This graph, from David Beckworth, is pretty hilarious. It’s unfair, but who minds unfair?
Thanks to commenter, uh, David Beckworth, for pointing this out.
I think what this picture best illustrates is the peril of getting drawn into a debate over the recovery question and the recovery question alone. The New Deal did a lot more than just set about a plan for recovery. For one thing, it saved some considerable number of people from starving, which is a nice thing. For another, it gave us a significantly reformed system of regulating economic downturns: a re-drawn Federal Reserve System, the FDIC, the SEC (which, prior to its gutting, was a pretty good thing), Social Security (which includes not only old-age but also unemployment insurance), and a variety of other similar measures. For yet another, it set about hauling the South out of poverty—a project at which nobody had succeeded despite considerable effort since the Civil War. So it was a lot more than just a recovery program.
But also, thinking just of the recovery program, I guess I think this graph isn’t so terrible at illustrating what I think we ought to notice: there was a deep, deep hole to climb out of in 1933, and the rate of climb was pretty quick. Could you really, realistically, expect much quicker, as broken as things were?
One question for David: what numbers did you use to establish trend? Did you include 1928-9, or just 1923-7 (which is one of the methodological disputes here)?
And one further point: the person who really should be on the list with Krugman, Sirota, and me, is the currently-rather-important Christina Romer. Probably also Gauti Eggertson, of the New York Fed, come to think of it.
The Edge of the American West 
18 comments
January 12, 2009 at 10:52 am
jazzbumpa
To my eyeballs* the bottom of the trend line seems to be pretty firmly anchored in the entire 1920 to ’29 period. The graph looks like a really good attempt to fit the proper best-fit straight trend line to the data of the 20’s and the 50’s. I’m guessing a least squares regression, based on the ’20 to ’29 and about ’49 on values, omitting the (aberrant in this context) data from the 30’s.
*Full disclosure: I’ve been in bifocals for about 20 years.
January 12, 2009 at 10:57 am
Vance
I’m guessing he eyeballed it, noting that the two straightish bits of the curve were collinear. But perhaps that’s ungenerous.
January 12, 2009 at 1:43 pm
Kieran
I think for proper historical perspective the trendline needs to extend forward to 2008 and backward to about 4000 BCE.
January 12, 2009 at 2:04 pm
Ahistoricality
By using BCE, Kieran, instead of the traditional BC, you’re making assumptions which bias your results in favor of flaming communist hell.
January 12, 2009 at 2:35 pm
andrew
Proper historical perspective.
January 12, 2009 at 5:31 pm
jazzbumpa
Where did you get GDP values prior to 1929?
That’s when the Bureau of Economic Analysis data set begins.
http://www.bea.gov/national/index.htm#gdp
January 12, 2009 at 6:19 pm
jazzbumpa
The actual nat log(GDP) line is very close to being straight, outside of the 1930 to ’46 range. I just did the simple-minded exercise of dropping a straight line across 1946 to 2000, and projecting it back to 1929. The the real values tend to wobble along just slightly above the line. It hits ’29 pretty well. I don’t have earlier GDP values.
I willing to bet Mr. Beckworth used a valid mathmatical methodology rather than eyeballs or something cooked. Outside of the 1930 to 1946 curvy segments, the distribution of data points above and below the trend line looks really good.
January 12, 2009 at 6:20 pm
jazzbumpa
Here’s a picture.
http://jazzbumpa.tumblr.com/post/70095517/real-gdp-in-constant-year-2000-dollars-pink-line
January 12, 2009 at 7:18 pm
eric
Where did you get GDP values prior to 1929?
Almost certainly Historical Statistics of the United States; you can get a series here, too: http://www.measuringworth.org/usgdp/
January 12, 2009 at 7:18 pm
eric
JFTR, I’ve not suggested he cooked anything.
January 12, 2009 at 7:50 pm
jazzbumpa
Aha. Good find, eric. Thanx.
For 29 on, it’s the same data set I cited above, in constant Y2K $.
And I didn’t mean to suggest you were suggesting anything. Sorry that was misleading.
January 12, 2009 at 8:08 pm
jazzbumpa
I just did a least squares regression on the portion of the dataset from 1900 to 2000. The curves look identical to those in Beckworth’s graph.
January 12, 2009 at 9:11 pm
Michael Turner
The New Deal did a lot more than just set about a plan for recovery. For one thing, it saved some considerable number of people from starving, which is a nice thing.
Silly liberal. “Nice” isn’t necessarily the same as “necessary.” Why, with something like the funding levels and ingenuity employed in the Manhattan Project, we might have invented some way to reanimate those dead people (I prefer to say “terminally unmotivated”, but whatever) and put them back to work when job growth renewed. As I proposed
. . . thinking just of the recovery program, I guess I think this graph isn’t so terrible at illustrating what I think we ought to notice: there was a deep, deep hole to climb out of in 1933, and the rate of climb was pretty quick.
Make sure Jonah Goldberg gets a copy then, with an added arrow labeled “Start of New Deal”. In a recent column Goldberg said something about how it was more down than up, from 1929 to full recovery, rather as if it had been FDR that whole time. I’d go go find the precise place he said that, but I really can’t be bothered with steam pouring out my ears at the moment.
January 12, 2009 at 9:23 pm
Michael Turner
Well, here’s your Jonah anyway. Be warned: the last three paragraphs are thermonuclear idiocy.
January 12, 2009 at 9:45 pm
Vance
It’s tempting to read this:
as telling (though of course he’s just writing fast). If it were telling, it might expose the personal foundation of his brand of conservativism — a conscious leap of faith, in the hope of reassurance.
January 12, 2009 at 10:01 pm
Brad
The problem with that lovely plot is that it paraphrases Higgs et al. as “a large output gap, how horrible” instead of the “look, this is when the New Deal happened, FDR caused the Great Depression!” argument that they seem to be making.
-5 for insufficient sarcasm.
January 13, 2009 at 6:47 am
Chris
IMO it needs one more caption: the 1929-1933 downslope should be labeled
Herbert Hoover: “Oops.”
There would just be room in the lower left, I think.
Snark aside, if we are indeed at the analogue of 1929, and not 1933, then it’s possible that we won’t be forced to go through a four year process of *making things radically worse*.
January 13, 2009 at 12:53 pm
David Beckworth
All:
Michael Turner is correct: I used the entire 20th century to estimate the trend. I just posted figures showing the data for the entire period at my blog. Also, for those who are interested I got the data from the historical GDP database at EH.Net.