust. Modifier	-0.002	0.255	0.413
	(-0.003)	(0.34)	(0.52)
CO₂ Emissions at (t-1)	0.423^***	0.267^***	0.147^*
	(4.56)	(2.85)	(1.73)
Log(Assets)	-0.042	-0.02	0.063
	(-0.345)	(-0.143)	(0.41)
Book Leverage	-0.731	-0.942	-0.866
	(-1.332)	(-1.377)	(-1.254)
Log(Cash/Assets)	0.052	0.021	0.041
	(1.61)	(0.49)	(0.86)

Observations	1,543	1,543	1,543
R-squared	0.797	0.751	0.738

21 of 42

Table 5B. CO₂ Emissions and Sustainability Targets in Annual Incentive Plans – Cross-Sectional Differences

	(1)	(2)	(3)
VARIABLES	t+1	t+2	t+3

Sust. Target	0.168	0.382	0.453
	(0.69)	(1.39)	(1.33)
CO₂ Emissions at (t-1)	0.467^***	0.000	0.000
	(5.51)	(3.50)	0.00
Sust. Target * CO₂ Emissions at (t-1)	0.006	-0.062	-0.115
	(0.12)	(-1.062)	(-1.638)
Sust. Target with Weight	0.267	-0.083	-0.212
	(0.80)	(-0.281)	(-0.500)
Sust. Target with Weight * CO₂ Emissions at (t-1)	-0.211^***	-0.142^**	-0.119
	(-3.022)	(-2.081)	(-1.367)
Sust. Modifier	0.556	0.655	0.767
	(0.82)	(0.86)	(0.85)
Sust. Modifier* CO₂ Emissions at (t-1)	-0.197^***	-0.149^***	-0.139^**
	(-5.885)	(-3.229)	(-2.208)

Controls	Yes	Yes	Yes
Observations	1,543	1,543	1,543
R-squared	0.8	0.754	0.742

The inclusion of sustainable target with weights and sustainable modifier in the AIP would improve the environmental performance of a firm if and only if the firm is a heavy polluter in the past.

Within firms including sustainable target with weight in their AIP, the carbon emission reduces by 0.211% when the past carbon emission increases by 1% in the following year.

Within firms including sustainable modifier in their AIP, the carbon emission reduces by 0.197% when the past carbon emission increases by 1% in the following year.

However, we do not observe a similar effect for the sustainability targets without weight.

The inclusion of sustainable target with weights and sustainable modifier in the AIP would improve the environmental performance of a firm if and only if the firm is a heavy polluter in the past.

Within firms including sustainable target with weight in their AIP, the carbon emission reduces by 0.211% when the past carbon emission increases by 1% in the following year.

Within firms including sustainable modifier in their AIP, the carbon emission reduces by 0.197% when the past carbon emission increases by 1% in the following year.

However, we do not observe a similar effect for the sustainable target.

We attribute this non-significant result to the multi-collinearity issues of our three sustainability related incentive measures. Therefore, we test the separated estimates for all three measures. The results indicate that all three sustainability related incentive measures are effective if and only if the firm is a historically heavy polluter.

22 of 42

Table 6A. CO₂ Emissions and Sustainability Targets in Annual Incentive Plans – Instrumental Variables Analysis

	Sust. Target			Sust. Weight			Sust. Modifier
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
VARIABLES	t+1	t+2	t+3	t+1	t+2	t+3	t+1	t+2	t+3
Sust. Target	-0.853	-1.547	-2.469^**
	(-1.157)	(-1.480)	(-2.287)
Sust. Target with Weight				-1.59	-2.526^**	-2.891^**
				(-1.693)	(-2.070)	(-2.158)
Sust. Modifier							0.548	0.713	2.475
							(0.54)	(0.47)	(1.45)
CO₂ Emissions at (t-1)	0.417^***	0.265^***	0.149*	0.426^***	0.278^***	0.161^*	0.423^***	0.267^***	0.152^*
	(4.55)	(2.93)	(1.81)	(4.77)	(3.08)	(1.92)	(4.52)	(2.80)	(1.76)

Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	1,546	1,546	1,546	1,548	1,548	1,548	1,547	1,547	1,547
R-squared	0.136	0.005	-0.11	0.136	-0.01	-0.062	0.152	0.058	0.001
Cragg-Donald	16.6	16.6	16.6	30.4	30.4	30.4	36	36	36
Stock-Yogo 10%	9.1	9.1	9.1	9.1	9.1	9.1	9.1	9.1	9.1
Hansen J-Stat p-value	0.18	0.18	0.49	0.28	0.38	0.82	0.18	0.18	0.49

Among the three sustainability related incentives, only sustainability targets with weights shows significantly negative and consistently increasing effect on future carbon dioxide emissions at 5% level.

Sustainability targets without specified weights only reveal negative effect at the third year.

Companies with sustainability modifiers will expect high CO2 emissions.

One possible explanation is that companies which employ sustainability modifiers do not concern about their environmental performance.

