Income and Income Inequality among Indian Rural Households

Preparing to load PDF file. please wait...

0 of 0
Income and Income Inequality among Indian Rural Households

Transcript Of Income and Income Inequality among Indian Rural Households

Income and Income Inequality among Indian Rural Households
This chapter analyses the changes in sources of income and income inequality of rural households using the Indian Human Development Survey (Data) of 2004-05 and 2011-12. We find that nominal incomes increased by 177% during the period. Income from agricultural labour, business and cultivation increased moderately by 117%, 132% and 152% respectively while income from casual labour and remittances grew by 237% and 528% respectively. Consequently, the average shares of agricultural labour incomes and cultivation income decreased from 35% and 11% to 33% and 9% respectively. The same for casual labour and remittance increased from 11% and 3% to 14% and 6% respectively. In 2011-12, 35% households were involved in agricultural labour compared to 40% in 2004-05. Also, proportion of households earning from casual labour and remittances increased from 28% and 6% in 2004-05 to 47% and 18% respectively in 2011-12. The share of casual labour increased more than that of remittances among SCs, STs, households with highest adult education as primary to higher secondary and landless households with salaried income. The share of remittances went up higher than casual labour for Brahmins, Forward Caste, OBCs, illiterate, above higher secondary, landless labourers, absentee lessors and all those who possessed land. Among sources of income inequality, we find remittances as the only income source which increased its share in income, decreased its gini and was favourable to lower quintiles. This made remittances more inequality decreasing in 2011-12 as compared to 2004-05 with an elasticity of gini to share of remittances at -1.2%.
Key Words: Income Sources, Income Inequality, Gini Decomposition, Indian Human Development Survey (IHDS), Remittances.
1 Assistant Professor, Agricultural Economics Research Unit (AERU), Institute of Economic Growth (IEG), New Delhi. E-mail: [email protected] 2Assistant Professor, Agricultural Economics Research Unit (AERU), Institute of Economic Growth (IEG), New Delhi. E-mail: [email protected] 3 Research Analyst, Institute of Economic Growth (IEG), New Delhi. E-mail: [email protected]

1. Introduction
Rural households earn their incomes from various sources including cultivation, livestock, agricultural wage labour and other non-farm occupations. Apart from cultivation which is considered the mainstay of Indian rural household income, livestock and agricultural labour income supplement the incomes of households wholly dependent on farm based activities. As economies grow and diversify into various non-farm activities, rural households also earn a larger chunk of their incomes from non-farm casual labour, migration to other rural or urban areas and from salary based activities. The participation of rural households in non-farm activities could be either due to “push” factors (e.g. risk reduction, land constraints, response to a crisis) and/or “pull” factors (eg. complementarities with existing income activities, higher profitability of the activities) (Barrett, Reardon and Webb, 2001). Various factors like education, skills, caste, religion, asset ownership, household size, and credit availability influence the decision of a household to participate and also decide on the extent of participation in a particular non-farm activity ( Srivastav and Dubey, 2002; Chadha and Sahu, 2002; Reardon et al., 2007; Jatav, 2010; Jatav and Sen, 2013). Understanding the relation between participation in rural non-farm sector and well-being of a household and relation between non-farm sector expansion and macro-economic indicators is of vital importance. Recent trends of a ‘stunted’ structural transformation in the Indian economy where the decrease in agriculture’s share in GDP has not been accompanied by subsequent accelerated growth of labour intensive manufacturing (Binswanger-Mkhize, 2013) has made understanding of rural nonfarm sector extremely pertinent. Members of rural households also migrate to other rural or urban areas for employment and the internal remittances also become an important source of income for rural households. This migration could also be distress driven or driven by the good opportunities provided in the other areas. In summary, a growing economy with a not so fast growing manufacturing sector might see a high level of income diversification among rural households. This is what we expect to see among rural household in India during the study period of 2004-05 to 2011-12. The chapter explores issues related to dynamics of income diversification of rural households. The extent of incomes that households generate from farm based activities like their own cultivation and agricultural labour and those from nonfarm casual labour, migration and salaried employment are explored during the period from 2004-05 to 2011-12 using a panel data.
There has been a notable change in sources of income among Indian rural households over the past few decades. The predominance of agriculture-based activities in total income has

reduced in terms of employment and share of income. Participation of rural households in non-farm sector has expanded greatly over the last two decades or so (Coppard, 2001; Srivastav and Dubey, 2002; Bhalla, 2002; Bhaumik, 2002; Chadha, 2002; Sahu, 2003). Particularly, in the decade spanning 1993-94 to 2004-05, non-farm employment in rural India grew rapidly than farm employment (Kashyap and Mehta, 2007; Abraham, 2009; Jatav, 2010; Chowdhury, 2011; Himanshu et al., 2013). In this period, 60% of the 56 million rural jobs that were created were in the non-farm sector. The share of non-farm income in the total income of rural households has also become substantial during this period. Studies estimate that farm households earn 46% of their incomes from non-farm activities and rural households earn 48% of their income from non-farm activities (Himanshu et al., 2013). For rural India, this figure was estimated to be around 35% in the year 1993-94 (Lanjouw and Shariff, 2004; Reardon et al., 2007)
This increased diversification of income generating activities could be seen as an indicator of expanding economic opportunities and thus expected to help in reduction of poverty and inequality. Income from agriculture is largely related to land ownership and since land distribution is highly unequal in India, we would expect this income to be highly unequal. So, an expansion to nonfarm sector could pave way for reducing income inequality. But, this may not always be true as the accessibility to non-farm opportunities are far from universal (Reardon et al., 2000; Kundu et al., 2003; Jatav, 2010). If profitable non-farm opportunities end up being exploited largely by the already well-off or elite in the population, we could see an increase in inequality due to expansion of non-farm sector. Given such contradictory possibilities, there might also be other factors that might explain the linkage between nonfarm sector expansion and income inequality of rural households. For instance, unequal access to non-farm economic opportunities is sometimes exacerbated by social factors like caste and religion. These factors would put constraints other than economic factors like credit and assets for a particular section of people in pursuing opportunities in the non-farm sector. There have been some recent studies though which seem to suggest that social factors are affecting occupational mobility lesser than before (Kapur et al., 2010; Himanshu et al., 2013). Different sources of income could also contribute to inequality differently over time. For example, with time the nonfarm sector could employ people who were excluded previously and as they also earn incomes, it might contribute to decreasing of income inequality over time. But, there has been no detailed study into looking at the dynamics of sources of income inequality in Indian context and this chapter attempts to do that.

2. Data and Methodology
Study of income inequality and its composition has largely been of two types. On the one hand, there have been studies that have looked at influence of population subgroups defined by age land ownership, caste, religion, education, etc. These studies look into the impact of these characteristics of population on the resulting inequality in the economy. There has been wide interest in such studies and they have largely been pursued using regression based approaches (a few examples would involve Oaxaca, 1973; Bourgignon et al., 1998; Fields, 2003). Another area of inequality studies which is of larger importance to our current pursuit involves studying the impact of different components of income on inequality. Here, the question is concerned with how a particular income source (say income from livestock, migration or investment income) affects the overall income inequality. Here, the most commonly used methodologies involve decomposition techniques. Some of the studies integrate regression based approaches with decomposition techniques to analyze the role of population characteristics in affecting inequality (eg. Morduch and Sicular (2002)). For our current purposes, where we intend to study the impact of different sources of income on income inequality, we use a decomposition based approach in the tradition of Shorrocks (1982). We first analyze income diversification strategies of rural households using a large scale nationally representative data. We first look into the different diversification strategies of different population groups defined by caste, education, land holding and landholding classes.
After analysing the income diversification strategies, we look at dynamics of different income sources over the period 2004-05 and 2011-12. We look into how the income source shift has happened among the rural households. Then we perform an interquintile analysis to analyse the sources of inequality. To identify the changes in sources of rural income inequality, we then decompose the contribution of different income sources to overall income inequality using a Gini coefficient decomposition analysis method for the two years 2004-05 and 2011-12.
Gini coefficient is the most widely used statistic in estimating income inequality and it varies between 0 and 1 with 0 indicating perfect equality and 1 indication very high income inequality where only one person gets all the income. Though theoretically possible, none of the countries in the world have values near to the extremes. Typically, Gini coefficients of income and consumption in the range of 0.3 to 0.35 are considered to be present in egalitarian

