MJ Perry posts this graph on his blog yesterday.
Perry quotes Scott Grannis who argues this chart lends support that Keynesian ideas are wrong. Expansion of the government leads to more unemployment and this is because the government is inefficient. It mostly takes money from one group and gives it to others.
But precisely because that's true, we should be suspcious of that interpretation based on this graph. Causation could easily run the other way. When unemployment increases, that puts more pressure on unemployment insurance, Medicaid, and other programs. It also increases the demand for fiscal stimulus.
Causation could also be confounding as well: something that's causing both higher unemployment AND a rise in government spending as a percent of GDP. This seems very likely as the "as percent of GDP" means the value could rise even if all that happens is GDP falls. Which is exactly what you'd expect to see if people are losing their jobs.
As sympathetic as I am to this anti-Keynesian take, this graph doesn't actually tell us anything useful.