Let's start with yet another defense of the stimulus by Christina Romer.
The piece starts with a remarkable display of selective amnesia. Romer says,
After listening to Representative Paul Ryan in the vice-presidential debate, you might think that careful evaluation isn’t needed. In his view, we spent $800 billion on the stimulus, yet unemployment still rose to 10 percent — so obviously it wasn’t helpful.
She then goes on to (correctly) point out that to evaluate a policy, we actually need the counter-factual, what would have happened without the policy in question.
But she ignores the counterfactual elephant in the room:
Which of course arrived to us in a document authored by... Christina Romer (and Jared Bernstein)!
Even more amazingly, she says the stimulus would have worked better if people would have believed in it more! In most circles, this is known as blaming the victim, not economic analysis.
And yet, who could fault people for not believing in the stimulus when by the second quarter of 2009 it was obvious that it would not accomplish anything nearly close to what the government had claimed it would accomplish?
Case #2 is after the jump!
Now let's switch to good old Thomas Friedman, who informs us that the President is making a mistake in not campaigning on his signature achievements: the Race to the Top in education and his increases in CAFE standards (I am not making this up).
Yes, the Race to the Top, "has already set off a nationwide wave of school reform", and is a "gamechanger".
Got any evidence Tom?
Some 4,500 state and local teachers’ union affiliates have signed onto their state’s reform proposals, showing they want to be partners. Roughly 25 percent of the turnaround schools, Duncan said, “have already showed double-digit increases in reading or math in their first year and about two-thirds showed gains.”
Yikes!
Umm, what about the control group, Tom (or I guess I should say Arne, since the quote is from the Secretary of Education)? What percent of the non turn around schools had double digit increases in reading or math? What percent had some gain?
Of course even that is not what we really need. We really need a good old fashioned RCT, you know, like the ones we pay to have done in Africa and Asia, to determine whether the intervention is effective.
Live by the counterfactual, die by the counterfactual.
4 comments:
A large part of the column argues that graph in your post shows the wrong baseline (and it explains why they got it wrong). That is, the whole point is that the baseline in the graph should not be used to evaluate policy, but you present it as if it should, and then call it the elephant in the room? Huh?
Lot of chutzpah on this Romer lady.
Mark, I'm pretty sure there is one and only one paragraph that "explains" why the baseline was wrong. It's this one:
"As it turned out, even with the stimulus, we lost almost nine million jobs. Indeed, because of horrific job losses in late 2008 and early 2009, we’d nearly passed the six-million mark before the Recovery Act was even signed. Adding in the estimated effect of the act, the correct no-stimulus baseline was a total employment fall of nearly 12 million. With a loss that big, creating three million jobs was helpful, but not nearly enough."
And that's not an explanation...unless you consider "we were wrong" to be an explanation.
Furthermore, and I believe this is to Angus' point, if a macro person (people in this case since there were many forecasters) can't even get the first order estimate correct (the # of job lost) then why in the world should any estimate of something like the second or third order effect (the # of job created or saved due to the stimulus spending) be believed (especially given that no control experiment was done)?
It never ceases to amaze one how dismissive the liberal media is out Pres. Obama's "accomplishments" (said tongue-in-cheek) yet put a negative story about Romney's wife on the front page of the New York Times as if it was an exposé on rich women… or something.
Post a Comment