Working Paper In Progress

Where Art Lives

Density, Income, and the Geography of Arts Infrastructure in America

March 2026 Working draft Joshua Baek
Institutions
60,643
Arts organizations linked to Census data
Geography
33,120
ZCTAs nationwide with demographic coverage
Arts deserts
938
ZCTAs significantly below predicted institution counts
Population
24M
People living in arts desert communities

Abstract

Where do America’s arts institutions concentrate, and what determines which communities have them? We link 60,643 arts organizations identified in IRS filings to Census American Community Survey data at the ZIP Code Tabulation Area (ZCTA) level and find that the conventional framing — wealthy areas attract the arts — obscures a more fundamental pattern. Population density, not household income, is the primary predictor of arts institution density, explaining 66% of variation in a log-linear model compared with income’s modest incremental contribution.

The urbanicity gradient is steep: metro core ZCTAs average 61 percentage points more arts institutions per capita than rural areas, while a one-standard-deviation increase in median household income predicts only an 11% increase in institution count after controlling for density. The relationship between income and arts infrastructure is heterogeneous across art forms: museums exhibit a negative income elasticity of −0.77 after controlling for urbanicity, while media arts organizations cluster in higher-income areas. When we turn from institutional presence to funding flows, income reasserts itself: the highest-income quintile receives 1.32 times its proportional share of grant dollars, and an Oaxaca-Blinder decomposition attributes 148% of the funding gap to unexplained location premiums. We identify 938 arts desert ZCTAs — not rural, but dense suburban communities where population growth has outpaced cultural infrastructure. A difference-in-differences analysis of the 2022 e-filing mandate confirms the visibility shock was geographically neutral.

These findings reframe the geography of cultural access: density creates the infrastructure, funding networks concentrate its benefits toward wealth, and the binding constraint on cultural participation is not income but proximity to the institutional apparatus of legitimation.

Key Findings

Finding 1
Density dominates income

Population density, not household income, is the primary predictor of arts institution density. Urbanicity explains 66% of variation in a log-linear model. Metro core ZCTAs average 61 percentage points more arts institutions per capita than rural areas. A wealthy exurb at the 90th percentile of income has fewer arts institutions per capita than a moderate-income urban neighborhood. Income operates at the margin; density is the binding constraint.

Finding 2
Not all art concentrates alike

The aggregate density gradient masks striking heterogeneity across art forms. Museums exhibit a negative income elasticity of −0.77 after controlling for urbanicity — they concentrate in lower-income urban cores, anchored to historic cultural districts. Media arts organizations show a positive elasticity of +0.25, clustering where creative professionals live. The z-test comparing these coefficients is highly significant (z = 11.43, p < 0.001). Any unitary account of “arts access” obscures meaningful structural variation.

Finding 3
Funding amplifies geography

While density determines where arts institutions are, income determines how much money they receive. The highest-income quintile of arts-bearing ZCTAs receives 1.32 times its proportional share of grant dollars. An Oaxaca-Blinder decomposition attributes 148% of the Q5–Q1 funding gap to unexplained location premiums — the network amplifies geographic advantage beyond what institutional composition can explain. Middle-income ZCTAs (Q3) are the most underfunded at 0.79x, falling through both targeted philanthropy and proximity-based giving.

Finding 4
Arts deserts are suburban

We identify 938 arts desert ZCTAs where observed institution counts fall significantly below population-predicted levels. These deserts are not rural — they are dense suburban communities: Houston exurbs, Southern California inland cities, Bay Area suburbs. ZCTA 77449 (Katy, Texas) has 122,000 people and one arts organization; it should have nearly six. The 938 desert ZCTAs are home to 24 million people, a population equivalent to the ten largest American cities combined.

Finding 5
The mandate changed nothing

The 2022 federal e-filing mandate doubled the number of visible grant-making foundations. A Kolmogorov-Smirnov test shows the income distributions of entrant and incumbent recipient ZCTAs are nearly identical (D = 0.024). The difference-in-differences specification shows only marginal shifts: the largest is a 1.8pp decline in Q1’s funding share. The geographic distribution of arts funding is not an artifact of selective observation; it is a structural feature of the system.

Finding 6
Network dynamics are geographically patterned

Preferential attachment is sharply concentrated by geography: the correlation between current in-degree and new edge acquisition is 0.87 in the wealthiest communities (Q5) but only 0.36 in Q3. The “rich get richer” dynamic is nearly 2.5 times stronger in high-income areas. Edge persistence, by contrast, is stable across quintiles (57.6%–60.4%). The network amplifies geographic inequality not through differential persistence but through differential attachment.

Figures

The Density Gradient

Binned scatter plot showing arts institution density by income quintile and urbanicity tier. Metro core ZCTAs dominate at every income level.
Figure 1. Income-density gradient. Binned scatter of arts institution density (per 10,000 population) against median household income, stratified by urbanicity tier. The vertical spread between tiers dwarfs the income gradient within any tier. Metro core ZCTAs average 61pp more institutions per capita than rural areas at equivalent income levels.
Partial regression plot showing the modest residual effect of income after controlling for density.
Figure 2. Partial regression plot. After partialling out the effect of population density, education, age, and region, income retains a positive but modest relationship with arts institution count. A one-log-unit increase in income predicts an 11% increase in institution count.

Art Form Stratification

Small multiples showing income elasticities by art form. Museums negative, media arts positive.
Figure 3. Type stratification (small multiples). Income elasticities of institutional density by NTEE art form category. Museums (β = −0.77) concentrate in lower-income urban cores; media arts (+0.25) and general arts (+0.22) cluster in higher-income areas. The heterogeneity is significant: z = 11.43 comparing museum and media arts coefficients.

