Socially Responsible Divestment

Responsible investing is becoming increasingly mainstream. In 2006, the United Nations established the Principles for Responsible Investment (“UN PRI”), a commitment to invest responsibly, which was signed by 63 investors managing a total of $6.5 trillion. By the end of 2021, this had grown to 4,375 investors, representing $121 trillion.

Blanket exclusion of “brown” industries, such as tobacco, gambling, and fossil fuels, is often seen as the purest and most effective form of responsible investing. Divesting starves them of capital, the argument goes, preventing them from creating further harm. Accordingly, practitioners and the public hold investors accountable for their holdings of brown firms. In 2020, Extinction Rebellion protesters dug up a lawn outside Trinity College, Cambridge to protest at its investment in fossil fuel companies, and many asset owners evaluate asset managers according to whether they manage a “net zero” portfolio. Beyond climate, Morningstar’s “globe” ratings of funds are based on the Sustainalytics ESG ratings of the stocks they hold and are thus boosted by divesting from brown stocks. Those with low ratings are often accused of greenwashing.

However, this argument considers only one channel through which divestment can affect a company’s real actions—the primary markets channel, whereby divestment affects the terms at which a company raises new capital. As the survey of Bond, Edmans, and Goldstein (2012) points out, stock market trading—and thus divestment strategies—can also have real effects through a secondary markets channel. Specifically, trading leads to the stock price reflecting a manager’s real actions, thus rewarding or punishing him for taking them. Even if a firm is in an irremediably brown sector, which unavoidably produces negative externalities, the manager may be able to take corrective actions to mitigate these externalities. Blanket exclusion fails to reward such actions because the firm is divested no matter what. Thus, it may be optimal for a responsible investor to pursue a “tilting” strategy, where she tilts away from a brown industry but is willing to hold firms that are best-in-class, i.e. take corrective actions.

We build a model in which responsible investment affects firm behavior through both above channels. There is a single brown firm that emits negative externalities. The firm’s manager can take a non-contractible corrective action, such as investing in clean energy, that reduces both externalities and also firm value. The firm also raises capital which it uses to fund an expansion, increasing both firm value and externalities. The firm’s manager is concerned with both fundamental value and the stock price; the latter may arise through takeover threat, termination threat, or reputational concerns.

The firm is owned by a continuum of risk-averse, profit-motivated, atomistic investors (“households”) and a risk-neutral responsible investor. Her objective is to minimize the externalities produced by the brown firm. To do so, she announces an investment strategy that depends on whether the brown firm takes a corrective action. Under exclusion, the investor never holds the firm; under tilting, she invests if and only if it takes the action.

We show that the optimal divestment strategy balances two forces. On the one hand, since the brown firm continues to produce negative externalities even under the corrective action, the investor wishes to minimize its size. She does so through blanket exclusion—by holding none of the brown firm’s shares, they have to be held entirely by risk-averse households, who require a risk premium for doing so. This minimizes the stock price, and thus the new funds the firm can raise. On the other hand, the investor wishes to incentivize the manager to take the action. Exclusion provides no such incentives, since the firm is always divested. Tilting rewards the manager for taking the action—by buying shares, the investor reduces the number that must be held by households, thus increasing the stock price. This echoes the “governance through exit” models of Admati and Pfleiderer (2009), Edmans (2009), and Edmans and Manso (2011), where investor trading causes a manager’s actions to be reflected in the stock price.

Intuitively, the responsible investor’s strategy is analogous to an incentive contract. Exclusion corresponds to paying the manager a flat salary, which minimizes the cost to the firm but provides no incentives. Tilting incentivizes the action, but is costly—in a contracting setting, the cost is the monetary value of the incentive; in a responsible investment setting, the cost is financing the expansion of a brown firm. This analogy highlights how exclusionary strategies may be suboptimal, despite being widely advocated—they are tantamount to giving the manager zero incentives.

We show that the optimal investment strategy involves tilting if the corrective action is effective at reducing the externality, because incentivizing the action is particularly important compared to stifling capital raising. Tilting is also optimal if the action is less costly and if the manager’s stock price concerns are high, as then the investor does not need to offer large rewards (in the form of share purchases) to incentivize the action; thus, the additional expansion and externalities created are low. These results suggest that exclusion may be optimal for industries such as controversial weapons, where it is relatively difficult to reduce the harm produced. In contrast, tilting may be preferred for fossil fuels, where managers can take corrective actions such as investing in clean energy.

We extend the model to the case in which the corrective action is unobservable, so the investor is unable to condition her investment on it; instead, she can make it contingent on a noisy signal of the action. The noisier the signal, the greater the reward the investor needs to offer to induce the action, and the more likely she is to choose exclusion. This result highlights a new benefit of ESG disclosure—it allows investors to induce corrective actions without having to promise large amounts of capital, thus financing the expansion of brown firms.

Importantly, even if the investor can gather perfect information about the manager’s action at an arbitrarily small cost, she may not do so. It may seem that such information will allow her to induce the action at lower cost, i.e. promising a lower investment—since the investor will always make the investment if the manager has taken the action, he will do so even if the investment is small. However, the investor may end up buying a company that has taken the action even though the noisy public signal suggests that it has not. Doing so may lead to the investor being accused of greenwashing—buying a brown company even though, in the eyes of the market, it has not taken any corrective action. If the investor suffers a sufficiently large reputational cost from buying such a company, she will not base her purchases on her private information. This reduces her incentives to gather it in the first place, and may deter her from inducing the corrective action.

A common criticism of divestment is that arbitrageurs can buy brown stocks at depressed prices, reversing the impact of divestment. In our final extension, an arbitrageur appears with positive probability, and has a single objective of maximizing trading profits. The arbitrageur buys half the shares that are not purchased by the responsible investor, thus reducing the impact of her trading decisions. On the one hand, this makes tilting less effective—since arbitrageur partially offsets the investor’s trades, he needs to promise an even larger purchase to induce the corrective action, making tilting more costly to implement. On the other hand, the arbitrageur makes exclusion less effective, since she buys up underpriced stock and reduces the impact of exclusion on the cost of capital. Since the arbitrageur buys half of the free float, his impact is greater on exclusion (where the investor’s trade is zero and the free float is the total shares outstanding) than on tilting. Thus, the greater the probability of the arbitrageur, the more likely the investor is to tilt.

The complete paper is available for download here.