Ch. 7: IPO Underpricing 413
their bids on the bids of earlier investors, rationally disregarding their own information.
Successful initial sales are interpreted by subsequent investors as evidence that earlier
investors held favorable information, encouraging later investors to invest whatever their
own information. Conversely, disappointing initial sales can dissuade later investors
from investing irrespective of their private signals. As a consequence, demand either
snowballs or remains low over time.
The possibility of cascades gives market power to early investors who can ‘demand’
more underpricing in return for committing to the IPO and thus starting a positive cas-
cade. It is in this sense that cascades may play a role in explaining IPO underpricing.
But cascades are not inevitable. In bookbuilding cascades do not develop because the
underwriter can maintain secrecy over the development of demand in the book. Less un-
derpricing is therefore required. Bookbuilding also offers the issuer the valuable option
to increase the offer size if demand turns out to be high (either unconditionally, by is-
suing more shares, or conditionally, by giving the underwriter a so called overallotment
option).^19
If investors can communicate freely, cascades also do not form, for then investors can
learn the entire distribution of signals. YetWelch (1992)shows that issuers are better off
with cascades than with free communication, because free communication aggregates
all available information which maximizes the issuing company’s informational disad-
vantage compared to investors. Moreover, preventing free communication reduces the
chance that one investor’s negative information becomes widely known, and so reduces
the likelihood that the IPO will fail.
6.1.1. Testable implications and evidence
Arguing that underwriters with national reach can more easily segment the market
and so prevent communication among investors than can local or regional underwrit-
ers,Welch (1992)derives several testable implications. Most importantly, compared to
locally or regionally distributed IPOs, IPOs managed by national underwriters are pre-
dicted to be less underpriced. While this implication has not been tested explicitly, it
relates to the literature on the relation between underpricing and underwriter reputation
discussed earlier, at least to the extent that market-share or tombstone-ranking measures
of reputation correlate with the bank’s geographic reach. Recall that the sign on the re-
lation between underpricing and underwriter reputation has flipped since the 1970s and
1980s, which implies mixed support for the cascades model.
On the other hand,Welch (1992)also stresses the factors determining which issuer
chooses which type of underwriter. Specifically, in the presence of fixed costs, the more
risk averse and capital-constrained the issuer, the greater the benefits of national distrib-
ution. Thus the choice of underwriter is not random, implying that simple OLS estimates
(^19) Overallotment options entitle the underwriter to purchase additional shares (usually 15% of the offer size)
from the issuer at the IPO price. Such options are sometimes called ‘green shoes’.