What Counts as Opportunity?
The geography of career opportunity cannot be read from one map or one number. This chapter separates the outcomes that are often collapsed into the word “opportunity”: jobs, labor income, disposable household income, mobility, homeownership, housing wealth, and total wealth. The distinction matters because a place can perform well on one dimension and poorly on another. The reader’s task is to match the measure to the question, then interpret the result with enough caution to avoid turning a population pattern into an individual forecast.
The first question is whether a statistic describes what a region produces, what its residents receive, or how their position changes over time. The second is whether it describes a level of opportunity or the relationship between starting resources and later outcomes. Those are different questions, and geography can answer them differently.
A Place Can Be Productive Without Making Residents Richer
GDP per capita is a measure of economic production, not the income of a typical person. It can be high where capital-intensive firms produce a great deal, where a large company’s activity is assigned to its headquarters, or where commuters generate output in one region while living in another. A resident’s disposable household income is measured after taxes and transfers and adjusted for household size. Labor income is narrower: it captures earnings from work and therefore depends directly on whether people can access employment and what kinds of jobs are available.
These measures often move together, but not reliably enough to substitute for one another. The OECD finds that regional median disposable incomes are more closely associated with regional employment rates than with GDP per capita in many countries. That pattern makes sense: employment determines how many working-age adults have access to labor income, while GDP includes forms of production that may not reach resident households in the same way. Taxes and transfers also compress differences between places.
Across the 24 OECD countries in the analysis, median disposable income in the highest-income small region is, on average, about 60% higher than in the lowest-income small region. That is a substantial difference, but it is narrower than the corresponding spread in GDP per capita. In some countries, the relationship is especially weak. A region can look prosperous in production statistics while its residents’ median income is less exceptional, particularly where commuting crosses regional boundaries or redistribution is strong.
The implication is not that GDP is useless. It can indicate the scale and productivity of the local economy. It is simply a poor answer to the question, “What can a typical resident earn and afford?” That question also requires attention to prices and services. The OECD’s income data do not fully adjust for regional cost-of-living differences, especially housing. A higher median income may buy less in a high-cost region. Nor does income alone capture the quality or accessibility of schools, care, health services, transport, or employment support.
Access to services illustrates how a region’s opportunity set can be larger or smaller than its wage statistics suggest. In the OECD’s European comparison, almost all residents of the highest-GDP-per-capita regions in France and Germany can reach a public employment service within 15 minutes by car, compared with fewer than half in the lowest-GDP-per-capita regions. This is a measure of physical travel time, not service quality, capacity, opening hours, or a causal effect on employment. It nevertheless shows why opportunity is partly infrastructural. A job-search service that exists but cannot be reached easily is not equivalent, in practice, to one that is readily available.
A second distinction concerns time. A regional income level is a snapshot; mobility asks what happens after a person starts from a given position. In the OECD analysis, working-age people near the 15th percentile of their national income distribution who remain in the same region over a five-year period advance less, stagnate more often, and remain near the bottom longer when they live in low-income regions. In Finland, people in low-income regions advanced about 10 percentiles on average, compared with 17 percentiles for counterparts in the Helsinki-Uusimaa capital region. In Belgium, 31% of low-income earners in the low-income region of La Louvière experienced no real income growth over six years, compared with 22% in the higher-income region of Roeselare.
These comparisons are more informative than a simple regional average because they hold the starting position in the income distribution roughly constant. They still do not amount to a randomized test of place. People and families differ across regions, and the countries with usable longitudinal data are a selected group. But the pattern is consistent with an important interpretation: local opportunity can compound. A place with fewer accessible jobs, weaker firm demand, or poorer services may not only produce lower income today; it may offer fewer ways to move upward tomorrow.
The U.S. Census study adds another layer by measuring mobility across generations and across several forms of economic resources. It uses linked records for more than 3.4 million parent-child pairs and estimates county-level outcomes for total income, labor income, gross housing wealth, homeownership, and gross total wealth. The study’s rank-based framework distinguishes two concepts that are easy to confuse.
Absolute, or upward, mobility asks about the average outcome of children whose parents began near the bottom of a resource distribution. It is close to the practical question, “How far can a child from a low-resource family rise?” Relative mobility asks how strongly parental position predicts the child’s position. In the study’s rank-rank model, a steeper relationship means greater intergenerational persistence and less relative mobility.
A place can perform differently on these two dimensions. Suppose children from low-resource families achieve somewhat higher outcomes in one county, but children from high-resource families rise by an even larger amount. Absolute mobility may improve while the relationship between family starting position and adult outcome remains strong. Conversely, a flatter relationship does not guarantee high outcomes if the whole distribution is shifted downward. A level and a gradient answer different questions.
