I'm sure someone else has thought of it. But I can't remember reading it anywhere else.

There is a flaw here too though.

Regulators could game this system by taking their lump sum and not doing the work involved with sniffing out fraud. If there's no fraud, they win, and didn't have to do any work.

If there is fraud, maybe they can squirrel some of that bounty into an offshore bank account and sneak out the back door like bernie madoff was hoping to do.

Darn.

This is a hard problem.

That's why I added the bounty system.

You'd probably end up with layers of insurance. So one company might give IBM's financial statements a Seal of Approval, stating that any investor can pay the company, say, $1,000 a quarter to insure up to $1 million in restatement-related losses. Then, they double-reinsure this (every $1 million they lose gets them $2 million from a reinsurer) and offer a bounty for anyone who catches fraud.

Or you could just expect people who suspect fraud to short the relevant stock, announce their findings, and make money if they're judged correct.

You're right that it's hard, by the way. I just think it would be easier if we were able to have multiple voluntary levels of insurance against fraud, instead of a single insurer -- especially because, in many cases, the SEC fines a company when management lies to shareholders. So if XYZ Co. overstates earnings by $100 million, investors are poorer -- and then the SEC makes the company give up another $20 million in restitution.