Divided Landscapes of Economic Opportunity: The Canadian Geography of Intergenerational Income Mobility

Transcription

1 Divided Landscapes of Economic Opportunity: The Canadian Geography of Intergenerational Income Mobility DATA APPENDIX REPLICATING THE RESULTS USING ONLY THE AGE COHORTS 16 TO 19 YEARS OF AGE IN 1986 Miles Corak University of Ottawa Miles Corak, all rights reserved. June 2017 This is an appendix to the paper Divided Landscapes of Economic Opportunity: The Canadian Geography of Intergenerational Income Mobility, offering information replicating all the tables and figures in the original paper using a subset of the original sample, those 16 to 19 years of age in 1986, the year in which the Census geography is defined. This is intended to examine the robustness of the results to a data quality issue highlighted in the data appendix attached to the paper. The possibility of geographic mobility may imply that some children may not be ascribed the geography of the region where they lived as teenagers if the algorithm creating the data links them to their parents when they are older. The original paper and all supplementary information is available at MilesCorak.com/Equalityof-Opportunity. The table and figure numbers in this appendix correspond to those in the original paper, including eight additional tables offering direct comparisons of each of the following eight intergenerational mobility statistics. Income mobility Regression to the mean in parent-child incomes 1. Absolute α j least squares estimate of intercept 2. Relative β j least squares estimate of slope 3. Average income Ȳ j average permanent income of parents Rank Mobility Percentile rank-rank regression 4. Absolute a j least squares estimate of intercept 5. Relative b j least squares estimate of slope Directional mobility Cells of qunitile transition matrix 6. Rags to riches P 1,5 conditional probability of top income 7. Intergenerational low income P 1,1 conditional probability of bottom income 8. Intergenerational privilege P 5,5 conditional probability of top income 1

2 Table 2: Number of children born between 1963 and 1970: weighted and unweighted Intergenerational Income Data sample sizes compared to estimates from the 1986 Census Administrative Data 1986 Census Ratio weighted Province / Territory Weighted Unweighted Total to Census Newfoundland and Labrador 41,600 30,900 Prince Edward Island 7,800 5,850 Nova Scotia 50,050 38,250 New Brunswick 41,950 32,450 Quebec 334, ,350 Ontario 462, ,650 Manitoba 54,800 43,250 Saskatchewan 55,000 38,500 Alberta 122,750 88,600 British Columbia 136, ,600 Yukon 1, Northwest Territories, Nunavut 3,300 1,800 Canada 1,312,600 1,005,000 Note: All numbers are rounded to the nearest 50. Table 3: Intergenerational income mobility: absolute income mobility, relative income mobility, and average parental community income Absolute Relative Average Province / Territory mobility mobility income (α j ) (β j ) (Ȳj) Newfoundland and Labrador ,635 Prince Edward Island ,439 Nova Scotia ,383 New Brunswick ,482 Quebec ,309 Ontario ,178 Manitoba ,550 Saskatchewan ,116 Alberta ,137 British Columbia ,029 Yukon ,871 Northwest Territories, Nunavut ,845 Canada ,808 Note: First two columns are least squares estimates. 2

3 Table 4: Intergenerational rank mobility: absolute rank mobility, relative rank mobility, and the estimated percentile rank for a child raised by the average parents in the bottom half of the income distribution At 38 to 45 years of age At 31 and 32 years of age Province / Territory Absolute Relative Expected Absolute Relative Expected (a j ) (b j ) Rank (a j ) (b j ) Rank Newfoundland and Labrador Prince Edward Island Nova Scotia New Brunswick Quebec Ontario Manitoba Saskatchewan Alberta British Columbia Yukon Northwest Territories, Nunavut Canada Source: Least squares estimates using Statistics Canada, Intergenerational Income Data as described in text. 3

4 (a) Children of bottom one and of bottom fifth percentile parents 5 Percentage of children (%) Bottom percentile parents Fifth percentile parents (b) Children of top percentile parents 10 Percentage of children (%) Child's income rank in adulthood (percentile) Figure 1: Intergenerational Directional Mobility: Percentile ranks of children raised by bottom ranking and top ranking parents 4

5 Table 5: Intergenerational directional mobility based on selected transition probabilities: rags to riches, intergenerational cycle of poverty, and the intergenerational cycle of privilege Rags to riches Cycle of poverty Cycle of privilege Province / Territory (P 1,5 ) (P 1,1 ) (P 5,5 ) (P 3,3 ) Newfoundland and Labrador Prince Edward Island Nova Scotia New Brunswick Quebec Ontario Manitoba Saskatchewan Alberta British Columbia Yukon Northwest Territories, Nunavut Canada Source: Selected entries from the quintile transition matrix calculated using Statistics Canada, Intergenerational Income Data Rags to Riches Rags to Riches (a) Cycle of Privilege (b) Cycle of Poverty Figure 2: Rags to riches mobility across Census Divisons and its correlation with intergenerational cycles of privilege and poverty, as measured with quintile transition probabilities 5