23 of 42

Table 6B. CO₂ Emissions and Sustainability Targets in Annual Incentive Plans – Cross-Sectional Instrumental Variables Analysis

	Sust. Target			Sust. Weight			Sust. Modifier
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
VARIABLES	t+1	t+2	t+3	t+1	t+2	t+3	t+1	t+2	t+3
Sust. Target	3.793^**	4.416^**	3.376^*
	(2.32)	(2.33)	(1.91)
Sust. Target * CO₂ Emissions	-0.901^***	-1.042^***	-0.963^***
	(-3.250)	(-4.191)	(-4.254)
Sust. Target with Weight				6.603^**	5.052^**	3.500^*
				(2.47)	(2.16)	(1.72)
Sust. Target with Weight * CO₂ Emissions				-1.526^***	-1.319^***	-1.078^***
				(-3.816)	(-4.141)	(-4.537)
Sust. Modifier							2.398	2.627	2.969
							(0.79)	(0.78)	(0.88)
Sust. Modifier * CO₂ Emissions							-0.596	-0.641	-0.252
							(-0.817)	(-0.862)	(-0.394)
CO₂ Emissions at (t-1)	0.764^***	0.663^***	0.516^***	0.667^***	0.482^***	0.327^***	0.461^***	0.307^***	0.167*
	(6.32)	(5.23)	(4.40)	(6.59)	(4.63)	(3.38)	(5.25)	(3.17)	(1.86)

Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	1,546	1,546	1,546	1,546	1,546	1,546	1,545	1,545	1,545
R-squared	-0.186	-0.323	-0.225	-0.265	-0.22	-0.142	0.137	0.035	0.004
Cragg-Donald	9.1	9.1	9.1	15.8	15.8	15.8	13.3	13.3	13.3
Stock-Yogo 10%	9.5	9.5	9.5	9.5	9.5	9.5	9.5	9.5	9.5
Hansen J-Stat p-value	0.53	0.78	0.57	0.33	0.16	0.11	0.16	0.17	0.23

Consistent with the separate estimates of Table 5B, sustainability target, with or without weight, will reduce the carbon emissions during the following three years if and only if the adopting firms are historical heavy polluters.

However, the result of the sustainability modifier disappear after instrumented.

Our instrument are mostly strong, with the sustainability target slightly below the 10% maximum bias critical value but below the 20% maximum bias critical value.

Hansen J test suggests there is no over identification issue.

24 of 42

Table 7A. Environmental Violation Penalties and Sustainability Targets in Annual Incentive Plans

	(1)	(2)	(3)
VARIABLES	t+1	t+2	t+3

Sust. Target	0.262	0.385	0.402
	(0.48)	(0.65)	(0.84)
Sust. Target with Weight	-1.126^**	-0.867	-1.374^***
	(-2.176)	(-1.425)	(-3.505)
Sust. Modifier	1.612^**	1.543^*	1.78
	(2.14)	(1.96)	(1.34)
Environmental Penalty at (t-1)	0.163^***	0.104^*	0.026
	(3.16)	(1.98)	(0.53)

Controls	Yes	Yes	Yes
Observations	1,544	1,544	1,544
R-squared	0.624	0.603	0.581

Among all three environmental related incentives, only sustainability target with weights reduces the environmental penalties in the following three years.

A counter intuitive finding is that the environmental related penalties increases when a firm decide to use sustainability modifier as the managerial incentive.

To address potential multi-collinearity issues, we report the separated estimates of the three sustainability related incentive measures in Panel II, and the results are consistent.

25 of 42

Table 7B. Environmental Violation Penalties and Sustainability Targets in Annual Incentive Plans – Cross-Sectional Differences

Only the inclusion of sustainability target with weights in the AIP might reduce the environmental related penalties of a firm, and that happens if and only if the firm is a heavy polluter in the past.

However, we do not observe any similar results for sustainability target and sustainable modifiers.

We also estimate the three environment related incentive measures separately, and the results are consistent.

	(1)	(2)	(3)
VARIABLES	t+1	t+2	t+3

Sust. Target	-0.222	0.183	0.577
	(-0.356)	(0.29)	(1.45)
Env. Penalty at (t-1)	0.185^***	0.151^**	0.104
	(3.65)	(2.40)	(1.56)
Sust. Target * Env. Penalty at (t-1)	0.086	0.018	-0.066
	(1.31)	(0.19)	(-0.709)
Sust. Target with Weight	0.008	0.264	-0.414
	(0.01)	(0.41)	(-0.890)
Sust. Target with Weight * Env. Penalty at (t-1)	-0.191^**	-0.181^*	-0.14
	(-2.088)	(-1.847)	(-1.659)
Sust. Modifier	1.976	1.458	1.909
	(1.71)	(1.65)	(1.46)
Sust. Modifier* Env. Penalty at (t-1)	-0.104	0.038	-0.019
	(-0.514)	(0.29)	(-0.166)

Controls	Yes	Yes	Yes
Observations	1,544	1,544	1,544
R-squared	0.627	0.607	0.586

26 of 42

Table 8A. Environmental Violation Penalties and Sustainable Targets in Annual Incentive Plans – Instrumental Variables

Neither sustainability targets without weight nor sustainability targets with weight have any significant effect on future environmental violation penalties after being instrumented.

