This graph contains pretty dramatic news: the lowest quarter of VCs used to make small positive returns up till 1997, and since then they've consistently lost more and more money. It looks like a lot of dumb money arrived in 1997.
The graph is a visual illustration of the standard VC complaint: too much money chasing too few deals.
There's very little I dislike more than graphs with unlabeled (or ambiguous) axes.
Here's a relevant page from Cambridge Associates (the source for this data): https://www.cambridgeassociates.com/Indexes/
I couldn't find the exact stuff Kedrosky graphed but most of their stats are for net end-to-end return to LPs, which would be value actually returned to investors.