A tale of two tails

Income inequality is increasing. A very large lower middle class got squeezed while the top 10% gained enormously. The concentration of high paid jobs in the upper class, abysmal real wage growth, and a major shift in employment from industry to agriculture account for the squeeze. The increased income share of the top 10% was driven by return on capital beating the national income growth rate by a handsome and growing margin.   

The income Gini coefficient, a synthetic and omnibus measure of inequality, increased 3.5% to 0.499 from 0.482 in 2016. Where in distribution did the increase occur? Is it due to the relatively many poor and vulnerable individuals or the few wealthiest individuals? A detailed look at the data suggest the recent increase in inequality is a story of prosperity on steroids at the top 10% and the dehydration of income of a large lower middle class.

Most Recent Facts

Using statistical concepts such as deciles allows rigorous comparisons across time since they are defined exactly the same way. I follow Piketty (2014) in defining the middle class as 40% above the bottom 50% and the elites as the top 10%.  

An overwhelming 90% of the population lost share in total income pie. The income share of the bottom 50% declined 0.56 percentage points and the middle 40% lost 2.21 percentage points between 2016 and 2022 (Table-1). The former subsumes 0.33 percentage points gained by the bottom 20%, which coincided with a decline in poverty from 26.4% in 2016 to 18.7% in 2022. Such a coincidence has not always been the case. For instance, the income share of the bottom 20% decreased from 5.22% in 2010 to 3.85% in 2016 while poverty declined by about 7 percentage points.

Seven out of ten do not, on average, make anything close to the national income per capita (Table-2). Per capita income of the bottom 50th percentile (a proxy for median income) was 42% lower than national income per capita in 2022; another indicator of inequality increasing from 38% in 2016. The 70th percentile in 2022 had per capita income 16% lower than the national income per capita; their shortfall up from 9% in 2016. 

It is possible to distinguish a "lower middle class" comprising of the 31.3% population above poverty in the bottom 50% and 20% population in the next 40% with per capita income below national per capita income. Sandwiched between the impoverished and the rather steep ladder to prosperity from the 80th decile onwards, these 51.3% of the population had the lowest per capita real income growth.  

Most rich got richer. The top 10% gained the 2.7 percentage points that the bottom 90% lost. BBS does not provide data on income split within the top 10%.  According to the World Inequality Database (WID), the income share of the top 1% in 2020 in Bangladesh was 16.2%, constituting over 42% of the share of the top 10%. At the very top 1%, monthly income per capita was Tk3 lakh 28 thousand while that of the top 10% was a modest Tk43,735. The top 10 and 1% managed to grow income respectively at 4.9% and 12.4%, miles ahead of the 3.6% annual increase in real national income per capita.                             

In the most egalitarian countries, the bottom 50% receive nearly twice the total income of the top 10% (Piketty 2014). It is the opposite in Bangladesh. The top 10 to bottom 50 total income ratio was 2.2 in 2022, moving in the opposite direction from 2 in 2016. 

The HIES misses the super-rich proportionately more in Bangladesh. Per capita monthly income based on national accounts is 2.6 times the per capita income reported from HIES 2022. Estimates of average per capita income from national accounts generally differ substantially from household survey based income estimates. But the discrepancy between the two in Bangladesh far exceeds the upper 50% bound in the gap with GNI per capita observed internationally.

Why are differences so sharp and rising? 

Income has two parts: labor and capital income. The bottom 50% typically have little capital income, owning 4.8% of national wealth (WID). The next 20% own 11.2%. The change in their share in the pie primarily depends on their labor income growth. Overall, real labor income grew 3.8% during 2016-22, well short of the 6.4% annual output growth. This necessarily implies a decrease in labor income share, affecting the bottom 70 the most. A 0.7% real wage growth added feebly to a decent 3.1% employment growth (Table 3). 

Faster labor income growth in agriculture boosted the income share of the bottom 20. A decrease in industrial employment and weakest real wage growth hurt the income share of the lower middle class. Fast industrial growth did not benefit workers with no specialized technical skills. Textbook economic models generally suggest that productivity growth should ultimately translate into wage growth, but that has not been happening in industry and services. Labor income in services pulled the share down by growing slower than output. This too affected the income share of the lower middle class more than proportionately.  