The branches are related but not interchangeable. Labor income may improve without producing ownership. Housing wealth may be high because prices are high, even when the probability of becoming a homeowner is low. Total wealth also includes financial resources that do not follow the same geography as work or housing. A serious claim about opportunity must say which branch it is describing.
Why Income, Housing, and Wealth Maps Disagree
The Census paper’s county maps show a recognizable geography for income mobility: lower relative and upward mobility across much of the Southeast, and higher income mobility across the Great Plains and parts of the upper Midwest. But the wealth measures do not reproduce that pattern exactly. Wealth mobility is less geographically clustered than income mobility and shows stronger concentration by density. Many large urban areas—including New York, Los Angeles, Chicago, Boston, Washington, Miami, Seattle, Atlanta, and Denver—show greater intergenerational persistence in housing wealth than in income.
The contrast is not a statistical contradiction. Income is closely tied to work and the local labor market. Housing wealth depends on ownership, property prices, parental assistance, borrowing conditions, and the timing of purchase. A city can offer more high-paying jobs and still make entry into ownership difficult. Families who already own valuable property may accumulate wealth as prices rise, while renters and would-be first-time buyers face a higher barrier. The same regional price movement can therefore look like wealth creation for owners and blocked mobility for those outside the market.
California illustrates the distinction in the Census study: some areas combine high housing values with low absolute mobility in homeownership. In other places, including parts of Pennsylvania, western Michigan, and Georgia, homeownership mobility can look relatively strong even when income and gross-wealth mobility are weaker. A lower home price can make ownership attainable for more people while limiting the wealth return attached to the property. High housing wealth and high homeownership mobility are not the same achievement.
The cross-county correlations make the point more formally. Relative mobility in total wealth correlates only 0.44 with relative mobility in labor income, while housing wealth correlates 0.41 with labor-income mobility. These are not negligible relationships, but they are far from identity. Relying only on an income-based measure would miss substantial variation in how family resources persist through property and wealth.
Local correlates also change with the outcome. The Census regressions find that local inequality is a strong negative predictor of both income and wealth mobility, extending the familiar relationship between inequality and intergenerational persistence beyond income. But income segregation is more strongly associated with housing and total-wealth mobility than with labor-income mobility. Local labor-force participation is not significantly associated with income mobility in the reported models, yet it is positively associated with upward wealth mobility. Median home values predict lower absolute homeownership mobility while being positively associated with housing wealth and income mobility.
These patterns are useful precisely because they resist a single explanation. If an area’s home values are high, that may reflect a strong local economy, desirable amenities, limited housing supply, or some combination. The result may be positive for existing owners and negative for people trying to enter. If income segregation is related more strongly to wealth mobility than to labor-income mobility, the relevant mechanism may involve access to property, neighborhoods, family assistance, or institutions rather than only access to well-paid jobs. The data suggest these possibilities; they do not prove which one is decisive.
The evidence also has to be read at the right geographic scale. County maps can make sparsely populated rural areas occupy more visual space than densely populated urban counties. The Census authors address this by mapping population or sample-size-weighted percentiles, so the midpoint reflects the experience of the median child in the underlying data rather than the midpoint of land area. Even then, a county statistic remains an average across diverse neighborhoods and households. It cannot tell a particular worker whether a specific employer will provide learning, stability, or advancement.
The OECD faces a related problem when comparing small regions across countries. Region size and population differ, and employment by residence can diverge from production by workplace. The Census estimates add their own boundaries: they are rank-based, model-assisted in small areas, focused on gross rather than net wealth, and incomplete for assets or debts that are not visible in the records. The paper is a Center for Economic Studies working paper, and its authors explicitly state that it has not undergone the review accorded to Census Bureau publications. These limits do not erase the patterns. They define how far the patterns can be carried.
A useful reading habit follows. Before asking whether a place offers opportunity, ask: opportunity for whom, measured how, over what period, and through which resource? If the concern is finding work, local employment and labor income may be central. If the concern is long-run family security, ownership and wealth matter. If the concern is moving up from a low starting point, absolute mobility may be more relevant than a region’s median. If the concern is how strongly family background persists, relative mobility is the right lens.
This is why no single map can identify the best place for a career. The map changes when the outcome changes, and the outcome changes when the mechanism changes. The next question is practical: if different places offer different combinations of opportunity, why do people not simply move toward the most favorable combination? The answer begins with the costs and constraints of movement itself.