societies and values exceeding 0.4 are considered inegalitarian (World Bank, 2015). This Gini coefficient of total income is then further decomposed using a method proposed by Lerman and Yitzhaki (1985). The total income is derived from different sources which we categorize as follows – income from cultivation and livestock (own-farm income), income from agricultural labour (agricultural wages), income from casual labour (nonfarm labour), salaried employment (salary) and employment in businesses (business) and income from rent, pensions, scholarships and welfare benefits from Government (other income). Each of these sources will generate inequality. Typically, these individual inequalities would be higher owing to the fact that not all of the households derive income from each of these sources.
The share of each of these sources in total income and the correlation between the source income and total income is used to decompose the Gini coefficient as follows:
G  [mk / m][2Cov( yk , Fk ) / mk ][Cov( yk , F ) / Cov( yk , Fk )] ... (2.1)
k 1
G  SkGk Rk ... (2.2)
k 1
Where G is the Gini coefficient of total income inequality and Gk is the Gini coefficient indicating the inequality in incomes from source k , m is the mean income of the population, mk is the mean income from income source k , Cov( yk , Fk ) is the covariance between income component k and its cumulative distribution, Cov( yk , F) is the covariance between income component k and cumulative distribution of the total income, Sk is then the measure of component k ' s share in total income and Rk is the “Gini correlation” between income component k and total income. This decomposition provides three important statistics of relevance - Sk , which suggests how high the share of a particular income component is in total income; Gk , which suggests how equally/unequally a particular income component is distributed and Rk , which suggest how much a particular income component and total income distribution are correlated. All these three are important for an understanding of income inequality of rural households. Just because a particular income component (say wage income) forms a high share of total income, it may not influence inequality in a particular way. If it is equally distributed and also has no correlation with the distribution of total income, it might not have any say in the final income inequality of the population. Similarly,

an income component which is very unequally distributed might not necessarily have a bad impact on total income distribution of the population. It might so happen that the share of it in the total income is quite less and also is negatively or not correlated with the total income distribution.

As we observe, only by having an idea of the three factors for each income component will we be able to say whether a particular income component has an inequality increasing or inequality decreasing effect on the population. Lerman and Yitzhaki (1985) further estimate the effect of small changes in a particular income source on inequality, holding income from all other sources constant. They derive the percent change in inequality resulting from small percent change in income from source k as follows:

G e  SkGk Rk  S ...(2.3)




Where G e is the change in Gini coefficient to a marginal percentage change ( e close to 1) in income source k .

We use the above two formulations – decomposing income inequality and estimating marginal effects on income inequality for the two years and identify the changing sources of income inequality among rural households. From the decomposition analysis, we estimate the share of different income sources in total income, the inequalities within different income sources and Gini correlation between different income source distribution and total income distribution. From the marginal analysis, we estimate the impact of percentage change in any income source on Gini coefficient of total income. This will tell us whether expansion in any income source is inequality increasing or inequality decreasing. Also, we decompose inequalities within the nonfarm incomes based on employment types. This will provide us with the structure of inequality in the nonfarm sector. This analysis is done for the period of 2004-05 and 2011-12 and the changes in these two periods are observed. The changes will provide an indication of whether a particular income source has become more inequality increasing or more inequality decreasing. This also changes with time and understanding these dynamics is also crucial. For instance, the relation between non-farm sector expansion and inequality might not be static and the dynamics of the relationship might be dependent on the stage of the non-farm sector expansion in the rural areas. For the economy as a whole Kuznets (1955, 1963) had suggested that as the economy moves towards more modern sectors, inequality will initially increase as the modern sectors would initially benefit the elite

