27 August 2014
The last few months have seen major research activity in the area of inequality at UNU-WIDER. An update of the World Income Inequality Database (WIID) was first published in June. This is the most comprehensive database on inequality currently available. Considerable and meticulous work has been done on checking all the data points over the last year or so. And at the start of September our WIDER Development Conference is devoted to the topic of inequality. One of the key topics to be discussed at the conference is data. What kind of datasets are available for cross-country comparisons of inequality? What should one keep in mind when using such datasets? And what do these data reveal about the extent and changes in income inequality?
The WIID update
The WIID collects and maintains data on income and expenditure for all countries for which this is feasible. The basic philosophy behind WIID is to provide users with all estimates of monetary inequality which are expressed either as Gini coefficients or as percentile share of income. But even if one restricts attention to these indices, inequality can still be measured in a myriad of ways, it can be before or after redistribution, it can refer to either income or consumption, the measures can have been estimated either using the whole population or for certain subsets such as only the urban population. All these choices make a huge difference to the estimated levels of inequality. Which is why WIID provides information about all these issues and, with due diligence, the user can make sure they are not comparing apples with oranges when making international comparisons of inequality.
Data quality varies enormously over countries. For some countries, we have only very limited knowledge about the data coverage or survey methods, whereas for other countries, the observations are based on high-quality standardized datasets (such as the Luxembourg Income Study database, LIS). The WIID offers a three-scale quality assessment of inequality observations. Then the user can limit the analysis only to observations for which the data quality is the highest or, to increase the scope of the countries covered, lower-quality observations can also be added to the analysis. The choice is for the user to make: the database allows (and encourages!) the user to make these decisions deliberatively and, when reporting results, advise their audience of the exact definition of inequality used.