Friday, June 05, 2009

Up, down and all around

Some days even my head spins. These are the news stories on the same exact news, literally the same exact datapoints released by the state in just this one press release reporting April's labor force information for the region. Keep in mind that that one press release is the sole source of data behind each of the following headlines:

One PBT version yesterday said:

Pittsburgh unemployment rate falls to 6.9 percent

The Trib headline earlier in the week says:

Unemployment rate in Pittsburgh region inches to 7.3 percent

Up? Down? Are they looking at different numbers? Nope, see above for source. The PG version parsed a bit:

County April jobless rate falls

and even the earlier PBT version of the story lead with a different spin on the data:
Unemployment rate falls in Allegheny, Beaver counties.

County? Region? Some counties? Again, this is all reporting on the same monthly report. All the versions are true of course, but you would never be able to tell what the story was if you just read the headlines. Numbers out show the region's seasonally adjusted unemployment rate went up a tenth of a percentage point, while the comparable unemployment rates in Allegheny and Beaver counties are down. The more recent PBT headline is the oddest to a degree. Reporting the unadjusted number is fine, but you want to be careful about what it means. I think it is literally always the case that the unadjusted unemployment rate falls between March and April in Pittsburgh as the cold recedes and employment in things like construction picks up. Literally every year it happens, so you need to take that into account when discussing what it means.

Whatever you want to call ground truth for the region, the nation went up 4 tenths of a percent between March and April. (update: and I forgot today we learn May numbers, up another half percentage point) Given the GM news, hard to see that national trend not continuing some more. But the local unemployment rate has been below the national level for 30 months now. The 1.6 point difference between the two is the largest the local unemployment rate has been below the national rate since June of 1975. The local unemployment rate was below the nation's during most of the recession of the early 1990's, but the most the two trends diverged was 8/10ths of a percentage point for one month. If the divergence now goes beyond 2.0 percentage points (which may well be soon, see update above) it will be the most in the current data we all use going back to 1969.

But for those who do not read much past the headline, different folks will come away with different stories depending on which version they saw. Truth is.. all of these tenth of a point changes could easily be revised by next month and well within what counts as sample error in these numbers. If you parse more than you should, the current unemployment rate went up to 7.27%, so the 7.3% is rounding. But to suggest that means anything is much like all of this this monthly parsing of the data. At best it's overinterpretation.

For the longer term trend there is my (google-enhanced) rust belt watch. Even non-Rust-Belt Charlotte has tipped over 12% unemployment. Even once envied Portland has one of the biggest jumps in unemployment around. Charlotte still is amazing to me in that it was exactly 10 years ago it had a unbelievable 2.2% unemployment rate.

I also found this sentence in the PG version a bit curious:
Manufacturing, which lost 100 jobs in the production of nondurable goods in April, lost a total of 7,700 jobs since the same month last year and is at its lowest level of employment since 1990. (emphasis added)
Since 1990? This is one of those cases that the data source reports back to 1990 for some technical reasons, but there is nothing important about 1990 in the manufacturing trends. It just happens to be the time series easiest to look at. But manufacturing employment is clearly the lowest here since when? the 1950's. That 90K number for manufacturing employment was once nearly 400K. Ponder that some.

If there is an unreported story anywhere in the latest numbers. One of the more remarkable numbers being reported is that unemployment just for City of Pittsburgh residents is being shown at 6.2%. So not just below the rest of the county and region, but more than two percentage points below the nation. The unemployment rate in the City of Detroit for comparison is currently measured at 22.8%, not just marginally above, but several times higher than that for the nation, almost 4 times that for Pittsburgh. There is an unreported story.


Anonymous Anonymous said...

Of course, it barely bares noting, but note that three of the headlines are positive and one puts a negative spin on the report. Guess which Pittsburgh news outlet went negative with news about Pittsburgh?

Friday, June 05, 2009 8:52:00 AM  
Anonymous MH said...

The Onion is getting into the Pittsburgh/Detroit game. Sorry, but I don't know how to make something clickable.

Friday, June 05, 2009 10:38:00 AM  
Anonymous Anonymous said...

Can you explain the difference between the "Employment" figure under "Labor Force Data" and the "Total Nonfarm Employment" figure under "Nonfarm Wage and Salary Employment"? In Pittsburgh, "Employment" is greater than "Total Nonfarm Employment" by 4k for April (though 14k back in Jan.)

Now compare to Phoenix...

I was shocked to see they have an unemployment rate lower than PGH (6.8%) despite experiencing the 2nd highest percentage decline in "Nonfarm Employment" amongst the Top 40 MSAs over the past year (-6.8)... Phoenix has both an expanding labor force and has lost a ton of "Nonfarm Employment" (130,000) over the past year... how could it have such a low unemployment rate?

Then I noticed on the BLS website that there is a huge discrepancy between "Employment" (1,972,700) and "Nonfarm Employment" (1,767,800)... that's 205k! Also, the decline in "Employment" for Phoenix was only 42k year-over-year for April. What could account for such dramatic discrepancies both in terms of "Employment" and the rates of decline? I doubt "Farm" employment could number 205k... and I doubt there's a significant amount of geographic mismatch in terms of residence of employed and place of work since Phoenix is so isolated.

Another case that reverses these numbers... Charlotte

In Charlotte, the "Total Nonfarm Employment" (813k) is greater than "Employment" (755k) by 58k.

Friday, June 05, 2009 2:36:00 PM  
Blogger C. Briem said...

more going on than that even. The Nonfarm employment number is a count of jobs. The labor force emplopyment number is a count of people. roughly 5% of folks have multiple jobs so you have to take that into account. May make one of your questions more extreme. One series counts self employed the other doesn't which I am guessing may be a bigger thing in greater Phoenix. One counts certain unpaid family workers, the other doesn't. There are some farm workers in the mix. It goes on. But basically they are very different data collections. One comes from a survey of employers, the other comes from a survey of people. Makes the two numbers difficult to compare against each other.

and as you allude... being an establishment survey, one is employment by place of work. The other being employment by place of residence. Commuting is a big thing affecting the difference between the two. I dunno about whether Phoenix being isolated makes that higher or lower. Being isolated it may mean that the job center is a bigger pull on jobs from farther out. But that is just a guess.

and then I am just guessing on this for Phoenix. Would employers be less willing to disclose undocumented workers that dilligent survey workers may find more of.

Friday, June 05, 2009 3:43:00 PM  

Post a Comment

<< Home