Posted on 11/10/2006 8:33:03 AM PST by GodGunsGuts
-which only leaves them more and more frustrated.
Listen Todd, every time you deprive them of a pain source, they need to run back to their rooms to watch reruns of "Meter Maids in Bondage" and "Leather Stewardesses".
--which flatly contradicts BEA numbers for personal income.
This Census/BEA dispute almost forces us to choose between who we're supposed to believe --is it the BEA's GDP or the Census' poverty levels?
Almost, but not quite. The Census Bur. does cite the BEA, except they don't really say where. I complained about it on one of their online questionnaires but it's not a problem for me. I'm a big kid and I can figure the numbers out myself.
What does chained 2000 dollars mean? I think I know but I want to be sure. Thanks--GGG
This is from the BEA's glossary:
Chained-dollar estimate. A measure used to approximate the chain-type index level and is calculated by taking the current-dollar level of a series in the base period and multiplying it by the change in the chain-type quantity index number for the series since the base period. Chained-dollar estimates correctly show growth rates for a series, but are not additive in periods other than the base period. Related terms: chain-type index, current-dollar estimate.
My take is that it's an inflation adjustment that recalculates every time period instead of using a constant basket of numbers over a series of years like the cpi uses.
This is great for political arguments because we can pick and choose some specific combination to "prove" that America was better or worse off which ever year we want. On the other hand, if we're trying to know what's actually going on (say, for running a business), then it means we have to work harder. My take is that those census numbers you mentioned were intended for politics, because they just don't add up. The 2005 total of all the average quintile incomes was $317 billion, but the BEA says that Americans earned two or three times that much:
Where did you get the $317 billion figure from? Table H-1 and Table H-2 list the number of households in 2005 to be 114,384,000. That works out to 22,876,800 per quintile. Table H-3 then lists the mean income of the five quintiles as $10,655, $27,357, $46,301, $72,825, and $159,583. So, by my calculation, the total 2005 income is calculated as follows:
22,876,800 * (10,655 + 27,357 + 46,301 + 72,825 + 159,583) = 22,876,800 * 316,721 = 7245.6 billion
That's actually about 3% more than the total employee compensation of $7030.3 billion and about 28% more than the wage and salary disbursements of $5664.8 billion from the BEA.
When I'm interested in actually knowing what's going on, I take the percents from the Census Br. and multiply them by one of the BEA's numbers. I still have use rubber gloves and wear a gas mask, but at least the numbers add up.
As I mentioned above, the numbers don't seem that far off. I suspect that some of the discrepency comes from the fact that they likely come from different sources. Remember how much controversy there was about the recent differences between the payroll and household job surveys? In any case, I'm a little surprised that you are complaining that the numbers "don't add up". I became involved in this discussion because the graph that you posted in post #77 contradicted all other numbers that I had seen. Despite their differences, the three graphs that I posted all show real wages to be stagnant or declining from 2000 through 2004. Yours, however, shows a rapid increase. Of course, I've seen so much bad data posted on the Internet, I never believe anything unless I can validate it against a source. At the very most, unsourced data may just motivate me to check out some credible sources. For that reason, I always include the source with my data. It not only makes it more credible and useful to others but it makes it easier for me when and if I decide to update it.
Yes, if you look at the original source at http://www.census.gov/hhes/www/income/histinc/h03ar.html, you'll see that it lists the mean household income received by the middle quintile to have been $16,428 in "current dollars" and $40,933 in "2005 dollars". That's just to say that, according to some set measure of inflation, $16,428 was worth as much in 1979 as $40,933 was worth in 2005.
Interesting chart (#2). It seems to me that the median would be a much more important number than the mean. The mean allows those at the very top (whose incomes have nearly doubled) to pull up the overall average. Whereas the median is the number right in the middle. If your chart is correct, male incomes (right smack in the middle) have stagnated for the last thirty years. And yes, you're darn right, a stat like that definitely deserves further attention.
As the very least, the chart shows that the mean and median can give very different results. This may not be that surprising in measuring incomes since they cannot fall in a perfectly normal distribution. Afterall, the minimum income cannot be less than zero but the maximum income is obviously much more than twice the mean. I agree that the median is likely the better measurement in this case, at least if you want to track the incomes of the "middle class".
That last chart is a doosy. It's probably nothing, but I find it interesting that the rate of increase of male incomes drops (not to mention the decline in weekly incomes) right after Nixon pulled us off the gold standard. Like I said, it's probably nothing, but it is something I will definitely be looking into.
I agree it's probably nothing but worth looking into. On a related topic, I recall that 1975 was the last year that the U.S. ran a trade surplus.
I am happy that we've at least got the Census Br. numbers adding up to $7 trillion now. We still want it to come out at $10.2 trillion, but we wouldn't to just fudge up the Census numbers by half -- that might work for 2005 and not for other years. Sure, what we got is good enough for say, political debates, but you and I couldn't use them in something serious like developing business decisions.