Wage growth was modestly equalizing between 2003-16 with higher paying sectors experiencing lower growth rates. This subsequently changed in the opposite direction, sharpening the differences. The source of productivity growth and the manner in which wages are set matters. Using machines to do what workers used to do improve productivity. But employers and managers may choose to keep more of the gains for themselves, using automation to undercut workers' bargaining power whenever the institutional framework allows.

Why did the share of the top 10% rise?

Capital income is a dominant component of the total income of the top 10% who owned 58.4% of net personal wealth in 2022, of which half by the top 1% (WID). The concentration of wealth at the top 10 is 1.5 times the concentration of income.  

Piketty's definition of the share of capital income in national income as the product of the rate of return on capital and the capital-to-income ratio is a useful accounting frame for understanding the rise of income concentration. Share of capital income increases when the return on capital (r) exceeds the GDP growth rate. 

This "excess return" (r>g) has increased in Bangladesh. The income share of the top 10% looks like a reasonable proxy for capital income share. Data on the wealth income ratio is reported in WID (2023). These allow computing r. The results (Table-4), which only indicate orders of magnitude, show the rate of return is considerable. It increased from 13.3% in 2016 to 13.6% in 2022, well ahead of the 6.2% GNI growth rate. Excess return (r-g) increased by 30 basis points from 7.1% in 2016 to 7.4% in 2022. 

Historically, the hyper concentration of wealth in traditional agrarian societies was due to low growth (0.5-1% a year) with the rate of return on capital markedly higher (4 or 5%). Bangladesh recently has been different with higher growth, even higher and rising return on capital but the same result – rising inequality. What matters for capital income share is the excess, not r or g per se. 

How much this divergence translates into a higher wealth income ratio depends on what the capital owners do with the excess return. If they recapitalize all of it, their wealth ratio increases. Reinvesting 7% doubles the wealth ratio every ten years. Consuming half of 7% doubles wealth every 20 years. Typically, the propensity to consume capital income is below 0.5.
Excess returns (r-g) can be high for the wealthy without necessarily being high for the economy as a whole or all its parts. Bangladesh's agriculture during 2016-2022 had low growth and apparently even lower return on capital. Industry and modern services were the opposite where wealth concentration was larger to begin with. An unequal return on capital amplifies the inequality effects of the excess return on capital. 

A high r is hardly surprising in an economy with a wealth ratio 40-50% below the global 5-6 years of national income.  However, it is supposed to converge in an open economy as return seeking capital from abroad take advantage, driving r down. That has not happened in Bangladesh because of protection provided to domestic capital and reverse capital flows. 

The protection to domestic capital is provided through fiscal, financial, and regulatory instruments. Their application is differentiated according to the bargaining power of parties in the transaction. But the net effect is to shift up the distribution of r in favor of the protected or, if you will, connected. 

The missing wealth

The high excess return makes one wonder why the measured net wealth ratio (Table-4) increased so little, just 1% annually, in 2022 relative to 2016 notwithstanding r-g = 7.4%! If capital owners consumed half the excess returns, the predicted ratio for 2020 is 3.4 against the measured ratio of 3. 

The difference, equivalent to 43.2% of national income, cannot be because g is underestimated. It is also not due to underestimating the propensity to consume capital income. Even at 2% excess return net of 5% consumption, we have net wealth equivalent to 16% of national income missing. May be a significant part of the equivalent of $74-200 billion (based on $465 billion total GNI in FY22) is missing because they left the country under cover to safe havens. 

The undercover assumption is plausible in light of evidence on annual illicit outflows from the Global Financial Integrity Report ($5.5 billion on average during 2004-13). At this rate we would have illicitly exported $100 billion in the last 18 years. Chances are the annual average increased since 2013. 

Illicit capital flight is equivalent to consuming capital. However, the wealthiest not only take money, many change nationality. Domestic r rises as the availability of capital domestically is reduced. This slices off a few basis points from income growth. Capital flight nurses r>g by increasing r and decreasing g. The wealth ratio of those remaining rises, increasing their share in total national income.

Zahid Hussain, former lead economist, World Bank's Dhaka office.