2.1 Definition of Inclusive Growth and Pro-Poor Growth

As pointed out above, drawing from the recent literature on inclusiveness of growth, the Twelfth Plan Approach Paper

Economic & Political Weekly

EPW

includes several outcomes that would constitute “inclusive growth”. Inclusive growth presupposes growth in income. But not all growth scenarios are considered inclusive. Therefore, there is a need to differentiate the growth processes that are inclusive from those that are not. Some researchers argue that inclusive growth is broad-based and benefits everyone in society – the poor, middle income groups and even the rich (Klasen 2010). In this sense pro-poor growth, in which the focus of outcome is on the poor, constitute a subset of the broad concept of inclusive growth. For pro-poor growth, the “pro-poorness” is to be embedded in growth with a policy bias in favour of the poor (resulting in a relatively higher increase in the income of the poor). This signals a clear departure from the “trickle-down development” doctrine of 1950s and 1960s that meant a gradual top-down flow from the rich to the poor (Pernia 2003). Grinspun (2009) summarises the current debate on definition and measurement of inclusive growth. One view suggests that pro-poor growth is any growth in mean income that benefits the poor (Ravallian 2004).

Since this criterion would encompass a vast majority of growth episodes, alternatively it is proposed that for growth to be pro-poor, it should benefit the poor proportionately more than the non-poor. With this, the focus shifted to the extent of income gains of the poor from growth (Kakwani 2004). While agreeing that the true test of pro-poorness is the existence of a policy bias favouring the poor proportionately more, this needs to be measured with reference to the country’s past record of poverty reduction. By this definition for growth to be pro-poor, the increase in income in current period should be necessarily greater than the preceding period (Osmani 2005). The concept got further extended with the new criterion of pro-poorness, namely, that the share of the poor in the income growth in the current year should exceed their share in the previous year, the share of the poor in incremental growth surpasses their share in population and that the share of poor in incremental growth exceeds some international norms.

Since the pro-poorness involves income gains for the poor, the trade-off between income growth and its distribution becomes important for pro-poor outcome (Ahluwalia 1976, 1978). In this context pro-poor growth would necessarily involve growth with declining inequality in income distribution (Rauniyar and Kanbur 2010). Initial inequalities also matter – the rate of decline in poverty tends to be less pro-poor in a situation where initial inequality is high. Inequalities caused by social exclusion, discrimination, and constraints on human development particularly limit the prospect of poverty reduction among certain excluded groups (Ravallion 2009). Therefore, Klasen (2010) emphasised non-discriminatory participation and disadvantage reducing features as a necessary condition for inclusive growth.

2.2 Economic, Caste, Ethnic and Religious Groups

In the hierarchical structure of the Indian population, it is well documented that there are certain groups that lag behind on a range of development outcomes – income, poverty incidence, education, health, and so on – and the inclusive approach in the Eleventh and Twelfth Plans focuses on gains from growth for these groups (SRGs).1 We have identified economic as well as socio-religious groups in the NSS data and calculated the incidence of poverty and real mean monthly per capita consumption expenditure (MPCE) for the rural and urban areas. The economic groups in the rural sector are the self-employed in agriculture (farmers, SEAG), self-employed in non-agriculture (non-farm production and business, SENA), wage labour engaged in agriculture (AGLA) and wage labour in non-agriculture (OLAH) and households which have more than one income source (OTHER). For urban areas, the economic groups are the self-employed (SEMP), wage/salary earners (RWSE), casual labour (CALA) and other households (with multiple sources of income, OTHER).

Among the social groups, the identifiable groups from the data include STs, SCs, OBCs and higher castes (non-SC/ST). Among the religious groups, data reports household’s religious denomination as Hindus, Muslims and several Other Religious Minorities (ORMs). For purpose of analysis, we club Christians, Sikhs, Jains and other religious minorities into one group, ORM (Thorat 2010).2

2.3 Data and Poverty Lines

For measuring growth, incidence of poverty and inequality, unit record data from three quinquennial rounds of consumption expenditure surveys (CES), conducted by the National Sample Survey Offi ce (NSSO) have been used. These surveys were conducted during the agricultural years 1993-94 (July 1993 to June 1994), 2004-05 (July 2004 to June 2005) and 2009-10 (July 2009 to June 2010), respectively. The NSS in these surveys follows a stratifi ed sampling design and the weights or multiplier for the surveyed households are used in the calculations.

