LIS Key Figures

LIS Key Figures - Methods

There are a number of methodological decisions confronting researchers of income inequality and poverty. We recognize that our users make different choices, for very good reasons, when measuring income inequality and we encourage you to continue doing so. However, we adopt the following approach in an effort to standardize the measures of income inequality that we report on our Web Page: Gini Coefficients, Atkinson Coefficients (epsilon=0.5 and epsilon=1.0), Percentile Ratios (p90/p10, p90/p50 and p80/p20) and Relative Poverty Rates (poverty lines computed at 40, 50 and 60 percent of the median equivalent income). For these final sets of measures, we also report the Poverty Rate for children and for the elderly.

Population coverage
All surveyed households and their members are included in our estimates of Gini and Atkinson Coefficients, percentile ratios and poverty lines (important: we exclude families from multifamily households in order to avoid double counting. Please refer to LIS Variable Definition List and the variable d5). It is important to note that after the poverty line has been calculated, only those households including either members under the age of 18 (for the child poverty figures) or over the age of 64 (elderly figures) are included when computing the proportion of the population (subgroup) living in poverty. In other words, the poverty line remains the same whether one examines the entire population, children, or the elderly but the households included in calculating the percentage below the threshold changes with each subgroup. Therefore, one must bear in mind that the sample sizes upon which these subgroup rates are based are fractions of the entire sample and treat these figures with discretion.

Income concept
All of the LIS Key Figures are for Disposable Income (the LIS aggregate income variable DPI). For more information about the sources of income included in this income concept, you can go to the following page: summary income.

Median Equivalent Income
We calculate the median equivalent income which is the middle observation of the considered distribution. Here, we used the Equivalised (See. below) Disposable Income (See. above)

Equivalence scale
The equivalent income of an individual is defined as the household income of the individual adjusted for differences in household size. Unrelated individuals are considered to be 1-person families. Each member of a given household has the same equivalent income, regardless of age or family relationship.
The square root of the number of persons in the household is used as the equivalence scale in all of the figures we report (see reference #1, page 21).

Weighting
We use person weights (HWEIGHT*D4) when measuring income inequality for the total population (see reference #1, page 21). Child poverty and elderly poverty rates are also reported among our tables. When computing the rate of poverty among children, a child weight is constructed by multiplying the household weight by the number of members under the age of eighteen (HWEIGHT*D27) (see reference #3, page 9). Our examination of poverty within the elderly population follows the same logic and applies a weight multiplying the household weight by the number of members 65 years old and older ((NUM6574+NUMGE75)*HWEIGHT). In many early LIS datasets, identified in the tables, one must adopt a different approach when constructing the weighting variable (please refer to the program "eldpov2").

Top and Bottom Coding
We bottom code the LIS datasets at 1% of equivalized mean income and top-code at 10 time the median of non-equivalized income (see references #2, page 661).

Missing values and zero incomes
All missing values and zero incomes are excluded. Due to the recoding of some income variables, zero represents both missings and actual zero incomes for some older LIS datasets, whereas for all wave 5 datasets missing values can be distinguished from actual zero incomes. Since interpretation of actual zero DPI is difficult, we exclude them as well.

Warnings
1. Results might not always be fully comparable as for some countries, datasets may be based on different surveys or variable construction may varies. For more information, please see the list of surveys or documentation by country.

2. Publication rules
** Estimates based on less than 15 observations are not published
* Estimates based on 15 to 30 observations only

3. Note that In LIS, expenditure or income variables are expressed in units of national currency in use at the time of the data collection period except for a few exceptions. Please look at LIS Policy on the Treatment of the Currency to properly interpret the LIS key Figures.

References

#1. Atkinson, A. B., L. Rainwater, and T. M. Smeeding. 1995. Income Distribution in OECD Countries: Evidence from the Luxembourg Income Study (LIS). Paris: OECD.

#2. Gottschalk , P. and T. M. Smeeding . 1997. "Cross-National Comparisons of
Earnings and Income Inequality," Journal of Economic Literature, XXXV (June): 633-687.

#3. Lee Rainwater, Timothy Smeeding and John Coder. 1999. Poverty Across States, Nations and Continents" A revised version of this paper appears in Vleminckx, Koen & Tim Smeeding (eds.) Child Well-being, Child Poverty and Child Policy in Modern Nations. What Do We Know? The Policy Press, Bristol, March 2001.