So, I'll go on record here and say that I think Obama's proposed "financial transactions tax" and "bank tax" are poorly designed.
I've already outlined what I think a better alternative to the financial transactions tax would be: a turnover tax. An explanation of how it would work, and links to why it's solving a problem that people overlook, is here: http://tfideas.blogspot.com/2010/01/how-to-deal-with-high-frequency-trading.html
I've listed other financial reform measures here: http://tfideas.blogspot.com/2010/01/how-asset-bubbles-and-rational-casino.html
I write here about a better version of the "bank tax".
The current bank tax targets all bank assets, ignoring insurance companies, auto companies and other recipients of the bailout.
The problem with this is it actually encourages everyone to seek bailouts in the future. Non-core companies (autos and insurers, here) can seek bailouts knowing they won't be punished for it. If banks will be forced to "pay to recoup bailouts" in the future whether or not they accept bailouts, they are likely to accept more cheap capital from the government in order to facilitate expansion, knowing they'll pay the tax on it later no matter whether they take it or not. Even with moderately high interest rates, if asset prices are very depressed and the bailout is the only source of credit, that credit is worth it to pick up bargain basement assets.
You don't want banks treating the government as a normal source of financing, albeit a countercyclical and highly unsophisticated one. Banks receiving bailouts should NEED it. Taxing banks that didn't need bailouts creates a bizarro moral hazard by which banks seek bailouts even when they DON'T need it. That's a real handout to the financial sector, and a completely unwarranted one.
The logical response is then that the government should enact a bailout tax. Any companies that received bailouts need to pay for it in the long run. Companies that don't take government money, don't need to pay.
There are nuances with this. Firstly, the banks functionally already pay a tax on their bailout money - all have to pay significant interest and equity on the bailouts they take. If the gvt were a private sector company, the fact that they're not recouping the cost means that the government didn't charge interest high enough. Of course, if the gvt were a private company, then they'd be pricing in defaults, which the government could not allow because it's a massive adverse selection problem - nobody takes bailouts because the interest rates are too high, so the only ones who take it are the ones who are very unlikely to be able to pay it back anyway, and the government needs to avoid defaults because they can bring the system down. Thus, allowing overall higher interest rates to account for default becomes de facto nationalization of weak banks and nobody else takes anything, even if they need it. The system still isn't stable.
So if the government can't just raise interest rates (functionally the same as a regular bailout tax), what does this mean? It means that you want variable interest terms for each bailout recipient. However, you can't do that during the bailout because it'd just sink the weak companies, both because they can't pay the interest immediately and also because a perception of weakness can drive their equity balance downward, forcing more capital calls. You also are relying on the government to accurately assess the strength of each company individually, under crisis pressure.
Two more practical problems: firstly, the government forced all banks, healthy or not, to accept a bailout so that the weaker ones don't look too weak. Secondly, here is the list of companies that the government has committed the most money to as of today (source: http://bailout.propublica.org/main/list/index):
AIG, Fannie Mae, Freddie Mac, General Motors, Citigroup, GMAC, Chrysler, BofA+Countrywide, PNC Financial Services, JP Morgan and SunTrust.
With the exception of JP Morgan, this is a pretty representative list of the companies on which the government is set to lose the bulk of the money.
Do those companies (other than JPM) look like companies that are currently capable of paying a tax on their portion of the bailout? One day, they SHOULD pay it all back, and for any notion of moral hazard, they're the ones you're concerned about (everyone ELSE paid back with interest, indicating they should have had lower rates to begin with), but they're very weak today.
So how do you achieve variable interest rates for different credit quality companies, only tax companies that sincerely took bailouts (as opposed to taking them from public pressure), and still not sink your weak companies in a fragile environment?
The answer to me seems to be accelerating interest rates (which could also be done retrospectively as a growing tax on the actual amount of bailout dollars outstanding).
If you had a situation where bailout dollars has no tax and reasonable (but nontrivial) interest rates in the first year (as actually happened), and then accelerated the interest rates over time, then you'd be in a situation where you don't sap crappy companies of even more money while they're weak, but eventually they are forced to pay everything back with interest. For some companies, the bonds could even be "bunny bonds" for the first few years of payments (bonds where instead of paying cash interest, the recipient just adds the interest to the principal of the debt until they can pay it all off)
This has a fairness about it (the companies you're doing it to would have taken all the exact same money anyway, cuz they're the ones who needed it most), and the companies who can pay back quickly with interest and who didnt get themselves into the mess in the first place aren't hit too hard. The companies that really screw up pay for it.
This also has the benefit of not promoting zombie companies that persist forever in a weak state. At some point, the truly weakest companies, the ones that will never return to effectiveness, are going to default because interest rates rise. However, the default is delayed and the major remaining creditor is the government. At that point, the bank can go through the US's usual swift (and often severe) bankruptcy process, removing a "zombie" from the field and leaving real competitors, with a competitive pool that is much stronger than it was during the crisis (and thus may be better able to absorb a bankruptcy).
Thus, an accelerating "outstanding bailout" tax has the feature of avoiding zombie socialism much more than a straight bank tax would (a bank tax may even cause zombies to persist longer, as all the banks are weakened but the non-accelerating part means they're rarely weakened to the point of dissolution, and if they are, it's slow).
True acceleration of the interest rates also means that companies are less likely to use the government as a source of cheap capital for expansion, because of the potential risk of not being able to pay it all back.
It's a system that's both fairer and more aligned with proper incentives than a comprehensive bank tax.