There is no single definition of manipulation, to be sure—so no way of definitively answering the question. Over the last ten or so years, manipulation has been equated with “buying foreign exchange in the market to block appreciation.” That definition is certainly built into the criteria laid out in the 2015 Trade Enforcement Act. But “buying reserves to block appreciation” wasn’t hardwired into the 1988 act, which has a much more elastic definition of manipulation.
Yet even if the 2015 Trade Enforcement Act isn’t the only possible definition of manipulation, it still provides a bit of guidance – as President Trump implicitly recognized today: “Mr. Trump said the reason he has changed his mind on one of his signature campaign promises is that China hasn’t been manipulating its currency for months.”
The thresholds of being called out for “enhanced analysis” that the Treasury was required to set out in the 2015 act aren’t perfect—no measures are. The threshold for the bilateral trade balance is genuinely problematic. It lets small countries with a propensity to intervene in the foreign exchange market off the hook for one. And even if you think there is sometimes valuable information in the bilateral trade data (many don’t), the bilateral balance really should be assessed on a value-added basis.*
But the current 3 percent of GDP current account surplus and 2 percent of GDP in intervention thresholds are certainly reasonable. Those criteria show that China should have been singled out for “enhanced engagement” from 2005 to roughly 2012, but not since.
But all criteria can be gamed. And I worry a bit that China has been revising its current account data with the goal of keeping the headline external surplus down—it is hard to overstate the number of times the details of China’s services data have been revised since 2014.***