Economic Insights
Doing as Well as One’s Parents? Tracking Recent Changes in Absolute Income Mobility in Canada

View the most recent version.

Archived Content

Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.

Archived

This page has been archived on the Web.

by Yuri Ostrovsky
Social Analysis and Modelling Division, Statistics Canada

Release date: May 23, 2017 Correction date: October 5, 2017

More information PDF version

Skip to text

Text begins

Start of text box

End of text box

Introduction

This Economic Insights article examines recent changes in intergenerational income mobility in Canada, focusing specifically on changes in the rate of absolute income mobility, i.e., the share of young Canadians whose family income was at least as high as that of their parents when they were the same age. The analysis focuses mainly on children who were born from 1970 to 1984 and compares their family income at age 30 (observed from 2000 to 2014) with that of their parents when they were the same age. It uses a unique Canadian database that directly links children and parents, providing information on their family incomes over almost four decades. The analysis is repeated at age 40 for earlier cohorts of children and parents.

A recent New York Times article (New York Times 2016) drew attention to a study of absolute income mobility in the United States (Chetty et al. 2016). According to the study, during the latter half of the twentieth century a declining share of 30-year-old Americans had a family income that was at least as high as that of their parents when they were the same age. Chetty et al. estimate that at age 30, 90% of Americans who were born in 1940 had incomes that matched or exceeded those of their parents when they were 30 years old, while this was the case for just 50% of those born in 1980. Based on this trend in intergenerational income mobility, the authors conclude that generational prospects have declined markedly in the United States.

Although this Canadian study also analyzes rates of absolute income mobility, the results cannot be compared directly with those of the U.S. study because of important data and methodological differences.^Note 1 Nevertheless, the attention drawn to Chetty et al. underscores the importance of the issue, and some of their findings provide a partial reference point for this analysis. In particular, in contrast to the significant decline observed for the 1980 birth cohort relative to the 1970 birth cohort in the United States, little change in rates of absolute intergenerational income mobility was evident across the Canadian birth cohorts, aside from mild pro-cyclical fluctuations.

Data, sample and method

The data for this study are from the Intergenerational Income Database (IID), a linked administrative database composed of two main components. The first component is a family file in which children who were aged 16 to 19 in 1982, 1984, 1986, 1991, 1996 or 2001 were matched with their parents. To be included in the IID, a child had to: (a) be aged 16 to 19 in one of those years; (b) have a social insurance number (SIN) at that time; and (c) reside with at least one parent at that time. Weights were applied to ensure the representativeness of the sample.^Note 2

The second component of the database is information from income tax returns over the 1978 to 2014 period. The direct matches between children and parents, combined with 37 years of income data, makes it possible to directly compare the incomes of (adult) children and their parents.

This analysis begins with an examination of the incomes of children and their parents at age 30. The period for which income data are available―1978 to 2014―necessarily restricts the sample to children whose parents were born in 1948 or later, as the oldest of them turned 30 in 1978—the first year for which income data are available. (A small number of parents who were younger than age 19 when the child was born are also excluded.) Thus, the first cohort of children consists of those born in 1970, who turned 30 in 2000. Their parents were born from 1948 to 1951, were aged 19 to 22 when the child was born, and had their 30th birthday in the years from 1978 to 1981 (Table 1). Subsequent cohorts of children and parents were identified, usually at two-year intervals. The last cohort of children were born in 1984, and hence, were aged 30 in 2014—the last year in which income is observed. Their parents were born from 1951 to 1965, were aged 19 to 33 when the child was born, and had their 30th birthday in the years from 1981 to 1995.

Given the structure of the data file and sample selection, the parents of the 1970 cohort were younger when their child was born than were the parents of the 1984 cohort. This issue is addressed below.

Adjusted family incomes at age 30 were calculated from income tax returns. For (adult) children who were married^Note 3 by age 30, family income was the sum of the spouses’ total before-tax income divided by two. If the child was married at age 30 but his or her spouse’s SIN was not provided or could not be matched to a tax return that year, the spouse’s income was set to zero. The family income of single children at age 30 was simply their total before-tax income that year.

The family income of parents at age 30 was calculated in much the same way. If two parents were present, total income was the sum of their total before-tax income divided by two. If the father was aged 30 in 1978 and the mother in 1980, the higher of these two family incomes was used for the calculation. If a 30-year-old parent reported being married but his or her spouse did not file a tax return, the spouse’s income was set to zero. If only one parent was present, the total family income was the total before-tax income of that parent.

