According to the well-known Kuznets (1955) inverted-U hypothesis,
income inequality increases during the early stages of economic development and after reaching a turning point declines.
The relationship has been subject to extensive empirical testing [see Bruno et al.
Most studies divide countries into two groups (developed and less developed) as a prerequisite to
testing the Kuznets hypothesis.
The implicit assumption is that the nature of the relationship differs according to a country’s level of economic development.
Therefore, the sample should be split into groups reflecting the level of economic development.
It is important to note, however, that the criterion according to which countries are divided into two groups depends on a researcher’s subjective notion of the category (LDC or otherwise) to which each country belongs or on a notional level of per capita income that divides the two groups.
The purpose of this paper is to reexamine the inequality–development relationship using recently
developed econometric techniques that test endogenously the existence of a threshold level and, thus,allow for the possibility of endogenous sample separation.
The threshold regression (TR) model allows the level of economic development (per capita income) to determine the existence
And significance of a threshold level in the Kuznets relationship rather than imposing a priori an arbitrary classification scheme.
If indeed there exists a well-defined relationship between income inequality and
per capita income that depends on the level of economic development,
theTRmodel can identify the threshold level and test simultaneously for such a relationship above and below the threshold.
The following section describes theTRmodel as it applies to the Kuznets hypothesis. Section 3
presents the empirical results.