2
Changing Contours of the Economic Divide In Sri Lanka: The Post–liberalization EXPERIENCE 1977-2004
Upali Vidanapathirana
Department of Social Studies
The Open Universtty of Sri Lanka
Abstract
The economic divide is inimical to long-term development. Experience of South Korea and Taiwan in the 1970s demonstrated how different combinations of policies engendered equality. Similarly, lessons drawn from Sri Lanka, and the Kerala State of India in the 1960s supported this claim further. However, the economic policy reversals in 1977 changed the public action based social welfare gains in Sri Lanka significantly. They worsened the economic divide continuously; its impact covered both income inequality and wider forms of capability deprivation. The findings of this paper refute the ‘hypothesis of convergence’ in the case of Sri Lanka. The paper identifies varying dimensions of the economic divide with a view to ascertain its underlying causes.
Introduction
Inequality is inimical to long-term development; Hence, redistribution of assets, income and opportunities functions as an important policy instrument of ensuring ‘inclusive’ growth.[1] The experience of South Korea and Taiwan and lessons drawn from Sri Lanka, Cuba and the Kerala State of India in the 1970s showed how specific policy initiatives of the state and/or programmes of ‘public action’ mitigate inequality implications of growth (Birdsal, 1995; Wade, 1990; Sen 1981 & 1988). Nonetheless, Kuznets in the early 1950s argued that income inequality tends to increase in the early stages of growth although it eventually converges in economies that grow fast (Kuznets, 1955). The hypothesis of convergence in income distribution however became somewhat debatable in more recent literature (Barro 1992; Ravallion, 2003). The post-reform experience of growth and redistribution of growth dividends in Sri Lanka has refuted the two views stated above. This paper uses the theoretical and empirical underpinnings of the arguments outlined above to revisit the impact of the economic liberalization of 1977 on changing the contours of the economic divide in Sri Lanka.
The objectives of this paper are to assess how the reforms of 1977 affected the patterns of the economic divide across different income deciles, sectors and regions and its wider implications on the well being of the people of Sri Lanka.[2] The tools of analysis of the divide include the decile distribution of income, Lorenz curve, Gini coefficient and Theil index among others. This paper is primarily based on secondary information elicited from two survey series i.e., the Consumer Finances and Socio Economic surveys (CFSES) and the Household Income and Expenditure (HIES) Surveys. The paper uses an eclectic approach as the causes and consequences of the economic divide cover domains other than income and expenditure. The paper focuses largely on the first phase of reforms (1979-1989) and the second phase of reforms (1990-2004) although data pertaining to the early 1970s are also used sparingly to gain a comparative insight.[3]
Presentation and Analysis of Income Data
The post-reform score-card of Sri Lanka reveals that the share of household income of the bottom most quintile declined persistently since 1978/79 except for a temporary turnaround in 1996/97. The income share of the bottom-most quintile that was 7.17 percent in 1973 declined to 5.02 by 2004, a fall of more than 30 percent. Much of this decline occurred during the height of the first phase of reforms (1981/82-1986/87). The top-most quintile on the other hand increased its share from 42.95 percent in 1973 to 52.06 (21.percent) by 12003/04.
Table 1-Quintile distribution of spending units[4] for 1973-2003/04 using income data
Quintiles / 1973 / 1979 / 1982 / 1987 / 1997 / 20041st quintile / 7.17 / 6.26 / 5.72 / 5.06 / 5.74 / 5.02
2nd quintile / 12.12 / 10.38 / 9.53 / 9.08 / 10.01 / 9.13
3rd quintile / 16.2 / 13.79 / 13.35 / 13.38 / 14.15 / 13.36
4th quintile / 21.56 / 20.49 / 19.4 / 20.09 / 20.78 / 20.43
5th quintile / 42.95 / 48.94 / 51.96 / 52.3 / 49.31 / 52.06
Source: Based on the Report on the Consumer Finances and Socio-economic Survey, 1999; 75 and 2005 Part II
The data on quintile distribution is graphically presented in Figure 1 of this paper. The income share of the top 40 percent that stood at 64.5 1973 increased to 72.5 by 2004. Conversely, the income share of the bottom 40 percent declined from 19.3 in 1973 to 14.2 by 2004. This trend refutes the convergence of income hypothesis promoted by the pro-reform lobby.[5] In fact, the income share of the top 40 percent which was about 3.3 times higher than the bottom 40 percent in 1973 increased to almost 5.12 times by 2004.