Sustainability modifiers have a negative effect on environmental violation penalties only in the third future year.

The instruments should be strong without the dual fixed effects of firm and year.

	ESG-Target			SUST. Weight			SUST. Modifier
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
VARIABLES	t+1	t+2	t+3	t+1	t+2	t+3	t+1	t+2	t+3
Sust. Target	3.977	3.037	3.433
	(1.62)	(1.60)	(1.21)
Sust. Target with Weight				-0.577	-1.750	-1.457
				(-0.362)	(-0.988)	(-0.577)
Sust. Modifier							-0.853	-1.547	-2.469^**
							(-1.157)	(-1.480)	(-2.287)
Env. Penalty at (t-1)	0.159^***	0.102^*	0.023	0.163^***	0.103^*	0.025	0.417^***	0.265^***	0.149*
	(3.07)	(1.97)	(0.44)	(3.13)	(1.96)	(0.51)	(4.55)	(2.93)	(1.81)

Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	1,546	1,546	1,546	1,546	1,546	1,546	1,546	1,546	1,546
R-squared	-0.072	-0.036	-0.067	0.041	0.019	0.02	0.136	0.005	-0.11
Cragg-Donald	16.6	16.6	16.6	30.5	30.5	30.5	36.7	36.7	36.7
Stock-Yogo 10%	9.1	9.1	9.1	9.1	9.1	9.1	9.1	9.1	9.1
Hansen J-Stat p-value	0.18	0.15	0.24	0.22	0.15	0.25	0.49	0.58	0.7

27 of 42

Table 8B. Environmental Violation Penalties and Sustainability Targets in Annual Incentive Plans – Cross-Sectional IV Analysis

Companies currently suffering from environmental penalties and having sustainability targets, whether with weights or not, are less likely to receive penalties due to their environmental performance in three years.

Sustainability modifiers reduce the environmental related penalties in the following year, and then the results decay.

Again, the instruments should be strong without the dual fixed effects of firm and year.

	Sust. Target			Sust. Weight			Sust. Modifier
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
VARIABLES	t+1	t+2	t+3	t+1	t+2	t+3	t+1	t+2	t+3
Sust. Target	2.292	4.494^**	5.605^**
	(1.132)	(2.669)	(2.365)
Sust. Target * Env. Penalty	-0.06	-0.238	-0.521^***
	(-0.377)	(-1.366)	(-2.864)
Sust. Target with Weight				-0.568	1.369	3.012
				(-0.381)	(0.733)	(1.243)
Sust. Target with Weight * Env. Penalty				-0.152	-0.307^*	-0.523^***
				(-1.008)	(-1.970)	(-3.159)
Sust. Modifier							14.215^**	10.646^*	6.261
							(2.665)	(1.777)	(1.203)
Sust. Modifier * Env. Penalty							-1.244^*	-0.066	0.878
							(-1.775)	(-0.086)	(0.832)
Env. Penalty at (t-1)	0.191^**	0.220^**	0.282^**	0.209^***	0.200^**	0.190^**	0.225^***	0.114	-0.01
	(2.300)	(2.085)	(2.549)	(3.071)	(2.509)	(2.449)	(3.721)	(1.685)	(-0.133)

Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	1,546	1,546	1,546	1,546	1,546	1,546	1,545	1,545	1,545
R-squared	0.004	-0.053	-0.097	0.046	0.026	-0.01	-0.115	-0.066	-0.078
Cragg-Donald	10.6	10.6	10.6	18.8	18.8	18.8	9.6	9.6	9.6
Stock-Yogo 10%	9.5	9.5	9.5	9.5	9.5	9.5	9.5	9.5	9.5
Hansen J-Stat p-value	0.36	0.42	0.41	0.27	0.48	0.36	0.9	0.6	0.46

28 of 42

Table 9A. TRI Toxicity and Sustainable Targets in Annual Incentive Plans

Only sustainability modifier reduces the toxic chemical releases in the following year. In addition, the effect last for only one year.

We also report the results where we estimate the impact of our three sustainability related incentives separately, and the results are consistent.

	(1)	(2)	(3)
VARIABLES	t+1	t+2	t+3

Sust. Target	0.020	0.031	0.010
	(0.730)	(0.988)	(0.309)
Sust. Target with Weight	0.038	-0.013	0.036
	(1.468)	(-0.358)	(1.053)
Sust. Modifier	-0.202^**	-0.143	-0.217
	(-2.205)	(-1.048)	(-1.263)
TRI Toxicity at (t-1)	0.410^***	0.265^***	0.198^***
	(18.840)	(10.989)	(8.473)

Controls	Yes	Yes	Yes
Observations	23,667	19,069	18,361
R-squared	0.94	0.936	0.934

29 of 42

Table 9B. TRI Toxicity and Sustainable Targets in Annual Incentive Plans – Cross-Sectional Differences

All three sustainability related incentives reduce the release of toxic chemicals in the next three years if and only if the firm adopting these incentives has heavy toxic chemical releases in the past.

