2.1. Saving propensity differential and the Kaldorian mechanism It is somewhat improper to refer to this channel of transmission of inequality to growth as “Kaldorian” because the causality is opposite to that of Kaldor’s original contribution, where it runs from growth to the functional distribution of income between capital and labor. 4 In fact, the modern literature essentially refers to the idea in Kaldor’s work that capitalists save more than workers. Furthermore, this difference in saving propensity between capital and labor income is taken to be equivalent to there being a higher propensity to save among richer people. Given this argument, and if savings determine investment, more inequality should be associated with faster growth. Then, an important empirical issue is whether it is valid to extend the saving propensity differential between capital and labor income at the macro level to individual incomes. If a substantial part of savings arises from undistributed profits, the observed level of inequality among households will have a negligible impact on savings, investment, and growth. It would then be better to consider aggregate factor shares than inequality measures. This potential negative impact of too much inequality, via the saving propensity gap, generalizes to the effect of progressive redistribution upon the rate of growth. Transferring income from the top to the bottom of the distribution through taxation may reduce the overall propensity to save of the economy through the preceding mechanism. More directly, however, income taxation may reduce the rate of return to capital and the incentives to save and invest. This is the traditional argument for the existence of the tradeoff between inequality and growth. Economies that redistribute more must be less unequal than others, but they may not grow as fast. This hypothesis has been the focus of an important part of the empirical literature on inequality and growth. Endogenous redistribution Redistribution may also be endogenous, and a response precisely to too much inequality in market incomes. This provides another channel through which income inequality may affect economic growth. If redistribution does reduce investment incentives and entrepreneurship, inequality may indeed be responsible for slower rather than faster growth. The difference with the preceding case is that the relationship now is between the inequality of market incomes and growth, rather than disposable incomes (i.e., after taxes and transfers) and growth. The same can be said of Piketty , which considers steady states where both the growth rate and the rate of return on wealth are exogenous. Thus, the causality again runs from growth to distribution, rather than in the opposite direction.