How to Interpret the Latest Employment Number
While we will never know what the Household survey would have reported before Wednesday night's presidential debate (h/t Jack Welch), there are a few conclusions we can derive from the data as it was presented. It is easy to get caught up in the month-to-month changes but that is ultimately a fruitless exercise since there is so much volatility built in both BLS surveys. That volatility has actually increased in the "new normal" of the recoveryless recovery, even to the point that survey results ironically have been exhibiting a rather obvious seasonal variation in the seasonally adjusted data.
Part of this is explained by the paradigm shift that has yet to have been incorporated in the BLS data series and imputation factors. Remember that seasonal and other adjustments continue to be based on a steady data series of historical events that cut across two obviously different economic archetypes. The old assumptions (and economic "laws" for that matter) for the pre-crisis period just may not be appropriate for the current economic age of central bank activism-writ large, but they remain stubbornly embedded in macroeconomics.
With that in mind we need to be careful about how we analyze and interpret these statistical results. Month-to-month noise can be overcome through the analysis of trends. It is here where the data perhaps provides an additional element of clarity.
Starting with August's Household survey, there were four monthly declines in the past six months. That denoted a quite obvious change in the trend for the employment outlook. The last time this survey exhibited as many monthly declines in such a compressed space (outside of the Great Recession) was the middle part of 2007. Therefore the trend change potentially identifies a phase shift in employment.
Given the September results of +873k, we might be tempted to conclude the trend change was premature and has perhaps reversed. Part of the change in employment trend, however, stretches beyond that month-to-month variation. Even though, by August, the Household survey had published four monthly declines in six months, overall employment over that six-month period was still higher (+67k). What was important was not whether the overall level of employment had fallen throughout the period, but that variation in the Household survey had noticeably risen.
Any time there is a phase shift we expect to find greater variation, particularly in statistical models not expecting nor mathematically able to capture inflection points or periods. Therefore, in terms of variation or volatility, any increase portends the distinct possibility of such phase shifts.
First, a large increase in the level of employment is not at all inconsistent with an economy falling into recession. In August and October 2007, the Household survey yielded very strong gains.
Second, notice the increase in variation of survey results around each phase shift.
Increased or elevated standard deviations in the Household survey align very closely to inflection points in the overall employment picture. In both charts above, the artificial "boom" period shows extremely low variance, which is what you would expect from any period where the overall economy was advancing at a steady pace, ie, an economy actually growing.
As the underlying economic trends undergo significant change in degree and direction, you would expect to find alternating or confusing results as both the statistical models and the human survey respondents find difficulty in making sense of what is taking place. Random walk statistical models, in particular, are uniquely incapable of detecting or identifying changes in trends or inflection because they will always be captured by the historical data series upon which they are constructed. But as the economic foundation shifts beneath them, we see these changes in trend occur.
The next chart below demonstrates the extremely close alignment of our variation-driven phase shift periods with the overall employment picture.
One of the prime reasons that statistical models have had such a difficult time keeping up with the "new normal" has been the paradigm shift in nearly every economic space. While various economic "laws" and theories are predicated on the cyclical recession/recovery cycle (including those driving monetary models that ultimately decide monetary policy at global central banks), it is clear that structural changes have taken place post-2008. In terms of national income, specifically wages, there has not been the typical or expected reversion to the previous housing bubble trendline.
The reason for that disruption is obvious and intuitive - no housing ATM's, no cyclical recovery. Thus we end up seeing data series like the one below: the stubborn elevation of part timers for economic reasons that does not come remotely close to reverting to the pre-crisis era.
Note not only the structural change in how this class of workers is used in the new economy, but also the variation in the results of the Establishment survey compared to the pre-crisis period. Despite the BLS attempts at seasonal adjustments, this data exhibits a very seasonal and recurring spike (which should not occur to such a high degree if seasonal adjustments were keeping pace with economic shifts) in those working part-time for economic reasons in the July - September period. In both 2010 & 2011, the July - September period saw large seasonal fluctuations in part timers for economic reasons, meaning that workers were "demoted" from full-time positions (the majority of the increase in these workers were in the class of "slack work or business conditions" rather than "could only find part time work"). This potentially means that businesses are using the full-time/part-time classification as a variance plug on their workloads, and thus this gives us an idea what the marginal economy is doing in a way not incorporated by other data series such as Jobless Claims.
Superficially, it would be easy to attribute these mini-cycles to the Federal Reserve's monetary programs since they coincide rather nicely. There is probably some impact in employment perhaps through inflation expectations, but since monetary policy in the new normal is incapable of extending into sustainable economic conditions we might expect such "stop-go" results, thus partially explaining these new mini-cycles as some marginal agents and businesses react to Fed attempts to cajole the "right" expectations.
Consistent with the change in variation from the Household survey, however, the mid-year lull or 2012 mini-cycle appears to have started earlier and has potentially lasted much longer than 2010 and 2011 - meaning 2012 may not be like the "stop-go" of the last two years.
There is no doubt that central bank activism has left imprints on various markets, but the spillover or transmission into the real economy has been much less dramatic and far from uniform. In my opinion, this is the reason the Fed has moved beyond limited and finite monetary programs into unlimited debasement. It seems clear that Bernanke & Co. have fixated on these "stop-go" mini-cycles as the pathology of monetary failure, and thus intend to "fix" that misstep through unlimited and ongoing intervention.
The great unknown now is the efficacy of this new monetary experimentation. Count me among those less impressed by this "upgrade" to debasement theory. The problem in the real economy, echoed by the volatility in employment surveys and the structural problems they allude to, cannot be alleviated by "liquidity" - it is lack of sustained employment growth amplified by these continued appeals to forcing inflation into broken monetary channels. Unless the Fed can find a way to circulate real money (as opposed to encumbering credit and debt) in the general population (hello helicopter) there is no method of imposing monetary largesse upon where it is actually needed. The economy needs all this "slack" labor to receive money in the form of wages, not money in the form of new, encumbering debt, or even government transfer payments.
Clearly, central banks are gambling that businesses will appeal to cheap credit for use in productive investments, and that business owners will be tempted by the fruits of asset inflation. The operative theory of asset inflation and the wealth effect is simple - get those that own price assets to hire those without. The problem is that those with price assets that might feel wealthy through additional asset inflation (the top 20% that hire the bottom 80%) happen to notice that the bottom 80% don't seem to be buying as robustly as they once did, and so don't feel much like hiring as much as they once did - and certainly not on a full-time basis. There is no way in 2012 for "money" to get from the Fed to the real economy. All the usual tricks are historical artifacts now (ie, low interest rates always stimulates the economy), so new "money" and wealth effects are increasingly wasted in speculative efforts (stock repurchases vs. capex, the US$ carry trade, etc.).
Economists believe that recessions do not occur without exogenous shocks; that is, an economic system's default setting is growth. That is a mistake. Any system that yields results consistently above its "potential" or natural rate is one that will eventually arrive at a point where the default setting is reversion - there is an elegance of entropy in the ultimately futile attempts to create sustainable economic growth through exogenous monetary inputs. The geometric rise in credit money and central bank activism (for decades) has pushed the economic system to just such a point where even trillions in new currency units all over the globe have minimal, at best temporary, effects. The old playbook is inoperable and even contemporary, poorly suited statistical analysis will eventually catch on to that.