Funding Amplification

Bar chart showing funding-to-institution ratio by income quintile. Q5 receives 1.32x its proportional share.
Figure 4. Funding amplification by income quintile. The ratio of funding share to institution share for each ZCTA income quintile. Values above 1.0 indicate disproportionate funding. The sharpest discontinuity is between Q3 (0.79x) and Q4 (1.13x), a 34pp jump suggesting a threshold in which above-median income unlocks a funding premium.
Oaxaca-Blinder decomposition showing 148% of the Q5-Q1 gap attributed to unexplained location premiums.
Figure 5. Oaxaca-Blinder decomposition. Decomposition of the Q5–Q1 funding gap into explained (institutional composition) and unexplained (location premium) components. The unexplained share exceeds 100% because the explained component is negative: if institutions were identical in type and number, the funding gap would be larger.

Arts Deserts

Map of 938 arts desert ZCTAs, concentrated in suburban rings of Houston, Los Angeles, and the Bay Area.
Figure 6. Arts desert map. Geographic distribution of 938 ZCTAs classified as arts deserts (observed institutions < half predicted, p < 0.05). Deserts cluster in fast-growing suburban rings — not in rural America. Major concentrations appear in the Houston exurbs, Southern California’s Inland Empire, and Bay Area suburbs.
Comparison of demographic characteristics between arts desert and non-desert ZCTAs.
Figure 7. Desert vs. non-desert characteristics. Arts desert ZCTAs are populous, suburban, and demographically distinct: lower educational attainment (28% vs. 34% bachelor’s), higher Hispanic population share, newer housing stock, and longer commute times. They are communities where population growth outpaced institutional development.

E-Filing Mandate

Income distribution of mandate entrant vs. incumbent recipient ZCTAs, showing near-complete overlap.
Figure 8. Mandate entrants income distribution. The income distribution of ZCTAs hosting newly visible grant recipients (entrants) versus incumbents. The Kolmogorov-Smirnov statistic D = 0.024 confirms near-identical distributions. The mandate revealed thousands of previously invisible relationships, but they mirrored the geographic patterns of the incumbent network.
Difference-in-differences showing marginal funding share shifts by income quintile after the e-filing mandate.
Figure 9. Mandate difference-in-differences. Post-mandate shifts in funding shares by ZCTA income quintile. The largest change is a 1.8pp decline in Q1’s share and a 1.4pp increase in Q4’s share — substantively negligible. The geographic distribution of arts funding is structural, not an artifact of selective observation.

Network–Geography Interaction

Edge persistence rates by ZCTA income quintile, showing remarkable stability across quintiles.
Figure 10. Edge persistence by income quintile. Funding relationships persist at roughly the same rate regardless of community income (57.6%–60.4% across quintiles). This null finding is itself substantive: the temporal stability of the network does not discriminate by income geography.
Funding Gini coefficient and total funding by income quintile, showing higher concentration in wealthier areas.
Figure 11. Funding concentration by income quintile. The Gini coefficient of funding received rises from 0.913 (Q1) to 0.946 (Q4). Total funding flowing to Q5 ($3.65B) exceeds Q1 ($2.46B) by 48%, despite nearly identical numbers of recipient institutions per quintile. Preferential attachment correlation ranges from 0.36 (Q3) to 0.87 (Q5).

Robustness

Forest plot showing robustness of key coefficients across alternative specifications.
Figure 12. Robustness forest plot. Key coefficients across alternative specifications: OLS vs. Poisson, alternative urbanicity thresholds, with and without education controls, and state vs. region fixed effects. The qualitative ordering is preserved across all specifications, though effect sizes vary modestly with the choice of urbanicity cutpoints.

Interactive Companion

Arts Infrastructure Map

Explore the geographic distribution of arts institutions interactively. The map links every arts organization in our dataset to its ZCTA, income quintile, and art form category — allowing you to see the density gradient, funding amplification, and arts deserts at the community level.

Open the Arts Infrastructure Map →

Methodology Note

This analysis links three data sources. Arts institution data comes from the IRS Business Master File (NTEE major group A) supplemented by the grant network entity resolution pipeline from The Legitimation Economy, yielding 60,643 unique arts organizations. Census ACS 2022 5-Year Estimates provide median household income, population, population density, educational attainment, race/ethnicity, and median age at the ZCTA level for 33,120 ZCTAs. Grant flow data comes from 384,543 IRS Form 990-PF filings across 2019–2024.

Six analytical approaches are employed: (1) OLS and Poisson regression of institution density on income and urbanicity; (2) art form-specific Poisson regressions for income elasticity heterogeneity; (3) Oaxaca-Blinder decomposition of funding gaps by income quintile; (4) Poisson-based arts desert identification (observed < 0.5 × predicted, p < 0.05); (5) Kolmogorov-Smirnov and difference-in-differences tests of the 2022 e-filing mandate; (6) network parameter estimation (edge persistence, preferential attachment, Gini) by ZCTA income quintile.

Key limitations include the ZIP-to-ZCTA crosswalk imprecision (5–8% of rural organizations potentially misassigned), a 44% join rate between arts organizations and Census data, and 12% classification noise in keyword-matched art form assignments. Full methodological details are in the paper draft.

Connection to The Legitimation Economy

This paper and its companion describe the same system from two vantage points. The Legitimation Economy documented who gets funded — the network of grants with its preferential attachment, edge persistence, and geographic modularity. This paper asks where they are. The two converge on a single conclusion: the selection apparatus that governs American arts philanthropy is spatially concentrated, institutionally persistent, and self-reinforcing. The legitimation economy has a zip code.