America's economy is a mosaic of puzzles and contradictions that has economists and bloggers scrambling for explanations and scrutinizing the data for quirks and flaws. Lately, I've been thinking dark thoughts: what if all it takes is a single explanation that assumes all the data are correct?
The first puzzle: why is GDP growing so slowly? Since the recession ended, growth has averaged 2.5%, roughly around its pre-recession trend-rate, which means no progress closing the massive gap between actual and potential output that opened over the course of the recession. We know recoveries are weaker after financial crises but they are still supposed to be recoveries, i.e. the output gap should close, albeit slowly.
The second puzzle: unemployment is falling more quickly than the GDP data can explain. I asked Ben Herzon at Macroeconomic Advisers what sort of economic growth the recent drop would normally entail. He constructed an Okun's Law from quarterly data:
Let's consider the 0.8 percentage point decline in the unemployment rate over the six-month period from August through February. Call this 0.4 percentage point per quarter. The equation below shows that over the last 27 years, on average, a 0.4 percentage point decline in a quarter has been associated with annualized GDP growth of 5.1% ( = -0.4 x -5.888 + 2.735).
So in a six month period when GDP has probably grown roughly 2.5%, annualised, unemployment has fallen at a rate fast enough to justify 5.1% growth.
The drop cannot be due to discouraged or underutilised workers not being counted: measured to include such workers, unemployment rate has fallen just as fast.
The third puzzle: why hasn't inflation fallen further? Phillips-curve based models that prevail in most forecasting shops would have projected much lower inflation given the size and persistence of the output gap than has actually occurred. To give just one example: in June, 2010, the Federal Open Market Committee projected that core PCE inflation would be between 1% and 1.5% this year. Right now, it's 1.9%.
There are myriad, credible explanations for these puzzles; at one time or another I have advanced all of them: GDP has been held back by bad luck and the usual headwinds of post-crisis deleveraging (a point discussed here in this week's print edition.) Unemployment is simply correcting for an earlier overshoot. The divergence between employment and GDP may be due to weather, seasonal adjustment, a temporary productivity reversal, or one of the two sets of data being wrong, most likely GDP, which will be revised up. Core inflation has been underpinned by extremely stable expectations, the pass-through of high energy prices, and the anomalous behavior of rents. Tim Duy nicely notes here how often Okun's Law breaks down.
Or you could go with a simpler but more pessimistic explanation: both the level and growth rate of American potential output is much lower than we think. This would resolve all these puzzles: GDP growth of 2.5% is above, not at, trend, the output gap is closing, and it was probably smaller than we thought to begin with. That would explain why unemployment is falling so quickly, and why core inflation hasn't fallen further. The excess supply of workers and products that ought to be holding back prices and wages is not as ample as we thought.
So this explanation has the merit of simplicity; is it plausible? Adam Posen likes to dismiss supply-side explanations for Japan's economic underperformance by saying the Japanese did not one day wake up and find their left arms had fallen off. I have long agreed with the American analogue: while the crisis and recession have set back America's output, its productive capacity remains largely intact. The rise in unemployment does not seem to be sectorally concentrated. The ratio of unemployed to job vacancies is falling, but not by much more than is typical at this stage of the cycle.
That's still my view. But lately, it is starting to look more like some left arms have gone missing.
Labour force participation seems to have settled at 64%, two percentage points lower than its pre-recession level. If that drop is permanent, it would alone entail a significant decline in the level of potential output. What about productivity? My colleague notes there's a healthy debate going on about whether trend productivity has slowed. I don't know the answer, but it's clear that actual productivity has been pretty unimpressive for this stage of recovery (see the nearby chart from Barclays), and certainly compared to a decade ago. For all the talk of social media IPOs, Apple's market capitalisation and the money pouring into alternative energy, none of these represent transformative technologies with much impact on productivity.
In the early 1990s Japan repeately disappointed forecasters who kept expecting it to return to its pre-1990 era of 4%+ growth. What we realized only much later is that its early 1990s financial crisis coincided with a slowdown in potential growth; Japan simply couldn't grow as quickly as it used to. The nearby charts from the IMF illustrate that the path of potential output commonly downshifts after crises; is it really different this time?
You don't need to be a supply sider to believe that potential has fallen; you could equally worry that actual output has been depressed for so long, that hysteresis has set in and dragged potential down with it. That would make the failure to have delivered more monetary and fiscal stimulus before now all the more tragic (Brad DeLong shows how hysteresis raises the benefit of fiscal stimulus here).
I still find the demand-side explanations for the depressed state of the economy more convincing and my base case is that the economy will outperform the 2% to 2.5% consensus this year, resolving some of these puzzles in the process. But the longer the contradictions persist, the more we owe to ourselves to consider the alternative.
Note: Here are the econometrics behind Ben Herzon's Okun's Law equation:
