I’d like to thank John Mori and Mckay Jensen for their helpful feedback. Any errors are my own.
Note: I am not an expert on this topic, and I have little background knowledge on corporate governance. Please be skeptical and point out mistakes.
This post is about doing good by changing the behavior of for-profit companies. The main examples of this within EA are corporate animal welfare improvements. For popular EA causes, plenty of other possibilities exist, such as improving the labor standards of companies operating in poor countries or improving the safety standards of AI and biotechnology companies.
One possible way to bring about corporate behavior change is through shareholder activism. Shareholder activism has been covered in a few recent EA articles. Relative to those articles, the contribution of this essay is to bring up a few new theoretical arguments in favor of shareholder activism, present some (weak) evidence on its costs, and present a rough cost-effectiveness analysis. In the end, I tentatively conclude that shareholder activism is a promising way to do good, and that further investigation would be valuable.
The rest of the article proceeds as follows: I first argue that theoretically we should expect shareholder activism to be a cost-effective way to do good, even relative to other ways to cause corporate behavior change such as outside advocacy campaigns. I then review some empirical evidence on shareholder advocacy, and, based on that evidence, conduct a rough cost-effectiveness estimate. Finally, I address some possible objections and conclude with a few recommendations for future research.
Shareholder advocacy is likely more effective than outside advocacy, at least in some cases.
The empirical literature on shareholder advocacy is promising, but messy. There isn’t evidence that it significantly reduces stock prices, and it seems to significantly change firm behavior.
Simple cost effectiveness estimates suggest that shareholder activism on climate change has the potential to be as cost effective as top climate interventions recommended by EA organizations. (Climate change is used as a case study because that’s where the evidence base on shareholder activism is strongest.)
EAs should look more seriously into shareholder activism as a way of doing good.
Brief overview of the shareholder engagement process
First, this is my understanding of what shareholder activism looks like in practice:
A shareholder (call them the “leading shareholder”) who wants a company to change their behavior first approaches company management and asks them to make the change. Note that you do not have to be the largest shareholder (or even close to that) to take on this role. See the “capital requirements” section below.
If management refuses, the leading shareholder can choose whether to put a new topic or list of options on the ballot (often this includes adding their own candidates for board positions. Sometimes this involves adding resolutions which ask the company to adopt new rules or practices).
The leading shareholder and management will campaign for their preferred choices in the election. Often this process, called a proxy fight, can be very costly for both sides (around $2 million for the shareholder and $4 million for the company).
All shareholders vote in the election.
From what I understand, most shareholder engagements (step 1) don’t end up going to the ballot (steps 2 through 4). Because of the large cost of a proxy fight, management usually agrees to a behavior change beforehand, especially when they have a good chance of losing a potential election.
I would also like to define different forms of shareholder activism:
Leading shareholder activism is when you are the leading shareholder in steps 1 through 3. This involves higher costs but potentially larger benefits.
Passive shareholder activism is when you don’t make any proposals yourself, but you exercise your right to vote in step 4. This will be cheaper than being a leading shareholder, but probably have smaller benefits.
This article is mainly concerned with leading shareholder activism. I also did a back-of-the-envelope cost-effectiveness estimate of passive shareholder activism, which is not included in this article. Reach out to me if you’re interested.
Finally, I think there are two types of interventions that EA organizations could consider:
Practicing shareholder activism with existing EA investment funds. For example,
Open PhilanthropyGood Ventures could start engaging in shareholder activism with their current investments.
Starting new funds which engage in shareholder activism (this might be similar to Engine no 1′s etfs). This would allow many small EA investors to pool their resources together, and may be able to attract funding from non-EA investors.
This article is mainly concerned with the first option. I have not looked carefully at the cost effectiveness of the second option, although intuitively it seems promising.
Changing corporate behavior from the inside vs the outside
There are multiple ways to try to get companies to do good. So far, EA organizations have mainly focused on lobbying firms from the outside (such as through cage-free egg campaigns). Is there any reason to believe that shareholder activism would sometimes be more cost-effective than lobbying firms from the outside? A few considerations come to mind.
Note, however, that it also doesn’t have to be either-or. If both approaches have diminishing returns, then the optimal strategy may be to invest in both. I’m just arguing that, at least in some cases, it may be optimal to spend some or even most of our resources on shareholder activism instead of outside lobbying.
