If the penalty is meant to be a deterrent, then it should hurt. Otherwise it will not be a deterrent. You have to consider the bank's annual profit if you want it to hurt.
Imagine if See's Candy has a product recall due to malfeasance or NetJets has a crash based on pilot error.
Should any penalties be based on the facts and circumstances of the situation or on the entire profits of Berkshire Hathaway?
also, as we see in criminal law, setting huge penalties doesn't work very well if you can only get a few convictions. it just turns into a game of hot potato where everyone still does the illegal thing to stay competitive and some unlucky firm occasionally gets stuck holding the bag.
a free society should strive for reasonable penalties that are consistently enforced.
The large bank can continue to do damage if they are caught, while the small one cannot. Thus they definitely do more harm, and at the very least will be perceived to have effective immunity from the law.
If you only fine thieves proportional to the value of the stolen goods, then all you're doing is saying it's ok to steal so long as you are sufficiently wealthy. (And if you became wealthy through theft... then you've made a legitimate living through damage to society.)
You don't need the equivalent of capital punishment for stealing and, by extension, don't need to take away all of a caught thief's money, just a sufficient multiple of their stealing to make it unattractive to enter into a career of it.
So the point is not simply that you need to defer thieves from entering a career of it, but that you need also to deter thieves from considering it as a one-time option for when their quarterly earnings are due, and that you need to do so sufficiently well that even a highly skilled thief with a multi-million dollar company salary won't consider it. So either it can't just be fines (asset seizure/prison, maybe for repeat offenders) or those fines must be legitimately severe as measured by the one committing the act.
you might reasonably assume that if you catch a company engaging in one instance of illegal activity, you have only seen the tip of the iceberg and there is much more that you couldn't prove. I would prefer the government not to be setup on this premise, but as I said, this is a philosophical position.
This of course is still proportional to the fraud (and the societal damage), not to the overall revenues or profits of the taxpayer in question.
> If the revenue office estimates that approximately 10% of tax fraud is caught, then the penalty for tax fraud should be at least 10x the amount of the fraud.
this, imo, is the exact situation that creates the game of hot potato. people stop making decisions based solely on EV when the variance gets too high. setting a huge penalty to create -EV may discourage the largest, most forward-looking institutions from engaging in the bad behavior (or at least directing it in a top-down fashion), but most companies and individuals with a shorter time horizon are just going to ignore it. and since everyone does it but almost no one gets caught, it creates an easy way to single out politically useful targets and throw the book at them.
rather than just increasing the penalty to gain ratio, I think it is more productive to ask why it is so hard to enforce these laws in the first place. tax law, for instance, can be so complicated that it is unclear to everyone involved (enforcement, accountants, defense counsel) whether a particular strategy is legal. when wrongdoing is difficult to understand, let alone prove, the government should simplify the laws, not make them more draconian. it wouldn't hurt to spend a little more on enforcement either.
Sarbanes Oxley demonstrated with clarity how to fix this: pierce the corporate veil and send executives who commit crimes (or allow crimes to be committed under them) to jail.
It's a matter of political will and power, nothing more.
If a small bank does $100M of tainted transactions and a large bank does the same, they should be fined the same, IMO.
Your point about the small company is not an analog in my opinion. If you have $500K in profits and a tax cheating penalty of $1MM, it does not follow that the precedent is that tax penalties are 2x annual profits, but rather that tax penalties are proportionate to the tax cheating. If another company has a tax fraud of 50x the scope and 1000x the profits, their tax fraud fine should be 50x yours, not 1000x yours.