The “seemingly crazed idea” that the US dollar will collapse against other major currencies in the post-pandemic global economy is not so crazy anymore, the economist Stephen Roach told CNBC’s “Trading Nation” on Wednesday.

Roach, a former chairman of Morgan Stanley Asia, also said he sees a more than 50% probability of a double-dip recession in the United States.

He based that prediction on historical evidence, saying that in eight of the past 11 business-cycle recoveries economic output has risen briefly and then fallen.

“It’s certainly something that happens more often than not,” he said.

Roach last predicted a dollar crash in June, saying it would collapse 35% against other major currencies within the next couple of years. At the time, the dollar index traded at about 96. On Thursday, the index traded at about 94.41.

He said on Wednesday that he expected the collapse to happen by the end of 2021, but he did not say by how much.

“We’ve gotten data that’s confirmed both the saving and current-account dynamic in a much more dramatic fashion than even I was looking for,” he said.

Explaining his outlook, Roach pointed to dire second-quarter data.

“The current-account deficit in the United States, which is the broadest measure of our international imbalance with the rest of the world, suffered a record deterioration in the second quarter,” he said.

“The so-called net national savings rate, which is the sum of savings of individuals, businesses, and the government sector, also recorded a record decline in the second quarter, going back into negative territory for the first time since the global financial crisis.”

Lingering vulnerability and the aftermath of the initial decline are two factors driving the dollar’s ominous future, he said.

“Lacking in saving and wanting to grow, we run these current-account deficits to borrow surplus saving, and that always pushes the currencies lower,” Roach said. “And the dollar is not immune to that time-honored adjustment.”

Additionally, Roach said, new COVID-19 infections and higher mortality rates must be part of assessing the risk of an aftershock, Roach said.

“As we head into flu season with the new infection rates moving back up again, with mortality unacceptably high, the risk of an aftershock is not something you can dismiss,” he said. “So that’s a tough combination. And I think the record of history suggests that this is not a time, unlike what the frothy markets are doing, to bet that this is different.”