Wed 13 Oct 2010
I’ve been getting some mail over yesterday’s column, with angry correspondents posting charts like this, showing government spending as a percentage of GDP, to claim that government spending has too surged:
He says this is dishonest, since:
What’s going on? Yes, that’s right: it’s what happens when you divide by GDP in a time of terrible economic performance. Spending hasn’t surged; in fact, it grew more slowly in the two years after Lehman collapsed than in the two previous years, despite a sharp rise [what's the difference between "a surge" and a "sharp rise" -- apparently Krugman knows] in spending on safety-net programs. Instead, GDP growth has plunged.
This is a totally fair critique — you could show a surge in spending even with a decline in real dollars spent, if GDP had dropped enough — and of course is has dropped over the last 2 years. But here’s where the mindboggling, balls-out dishonesty comes in. Here’s the graph that Krugman posts to refute it:
But if you look at the raw numbers on government spending, here’s what you see:
I’ll take a moment to pause, because everyone who has ever taken a stats course or submitted a research paper has just violently thrown up on their computer screen. What Krugman’s done here is completely unacceptable — he just compared two graphs, after resizing the Y axis. This is 100%, totally, fundamentally dishonest.
Yes, I know that what he said he did was get a graph that showed real spending increases vs. spending-as-a-percentage-of-GDP and he did indeed do that — but the problem is that if that was all he did, the graphs wouldn’t look the way he wanted them to — the 2nd would actually show a larger change (see below), with similarly scaled Y axes. But when he resized the Y, he’s stepping into cheating land. To get a good grasp of how dishonest this is, you have to look at the actual numbers.
The top graph which, according to Krugman, dishonestly shows “a surge,” has a total increase (going purely on my visual estimations, instead of referencing the actual numbers, since Krugman is asking his readers to rely solely on the graph) of ~30.8 to 36.1 — the latter number is about 117% of the former. In the lower graph, the left most number is ~4,100 and the final number about ~5,300, which is about 129% of the former. I’ll let that sink in. Krugman has “refuted” his opponent’s claim of there being a “surge” in government spending by reproducing a different graph with a method of calculating spending that actually shows a higher percentage increase — which, no matter the subjective definition of “surge” or “sharp rise,” I think we can all agree that 129% of something represents a larger “surge” than 117% of something.
The entire reason the second graph looks like a less drastic increase is the resizing of the Y axis, which is completely dishonest. I feel like a broken record, but this is simply unacceptable. You can’t do that and expect to be taken seriously. To get a good feel for how crazy Krugman’s claim that “real spending vs % of GDP shows there was no spending surge” is, you have to understand that it’s usually the “cut spending” (i.e. Krugman’s enemies) crowd that revises the graphs like this — there’s a reason Nick Gillespie, in his “we are out of money” rants at Reason.com, always encourages using the real dollar amounts, because it makes the spending increase look larger!
This is flat out dishonesty.
Edit: also, though I don’t fault Krugman for this, since the graph he was responding to started in 2006, to get a real feel for how much spending has gone up you have to go back to 2000 — if you do, then it really looks like a surge, almost no matter how much you cheat on the Y axis. Bush was guilty of over spending in his first 6 years in office too.