However, only the sustainability targets at the second and third year, and the sustainable modifiers at the first year is statistically significant.

We again argue that this is because of multi-collinearity and report the results where we estimate the three sustainability related incentives separately. The consistent and statistically significant estimates suggest our argument is worth considering.

	(1)	(2)	(3)
VARIABLES	t+1	t+2	t+3

Sust. Target	0.053	0.085^**	0.074^*
	(1.547)	(1.989)	(1.761)
TRI Toxicity at (t-1)	0.421^***	0.278^***	0.212^***
	(17.969)	(10.710)	(8.265)
Sust. Target * TRI Toxicity at (t-1)	-0.014	-0.023^*	-0.027^**
	(-1.174)	(-1.716)	(-2.039)
Sust. Target with Weight	0.081^**	0.037	0.064^*
	(2.517)	(0.941)	(1.732)
Sust. Target with Weight * TRI Toxicity at (t-1)	-0.023	-0.026	-0.016
	(-1.611)	(-1.445)	(-1.057)
Sust. Modifier	-0.132	-0.047	-0.035
	(-1.478)	(-0.515)	(-0.359)
Sust. Modifier* TRI Toxicity at (t-1)	-0.143^*	-0.201	-0.351
	(-1.780)	(-1.455)	(-1.458)

Controls	Yes	Yes	Yes
Observations	23,667	19,069	18,361
R-squared	0.94	0.937	0.934

All three sustainability related incentives reduce the release of toxic chemicals in the next three years if and only if the firm adopting these incentives has heavy toxic chemical releases in the past.

For instance, compared to the firm without sustainable modifiers, firms with sustainable modifiers in their AIP reduces their toxic chemical releases by 0.143% when the historical toxic chemical releases increase by 1%.

However, only the sustainable targets at the second and third year, and the sustainable modifiers at the first year is statistically significant.
We again argue that this is because of multi-collinearity and report the results where we estimate the three sustainability related incentives separately. The consistent and statistically significant estimates suggest our argument is worth considering.

30 of 42

Table 10A. TRI Toxicity and Sustainability Targets in Annual Incentive Plans – Instrumental Variables Analysis

The unconditional effect of sustainability modifier we find in Table 9A are not stable. None of the three sustainability related incentives have any statistically significant unconditional effect after instrumented.

But this is consistent with our argument that there is a disconnect between the sustainability related incentives and the unconditional environmental outcomes of firms.

Our instruments turn out to be pretty strong. Meanwhile, we also recognize the potential over identification issue, especially for ROE, as suggested by the Hanse J test.

	Sust. Target			Sust. Weight			Sust. Modifier
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
VARIABLES	t+1	t+2	t+3	t+1	t+2	t+3	t+1	t+2	t+3
Sust. Target	-0.019	-0.048	-0.012
	(-0.558)	(-1.128)	(-0.292)
Sust. Target with Weight				-0.02	-0.084	-0.036
				(-0.495)	(-1.501)	(-0.661)
Sust. Modifier							0.049	0.558	-0.195
							(0.071)	(0.746)	(-0.265)
TRI Toxicity at (t-1)	0.408^***	0.261^***	0.197^***	0.410^***	0.266^***	0.198^***	0.408^***	0.261^***	0.197^***
	(19.653)	(11.232)	(8.699)	(18.833)	(10.992)	(8.476)	(19.646)	(11.234)	(8.693)

Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	25,029	20,325	19,556	23,667	19,069	18,361	25,029	20,325	19,556
R-squared	0.183	0.075	0.044	0.186	0.079	0.045	0.183	0.074	0.044
Cragg-Donald	7455.6	10,000	6234.6	13,000	12,000	8222.4	549.4	1125	735.3
Stock-Yogo 10%	9.1	20	9.1	9.1	20	9.1	9.1	20	9.1
Hansen J-Stat p-value	0.52	0.69	0.81	0.59	0.94	0.93	0.47	0.81	0.82

31 of 42

Table 10B. TRI Toxicity and Sustainability Targets in Annual Incentive Plans – Cross-Sectional IV Analysis

Firms with sustainability target or sustainable target with weigh in their AIP reduce their toxic chemical releases if they are historically heavy toxic chemical emitters.

Meanwhile, we do not find similar trends or any other impact for the sustainable modifier.

All these results are consistent with our findings with the OLS estimation.

The instrument here does not seem to be very strong, but that likely due to the dual firm and year fixed effect. Removing the firm fixed effect would provide us well identified results.