In the U.S. study, family income was not adjusted for the number of adults in the family. This measure―unadjusted family income―was also considered in the current analysis. Unadjusted family income is the combined income of all adults in the family. However, for two reasons, family income adjusted for the number of adults in the family is preferred. First, age at first marriage has risen, so the percentage of individuals who were not married or in a common-law union at age 30 was higher among the children than among their parents.^Note 4 Because of this generational difference in marriage rates, adjusted family income is likely to be a more accurate measure of a family’s financial well-being. The second reason relates to a fundamental characteristic of the sample: by definition, all parents in the sample had children when they were 30, so it was likely that they were married at that age. In contrast, there was no requirement for the children in the sample to have offspring, so at age 30, they would be less likely than their parents to be married. For these reasons, the study primarily focuses on family income adjusted for the number of adults.

All income figures were converted to constant 2015 dollars using the all-items Consumer Price Index (CPI) series.

The absolute income mobility rates for each birth cohort of children $M_c$ , were calculated directly as the share of children whose family income was at least as high as that of their parents $$M_c=100\%\times N_c^{-1}{\displaystyle \sum _{i}1\{y_{ic}^k\ge y_{ic}^p\} ,}$$ where $N_c$ is the number of children in cohort $c$ , $y_{ic}^k$ is the child’s family income at age 30, and $y_{ic}^p$ is the parent’s family income at the same age. To calculate absolute income mobility rates corresponding to each percentile of parental income, the parents of the children in each cohort were ranked by their income and assigned to percentiles.^Note 5 Percentile-specific rates of absolute income mobility were then calculated by applying the equation above to the parents and their children within each percentile.

Results

Chart 1 shows absolute income mobility rates for children born between 1970 and 1984. Based on adjusted family income, absolute income mobility was higher for each cohort born between 1970 and 1977 than for the preceding cohort: 59% for the 1970 cohort and 67% for the 1977 cohort. In other words, when they reached age 30, 59% of children born in 1970 had a family income that was equal to or higher than that of their parents when they were 30 years old, while for children born in 1977, the corresponding figure was 67%.

For children born in 1979, 1980, 1982 or 1984, absolute income mobility rates by age 30 ranged from 66% to 64%. The modest decline across these later cohorts was likely related to the economic recession in 2008 and 2009. The latter is the year in which the 1979 birth cohort reached age 30. Because the incomes of all children in a particular cohort were measured in the same year (the year they turned 30), their incomes were more sensitive to business cycle fluctuations than were those of their parents.^Note 6

The cross-cohort pattern of changes in the rate of absolute income mobility was similar when the computations were based on the unadjusted family income (Chart 1). However, mobility rates were lower compared with those based on the adjusted family income: 48% for the 1970 cohort, 55% for the 1977 cohort, and 53% for the 1984 cohort. As mentioned above, unadjusted family income is strongly correlated with marital status, so the lower rates likely reflect the lower percentages of children who were married or in a common-law union at age 30, compared with their parents.

These results suggest that overall rates of absolute income mobility for the 1970 cohort were somewhat lower in Canada than in the United States.^Note 7 However, the U.S. estimates show a substantial decline in rates among later cohorts, whereas Canadian rates remained stable. As a result, mobility rates for the 1984 cohort were similar in Canada and the United States when calculated using either income definition.

In addition to the overall rate of absolute income mobility, rates were calculated for children in the 1970, 1975, 1980 and 1984 cohorts according to their parents’ location in the income distribution (i.e., by income percentile) when they were age 30 (Chart 2). Within each birth cohort, children with parents in the lowest income percentiles were the most likely to have a higher family income than their parents at age 30, while the opposite was true for children whose parents were in the highest income percentiles. Furthermore, for children whose parents were between the 20th to 80th percentiles, rates of income mobility were higher among the 1975, 1980 and 1984 birth cohorts than among the 1970 cohort. This suggests that Canada did not experience the marked decline in absolute income mobility among the 1980 cohort relative to the 1970 cohort that was observed in the United States.^Note 8

Two sensitivity analyses were performed to address concerns that, given the composition of the sample dictated by the structure of the IID, parents of earlier birth cohorts were generally younger when their child was born than were parents of later birth cohorts (Table 1). Even though the incomes of all parents were observed at age 30, the lifetime income profiles of parents may vary systematically across cohorts because of differences associated with the timing of family formation.^Note 9

