Saturday, October 11, 2008

More on the "Decline of the Middle Class" Myth

On this CD post about this Skeptical Optimist post, Bobble writes (with grammar/punctuation corrected): You might want to check Steve's latest revisions. They are inflation adjusted and the income trends look pretty flat. Real income increased from 1994-2007 at approximately 1%. He still compares the period of 1994-2007 with the period 2001- 2007. If he showed the increase from 2001 to 2007 I *think* it would be around zero. He still doesn't show the increase for the top quintile. So how can we tell how the middle class did compared to the top earners? I think this guy and his charts are busted. That you are ignoring this reflects negatively on you and your website. I hope you are more diligent in your teachings.

1. If you look at the actual Census data, you'll see that: a) the datasets for household income are available from 1994 to 2007, and b) household income is NOT reported by quintile or decile, but by income categories in increments of $2,500 UP TO $100,000: e.g. under $2,500, $2,500 to $4,999..... $97,500 to $99,999, and THEN $100,000 AND OVER.

As Steve explains, "The bottom 4 quintiles of household income are easy to analyze, but the 5th (top) quintile is not. Each quintile has the same number of households in it, but only 4 out of 5 quintiles have easy-to-calculate, weighted-average income per household, and income per earner. That's why the charts don't show the highest one."

Certainly, if Census had reported income by quintile, and Steve Conover left that quintile out of his analysis, that omission would be subject to criticism. But if you look at the data, and read Steve's explanation, you'll see that it was not possible to report the top quintile because of the way Census organized the income data.

And adding the top quintile would not change the fact that "income per earner" for all four of the bottom quintiles increased between 1994-2007. In other words, the "middle class" did not disappear and income for that group and even income for the "lower class" did NOT stagnate. And if income inequality per earner did increase during that period (which we can't tell without the top quintile), it was NOT because the income of the middle and lower income groups stagnated or declined.

2. Adjusting for inflation doesn't change the original analysis that showed "income per earner" for all four bottom quintiles increasing between 1994 and 2007, at about the same rate, but with a slightly higher rate for the bottom quintile than the other three (see top graph above). Subtracting 2.5% average annual inflation from each quintile doesn't change the facts that: a) real income per earner for each of the four groups increased between 1994 and 2007 at about the same rate (1%), except that the LOWEST QUINTILE increased at a significantly HIGHER rate of 1.5% (see bottom chart above).

Bottom Line: The value of Steve Conover's analysis is that he has converted the Census Bureau's raw data on "Household Income" from 1994-2007 to INCOME PER EARNER, BY QUINTILE, to investigate the often-reported stories about "the decline of the middle class" (305,000 Google hits), "the war on the middle class" (515,000 Google hits), etc. Whether we look at nominal income or real income, the result is the same: all income groups have experienced gains in "income per earner" from 1994-2007, and the lowest income quintile did even better than the next highest 3 quintiles. Therefore, the rich got richer, the middle class got richer, and the poor got richer.

Any analysis of household income over time will always be distorted by the facts that: a) the number of "persons per household," and b) the number of "earners per household," vary significantly by quintile, and change over time.

My own analysis showed this in a previous post, when I adjusted real median household income over time by the average number of persons per household, which has declined significantly from about 3.3 persons per household in 1967 to about 2.55 in 2007. After adjusting for household size, the real median income per person reached an all-time high in 2007, see charts below:


At 10/11/2008 11:00 AM, Anonymous Anonymous said...

Clearly real median household income per household member fell from 2000 to 2006, with a rise above the post-2000 peak in 2007.

Draw a another colored line from the 2000 peak to the 2007 peak. All the white on the chart below that new colored line and above the blue line is a loss or a fall in household member purchasing power.

At 10/11/2008 11:34 AM, Blogger OBloodyHell said...

As you note, bobbie stopped reading as soon as it violated his prejudices.

Bobbie used to post questions worth reading -- I didn't always agree, but at least the questions and observations were somewhat reasonable. Lately they're almost always based on assuming that the far left always tells the truth, and anyone who disagrees is lying and/or a shill.

At 10/11/2008 11:44 AM, Blogger Ironman said...

