there are many theories about IPO underpricing: asymmetric information, principal-agent problem, or sth else. You may find a wonderful literature review concerning IPO in Handbook of Corporate Finance, just in this BBS.
My theory about IPO underpricing in China follows principal-agent theory which claims: the price of IPO is a trade-off between the influence of issuer and the muscle of stock bidder (usually mutual fund or pension fund), and investment bank will be a coordinator. It will benefit the original owner of the firm to raise the initial price, while the bidder will more likely suffer a loss. To lower the intitial price, the original owner will be undermined meanwhile those fund managers will be happy. So the investment bank, if it want to please those firm owner rather than those fund managers, should try to raise the initial price, and this will lead to "破发“.
In previous years, many firms which are state owned, want to go IPO and some IPO are political task………..they have to be sold (你懂的). So definitely, they are sold cheaply, recently, domestic critics are accusing such cheap sell, but you know, no more than a gesture. Also, there are not so many investment banks and not so many funds in China just a few years before, so the demand side of IPO market is some kind of monopoly which also lowers the initial price of not-state-owned firms. I am wondering whether we could find the same phenomenon in other developing countries, say, countries in Africa also established IPO system. Basically, things have changed in China, the SOE are almost sold out and there are now too many investment banks and funds, the situation will change fundamentally, benefits of those firm owners will be assured. It is not some sort of China-specific problem, a word we too often assert. It is just a step every financial market has to go through, the IPO history is anything but long, and SOE IPO may exaggerate your observation into IPO underpricing.
Just a final tip, there are too many theories in this field and asymmetric information theory seems enjoy the greatest popularity. You may get your own idea. All in all, we need robust empirical evidence to prove any of them.