Tuesday, January 22, 2008
The General Importance Of Logic
The study was on Danes, and it claims that for each activity level, not drinking puts you at a higher risk of heart disease. Okay, obviously, people who have problems with alcohol are way more likely to live longer by not drinking. But in that case, you'd better exercise moderatey, ESPECIALLY IF YOU ARE A WOMAN. Here's two tables from this thing:
|Outcome||Inactive||Low activity||Moderate/high activity|
|Fatal IHD (men)||1.0||0.67||0.71|
|Fatal IHD (women)||1.0||0.76||0.72|
|All-cause mortality (men)||1.0||0.79||0.73|
|All-cause mortality (women)||1.0||0.74||0.75|
|Outcome||No alcohol||Moderate alcohol||High alcohol|
|Fatal IHD (men)||1.0||0.77||0.79|
|Fatal IHD (women)||1.0||0.72||0.65|
|All-cause mortality (men)||1.0||0.87||1.05|
|All-cause mortality (women)||1.0||0.93||1.03|
I'm saving this down for use with Baptists. I like Baptists. Most of them I have known have been fine people with good senses of humor, and surprisingly tolerant in daily life, but some Baptist ministers have a poor sense of logic. This is an exact argument I heard from one Baptist:
- Scripture is inerrant and .literally authoritative.
- Jesus was God. He could not sin, or aid sin. He came to deliver man from sin.
- Drinking is sinful.
- If Jesus had turned water into wine at the Cana wedding, he would have been aiding sin.
- Therefore, Jesus did not turn water into wine at the Cana wedding. He turned it into GRAPE JUICE. There's just a mistake in the translation when it says wine. They meant new wine, or grape juice.
Now let's talk inerrancy and BLS, which publishes employment statistics. There are those who insist upon the inerrancy of BLS, but I have a wee problem with that.
The reason is FUT or FUTA. FUT= Federal Unemployment Tax (IRS 940) which applies on the first $7,000 paid in wages to each employee in the calendar year. You have to pay FUT on any employee to whom you pay at least $1,500 in a quarter. Therefore total FUT receipts to the Treasury are correlated largely with numbers of employees, unlike income and employment withholding, which is highly correlated with wage levels. Governments do not pay FUT. Farmers do pay FUT, construction firms would pay FUT, and so would most seasonal employers.
There should be a good correlation between FUT and BLS private employment figures. After all, no one's going to start feeling generous and send in extra FUT to help the government out. Nor is this tax one it pays to cheat on. Not only that, but employers get credits for state UI paid and also for a good state experience rating, so this is quite a stable tax except for actual employment changes. FUT is, however, very sensitive to changes in employment. It can rise sharply when temporary employment is rising at the beginning of an up cycle (because if an employee is moving from temp job to temp job, he may be counted several times compared to a permanent employee), and it will show the first drops in employment. You often cut temps and seasonals in downturns first, so the change can be amplified.
FUT taxes can be obtained from Treasury Daily Statements on a timely basis. I do not actually pull statements about the loss of so and so many jobs out of my butt, or from staring at a crystal ball. A few years ago I became concerned about the reality that rapidly increasing employment of illegal aliens was totally FUBARing BLS data, so I compiled FUT and other data from the Treasury Daily Statements and now I use that instead of BLS data. Here are some graphs for you:
As you can see, there is a pretty definite change from the end of 2005 through the end of 2007. Here is a second graph which correlates BLS total private employment for December of each year with FUT:
According to BLS, total private employment is still rising. It's just rising more slowly.
Now if anyone can convince me that total employment is still rising when FUT has dropped with such an amplitude (without using Baptist logic, i.e., BLS must be correct, therefore FUT is wrong), you would indeed be a miracle worker. Have at it, if you dare.
In this graph you see the answer to the puzzle of CR of Calculated Risk's missing construction employment losses. BLS is not picking up the recent immigrants who concentrated in construction in the household survey. They didn't have them at any time in the household survey, but they did pick quite a few of them up in the establishment survey. The other thing wildly off is the Birth/Death adjustment. The reason I picked figures at this time of year is that it minimizes the total construction impact, and it does appear that BLS is off by more than just construction jobs. The other problem with BLS data is probably cell phones. Cell phone usage is concentrated among the young and the recent immigrant population, which is also the group most likely to work temporary jobs.
The next BLS B/D adjustment will be based on an analysis of data from state unemployment tax filings through March 2007. Therefore it isn't going to pick up much of the 2007 trend. We probably will not get an accurate idea of 2007 employment until 2009, which should give everyone pause to ponder.
In fact NBER does not always rely on current GDP estimates, as you can see in this 2003 explanation.
From the Fed's POV, they'd be better off paying attention to significant FUT changes than GDP than BLS releases. The Fed cannot afford to wait 2 years or so for "real" employment and GDP figures to inject stimulus to correct dropping growth patterns.
You make such a great case for using FUT, it makes me wonder -- what's the catch? There must be some reason economists don't use this data. Maybe its just hard to compile (sounds like it from your post). I guess economists are lazy, at the end of the day. Anyway, thanks -- a great piece of analysis!
David - FUT is really a hassle to deal with, but there's a lot of potential there too. It is an excellent lens with which to examine some parts of employment activity on a very timely basis. I actually compiled 3 different sets of data to check myself, since FUT is reported by day and there are bulges and dips based on calendars and tax filing dates. The same is really true of all tax receipt data. Comparisons are best done YoY.
The correlations between FUT and trucking/rail activity over the last couple of years are quite strong. When I examine data from a number of differen sources, it appears that BLS is the outlier.
When BLS revises its benchmarks, it revises 21 months of data backwards. When it revises methodology, it revises five years. Economists should factor in that reality. Neither Census' housing data nor BLS employment data is published as a black box - it's just that many analysts treat it that way.
Here's Internet Monk's coverage, as of last November. (It also has one of the longest comment threads on his site, as the Battle of the Booze spilled over into his blog.)
People are entitled to their beliefs, but you'd have to figure that the cognitive gap between mankind and God is bigger than the gap between, say, housecat and human. Any time I get the notion to tell someone else about what they should believe I end up thinking about the quality of my cat's knowledge, and button up.
As for housecats vs God, I think that's a great metaphor, especially because housecats can be very arrogant!
Links to this post: