A Model Of Net Wealth Distribution Specified For Negative, Null and Positive Wealth. A Case Study: Italy

Abstract

Research on wealth distribution deals with two major categories of variables: (a) total (gross) wealth, and (b) net wealth, which is equal to gross wealth minus total debt. By definition, the economic units’ total wealth takes non-negative values. Pareto’s (1896) seminal monograph Ecrits sur la courbe de la repartition de la richesse does not deal with wealth but with income distribution. Although several models, including Pareto’s, were also proposed to fit wealth distributions, they are restricted to describe only the positive range of wealth since they are not defined for null and negative net wealth and for null total (gross) wealth. The only exception to them is the four-parameter Dagum (1977, 1980a) model which extends the range of income to include the fitting of the total frequency of economic units with null and negative incomes. The fourth parameter a = F(0) = P(Y 2264; 0) estimates this frequency. Although the sample surveys of income distribution in the U.S.A. and Canada present a finite and relatively small frequency of economic units with negative and null incomes, it was not possible to model it because of the lack of information about this part of the distribution. Unlike the net wealth distribution, this was not a major issue because of (i) the observed frequencies of economic units with null and negative incomes are smaller than those of wealth distributions, and (ii) the fourth parameter a in Dagum model gives an accurate and statistically significant estimate of the cumulative frequency F(0).I am grateful to Dr. Luigi Cannari of the Servizio Studi of the Bianca d’Italia, for his important help in providing me with the Italian net distribution data. Research grants from the University of Ottawa (Canada) and the University of Sienna (Italy) are gratefully acknowledged.KeywordsIncome DistributionIncome ElasticityWealth DistributionTotal WealthEconomic UnitThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.