Fixing Herding in Equity Crowdfunding: How Better Platform Design and ECSPR Can Protect Retail Investors
Equity crowdfunding promises democratised finance, yet platform mechanics often turn social proof into distortion. Building on recent HEC Paris research (Åstebro, Fernández, Lovo, Vulkan), this piece argues that silence, not hype, is the hidden killer via “abstention cascades”. We translate the evidence into a regulator-ready design agenda: sequence disclosures for learning rather than FOMO, surface anchor-quality signals over raw counts, use call-window matching to reduce end-rush herding, and align ECSPR/ESMA guidance with behavioural reality. The result is a credible path to protect retail investors, improve selection, and scale Europe’s crowdfunding markets without sacrificing integrity.
Retail capital at scale will live or die by design. We admire the promise of crowdfunding because it widens the aperture of judgement: thousands of small investors weighing in on ventures that once depended on a handful of gatekeepers. Yet the very mechanism that makes this democratic (sequential observation of others’ actions) can corrode it. If platforms amplify social proof in the wrong way, what looks like learning becomes theatre; what looks like collective intelligence becomes coordination failure. The question is not whether retail capital should scale, but whether its scaffolding helps the crowd learn from each other without being led by each other.
Consider an ordinary scene on a platform. It is 08:43 AM on a Tuesday. A promising climatetech venture opens its raise. For forty minutes, nothing. A single €10,000 ticket then lands from an account whose biography reads simply “long-time member”. Within an hour, small pledges follow like falling dominos. Two days later the large ticket disappears; the page shows no trace of the brief jolt that catalysed others. The company closes under target, not because fundamentals were poor but because the first hour was quiet and the one salient signal proved evanescent. What did the crowd actually learn? In such moments, disclosure cadence, the salience of “who” signals, and the matching mechanism matter as much as legal compliance.
“Thresholds can make early support more meaningful and reduce bad-path dynamics, provided the surrounding information environment does not drown signal in spectacle.”
The most careful empirical window we have into these microdynamics comes from the work of Thomas Åstebro, Manuel Fernández, Stefano Lovo and Nir Vulkan. Using the full universe of 710 Seedrs campaigns and 69,699 pledges from 2012 to 2016, they show that the size of the most recent pledge is strongly predictive of the next one: in their preferred specification, doubling the last pledge is associated with a 10.6% increase in the subsequent amount; about £127 on a typical £1,200 ticket. They also document that time since the last pledge depresses the probability and size of the next, and explicitly allow that “abstention information cascades” can arise early, pushing viable projects to miss by a wide margin. Seedrs’ permissive withdrawal feature is crucial context: the authors note roughly 12,373 pledges later withdrawn in their window, underlining how apparent momentum can evaporate. These are within-campaign relationships, not cross-campaign selection artefacts, estimated with instruments that exploit hidden investor attributes and “money returned” events to address simultaneity. It is an unusually rigorous anatomy of how platform-visible cues travel.
Theory reinforces and sharpens the distinction between informative momentum and performative buzz. Lin William Cong and Yizhou Xiao show that introducing an all-or-nothing threshold into the classic cascade environment changes the game: early supporters effectively install a gatekeeper, producing uni-directional cascades that cut down the risk of downward herding on rejections and, importantly, can support efficient selection in large crowds. Thresholds, in other words, can make early support more meaningful and reduce bad-path dynamics, provided the surrounding information environment does not drown signal in spectacle.
Europe has, wisely, built the architecture for a single market in crowdfunding through Regulation (EU) 2020/1503 and its technical standards. The regime sets a baseline: a standardised Key Investment Information Sheet; a knowledge test and loss-bearing simulation for non-sophisticated investors; and a four-day reflection period during which such investors can revoke their offers without penalty. It also embeds a principle of channel neutrality: providers “shall not pay or accept any remuneration, discount or non-monetary benefit for routing investors’ orders to a particular crowdfunding offer.” Yet the law is deliberately light on ergonomics (the micro-presentation of contribution flows, the cadence of updates, the permissibility of auction-like matching on primary pages). That is where most of the behavioural action lives, and where design can quietly do as much for investor protection as any risk warning.
