8. Do other states exhibit this puzzle?

North Carolina followed two paths with regard to per-capita income and per-capita gross state product (GSP) after 1960. The first path was one beginning with low average per capita income and with convergence to the US average by 1997. The second path was one beginning in 1997 and diverging from the US average towards relatively lower per capita GSP in the period until 2019.

This raises an important question. How common is this two-part development path among the states of the US? I address this using data on real gross state product (GSP) from the Bureau of Economic Analysis for the period 1977–2019. This measures the value of total production in each state and serves well as a proxy for the total income of the residents. The US Census provides annual estimates of population by state for the same period. The ratio of real GSP to population yields our measure of per capita income by state. We can use the same sources to define the US average per capita real GSP for each year.

I calculate the cumulative growth rate in per-capita real GSP for each state for two periods: 1977–1997 and 1998–2019. I represent the results in Figure 17.

The horizontal axis measures the growth rate for the earlier period while the vertical axis measures the growth rate for the more recent period. The state observations are marked by the state acronym. The US average is indicated by a red US, while the North Carolina average is indicated by the green NC.

The steepest gold line on the graph is drawn through combinations of growth rates that are equal in the two periods: states illustrated as above this gold line (e.g., California,North Dakota or Washington state) had more rapid growth in per capita income in the 1998–2019 period than in the 1977–1997 period. The next gold line illustrates growth-rate combinations with cumulative growth in 1977–1997 twice the rate observed in 1998–2019: Vermont, for example, illustrates this. The flattest gold line illustrates growth rates in 1977–1997 that are four times as high as in 1998–2019. North Carolina experienced growth of this sort: 63 percent vs. 16 percent. The US as a whole experienced 48 percent growth in real per-capita GSP during 1977–1997, but only 33 percent growth during 1998–2019.

If we consider those states with real per capita GSP growth rates higher than that of the US average in 1977–1997 and lower than the US average in 1998–2019, we have the 20 states in the lower right corner of the chart.

— States that grew twice as fast in the earlier period include the manufacturing states of the Midwest (Wisconsin, Minnesota, Ohio, Illinois, Indiana) and two southern states (Virginia, Alabama).

— North Carolina is among the states that grew four times as fast in the earlier period. Also in that group is a collection of southern states (Mississippi, Kentucky, Tennessee, Georgia, South Carolina and North Carolina) as well as New Jersey and Connecticut.

While the graph provides a nice illustration, it is based on evidence only from the beginning and ending years of each period. Regression analysis provides a more complete picture. I specify a simple model of the evolution of per-capita real GSP over time and estimate it separately for each state over the period 1977–1997. I then use those results to predict what real per-capita GSP would be in each state for each year from 1998 to 2019 inclusive.

For the puzzle to hold, a state should have at least 90 percent of its predictions be higher than the actual values: in at least 20 of those years, the predicted value should be higher than the value actually observed. The estimation results and a more technical write-up are available on request. Of the 50 states, only five satisfy this quantification of the puzzle: North Carolina, South Carolina, Alabama, Missouri and New Jersey.