Update: Well that was an embarrassing error on my part. It’s actually a post by David Beckworth, which was “borrowed” by the Malaysian blogger.
Nick Rowe directed me to an excellent graph from Economic Mind (produced by an anonymous Malaysian blogger):
Scott Sumner’s case, however, gets even stronger if we look beyond mid-2008 and compare it to the growth rate of nominal spending. The figure below shows layoff and discharges again, but now it is graphed against the growth rate of monthly nominal GDP. Now we see there is a surge in layoffs and discharges but it coincides with the collapse in the nominal GDP growth rate.
The data seems very clear to me. It indicates there was a housing bust that was putting a damper on economic activity, but by itself was not large enough to create the Great Recession of 2007-2009. Rather, that required the failure of Fed officials to stabilize nominal spending in 2008.
I love the graph. But as this old post shows, the case against recalculation is even stronger. During the period after the housing market peaked, non-residential real estate construction and employment continued to rise. Hence the drop in residential construction employment was much sharper than the drop in overall construction employment. Non-residential construction employment plunged after mid-2008, when the economy tanked.
Take a look at the graph and ask yourself what the layoffs numbers would have looked like if the Fed had kept NGDP growing at 5% per year.
PS. I’ll put the following in a postscript, because I know many readers are sick of hearing about the Wren-Lewis/Krugman mistake. But for those who still think I don’t understand accounting identities, I regret to inform you that there seems to be an epidemic of confusion among some of our most brilliant bloggers, especially those named of “Smith.” Here’s a new post from Karl Smith:
I took Scott’s point to be that one must invoke the old Keynesian model in order for Wren-Lewis to have been correct. Its not simply that once one acknowledges consumption smoothing that even a child can see Cochrane was wrong.
Even worse, if you do invoke that model (where I is unaffected), then the balanced budget multiplier is one and there is no consumption smoothing. And smoothing was the factor that allegedly proved John Cochrane wrong.
And here’s several observations that Noah Smith left in my comment section yesterday.
In the classic Keynesian model of a “balanced budget multiplier,” government taxes people X and spends X on building a bridge. People’s income goes down X from the taxes and goes up X from being paid to build the bridge. Hence, there is no change in their after-tax income, so their investment and consumption behavior are unchanged. Meanwhile, output rises by X because now we have a bridge that we didn’t have before (or, alternatively, because now people’s pre-tax incomes are higher by X). Thus, the balanced-budget multiplier in this model is 1. Taxing and spending X increased output by exactly X.
So, basically, Wren-Lewis’ example does not correspond to the classic Keynesian model; he understates his case by forgetting about the extra income that people will receive from being paid to build the bridge, and arrives at a balanced-budget multiplier of 0.8 instead of 1.
In fact, I believe that Scott made exactly this point in his first “S=I” post, where he said that if income doesn’t fall, then no consumption smoothing actually occurred, or something like that.
The more general point is, I can’t think of any Keynesian type of model where the multiplier is inversely related to the elasticity of intertemporal substitution. Multipliers in old Keynesian models would be bigger if people consumed a fixed percent of their income instead of smoothing it. In the main New Keynesian models I don’t think it matters…
Noah Smith is of course the highly respected Keynesian blogger that Paul Krugman linked to as an expert on the issue of misinterpreting identities. Who says I always disagree with Paul Krugman?