This is timely because the EU market is both young and large. ESMA’s first market report collating 2023 activity finds more than €1 billion raised across the sample of providers, with fully 87% of investors classified as retail. Loan-based models dominate volumes, but equity raises carry the most acute social-proof externalities; precisely where design choices about visibility and sequencing matter most. As the register of authorised providers expands (159 by end-2023) and cross-border activity grows, the cost of small design errors scales with it.
“The lesson is scope: disclosure and mechanism tweaks should be calibrated to platform and jurisdictional realities, not air-dropped as one-size-fits-all.”
It is not enough, however, to staple UK evidence to a pan-EU argument. German data remind us that signals can be double-edged. Felix Reichenbach and Martin Walther hand-collect 88 Companisto campaigns with more than 64,000 investments and 742 updates to ask a hard question: do the signals that raise participation also predict post-offering survival? Some do; others, like large, attention-grabbing investments, are not uniformly benign and can be associated with higher subsequent failure risk, depending on who is behind them. The German literature more broadly, on investor communication dynamics and local bias, shows how national ecosystems inflect crowd psychology. The lesson is scope: disclosure and mechanism tweaks should be calibrated to platform and jurisdictional realities, not air-dropped as one-size-fits-all.
What, concretely, should change? Begin with the most sensitive variable in the Åstebro–Fernández–Lovo–Vulkan analysis: recency and salience. Today, many platforms default to a fully granular, always-on feed of incoming pledges. That is a design choice, not a legal mandate. The plausible alternative is to pace the visible feed for non-sophisticated investors, aggregating updates at fixed intervals (for example, five or thirty minutes) or into size buckets, while preserving a raw, continuous “professional mode” for users who opt in after passing the knowledge test. Randomised trials on Seedrs have already shown that the order in which information is presented can materially shift investor attention and willingness to engage; pacing a live feed is, in effect, a choice about the order in which social information arrives. This is filtration, not concealment.
A second lever is mechanism design. Lars Hornuf and Armin Schwienbacher compare first-come-first-served allocation with second-price auction formats and show that funding paths differ markedly: L-shaped under FCFS, U-shaped under auctions. Translating the spirit, not the letter, of this finding to primary equity raises suggests a simple intervention: intra-day call windows on the primary page. Rather than giving a single ostentatious pledge pride of place in a continuous stream, orders could batch into two or three daily clearings. If you are worried about gaming the boundary, clear each window at a random time within a short band. The aim is not to extinguish momentum but to partition it, so that one large ticket cannot yank beliefs alone.
A third lever sits at the boundary with secondary markets. For years Seedrs ran monthly secondary windows; in May 2025, Republic Europe announced a shift to a 24/7 bulletin-board marketplace. Liquidity changes what investors can rationally infer at issue. If investors know there is a credible path to exit or re-price, they may rely less on the early social proof of primaries. But continuous bulletin-boards import their own microstructure risks; here, basic guardrails (opening auctions, volatility interruptions, transparent “last trade” histories) can cool the heat of the scroll without chilling participation. The policy point is simple: primary and secondary design are complements. Getting one right helps the other.
“None of this denies that momentum sometimes is information. The question is not whether to mute this entirely, but whether to ensure that the part of momentum that moves markets is the part tethered to interpretable signal. That is what pacing, batching, and labelling try to do.”
At this juncture a principled libertarian will object that pacing and batching infantilise investors. Better to show the world as it is and let people decide. The answer is twofold. First, the law itself already differentiates between investor classes and requires tests, warnings and reflection periods - a recognition that presentation affects outcomes. Second, the alternative to pacing is not a laboratory of pure truth but a parade of performances that exploit recency and salience, compounded by reversible “warm-up” pledges. Far from constraining autonomy, dampening performativity enlarges it by giving investors a cleaner informational surface on which to exercise judgement.
