HOW TO INTERPRET INEQUALITY DATA AND PERCENTILES ON WID.WORLD?

Some of the data in WID.worl d refers to specific parts of the distribution. That is, we rank people according to their income or wealth, group them according to their rank, and report statistics on these different groups. They are coded pXpY, where X and Y are both numbers between 0 and 100. X correspond to the percentile for the lower bound of the group, and Y to the percentile for the upper bound (hence X < Y).

WID.worl d defines the following groups:

key groups are the most widely used. They include p0p100, p0p50, p50p90, p90p99, p99p100, 9p100, p99.99p100, in addition to all deciles (p0p10, p10p20, p20p30, …, p90p100). For example, p0p50 refers to the bottom 50% of the population, and p99p100 to the top 1%.

detailed percentile groups are narrow groups that cover the whole distribution. They include all percentiles (p0p1, p1p2, …, p98p99, p99p100), but also tenths of a percentile in the top 1% (1, p99.2p99.3, …, p99.8p99.9, p99.9p100), hundreds of a percentile in the top 0.1% (p99.9p99.91, p99.92p99.93, …, p99.98p99.99, p99.99p100), and thousands of a percentile in the top 0.01% (p99.99p99.991, p99.992p99.993, …, p99.998p99.999, p99.999p100). For example p99.5p100 correspond to the top 0.5% of the population.

detailed top groups also cover the whole distribution but refer to the entire population above a percentile. They take the form p0p100, p1p100, p2p100, …

Some population groups can belong to several of the above categories (for example p99p100 is in all three). When that happens, the group appears only once in the data.

The population group p0p100 refers to the entire distribution and is used for aggregate series for which there is no distribution available, and for economic series with no individual meaning such as price indices and exchange rates.

For all these population groups, we provide:

wealth or income share: these correspond to the income or wealth of the group, divided by the total for the whole population. For example, the income of the group p99p100 is the top 1% income share.

wealth or income averages: these correspond to the average wealth or income of the people in the group. For example, the wealth average of the group p90p99 is the average income of the top 10% excluding the top 1%.

wealth or income thresholds: they correspond to the minimum level of income or wealth that gets you into a group. For example, the income threshold of the group p90p100 is the income of the poorest individuals in the top 10%. By definition, it is equal to the income threshold of the groups p90p99 or p90p91.

For all top groups, we also provide the inverted Pareto coefficient. It is the ratio of the group average over the group threshold. They are a measure of the fatness of the tail of the distribution of income or wealth. See Thomas Blanchet, Juliette Fournier and Thomas Piketty, “Generalized Pareto Curves: Theory and Applications” (2017 ) for details.

Beware that people are not always ranked with respect to the quantity that the variable measures. That is true in particular for subcomponents of income or wealth. For example, a variable for “capital income” may rank individuals according to their total income (hence p99p100 would refer to the capital income of the top 1% of both labor and capital income earners), or it may rank them according to their capital income (hence p99p100 would refer to the capital income of the top 1% of capital income earners). The description of the variable specifies which ranking is used.