June 2022

Sun Mon Tue Wed Thu Fri Sat
      1 2 3 4
5 6 7 8 9 10 11
12 13 14 15 16 17 18
19 20 21 22 23 24 25
26 27 28 29 30    
Blog powered by Typepad

« The Good Old Days are Now, continued | Main | Quote of the Day: Milgrom and Roberts (1987) AER paper »


Feed You can follow this conversation by subscribing to the comment feed for this post.

I tried to find how many of the roughly half of the bottom quintile that moves up were students that finished school.

Do you have that info?

Spencer: I believe that the taxpayer studies excluded such people because to be in the database you had to have filed your own (or joint) taxes for all 9 years. That would exclude that scenario.

And, if I recall, that's one of the reasons the Cox and Alm data is skewed toward more mobility: folks like that were counted in some form or another.

I don't have the sources or links right now (and am not going to chasing after them), but I have seen quite a few studies recently suggesting that in contrast to the self-image held in the country, income mobility in the US has declined in recent years and is actually lower now than in many European countries, which, of course, pretty much all have more equal distributions of income as well.

Steve - that is true of Cox and Alm (or at least of the Treasury study) but recall that the static inequality numbers ALSO include those people.

Barkley - relative mobility, yes. Absolute mobility, no. (For data prior to and hence excluding the recession, of course).

Let's take this a step further with a hypothetical. Suppose a typical person in this society started in the lowest quintile and then rose up through the quintiles over his lifetime. Then the typical person would actually prefer a society with a greater income distribution, other things equal. I'd rather get {$20K, $40K, $70K, $110K, $160K} than {$20K, $40K, $60K, $80K, $100K}, for example.

Or to use the hotel analogy, if I were going to spend a year on each floor, I'd be in favor of larger differences between floors, because larger differences mean larger improvements in my lifestyle over time.

This is obviously just a hypothetical, because in reality it's not true that everyone advances through all the quintiles. But it demonstrates that, once you start thinking in term of lifespans instead of time slices, higher income inequality can actually be desirable in itself.


I am confused. So which one is which? Who has the greater or lesser relative versus absolute mobility?

Barkley - Relative mobility in the US has been falling somewhat (according to some sources) and is lower than in Europe - has been for a long time, maybe always. However, absolute mobility has not been falling and is higher than in Europe - has been for a long time, maybe always.

This is easy to explain, and should be obvious. I will give you an example of two countries to illustrate why this is. Lets call the two countries FreeMkt and WelfMkt, to make the point obvious.

FreeMkt has 5 quintiles:
$10k, $30k, $60k, $100k, $150k

WelfMkt has 5 quintiles:
$12k, $20k, $30k, $50k, $80k

Now, when a twenty-something in FreeMkt gets her first good job after college, she starts in the first quintile, earning $15k. Her counterpart in WelfMkt may start out slightly higher, given the union bargaining power, at $18k, but this is still the first quintile.

In any case, when both of them 5 years later have been promoted, the one from FreeMkt now earns $45k, which puts her in the 2nd quintile, while the one from WelfMkt earns only $35k, but this puts her in the 3rd quintile. Hence, relative mobility is higher in WelfMkt (and in Europe) but absolute mobility is higher in FreeMkt (and in the US).

Relative mobility may also be falling in the US, but is this in conjunction with higher absolute mobility?


In Intellectuals and Society, Thomas Sowell has a lengthy discussion of the issue and the literature on income distribution. He contrasts studies that follow changes in statistical categories with those using data that follow specific, real-world people.

When using tax return data tracking individuals between 1996 and 2005, one finds that the income of the bottom quintile rises much more rapidly than those in the top quintile (91% versus 10%). The income of those in the top 5% and 1% actually declined.

His explanations are entirely microeconomic and commonsensical. They rely heavily, though not exclusively, on earnings over life cycles of real people.


This looks like a bad case of small sample bias, given that the aggregate data definitely does not support that. How did Sowell select his people to track?


