Table 2.

Theoretical evaluations and empirical patterns of the four income definitions.

Approach	Conceptual evaluation	Empirical differences and similarities	Gender differences
Conceptualizations
Household (disposable)	(1)Considers economies of scales; (2)Better reflects material conditions; (3)Better captures absolute income; (4)Unpartnered and unemployed individuals fall into lower income groups.	(1)Larger mortality inequalities at working ages are found when individual income is used than when household income is used; (2) Larger mortality inequalities are found at older ages when household income is used than when individual income is used; (3)The age patterns of inequalities are similar for both, but the use of household income leads to later peaks.	(1)For women, larger differences in life expectancy are found when household income is used than when individual income is used; for men, when population-wide grouping is applied, larger differences in life expectancy are found when individual income is used than when household income is used; (2)Individual income ranks the lowest-quintile women higher than some higher-income groups in recent periods (due to differing household composition); (3)The period patterns for men are similar for both, but the patterns are different for women; (4)Differences in inequality levels depending on the income definitions are larger for women and for men.
Individual	(1)Better reflects earning ability and relative social position; (2)Not able to capture absolute income for partnered individuals who do not work
Operationalizations
Population-wide grouping	(1)Better reflects absolute income (not wealth or resources); (2)Young and old individuals are less likely to be in high-income groups, and individuals at prime working ages are more likely to be in higher-income groups.	(1)Larger mortality inequalities across ages are found when population-wide grouping is used than when age-specific grouping is used; (2)Larger differences in life expectancy are found when population-wide grouping is used than when age-specific grouping is used; (3)The age patterns are similar for both.	(1)For individual income, the gaps in life expectancy are much smaller when age-specific grouping is used than when population-wide grouping is used, and population-wide grouping ranks the lowest-quintile women much higher. These patterns are not found for men; (2)The period patterns are similar for men, but are different for women; (3)Differences in inequality levels depending on the income income definitions are larger for women and for men.
Age-specific grouping	(1)Better reflects relative positions on income and social ladders.