For calculating the incidence of poverty, we use poverty lines (PLs) published by the Planning Commission. These are the PLs originally derived by the 1979 Task Force (GOI 1979) and modifi ed by the 1993 Expert Group (GOI 1993)3 for calculating state-level PLs by adjusting for price variation across states. For the years 1993-94 and 2004-05, the state-wise PLs have been

march 10, 2012 vol xlvii no 10

ever, since the submission of the Re-

port of the Expert Group to Review the

(GOI 2009),4 the Planning Commis-

sion has not specified a set of poverty

lines for India and the states for

2009-10 so far.5 Consequently, we

have updated the poverty line of

2004-05 as reported in GOI (2007)

1993 Expert Group. Thus, the inci-

dence of poverty reported in this paper has been calculated using the “old offi cial poverty line”.6

The incidence of poverty is measured as the percentage of population below the poverty line, also known as the Head Count Ratio (HCR).7 In addition, the NSS CES data report consumption expenditure of the households in nominal rupees. We have converted the nominal expenditure at constant (1999-2000) prices. The price deflator that we used to convert the household expenditure at constant prices is the implicit price defl ator derived from the state-wise PLs for rural and urban areas separately. The growth of MPCE and the summary measure of inequality, Gini coefficient, has been calculated using defl ated MPCE data.

3.1 Aggregate and by Social and Religious Groups

Between 1993-94 and 2009-10, rural poverty declined at 2.5% per annum, which amounts to the decline by 15 percentage points (Tables 1 and A1). Across social groups, the rate of decline in rural poverty has been higher for the upper castes, followed by SCs and STs – the per annum decline being 2.7%, 2.4%, 2.1%, respectively. In the case of religious groups, rural poverty has declined at a higher rate for Muslims and ORMs as compared to Hindus – the per annum rate of decline being 2.4% (Hindus), 3.4% (Muslims) and 3.6% (ORMs). Thus, Muslims and higher castes and ORMs have done better compared to rest of the groups.

3.2 Changes during Two Sub-Periods

Rural poverty declined at the rate of 2.2% annually during the first period, 1993-94 – 2004-05. Among the SRGs, the per annum rate of decline has been the highest for upper castes and Muslims, followed by SCs and STs – the decline being the lowest for STs. Thus, during the first period, the upper castes and Muslims did better in reducing poverty as compared to the rest, with STs and SCs lagging behind (Table 1).

Economic & Political Weekly

EPW

In the second period, 2004-05 to 2009-10, there has been a significant acceleration in the annual rate of decline of poverty across SRGs (Table 1). Rural poverty declined 4.4% annually. The rate of decline is the highest for the STs (5.2%), followed by the upper castes (4.5%) and the SCs (4%) among the social groups. Among the religious groups, Muslims experienced the largest decline at 7.6% per annum followed by ORMs at 7.2% (Table 1).

NSS used codes for OBCs during the 2004-05 and 2009-10 surveys. The exclusion of OBCs from the upper castes shows improved performance of the forward castes (non-ST/SC/OBC) in reducing poverty, from 4.5% to 6.6% annually (Table A2). Thus, the overall ranking changes to some extent. In terms of per annum decline during 2005-10, Muslims, upper castes and STs have a higher ranking, the rate of decline being 7.6%, 6.6% and 5.5%, respectively, while the SCs lagged behind with a 4% decline.

3.3 By Livelihood Categories and Socio-Religious Groups

It is important to look at the livelihood categories as these are similar to the Indian socio-religious structure that traditionally put the households in different livelihood pattern.8

Table 2 (p 46) presents the changes in the poverty incidence for the economic groups by SRGs in the rural areas. This enables us to know the performance of self-employed households engaged in farm and non-farm activities vis-à-vis the wage labour households in reducing poverty in the rural areas.

