The Gini coefficient and the case of negative values

The contribution of smallholders' livelihood activities on income inequality and poverty: Case study from rural Tanzania

“…First, if the number of households with negative income is high, a significant proportion of information would get lost. Second, ignoring households with negative values may lead to insufficient comparisons between different distributions (de Battisti et al, 2019). Hence, ignoring households with negative income, which are generating most of their income from agricultural activities, thus is not the optimal solution because reporting negative incomes is common and ignoring them also means ignoring key features of rural household's income (Rawal et al, 2008;Zhang et al, 2019).…”

“…Another common way to deal with negative incomes is to set them to zero. Many researchers (Burkhauser & Simon, 2010;de Battisti et al, 2019;Feng et al, 2006;Ferreira & Gomes, 2015;OECD, 2017) applied this truncation of data with negative values. The truncation can lead to a loss of information, but studies from Seidl et al (2012) and Bray (2014) showed consistency in the results setting negative values to zero.…”