Brian Nolan, Ive Marx and Wiemer Salverda
The Time Period and Unit of Analysis
In studying the distribution of earnings, measures over a variety of time-periods may be the focus. Earnings
over a week, month or year may be employed, and sources vary as to which is covered. it is also common to ex-
press earnings in hourly terms by dividing total earnings over the period by the number of hours worked. This has
a number of advantages, especially in allowing full-time and part-time employees to be compared on a meaningful
basis - which is particularly important given the growth in part-time employment in some countries and the diver-
sity in patterns across countries. Variation across employees in earnings over the week, month or year are clearly
affected both by differences in hourly rate of pay and in hours worked during the week, the month or the year.
Table 2 demonstrates the quantitative importance that taking different bases can have for the level of in-
equality, using the USA as example.
Table 2: United States wage inequality ratios, highest to lowest decile, 1979
Study |
Wage concept |
All |
Men |
Karoly, 1993, Appendix 2B2 |
Annual |
21.50 |
13.79 |
Weekly |
8.62 |
7.00 | |
Hourly* |
4.75 |
4.72 | |
Mishel et al.. 1996, 143—4 |
Hourly |
3.50 |
3.67 |
OECD, 1996, 62 |
Weekly |
3.18 | |
Davis. 1992. figure 1A |
Weekly |
3.37 | |
Katz. et al.. 1995. 58 |
Hourly |
3.42 | |
Erikson & Ichino. 1994. 31 |
Annual |
5.61 |
* Used in OECD, Employment Outlook 1993
Source: Salverda (1998, Table 3.3)
The use of a longer time period clearly expands the number of persons over whom one is calculating inequality
measures - from all those currently in work to potentially all those who have worked, if only for a very short pe-
riod, during the year. A related issue is that the individual earner is not necessarily the only possible unit of analy-
sis: one can for example analyse the variation in hourly earnings across all hours worked rather than all earners.
Studies of wage inequality, which most often have a labour market focus, generally concentrate on gross earn-
ings before tax or social insurance contributions are deducted. Earnings net of such deductions are also of interest,
though, both from a labour market and a distributional perspective. The timing of tax deductions and tax rebates in
particular may then also affect net earnings.
Summary Measures
Summary measures are often used to capture earnings inequality, but for reasons that are not obvious com-
mon practice differs from the income inequality field - reflecting the fact that these have been two largely separate
Page • 28