If it helps, here's the overall change in the income profile of U.S. individual income earners from the years 1995 to 2005, between $0 and $95,000.

Here's the post describing the method by which those income distributions shown in the link above were created. Here's the tool we created so that you can have access to that data, which breaks the data down by age group. To get the whole U.S. workforce, you would just need to add the data for the individual age groups together.

Finally, here's the final post in the series we did in examining how the distribution of income is changed by age group from 1995 to 2005, complete with links to all the other posts in that series.

Now, if you want the full distribution of household income for 2005, rather than the income distribution of individual income earners below $100,000, we can accommodate you!...

At 10/11/2008 11:53 AM, Blogger OBloodyHell said...

Doc: One thing you might want to do when including multiple charts is to add a "figure number" or something for commenters to refer to. Just a thought.



So, you figure that 2000-2006 sucked because, for six whole years, income was above all of the previous history except for about a year or maybe two that immediately preceded it? That people made, what, $3000 more than they did over Clinton's final two years, as opposed to making, oh, what, $8000 more than they did during Clinton's first six years?

Yeah, it sucks to be you.

For six whole years, about 18 months of which was in the middle of a recession the MHI/HM was below what it was for a whole year, maybe two, but above all the rest of the time in the graph.

Oh, yeah, that just SUCKED! We should never, ever let people make so much EVER again. And Obama is JUST the man to lead the charge!!

Also, mind you, note when the drop off occurred... well before 1/21/2001.

So the situation there was not Bush's to make, but Bush's to inherit, from that guy who preceded him... what was his name and party affiliation, again?

I wonder if you'll be as quick to blame the next admin for the situation they are in, or if you'll write it off as an inheritance...?

I guess that'll depend on whether or not the winner is McCain or Obama, won't it?

Dang! That argument of yours is looking positively Hindenbergesque. I hate to tell you, but I don't think it floats with holes that numerous and that large in it.

At 10/11/2008 3:22 PM, Blogger bobble said...

first of all, MP, thank you for addressing my question. if i'm wrong, that's fine. by having someone point out where i have erred, at least i've learned something.

now, here's the quote you used to start your blog entry: "Median household income since 2001 looks stagnant at best, doesn't it (see chart above)?" and there is a chart with the years 2001-2007 circled. sounds like we are talking about the years 2001-2007 doesn't it?

so, tell me again, why are we using growth rates from 1994-2007?

eyeballing the new, inflation adjusted, charts that proclaim 1% growth between 1994 and 2007, the slope of the line between 1994 and 2000 is much steeper than the slope between 2001 and 2007. this looks a lot like most of the 1% growth occurred 1994-2000. there may be growth there but its going to be far below the claimed 1%, bordering on, well, stagnation.

regarding the missing top quintile. i'm sorry that the way the census reports the data prevents him from 'easily' computing the growth rate. but, if the top quintile grew at, say, 10% (this is just for argument sake,i have no idea what the actual rate was) and the lower quintiles grew at 1%, i think you could make the argument that the lower quintiles stagnated. however, since we don't know what the top quintile growth rate was, its hard to make a statement one way or the other.

At 10/11/2008 4:28 PM, Anonymous Anonymous said...

Yeah, it sucks to be you.

No. It sucks to be you. I don't agree with data miners that use specific dates for proof of anything. If Carpe Diem could instruct a grad student to log real disposable income per household member on the charts from business cycle peak to business peak, it may be instructive.

It appears from Carpe Diem's charts that Bush41 and Bush43 were unlucky. Well, Bush43 appears to be really unlucky. Maybe it is not so if the data is logged and recession troughs and expansion peaks are noted.

Bottom line: Steve Conover of The Sceptical Optimist didn't demonstrate anything worth remembering with the earner classification.

When it comes to income, Saez& Picketty, using IRS data, are the cream of the crop. No personal offence to you, ironman.

Remember, Conover is the national debt guru whose charts were so off the map that he ceased to publish them. It is $1 trillion in additional debt so far this year and calender 2008 is not over.

At 10/11/2008 6:06 PM, Blogger the buggy professor said...

