Income and the outcomes of children

Aperçu

Résumé (français)

Veuillez noter que les résumés n'apparaissent que dans la langue de la publication et peuvent ne pas avoir de traduction.

Résumé (anglais)

Abstract This report re-investigates the connection between income and child well-being for a broad range of outcomes. The report attempts to address four research questions: 1. Which measure of income is most appropriate to examine the relationship between income and child outcomes? 2. What is the functional formof the relationship between income and child outcomes? 3. Are associations between incomeand child outcomes larger for younger or older children? 4. Do income changes or income levelsmatter most for child outcomes? In order to understand how income levels and/or income changes may affectchildren at different stages of development, regression equations are estimated using alternative income concepts and hypothesized functional forms. The analysis uses data from the Statistics Canada National Longitudinal Survey of Children and Youth (NLSCY), cycles 1, 2 and 3 1 . The report focuses on the longitudinal sample of children who are present in all three cycles. This allows for a comparison of associationswhich exist between current measures of income and/or low-income status and current child outcomes with associations which exist between current outcomes and measures of past income. Analyses are conducted for a broad list of child outcomes which can be categorized into four developmental domains: 1) cognitive, 2) social/emotional, 3) physical, and 4) behavioural. The major results derived from the regression analysis are summarized below: 1. Higher income is almost alwaysassociated with better outcomes for children. This is true regardless of the measure ofincome employed, the assumed functional form of the relationship between income and child outcomes, the age of the child, orthe type of child outcome being studied. It is also apparently true using either the NLSCY or the Youth in Transition Survey (YITS)/Programme for International Student Assessment (PISA) data. 2. The sizeof the association between income and child outcomes varies with developmental domain. Thus, for example, income has particularly strongassociations with cognitive outcomes (e.g., Peabody Picture Vocabulary Test (PPVT) scores or math and reading scores) and behavioural outcomes (e.g., hours spent watching television). Physical health outcomes also have quite consistent positive associations with family income. Associations are generally smallest with ‘social/emotional’ outcomes (though hyperactivity is an exception to this ‘rule’). Again, descriptive evidence from the YITS/PISA are consistent with these findings. 3. A three-period average of family equivalent income consistently has the largest associations with child outcomes. This is true across almost all kinds of outcomes and all ages of children. It is also true for children living inmarried-couple or lone-mother families. Thus, in general, it appears advisable to use an income measure averaged over as many years as are available in the data. 4. The functional form of the relationship between family income and child outcomes varies considerably across developmental domains. Estimating an inappropriate functional form may lead a researcher to the (false) impressionthat income does not matter, so that it is important to test a variety of alternatives. 5. It is almost never true that beyond the low-income threshold, income is unimportant for children’s outcomes. (Non-nested hypothesis tests reject ‘low-income’ specifications in favour of continuous income specifications in almost every case.) 6. While over half of the outcomes studied here increase more quickly with income at lower than at higher income levels, itis almost always true that there is not a ceiling above which income no longer matters for child outcomes. 7. The relationship between incomes and outcomes appears to ‘flatten’ out toward the linear for older as compared to younger children. Thus, increases in income at very low-income levels are particularly important for the youngest children. 8. We find little evidence of differences in the size of the income/outcome associations for children of different ages. However, only a subset of outcomes are measured identically across age groups. 9. For middle-aged and for older children, changesin family income appear to be less important for child outcomes than levelsof family income. However, incomechanges are more important for younger children, particularlyif they happen earlier in life (i.e., between 1994 and 1996 rather than between 1996 and 1998). Income ‘ups and downs’ are particularly important for child emotion scores, an outcome for which income levels appear to play a less important role. 10. Results using the Youth in Transition Survey show a positive correlation between socioeconomic status and child outcomes, in general. While income was not available in this survey at the time of analysis, direct comparisons by incomeare not possible. Therefore a proxy for income in a socioeconomic status (SES) index is used with outcomes which can be categorized into the same categorical domains as the NLSCY.It is interesting that the differences across SES quintiles are particularly large for o