Help me find a graph linked to on metafilter somewhat recently

January 23, 2009 11:26 AM

Within the past few months someone posted a link to either a graph or (an article containing a graph) that showed that GDP calculated within a few months of the time period in question bears no relation (no statistical correlation) to GDP calculated with more complete non-estimated data two years down the line. I would like to find this graph.

It believe the link I followed was in a comment, not in a post. I actually don't know if it was on the blue or in AskMe. There is some small possibility that the graph was not actually on the linked page but something you got by following a link on the linked page, so if you know of such a graph, please point it out even if you can't find a metafilter link to it.

posted by If only I had a penguin... to MetaFilter-Related at 11:26 AM (4 comments total)

I don't know anything about the link, but if I understand your question what the graph was trying to show is not correct (at least for the USA). The Philadelphia Fed keeps track of these kinds of data revisions. If you scatterplot the first available estimate of real GDP growth for each quarter against the fully revised estimate (what we have as of January 2009), you get this picture. Clearly the initial estimate is correlated with the final estimate, as it should be. A plot of the two time series is here. In these pictures the unrevised numbers are what is available the month after the end of the given quarter.
posted by thrako at 1:12 PM on January 23, 2009

Hmm...thanks...the graph as I recall it looked a lot like the second link, though I think it was for actual GDP, not growth. Where/how did you find the two graphs you linked to?
posted by If only I had a penguin... at 3:10 PM on January 23, 2009

Come to think of it, the ShadowStats site has been link to a few times, so the graph you're thinking of may be there (though I am skeptical about that guy's results).
I made the graphs I posted, using the data from the Philly Fed link under "Quarterly Vintages". If you plot the levels of real GDP you do get something that looks stranger. This is because of the definitional change from GNP to GDP, as well as (I believe) the use of different base years for prices.
posted by thrako at 3:53 PM on January 23, 2009

Even the real GDP graph looks pretty clearly correlated. If I'm misremembering the graph that would explain why I can't seem to find it, though I would swear it had an explanation/caption essentially saying that there was no correlation. Thanks for going to all the trouble of making the graphs!
posted by If only I had a penguin... at 6:38 PM on January 23, 2009