	Sust Target			Sust. Weight			Sust. Modifier
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
VARIABLES	t+1	t+2	t+3	t+1	t+2	t+3	t+1	t+2	t+3
Sust. Target	0.125^*	0.181^*	0.279^***
	(1.729)	(1.949)	(2.840)
Sust. Target * TRI Toxicity	-0.063^**	-0.101^***	-0.127^***
	(-2.312)	(-2.831)	(-3.248)
Sust. Target with Weight				0.296^**	1.142^***	0.584^***
				(2.248)	(3.207)	(2.589)
Sust. Target with Weight * TRI Toxicity				-0.164^**	-0.615^***	-0.316^***
				(-2.507)	(-3.403)	(-2.834)
Sust. Modifier							-2.293	-1.246	-1.366
							(-1.347)	(-0.704)	(-0.793)
Sust. Modifier * TRI Toxicity							3.408	2.74	1.123
							(1.321)	(1.045)	(0.690)
TRI Toxicity at (t-1)	0.439^***	0.303^***	0.251^***	0.439^***	0.319^***	0.231^***	0.406^***	0.260^***	0.195^***
	(16.745)	(10.104)	(8.063)	(17.021)	(10.525)	(8.515)	(19.406)	(11.175)	(8.607)

Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	25,029	20,325	19,556	23,667	19,069	18,361	25,029	20,325	19,556
R-squared	0.181	0.069	0.032	0.174	-0.1	-0.004	0.134	0.034	0.024
Cragg-Donald	1265.7	1316.4	861	287.5	126.1	154.5	20.1	34	18
Stock-Yogo 10%	9.5	7.6	9.5	9.5	7.6	9.5	9.5	7.6	9.5
Hansen J-Stat p-value	0.43	0.09	0.19	0.25	0.9	0.01	0.14	0.02	0.01

Firms with sustainable target or sustainable target with weigh in their AIP reduce their toxic chemical releases if they are historically heavy toxic chemical emitters.

According to our instrumented results, for firms with sustainable target in their AIP, a 1% increase in the historical toxic chemical releases translate into a 0.063% decrease in the toxic chemical releases in the following year, a 0.101% decrease one year later, and 0.127% decrease in the third year. Similarly, for firms with sustainable target with weight in their AIP, a 1% increase in the historical toxic chemical releases translate into a 0.164% decrease in the toxic chemical releases in the following year, a 0.615% decrease in the following year, and a 0.316% decrease in the third year.

Meanwhile, we do not find similar trends or any other impact for the sustainable modifier.
All these results are consistent with our findings with the OLS estimation.
The instrument here does not seem to be very strong, but that likely due to the dual firm and year fixed effect. Removing the firm fixed effect would provide us well identified results.

32 of 42

Table 11A. Green Patenting and Sustainable Targets in Annual Incentive Plans

No effect of any one of these three incentive measures at any time on the green patenting.

This means that even though some companies seem to have sustainability incentives, they are not incentivized to do anything innovative in the environmental aspect.

Evidence of green washing.

Meanwhile, past carbon dioxide emissions also have nothing to do with future green patenting.

	(1)	(2)	(3)
VARIABLES	t+1	t+2	t+3

Sust. Target	0.096	0.091	0.086
	(1.462)	(1.281)	(0.828)
Sust. Target with Weight	-0.076	-0.07	-0.096
	(-0.938)	(-0.790)	(-0.776)
Sust. Modifier	0.037	0.026	0.025
	(1.302)	(0.955)	(1.079)
CO₂ Emissions at (t-1)	0.008	0.006	-0.002
	(1.402)	(1.096)	(-0.648)

Controls	Yes	Yes	Yes
Observations	1,431	1,286	1,139
R-squared	0.518	0.5	0.525

33 of 42

Table 11B. Green Patenting and Sustainability Targets in Annual Incentive Plans – Cross-Sectional Differences

The impact of sustainability related incentives on green parenting has no cross-sectional difference.

Neither the historical polluters nor the historical non-polluters are incentivized to innovate after the adoption of any sustainability related incentives.

	(1)	(2)	(3)
VARIABLES	t+1	t+2	t+3

Sust. Target	0.078^*	0.048	0.09
	(1.893)	(1.405)	(0.773)
CO₂ Emissions at (t-1)	0.006	0.000	0.000
	(1.456)	0.000	0.000
Sust. Target * CO₂ Emissions at (t-1)	0.006	0.015	-0.003
	(0.424)	(0.855)	(-0.434)
Sust. Target with Weight	-0.050^*	-0.025	-0.046
	(-1.989)	(-0.776)	(-0.698)
Sust. Target with Weight * CO₂ Emissions at (t-1)	-0.007	-0.013	-0.009
	(-0.342)	(-0.511)	(-0.752)
Sust. Modifier	0.035	0.024	0.012
	(1.058)	(0.701)	(0.376)
Sust. Modifier* CO₂ Emissions at (t-1)	0.002	0.002	0.004
	(0.453)	(0.625)	(0.878)

Controls	Yes	Yes	Yes
Observations	1,434	1,289	1,143
R-squared	0.518	0.501	0.526

34 of 42

Table 12A. Green Patenting in Annual Incentive Plans – Instrumental Variables Analysis

The results are consistent with the raw OLS estimation.

