Human Capital Follows the Thermometer

Originally published in The New York Times

April 19, 2011
Edward Glaeser (Glimp Professor of Economics, Harvard University)

Over the last decade, population growth in the fifth of American counties where January temperature averaged above 43 degrees was over 9 percent, while the population growth in the fifth of American counties where January temperature average below 22 degrees was less than 2 percent. Population growth was over 13 percent in the fifth of counties where more than 21 percent of adults had college degrees in 2000, while growth in the least educated three-fifths of counties was below 3 percent.
The powerful pull of skills reminds us that human capital is the bedrock of local and national success. The message of the Sun Belt is more complicated. Its success tells us a bit about the pleasures of warmth, and a bit about the importance of natural resources and a bit about the impact of limited government.

http://graphics8.nytimes.com/images/2011/04/19/business/economy/19economix-population-growth-counties/19economix-population-growth-counties-blog480.jpg

Population data is from the 2000 and 2010 Census.
Skills data (share of 25+ population with a college degree) is from the 2000 Census.
January temperature comes from ICPSR (Interuniversity Consortium for Political and Social Research) Study No. 2896, "Historical, Demographic, Economic, and Social Data: The United States, 1790-2002," by Michael R. Haines, which compiles data from various Census sources over many years.
The chart shows population growth across American counties between 2000 and 2010. I have ranked counties both by average January temperature and by share of the adult population with college degrees as of the year 2000. Each point represents one-tenth of America’s counties. The blue line shows the powerful connection between skills and population growth; the red line shows the also-strong connection between January temperature and population growth. Both trends represent longstanding patterns.
Last week, I discussed a new paper of mine jointly written with Giacomo Ponzetto and Kristina Tobio, looking at population growth over the last two centuries. Our longer-run investigation focused on counties in the eastern United States, roughly bordered by the Mississippi, in order to focus on an area that was populated at the time of the Civil War. For more recent decades, we also look at metropolitan areas throughout the United States.
We cannot look at the correlation between growth and historic skills — at least as typically measured by the share of the population with college degrees — over very long time horizons, because it was only in 1940 that the Census began measuring educational attainments at the county level. We are therefore limited, like earlier researchers, to looking at 1940 education levels. This is somewhat problematic since growing areas might have attracted more educated people.
Looking only at counties in the eastern United States, we find that education as of 1940 is essentially unrelated to county growth during the 19th century. Perhaps education in 1940 is not very correlated with the relevant skill level as of 1830 or 1860 or perhaps skills just weren’t important in generating local success during the era of the railroad and the mechanical reaper.
Starting in 1900, however, skills, as of 1940, predict faster county population growth, during eight of the next 10 decades. During the decades when skills don’t predict county growth in the eastern United States — the 1970s and the 1990s — skills are still powerful predictors of metropolitan area and city population across the entire nation.
Our paper then seeks to understand why skills predict metropolitan area population growth throughout the United States since 1970. Like previous papers on this topic by Jesse Shapiro, we find that the skills-growth link occurs primarily because more educated places have become steadily more productive. In the West, there is also some evidence that suggests that skilled areas have quality-of-life amenities that people increasingly value, but that that isn’t true across the nation more generally.
The impact of sunshine, as measured by average January temperature, on population growth is more complicated. From the 1790s to the 1860s, population grew more quickly in the colder counties east of the Mississippi, which helped ensure that the Union enjoyed a healthy demographic advantage at the start of the Civil War. But after 1870, warmer states grew more quickly for four decades. Between 1870 and 1910, the population of the South increased by nearly 140 percent, while the population of the Midwest increased by 130 percent and the Northeast increased by 110 percent. The growth was particularly strong in the least dense Southern states, which may reflect the increasing spread of railroads into those areas after the Civil War.
But after 1910, the connection — within Eastern counties — between January temperature and population growth disappeared until the 1960s. During those decades, the Great Lakes areas expanded. Areas that had once been centers for shipping natural resources grew as great manufacturing hubs. Proximity to the Great Lakes predicts population growth before 1870 and then after 1910, but not between. Chicago, Cleveland, Detroit and their surrounding areas all boomed as Americans moved from farms to factories.
After 1960, trends changed again. The South came roaring back after World War II and the Great Lakes region became known as the Rust Belt. While the pre-World War II South was hardly known for being friendly to outsiders, the post-World War II South adopted right-to-work laws that helped lure manufacturing to its lower costs. The decline of Jim Crow, a victory for racial justice, also made the South more politically competitive and less frightening to outside investors.
Over the last decade, January temperature continues to predict growth throughout the entire United States. This connection reflects economic productivity, but also the ease of construction in less regulated areas. Atlanta, Dallas, Houston and Phoenix grew more than any other metropolitan areas since 2000 because they combined economic productivity with a regulatory environment that encouraged, rather than stifled, new building.
Sun and skills are not opposites. There are plenty of skilled metropolitan areas, like Atlanta and Charlotte, in the Sun Belt. But at the extreme, sunshine and skills do represent two models of American success. Colder, skilled areas, like Boston, succeed because education makes up for a difficult regulatory environment, especially toward new construction. Warmer, less skilled areas succeed because limited regulation and natural resources make up for limited human capital.
For America to be successful in the 21st century, it is going to need the power inherent in both Houston and Boston. It will need unleashed human capital, and that’s why our nation needs to invest heavily in our children and in policy reforms that will make entrepreneurship easier and less expensive.