maybe that's an interesting idea.  I tend to detest salary caps in sports (I tend to get outraged and how underpaid NFL players are--yeah, i'm idiosyncratic like that, i guess), but maybe there is an analogy here.  Whether, assuming Congress could pass such a law, it's a good idea, I don't know.  Sports salary caps are wealth transfers from players to owners.  Corporate caps would have the same effect, I guess--wealth would transfer from executives to shareholders, presumably. 

 but, even assuming that (a) sports salary caps are good and (b) corporate ones are practical, it doesn't really deal with the problem.  Mainly, executives are overpaid precisely because those setting their salaries are not really the "owners"--they are the board, which, seemingly, is not being responsive enough to shareholders.  So, it seems to be that efforts to strengthen shareholders, either through SOX-type disclosure rules or shareholder activism (like that seen at CalPERS and whatnot) are better for solving the root problem and also not as disruptive to the market (i mean, a salary cap really distorts the market, and i'd say in bad ways, even for Bill Ford and that Wagoner guy).

Well, we tried it once--it has been a disaster.  Maybe an improved version would work, but it isn't an easy task to design one.
Current tax law (dating, IIRC from the late 70's/early 80's) makes salaries greater than $1mm non-deductible for corporate income tax purposes.  However, benefits and performance-based pay are not included under the cap.  The impact has been to push compensation toward bonus schemes (based on various sorts of targets) and benefits (like deferred compensation arrangements, use of corporate aircraft, etc), which cost the company more for the same value to the executive and are far less transparent; some of them also lead to more temptation to cook books, fudge numbers, etc.
One big difference from sports is that the salary cap is team-based; you can squeeze salaries a little for the bulk of the players and hire a few superstars.  Since the problem is that CEO’s are paid much more than everybody else, the logistics would be quite different.
 

The other obvious problem is that it could create a talent outflow to countries without a cap.

If you want to play pro football and make any money at all, you have to play in the NFL. If you want to run a large company, you have plenty of options outside the US.

And the danger is not just at the CEO level - it's there for anyone who ASPIRES to be a CEO. If you were very talented and very ambitious, and thought you could be a CEO, would you start your career in a US firm where the most you could ever make no matter how high you went was, say, $2M a year, or in a British firm, where you could make, say, $25M a year if you got to the top? 

well, the answer is pretty obvious:

a cap on salaries would anger our all-powerful lord, The Invisible Ham of the Market!  As the Ham is all-powerful and all-wise, it would bring great misfortune to challenge him as his divine will is exercised in the Marketplace.  The Ham has decreed the value of certain people, as reflected by the salary they earn (or money they inherit).  If a company that pays its CEO $2 million per year is failing, then clearly they aren't paying enough money to attract a CEO good enough to allow that company to succeed.  If the board and stockholders of a company don't want to spend the money to pay a CEO worth $4 million per year, who's fault is it really when the company fails?

Note: formerly the Ham was believed to be an invisible hand, but an invisible hand is clearly absurd.

Commentarlian is right too - a point I made in the comments to your last post on this issue.

This would be a problem for many reasons.  There differences in the skills needed to run a small company vs. a large one.  So you would have to make the cap large enough that large company CEOs could be 'properly' compensated.  In doing this you set the cap way too high for small and mid-sized companies.  This would create perverse incentives in lots of unpredictable ways.  If you scale it to some measure of company worth, you end up with huge valuation problems--far worse than the income and wealth measurement problems that cause many liberals to resist means testing for Social Security.  Some companies would end up deforming themselves to fit certain valuation models that would not be otherwise efficient.  As an example of the problem:

Which company is more valuable, a company of 200 employees in a low-labor  high-margin niche market which makes $50,000,000 in profit or a company of 25,000 employees in a high-labor low-margin general market which makes $50,000,000 in profit?  The second job takes much more management skill and at the moment is likely to have a higher paid executive.  But if you evaluate it on profit they are identical.  So you can't measure it just on total profit.

Which is more valuable, a trading company of 15 people that sells hundreds of millions in real estate for a medium total profit, or a small grocery chain of 20 stores which sells hundreds of millions in groceries for the same profit?  Right now the CEO of the small grocery chain could make more money.  You probably shouldn't tie to toal sales.....

You can't measure it on number of employees because that would overvalue jobs which are labor intensive compared to jobs that are automated.  This would tend to pervert incentives to making dead-weight jobs.  This might sound good in the short run, but ask Saudi Arabia how that works in the long run.  (Especially ask them if oil ever becomes significantly less important). 

In the Social Security arena I can at least conceive of where to start if you want to benchmark wealth and income.  But what the 'proper' measure of company size to index an executive cap isn't something I can even figure out where to tackle.