It never worked that way.
Yes, it did. The top marginal tax rate was between 70% and 92% during the entire four decades from 1940 to 1980 in the U.S. During those same decades, investment in business expansion was always fully deductible. During those same decades, per capita GDP growth was above 4%. Since 1981, when the top marginal rate dropped to 50%, and especially since 1986, when it dropped to 35% with the tax reform act of that year, per capita GDP growth has averaged a little above 2%, so the high-tax period outperformed the current years by more than two to one.
While the high taxes were certainly not the only thing helping this to happen, they were an important factor for exactly the reason the OP stated: because rather than pay the taxes, people with incomes above the confiscatory bracket invested in business expansion. The higher taxes discouraged having wealth pile up or be invested in quick-payoff, non-job-creating instruments.
You don't increase business investment by taxing away the thing to be invested.
In the first place, you apparently didn't understand what you were reading, because most people didn't actually PAY these taxes. They invested in tax-deductible business expansion, thus creating jobs, and so the capital WASN'T taxed away.
In the second place, even if it had been you'd still be wrong, because this would be money that wouldn't otherwise have been invested productively.
The limit on investment in job-creating ventures is NOT the amount of capital available to invest. It's consumer demand for the products or services the investment is supposed to produce and put on the market. Capital beyond this limit
will not be invested, and so taxing it away is no loss.
Also, when people have to pay high marginal tax rates, they simply work less. That last thing you want in an economy is your most productive people cutting back on the number of hours they work.
That theory is beyond stupid.
Let me acquaint you with a method commonly used to find out truths about the observable world and how it works. It's called the scientific method. Using the scientific method, what we would do with a hypothesis such as you have expressed above is to make a prediction on the basis of it, and then look at what actually happens to see if the prediction comes true. If it does, then we find your theory has been supported by the evidence, but if not, then we know you were wrong and your theory should be modified or scrapped.
Now, on the basis of what you say here, we would predict -- if you are right -- that during times when taxes were higher, the economy would be less productive, not grow as well, have higher unemployment rates, worse recessions, etc., as the "most productive people" cut back their hours worked.
As noted above, though, the opposite is true. During the four decades when the top marginal tax rate varied from 70% to 92%, the economy outperformed the thirty years since when top marginal rates have ranged from 28% to 39.6% by more than two to one.
Looks to me like my "stupid" theory is the one that matches observable reality, while your "smart" theory is the one that doesn't. And it's whether the theory matches observed reality that matters, not whether you, on the basis of some arbitrary presupposition, judge it to be "stupid."