People may be less willing to participate in an outside advocacy campaign
Both strategies involve engaging with companies to ask them to change their behavior and then, if they refuse, punishing them in some way. In the case of outside advocacy, this punishment is often in the form of protests or boycotts, which aim to hurt a company by reducing their sales. For shareholder advocacy, the punishment is to initiate a costly proxy battle to change the board members of a company. Both of these punishment strategies require other people to join your campaign (in the outside advocacy case, you will need consumers to choose to not purchase the company’s products. In the shareholder activist case, you will need other shareholders to vote for your proposals). There are multiple reasons to expect that, at least in some cases, people will be more willing to participate in a shareholder activism campaign than in a boycott.
The first is simple: sometimes a companies’ consumers might not care about your objective, and thus they won’t be willing to participate in a boycott. For example, if consumers of Ford trucks don’t care about climate change, then it will be very costly to persuade them to join a boycott. Ford’s shareholders may be relatively more concerned about climate change, so they might be willing to vote to reduce emissions
In some cases, even if consumers care about your objective, it may be very difficult for them to substitute away from a product. For example, utility companies in the US are often local monopolies, and demand for electricity is fairly inelastic, so it would be very difficult for a consumer to boycott their electric company over their bad climate change practices.
Finally, even if it is equally costly to participate in an outside advocacy campaign vs a shareholder activism campaign, there is a reason to expect that people will be more likely to participate in shareholder activism campaigns: shareholder voting allows for a form of donor coordination.
As has been noted elsewhere, moral objectives are public goods. I may not be willing to unilaterally donate $1 to charity, but if I’m in a large group, I may be willing to vote for everyone in the group (including myself) to donate $1 to charity. This is because if the vote passes, my donation will be effectively matched by everyone else in the group.
Shareholder voting is like a large group voting to give to charity. All shareholders are paying a cost in terms of losing returns on their shares in exchange for doing good, but they will only pay that cost as long as everyone else does. Effectively, their “donation” to the cause is matched by all other shareholders in the event that the vote passes (this argument is formalized in a recent economics paper). This suggests that there may be many individuals that are willing to vote for impactful shareholder proposals but not to participate in an outside advocacy campaign against a company.
Companies may respond more to a shareholder activism campaign
Even if outside advocacy and shareholder activism would attract the same number of supporters, in some cases firms may respond more to shareholder pressure than to outside pressure.
A clear case of this is when a firm’s bottom line won’t be affected much by a boycott. This would happen if the demand curve a firm faces is fairly elastic, in which case cutting demand by even a large fraction won’t have a big impact on prices and quantity sold.
A less obvious reason has to do with principal agent problems. In general, corporations face a principal agent problem where managers and board members are under-incentivized to maximize profits. Typically outside advocacy threatens to punish a firm with lower profits if they continue to behave badly. Shareholder activism, instead, threatens to punish a firm’s board with a loss of their jobs. To the extent that the principal agent problem is unsolved, board members will react more to a threat of losing their jobs than to a threat of lost profits. So this pushes in favor of shareholder activism.
Here I will look at some empirical evidence on shareholder activism. I’ll cover both academic papers and some specific cases reported in the news. Note that this is not a comprehensive literature review. There are likely many relevant papers that I’m missing. Also note that I focused on papers on socially responsible shareholder activism. There is a much larger literature on “standard” shareholder activism, where the goal is just to increase profits. I have not spent much time investigating that.
Can shareholder activism change firm behavior
The first question is how much does shareholder activism change firm behavior to be more beneficial to society. The main source I’ve found on this is Naaraayanan et al (2021). They look at the effect of the Boardroom Accountability project, which is an activism campaign by the New York City Pension System starting in 2014. The campaign first pushed companies to adopt proxy access, a corporate governance reform which makes it easier for shareholders to vote out incumbent board members. The campaign then used this increased access as leverage to push for changes in corporate behaviors, especially on climate change and board diversity. Naaraayanan et al (2021) estimate that the campaign caused firms to reduce carbon emissions by 7%, and to reduce local air pollution by 19%.
Giving Green’s report finds more empirical evidence that shareholder activism can change corporate behavior on climate change.
The track record of animal welfare shareholder activism is less promising. Hallaj (2013) finds that shareholder proposals related to animal welfare between 2010 and 2013 averaged support in the low single digits when they went to ballot. Carl Icahn’s recent failed attempt to get McDonald’s to adopt better welfare standards also got only 1.8% support from shareholders. However, Hallaj (2013) also argues that many of animal welfare related shareholder engagements were able to get concessions from companies without having to go to the ballot .
