Research ArticleSOCIAL SCIENCES

Urban scaling and the regional divide

See allHide authors and affiliations

Science Advances  30 Jan 2019:
Vol. 5, no. 1, eaav0042
DOI: 10.1126/sciadv.aav0042
  • Fig. 1 Scaling relations of urban indicators across Sweden’s 75 labor market areas in 2012.

    (A) In line with previous research (13), we find superlinear growth in indicators of economic output such as total company turnover [blue: β = 1.196 ± 0.048 (95% confidence interval), R2 = 0.945] and the property tax collected in each labor market area (red: β = 1.131 ± 0.050, R2 = 0.977), both measured in millions of Swedish krona. Gray lines indicate proportional relations (β = 1); the colored lines show estimates of β from a linearized model (see Eq. 1 in Materials and Methods). (B) Acceleration of the pace of life is apparent from the number of residential moves (blue: β = 1.121 ± 0.022, R2 = 0.995) and the number of divorces (red: β = 1.146 ± 0.051, R2 = 0.976). (C) Cities also differ in their composition: The number of college graduates (blue: β = 1.114 ± 0.019, R2 = 0.996) and of employees in creative jobs (red: β = 1.112 ± 0.017, R2 = 0.996) likewise follow scaling relations.

  • Fig. 2 Sweden’s 75 labor market areas.

    Labor market boundaries reflect commuting patterns. We colored each labor market area according to its population size (2673 to 2.51 million inhabitants). The gray ties indicate migration flows from smaller to relatively larger labor markets, weighted by the numbers of movers in 2012. In absolute terms, most movers to denser urban environments appear within the country’s largest labor market areas, reflecting their overall population size. The inset plots yearly net-migration flows (inward movers–outbound movers) during 1990–2012 as a percentage of the local working-age population against the size of labor market areas [the blue line indicates a linear best fit (slope 0.136 ± 0.023, R2 = 0.623)]. We exclude Gällivare (18,307 inhabitants) and Kiruna (22,968), the mining areas in the far north, from our individual-level analyses, as their economies depend almost exclusively on the extraction of natural resources.

  • Fig. 3 Composition effects on the scaling of wages.

    (A) Total wages of Swedish males, measured in millions of Swedish krona, scale superlinearly across labor market areas (blue: β = 1.082 ± 0.022, R2 = 0.993). So does labor market participation, measured as the total number of employees (red: β = 1.035 ± 0.019, R2 = 0.995). We exclude the mining areas Gällivare and Kiruna [gray dots (see the Supplementary Materials for robustness analyses)]. (B) Per-capita wage (blue) also relates above proportionally to labor market size (β = 0.047 ± 0.008, R2 = 0.678), carrying the remainder of the total scaling relation (1.035 + 0.047 = 1.082). The gray line indicates a proportional per-capita relation [β = 0 (see Eq. 2)]. (C) Statistically controlling for human capital, cognitive ability, and creative job characteristics further reduces the per-capita scaling relation to β = 0.028 ± 0.009 (see Eq. 3). The vertical lines indicate 95% confidence intervals, and the dashed line stands for the per-capita scaling parameter β = 0.047 without composition controls.

  • Fig. 4 Urban wage premium approximates the upper bound of the interconnectivity effect.

    (A) Urban wage premium for movers from smaller labor market areas to the four largest in 1993–2012. The horizontal line represents movers’ counterfactual wages (at year t counted from the year of move) had they remained (see Eq. 4). Both the immediate (t = 1) and the long-term urban wage premium (t = 10) relate positively to population size and are largest for those entering the Stockholm labor market (+29.8% ± 2.1% at t = 1 and +37.2% ± 2.3% at t = 10); dashed lines indicate 95% confidence intervals. (B) There exist (72 × 73)/2 = 2628 potential combinations of origin and target labor markets in moving from a smaller to a relatively larger area. Relating the mean urban wage premium for the 100 labor market pairs with at least 200 movers to the logarithm of their difference in population size reveals a scaling relation of β = 0.050 ± 0.014, R2 = 0.351. (C) Our two complementary analyses reduce 34 to 61% of wages’ scaling parameter to interconnectivity effects (red). Most likely, interconnectivity explains about half of the scaling relation.

  • Fig. 5 The social gradient of urban scaling.

    (A) The highly educated (per-capita scaling parameter β = 0.070 ± 0.037) and those with high cognitive ability (β = 0.054 ± 0.013) benefit most from living in urban environments. We split the study population into three groups consisting of those with relatively little (<25th percentile), intermediate (25th to 75th percentile), or high (>75th percentile) education or ability, respectively. The vertical lines indicate 95% confidence intervals and the dashed line represents the net-agglomeration effect β = 0.028 ± 0.009 from Fig. 3C. (B) Long-term urban wage premium is smallest for the least-educated (+17.0% ± 2.7%) and the least-able (+25.3% ± 4.3%), who thus benefit least from moving into urban environments. The dashed line is the unconditional long-term urban wage premium averaged over the trajectories shown in Fig. 4A.

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/5/1/eaav0042/DC1

    Section S1. Full population data, metropolitan areas, and regional composition

    Section S2. Outlier analysis of urban scaling parameters

    Section S3. Replication of the scaling relation’s decomposition with U.S. data

    Section S4. Wage data and measures of individual productivity

    Section S5. Full tabulation of cross-sectional results

    Section S6. Full tabulation of the urban wage premium

    Fig. S1. Scaling relations of urban indicators excluding the three largest labor market areas.

    Fig. S2. Decomposition of the total scaling relation for wages across U.S. Metropolitan Statistical Areas.

    Fig. S3. Complementary analyses of the urban wage premium.

    Table S1. Description of Sweden’s full working-age population.

    Table S2. Creative jobs and the corresponding occupational codes.

    Table S3. Composition effects on the scaling of wage income.

    Table S4. Urban wage premium following a move from one of Sweden’s smaller labor market areas to one of the four largest.

  • Supplementary Materials

    This PDF file includes:

    • Section S1. Full population data, metropolitan areas, and regional composition
    • Section S2. Outlier analysis of urban scaling parameters
    • Section S3. Replication of the scaling relation’s decomposition with U.S. data
    • Section S4. Wage data and measures of individual productivity
    • Section S5. Full tabulation of cross-sectional results
    • Section S6. Full tabulation of the urban wage premium
    • Fig. S1. Scaling relations of urban indicators excluding the three largest labor market areas.
    • Fig. S2. Decomposition of the total scaling relation for wages across U.S. Metropolitan Statistical Areas.
    • Fig. S3. Complementary analyses of the urban wage premium.
    • Table S1. Description of Sweden’s full working-age population.
    • Table S2. Creative jobs and the corresponding occupational codes.
    • Table S3. Composition effects on the scaling of wage income.
    • Table S4. Urban wage premium following a move from one of Sweden’s smaller labor market areas to one of the four largest.

    Download PDF

    Files in this Data Supplement:

Navigate This Article