Methodology for the evaluation and adjustment of income distribution data

OF THE THREE, INCOME TAX DATA ARE THE LEAST USEFUL. TAXPAYERS NORMALLY CONSTITUTE LESS THAN 10 PERCENT OF THE POPULATION; THEIR INCOMES LIE IN THE UPPER END OF THE DISTRIBUTION; AND DECLARED INCOME CAN CONSTITUTE AS LITTLE AS 10 PERCENT OF ACTUAL INCOME BECAUSE OF EVASION, FRAUD, AND PREVAILING FORMS OF PRODUCTION. THUS PERIPHERAL DATA SOURCES MUST BE USED TO BUILD UP DISTRIBUTION DATA "FROM THE GROUND UP." BUT THE PERIPHERAL DATA MUST BE EXAMINED FOR ITS RELIABILITY, COMPLETENESS, AND DETAIL BEFORE IT IS USED AS A MEANS OF ARRIVING AT ACCURATE INCOME DISTRIBUTION ESTIMATES. SURVEY DATA MUST BE ASSESSED FOR ITS SAMPLING PROPERTIES AND POSSIBLE SAMPLING ERRORS. NON-SAMPLING ERRORS MAY BE LARGE. DEFINITIONS OF RECIPIENT UNIT, INCOME, AND EXPENDITURE MUST BE EXAMINED FOR THE POSSIBILITY THAT ADJUSTMENT TO ANOTHER TYPE OF UNIT WILL FACILITATE USE OF INDIRECTLY RELATED DATA AS A CROSSCHECK. EXPENDITURE DATA MUST BE SUFFICIENTLY DETAILED TO PERMIT ITS ACCURATE CONVERSION INTO INCOME DATA. THE POSSIBILITY THAT PRICES FACING DIFFERENT INCOME GROUPS HAVE CHANGED AT DIFFERENTIAL RATES MUST ALSO BE EXPLORED.