"Understanding the employment measures from the CPS and CES surveys." by Bowler and Morisi is a good use of time.
What becomes clear is that the most useful read of employment, or at least a good starting point, is not the monthly change or current level of either array (I am now going to call the established survey the CES and the household the CPS for brevity) on its own, but how the CES and CPS are playing off each other. That way the quality of the current phase of the cycle can be identified and the quality of the recovery or recession understood. There is little information in any one month of NFP payroll or the unemployment rate or household change. Frankly the drama around the monthly employment data day is undeserved and a waste, unless one is at a unique critical point experienced once every 30 or 50 years (1983 and 2009) or if the number surprises on magnitudes of at least 2X to 3X of expectations. However considering the differences between the CES and CPS dynamics both short and long term is very useful.
First consider the Bowler and Morisi grid outlining the differences between the CES and CPS:
The most important differences are that small business proprietorship with family contribution of work often unpaid is in the CPS but not the CES. In addition agriculture, domestic employment, moonlighting military, and workers on leave are in the CPS but not the CES. The worker on leave was an important difference during the furlough "crisis" of last quarter. Multiple job holders are counted per job held in the CES but only once in the CPS. The CES is centric to the establishment structure - hence its label - while the CPS is centric to the entire nations demographics, or CES is impacted by birth/deaths of the firm while CPS is impacted by birth/death of the labor force.
Since all the above tend to respond specifically to certain phases of the economic cycle in a correlated way, the CES will do better in the last stage of a well rooted recovery of the economic cycle, but will do worse than CPS in the initial phases of a recession. During the reorganization and turmoil (Schumpeter's "creative destruction" phase), the firm population dynamics will make the CES volatile and confusing, while the non impacted sample characteristics of the CPS will provide a clear picture of trend and the quality of growth. And the best concurrent indicator is initial claims (which the CES folks use to revise the CES.)
Bowler and Morisi portray these differences by tracking the ratio of CES to the CPS non agriculture and wage component:
In general CES ratio to CPS adjusted is doing worse is an important bearish flag and at times predicting but always confirming of a downturn and is shown by the ratio decreasing.
But more information is found by not seeking equivalency with the Bowler Morisi adjustment, but plotting the ratio without their adjustment as the differences between CES and CPS provide more insight for understanding the macro dynamics underway and the "nature" of the recovery or downturn.
This shows that when the economy is in downdraft the CES quickly deteriorates as large establishment businesses go into a defensive mode and reduce employment. This is shown that when a establishment downturn starts, the most obvious one being the DotCom bust leading into 9/11 and then the corporate fraud crush, household is not as impacted and the bottom falls out of CES and the ratio to CPS plummets. Otherwise in a business cycle downturn from monetary policy tightness (usually) the CES leads but also CPS drops as well, making for a more gentle downturn in the ratio. Therefore the dynamics of this ratio can tell us a lot of the nature of the business cycle or recession or recovery.
The largest cumulative difference in both the Bowler Morisi adjusted and the adjusted was in the 1990s (labelled "III") which indicates to my read that a profound change was underway within the establishment data as new technology surged the CES index. I put this to a Schumpeter Creative Destruction phase wracking the economy and redefining the entire makeup of employment. This was massive in Phase III above but also shown in Phase I, II and perhaps is starting now with a Phase IV.
It is interesting to note that the ratio of CES/CPS discretely, perhaps gently, rolled into the 2009 nadir which indicates the impact on employment was not differentiated between sectors during the crisis but was visited equally upon all sectors at the same time. Also the deterioration as far as the quality of the economy began after 9/11. A close up of the above set from 1/2000 on shows this well where there was some improvement post DotCom but then that steadily eroded after 9/11.
It is my belief that the most significant event for the last decade and the current economy was 9/11. That the USA is still in war, though that is ending, and the pressures and strategic arrangements required of war, especially allowing China to gain massive beneficial terms of trade and market access to the USA so as they would not be a problem as USA fought the war, is the near entire story of the last 13 years.
That the solvency crisis was the final straw of a deterioration visited on the USA economy from war as the surfeit of trade imbalances led to a hyper liquidity that found its signature asset in mortgages. So with the end of war and with the addressing of the China sourced imbalances, the quality of the economy is not only recovering from the solvency event but also from the strain of war and the near belligerent trade policy of China.
If CES to CPS ratio is a solid indication of the quality of the economy, then for the first time since the 1990s the USA is experiencing a revival of immense size. When one look sat the long term graph above the previous, it is obvious the USA has experienced three such surges (marked I,II, and III) and not until the last two years are we seeing patterns that are similar to the three prior surges and may be surge IV. During these three surges prior the great gains in US real GDP productivity and wages growth occurred, except perhaps wage growth was muted in the III surge of the 90s (but I suspect further work will show that it does not account for the Pareto Distribution in context of a immense economy), and the three surges were the engines that expanded US GDP ten fold. note as well that all three surges have war as their main inspiration - WW II, Vietnam and the end of the Cold War. I think this is the story underway for the current recovery that it is not a routine cycle but is fueled by epic historical process.
The ratio of CES and CPS also disclose massive movement of payroll (the end expression of CES), for if this ratio does capture the qualitative aspects of the US economy, September was a one million leap in payroll, effectively. Now most of that was from the September, the actual payroll was reported as a large 237,000 but nowhere near the one million qualitative change. The falloff in Oct and Nov, almost totally offsetting the September qualitative gain of one million, is "faux" as is die to how the 800,000 furloughed workers were to be accounted for. The take home is that payroll, below the surface reported, is undergoing massive positive qualitative movements that will be shown to be permanent and might be a historical Surge IV. Also this surge in qualitative CES will always have large bottlenecks and frictions and spark wage inflation - we are obviously past the NAIRU level. The CES surging shows that unless the FOMC raises Fed Funds, not to prevent this qualitative change but to be able to fulfill their mandate, the FOMC will lose all control. Fed Funds should have moved when this surge started, in the 4th Q 2013 but has been kept at bay by goofy monetary theory, the phony budget crisis, and now the unusual weather - which dont really have strategic impact but for obfuscation of the data.