Figure 1:- Income distribution trends by quintiles for 1973-2004
Source: Based on the Report on the Consumer Finances and Socio-Eonomic Survey, 1999; 75 and 2005 Part II
One of the most widely used techniques of income distribution is the Lorenz Curve (LC).[6] It graphically presents the patterns of income distribution by deciles. The subtle differences in the inequality curves of 1982, 1986, 1996 and 2004 make the Lorenz curves overlap, thus making the picture crowded. Therefore this paper selected curves for 1973, 1978/79 and 2003/04 to highlight the distinct changes.
Figure 2:-Lorenz curves for 1973, 1978 and 2004
Source: Computed using data from Table 3.1 above.
The 1973 curve represents the state of income distribution during the pre-reform era: it was relatively closer to the line of equality. The conditions changed for the worse in 1978/79 in the aftermath of reforms where the income distribution curve shifted outward; the outer-most curve depicts a further deterioration of inequality in 2004. These curves underline the pattern of income distribution that was skewed towards the richer quintiles. Table 2 below summarizes the above trends in terms of Gini ratios, decile distribution ratios and Theil Indices compiled for the period 1973-2004. It cogently supports our basic tenet that income distribution trends worsened continuously during the post-reform period. Accordingly, the Gini ratio that stood at 0.35 in 1973 increased to 0.43 by 1978 and reached the level of 0.46 by 2003/04.[7] All the other measures of inequality listed in the paper confirm that the economic divide widened during the post-reform era.
Table 2- Other Measures of Income Distribution for 1973-2003/04
Description / 1973 / 1978/79 / 1981/82 / 1986/87 / 1996/97 / 2003/04Gini ratio / Spending units / 0.35 / 0.43 / 0.45 / 0.46 / 0.43 / 0.46
Income receivers / 0.41 / 0.50 / 0.52 / 0.52 / 0.48 / 0.50
Quintile distribution / (Q1 +Q2) /Q5 / 0.45 / 0.32 / 0.29 / 0.27 / 0.32 / 0.27
Q5 /Q1 / 5.9 / 7.8 / 9.0 / 10.3 / 8.6 / 10.3
Theil Index / 0.27 / 0.35 / 0.39 / 0.39 / 0.33 / 0.36
Source: Computed using data from the Central Bank of Sri Lanka (2004), Sri Lanka, Key Economic Indicators, p. xx and Table 6.2.
Presentation and Analysis of Expenditure Data
It is claimed that expenditure data portray income distribution trends better because income data are affected by a combination of reporting biases where the rich tend to under-report income to evade tax obligations while the poor tend to under-report income to gain eligibility for social security benefits. However, the expenditure data too are not altogether free of reporting biases. Here the rich-poor gap tends to fall because the poor over-report their expenditure on food while the rich under-report their non-food expenditure.[8]
Table 3 below presents data on the distribution of expenditure for 1980/81-2002. Although the picture that emerges from these data sets is somewhat dissimilar to that of the CFSES series, the argument that income inequality worsened during the post-reform era is corroborated by the evidence presented. The table shows that the expenditure share of the bottom quintiles fell continuously paving the way to increase the expenditure share of the top-most income quintiles. The only exception for this trend was 1990/91 where inequality fell in response to the poverty alleviation programme of 1989.[9]
Table 3- Expenditure share by quintiles and income growth by quintiles
Year / 1980/81 / 1985/86 / 1990/91 / 1995/96 / 2002Bottom / 8.915 / 8.498 / 8.88 / 8.434 / 7.708
2 / 12.98 / 12.53 / 12.9 / 12.18 / 11.54
3 / 16.45 / 16.24 / 16.47 / 15.68 / 15.44
4 / 21.29 / 21.45 / 21.52 / 21.12 / 21.54
Top / 40.37 / 41.28 / 40.23 / 42.58 / 43.77
Income growth / 1980-85 / 1985 -1990 / 1990-95 / 1995- 2002
Bottom / -4.62 / +4.49 / -5.0 / -8.56
2 / -3.46 / +2.95 / -5.58 / -5.25
3 / -1.27 / +1.41 / -4.79 / -1.64
4 / +0.75 / +0.32 / -1.85 / +1.98
Top / +2.25 / -2.54 / +5.84 / +2.79
Gini ratio / 0.31 / 0.33 / 0.31 / 0.34 / 0.36
Theil Index / 0.18 / 0.2 / 0.17 / 0.21 / 0.23
Source: Department of Census and Statistics, LFSS (1980/81, 1985/86), HIES (1990/91, 1995/96, and 2002).