To shed light on this, income mobility rates were estimated for a subsample of parents who were the same age when their child was born. Because it is not possible to expand the age range of parents in earlier cohorts, it is necessary to restrict the age range of parents in later cohorts (“restricted” sample). The first set of columns in Chart 3 shows rates of absolute income mobility for the restricted sample across cohorts of children whose parents were aged 19 to 21 when the child was born. The income mobility rates for this restricted sample ranged from 59% to 64%. Despite some variability before and after the 1979 cohort, no longer-term trend is discernible. When the analysis was repeated for the broader group of parents who were aged 19 to 24 when their child was born, the results were much the same―rates ranged from 63% to 66%, and no trend is discernible across the decade separating the 1975 from the 1984 cohort. Nonetheless, simply restricting all cohorts to children with younger parents does not eliminate the selection issue in the original sample.

Comparing the family incomes of children and their parents at age 40 is another way to verify the results. The sample for this analysis consisted of children born from 1963 to 1974 who reached age 40 in the years from 2003 to 2014, at which time their family income was observed (Table 2). Their parents had been born between 1938 and 1955, and reached age 40 in the years from 1978 to 1995. When their child was born, the ages of these parents ranged from 19 to 25 among the 1963 cohort, and from 19 to 33 among the 1974 cohort. Hence, the selectivity issue regarding parents’ age when the child was born is more modest. Another advantage of comparing children’s and parents’ incomes at age 40 is that it addresses the “lifecycle bias” in mobility estimates that arises when individuals’ incomes are observed when they are still young. Annual income at age 40 to 45 is most representative of lifelong income (Chen, Ostrovsky and Piraino 2017).

Chart 4 shows rates of absolute income mobility for children in the 1963-to-1974 birth cohorts. By the time they reached age 40, 61% of children born in 1963 had a family income that was equal to or higher than that of their parents when they were the same age. For children born from 1967 to 1974, the figure was 67%. Across the percentile distribution for the 1963, 1969, and 1974 cohorts (Chart 5), rates were higher among the 1974 than among the 1963 birth cohort.

Conclusion

This study examines changes in the rates of absolute income mobility in Canada since 2000. Across the cohorts born from 1970 to 1984, and whose income was observed from 2000 to 2014, the percentages of those whose incomes matched or exceeded their parents’ at age 30 remained fairly stable. Likewise, for cohorts born from 1963 to 1974, this was the case at age 40. The findings complement recent estimates of relative income mobility in Canada (Chen, Ostrovsky and Piraino 2017; see also Chen, Ostrovsky and Piraino 2016) and are a first step in examining changes in economic opportunity in Canada over time.

Differences in data sources and methodology mean that the results of this study cannot be compared directly with the U.S. study, which used census data from 1940 to 1980 to measure parents’ income. Canadian census data for the same period do not have sufficient detail to construct a comparable income measure. Consequently, the historical span of this study is narrower. Also, this study and the U.S. study used different methodologies to establish links between children and parents.

Despite such differences, the results from the United States provide a partial reference point. At age 30, rates of absolute income mobility for the 1970 cohort were somewhat lower in Canada than in the United States. However, subsequent trends in the two countries appear to be quite different. The U.S. estimates show a substantial decline in absolute income mobility, but Canadian rates appear more stable.

Appendix tables

References

Chen, W.-H., Y. Ostrovsky, and P. Piraino, 2017. “Lifecycle variation, errors-in-variables bias and nonlinearities in intergenerational income transmission: new evidence from Canada.” Labour Economics, 44 (January): 1–12.

Chen, W.-H., Y. Ostrovsky, and P. Piraino, 2016. “Intergenerational Income Mobility: New Evidence from Canada.” Economic Insights, no. 59. Statistics Canada Catalogue no. 11-626-X. Ottawa: Statistics Canada.

Chetty, R., D. Grusky, M. Hell, N. Hendren, R. Manduca, and J. Narang. 2016. The Fading American Dream: Trends in Absolute Income Mobility since 1940. NBER Working Paper Series, no. 22910. Cambridge, Massachusetts: National Bureau of Economic Research.

Cook, K., and A. Demnati. 2000. Weighting the Intergenerational Income Data File. Ottawa: Statistics Canada. Unpublished report.

New York Times. 2016. “The American Dream, quantified at last.” December 8. Available at: http://www.nytimes.com/2016/12/08/opinion/the-american-dream-quantified-at-last.html) (accessed March 31, 2017).

Notes