...the three graphs that I posted all show real wages to be stagnant or declining from 2000 through 2004. Yours, however, shows a rapid increase.... ... I never believe anything unless I can validate it against a source
You and I have different styles; they're both good, but different. I see your posts plodding along with enormous detail which I find excessive and hard to follow. Evidently my posts are so skimpy that you've not been able to find any meat. I'll expound a bit, and please let me know if you still miss something. In post 183 there's a link with the words "BEA numbers" which leads to a BEA site with "Table 2.6. Personal Income and Its Disposition, Monthly." In line 31 of that Table are some of the numbers that I plotted into the .gif file in my post. To get the rest of the numbers, you can change the input in the box "First Year" and then refresh the table.
Cheers!
Why do we "want it to come out at $10.2 trillion"? This seems to assume that the bigger number is the better number. Do you have any evidence that the BEA numbers are "better" than the Census Bureau numbers? Or are you willing, like I am, to look at them both?
You and I have different styles; they're both good, but different. I see your posts plodding along with enormous detail which I find excessive and hard to follow. Evidently my posts are so skimpy that you've not been able to find any meat.
I'm not bothered by "skimpy" so much as I am with "incorrect". As I said before, I became involved in this discussion in response to the apparently incorrect graph that you posted in post #77. I'm glad to see that you've removed it now. In any case, I'll concede that your data is generally good and that you do make some good points. However, you also seem to jump to conclusions in some areas such as the idea that I am one of those "doom-and-gloomers". I simply believe that discussing possible problems is the best way to insure that they are eventually addressed.
I'll expound a bit, and please let me know if you still miss something. In post 183 there's a link with the words "BEA numbers" which leads to a BEA site with "Table 2.6. Personal Income and Its Disposition, Monthly." In line 31 of that Table are some of the numbers that I plotted into the .gif file in my post. To get the rest of the numbers, you can change the input in the box "First Year" and then refresh the table.
Following is a graph showing the numbers that you plotted along with employee compensation and wage and salary disbursements:
The numbers come from your same BEA source. The per capita figures in chained 2000 dollars for compensation and wages came from adjusting them by the same deflators and population figures that the personal income figures were adjusted by. As you can see, the blue line looks the same as the one in your graph. However, wages (the green line), showed a decline from 2000 to 2003 from which they just recently recovered. If we wish to estimate the condition of the so-called "middle-class", I think that we need to concentrate on wages as that is where the vast majority of their income comes from. Alternately, we can look at all income but look at the median income. Looking at the average personal income, however, tells us next to nothing about the personal income of those in the "middle class".
Is there a chance that you could share the point of your post 189 in say, 25 words or less?
In estimating the condition of the middle class, the median personal income and average wage are better measures than the average personal income.
We got sidetracked when GGG mentioned male income because your chart made his look bad.
We got sidetracked when GGG mentioned male income because your chart made his look bad.
I likewise noticed the discrepancy and I think that I finally figured it out. After not finding any reference to the source on the graph (b-33 2/06) on the Census Bureau site, it occurred to me to check the 2006 Economic Report of the President which came out in February. It turns out that table B-33 in that document contains the median income data that Hodges plotted back to 1991. That data exactly matches the data that I plotted from Table P 36 on the Census Bureau site. However, I believe that he got the data from 1977 to 1991 (or possibly 1995) from the 1997 Economic Report of the President and adjusted it to sync up with the beginning of the 2006 ERP data (1991) or the end of the 1997 ERP data (1995). Those two possibilities are shown in the purple and red lines in the following graph:
As can be seen, the red line comes pretty close to matching Hodges graph. However, I was initially stumped on how the estimates of male median income could change so much between the 1997 and 2006 ERPs. As it turns out, I believe that the difference is due to a change in the index used. The 1997 ERP states that it uses the CPI-U-X1 and the 2006 ERP states that it uses the CPI-U-RS. I then went back and found that, if you use the CPI-U before 1995 instead of the CPI-U-RS, you end up almost exactly with the red line. Yes, yes, I know that I'm once again "plodding along with enormous detail"! The point is that the use of a slightly different index can make a significant difference over time. However, without any evidence that the previous index was better, I would have to stick with the data as shown in my graph in post 174.
Very impressive detective work! Thank you.
Guns is right, everyone agrees that this was "impressive detective work!"
I was tweaking a bit, and it seems that full-time-year-round workers are 2/3 of all employed ([60,095+42,333] / 138,472 ), just as total quintile incomes added up to 2/3 of total income. The thought of two thirds of two thirds left me uncomfortable because my forecasting has been coming out a lot better by using solid estimates. Like, for mid-fifth incomes I've had to use the product of total personal income times the Census bur. mid-quintile share, but that kind of effort might not be needed in all situations..
Guns is right, everyone agrees that this was "impressive detective work!"
Thanks for the comments and applause! I was afraid that spotting a source by the name of one of its tables might seem a little too wonkish. Have a Happy Thanksgiving!
Having quoted you, I meant to include you as a recipient on the previous post. Anyhow, have a Happy Thanksgiving!
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.