Among the livelihood categories, the farm and non-farm wage labourers (AGLA and OLAH) are the most poor. In 2009-10, about 35% of AGLA and 26% of OLAH population were poor. In comparison, the poverty level for SEAG and SENA households is about 17%. So the HCR of the AGLA households is twice the SEAG households.

Table 2 shows that during 1993-94 – 2009-10, the annual rate of decline in poverty has been relatively higher for SEAG (2.8%) and SENA (2.9%), but relatively lower for OLAH (2.5%) and AGLA (2.3%). A similar pattern is observed in both the subperiods, except that while the poverty incidence among the SEAG and SENA households declined by the same rate (2.5% per annum) in the first period, in the second period the SENA households did better (5.5%) than the other households, namely, the SEAG, farm and non-farm wage labour (about 4.5%).

Table 2: Average Annual Change in HCR by Household Type and SRGs in the Rural Sector (in %)

SRG ST SC OTHERs Hindu Muslims ORM Total

1993-94 to 2004-05 SENA -1.9 -1.8 -2.8 -2.6 -2.2 -5.3 -2.5

AGLA -0.6 -1.9 -1.9 -1.7 -1.8 -2.3 -1.7

OLAH -1.1 -1.7 -2.4 -1.8 -2.7 -4.0 -2.0

SEAG -1.2 -2.5 -2.9 -2.5 -2.9 -1.6 -2.5

OTHER -1.1 -1.8 -2.0 -1.9 -2.1 -4.5 -1.9

All -1.1 -2.2 -2.5 -2.2 -2.5 -3.0 -2.2

2004-05 to 2009-10 SENA -5.9 -5.4 -5.5 -4.4 -8.2 -3.1 -5.4

AGLA -4.1 -4.3 -4.2 -3.4 -10.2 -7.4 -4.3

OLAH -4.8 -3.6 -4.7 -3.9 -7.0 -12.5 -4.4

SEAG -5.9 -2.0 -5.1 -4.6 -4.1 -1.1 -4.5

OTHER -7.9 -1.6 -10.1 -6.4 -10.4 -14.0 -7.8

All -5.2 -4.0 -4.6 -3.8 -7.6 -7.0 -4.4

1993-94 to 2009-10 SENA -2.8 -2.6 -3.1 -2.8 -3.5 -4.0 -2.9

AGLA -1.6 -2.3 -2.4 -2.0 -3.8 -3.3 -2.3

OLAH -2.1 -2.1 -2.7 -2.2 -3.4 -4.9 -2.5

SEAG -2.4 -2.2 -3.1 -2.8 -2.9 -1.4 -2.8

OTHER -2.9 -1.7 -3.8 -2.9 -3.9 -5.3 -3.2

All -2.2 -2.4 -2.8 -2.4 -3.4 -3.5 -2.6

However, the pattern varies across the SRGs within the economic groups. Taking the self-employed agricultural households first, during 1993-94 – 2009-10, poverty declined among all social and religious groups, but at a lower rate for the SCs and STs as compared to upper castes and Muslims.

There are important differences in the rate of decline in poverty in SEAG households between the two sub-periods. During the first period poverty among the ST farmers declined at a much lower rate as compared to SCs, upper castes and Muslims (although upper castes and Muslims did better than SC self-employed households). However, the story changes completely during 2005-10. The ST SEAG households experienced significant acceleration in poverty reduction, while the upper castes and Muslims continued to do better. The SC farmers lagged behind in reducing poverty in the second period.

In case of SENA households, during the overall period of 1993-94 to 2009-10, all SRGs experienced decline in poverty. However, upper castes and Muslims benefited more from the growth in the rural non-farm sector, and the SCs and STs lagged behind. A similar pattern is ob

households declined at the lower rate than the self-employed households. Among the farm and non-farm wage labour households, poverty declined at a higher rate among the upper castes and Muslims and at a lower rate among SCs and STs.

