Well in an ideal world, companies shouldn't have to haggle at all -- a public auction should determine the IPO price. Or better yet, companies should be able to just submit sell orders as they see fit, without being required to make any volume or price commitments beforehand.
An IPO is (usually) for a small amount of a company (5-10%) - hence the term "initial". Once the stock is traded, the market is continuously setting prices (essentially a continuous auction) and companies are able to sell more stock if they so wish. The reason IPOs are not also auctions (or market priced) is because that model has been tried and it doesn't work.
" The offer price of $85 was well below initial expectations" (https://www.vcexperts.com/buzz_articles/252)
That's a good point, and applies to all of the previous poster's examples. Does your underwriter work 10x as hard to underwrite a 1B IPO than a 100M IPO? Is it 10x harder for a realtor to sell a $1M house than a $100K house? Does Visa work 10x harder on processing a $100 payment than it does on a $10 payment? It's crazy that these "services" charge a percentage of the transaction price. Why don't these guys get paid by the hour like most professional services? I'd love to get paid a fixed percentage of the price of the products that I work on, but that's ridiculous!
If you are selling before any price discovery is on, you are either getting screwed; or your investors are. If the investment bank could figure out the right price for the shares, why have the stock exchange in the first place.
If this was true in every circumstance, please consider why "market makers"[1] exist within the NYSE and London stock exchanges. If the stock market itself provides the value, buyers and sellers should find "market makers" redundant with zero value-added -- but that doesn't happen.
Therefore, I propose that one doesn't compare investment bank IPO underwriters to a naked stock exchange. Instead, compare the underwriter to a "market maker". The underwriter can be seen as the first market maker for the initial offering. That's worth something to many owners of companies looking to get liquidity.
[1] https://en.wikipedia.org/wiki/Market_maker#In_stock_exchange
The underwriter is like a market maker only in the fact that it is taking principal risk when it purchases the shares. It typically already has buyers lined up for all the shares and it does not face risk from the price changing between when the IPO is priced and trading starts. That is not to say that the bank is not without risk. It has obligations like price support for the IPO.
The underwriter provides value by being _one_ convenient transaction to sell my company's shares. Instead of haggling with 1000 individual investors on "price", I haggle with just the bankers at Goldman Sachs or Merril Lynch. The investment bank then wires me the money and I get to concentrate my efforts elsewhere (like the product) instead of fundraising. (Damnit Jim I'm a tech CEO and not a hedge fund portfolio salesman!) The investment bank is now on the hook for selling the shares to those 1000 investors. Also, the investment bank has the prestige and the rolodex of potential buyers. For me (say a young 20-something) to replicate that would require many golf games and champagne dinners that take my time away from improving the product. Saving me time is worth something, right? (Maybe not a 3% underwriting fee but it's definitely worth something.)
Cutting out the underwriter also means I leave "money on the table" because many investors would rather buy initial shares from a reputable intermediary like Goldman Sachs rather than a startup like me.
If you as a company, prefer to _do_ the same investor relationship work that underwriters do, by all means, skip the underwriter fees.