Income tax is often regarded as a major disincentive for work imposed by a government. In theory, when income tax rises, people’s income decreases and therefore people work more to maintain the living standard, known as income effect, but at the same time they substitute away from work to leisure because the opportunity cost of work is now lower, known as substitution effect. Hence, theoretically, the direction of the overall change is ambiguous. Nevertheless it is generally assumed empirically that income tax overall decreases work effort. However, this approach implicitly assumes that people’s work is rewarded by their absolute incomes. But this may not be the case. This paper presents three models that consider other factors that may affect people’s work decision. The first model looks at consumption. The second model considers the case where people’s productivity is affected by the society’s income equality and the third model takes the idea from comparison income effect, which argues that people’s satisfaction is determined by relative income rather than absolute income. The second and the third models take into account past papers’ results. In all cases the models suggest that redistribution or income tax alone can improve the overall work effort in a society when the initial income inequality is high. The three models are entirely original. This paper also provides empirical tests with two data sets from the OECD countries and other countries in general. Taking a long term economic growth rate as a measure of people’s work effort, this paper finds strong empirical support for the second model. I will first present in Section 2 the thought experiment that led me to investigate the relationship between income tax and work effort. Section 3 will therefore review past papers about the impact of income tax on work effort. Section 4 will present the models with reference to the past papers on which the models are based. Section 5 will examine the hypotheses for the first and the second models. Section 6 will discuss and conclude the results.