During the first period (1993-2004), the per annum decline in case of AGLA households was particularly low among the STs (0.6%), with minimum differences in case of other groups (the rate varies between 1.7% and 1.9%). In the case of OLAH households, both the STs and SCs performed poorly in reducing poverty as compared to others. The upper castes and Muslims did relatively better. Thus, SC and ST households in farm and nonfarm wage labour lagged behind in reducing poverty as compared to other groups during 1993-2004. During the second period (2004-10), the situation in the farm wage labour had significantly improved for all the groups, including the STs who had performed badly in the first period. The rate of decline has been particularly high for Muslims (10.2%) (Table 2). A similar acceleration in poverty reduction was experienced by all groups for OLAH households as well. The acceleration was of higher magnitude for upper castes (4.4%) and Muslims (7%). The rate of decline was relatively low for the SCs (3.6%) (Table 2).

3.4 Urban Poverty

4.1 Temporal Change in MPCE

Since, a comparable household income distribution data is not available; in this section we examine the changes in real MPCE for the SRGs and economic groups. Figures 1 and 2 show the distribution of MPCE at real (1993-94) prices at three time points for the rural and urban areas, respectively. First, there is

Economic & Political Weekly

EPW

march 10, 2012 vol xlvii no 10

discernible shift in the distribution towards right in both the fi gures. Second, given the nature of the distribution in both the sectors, the use of “a particular” poverty line is unlikely to affect the changes in poverty incidence over time or comparison of poverty incidence across SRGs or economic groups.

The scrutiny of the two distributions also suggests that in the rural areas the change in MPCE between 1993-94 and 2009-10 or for that matter between 2004-05 and 2009-10 is moderate. In case of urban sector, though, it is far more pronounced and there is significant shift to the right. Another noticeable feature is a spike in MPCE among the higher income group of households.

4.2 Changes in Real MPCE-Rural

In Table 4, the growth of real MPCE at an aggregate level and across SRGs is reported. The MPCE increased at a per annum rate of 1.5% during 1993-2010. The per annum rate varies among the groups in a narrow range of 1.3% to 1.8%, with ORMs showing the highest per annum increase.

During 1993-94 – 2004-05, MPCE increased at the rate of 1.3%. The STs experienced the lowest annual increase at 0.5%, followed by 1.3% for the SCs and 1.4% for the upper castes. Among the religious groups, Muslims experienced a higher increase (1.7%). In the second period, rate of growth of MPCE increased to 1.7%. Among the social groups, STs showed marked improvement in the rate of increase from 0.5% to 3%. In case of other groups, the per annum increase was more or less similar and varied in a narrow range between 1.6% for SCs and 1.8% for Muslims (Table 1).

Exclusion of OBCs (Table A2) from the OTHERs has accelerated the per annum increase in the MPCE for upper castes from 1.7% to 2.1% during 2005-10, the second highest after the STs. The rate of increase for OBCs is close to average for all (1.5%).

4.3 Changes in MPCE by Livelihood Categories

Across the economic groups, the growth of MPCE is reported in Table 5. At the aggregate level, among the five household types in rural areas, SENA households (engaged in non-farm production and business activities) experienced a relatively higher increase in MPCE (1.7%) and for the rest it is around 1.3% per annum. A similar pattern is observed during 1994-2005, MPCE grew at a higher rate for SENA (1.6%) and at a lower rate for farm and non-farm wage labour and self-employed in agriculture (between 0.9% and 1.1%).

During 2005-10, all households experienced an acceleration in the growth rate. While SENA households have a similar rate of growth as in the first period, for the SEAG and wage labour households, the growth rate accelerated. During 2005-10, the rate of increase is higher for wage labour households.

For the SRGs across the livelihood groups during 1993-94 – 2009-10, SENA households have done better compared to rest of the households. It emerges that STs and upper castes have recorded higher growth of MPCE compared to other SRGs.