Table 3 also shows the Gini and Theil ratios computed using HIES data series. One conspicuous feature is that the Gini and Theil ratios hold lower values than that of the ratios developed based on the CFSES series. However, the declining trend of poverty and inequality reported in 1990 was reversed in 1995/96 indicating that the SAP has more than offset the beneficial effects of the poverty alleviation programme.
Figure 3-Findings of the CFSES and HIES data series compared
Source: HIES and CFSES data series.
Figure 3 compares the trends of Gini ratio compiled using the CFSES and HIES databases for the entire period. Three caveats deserve to be noted at this stage. First, the two surveys have been conducted at different time intervals which makes comparison problematic. Second, there is no regularity in terms of the years especially with respect to CFSES surveys. Third, the Gini values for the two different data sources show a significant gap. Subject to these caveats the following section examines some of the salient features emanating from Figure 3.
First, the gap between income and expenditure measures of inequality was low during the 1970s; it widened towards the latter part of the 1980s. Second, the CFSES series has shown a relatively persistent trend of increase in Gini values where it increased from 0.35 in 1973 to 0.46 by 2004. This compares with the Gini values of the HIES data set which was 0.31 in 1980 but showed a trend of worsening only after 1990. It reached 0.36 by 2002.[10] Third, it is argued that the CFSES data set depict the effects of redistributive policies of the early 1970s and the adverse impact of reforms after 1977 although the HIES data especially for 1980/81 do not support this argument.[11] Fourth, in the case of the CFSES data much of the inequality occurred in 978/79 while in the case of the HIES series the much of the worsening of inequality occurred in 2002.
Notwithstanding these differences in the two data sets, the overall picture that emerges from figure 3 agrees with the main tenet of this paper, i.e., income inequality did not show signs of convergence after the reforms; conversely the divide was widening. More significantly the inequality trends of expenditure data continued to deteriorate in the new millennium.
An underlying assumption of the convergence hypothesis is that income levels of the poor should grow at a faster rate than the income levels of the rich. It is therefore important to examine briefly the trends of household income by quintiles and also by sectors. With respect to the trends of household income, the CFSES data showed that the mean income of households in real terms have eroded both in the 1980s and 1990s. For instance, the rate of change of mean income of the bottom most decile has been negative for the period 1981/82-1986/87 (CFSES, 1999, Table. 5.7). The extent of erosion of real income recorded during this period was Rs.36 per month per household (-6%). A similar pattern of erosion of real income was evident for the bottom most quintile for the period 1996/97-2003/04 where it fell by Rs. 117 (-4.3%).
Similarly, this paper anchored the mean income of households of the estate sector in real terms in 1973 at Rs. 100, to compare the ratio of real income of the rural and urban sectors that were as high as Rs.197 and Rs 267 respectively. It shows that the real income of the urban sector even in 1973 was substantially higher than the estate sector. However, the index of per-capita earnings for the urban sector increased continuously from 275 (1978/79) to 368 by 1996. The estate-urban gap in terms of household earnings was widening persistently. The pattern that emerged indicates that the income levels of the poorer households by quintiles and sectors have plummeted during this period.
Figure 4- Sectoral Income Widening for 1973- 2003/04
Source: CFSES data, various reports.
These differences in mean income confirm that both inter and intra-sectoral income inequality has widened over the years. Table 5 below shows how these differences are reflected in the Gini ratios of inequality. Firstly, the pattern of worsening intra-sectoral income inequality was more evident for the urban sectors. The Gini ratio for the urban sector rose from 0.4 in 1973 to 0.56 in 2003/04. Ironically, the sector that has recorded the highest rate of income growth and poverty reduction has strangely become the sector that has recorded the highest level of income inequality.[12]