but gradually over time, will start benefiting the lower strata as well resulting in reduced inequality. Himanshu et al. (2013) suggest that this kind of an inverted U-curve could be observable even in the rural economy as well and non-farm sector expansion might initially have an inequality-increasing tendency. All these aspects need to be explored while studying the linkage between different sources of income and income inequality.
We use the data collected from two rounds of Indian Human Development Survey (IHDS) conducted in 2004-05 and 2011-12. The survey is a large scale nationally representative and was conducted under the supervision of National Council of Applied Economic Research (NCAER) in collaboration with University of Maryland. The detailed methodology of the two rounds of survey, along with some summary information and preliminary findings, could be obtained from Desai et al. (2010) and Desai et al. (2015). We present a brief note on the survey here. The survey covers almost all the states and union territories of India except Andaman and Nicobar and Lakshadweep. The survey used two-stage stratification and was conducted over a sample of 27010 rural households (from 1503 villages) and 13126 urban households (over 971 urban blocks) in 2004-05. In the year 2011-12, the survey team reinterviewed around 83% of the households as well as split households (if located within the same village or town). It also selected an additional replacement sample of 2134 households in this round. Totally, the 2011-12 survey was conducted among 42,152 households. For the purpose of our analysis, we use only those rural households which were surveyed in both 2004-05 and 2011-12. There were household that were split after 2004-05 and the split households in the same villages were resurveyed in 2011-12. Since the survey did not mention whether there were households that were split and not in the same village, we did not include any of the split households for our analysis. In all, there were 19,831 households that fit the above criteria and they have been included for the purpose of our analysis.
The survey has collected data related to various social, economic and political aspects. The survey has data related to income from different sources for all the households. It has various unique modules which collect data on education, health, occupation, economic status, marriage, fertility, gender relations, and social capital. It also collects information on income from various sources for every household. In particular, information is collected on household’s income from cultivation, livestock, agricultural wages, non-agricultural wages, salaried employment, businesses, sale or rent of agricultural property, remittances, sale of property and welfare benefits. We include incomes from cultivation, livestock and lease of agricultural property as income from own-farm activities. We consider agricultural wages

which are obtained from agricultural labour in other farms separately as agricultural labour income. Income from wages in non-agricultural activities, remittances, businesses and salaried employment are considered as different sources of income. Income from rent of property, pension, scholarship and welfare benefits (like old age pension, widow pension, maternity benefits, disability schemes, Annapurna, income support other than NREGA and assistance from NGOs/charities) are combined as other income for the purposes of our analysis. There were two other facets of income that needed attention in our analysis. The net income from cultivation, livestock and nonfarm business activities could be negative and because of that some of the households had a total income that was negative. When calculating the shares of a particular income source in total income, these estimates biased the average shares. For instance, if the household earned a net negative income of INR 45,000 in cultivation and earned INR 36,000 from other income sources, the total income of the household income turns out to be negative INR 9,000. If share of agricultural income is calculated for such household, it comes out to 500%. Such estimates were inducing a bias in the average share of agricultural income in the household. To avoid such biases, we have only considered households with positive incomes when analysing the shares of different income sources in total income. Since there were a significant number of households with negative incomes, this analysis is somewhat limited. The negative income households will have to be analysed separately to understand their diversification patterns. We also find that there were more than 40% of the rural households that did not possess any land. The diversification patterns of such households differed largely from households that had any kind of land. We have tried to separate analysis for these two types of households in appropriate places.