1) This post by Mark --- like his earlier one on household income per household member in a longer-time period --- seems to me to make a sound point: a chunk of the slow growth of income between 1994 and 2007 for households per quintile (for the bottom four anyway) looks differently when recalculated by the Skeptical Optimist. That is, when income earners' are plotted against each of the bottom four household quintiles.


2) The trouble is, as anonymous noted at 4.28 PM --- why are the dates 1994 to 2007 used?

These dates take two different presidential eras, put the income advances in the Clinton and Bush-W eras together, and then leave the matter that way without any clear explanation.


3 So I went to the latest work by two scholars, Saez and Picketty, and it was an updated version of their work that Saez put out in March of this year.

The dates that track average (real) income growth are for the Clinton expansion (1993-2000), the Bush expansion (2002-2006 end), and the whole period 1993-2006. And they look at three different wage increases: average, top 1%, and bottom 99%.

These dates, as you can see, don’t jumble together two different periods of growth in two different eras the way Skeptical-Optimist does (for whatever reason). The data in the latest 2008 paper only traces the data for the top 10%, top 5%, and top 1% earners in the executive summary I used --- and, in particular, for the following table to repeat, the income earnings are confined to them. So they’re not fully compatible with Skeptical-Optimists’ quintiles, but do include at least the very affluent and rich and super-rich as well as the average and the bottom 90% categories.


4) Damn! Mark's site won't allow for the use of a paragraph HTML tag, and hence no table can be used. Yikes, it won't then take a href html formatting for a click-link.

Here is the table --- which, I trust, taken directly from Saez’s paper (table 1) and converted by me from PDF to HTML, will show up properly when posted here.

If not, please paste this in your browser for the table:".

1993-2006: 1.9% 5.7% 1.1%

3.7% 10.1% 2.4%

2.8% 11.0% 0.9%

column 1 is for average real wage growth (annual)

Column 2 is for average real wage growth for top 1%

Column 3 for average real wage growth for the bottom 99%


5) The results, to repeat, are not directly comparable to the Skeptical Optimist’s work. They do support, though, anonymous’ and Bobble’s key point, I believe: average earner growth in the Bush expansion era has to be inescapably very low, if not near to stagnation.


Quite simply, because the yearly growth of average income for the top 1% --- which includes all their sources --- grew about 13 times faster between 2002 and 2006 than the growth rate for the entire bottom 99%. You can also see, then, how the growth of income for the average earner in that period is heavily lopsided because of the top 1% rate.

To put it differently, other studies --- not Saez and Picketty’s (which stops too early in the Bush period) --- show that the top 10% income-earners in the Bush era gained 90% of the fairly fast growth between 2002 and 2007. Which means that the bottom 90% got 10% of the national income growth.


6) Is all this a matter of an ideological bent by Saez and others?

Generally, economists don’t operate that way.
They may selectively choose data-sets --- especially by years --- and may selectively omit key variables to bring about desired theoretical results; but if they do, that kind of work is professionally sloppy, and it will be given the raspberry-treatment it deserves.

Michael Gordon, AKA, the buggy professor

At 10/12/2008 4:29 AM, Anonymous Anonymous said...

I would like to see data going back to the 70's. I remember back in the 70's anybody when a tradesman i.e plumber, carpenter, welder, mill-wright, electricain could buy a home and make a decent living. Today these same people are not able to afford what they could 30 years ago and more couples are working than 30 years ago.

I guess I should read a graph to tell me if my friends and family are taking more family vactions now or in the past.

Next some fool will be telling us that population density in Southern California is less now and we also have less traffic and more time to enjoy our lives.

At 10/15/2008 7:04 PM, Blogger OBloodyHell said...

> Remember, Conover is the national debt guru whose charts were so off the map that he ceased to publish them. It is $1 trillion in additional debt so far this year and calender 2008 is not over.

And this relates to the charts at issue HOW?

Do his feet smell too much for the charts to be correct?

Is his car the wrong shade of blue for the points he makes to be accurate?

I suppose we should all expect this sort of irrelevant response from your rather lame, "NaaaaUhhh, It sucks to be youuu" retort.

Hint: This isn't Middle School any more. And if it is for you, then... yeah, it shows.

/snark on

I must say, I don't know how anyone could EVER dare challenge such a brilliant debater such as yourself...

/snark off


Post a Comment

<< Home