In general, firms are not incentivized to do anything innovative in the environmental aspect with the sustainability related incentive measures.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
VARIABLES	t+1	t+2	t+3	t+1	t+2	t+3	t+1	t+2	t+3
Sust. Target	0.592^*	0.569	0.379
	(1.625)	(1.353)	(1.016)
Sust. Target with Weight				0.141	0.117	-0.002
				(0.879)	(0.672)	(-0.015)
Sust. Modifier							-0.103	0.074	0.523
							(-0.855)	(0.994)	(1.016)
CO₂ Emissions at (t-1)	0.007	0.002	-0.003	0.007	0.005	-0.003	0.007	0.006	0.000
	(0.822)	(0.275)	(-0.587)	(1.366)	(1.170)	(-0.911)	(1.349)	(1.215)	(0.474)

Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	1,435	1,290	1,145	1,435	1,290	1,145	1,435	1,290	1,144
R-squared	-0.286	-0.237	-0.109	-0.017	-0.01	0.004	-0.001	0.002	-0.054
Cragg-Donald	14.6	12.5	11.5	26	21.5	16.2	36.1	34.6	32.2
Stock-Yogo 10%	9.1	9.1	9.1	9.1	9.1	9.1	9.1	9.1	9.1
Hansen J-Stat p-value	0.46	0.6	0.56	0.44	0.65	.	0.15	.	.

35 of 42

Table 13A. Profitability and Sales Growth and Sustainability Targets in Annual Incentive Plans

The unconditional effect of sustainability related incentives on sales and ROE are generally statistically identical to 0 with a few exceptions.
The existence of sustainability target with weight in AIP lead to a decrease in sales and increase in ROE in the third year after the target is set.
The inclusion of sustainability modifier increases the sales in the following year.

	(1)	(2)	(3)	(4)	(5)	(6)
	Sales Growth	Sales Growth	Sales Growth	ROE	ROE	ROE
VARIABLES	t+1	t+2	t+3	t+1	t+2	t+3

Sust. Target	0.051	0.028	0.038	0.033	0.007	-0.022
	(1.018)	(0.685)	(0.928)	(0.665)	(0.175)	(-0.608)
Sust. Target with Weight	0.006	-0.033	-0.126^**	-0.041	0.053	0.111*
	(0.110)	(-0.603)	(-2.245)	(-0.582)	(0.597)	(1.829)
Sust. Modifier	0.144^**	0.034	0.078	0.026	0.048	-0.087
	(2.445)	(0.851)	(1.705)	(0.514)	(0.958)	(-1.027)
CO₂ Emissions at (t-1)	-0.015	0.002	0.008	0.017	0.000	0.01
	(-1.490)	(0.163)	(0.634)	(1.562)	(-0.029)	(1.011)

Controls	Yes	Yes	Yes	Yes	Yes	Yes
Observations	1,426	1,282	1,137	1,430	1,283	1,139
R-squared	0.516	0.527	0.536	0.282	0.271	0.268

36 of 42

Table 13B. Profitability and Sales Growth and Sustainability Targets in Annual Incentive Plans – Cross-Sectional Differences

The main take-away from Table 13B is that after introducing the sustainability target into AIP, the historical heavy polluters experience a statistically significant reduction in their sales growth and ROE compared to their non-heavy-polluting peers.
Although not statistically significant, we find similar patterns in the other two sustainability related incentives, the sustainability targets with weight, and the sustainability modifier, as well, except that the sustainability target with weight will increases the sales growth in the third year.
Overall, we believe our evidence is consistent with the story that the sustainability related incentives increase the environmental outcomes of a firm at the cost of financial performance.

	(1)	(2)	(3)	(4)	(5)	(6)
	Sales Growth	Sales Growth	Sales Growth	ROE	ROE	ROE
VARIABLES	t+1	t+2	t+3	t+1	t+2	t+3

Sust. Target	0.079	0.082	0.115^*	0.071	0.003	0.033
	(1.353)	(1.683)	(2.050)	(1.224)	(0.055)	(0.583)
CO₂ Emissions at (t-1)	-0.01	0.008	0.012	0.026^**	0.005	0.019^*
	(-0.835)	(0.498)	(0.842)	(2.317)	(0.436)	(1.907)
Sust. Target * CO₂ Emissions at (t-1)	-0.011	-0.019^*	-0.02^***	-0.015^*	-0.001	-0.021^*
	(-0.874)	(-1.841)	(-2.948)	(-1.938)	(-0.164)	(-1.853)
Sust. Target with Weight	0.057	-0.071	-0.33^**	0.043	0.188	0.143
	(0.481)	(-0.730)	(-2.591)	(0.415)	(1.582)	(1.112)
Sust. Target with Weight * CO₂ Emissions at (t-1)	-0.007	0.012	0.047^**	-0.014	-0.028	-0.002
	(-0.302)	(0.754)	(2.309)	(-0.704)	(-1.086)	(-0.075)
Sust. Modifier	0.128^*	0.029	0.141^**	0.096	0.122	-0.041
	(1.783)	(0.442)	(2.580)	(1.422)	(1.644)	(-0.352)
Sust. Modifier* CO₂ Emissions at (t-1)	0.001	-0.001	-0.021	-0.027	-0.028	-0.021
	(0.114)	(-0.077)	(-0.995)	(-1.291)	(-1.046)	(-0.546)

Controls	Yes	Yes	Yes	Yes	Yes	Yes
Observations	1,426	1,282	1,137	1,430	1,283	1,139
R-squared	0.517	0.529	0.54	0.284	0.274	0.271

The main take-away from Table 13B is that after introducing the sustainable target into AIP, the historical heavy polluters experience a statistically significant reduction in their sales growth and ROE compared to their non-heavy-polluting peers.

