With the new weakness revealed in the non-farm jobs number, there is a concern about the potential for a double dip. Some are looking at the lower speed of job growth and some are wary that GDP growth has dipped down below 2% Yr/Yr, and that might be a signal of an oncoming recession.
We do not think the data supports any such conclusions. Forward-looking GDP is yet to be determined by events we still do not know about, like the potential for unraveling in Europe. Setting aside the unknown, it does not look at all as though we have already begun a slip into a double dip.
Click to enlarge
For those that want to look at the slippage of growth below 2% as a recession signal, let's make three points:
(1) Yr/Yr GDP growth was below 2% for three quarters running and it is now at 2%
(2) Potential GDP growth used to be a lot higher, now the CBO puts it at 1.8%, yes, potential GDP growth is below 2%. Thus believers in the 2% signal apparently accept that we are cursed to never have an expansion again!
(3) GDP is still volatile; the standard deviation of Yr/Yr GDP growth as a quarterly series is 2.7 percentage points. That means given the reduced potential GDP, much lower rates of growth than what we have seen historically are statistically in the picture on a regular basis.
The notion that we can take some fixed rate of growth from history and apply it to a trigger point for recession today, as potential GDP growth has ratcheted lower, does not make any sense at all. When potential GDP was at 3.6% in 1999, actual GDP growth of 2% would have been 1.6 percentage points below potential and would have started to create slack resources. The unemployment rate would probably have risen and that could have been the signal for a real recession. But in this economy, growth of 2% is actually above trend (to the CBO)! There is no slack created. The signal from 2% growth and its impact on the economy are totally different today.
We can look back at a different but somewhat more coherent metric, at episodes when growth fell below potential. Going back to 1951, using quarterly data on Yr/Yr growth rates and the CBO time series definition of 'potential GDP', we find there were 19 such episodes over that span. I define an episode as one of any continuous string of growth rates below the mark deemed to be potential. It could be one quarter or many quarters. The amount below potential is not an issue to the count.
Out of these 19 episodes, 10 of them did not presage a recession within the next year. But of these, ten did signal a recession that came five quarters after the signal was turned on, a period I deem to be too long to call it a hit. Four of the 19 episodes signaled a recession in advance, as GDP slipped below potential growth four quarters or fewer before the recession started. An additional two episodes signaled recession in the very quarter that recession began. Three episodes did not signal recession until after the recession had started.
So of the ten recessions we have had over this period, there have been four timely signals and two coincident signals that were correct. Thus there are correct signals for six of ten recessions. But then there are the other ten signals for recessions that did not come or for the one that came late. On balance, this does not seem to be a very powerful tool.
GDP, which is the comprehensive measure of the economy's performance, is not a very good leading indicator of itself. Given the lower growth potential for the economy, we simply are going to have to live with the idea that slow growth rates are going to be more common than they used to be.
We can also look at the jobs growth signal itself. Job growth has slowed markedly with the non-farm payroll gain at 69K in May, after a downward revision to 77K in April. But the unemployment rate, which had dropped in April, is back up in May. Yet household jobs in May expanded by a stunning 422K after falling by 169K in April and by 31K in March. Despite the May spurt in Household report jobs, payroll survey jobs have averaged larger gains over the last three months (96K) than the household report (74K).
We are left with any number of metrics on the economy to try and predict performance. One interesting metric is the unemployment rate. When the unemployment rate rises by more than 0.5% from its cycle low that has been a very reliable recession signal. And… the unemployment rate is actually very well correlated with non-farm job growth. It has a better correlation to the 12-month average than to the most recent monthly gain, and a better correlation than to household report jobs gains.
As a result, we are not in much danger just yet of driving the unemployment rate up, at least not according to this relationship. Since are having a repeat of out summer slowdown of last year, the current 12-month moving average of non-farm jobs growth includes June 2011 at 84K, July 2011 at 96K, and August 2011 at 85K. The hurdle for job growth does not go up until Sept 2011 when the gain rises to 202K. If we only create job growth in this vicinity for the next few months, it looks at though the pressure on the rate of unemployment from 'weak non-farm job growth over 12 months' might not develop.
Of course, the real unemployment rate is a function of the actual unemployed as a ratio to the size of the labor force. And the payroll statistic does not include either of those two measures, despite its correlation. As always, anything goes. But we are not arguing economic mechanics here, only statistics. And it is unlikely that the labor force would suddenly grow faster if non-farm jobs did not accelerate. Moreover, household jobs can go wild for a month or so without payroll jobs following, but the calculation we did above for the last three months is illustrative. The household and payroll reports can be different and by a lot in any single month, but over time, their growth tends to converge.
On balance I see risks to growth. I see risks in the US in politics, I see the risk of the fiscal cliff. I see the risk from Europe. But I do not see a recession baked in the cake. And I see some time and some room yet to maneuver.