The Kuznets curve: Evidence from Prussia revisited

To avoid confusion arising from mixing up methods, the methodological arsenal was restricted to the computation of percentiles as this was the favoured device in the Kuznets-paper. But with an increasing population the interpretation of rising percentiles is by no means straightforward (which is equally true with other measures of inequality). When the number of income-receiving units and their mean income are held constant, rising shares of the top-income receivers can be interpreted in terms of “rich people getting richer and poor people getting poorer”. But when both, the number of units and their mean income are rising, the interpretation of rising percentiles is quite a problem. For example, aggregating two populations with all percentiles alike (i.e. identical Lorenz curves) but different mean incomes would result in a greater population with increased shares of the top-income receivers and lessened shares of the low-income receivers (i.e. a Lorenz curve more close to the boundaries of the rectangle). It is obvious from this example that the welfare implications of increasing percentiles cannot be separated from the welfare implications of rising mean incomes. Especially, with rising populations and rising mean incomes the income shares of the top-income receivers may very well increase without any poor man getting poorer, so that not even Rawls [1971] would necessarily object to such a development. Bearing these limitations in mind, the outcomes of the previous sections may be summarized in confirming Kuznets’ surmise that income inequality in Prussia increased within the second half of the last century. Furthermore, it is emphasized that the peak was most probably reached around the turn of the century with the First World War merely fortifying the equalizing tendencies which originated from the economic development process itself.