3. Changes in Incomes of Rural Households As mentioned previously, we considered seven different income sources of rural households for our analysis. The distribution of the incomes from these different sources and total incomes for the years 2004-05 and 2011-12 are presented in the table 3.1 below.
Table 3.1. Incomes of Rural Households 2004-05 and 2011-12

Source Income

of Average (1)

Standard Deviation (2)

Coefficient of Variation (3)

Average (4)

Standard Deviation (5)

Coefficient of Variation (6)

CAGR (7) = ((4)/(1))^1/7-1


Average (8) = ((4)-(1))/(1)%

Standard Deviation(9) = ((5)-(2))/(2)%

Agriculture Agricultural Labor Casual Labor Salary Business Remittance Other Income Total

4414 4315 6751 4463 1064 1611 34248

8086 10171 22870 20930 7962 9062 52480

184% 236% 339% 470% 750% 563% 154%

9605 14534 17465 10360 6678 6893 94793

21274 31661 61560 69226 25907 40204 187635

221% 218% 352% 668% 388% 583% 198%

11.8% 19.0% 14.6% 12.8% 30.1% 23.1% 15.7%

118% 237% 159% 132% 528% 328% 177%

163% 211% 169% 231% 225% 344% 258%

From table 3.1, we observe that the average household income increased from INR 34,248 to INR 94,793, an increase of 177%. The CAGR for the period of 7 years turns out to be roughly 15.7%. The growth in income different income sources was not uniform and varied from 118% in agricultural labour to 528% in remittances. Income from salaried employment grew almost on par with rural incomes at 159% during this period. Agricultural income involving cultivation, livestock and lease income, agricultural labour and business income grew by only 152%, 118% and 132% during this period, much lesser than the growth rate of total income. Income from casual labour grew at 237%, higher than the growth of total income. Income from sale of property and benefits from government grew the least among different sources at 328%. The income source that saw the highest growth was remittances, which recorded a growth of 528% during this period. The growth of this along with casual labour income seems to indicate a movement of labour towards nonfarm employment. The reduction in coefficient of variation in these two income sources also indicates that these incomes were becoming more uniform among rural households. This could be either due to more households participating in these activities or the incomes from these sources becoming more uniform among the same participating households or combination of both. The coefficient of variation of household income increased by 44% suggesting an increase in income inequality during the study period. Apart from casual labour income and remittances, coefficient of variation of all the income sources increased indicating increase in inequality of different income sources in the study period.
4. Changes in Income Diversification of Rural Households As indicated in the methodology section, we are considering only the households with nonnegative incomes from different sources to get unbiased estimates of income diversification patterns and its changes over the study period for the same households. Table

4.1. provides the average shares of different income sources for different income quintiles of the rural households.

Table 4.1. Income Diversification Dynamics

Agricultural Labor

2004-05 32.9%

2011-12 29.6%

Changes -3.2%




Casual Labor Salary Business Remittance

17.0% 11.6%
9.8% 2.8%

20.1% 11.3%
8.0% 7.7%

3.1% -0.3% -1.8% 4.9%

Other Income




As table 4.1 indicates, agriculture is the dominant source of income for rural households in both 2004-05 and 2011-12. In 2004-05, the household earned an average of 32.9% from agriculture (cultivation, livestock and lease rent) while in 2011-12 they earned an average of only 29.6%. The same was the case with agricultural labour incomes. The average share of agricultural labour income in total income reduced from 22.1% in 2004-05 to 16% in 201112 a reduction of 6.1%. The shares in total income increased for casual labour income and remittances from 2004-05 to 2011-12. The average share of casual labour income increased from 17% in 2004-05 to 20.1% in 2011-12. This meant that casual labour income became the second highest important income source after cultivation in 2011-12 as compared to agricultural labour in 2004-05. The other important income source which has shown a high increase is that of remittances earned from migration to other rural and urban areas. As compared to 2004-05 in which household earned 2.8% of their household income on average, the households earned 7.7% of their household incomes on average from remittances in 201112.
The changes observed could be due to various reasons. It could be due to households opting out of a particular activity or give lesser time to a particular activity or reduced wages or returns from the particular activity. Table 4.2 provides the changing dependency on different income sources for the households in the surveys. It shows the percentage of people earning any income from a particular source of income in 2004-05 and 2011-12.

Table 4.2 Changing Dependency on Different Income Sources
IncomeHouseholdsInequalityIncome InequalitySources