Edit: after talking to Josh Balk of the Humane Society about his experience running activist campaigns, I’ve updated in favor of shareholder activism on animal welfare being effective. In fact, in Josh’s opinion, a substantial amount of the success of corporate animal welfare campaigns has been due to shareholder activism.
Overall, the evidence on climate change seems well developed and suggests that shareholder activism can be very effective. The evidence on animal welfare is much less developed
, but from what I’ve seen is less promising. I think a deeper dive into shareholder activism on animal welfare would be valuable.
The effect of shareholder activism on share prices
As was covered above, a potential cost of shareholder activism is a loss in profits, reflected in lower share prices. There are a few papers on this. Naaraayanan et al (2021) (mentioned in the previous section), found no significant effect on share prices from environmental activism. There may be a theoretical justification for this, since in many standard models firms can deviate from the profit maximizing strategy at near zero marginal cost.
Another paper, Dimson et al (2015), actually estimates significant positive effects on share prices from shareholder activism. This is surprising. The paper mentions one possible explanation: corporate short-termism. This is the idea that due to principal-agent problems, company management will be overly concerned with short-term profits relative to what shareholders would prefer. Forcing management to be more concerned with longer term issues, such as future damage to the firm from having a negative environmental impact, may thus benefit shareholders and raise share prices.
So, the results are mixed, but there’s no evidence of large negative effects on share prices due to shareholder activism. In the cost effectiveness section below, I assume that the effect on share prices is zero.
Other costs of shareholder advocacy
There are a number of other costs associated with shareholder activism. First, the cost of a proxy fight averages $1.8 million, but most shareholder engagements resolve without needing a proxy fight (in Naaraayanan (2021)’s dataset, only 1⁄3 engagements needed to go to a vote).
Next, there are the costs of engaging with companies and becoming informed on these issues. This has been the most difficult for me to get data on. For now, I just have this piece of suggestive evidence: the hedge fund Engine no 1, which has taken a leading role in multiple shareholder engagements on climate change, only has 28 employees. Many of those employees probably aren’t involved in the engagement process, so it may only take around 10 to 15 full time employees to engage in a large number of activist shareholder campaigns.
How many shares do you have to hold to be a leading activist shareholder? Not many. There is no capital requirement to run a proxy battle. “Any shareholder who’s willing to pay the cost can launch a proxy contest”. And there are examples of this being done with a very low number of shares. Engine no. 1 only held .02% of Exxon stock during their successful proxy fight.
However, to use proxy access, a newer strategy which is lower cost, you typically need to hold at least 3% of a company’s shares for at least 3 years.
For reference, a fund of $10B (which I would expect to be around the amount that Open Philanthropy or FTX already have invested) in the S&P 500 would give about a .025% stake in each of those companies. This would be enough to engage in proxy battles but not to use proxy access. However, if you chose to skew your investments toward smaller (and presumably especially influential) companies, then you could potentially acquire much larger percentages of those companies and use more effective engagement strategies.
Note: the estimates given in this section are very uncertain. They could be off by orders of magnitude. I view them as weak evidence in favor of shareholder activism being cost-effective, and a starting point for others to improve on.
In this section I look into the cost effectiveness of shareholder activism to address climate change. I look at climate change specifically for a few reasons:
There are somewhat reliable estimates on the effects of shareholder activism in this area.
There are cost effectiveness estimates of alternative strategies for addressing climate change, which can serve as a benchmark to measure the effectiveness of shareholder activism.
Remember, though, that shareholder activism can be used to address many cause areas. I would guess that the most cost-effective applications of shareholder activism will not be climate-related.
The top climate recommendations from Founders Pledge are the Coalition of Rainforest Nations and the Clean Air Task Force. These are estimated to cost between $.1 and $1 per ton of CO2 removed. This is the benchmark for climate related shareholder activism.
Cost effectiveness of a leading shareholder activism campaign
Below is a summary. See the appendix for details.
Here I estimate the cost effectiveness of the Boardroom Accountability Project, the shareholder activist campaign by the New York City Pension System to address climate change. As was mentioned above, this campaign was estimated to reduce targeted companies’ carbon emissions by about 7%. This corresponds to a reduction in emissions of 114M per year.