A different critique worries about innovation chilling: if ESMA begins to opine on UX, won’t it fossilise experimentation? The opposite is available. Because ECSPR is deliberately silent on micro-presentation, ESMA can encourage controlled experimentation through guidance: A/B tests of cadence, categorical labelling of large early tickets (staff/affiliate, lead with lock-up, returning investor, new retail), and sandboxed trials of primary call windows for raises above a threshold. The point is not to prescribe a single interface; it is to normalise measurement and publish what works.
Two further amendments would align incentives with information quality. First, extend ECSPR’s spirit of channel neutrality to in-platform attention neutrality. Live leaderboards and “latest investor” carousels are routing tools by another name. If they remain, their prominence should be justified on investor-protection grounds or default to opt-in. Second, require a low-resolution identity class for conspicuous signals. A €10,000 ticket from a staff affiliate is not the same as a €10,000 ticket from an independent lead bound by a six-month lock-up. The German survival evidence is explicit that “who” often matters as much as “how much.”
None of this denies that momentum sometimes is information. Cong and Xiao’s thresholds show why, and Vismara’s work on Crowdcube demonstrates that public-profiled early investors can draw in followers in ways that accelerate discovery. The question is not whether to mute this entirely, but whether to ensure that the part of momentum that moves markets is the part tethered to interpretable signal. That is what pacing, batching, and labelling try to do.
Design also has to respect platform economics. Real-time tickers and scarcity cues keep users engaged; attention is revenue. That is precisely why the public sector’s role should be to tilt experimentation towards designs that reduce noise without killing flow. There is precedent for this in adjacent markets. In public equities, the intellectual case for discrete-time matching (to curb arms-race dynamics and reduce the value of microsecond speed) rests on robust theory and growing empirical work on opening and volatility auctions. Crowdfunding is not the LSE, and retail primaries are not HFT battlegrounds. But the transferable intuition, that batching can reorient competition toward price and quality, points the way for primary crowdfunding as well.
“It requires owning the quiet truth that how the crowd sees each other is part of what the crowd knows. If we act on that truth now, retail finance in Europe can mature into what it promised to be at the beginning: not a lottery with good PR, but a curriculum in collective intelligence.”
If Europe wants a practical roadmap, it can be unglamorous and empirical. In the next twelve months, large equity platforms should pre-register field trials in three domains. First, cadence: assign non-sophisticated investors at random to raw, five-minute, or thirty-minute feeds; track campaign success, pledge withdrawals, investor-level regret proxies and satisfaction. Second, mechanism: run intra-day primary call windows on a subset of raises and compare pledge velocity profiles and withdrawal-induced whipsaw to matched FCFS cohorts. Third, market linkage: as continuous bulletin-boards bed in, measure whether the promise of secondary exit substitutes for social proof at issue by comparing the elasticity of early pledges to visible large tickets pre- and post-24/7 roll-out. The hypothesis is falsifiable: if pacing and batching do not reduce manipulation or improve ex-post quality, abandon them. If they do, normalise them.
Proportionality matters. Equity herding externalities are most acute on larger raises with conspicuous signals; pilots should begin there. Loan-based platforms, which dominate EU volumes, face different dynamics and should not inherit equity-centric burdens by default. And micro-providers should see sandbox obligations scale with activity, so that design upgrades do not kill the very diversity they are meant to preserve. ESMA’s first report already treats the market as heterogeneous; policy should match that reality.
Finally, a word about truth in numbers and humility in claims. The Seedrs evidence lives in a UK sample from 2012–2016; German studies on Companisto underscore that national context matters; post-offering outcomes vary with who leads and how campaigns communicate; and measured manipulation and withdrawals are part of the landscape. Causal language should respect these scope conditions. But taken together, the literature now supports a simple normative point: small, inexpensive design tweaks can reduce the likelihood that crowds mistake a stage-managed scroll for genuine information aggregation.
This is where HEC Paris’s contribution should be front and centre of the EU conversation. Åstebro, Fernández, Lovo and Vulkan did not set out to denounce the crowd; they modelled a market in which rational information aggregation coexists with brittle early-stage dynamics. Their estimates give regulators and platforms a ruler: how much an ostentatious pledge moves the next; how quickly silence corrodes belief; how often putative momentum reverses. Cong and Xiao supply the theoretical architecture for why thresholds can harness good herding while disabling the bad. Hornuf and Schwienbacher explain why mechanism choice shapes the path that capital takes. Reichenbach and Walther tell us to ask, with discipline: which signals that excite the crowd also predict survival? The rest is policy housekeeping.
