We might just want to admit that the BLS payroll data, imperfect as they may be, are preferable to anything else we have or can concoct.

That's kinda how I always looked at it - though who cares what I think .

The only justification for a 'blend' is if the different metrics measure the same thing but with a slightly different bias... then a blend will have a better shot at producing a meaningful measure of the attribute but at the cost of a slightly larger potential error (RMS of the individual errors if I recall my stats from a couple decades ago).

I think the failure of the blend is - as kharris points out - they don't really measure the same things, not exactly... To use the apples to oranges analogy... the blend measure fruits but is that what we want, or do we want apples or oranges? That is the first and most important question to ask & I don't see it asked.

I'm deeply suspicious of all these payroll reports now, but the ADP moreso after this month.

Maybe it will be vidicated by revisions.

I do think it's curious that the -40,000 ADP jobs report seemed to be substantiated by a couple of other indexes (that Hudson and Monster index). But that assumes they have merit...

Cal - interesting read. I think they got the idea right but are short on the time scale...

Looking at THIS chart I'd say it will take a couple years or more to get back to historical affordability... if the curves remain on that trendline. Obviously much sooner if prices drop a lot - I don't see interest rates going a lot lower or incomes increasing enough over 2007 to make a big difference.

It will take time & some pain. JMHO.

The Hudson survey and Monster report did not support the ADP number.

I guess because the headline numbers were all negative they got reported together in a couple of articles "ADP jobs decline 40k, Monster Employment Index declines 8 points, Hudson Index down 2.6 points!!!!"

I admit there was perhaps a lack of due diligence on my part.

I never would have suspected that it's so difficult to come up with a clear picture of employment changes, especially given how decisivly markets respond to these reports.

i actually work for adp and the sampling is from actual payrolls. remember folks we do almost 20% of all US-based payrolls so its got to be taken into consideration.

Thanks Richard - one question, does your sample 'map' accurately to the whole 'universe' of companies covered by the BLS? Do they test for fit of the 'map'?

I say that 'cause I would guess you do more payrolls for some types of companies than others (say differentiated by size, geographic location, SIC segment, etc.)...

I would also guess that there are differences between employment/layoff rates among these different segments.

Considering that 'job gains/job loss' is the sum of then difference between pretty large sets of numbers I would not be surprised at all that there might not be agreement - yet both methodologies could still be 'right' for the purpose they were intended.

Could you shed some light on this - without say getting yourself in hot water should this forum be 'monitored'?

TIA

dryfly, since we do 20% of all payrolls you can safely assume we have a good cross section of coverage of private companies in terms of size, geography and industry. also note that adp doesn't actually produce the algorithms that arrive at the #'s. that's done by a partner who takes our data. the adp number is but one of 3 data points that's blended together and then called the ADP report. I will admit some of it has to do with marketing. We do payroll, how much press coverage would that typically generate?