The increased interest for reliable information for restricted domain with reference to inequality measures is due to different reasons. One of the most relevant is to better plan policies to reduce inequality at local level. In this regards, an increasing gap in inequality and social exclusion among regions within the different EU member States has been observed in recent years. This issue is particularly relevant for Italy whose economic system is characterized by a strong territorial disparity.

Using data taken from the EU-SILC sample survey for Italy in 2013, we focus on the estimation of inequality measures for the Italian provinces. Small area models incorporate the design based estimates (direct estimates) as inputs, that are typically unbiased even though unreliable for small samples. Nevertheless, in the case of inequality measures, design based estimates are instead known to be biased for small sample sizes. In this work we focus on the search for a correction that can produce approximately unbiased direct estimators, taking into account the complexity of the survey design. Those modified estimators can then be used in small areas models.