Friend of The Agenda Scott Winship, a scholar at Brookings and a former manager of Pew’s Economic Mobility Project, kindly shared an email he sent to John Siliva and the economic team at Wells Fargo, authors of a new report on economic mobility recently excerpted by Barry Ritholtz. I actually found the authors’ conclusion very congenial:

In this paper, we examine statistics on economic mobility in America. Our findings are somewhat surprising and may cause some of the more thoughtful protesters among the 99 percent to reconsider the best ways in which to address income inequality. More than anything else, the message of our findings to policymakers is that, rather than a redistribution of wealth, what is really needed is broader access to affordable education, better essential nutrition, more stable early childhood development experiences and basic financial literacy training for all income groups.

Yet the paper relies on a PSID sample that, according to Scott, might offer a somewhat misleading portrait of U.S. economic mobility:

I received your new report through a former colleague and wanted to respond, as I think it could be valuable if one major problem is corrected (and the results do not change). As I understand it, your PSID sample includes only people who are household heads or spouses in every year from 1968 to 2009 (41 years). That means, first, that you are looking at a group who is roughly 18-45ish in 1968 (since most people who are older than that will have died by 2009). Within a period (e.g., 1968-1980) it is unsurprising that the share of income received by people who start at the top will decline—the younger people are just at the start of their careers, and more importantly many of them are students, in the late 1960s and early 1970s. That’s not a problem for comparing the 1968-80 period to the 1980-2009 period. What is a problem is that you don’t start over with people who are 18-45ish in 1980. Your sample is aged 30-57 in 1980. Therefore, you could get different mobility results for the two periods simply due to aging effects rather than to period effects. That is, there might be more mobility among older Americans than among younger ones rather than mobility having declined between the earlier period and the later one. Since by 2009 your older sample members will generally be retired, it is again unsurprising to find that the share of income held by those starting at the top has fallen. But it’s highly likely that what you’ve shown is that mobility is higher when you look at the younger half of adults than it is when you look at the older half.
To show that mobility has fallen over time, you need to compare 18-45ish people in 1968 to 18-45ish people years later, and you need to make the two periods of equal length (e.g., 1968-1980 and 1997-2009). Ideally, you’d also make the starting year for both periods at the same point in the business cycle and ditto for the end points. Try 1968-1979 and 1989-2000 for instance, or 1973-1982 and 2000-09. As it stands you are giving your customers unduly pessimistic analyses.

I share Scott’s instincts: the Wells Fargo team is asking the right questions. Assuming Scott is right, however, they need to take another pass.

It’s also interesting to note that the research arm of a major financial institution is producing work along these lines.