Making sense of the projected poverty decline
The projected 1.3 percentage point decline in poverty is most likely an upper bound
What does the estimated 1.3 percentage point reduction in poverty headcount rate in FY19, reported officially on December 19, really represent?
It is a projection based on the estimated 8.15 percent GDP growth rate and the observed responsiveness of the poverty headcount rate to the GDP growth rate. The latter, technically known as the elasticity of poverty reduction to GDP growth per capita, is the percent reduction in poverty divided by GDP growth per capita. In the official poverty projection, it is assumed to be the same as experienced during 2010-2017. So, the projected decline in poverty headcount is as good as the plausibility of the assumption that recent history on the poverty-growth nexus will repeat itself and the quality of the GDP growth estimate.
History does not provide much comfort in assuming a stable relationship between GDP growth and poverty reduction over time. This is evident from the fact that GDP growth accelerated while the pace of poverty reduction slowed in the past decade relative to the decade preceding it. The amount of poverty reduction, based on the upper poverty line, each percentage point of growth per capita delivered since 2010 fell from 0.88 to 0.73, according to the World Bank Poverty Assessment 2019.
In general, the elasticity of poverty reduction to growth per capita is higher at lower levels of poverty. The reason is partly arithmetic: it is easier to halve the poverty rate when going from, for example, 10 percent poverty (requiring a 5-percentage point reduction) than from 60 percent poverty (requiring a reduction of 30 percentage points). So, Bangladesh's progress in reducing poverty cannot explain the elasticity decline.
The decline is better explained by increased income inequalities. This is evident from the 5.2 percent increase in the Gini coefficient of income—a composite measure of income inequality—in 2016 relative to 2010. More striking is the fall in nominal income per household in the bottom 5 percent from Tk 5,149 per month in 2010 to Tk 4,610 per month in 2016 while income per household in the top 5 percent increased from Tk 35,695 per month in 2010 to Tk 45,172 per month in 2016, according to BBS's Final Report on Household Income and Expenditure Survey 2016.
Feeble wage and employment growth explain why income inequality increased. Employment growth declined from nearly 3 percent per annum in the decade preceding 2010 to about 1.7 percent annum during 2010-16. According to ILO's Global Wage Report 2018/19, real wage in Bangladesh grew by 3.4 percent annually during 2008-17, compared with 5.5 percent in India, 4 percent in Sri Lanka and 4.7 percent in Nepal.
Then, how reasonable is it to assume that the elasticity of poverty reduction has remained stable during last three years? We will not know definitively until the results of the next survey are available. However, there is an indirect way to gauge by looking at the distribution of GDP growth between labor and capital contributors to production. Note that capital includes all forms of explicit or implicit return-bearing assets: housing, land, machinery, financial capital in the form of cash, bonds and shares, and intellectual property. In jargon, such a breaking up of GDP growth is known as the "functional distribution" of growth.
Constructing the functional distribution requires data on nominal wage growth, employment growth, inflation and the share of labor in total income. These enable ball park calculations. The results (using BBS data on nominal wage growth and inflation and assuming employment growth to be the same as reported in the BBS's 2017 Labor Force Survey as well as 50 percent labor share in total GDP based on international data) are presented in the table below:
Distribution of growth between labor and capital
|Labor income growth||3.3||2.9||3.1|
|Capital income growth||11.3||12.8||13.2|
Source: Author's calculation based on BBS data
The growth of capital income far exceeded the growth of labor income and GDP growth. Capital income growth was 3.4 times labor income growth in FY17. This multiple increased to 4.3 in FY19. In his book Capital in the 21st Century, Thomas Piketty demonstrated that if the rate of return on capital remains permanently above the rate of growth of the economy, it might drive the share of capital in national income sky-high. As the share of capital in national income increases, not only do capital owners become richer, but more remain for them to reinvest unless they consume the entire return from their capital. The increased saving in turn makes the growth rate of capital exceed further the growth rate of national income. Thus, not only does higher capital lead to higher share of capital owners in national income but higher share of capital owners in total income leads to higher capital.
Income distribution is therefore likely to have worsened further. That capital incomes are more concentrated than incomes from labor is an uncontroversial fact. As a result, the personal income distribution i.e. the division of income among individuals, becomes more unequal as revealed from the analysis of HIES 2010 and HIES 2016/17 data. Such a trend appears to have continued since the last survey, thus weakening further the relationship between poverty reduction and growth.
It is therefore implausible to assume that the elasticity of poverty reduction has remained stable.
The official projection of the poverty decline is still useful. But the caveats must not be overlooked. Assuming the recent observed relationship to hold in the short run is not a bad simplification. Any other assumption will essentially be arbitrary. The projected decline in poverty reduction is therefore a useful indicator of the magnitude of poverty reduction that can be expected from 8.15 percent GDP growth.
Given the "evidence" on the distribution of growth in the table above, 1.3 percentage point reduction estimate is most likely to be an upper bound. All the more so when the plausibility of the growth estimate is also considered. The debate on the consistency of the latter with growth related indicators such as exports, capital machinery and intermediate inputs imports, remittance growth, private credit growth, rise in non-performing loans and low tax revenue growth has not gone away.
The author is an economist.