We took two of this year’s most interesting data-centric economic analyses related to equitable growth and started looking for insights that come from putting them together. Specifically, we used the data Raj Chetty, Nathaniel Hendren, Patrick Kline, and Emmanuel Saez compiled on economic mobility and the industry/trade data used by David Autor, David Dorn and Gordon Hanson in their work on the effects of the decline in manufacturing. A quick analysis of the data found some interesting relationships and we encourage graduate students to take a look at these data sets (or other data sets too) and send us any interesting stories about equitable growth that come out in at most three graphs and a few hundred words. We’ll ask the authors of the most interesting submissions to write them up to get posted on the Equitablog. Send submissions as a word document to nbunker at equitablegrowth dot org by the end of February 2014. We hope to make this a periodic feature.
Data and Methods
We used data from two sources:
1. Raj Chetty, Nathaniel Hendren, Patrick Kline, and Emmanuel Saez. The Equality of Opportunity Project. Website: http://www.equality-of-opportunity.org/index.php/data
From the Chetty et al. data, we used two measures of economic mobility:
“Expected.Child.Income.Rank.for.Parents.with.Below.Median.Incomes” is a measure of absolute economic mobility. It measures “the expected economic outcomes of children born to a family earning an income of approximately $30K (the 25th percentile of the income distribution).”
“Probability.Child.with.Parent.Income.in.the.First.Quintile.Reaches.the.Fifth.Quintle” is a measure of relative economic mobility: the probability a child born to parents in the bottom quintile of the income distribution would be in the top quintile by age 30. These were estimated by looking at children born in 1980-1981, measuring their income in 2010-2011 and assessing their parents’ household income in 1996-2000.
2. David Autor, David Dorn and Gordon Hanson. “The China Syndrome: Local Labor Market Effects of Import Competition in the United States.” American Economic Review, 103(6), 2121-2168, 2013. Website: http://www.cemfi.es/~dorn/data.htm
From the Autor et al. data, we looked at two variables: “d_pct_manuf” is the change in the percentage of the workforce in manufacturing from 1990 to 2007 and “l_shind_manuf_cbp” is the fraction of the workforce employed in manufacturing jobs in 1990.
We merged these two data sets by commuting zone. The Autor data did not include commuting zones in Alaska or Hawaii and so those were not considered. We also dropped 11 zones because of incomplete mobility information (these are primarily commuting zones with fewer than 2,500 people) leaving us with a total of 711 records.
As of December 16, 2013, you can get the data from the websites above in Stata form or you can contact me (cprice at equitablegrowth dot org) if you would like the Excel spreadsheet or R code used for this mash up.
As a quick test of how the manufacturing landscape has changed, for each commuting zone we plotted the relationship between the share of manufacturing jobs in 1990 and how that changed from 1990 to 2007, as seen in Figure 1. Most of the gains in manufacturing employment were in places where there hadn’t been much manufacturing in 1990 and most of the losses were in places with a heavy concentration of manufacturing jobs in 1990.
Figure 1: Manufacturing employment in 1990 vs. the decline of manufacturing’s share from 1990 to 2007
Next, we took a look at the income mobility measures and how they relate to the employment concentration in manufacturing (Figure 2) and to the change in manufacturing employment (Figure 3). Areas with a higher concentration of manufacturing jobs in 1990 tended to have lower economic mobility both in absolute and relative terms for the cohort born in 1980-81. Likewise, areas that saw large changes in manufacturing employment tended to have lower economic mobility. Because the share of the labor force working in manufacturing and the decrease in manufacturing’s share of the workforce are correlated, a more careful analysis will be needed to determine how these changes relate to each other.
Figure 2: Manufacturing employment vs. economic mobility
Figure 3: Change in manufacturing employment vs. economic mobility
Because these relationships were found in a quick mash-up, all of this analysis requires more scrutiny but it does raise a few questions for further investigation about the relationship between economic mobility and growth:
- Is the concentration of the workforce in one sector harmful to economic mobility—would analysis of sectors other than manufacturing lead to similar findings?
- Did the decline in manufacturing jobs cause the decrease in economic mobility?
- Should policymakers promote a diversity of industries in their economy? What are good policy responses to the changes?
- Were some responses to the decline in manufacturing more effective than others at preserving economic mobility and opportunity?
Clearly, this is an area that is rich with questions to study and problems to solve. We’d certainly like to hear your thoughts and see your submissions for this data mash-up challenge. We’re interested in analysis from either these questions or any others you come up with related to these data sets.