Launch of Wave IX – US 2013 added & revision of the US series

United States 2013 added to the LIS DatabaseLIS is delighted to announce the launch of Wave IX of the LIS Database.
The first dataset in Wave IX is the United States 2013, which is now available for use. This dataset is based on the Current Population Survey (CPS) carried out by the US Census Bureau.

In addition, the US datasets in Wave I through Wave VIII have undergone a substantial revision triggered by a methodological change in the topcoding procedures applied by the US Census Bureau. Starting with the 2011 CPS data (referred to at LIS as US10), the Census Bureau applied a new approach to preserving the confidentiality of high-income earners; they subsequently applied this new procedure to previous datasets going back to the 1970s.

In order to provide US data that are as comparable as possible over time, we re-harmonised all of our US datasets (with the exception of the Historical Wave), using these newly released data. When we re-harmonised the US datasets, we also utilized the newest LIS Database template (i.e., the set of LIS variables and definitions). As a result of both factors (i.e., the revised original data and the re-harmonisation carried out with the new LIS template), the US data in the LIS Database are now much more consistent over time — especially when analyzing inequality and incomes at the top of the distribution.

Furthermore, during the process of re-harmonising the US series, we discovered an error in our harmonised 2010 US dataset. We had inadvertently omitted Food Stamps from all the income variables. We have corrected that error.

As far as the Historical Wave is concerned, we have decided to remove the US69 dataset as it was the only LIS dataset which did not include DHI (and was thus the only one for which we had no Key Figures); the (non-revised) dataset is still accessible through LISSY, but it no longer appears in the documentation. On the other hand, the US74 dataset is still part of the LIS Database, but we were not able to rerun it with the newly released data due to the fact that it is no longer possible to retrieve the information on taxes and contributions for that year. The dataset was thus only slightly modified, alongside the rest of the database, to take into account the latest template (see “LIS DatabaseRevision Note).

As a result of these revisions, LIS strongly recommends that, for projects now underway that include US datasets from US79 through US07, users should re-run their analyses using this new series — especially if the study focuses on the upper end of the distribution. For projects now underway that include US10, users should re-run their results based on the corrected US10 dataset; the correction of the Food Stamps error may affect results at the lower end of the distribution.

For details on the US revision, see Revision Notes.

November 26, 2014 | Highlight