Abstract
Comparative poverty research flourishes,especially since comparable income data areeasily available through the Luxembourg IncomeStudy. However, widely-used income surveyssuffer from a number of methodological pitfallsfor comparative poverty research. This researchreport uses the data available in theLuxembourg Income Study for three countries –Germany, Sweden and the United Kingdom – toexemplify the limited comparability ofwidely-used income data used in povertyresearch. In a simple simulation exercise,entitlements to means-tested benefits areimputed for each household in the sample, basedon the institutional regulations in eachcountry. Compared to actual poverty rates inthe original sample, imputed poverty rates aremarkedly smaller, if not reduced to zero. Evenif one accounts for an incomplete take-up ofbenefits, a large gap between actual andsimulated poverty rates still remains, largelycaused by problems in survey design. The paperconcludes with a number of recommendations forimproving income surveys from the perspectiveof comparative poverty research.