Introduction
Education enhances employment opportunities, but not uniformly—location matters. In some areas, people without a high-school diploma have few opportunities; in others, they are just as likely to have a job as people who went to college (although their jobs and salaries will differ). If we study the spatial variation in employment rates for people with different educational levels, we discover that high employment rates for the most educated do not mean high employment rates for the least educated—a rising tide does not lift all boats. And while some places do well and some poorly, there are a few counties with very high or very low employment rates across all educational levels—these are places that are doing something very right or very wrong.
How much does employment vary?
More educated people are more likely to be employed. Nationally, only 53% of those who did not finish high school have a job, while the number for those with a bachelor's degree is 82% (these are five-year estimates for 2011—2015, for people aged 25 to 64). Yet this nationwide number conceals a lot of geographic variation. If we look at county-level data, employment rates for the least educated range from 5.8% to 100%—with a median of 48%. For high-school graduates, the median county is at 66.4%, with a similarly broad range (15.5% to 100%). Even the quartiles (top and bottom 25%) are wide: from 60.2% to 72.3% for those with a high-school education. The median continues to grow at higher educational levels, while the range gets smaller. The big picture is clear: there is a lot of spatial variation in employment rates across all educational levels.
| United States: County-level employment-population ratio, ages 25-64, by educational attainment | ||||||
|---|---|---|---|---|---|---|
| Educational attainment | Min | 1st Qu | Median | Mean | 3rd Qu | Max |
| Less than high school graduate | 5.8 | 39.1 | 48.0 | 48.3 | 57.7 | 100.0 |
| High school graduate (incl. equivalency) | 15.5 | 60.2 | 66.4 | 65.7 | 72.3 | 100.0 |
| Some college or associate's degree | 27.0 | 67.4 | 72.9 | 72.1 | 77.7 | 100.0 |
| Bachelor's degree or higher | 34.4 | 78.0 | 82.1 | 81.1 | 85.1 | 100.0 |
Source: U.S. Census Bureau, 2011-2015 American Community Survey 5-Year Estimates, S2301, Employment Status
Of course, not all counties are equal: Kalawao, Hawaii has an adult population of 57, whereas Los Angeles, California has 5.4 million people aged 16 and above. If we look at the number of people who live in counties with different employment rates, the results are broadly similar as those stated above—except in the case of the least educated: roughly 6.6 million people live in counties where employment rates are less than 50%—that's 33% of the total population without high-school diplomas. In this instance, the median quoted above (48%) understates the population-weighted average of 53%. But for all other educational levels, the differences are marginal. However one looks at it, there is clearly substantial spatial variation in employment rates.
United States: Million people by county-level employment-population ratio, ages 25-64
Source: U.S. Census Bureau, 2011-2015 American Community Survey 5-Year Estimates, S2301, Employment Status
Does opportunity cluster?
Is there a relationship between employment rates at different educational levels? Put differently: if a county has high employment rates for people with higher educational levels, is it likely to have high employment rates for people with lower educational levels? Is there a "trickle down" effect?
The short answer is: not really. There is a link between employment rates for the most educated (bachelor's degree or higher) and every other educational level: in statistical terms, we can reject the hypothesis that the coefficient is zero. But the employment rates for the most educated do not tell us much about employment for the least educated—again, in statistical terms, the r-square is low (for people with less than high school, the r-square is 0.1319). However the relationship becomes stronger as educational levels increase: for high-school graduates, the r-square is 0.2950, and for those with some college, it is 0.3625. Even so, the relationship is weak. Opportunity does not really cluster—at least not when measured in this way.
United States: County employment-population ratio, ages 25-64, vs. Bachelor's degree or higher
Source: U.S. Census Bureau, 2011-2015 American Community Survey 5-Year Estimates, S2301, Employment Status
What places do best—and worst?
Are there counties that do better or worse across all educational levels? In other words, a county might have high employment rates for the most educated but low employment rates for the least educated, or vice versa. Are there places that have high rates across all educational levels?
The answer is: yes. There are 188 countries that rank in the top quintile (top 20%) across all educational levels; and there are 152 counties that rank in the bottom quintile across all educational levels. The top performers cluster around the Midwest (North and South Dakota, Nebraska, Kansas, Minnesota, Iowa, Wisconsin, and Wyoming). The bottom performers are more scattered: California, Arizona, New Mexico, Texas, much of the south, and into Kentucky, Virginia, West Virginia, Pennsylvania and New York. Few states have both top and bottom performers.
United States: Counties with high/low employment rates across all educational levels
Source: U.S. Census Bureau, 2011-2015 American Community Survey 5-Year Estimates, S2301, Employment Status
What separates the top from the bottom performers? They vary in many ways: their racial make-up is different. The bottom performers are more rural, but the top performers are no more urban than the national average. Both the top and bottom performing counties are smaller than the average, but the bottom performers are much smaller. The bottom performers have more government employment, while the top performers have relatively more jobs in agriculture, forestry and mining. Yet these are just correlations—they do not necessarily explain why some places have better or worse outcomes. That requires a deeper investigation into state and local economies and policies. What is clear, however, is that some places are able to provide employment opportunities for people across educational levels; and some places are struggling no matter what a person's employment level. And that, on its own, is worth understanding—and studying further.
Notes
All data come from the U.S. Census Bureau, 2011-2015 American Community Survey 5-Year Estimates, Table S2301, Employment Status. For more information, please visit my github repository.