BFD. Sounds like a bad model, bad data or numerical analysis bug(s)! Or a poor attempt to support a political view by invoking math and data. Oooohhh! Scary!
Every time I have a problem like this, I question and examine my assumptions, data and model until I find the cause.Why would he believe their model & data should do anything reasonable at all?
The hubris is pouring off this article and I don't know whether it is from the author of the marketplace.org article, the researcher Kent Smetters, some unknown peer-reviewed paper(s) or what???
"They made a model of the entire U.S. economy, which required a lot of computing power. “This math problem was a big one,” Smetters said. “And the model computations are about 20,000 times bigger than our standard model.”
Smetters borrowed some computing help from Amazon and NASA and then he and his colleagues then fed the entire U.S. economy in all of its complicated glory into this mega-model. And the U.S. economy… could not compute.
“Their economic models crashed when trying to project out the economy over the long term,” said Jessica Riedl, an economist with the Manhattan Institute who studies the budget. “We cannot even model out a functioning long-term economy under current debt projections.”
The crash itself: not super cinematic, said Smetters. No flashing red letters, no skull and crossbones, no lightning bolt, just a few words that make a macroeconomist’s blood run cold:
Model not converging.
“The model's trying to find what's called a fixed point where everything just adds up, everything's consistent, and it's not able to do that,” Smetters explained.
In other words, if the debt keeps rising at its current rate and we just do not deal with it, even thousands of NASA and Amazon computers all working together cannot get the math to math.
“Really, it's a question of how far can we go before the bond market says, ‘I just don't believe that you're gonna pay us back,’” explained Smetters. "
Every time I have a problem like this, I question and examine my assumptions, data and model until I find the cause.Why would he believe their model & data should do anything reasonable at all?
The hubris is pouring off this article and I don't know whether it is from the author of the marketplace.org article, the researcher Kent Smetters, some unknown peer-reviewed paper(s) or what???
The relevant part of the article:
-----------------------------------------------------
"They made a model of the entire U.S. economy, which required a lot of computing power. “This math problem was a big one,” Smetters said. “And the model computations are about 20,000 times bigger than our standard model.”
Smetters borrowed some computing help from Amazon and NASA and then he and his colleagues then fed the entire U.S. economy in all of its complicated glory into this mega-model. And the U.S. economy… could not compute.
“Their economic models crashed when trying to project out the economy over the long term,” said Jessica Riedl, an economist with the Manhattan Institute who studies the budget. “We cannot even model out a functioning long-term economy under current debt projections.”
The crash itself: not super cinematic, said Smetters. No flashing red letters, no skull and crossbones, no lightning bolt, just a few words that make a macroeconomist’s blood run cold:
Model not converging.
“The model's trying to find what's called a fixed point where everything just adds up, everything's consistent, and it's not able to do that,” Smetters explained.
In other words, if the debt keeps rising at its current rate and we just do not deal with it, even thousands of NASA and Amazon computers all working together cannot get the math to math.
“Really, it's a question of how far can we go before the bond market says, ‘I just don't believe that you're gonna pay us back,’” explained Smetters. "
-----------------------------------------------------
So they
1) Picked/built an economic model,
2) fed more data into the model than had been done before and
3) the model failed. So they then
4) drew hard conclusions about the US economy from the model's failure to converge! and
5) published (although I see no reference to a peer-reviewed article.)
-----------------------------------------------------