So that's somewhat similar to Marx's idea, with two key differences:
1. Capital only grows by getting returns on previous investment. If capital "starves all other parts of the economy of resources", that means that capital is wasting the resources, which means that the capital is not earning a return, which means that it stops growing.
2. The end of the cycle doesn't have to be horrible. Marx hadn't seen much in the way of soft landings, but the Fed has gotten fairly good at them. (1929 and 2008 are exceptions, not the rule.)
But isn't that the crux of the problem. At some point those with capital accrue enough power that they don't allow capital to be destroyed?
Instead of taking a loss they lobby for laws and institutions that allow them to socialize their losses. For example laws that prevent student loans from being discharged in bankruptcy; Federal guarantees on loans; the asymmetric response of the Federal Reserve with interest rates (slow up, fast down); suspending of mark-to-market accounting rules; inflation targets that paper over bad investments; and rigged markets (i.e. prescription drugs).
The result is a subsidy from labor to capital and from new capital to old capital, increasing wealth inequality and lower social mobility.
For example assume that electric cars are the way of the future and that companies like Tesla will dominate. In that case anyone who invested in internal combustion engines invested wrong and should lose their money. That is capitalism.
Instead what do we see? Thickets of laws to protect the business models of dealers of ICE cars and a market with thinner profit margins because companies were repeatedly bailed out rather then be forced to shrink or die.
What's interesting is that people might not notice for awhile that the capital isn't earning a proper return, and the realization can be in the form of a sudden price adjustment + crash (think tech bubble burst or housing bubble).
I think even standard economics is starting recognize this, even though the serious study of severe crashes has been somewhat neglected between WW2 and 2008. One problem is that data is hard to come by, given the rarity of the event. Another is that it's hard to come up with a good theoretical reason why asset prices can be persistently wrong for so long.
> (1929 and 2008 are exceptions, not the rule.)
What if it turns out that severe crashes to the tune of 1-2 a century are a rule? How many data points would we need to see before economists would be convinced?
The reason is well understood, but not by mainstream economists. I suggest reading about Hyman Minsky and Steve Keen's research. To summarize, increases in total debt (private or public) contribute to demand, while decreases in total debt subtract from demand. This results in a financial instability. More demand => more debt-fueled investment => more demand. Eventually, businesses that could never have survived at the bottom of the cycle can get huge loans; they might not even seem speculative, because demand is so high and keeps growing. However total debt rises too quickly, due to speculation and Ponzi financiers, and it can't rise at such rates indefinitely. Once debt growth begins to slow, businesses financed primarily by debt begin to fail, because demand slows and they can't convince anyone to give them more loans to cover interest. This triggers runaway deleveraging and fall in demand. In such a down-turn, even businesses that could have survived at the bottom of the cycle have trouble because they've taken out loans or sold shares to invest in growth, since they expected demand to continue to rise. Instead, demand shrinks, because much of it was debt-fueled and unsustainable.
What is required is a good theoretical explanation for why investors fail to predict the point they are in the bubble and rein in speculation, since obviously this cycle has happened many times in the past.
I think the problem is human nature. People forget. Sure, this has happened many times before, but the most recent time was, say, eight years ago. That's a long time in peoples' memories.
And, while there may be this nagging memory of trouble, right now there's money to be made...
If you want a theory for _why_ the future is unpredictable, read up on chaos theory.
During the 1800s, it was more like 5-10 severe crashes a century. Add in the crash of 1907, as well.
What changed? The Federal Reserve, created in 1913.