I’d like to further elaborate and pose a question to Paul and Larry (really, Larries—Summers and Ball) and, of course, anyone else who’d like to weigh in.
One reason Larry’s analysis is so compelling is that it’s framed quite cozily in neoclassical thinking (at least, of the Keynesian sort) and simple empirics. This is not the stuff of “bloody five-year planners” as Woodhouse labeled the Rooskies. Many years post-panic, we still have large output gaps and no evidence of price pressures. The zero-bound is constraining Fed policy, and thus we must do more with economic policy, not less.
In fact, at the heart of the argument is that even before the crash, a massive housing bubble wasn’t achieving full employment (and its corollary: more equitable growth), suggesting a level of secular stagnation that I and others have been worrying about for a very long time. It’s behind my conviction—really, my life’s work in this field of late—that left to its own devices, the market can’t be counted on to generate full employment.
So, given that this is the stuff of OTE since its inception, what’s the question?
It is this: embedded in Summers’ rap, and Ball’s, and Krugman’s, and my own in many places, is the conviction that one reason we’re stuck where we are (and where we were in the 2000s) is that the real interest rate has been above its equilibrium level. The nominal rate is of course bound by zero but the real rate is bound by the negative of the inflation rate, and it’s the real rate that hasn’t been low enough. If it were to fall to the needed level, we’d have a better chance of closing the gaps we’ve been stuck with for a long time.
To be clear, that is explicitly not where the people listed above have stopped. Everyone on that list has called for more aggressive fiscal policy as well. But, in the spirit of recognizing limits of interest rate policy, how certain are those of us who advocate this position—which given today’s ZLB means higher inflation to achieve lower real rates—that it would help much?
At first blush, this is simple IS-LM stuff—history is very clear that the IS curve slopes down and there’s a negative relation between real interest rates and output growth. Ball is most explicit about the growth benefits of lower real rates, as in this section from his recent paper advocating for a 4% inflation target (link above).
How would a larger interest-rate cut have affected the economy? Let us do a back-of-the-envelope calculation using a dynamic IS curve, as calibrated by Ball (1999):
y_t = -(1.0)r_t-1 + (0.8)y_t-1 ,
where y is the log of output, r is the real interest rate, and a time period is a year. Based on this equation, if interest rates had been two points lower during 2009, output in 2010 would have been 2% higher. If rates stayed constant after that, the effect would have grown because of the lagged output term. The output gain for 2013 would be 5.9%, and the cumulative gain over 2010-2013 would be 16.4% of annual output. Assuming an Okun’s Law coefficient of one half, the cumulative reduction in unemployment would be 8.2 percentage points.
Strong stuff, and surely the sign is right, if not the magnitudes, but there’s something about “secular stagnation” that has a way of messing with old rules. And I wonder if the key is “secular,” as in sector, as in sectoral misallocation. Many observers of the US economy have worried about the impact of financialization—the relative growth of the finance sector—on growth. Part of the concern is the bubble machine, and part is the devotion of considerable resources to non-productive activities.
And the misallocation is profound. Who out there thinks financial markets are playing their necessary role of allocating excess savings to their most productive uses? Anyone?…Bueller?…(do click that link—it’s relevant!)
So, I’m totally with the program re getting the real interest rate down to where it should be—in fact, I recently posted on the importance of measuring the output gap correctly (i.e., correcting for the unemployment rate bias) in this regard. But I’m nervous that it might not be as effective as historical correlations would suggest.
I’d be interested in Paul and Larries responses.