The paper investigates the robustness of some official estimates which show a dramatic decline in poverty in Bangladesh in the 1980s. HES data of the Bangladesh Bureau of Statistics shows a real per capita consumption growth rate of 9.9% per annum from 1981/82 to 1985/86. But, the paper estimates the growth rate of per capital real consumption to be only 0.5% per annum based on National Accounts (NA) data. The rate of growth of real consumption per capita based on HES data appears to be highly exaggerated. It true, it would be the highest rate of per capita real consumption growth for any country in the world during this period.  

Three alternative measures of poverty belonging to the FGT class of additively decomposable poverty measures are used in the analysis. The results show that an implied growth rate of 0.5% per year in real per capita consumption is the much more plausible estimate of the true rate. Assessments of growth during the 1980s consistent with national accounts data suggest that the proportion of poor population has remained fairly stable in recent years while absolute numbers of poor have increased. The paper shows that the growth rate of real consumption per capita has to be at least equal to the rate of population growth before the absolute number of poor can start to decline appreciably without a shift in relative inequalities. Further, measures of poverty particularly those that give more weight to the poorest of the poor, and those that use a lower poverty line would reveal greater proportionate impacts of expected economic growth in Bangladesh.