WID.world Data Quality

No data is available (imputed regional average). This is the lowest-quality category and is used only as a last resort to ensure coverage.

Estimates are based on the observed correlation between income inequality and wealth inequality. Wealth distribution is predicted using an income inequality weighted average of other countries’ wealth distributions, with adjustments for rich-list and national accounts.

No income and employment data is available (imputed regional average). This is the lowest-quality category and is used only as a last resort to ensure coverage.

The value is imputed by carrying the nearest available observation forward or backward, filling short gaps in an otherwise continuous series.

Estimates based solely on survey data, statistically adjusted to approximate missing top-end values. They include rich list and national accounts corrections when available (typically from the 1990s onwards).

Values are estimated using an OLS regression model with female shares of wage and self-employment as the main predictors; employment data are drawn from ILO modelled estimates. Unlike grade 2, no observed income data are available for the country, so the full series is model-generated.

Values are estimated by linear interpolation between two known data points, typically for short, internal gaps.

Wealth distribution estimates are imputed either via a simplified mortality-based multiplier method, from micro inheritance/probate data, or from wealth tax tabulations. These are supplemented by rich list and/or national accounts adjustments when available (typically from the 1990s onwards).

Values are estimated using an OLS regression model with female shares of wage and self-employment as the main predictors; employment data are drawn from ILO modelled estimates. Where observed income data exist for the country, the predicted trend is level-adjusted to match those observations, anchoring the series to real data points.

Only one type of source is used: either income/consumption surveys, or tax tabulations; the two sources are not jointly available. Tax tabulations are typically the sole source for the historical series (pre 1980) or fiscal income series.

Aggregate wealth figures are derived from national-accounts balancing items, then distributed according to survey shares.

Values are estimated by linear interpolation between two known data points, typically for short, internal gaps.

Tax and survey data are both available, but at least one source is in tabulated (not micro) form.

Wealth distribution is modeled using a hybrid of either income-capitalization techniques or wealth/estate tax tabulations and survey information, with rich list/national accounts adjustments.

Female and male wage and self-employment incomes are derived from a supplementary survey source. These observations are used to extend the time series beyond the years covered by the benchmark source, and are therefore of slightly lower priority but remain grounded in observed microdata.

Fully integrated micro-level data from tax records and household surveys are available.

High-quality wealth microdata from administrative registers or tax declarations, requiring only minor imputations for unobserved assets.

Female and male wage and self-employment incomes are directly observed from the benchmark survey source for each country. The benchmark source is selected based on breadth of coverage, methodological consistency, and systematic data collection procedures. These observations represent the highest data quality in the series.

No data are available. Values are assigned using a regional average or another assumed value solely to ensure coverage. This is the lowest-quality category and is used only as a last resort.

No data are available. Values are assigned using a regional average to ensure coverage. This is the lowest-quality category and is used only as a last resort.

No data are available; a static value is assumed. This is the lowest-quality category and is used only as a last resort to ensure coverage.

No observation is classified with this score.

Values are estimated by extrapolating nearby observations using simple assumptions, such as forward/backward projection, moving averages, regional growth rates, or stable relationships with a related country or former common jurisdiction.

Values are estimated by extrapolating nearby OLS predictions with a regional average.

The value is imputed either directly or by projecting an observation using a secondary variable—or the same variable—as a reference for growth observed in one or more neighbouring countries or countries with historical affiliations. For example, exchange rates may be imputed based on comparisons of Yugoslav GDP values expressed in different currency terms, with the resulting estimates applied to the former Yugoslav countries.

The value is imputed by projecting the nearest available observation forward or backward, assuming constant relative shares, estimated in a recent reference period, between the country and the aggregate of another country with which it was part of the same jurisdiction at some point in time—for example, the Czech Republic and Slovakia following the dissolution of Czechoslovakia.

Values are estimated using statistical or econometric methods, such as OLS regressions on relevant macroeconomic covariates or signal-extraction techniques (e.g., Hodrick–Prescott filtering).

Values are estimated by extending nearby directly observed data with regional average.

The value is imputed by projecting an observation using a secondary variable observed within the same country as a reference—for example, a KP PPP exchange rate calculated based on the comparison of GDP values expressed in different currency terms—or by carrying the nearest available observation forward or backward.

The value is imputed by projecting the nearest available observation forward or backward, using growth rates or age shares observed in the country’s available totals.

Values are estimated by linear interpolation between observed data points or derived through the simultaneous adjustment of macroeconomic identities to ensure internal consistency.

Values are estimated using OLS regressions on relevant macroeconomic country-specific covariates rather than past observations.

Values are estimated by linear interpolation between two known data points, filling short internal gaps.

Values are estimated using linear interpolation between adjacent data points to fill short internal gaps.

Values are directly sourced from older national accounting frameworks, benchmark academic studies, or historical compilations. Comparability with current standards may be limited.

Values are estimated by extending nearby directly observed data with OLS predictions.

Values are sourced from benchmark academic research, historical compilations, forecasts under the latest SNA (e.g., IMF WEO estimates of deflators for some countries in the past year), or from older SNA frameworks.

Values are directly sourced from historical compilations; comparability with current standards may be limited.

Values are directly observed from the most recent national accounts compiled under the latest SNA standards. This is the highest-quality category.

Values are directly observed in official balance sheets (from the most recent national accounts compiled under the latest SNA standards), global asset datasets, or validated academic studies. This is the highest-quality category.

Values are directly observed from the most recent national accounts under the latest SNA, representing the highest-quality data.

Values are directly observed from the UN World Population Prospects or from the most recent national accounts compiled under the latest SNA, representing the highest-quality data.