This morning's 8:15 release of the ADP number showed an addition of 170,000 private sector jobs vs a consensus estimate of 182,000, a very small "miss" of 12,000 jobs. (It is worth noting that last month's figure was revised downward from 325,000 to 292,000.) This release provides some information about job creation in the private sector, and also gives us an advance reading on the two key market-moving numbers on deck.
The monthly ADP employment number has been maligned by many as a poor predictor of the government's official monthly Non-Farm Payrolls number, released two days after ADP's. That criticism is likely overdone. In fact, the ADP number is reasonably good at foreshadowing the key government employment statistic--and can provide us with a hint as to what to expect from the monthly ISM number as well.
The ADP Employment Report is a national reading of private, non-farm employment that aggregates information from a sample covering 24 million employees. (There are 138 million workers in the U.S.) It should not be a surprise to see how this ADP information can be useful in estimating the government's official non-farm payroll numbers, though we need to adjust for the fact that the government looks at the entire population, and takes into account government jobs lost or created, as well as private sector jobs.
The ADP number can also shed some light (and I emphasize "some") on what to expect from the ISM Manufacturing data. There is a reasonable link between job creation in the private sector, and manufacturing activity. This relationship is borne out by statistics, and appears when we compare the history of the two readings.
click to enlarge
Above you can see a graph of the ADP data (white line) overlaid with the ISM series (yellow line). The correlation of the two over the past 36 months has been over .88, a significant result. Although the series diverged quite often between 2005-2007, most of the time the pair has been fairly consistent.
Since we now know that the ADP number was 170,000 jobs added vs the expected 182,000, we can estimate, based on this number alone, that the ISM manufacturing number should be close to what is anticipated. The ISM number is expected to come in at 54.5 vs 53.1 the previous month, and based solely on the ADP number, there is not much reason to adjust our expectations. Keep in mind that we are considering the ADP in isolation, so we should take our ISM guesstimate with a grain of salt.
The ADP is even more useful in forecasting the official Change in Nonfarm Payrolls, to be released this Friday. This "NFP" number is probably the most-watched of all financial releases, since the creation or loss of jobs is seen as the most accurate indicator of the country's economic health.
As you can see, the relationship between the two seems to be quite strong, and this is borne out by the .937 correlation over the past 36 months. Since the ADP number was about as expected, we have some evidence that supports the consensus of surveyed economists, that is, an addition of 145,000 jobs.
When the government releases its official NFP number this Friday, it will also release a breakdown of that figure, including the number of private-sector jobs added or lost. Remember, the NFP number itself also contains government-sector job creation. This private-sector jobs number is historically closer to the ADP figure than the NFP number itself, as we might expect, since both are measures of private sector jobs created or lost. The correlation of these two numbers has been .975 over the past 36 months.
We can now compare apples to apples. That said, we do need to keep in mind that the two series are derived in different ways using different data samples, and can sometimes exhibit considerable variation from month to month. To arrive at a rough estimate of what Friday's NFP will be, we can compare the actual ADP figure to the number that had been expected by the consensus, and adjust the expected private sector component of the NFP by the same amount.
We won't attempt to estimate the government sector portion of the NFP, since we have no information about that. We can add or subtract the variation between the actual ADP and the estimated ADP, and add or subtract that number to the expected NFP number. In other words, we won't look at the proportion of the variation, we will look at the number itself.
Since the ADP number was 12,000 smaller than expected, we can arrive at a rough estimate of Friday's NFP by simply adjusting the expected NFP +145,000 by that amount. Our rough estimate based on ADP alone is for a gain of 133,000 jobs. (This estimate ignores the ADP downward revision of the previous month.)
We have to keep in mind that these economic releases are very volatile and notoriously hard to predict. I stress that at best the ADP can hint at what is to come. At the same time, these series don't exist in a vacuum, and can help to shed light on what we can expect, especially if used with other information such as similar releases in other countries, econometric models, anecdotal evidence, etc. We cannot bank on the ADP number alone, but it may give us some indication of what is headed our way.
Disclosure: I have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.