During the two sub-periods, we find some differences in the rate of change in MPCE for SENA households across SRGs. During 1994-2005, for SENA households, the per annum increase is higher for Muslims (2.0%) and upper castes (1.8%)

48

Table 5: Growth of MPCE at Constant (1999-2000) Prices by Household Type and SRGs in the Rural Sector

SRG ST SC Others Hindu Muslims ORM Total

1993-94 to 2004-05 SENA 0.9 1.0 1.8 1.4 2.0 4.2 1.6

AGLA 0.2 1.1 0.9 0.8 1.2 1.2 0.9

OLAH 0.1 1.1 1.1 0.8 2.0 2.3 1.1

SEAG 0.6 1.1 1.1 1.0 1.2 1.6 1.1

OTHER 1.0 1.7 2.2 2.0 2.4 2.1 2.0

All 0.5 1.3 1.4 1.2 1.7 2.1 1.3

2004-05 to 2009-10 SENA 4.7 1.9 1.6 2.0 0.3 0.8 1.7

AGLA 2.8 1.7 2.0 1.8 4.4 2.4 2.0

OLAH 2.7 0.6 2.1 1.2 2.1 12.1 1.8

SEAG 2.4 1.8 1.6 1.4 1.6 3.0 1.6

OTHER 2.6 0.6 3.2 1.8 4.8 4.4 2.5

All 3.0 1.6 1.7 1.5 1.8 4.3 1.7

1993-94 to 2009-10 SENA 2.2 1.3 1.8 1.7 1.5 3.2 1.7

AGLA 1.0 1.3 1.3 1.2 2.4 1.7 1.3

OLAH 0.9 1.0 1.5 0.9 2.2 6.4 1.3

SEAG 1.2 1.4 1.3 1.2 1.4 2.2 1.3

OTHER 1.5 1.4 2.7 2.1 3.5 3.1 2.3

All 1.3 1.5 1.6 1.4 1.8 3.1 1.5

but lower for SCs (1.0%) and STs (0.9%). In case of SEAG households, the per annum rate varies minimally across the social groups, with the only exception being STs (0.6%).

The 2005-10 period has been a high growth one for MPCE across all the SRGs. In the case of SENA households, the STs showed signifi cant acceleration in the rate, followed by the SCs; while among the religious groups, Muslims and ORMs show a deceleration. The SEAG households have experienced acceleration in the growth of MPCE during 2005-10 for all social groups; the improvement has been particularly high for the STs (Table 5). STs, therefore, seem to be doing better in case of both farm and non-farm self-employed households during the second period.

During the period 1994-2010, the MPCE of farm wage labour households grew at 1.3% annually, except for Muslims (2.4%). During 1994-2005, the MPCE of the AGLA households increased at a relatively lower rate of 0.9%. However, the growth accelerated to 2% during 2004-10. Other SRGs report substantial acceleration in the MPCE, which has been particularly higher for STs and Muslims.

In the case of OLAH households, MPCE increased by 1.3% annually. However, growth has been lower for the SCs (1%) and STs (0.9%) as compared to Muslims (2.2%). During 1994-2005, growth of MPCE is higher for Muslims and much lower for STs. During 2005-10, the growth rate accelerated from 1.1% to 1.8%. This acceleration in MPCE growth has been felt across all social groups, with the exception of SCs, for whom it decelerated from 1.1% to 0.6%.

4.4 Growth in MPCE in the Urban Sector

During 1994-2010, the MPCE in the urban sector grew at the rate of 2.4% per annum, which is significantly higher than that observed in the rural sector. Among the social groups, the

march 10, 2012 vol xlvii no 10

After having analysed the changes in poverty and MPCE, in

cise are reported for two time periods, 1993-94 to 2004-05 and

Household Type ST SC OTHERs Hindu Muslims ORM Total

three social and three religious groups for the rural and urban

1993-94 to 2004-05

sector separately in Table 7.