In light of the theoretical and empirical evidence presented above, I will assume that there is approximately no drop in share prices due to the activist campaign. So, the remaining costs will be due to the cost of becoming informed, engaging with management, and running a proxy contest.
As far as proxy contests, these were only necessary for a third of the 61 targeted firms, according to Naaraayanan et al (2021). Since the average proxy contest costs $2M, this comes out to an upfront cost of $40M. Now, there’s a question of whether this is a one time payment or whether future proxy contests will be needed to sustain the emissions reductions. To be conservative, I’ll assume that these proxy contests have to be repeated roughly every 10 years, so that comes out to about $4M per year.
The remaining costs are from becoming informed and engaging with these companies. These are the terms that I have the least amount of evidence on. To start, though, I’d like to point out that as long as this cost is less than $112M per year, the cost effectiveness of this campaign will be within the range of Founder’s Pledge’s top climate charities. It seems likely that the costs will be significantly below that. As was mentioned earlier, one activist hedge fund, Engine no 1, has only 28 employees. In light of this, I will guess that the cost of engagement comes out to about $10M per year, which would put the total cost effectiveness at around $0.2/ton CO2, which is well within the range of Founders Pledge’s top climate charities.
One major caveat is that the Boardroom Accountability Project was done by a pension fund with over $100B of assets. Is it realistic to expect that any EA funder could run an activist campaign with a similar level of cost effectiveness? Maybe. Some evidence on this is that Engine no 1′s campaign against Exxon cost about $30M (and Engine no 1 had only $400M in assets at the time, which is certainly attainable by a large EA foundation). If we assume that this campaign had a similar impact on emissions as the estimates for the boardroom accountability project, then this campaign reduced emissions by 7M tons CO2 per year. Assuming that continued engagement with Exxon costs about $1M a year, this gives a cost effectiveness estimate of $0.6/ton CO2, also within the range of the Founders Pledge estimates.
Anticipated Objections and Questions
Is this Neglected?
An application of the ITN framework to shareholder activism is here. I agree with that article’s conclusion that although there’s a lot of money in socially responsible investing, very little of it is focused on effective socially responsible investing.
Is this effective for any cause area other than climate change?
Overall, my biggest concern is that many of the most important directions for shareholder activism may be too neglected. Shareholder engagement is more likely to succeed when shareholder proposals are likely to win more votes. So, if hardly anyone else cares about your cause area, shareholder engagement isn’t likely to succeed. While that’s not a big issue for climate change, it may be for other EA cause areas.
However, I would guess that in other cause areas it is still possible to find valuable changes that would get broad shareholder support. A lot of things related to global health and development (reducing costs for vaccines sold to poor countries, for example) seem like they could get broad support. Changes related to pandemic preparedness may also be broadly popular. Also, there may be more low hanging fruit in the more neglected cause areas (meaning that there are more cheap and easy corporate behavior changes that will take very little activist pressure to bring about).
A previous EA forum post brought up fiduciary duty as a potential barrier to socially responsible investing. The concern is that fiduciary duty requires many investors to maximize financial returns, which means that investing for some other goal wouldn’t be allowed. However, it seems like current US laws do not require private foundations to maximize financial returns. Current law does seem to say that public pension funds must maximize financial returns, but there is apparently a lot of latitude in this because many pension funds are engaging in socially-conscious shareholder activism (see the examples above involving the New York City Pension System).
Overall, it seems like fiduciary duty is not a meaningful barrier to impactful shareholder activism.
Conclusion and suggestions for future work
There are theoretical reasons to expect for shareholder activism to be an effective way to do good. Empirical evidence suggests that at least for one cause, climate change, shareholder activism is cost effective. I think this evidence justifies taking a closer look at shareholder activism.
Here are some directions for research that I think would be especially valuable:
Getting more evidence on the costs of engaging with companies. This is the most uncertain part of my current cost-effectiveness analysis.
Research on whether activism success on climate change can be replicated in other cause areas. Perhaps a good place to start on this is to look more closely at the track record of shareholder activism on animal welfare.
Here are some specific examples, which seem especially valuable, where companies can do good at low marginal cost:
Reducing the price of plant-based foods or any food which is a substitute for meat.
Raising the price of animal products.
Reducing the price of vaccines or mosquito nets.
Reducing the price of any consumer good purchased by people in extreme poverty (as long as the good doesn’t have a negative externality).
Raising wages for workers in poor countries.
