We study the viability of socially responsible (SR) funds when investors are small, have social preferences, and can cheaply invest in other ways. In a static setting, SR “impact” funds, which engage with dirty firms at a cost, cannot exist due to free riding, regardless of the type of social preferences. SR exclusion funds with reduced ownership of dirty firms can exist, but only if investors’ social preferences are “warm glow,” i.e., tied to the extent of ownership. In a dynamic setting, we demonstrate that such exclusion funds endogenously make impact funds viable. Impact funds profit by making dirty firms investable for exclusion funds.

We study whether U.S. mutual funds’ sustainability objectives predict behavior and real-economy outcomes. Using machine learning on prospectuses, we classify sustainable strategies as financial, moral, or impact-oriented. Among 1,523 funds managing $1.7 trillion in 2023, 88% of assets are financially motivated, 10% morally motivated, and just 2% impact-oriented. Financial funds hold greener, lower-carbon portfolios; moral funds rely on exclusionary screens and exhibit lower flow–performance sensitivity; impact funds support outcome-oriented proposals and are the only fund type whose purchases are followed by reductions in portfolio firms’ carbon intensity. Most sustainable capital is not allocated to strategies seeking measurable real-economy change.

We study the effect of ESG-induced divestment on access to financing, operating efficiency, and

CO2 emissions for long-lived, capital intensive assets: U.S. power plants. Using a shift-share

measure of firms’ exposure to banks’ coal divestment policies, we find that divestment leads to

a reduction in debt supply and subsequently lower capital expenditure. We then trace out the

full chain of real consequences of this financial shock. Using hourly operating data for all major

U.S. power plants, we show that capital rationing causes operational inefficiencies in generator

utilization, resulting in inefficiently high CO2 emissions. We show that policy choices play a

major role in these unintended consequences. Exploiting the staggered introduction of emissions trading schemes, we show that, unlike ESG divestment, carbon pricing does not lead to

inefficient asset utilization and excess CO2 emissions.