CALA 0.2 0.6 1.0 0.8 1.1 0.5 0.8

Socio-Religious Rural Urban

OTHER 3.4 2.9 4.1 3.8 3.4 6.9 4.1

Hindu -8. 2 -10. 5 2. 3 -6. 6 -10. 7 4. 1

SEMP 3.0 1.9 2.3 2.3 1.6 2.7 2.2

Muslim -11. 0 -14. 9 3. 9 -7. 2 -11. 9 4. 8

RWSE 3.8 2.0 2.9 2.8 2.9 2.5 2.8

OTHER RM -7. 9 -11. 4 3. 6 -7. 6 -8. 8 1. 2

CALA 1.1 1.3 2.0 1.7 2.3 1.6 1.8

2004-05 to 2009-10

OTHER 2.8 3.3 3.6 3.4 3.3 3.8 3.5

ST -13.67 -13.17 -0.50 -5.89 -12.04 6.15

The pattern varies across SRGs. The RWSE households

Hindu -5.93 -5.23 -0.70 -4.62 -6.53 1.91

appeared to be better off during 1994-2005, the growth being

Muslim -5.86 -2.27 -3.59 -7.74 -9.88 2.15

relatively higher for the STs and upper castes, but lower for SCs

OTHER RM -4.61 -5.13 0.53 -0.73 -3.06 2.32

and Muslims. In the second period, there is a signifi cant

As in Table 1.

Source: As in Table 1.

increase in the rate from 1.8% to 4.1%. The increase is relatively

Economic & Political Weekly

1993-94, while for OTHERs it should have declined by 11%. But because of the changes in the distribution, 2.4% for SCs and 3% for OTHERs were offset. In the case of STs, the loss due to distribution is about 4.6%.

Among the religious groups, the highest decline in poverty, 'H, has been for Muslims (11%), followed by 8.2% and 7.9% for Hindus and ORM, respectively. However, because of growth alone, these figures should have been 14.9%, 10.5% and 11.4%, respectively. But because of the unfavourable changes in the distribution, close to 4% for Muslims, 2.3% for Hindus and 3.6% for ORMs have been offset. Thus during the fi rst period, the loss due to distribution has been high for STs, Muslims and others among the social groups.

During 2004-10, the rural sector presents an interesting picture. In almost all the cases – for the entire rural, all the social groups (STs, SCs and OTHERs) and Hindus and Muslims in religious groups – the decline in poverty has been more than what could have been realised because of the growth effect. The decline in HCR by 13.7% for STs, 7.1 for SCs, 4.60 for others, 5.93 for Hindus and 5.86 for Muslims is higher than the decline expected because of growth. The distribution effect also contributed in reducing poverty albeit moderately (less than a per cent), except in the case of Muslims, where the contribution of distribution has been about 3.6%.

In the urban sector, during 1993-2004, because of the growth effect, the HCR should have declined by about 11% for all the SRGs, but lost out by 4.3%. Because of distributional changes, Muslims lost out by 4.8%, Hindus by 4.1%, upper castes by 4.5%, STs by 2.8% and SCs by 2.2%. Though there are modest changes in the inequality measure, the Gini coeffi cient, during this period (Table 3), the changes in distribution seem to have played a role in decelerating reduction in poverty incidence in the period, 1993-2005.

During 2004-10, the role of distribution is reversed. Because of growth, the overall decline in urban poverty should have been about 6.8%, but close to 1.9% were offset because of changes in distribution. Among the social groups, the highest reduction in poverty should have been for STs by over 12%, but more than half of it was offset because of worsening of the distribution among the ST households. The observed 'H is lower by about 1.8% for both SCs and upper castes among the social groups. Among the religious groups, Muslims had the highest 'H at 7.7% but because of growth alone, their poverty should have declined by about 9.9%. Among the large population groups, after STs, they have been the biggest loser in poverty reduction because of distributional changes. The Gini coeffi cient too grew during the second period by similar magnitude for all the SRGs, the effect of the rise in inequality on reduction of poverty seems to be different.

In Tables 1 and 3 we report the summary measure of inequality, Gini coefficient, calculated for each SRG separately. It shows only marginal increases during the two sub-periods, the adverse effect of inequality do seem to have affected the potential of growth in reducing poverty during the fi rst period where the increase in the Gini coefficient in both the rural and urban sectors has been relatively higher. It seems the reason for inequality increase is because of increase in the MPCE in the higher income groups (Figures 1 and 2). During the second period, the increase in Gini coefficient is comparatively lower in both rural and urban areas. The higher MPCE classes have experienced higher growth (Figures 1 and 2), but other interventions (such as wage employment, wage level and food security measures) would have helped those around the poverty line to cross it with marginal gains in their expenditure in the rural areas (as distribution effect is contributing in reducing poverty). In the urban areas the adverse effect of inequality appears to have moderated the distribution effect.