Fiscal redistribution is inequality reduction by way of public taxes and transfers---this is the most up to date methodology.

The figures described in this chapter update and extend our “Fiscal Redistribution Dataset,” which has provided data on a number of aspects of inequality reduction in developed countries by way of taxes and social transfers. The dataset was first compiled from LIS microdata in 2005. 4 In 2008, it was updated to reflect changes in LIS methodology and to include several newly available income surveys. 5 Some of our measures were further extended and updated in 2011 in a separate effort by Caminada and Wang.6 Our revised dataset, as well as that of Caminada and Wang, are available on the LIS website and have been widely used by researchers interested in income inequality and government redistribution.

The purpose of this chapter is to describe the results of a thorough update of our data on fiscal redistribution that we have recently undertaken, an update in which we have not only added new figures but also recalculated earlier ones to reflect recent changes in LIS methodology. For a number of reasons, we believe that this is a good time to update our dataset “from the ground up.” First, in mid-2011 the LIS implemented a new data template that made a number of changes in traditional LIS variables. One important revision was a new definition of post-government income, which is now called “disposable household income” (DHI). The main change is that most non-cash income (but not imputed rent) is now included in market income. This matters most for developing countries, but it also has some effect on the developed countries that are our focus. Since this new DHI income concept has become the basis for the LIS’s widely used “Key Figures7 ,” we believed that it was important for us to employ it as well.

Beyond this, there has been a broader effort in the new LIS template to improve standardization of income definitions across countries, and numerous smaller revisions and corrections have been made to various datasets in the last few years, some initiated by the LIS and others by the national statistical agencies that supply the original data. Finally, nearly all LIS variable names were changed in 2011, in an effort to develop a consistent nomenclature. In this revision we will use the new names, which should make it easier for other researchers to replicate, update or extend our calculations.

Aside from the new template just described, the LIS database has grown considerably since our last calculations. While our original data set included 59 country-years, the version described here includes 116. One reason for this is that LIS Waves have become more frequent: they are now conducted approximately every three years instead of every five. Of special interest is the fact that Wave VIII includes surveys conducted in 2010, after the onset of the 2008 global financial crisis. In addition, the figures described here include no fewer than seven countries that were not represented in earlier versions, either because the starting point of thenavailable surveys measured income net of direct taxes, or because they have recently joined the LIS project.

To be specific, this revision includes all currently available (as of June, 2015) LIS “gross income” datasets, that is, data sets whose starting point is pre-tax income.8 It does not include “net income” datasets, for which it is not possible to account for direct taxes, or “mixed” datasets, for which coverage of direct taxes is incomplete.9 Furthermore, our coverage is limited to the developed countries; it does not include LIS datasets for transitional or developing economies. In all, we include 20 countries for an average of 5.8 points in time ranging from 1967 through 2010. The exact countries and years, along with all data and details on household size equivalization, household weighting, survey weighting, the treatment of zero income, and top and bottom coding, are available in the on-line Appendix (available at http://www.lisdatacenter.org/resources/other-databases/).

Measuring Income Inequality and Government Redistribution

The starting point in computing summary figures for income redistribution is to measure the distribution of private sector income. The most important source is earnings, which are comprised of wages, salaries and income from self-employment, including (as much as possible) non-cash compensation. To this figure are added income from property, such as interest and dividends, rental income, royalties, non-cash income (as much as possible) and pensions received by private and public sector employees. The total of these sources of income is defined as “factor income.” Finally, we add to factor income three additional, relatively minor, sources of private but non-market income: merit-based educational transfers, transfers from non-profit institutions, and inter-household transfers such as alimony and child support.

In measuring the effect of direct state redistribution via taxes and transfers, it is first necessary to add to private sector income a number of public sector social transfers. As has been indicated, the coverage of such transfers in LIS income surveys is quite extensive. The main benefit modes include retirement pensions; child and family allowances; unemployment compensation; sick pay; accident pay; disability pay; maternity pay; “other social insurance”; and means-tested social assistance of various kinds. After summing private and public sector sources of income, we arrive at “total gross income.” The final step is to deduct from total gross income the most important taxes that are paid directly by households: income taxes and mandatory social insurance contributions. When this is done, we arrive at our measure of posttax and -transfer income, called “disposable household income”—the income households actually receive.