The Mathematical Trend
This trend is purely math-based. It takes no account of any additional sampled data, knowledge or information other than was contained in the prior month’s Payrolls Release. It is not to be taken as a forecast on our part, but rather as a measure of what the mathematical analysis says would be continued growth with, let’s say, the same “momentum” as has been seen in the recent past.
As such, it is useful for comparison purposes with what actually transpires. Any differences may point to significant acceleration or deceleration in employment – or, is often largely the case – to revisions to prior month numbers.
Click to enlarge
Having said that, let’s discuss the trend extrapolated out two months. This is shown for Total Non-Farm payrolls in the chart above. Note the sharp drop in growth built into October. This is a good example of how residual seasonal-adjustment effects move the growth numbers around. We will discuss the major source of this drop below, but first, a summary of the numbers.
Mathematical trend forecast taken from August Payroll data.
| Sector | September | October |
|---|---|---|
| Total Non-Farm | 181,000 | 70,000 |
| Private | 163,000 | 111,000 |
| Government | 18,000 | -41,000 |
October Downdraft
Seasonal adjustment should produce a smooth trend, but it is not perfect. October’s forecast shows the current residual seasonality which the model expects to remain in the adjusted numbers.
Click to enlarge
The August survey reported an unrealistic increase in local education jobs of 21,000 (See our observations on the August numbers), September is forecasting a similar amount.
These are seasonal residuals left over from a struggle with the highly seasonal teacher numbers. October is where the model is playing catch-up with the long-term undajusted growth.
Click to enlarge