Increasing R&D on clean energy or plant-based food.
Increasing R&D on any product which doesn’t have negative externalities.
Increasing AI or biotechnology safety standards.
Reducing efforts to protect IP.
Reducing harmful lobbying expenditures.
Increasing disclosure of information that would be useful to others.
Details of cost effectiveness estimate
I use the following formula for cost effectiveness:
Here cost-effectiveness will be in units of $ per ton of CO2 averted. To estimate the benefits term in the denominator, I start with the 7% estimated reduction in carbon emissions from Naaraayanan (2021). That paper also states that the 61 targeted companies make up about 25% of total US emissions (I don’t know if this number accounted for the targeted firms’ global emissions or just their emissions within the US. I assume in what follows that it is global emissions. If it turns out to only be US emissions, then the actual emissions reduction from the activist campaign could be much larger than what I estimate). Total US emissions were 6.59B tons of CO2 equivalent in 2015, so that would put targeted companies emissions at .25*6.59B = 1.65B tons CO2. Multiplying this by the .07 reduction gives total benefits of 115 million tons of CO2 averted per year.
There are three cost terms, which are calculated in the following way:
1. ReturnLoss is the cost you pay in terms of lower share prices due to the activist campaign. In light of the theoretical and empirical evidence presented above, I assume this is zero.
2. EngageInfoCost is the cost of being informed and engaging company management. I guess that this is about $10 million per year. Some justification for that guess is given in the body of the paper.
3. ProxyBattleCost is the per year cost paid in proxy battles. We know that the initial wave of engagement required about 20 proxy battles according to Naaraayanan (2021). Now there’s a question of whether future proxy battles will be needed to sustain the emissions reductions. I haven’t found any evidence that the NYCPS fought any proxy battles since the initial engagement, but to be conservative, I will assume that the same amount of proxy battles needed in the initial wave of engagements (20) will be needed roughly every 10 years moving forward (which implies that there will be on average 2 proxy battles per year against the 62 targeted firms). Under that assumption, the following formula gives the per year cost:
Where X is the cost of proxy battles during the initial wave of engagements, c is the cost per proxy battle, n is the expected number of proxy battles in each subsequent year, and β is the discount rate. We know that the average cost of a proxy battle c is nearly $2 million. There were 20 initial proxy battles, so X=$40 million. Under the assumption stated in the previous paragraph, n=2. And finally, I’ll assume a discount rate of .9 (the average rate of return on equity). Plugging this in gives
Plugging in all cost and benefit terms into the initial equation, we are left with an overall estimate of $0.16 per ton of CO2 averted.
Finally, in the example of the Engine No 1 campaign against Exxon, the benefits were calculated as follows. I assume that the campaign reduced carbon emissions by 7%, which is the estimate for the Boardroom Accountability Project campaigns, and seems plausible in light of Exxon’s behavior after the proxy battle. Exxon’s recent emissions are around 120M tons CO2 per year, but it is decreasing, so I assume that going forward it would have averaged around 100M in the absence of Engine no 1′s activism. This implies that the campaign caused a reduction in carbon emissions of about 7M tons per year.
A list of more examples is in the appendix.
So far I’m finding that, at least on climate change and with a fund of around $10 billion, it’s likely less cost-effective than leading shareholder activism and than top climate charities. But, due to economies of scale, this could change if you had a much larger fund.
According to this article, animal welfare advocacy against gestation crates combined both strategies.
The reverse of this will also sometimes be true. In some cases, shareholders will care relatively little about your objective while consumers are more supportive.
Conversely, shareholder activism is probably more likely to be effective when companies are monopolies because then it’s less likely that socially irresponsible competitors will crowd-out their efforts to do good.
If their actions lose the company money, most of those costs are borne by others (namely, the shareholders) rather than the manager. Companies try to reduce the extent of this problem through various incentive schemes like tying executive pay to share prices, but from what I understand most economists believe that these schemes don’t fully solve the problem.
They base this off of a few anecdotes. They don’t present much quantitative evidence
It is plausible that management might be willing to make significant concessions even in cases where the firm will very likely win a proxy battle. There are a few reasons for this. First, proxy battles are costly for firms even if they win. Second, if the firm loses the proxy battle, board members lose their jobs, which might make it beneficial for them to avoid even a small chance of losing.
The issue is that diminishing returns might not hold here, which is needed for neglectedness to predict marginal returns