We can, in short, choose design over drift. Treat disclosure cadence as a risk control, not a marketing setting. Turn leaderboards from routers into optional dashboards. Label conspicuous early money by category. Pilot primary call windows for larger raises. Build secondary bulletin-boards with opening auctions and basic collars, not just perpetual scrolls. None of this requires rewriting ECSPR. It requires owning the quiet truth that how the crowd sees each other is part of what the crowd knows. If we act on that truth now, retail finance in Europe can mature into what it promised to be at the beginning: not a lottery with good PR, but a curriculum in collective intelligence; one that Åstebro, Fernández, Lovo and Vulkan at HEC Paris, Cong and Xiao, Hornuf and Schwienbacher, Reichenbach and Walther, and ESMA’s own analysts are already helping us write.
Sources
Åstebro, T., Fernández, M., Lovo, S., & Vulkan, N. (2024). Herding in equity crowdfunding. The RAND Journal of Economics, 55(3), 403–441. https://doi.org/10.1111/1756-2171.12474
Budish, E. B., Cramton, P., & Shim, J. J. (2015). The high-frequency trading arms race: Frequent batch auctions as a market design response. The Quarterly Journal of Economics, 130(4), 1547–1621. https://doi.org/10.1093/qje/qjv027
Cong, L. W., & Xiao, Y. (2024). Information cascades and threshold implementation: Theory and an application to crowdfunding. The Journal of Finance, 79(1), 579–629. https://doi.org/10.1111/jofi.13294
Dambanemuya, H. K., Choi, E., Gergle, D., & Horvát, E.-Á. (2022). Beyond words: An experimental study of signaling in crowdfunding. arXiv (2206.07210).
European Commission. (2022, November 8). Commission Delegated Regulation (EU) 2022/2119 of 13 July 2022 supplementing Regulation (EU) 2020/1503 with regard to regulatory technical standards for the key investment information sheet. Official Journal of the European Union, L 287, 63–75.
European Parliament and of the Council. (2020, October 7). Regulation (EU) 2020/1503 on European crowdfunding service providers for business, and amending Regulation (EU) 2017/1129 and Directive (EU) 2019/1937. Official Journal of the European Union, L 347, 1–49.
European Securities and Markets Authority. (2025, January 8). Market Report—Crowdfunding in the EU 2024 (ESMA50-2085271018-4039). ESMA.
Hornuf, L., & Schwienbacher, A. (2018). Market mechanisms and funding dynamics in equity crowdfunding. Journal of Corporate Finance, 50, 556–574. https://doi.org/10.1016/j.jcorpfin.2017.08.009
Innovation Growth Lab. (n.d.). The wisdom of crowds in equity crowdfunding – A randomised field experiment. Nesta. (Retrieved September 29, 2025)
Lukkarinen, A., & Schwienbacher, A. (2024). Secondary markets in equity crowdfunding. In D. Cumming & B. Hammer (Eds.), The Palgrave Encyclopedia of Private Equity. Palgrave Macmillan. https://doi.org/10.1007/978-3-030-38738-9_244-1
Republic Europe (Seedrs). (2025, May 13). Secondary Market is opening for the last time in May.
Republic Europe (Seedrs). (n.d.). How to buy and sell shares on the Secondary Market. (Retrieved September 29, 2025)
Republic Europe (Seedrs). (n.d.). Introducing Seedrs’ Secondary Market. (Retrieved September 29, 2025)
Reichenbach, F., & Walther, M. (2021). Signals in equity-based crowdfunding and risk of failure: A follow-up on post-offering success. Financial Innovation, 7, Article 54. https://doi.org/10.1186/s40854-021-00270-0
Vismara, S. (2018). Information cascades among investors in equity crowdfunding. Entrepreneurship Theory and Practice, 42(3), 467–497. https://doi.org/10.1111/etap.12261