At this point, Donald Trump has only the thinnest of reeds to cling to in his effort to prevent his campaign from sinking out of sight. It’s not the fantastical claim that he’s going to capture the African-American vote. He’s not; especially not if he keeps doing things like this. It’s not the myth of “missing voters” who will magically appear to carry him to victory in November. And it’s certainly not the ongoing effort to convince voters that, against all evidence, Hillary Clinton is secretly battling a deadly illness.

It’s the economy, stupid.

Related: Is the Decline in Productivity Linked to Our Aging Population?

Trump is the candidate of the frustrated and the angry who feel that they have been left behind by a country and an economy that simply doesn’t look the way it used to. And the news Friday that the Commerce Department had downgraded the already dismal second quarter GDP numbers from an annualized 1.2 percent to 1.1 percent is the sort of thing that, if he had the self-control to avoid unnecessary distractions, he would be hammering on night and day.

“The downward revision to second-quarter growth was primarily a reflection of weaker state and local government spending, and inventories,” wrote Steve Murphy, U.S. economist for Capital Economics. “On the bright side, real consumption growth was revised up to show a 4.4% annualized gain, from 4.2% in the first estimate. Otherwise, the details of the report suggest the weakness in the second quarter may have been even more pronounced than the expenditure-based GDP measure implies. Gross domestic income, an alternative measure of economic activity, increased by a muted 0.2% annualized.”

The state of the economy is, at best, confusing to many well-known economists. A strong job market suggests there ought to be increased economic growth as well, but the GDP numbers remain stubbornly sluggish.

Related: A More Hawkish Yellen Clears the Way for A Rate Hike’

To be fair, there is substantial debate over whether the disconnect between things like worker productivity and job numbers are actually real. In a recent paper, Kemal Derviş and Zia Qureshi of the Brookings Institution lay out the case that we might simply be measuring productivity incorrectly.