Economic activities such as crowdfunding often involve sequential interactions, observational
learning, and successes contingent on achieving certain thresholds of support.
To analyze them, we incorporate an all-or-nothing (AoN) feature in a classical
model of information cascade. Relative to standard settings, we find that an AoN target
effectively delegates early supporters’ downside protection to a later “gate-keeper”, and
leads to uni-directional cascades and prevents agents’ ignoring private signals and imitating
preceding agents’ rejections. Consequently, information aggregation improves,
and issuance becomes less under-priced, especially with a large crowd of agents, and
even when agents have the options to wait. More generally, endogenous AoN targets
improve the financing efficiency of costly projects and the harnessing of the wisdom of
the crowd under information cascades.