A bunch of economic relationships are not behaving the way a lot of influential people say they should.

Sometimes that’s because, as Krugman stresses, such people have the wrong models.  For example, they expect monetary and fiscal stimulus to generate breakout inflation and spikes in bond yields because they fail to grasp unique characteristics of the current period: liquidity trap, ZLB, deleveraging cycle.

Other times, relationships between variables change for lots of other reasons that are harder to nail down, often slowly evolving structural changes like globalization (as markets become more open, their dynamics become more interdependent), climate (which can affect seasonal patterns and even crop yields in economically meaningful ways), and behavior (more elastic responses to higher gas prices).

For example, I’ve been struck by how steady core inflation has been in relationship to the large and protracted output gap we’ve been stuck with in recent years.  The figure plots the annual percent change in the core CPI against the output gap, measured as real GDP/potential GDP (the latter from CBO), 1980-now (e.g., at the end of the series, real GDP is about 5% below its potential).

Sources: BLS, NIPA, CBO

The first thing you might notice here is the V of the early 1980s recession versus the L of this one.  Both downturns were comparably deep in terms of the proportionate loss of real GDP relative to its potential, but in the earlier case, the Fed jammed interest rates up to whack inflation and so rates had a high perch to fall from when inflation receded.  In other words, no zero lower bound.

In our current slog, inflation was never untethered, and you see the GDP gap closing very slowly.  While there’s a little movement in core inflation, there’s not much.  Are inflation expectations so “well-anchored” that the there’s less deflation than you’d 87uu7u8 (my cat typed that!) expect given the persistent GDP gap?

GS researchers believe so (no link):

Despite a very large output gap, inflation has been fairly stable at levels only slightly below the Fed’s 2% target. This provides a strong refutation of a simple “accelerationist” inflation model, in which this year’s inflation is equal to last year’s inflation plus a term that depends on the level of the output gap. This implies that there is less risk of outright deflation than one might have thought before the crisis. However, note that the anchoring of expectations cuts both ways; there is less risk of deflation, but also less risk of inflation if the central bank does oversupply monetary stimulus.

When I run that regression through 2007q4, I get a coefficient of 0.098 (t-stat: 5.39).  But when I extend the equation through 2012q2, the coefficient falls by almost half (0.049, t-stat: 3.51).  If I use the first regression—through 2007q4—to forecast core inflation through last quarter, price growth turns negative (deflation) in 2009q3 and by now, given the existing GDP gap, the model predicts that core prices would be falling by 6%!  That’s more (simplistic) model failure than a plausible result, but there it is.   Something important has changed.

Whatever’s driving core price stability, whether it’s expectations or some other dynamic, including, of course, Fed actions themselves of late (the “unconventional” balance sheet expansions and the “twist”), a key point is that in the face of this large, remaining output gap and these very stable prices, it seems an awfully long reach to cite inflation as a reason for not doing anything.

It’s also an interesting economic question as to why core prices have remained so stable in the face of such historically large output gaps.  Feel free to speculate in comments.  Humans only.