More specifically, among all firms implementing a sustainable target in their AIP, a 1% increase in the historical carbon emission leads to a 0.011% decrease in the sales growth and 0.015% decrease in the ROE in the following year, 0.019% decrease in the sales growth and 0.001% decrease in the ROE one year after that, and 0.02% decrease in the sales growth and 0.021% decrease in the ROE in the third year.

Although not statistically significant, we find similar patterns in the other two sustainability related incentives, the sustainable targets with weight, and the sustainable modifier, as well, except that the sustainable target with weight will increases the sales growth in the third year.
Overall, we believe our evidence is consistent with the story that the sustainability related incentives increase the environmental outcomes of a firm at the cost of financial performance.

37 of 42

Table 14A. Profitability and Sales Growth and Sustainability Targets in Annual Incentive Plans – Instrumental Variables Analysis

The overall conclusion is consistent with Table 13A, that the unconditional effects of sustainability related incentives on sales growth and ROE are statistically indifferent from 0.

The only exception is the impact of sustainability target with weight on sales growth.

Our instruments turn out to be strong. Meanwhile, we also recognize the potential over identification issue, especially for ROE, as suggested by the Hansen J test.

Dep. Variable: Sales Growth	Sust. Target			Sust. Weight			Sust. Modifier
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
VARIABLES	t+1	t+2	t+3	t+1	t+2	t+3	t+1	t+2	t+3
Sust. Target	0.032	-0.045	-0.077
	(0.169)	(-0.212)	(-0.317)
Sust. Target with Weight				0.278^*	-0.322^*	-0.406
				(1.651)	(-2.022)	(-1.607)
Sust. Modifier							0.246	0.057	0.295
							(0.699)	(0.246)	(0.912)
CO₂ Emissions at (t-1)	-0.016	0.002	0.007	-0.016	0.004	0.010	-0.015	0.003	0.008
	(-1.531)	(0.172)	(0.589)	(-1.448)	(0.329)	(0.788)	(-1.419)	(0.196)	(0.654)

Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	1,427	1,283	1,139	1,427	1,283	1,139	1,427	1,283	1,138
R-squared	0.019	0.031	0.025	-0.011	-0.014	-0.012	0.017	0.032	0.022
Cragg-Donald	14.3	12.3	11.7	26.1	21.6	16.2	37.7	36.6	35.1
Stock-Yogo 10%	9.1	9.1	9.1	9.1	9.1	9.1	9.1	9.1	9.1
Hansen J-Stat p-value	0.64	0.63	0.24	0.63	0.47	0.45	0.44	0.68	0.35

38 of 42

Table 14A. Profitability and Sales Growth and Sustainability Targets in Annual Incentive Plans – Instrumental Variables Analysis

Dep. Variable: ROE	Sust. Target			Sust. Weight			Sust. Modifier
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
VARIABLES	t+1	t+2	t+3	t+1	t+2	t+3	t+1	t+2	t+3
Sust. Target	-0.213	-0.065	0.085
	(-1.114)	(-0.361)	(0.346)
Sust. Target with Weight				-0.203	0.090	0.132
				(-0.807)	(0.435)	(0.824)
Sust. Modifier							0.157	0.028	-0.028
							(0.613)	(0.135)	(-0.156)
CO₂ Emissions at (t-1)	0.016	0.000	0.011	0.017	-0.001	0.010	0.017	0.000	0.011
	(1.489)	(0.013)	(1.080)	(1.583)	(-0.078)	(0.991)	(1.604)	(-0.019)	(1.040)

Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	1,431	1,284	1,141	1,431	1,284	1,141	1,431	1,284	1,140
R-squared	-0.014	0.009	0.016	0	0.013	0.02	0.011	0.013	0.017
Cragg-Donald	14.5	12.4	11.9	25.4	21	16.1	36	34.5	32.8
Stock-Yogo 10%	9.1	9.1	9.1	9.1	9.1	9.1	9.1	9.1	9.1
Hansen J-Stat p-value	0.1	.	.	0.11	.	.	0.17	.	.

The overall conclusion is consistent with Table 13A, that the unconditional effects of sustainability related incentives on sales growth and ROE are statistically indifferent from 0.

The only exception is the impact of sustainability target with weight on sales growth.

Our instruments turn out to be strong. Meanwhile, we also recognize the potential over identification issue, especially for ROE, as suggested by the Hansen J test.

39 of 42

Table 14B. Profitability and Sales Growth and Sustainable Targets in Annual Incentive Plans – Cross-Sectional IV Analysis

Although not all statistically significant, historical heavy polluters experience decrease in the sales growth and ROE after introducing sustainability related incentives compared to their non-heavy-polluting peers.
This is consistent with our findings with the OLS estimation.
Again, the instrument here does not seem to be as strong as in Table 14A, but that likely due to the dual firm and year fixed effect. Removing the firm fixed effect would provide us well identified results.

