First, a Review of Our Data Series

We want to repeat our warning that simply using the monthly receipts reported by the Treasury Department is not useful.  This is explained in our Overview of Payroll Tax Receipts but to illustrate this, in Fig.1 below we chart the year-to-year change in tax receipts between each calendar month from 2010 to 2011. The eleven data points are plotted with large gray circles.

We also plot the year-to-year change in the receipts of corresponding "standardized" four-week periods produced by our proprietary calendar-adjustment model.  As you can see, the usefulness of the unadjusted data is limited.

payroll  tax deposits chart 

Figure 1.

 

Recent Trend Change

Again, a reminder that the tax receipts between mid-December and April contain two annual bonus periods, and these distort the picture of what is happening in the underlying employment situation.

Confining our attention to post-April then, we see some minor ups and downs in the growth in tax receipts up to October (Standardized Period data in Fig.1) but there seems to be a more significant dip in November.  Since the data points look-back over four weeks, we can mark the start of the decline in actual tax receipts to late October.

The significance of the dip is two-fold:  (a) Although we have seen growth levels of 0.5% y-y in August, we feel that a lot of the variation was due to short-lived seasonal shifts. The Oct-Nov decline is longer lived;  (b) The year-to-year growth in private payroll earnings as reported by the Bureau of Labor Statistics (BLS) has diverged from our tax series, as we shall see.

In Fig. 2 we show a longer-term view of the tax-growth series, overlaid with the year-to-year change in total earnings for private non-farm employment, as estimated by the BLS. (Official recession shaded gray.)

 

payroll  tax deposits chart

Figure 2.

 

Note that the relative levels of each series (Treasury tax and BLS earnings) change with each calendar year’s tax schedule.  They will differ in percentage levels and slope, but the trends should parallel each other.

What we are looking at is that drop-off in tax growth in the final quarter (to date) of 2011, and contrasting this with a continuation of a roughly 3.5% growth rate in earnings estimates from the BLS Payroll Survey. 

The BLS earnings estimates exclude farm and government workers, but some rough calculations show that it is unlikely that this can explain the discrepancy - unless there has been a dramatic (additional) decrease in hours worked by (remaining) state and local workers.  (The BLS provides estimates of their numbers, but not hours and wages.)

We are, however, using the seasonally-adjusted BLS series.  We do not use the unadjusted series here, since it contains reporting biases which are removed by the adjustment process.  Because this series does not have a long history (commencing in early 2006) it requires special, additional seasonal-adjustment efforts by the BLS.  Perhaps the seasonal adjustment process is having problems with the unusual volatility seen in the data’s brief history.  We wish to undertake a closer examination of this, as an extension of the work we have done in replicating the seasonal adjustment process of the BLS’s job numbers (to be published).

 

Social Security Wage Limit

We have received questions asking whether the drop off in November tax receipts, on a year-to-year basis, might be explained by more and more employees reaching the Social Security contribution earnings limit of $106,800, and at that point making no more payroll contributions. 

Although the year-to-year comparison cancels out the normal decline in social security contributions through the year, there is a "second order" effect whereby the 2011 drop in employee contribution from 6.2% to 4.2% will give a small residual, declining effect throughout the year (all else being equal, and absent actual growth).  However, the decline is too small in size, and extends so smoothly throughout the year, that it cannot explain the November drop we are seeing. (This observation comes from modeling social security contributions based on the Census Bureau’s personal income distribution data.)

There may be further small effects due to larger bonuses at the start of the year causing the contribution limits to be reached earlier, or due to changes in income distribution from 2010.  But given the slow tapering of the social security contributions over the year, we believe such effects must be neglible in size.

The tax deposit series is a wonderfully complex one though, and our analysis continues.

 

Coming Weeks

Unfortunately,  year-end bonus season is upon us, and it will be hard or impossible to separate bonus payments from normal earnings in order to monitor the down-trend we are interested in here.  Perhaps in January we may be able to make some conclusions about non-bonus levels, but we need to await the data. 

Any questions seeking clarification, or insights into what is involved in the tax deposit data series are most welcome:   Contact Form.