I wouldn't hurl accusations of bias around before reading the work in question. It is really quite outrageous.

It is not Sowell's study but that of the U.S. Department of the Treasury, "Income Mobility in the U.S. from 1996 to 2005." The study was widely reported in the media when it was released.

If you read Sowell's analysis (and he has written on the subject before), you would understand that there is no reason why the two data sources should correlate. His commentary appears on pp. 34-47 of the book.

His succinct statement of the issue (p.37) is as follows: "Internal Revenue Service data can follow particular individuals over time from their tax returns ... while data from the Census Bureau and most other sources follow what happens to stastical catgories over time, even though it is not the same individuals in the same catgories over the years."

"I believe that the taxpayer studies excluded such people because to be in the database you had to have filed your own (or joint) taxes for all 9 years. That would exclude that scenario."

How many of them graduated in the three or four years before the study begins?

And, a very interesting question: How high a percentage of the people moving up had *parents* in upper quintiles already? All around me in Brooklyn I see people in their 20s and early 30s, from middle class or higher backgrounds, spending a decade or so bartending, waitering, trying acting or music or writing -- spending all this time in the bottom quintile -- then getting married or giving up on painting and "getting a real job," and rapidly rising into the quintile their parents raised them in. These are all just delayed "adulthoods," and not real income mobility at all. (I.e., it's not ghetto or rural poor families moving up.)


That may not be "real" mobility in the sense of moving to a different quintile than their parents, but that's not what these studies were ever claiming they were measuring. There are separate studies looking at inter-generational issues. The question here was not that.

And the broader point is that normal life-cycle movement among quintiles matters because that is what shows why static comparisons of who has what percent of income or wealth can be so misleading: who comprises those quintiles changes every year for a variety of reasons, including life-cycle movement.

For me, the two key questions are:

1. How likely is it that poor folks can move up in X number of years? (The answer seems to be "fairly likely".)

2. Are the real conditions of those in the lower quintile improving over time? (The answer there is yes.)


"really outrageous"? I would say that is a bit overdoing it. Small sample bias happens a lot, and not necessarily because the person selecting the sample is doing so in a consciously biased way. Or did you not know that? For the record, I was not at all alleging any conscious bias in this matter by Sowell, and there may still be bias in the sample, even if the study was done by the US Treasury.

So, I have not read Sowell's analysis, but some points stick out. One is that when people are very poor, fairly small increases in income in absolute amounts translate into big percentage increases.

I can imagine that tracking individuals one can find quite a few in top categories who are hitting retirement, and so declining in income somewhat. It is, however, a fact that the percentage of income going to the top 1%, and even more dramatically to the top 1/10 of 1%, rose and quite substantially over the period in question. This has been the most noticeable change in the overall structure of income distribution in the US recently.

Sowell may have an explanation for this, and I am curious to see what it is. Perhaps it is a matter of more people moving down a bit, with a smaller number moving up a whole lot, thus dragging up the overall average at the top.

"That may not be "real" mobility in the sense of moving to a different quintile than their parents, but that's not what these studies were ever claiming they were measuring."

No, Steve, I'm not saying they did anything other than exactly what they said they did. But I think there is very interesting work to be done here in going further and seeing who is doing the moving.

"1. How likely is it that poor folks can move up in X number of years? (The answer seems to be "fairly likely".)"

Here's the key thing I see here: When these are "post college hiatus" folks moving up, then they were no more "poor" than someone taking a road trip between apartment rentals is "homeless." Now, I don't pretend to know in advance what the percentage of 20-somethings "chilling" is among these "mover-uppers," but it makes a big difference to answering question 1. of yours.

I hope Gene and Barkley are reading each other comments. For Barkley, the problem is people are very poor and small changes in income have a big percentage impact. For Gene, the issue is that people at the bottom aren't really poor.

This exchange actually illustrates the main point Sowell makes in his discussion. Experts cannot even agree on what the data show.

I would agree that there are disagreements among the experts on what the data mean.

The comments to this entry are closed.

Our Books