Dep. Variable: Sales Growth	Sust. Target			Sust. Weight			Sust. Modifier
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
VARIABLES	t+1	t+2	t+3	t+1	t+2	t+3	t+1	t+2	t+3
Sust. Target	-0.356	0.089	0.043
	(-1.240)	(0.415)	(0.152)
Sust. Target * CO₂ Emissions	0.023	-0.083^**	-0.054^*
	(0.532)	(-2.460)	(-1.921)
Sust. Target with Weight				0.372	-0.124	-0.471
				(1.152)	(-0.418)	(-1.143)
Sust. Target with Weight * CO₂ Emissions				-0.061	-0.057	0.003
				(-1.096)	(-1.310)	(0.051)
Sust. Modifier							0.228	0.061	1.029^*
							(0.322)	(0.139)	(1.717)
Sust. Modifier * CO₂ Emissions							0.024	0.007	-0.178^*
							(0.222)	(0.092)	(-2.064)
CO₂ Emissions at (t-1)	-0.024	0.034*	0.028	-0.007	0.011	0.010	-0.016	0.002	0.019
	(-1.248)	(1.847)	(1.476)	(-0.624)	(0.760)	(0.649)	(-1.323)	(0.138)	(1.395)

Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	1,427	1,283	1,139	1,427	1,283	1,139	1,427	1,283	1,138
R-squared	-0.057	-0.033	0.006	0.011	-0.053	-0.028	0.015	0.032	-0.04
Cragg-Donald	7.6	6.2	6.3	14.3	11.7	8.9	13.6	12.6	11.2
Stock-Yogo 10%	9.5	9.5	9.5	9.5	9.5	9.5	9.5	9.5	9.5
Hansen J-Stat p-value	0.5	0.42	0.46	0.16	0.49	0.53	0.22	0.68	.

40 of 42

Table 14B. Profitability and Sales Growth and Sustainable Targets in Annual Incentive Plans – Cross-Sectional IV Analysis

Dep. Variable: ROE	Sust. Target			Sust. Weight			Sust. Modifier
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
VARIABLES	t+1	t+2	t+3	t+1	t+2	t+3	t+1	t+2	t+3
Sust. Target	0.447	0.304	0.41
	(1.412)	(1.118)	(1.315)
Sust. Target * CO₂ Emissions	-0.072**	-0.045	-0.027
	(-2.073)	(-1.527)	(-0.855)
Sust. Target with Weight				0.126	0.455	0.424
				(0.386)	(1.157)	(1.223)
Sust. Target with Weight * CO₂ Emissions				-0.042	-0.080*	-0.037
				(-0.957)	(-1.804)	(-0.802)
Sust. Modifier							-0.012	0.239	-0.338
							(-0.024)	(0.548)	(-0.795)
Sust. Modifier * CO₂ Emissions							0.028	-0.075	0.043
							(0.314)	(-0.952)	(0.536)
CO₂ Emissions at (t-1)	0.044**	0.015	0.021**	0.022*	0.009	0.013**	0.015	0.004	0.007
	(2.509)	(1.021)	(2.334)	(1.905)	(0.780)	(2.295)	(1.221)	(0.339)	(0.820)

Controls	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	1,431	1,284	1,141	1,431	1,284	1,141	1,431	1,284	1,140
R-squared	-0.027	-0.006	-0.022	0.013	0.007	0.011	0.01	0.011	0.012
Cragg-Donald	7.9	6.4	6.4	13.9	11.3	8.7	13	12.1	10.8
Stock-Yogo 10%	9.5	9.5	9.5	9.5	9.5	9.5	9.5	9.5	9.5
Hansen J-Stat p-value	0.28	.	.	0.14	.	.	.	.	.

Although not all statistically significant, historical heavy polluters experience decrease in the sales growth and ROE after introducing sustainability related incentives compared to their non-heavy-polluting peers.
This is consistent with our findings with the OLS estimation.
Again, the instrument here does not seem to be as strong as in Table 14A, but that likely due to the dual firm and year fixed effect. Removing the firm fixed effect would provide us well identified results.

41 of 42

Conclusion

We find that only a small fraction of Fortune 250 firms include environmental and safety goals (i.e., sustainability) in their CEOs’ annual incentive plans.

Among those Fortune 250 firms with such goals, majority are oil & gas firms.

Moreover, among oil & gas firms, the effect of these goals in AIPs is emission-reducing ONLY for past high polluters.

Importantly, not all aspects of sustainability goal compensation packets work. Notably, the sustainability modifiers have an emission-increasing effect in some cases.

42 of 42

Conclusion

Overall, our results are consistent with a conditional beneficial effect of sustainable AIP goals – only for oil & gas firms, and among those, only for the highest polluters.

Our findings therefore suggest a more nuanced view of such goals in comparison with Flammer et al. (2019).

More precisely, the “one size fits all” approach to sustainability-centered CEO compensation appears questionable.