Earlier this week I did a post that criticized reporters for unquestioningly accepting the findings of a report from the Pew Research Center that purported to a show a growing gap in wealth between people over age 65 and people under age 35. I argued that this report misrepresented this gap and gave numbers on the change in wealth by age cohort that did not show as marked a gap as the Pew numbers.

It turns out that the numbers I gave in that post were incorrect. I had used the Federal Reserve Board's 1983 and 2009 Surveys of Consumer Finance (SCF) as the basis for my calculations. Paul Taylor, one of the authors of the study, pointed out to me that the 1983 data had been subsequently revised. The revised data leads to a lower increase in median wealth for several  age cohorts. Here is the rate of growth in wealth by age cohort using the revised data:

   Median Net Worth   
   (thousands of 2009 dollars)  
  1983 2009 Percent
Age of head (2007)     Change
Under 35 14.2 9.0 -36.6%
35–44 83.2 69.4 -16.6%
45–54 115.9 150.4 29.8%
55–64 141.0 222.3 57.7%
65–74 127.4 205.5 61.3%
75 or more 83.2 191.0 129.6%

Source: Survey of Consumer Finance, 1983 and 2009.

This gives us a different picture than the numbers I had in the earlier post, although it still gives a different picture than the Pew study. (The Pew analysis used a different survey, the Survey of Program and Participation.) The SCF data show the 55-64 cohort faring almost as well as the 65-74 cohort, whereas the Pew study showed them with just a 10 percent gain. The 45-54 cohort shows a gain of 29.8 percent in the SCF data whereas the Pew analysis showed them with a drop in real wealth of 10 percent.

I will also point out three of the points that I raised in objecting to the sort of comparison of wealth growth over time in the Pew report. First, this analysis takes no account of defined benefit pensions. It is likely that the median older household would have had at least some income from a defined benefit pension in 1983. This is becoming increasing rare now. This means that most of these households will have only their income from Social Security, and whatever income they can derive from their wealth to support them in retirement. (The discounted value of a defined benefit pension of $10,000 a year over 20 years of retirement is roughly $150,000.) 

The price of the median home is currently around $170,000. This means that the $205,500 held by the median household headed by someone between ages 65-74 would be enough to pay off the mortgage on the median home (remember, these numbers include home equity) and leave about $35,000 to supplement the household's Social Security income (@$1,300 a month) throughout their retirement. 

The second problem is that the under 35 group includes many people who are still in college. The rise in college enrollment over the last quarter century would almost certainly have the effect of pushing the wealth for this group downward. People in college will generally not be accumulating wealth; in fact they are likely to be accumulating debt. Still, a 28 year-old with $15,000 in debt and a college degree is almost certainly better off than a 28 year-old with $15,000 in the bank and just a high school degree. In other words wealth is not an especially good measure of living standards or well-being for the youngest age group.

Finally, I objected to the highlighting in the report of the ratios of wealth of the oldest cohorts to the youngest. This can create a misleading impression, since the young have so little wealth. (The story was more dramatic with Pew's data since it showed a substantial decline in the wealth of the young.)

When the denominator is small it is easy to have a large percent changes. For example, if a country's inflation rate goes from 0.5 percent to 1.0 percent, we can say that its inflation rate has doubled. However, it would be wrong to imply that this is somehow of greater concern than a rise in the inflation rate from 3.0 percent to 5.0 percent, even though the latter is just a 67 percent increase. 

The basic story is that young people rarely have a meaningful amount of wealth. Their well-being is going to be far more dependent on their employment and earnings prospects than the amount of wealth that they have at age 30. In the current economy the latter don't look especially good, but the wealth measure just is not giving us much information.

Finally, I should apologize to Paul Taylor and his co-authors. I think the study is seriously flawed for the reasons listed above and others. However, I should have given them credit for carrying through their research in good faith. I do not know that their intentions were to promote the idea of a generational war, even if others are using this research for that end.


Note: A post earlier this afternoon had incorrectly adjusted for inflation. Paul Taylor called this to my attention.