Some imbalances are corrected by the fall of civilization

I don't often read Greg Mankiw. But his First Thoughts on Piketty snagged me. In that post Mankiw asks:

What does r>g tell you in a standard overlapping generations model?

He gives this answer:

That the economy is dynamically efficient (that is, it has not over-accumulated capital).

I think this is what he means: If the economy had over-accumulated capital, then r would fall, becoming less than g.

Sort of like, if they print too many dollars, the value of the dollar falls.

Mankiw's answer is a little too brief and dismissive for me. For one thing, r might fall due to over-accumulation of capital without becoming less than g.

But here's what really bothers me:

Mankiw's answer assumes that the relation between r and g is the natural result of economic forces. Yet I know for a fact that economic policy encourages the growth of r. It encourages the growth of r as a way to encourage the growth of g.

Man interferes with nature. Policy distorts the "natural" economy. It's supposed to do that, yes. But you can't then turn around and argue, as Mankiw argues, that things that happen because of policy are natural results. That argument is used all the time, and it is wrong, all the time.

Because policy interferes, r>g tells us nothing about the over-accumulation of capital. Mankiw's assertion is baseless.

One more thing... Mankiw's answer appears to depend on what Anatole Kaletsky calls "an inbuilt assumption that the economy is self-stabilizing." Mankiw thinks everything comes naturally to balance in the economy, like water seeking its own level. If there is an over-accumulation of capital, he thinks, the economy will naturally change so that the over-accumulation dissipates and balance is restored.

Maybe... but on what terms, and at what scale? The business cycle is part of that natural process. So is the Kondratieff wave. And so is the Cycle of Civilization.