Publication Date



A sample of 860,000 individuals from the 1973 Colombia Census Population is used to study income determinants and income inequality. Men and women are analyzed separately, as are employees and employer. Within these groups, education, age, region, and rural/urban differences in income are distinguished using a variety of procedures including simple cross tabulations and decompositions of the log variance of income by analysis of variance and by regression techniques. By standard statistical conventions, the four way classification by educational attainment is much the most important determinant of the logarithm of monthly income, while the seven age categories are generally somewhat more significant than the six regions. The fourteen parameters used to model these main effects account for a third of the log variance in income of employees and a quarter of that of employers. Each year of schooling is on the average associated with about 20 percent more income for male employees and employers. The restricted specification of a conventional earnings function increases the standard error of estimate by only .1 percent. Within education and age classes relative dispersions of incomes across regions are larger for the less educated, and for the very young and old.


Suggested Citation
Fields, G. S. & Schultz, T. P. (1977). Sources of income variation in Colombia: Personal and regional effects [Electronic version]. Washington, DC: U.S. Agency for International Development.