6 70,000 Average income is higher for children 60,000 Average child adult income 50,000 40,000 30,000 20,000 10,000 Average income is higher for parents 10,000 20,000 30,000 40,000 50,000 60,000 70,000 Average parent income in each Census Division (2014 constant dollars) Figure 3: Average incomes of parents and children in each of 266 Census Divisions 6

7 60 Expected percentile rank of children Canada Parents in the top half Parents in the bottom half Parents in the bottom fifth Census Divisions ordered from lowest to highest rank mobility for bottom half children Figure 4: Expected rank outcomes for children from top half, bottom half, and bottom fifth parents by Census Division 7

8 Probability of moving to the top quintile for men and women having bottom quintile parents 0.20 or more 0.15 to to to to 0.05 less than Figure 5: Rags to riches mobility: Census Divisions classified according to the probability that children of bottom quintile parents have adult incomes in the top quintile 8

9 Table 6: Income and rank mobility statistics for the Census Divisons with a probability of moving from the bottom to the top quintile of less than five percent Number Income mobility Rank Mobility Census Division of Absolute Relative Average Absolute Relative children (α j ) (β j ) Parent Income (a j ) (b j ) Central Coast Regional District BC , Stikine Region BC , Division No. 19 MB , Kitikmeot Region NT , Brome QC , Charlevoix-Ouest QC , Charlevoix-Est QC 1, , Division No. 22 MB 1, , Frontenac QC 1, , L Islet QC 1, , Iles-de-la-Madeleine QC 1, , Huntingdon QC , Sunbury County NB 1, , Yamaska QC , Hants County NS 2, , Gaspe-Ouest QC , Division No.8 MB , Division No. 21 MB 1, , Shelburne County NS 1, , Gaspe-Est QC 2, , Kings County NS 3, , Division No.7 MB 3, , Carleton County NB 1, , Queens County NB , Division No.3 MB 2, , Division No. 23 MB , Maskinonge QC 1, , Skeena-Queen Charlotte Regional District BC 1, , Note: Census Divisions are ranked from lowest to highest transition probability conditional on being less than Numbers are rounded to nearest 50. 9

10 Table 7: Income and rank mobility statistics for the Census Divisons with a probability of moving from the bottom to the top quintile of at least twenty percent Number Income mobility Rank Mobility Census Division of Absolute Relative Average Absolute Relative children (α j ) (β j ) Parent Income (a j ) (b j ) Dufferin County ON 1, , Division No. 13 AB 3, , Division No.7 AB 2, , Division No. 16 AB 3, , Division No.2 SK 1, , Division No. 18 AB , Division No.8 AB 6, , Division No. 19 AB 3, , York Regional Municipality ON 19, , Division No. 14 MB , Halton Regional Municipality ON 15, , Division No. 14 AB , Division No.10 AB 4, , Division No. 13 SK 1, , Note: Census Divisions are ranked from lowest to highest transition probability conditional on being at least Numbers are rounded to nearest

11 Probability of staying in the bottom quintile for men and women having bottom quintile parents 0.40 or higher 0.35 to to to to 0.25 less than 0.20 Figure 6: Intergenerational cycles of low income: Census Divisions classified according to the probability that children of bottom quintile parents have adult incomes in the bottom quintile 11

12 Table 8: Income and rank mobility statistics for the Census Divisons with a 40 percent or higher chance that children of bottom quintile parents will be bottom quintile adults Number Income mobility Rank Mobility Census Division of Absolute Relative Average Absolute Relative children (α j ) (β j ) Parent Income (a j ) (b j ) Squamish-Lillooet Regional District BC , Kootenay Boundary Regional District BC 1, , Brome QC , Alberni-Clayoquot Regional District BC 1, , Annapolis County NS 1, , Territoire-du-Nouveau-Quebec QC 1, , Stikine Region BC , Sunbury County NB 1, , Haliburton County ON , Division No.8 MB , Division No.3 AB 1, , Baffin Region NT , Division No.1 MB , Keewatin Region NT , Division No. 18 MB 1, , Kitikmeot Region NT , Kitimat-Stikine Regional District BC 2, , Manitoulin District ON , Mount Waddington Regional District BC , Skeena-Queen Charlotte Regional District BC 1, , Division No. 21 MB 1, , Division No.7 MB 3, , Division No. 18 SK 1, , Central Coast Regional District BC , Kenora District ON 3, , Division No. 23 MB , Division No. 22 MB 1, , Division No. 19 MB , Note: Census Divisions are ranked from lowest to highest transition probability conditional on being at least Numbers are rounded to nearest

