Buying a wind developer in the secondary market does not finance its turbines. Someone else's capital paid for them; the trade merely transfers a share certificate between accounts. Busch, Bernard-Rau, Prüßner and Freiling make this distinction the foundation of their review. They are right to do so.
The distinction comes from Kölbel et al. (2020). Company impact means the change a company's activities produce in social and environmental parameters. Investor impact means the change an investor's activities produce in company impact. The quantities belong to different actors. A portfolio may score highly on company impact while contributing nothing to it.
Busch and co-authors then use a division from Busch et al. (2021). Impact-aligned investments own companies with positive impact, regardless of whether the investor caused it. Impact-generating investments involve an investor actively trying to influence the outcome.
How the review was assembled
The authors conduct a systematic literature review, following Tranfield et al. (2003) with PRISMA screening. Their search string combines "impact invest" with mechanism terms (shareholder engagement, dialogue, stewardship, shareholder activism, voting, cost of capital, field building, non-financial resource) and outcome terms. They searched JSTOR, ScienceDirect and Google Scholar. Data collection ran June 2024 to January 2025, while the publication window spans 2013 to 2024, beginning with Brest and Born.
The funnel starts with 3,502 records. Date and English-language filters leave 2,392, followed by 577 after 1,815 are removed for journal ineligibility. De-duplication reduces the count to 523. The authors screen 154 full texts and find 33 eligible academic papers. Nineteen industry working papers bring the final sample to 52 studies.
All four authors coded the material. They identify three strategies and eight mechanisms. Engagement/stewardship includes dialogue & voting and the provision of non-financial resources. Capital allocation includes provision of liquidity and cost of capital. Each mechanism is divided into direct and indirect variants, with separate ratings for each.
The pathway distinction comes from Marti et al.'s (2024, p. 2189) taxonomy: direct impact on companies, indirect impact through other shareholders, and indirect impact through the institutional context. Every variant receives a high, medium or low investor impact potential rating.
The organising equation is investor impact potential, IIP = L x M. L represents the probability that an investor's action influences company impact. M represents the magnitude of that impact. The formula is intended for ex ante assessment because impact unfolds over time and cannot be measured when the allocation decision is made.
Causal proximity sets the rankings
The ratings form a clear ladder. Direct dialogue & voting by a large owner receives a high rating. Two other mechanisms reach the same level: provision of liquidity when the single investor's capital is essential, and provision of non-financial resources such as expertise, networks and capacity building. Both belong chiefly to primary-market and early-stage settings.
Individual dialogue by a minority holder in a dispersed large-cap receives a low rating. Collaborative dialogue & voting, where small stakes are pooled into a bloc, reaches medium. A single investor changing the cost of capital through secondary markets rates low. Coordination by many investors through price signalling lifts it to medium.
All four field building mechanisms, stigmatization & endorsement, demonstration, lobbying, standards & benchmarks, are indirect and receive low potential ratings. The authors give a direct reason. Each mechanism depends on third-party reactions, leaving a very low likelihood that any one investor's action proves essential to a change in company behaviour. The potential scale may still be high.
Three conditions matter more
Governance engagements succeed significantly more often than environmental or social engagements. The review cites Dimson et al. (2015), supported by Gillan and Starks (2003) and Aggarwal et al. (2015). A mechanism judged high in principle may therefore perform best on the topics impact investors care least about.
Requested reforms become less likely to succeed as their cost rises (Dimson et al. 2015; Barko et al. 2022). Firms with lower ex-ante ESG ratings face targeting more often. When dialogue succeeds, those firms also record larger score improvements (Barko et al. 2022). Together, these findings mean the engagement literature measures cheap asks against low base rates. The selection problem lies in the evidence base and conditions every rating in the review. We did not find any treatment of publication bias in the underlying engagement studies.
Geography also matters. Slager et al. (2023) find that dialogue with a target firm improves greatly when the investor or a coalition member has a presence in the firm's home country. Home-country presence can be observed and allocated. We have seen little ESG engagement work that models it.
An equation without estimates
The paper states IIP = L x M, yet estimates neither L nor M for any of the eight mechanisms. It supplies no bounds and no ranges. The authors acknowledge that the measurement has yet to be developed. They write that "future research should prioritize the development of standardized metrics for measuring investor impact potential". The conclusion adds, "Our review highlights the necessity for future studies to refine the measurement of investor impact potential".
High, medium and low therefore come from qualitative judgments in a narrative synthesis. We did not find a scoring rubric or an inter-coder reliability statistic. The review extracts no effect sizes, t-statistics or confidence intervals from any of the 52 studies.
The omission matters in practice. The framework can support diligence and police attribution claims. When a manager claims impact from a small minority stake in a mega-cap, the review supplies language and citations explaining why that mechanism receives a low rating. It cannot rank two funds or forecast realised impact because "medium" and "high" carry no units.
The authors concede parts of the problem. They say the mechanisms and their potentials remain open to re-evaluation, while the conditions influencing impact potential are not exhaustive. The sample combines 19 industry working papers with 33 academic ones, and the authors acknowledge that some sources did not undergo peer review. Their defence appears in the abstract: the paper "provides a conceptual model for understanding investor contribution and identifies research priorities" rather than a measurement instrument.
Fair enough. Yet the model arrives with a multiplicative formula that invites arithmetic beyond what the evidence can support. The authors go furthest on secondary markets, where further research is required "where its existence is still questionable". On exclusion, they report that "the effectiveness of exclusionary strategies in increasing the cost of capital of unsustainable firms is still hotly discussed" (Landier and Lovo 2020). They side with Pástor et al. (2021), who demonstrate that lowering the cost of capital for sustainable firms raises it for unsustainable holdings and makes those holdings less attractive.
A review of impact investing that concedes public-market impact may not exist is doing something honest.
One methodological choice determines much of the sample. Every query requires the term "impact invest*". Yet the argument relies on Hirschman (1970), Heinkel et al. (2001), Gillan and Starks (2003) and MacLeod and Park (2011), all outside the stated 2013-2024 window. The mandatory term failed to capture the finance literature supporting the argument. We also could not find stated criteria for the 1,815 exclusions attributed to journal ineligibility, which prevents the funnel counts from being re-run.
We could not test any of this. The central claim requires company-level outcome data, including attributable emissions reductions and labour outcomes, linked with investor-specific engagement, voting and coalition records. We have neither. Price and fundamentals data cannot establish financing additionality. We have taken a related position before in our review of ESG momentum in EUR credit (/articles/esg-momentum-in-eur-credit-pays-only-where-investors-already-objected).
A meta-analysis assigning a success probability to each mechanism, conditioned on ownership stake and engagement topic, would change my view of the ratings. Until then, IIP = L x M gives the question a structure without answering it.