> "Well-established protocols" can result in IPO prices being systematically set too low. Yes, theoretically a bank should be able to break the ranks, but all they would get for their trouble is smaller profits, and a potential lawsuit from investors. After all, they did diverse from the "established accounting standards" when pushing the IPO price up.
Assuming a roughly competitive market with n players that do not engage in direct collusion, if IPO prices are being set too low from the perspective of the companies IPOing, there's room for an additional player (n+1) to set their prices slightly higher. Assuming their ability to predict the risk on the opening bell prices is the same as the other n players' ability to predict risk, that bank will produce IPOs that are consistently favorable for the companies IPOing, and companies will choose that bank as their underwriter. Ceteris paribus, their profits would grow, not shrink.
There are factors that impede this from happening perfectly in practice - such as barriers to entry for the underwriters - which is (part of) what explains why this disparity won't trend to exactly zero. But it's wrong to say that banks would get punished by either companies or their investors for responding to this disparity by raising prices - the exact opposite would happen. And in itself, that still doesn't point to widespread collusion between banks, or even any sort of implicit conspiracy.