13 Table 9: Income and rank mobility statistics for the Census Divisons with a less than 20 percent chance that children of bottom quintile parents will be bottom quintile adults Income mobility Rank Mobility Census Division Number of Absolute Relative Average Absolute Relative children (α j ) (β j ) Parent Income (a j ) (b j ) Montmorency No. 2 QC , Division No. 13 SK 1, , Division No. 14 AB , Division No. 19 AB 3, , Division No.3 SK 1, , Division No.2 SK 1, , Division No.10 AB 4, , Division No.8 AB 6, , Note: Census Divisions are ranked from lowest to highest transition probability conditional on being less than Figure 7: Preferred Dendrogram illustrating the grouping of 266 Census Divisions into two clusters, based on Complete linkage with Euclidean-Based Distance 13

14 Figure 8: The landscape of us and them: Census Divisions clustered into two groups based upon similarities between all eight intergenerational mobility measures 14

15 Table 10: Does the Manitoba border matter? Estimates from Linear probability models of whether Census Divisions are within the province of Manitoba (1) (2) (Intercept) (0.016) (0.015) Absolute income mobility (0.437) (0.431) Relative income mobility (0.407) (0.401) Absolute rank mobility (0.107) (0.105) Relative rank mobility (0.060) (0.060) Bottom to bottom mobility (0.041) (0.040) Bottom to top mobiility (0.036) (0.036) Top to top mobility (0.023) (0.023) Parent mean income (0.019) R Adj. R Num. obs RMSE Note: Boldface indicates a p-value of less than All regressors are standardized. 15

16 Census Division Clusters (determined by Hierarchical Agglomorative Clustering) Figure 9: The divided landscape of intergenerational income mobility: four clusters of Census Divisions as determined by agglomoarative hierarchial clustering 16

17 Probability of intergenerational low income ,000 Number of Children in bottom quintile families 10, , Poverty rate in the parent's generation (Percent of population in the Census Division below the LICO) Figure 10: The relationship between the intergenerational cycle of low income and Census Division poverty rates 17

18 Table 11: Correlates of intergenerational income mobility: correlation coefficients based on Census Divisons weighted by number of children Income mobility Rank mobility Directional mobility Absolute Relative Average Absolute Relative Rags to Cycle of Cycle of Census Division Characteristic income income income rank rank riches poverty privilege (α j ) (β j ) (Ȳj) (a j ) (b j ) (P 1,5 ) (P 1,1 ) (P 5,5 ) Poverty rate (below low income cut-off) Gini coefficient (parent incomes) Parents between 25th and 75th percentiles Parent incomes at 50th to 25th percentile Parent incomes at 75th to 50th percentile Parent incomes at 90th to 50th percentile Parent incomes at 99th to 50th percentile Less than high school Lone parents Not born in Canada English mother tongue French mother tongue Parents not married Population Population 25 years and older Parents filing in french Moved to Census division Moved to Census sub division Employment in manufacturing Employment in primary sector Employment in top ten 3-digit SICs Note: Column entries are weighted correlation coefficients. Census Division characteristics are derived from administrative tax data and the 1986 Canadian Census. 18

19 Estimate using only 16 to 19 year olds in Estimate using all cohorts Figure 11: Estimates of asbsolute income mobility in each of 266 Census Divisions using all available data compared to the subset 16 to 19 years old in

20 0.8 Estimate using only 16 to 19 year olds in Estimate using all cohorts Figure 12: Estimates of relative income mobility in each of 266 Census Divisions using all available data compared to the subset 16 to 19 years old in

21 Estimate using only 16 to 19 year olds in ,000 50,000 40,000 30,000 20,000 10,000 10,000 20,000 30,000 40,000 50,000 60,000 Estimate using all cohorts Figure 13: Estimates of average parent income in each of 266 Census Divisions using all available data compared to the subset 16 to 19 years old in

22 Estimate using only 16 to 19 year olds in Estimate using all cohorts Figure 14: Estimates of absolute rank mobility in each of 266 Census Divisions using all available data compared to the subset 16 to 19 years old in

23 Estimate using only 16 to 19 year olds in Estimate using all cohorts Figure 15: Estimates of relative rank mobility in each of 266 Census Divisions using all available data compared to the subset 16 to 19 years old in

24 Estimate using only 16 to 19 year olds in Estimate using all cohorts Figure 16: Estimates of rages to riches movement in each of 266 Census Divisions using all available data compared to the subset 16 to 19 years old in

25 Estimate using only 16 to 19 year olds in Estimate using all cohorts Figure 17: Estimates of intergenerational cycles of poverty in each of 266 Census Divisions using all available data compared to the subset 16 to 19 years old in

26 0.6 Estimate using only 16 to 19 year olds in Estimate using all cohorts Figure 18: Estimates of intergenerational cycles of privilege in each of 266 Census Divisions using all available data compared to the subset 16 to 19 years old in