Value-added disaggregated social accounting matrix for the Indian economy of the year 2007–2008

For the proposed disaggregated SAM, we have disaggregated only the labour input into 48 various types of labour. Here we have used region, social class, education and gender as four indicators to classify the labour input. The detail about these indicators is given in Table 2.

Once we classify the types of labour according to their location, social status, level of education and gender, our next task is to obtain data on payment for these types of labour by different sectors. In this case, we have estimated the distribution of labour payment across the SAM sectors by using the unit-level households survey data on employment and unemployment situation in India. The National Sample Survey Organization (NSSO) of India conducts detailed household survey in every 5 years separately for consumption expenditure and employment unemployment situation in India. The 68th round survey data for the year 2011–2012 are latest available quinquennial survey data for both consumption expenditure and employment–unemployment situation in India. Prior to the 68th round survey, immediate latest quinquennial survey data are available for the year 2009–2010 (66th round) and for the year 2004–2005 (61st round). Therefore, it was challenging for us to select a database to fulfil the data need for this proposed SAM. It is also worth noting that the Indian economy has experienced 9% GDP growth during the periods from 2005–2006 to 2007–2008. Although the real GDP growth dipped down at 6.7% level in the year 2008–2009, it has reached more than 8% level and continued at that level till 2011–2012 (CSO 2012). Now with the faster economic growth between 2004–2005 and 2011–2012, structure of the economy has also changed and hence the structure of employment. Table 3 presents the compound annual growth rate (CAGR) of employment between the year 2004–2005 and 2011–2012. Table 3 shows that the employment in agriculture sector has fallen during this period at a rate of 1.55%, whereas it has increased significantly at a rate of 13% in the construction sector. Apart from that, other sectors also experience positive growth in employment between the year 2004–2005 and 2011–2012.

Now since the base year of our proposed SAM is 2007–2008, the NSSO 61st round data will not capture the structural change happened during the golden periods of Indian economy during post-2004–2005 periods. Again as there was severe drought in India in the year 2009–2010, the 66th round quinquennial survey has been updated with 68th round survey for the year 2011–2012. Therefore, to capture the structural change in Indian economy during post-2004–2005 period, we have selected NSSO 68th round data to estimate required data for our proposed SAM of India for the year 2007–2008.

The NSSO employment survey classifies industries at the 5-digit level as described in National Industrial Classification (NIC) guideline. The households demographic and social profile data of the same survey have been first organized to classify the types of labour according to the description as mentioned in Table 3. Further, we have aggregated the industries according to their map of concordance with the sectors of our proposed SAM. In this context, the map of concordance between the sectors of our proposed SAM and the NIC 5-digit classification of industries is given in “Appendix 1”.

To make our value-added disaggregated SAM useful for poverty analysis, understanding structure of consumption expenditure across various social classes of households is important. Therefore, we have classified households into 80 different classes. In Table 4, we have listed various indicators which have been used to classify households.

After classifying households according to the above-mentioned categories of classes, we have estimated sources of income of these households from different types of labour and capital. It is also important to note that household’s typewise data on capital stocks are not readily available for Indian economy. Therefore, we have used the income of self-employed and own account category of households as proxy of the capital income. Now to estimate the capital income this way, we have cross-tabulated the NSSO 68th round employment–unemployment survey data between households categories (using demographic profile) and types of labour (using status of work profile of the households). Again to obtain the income of the households from different types of labour, we have prepared another cross-table between the household categories and the types of labour we have defined for our study purpose.

To complete the construction of consumption block of this proposed SAM, we have used National sample survey (NSS) 68th round data on households’ consumption expenditure. But the descriptions of commodities in these databases are different than the sectors of our proposed SAM. So to overcome this problem also, we have used the map of concordance between the NSSO items and SAM sectors as given by Pradhan et al. (2013).

In the above-mentioned steps, we have estimated the required data for our proposed SAM especially for those accounts which make this SAM significantly different from others. Now in this step, we have completed the construction by obtaining data on households’ income other than factor income. This includes transfer income from government, interest receive from public bonds and remittances from the abroad. To obtain these data corresponding to the different category of households, we have first taken the aggregate data from 2007 to 2008 SAM and then distributed it among the different class of households according to their share in aggregate consumption expenditure as obtained in the previous step.

Therefore, by following above-mentioned steps we have constructed a value-added disaggregated SAM for India for the year 2007–2008 which consists of 171 rows and column. The entire SAM is given as the supporting document along with this paper (Additional file 1: Appendix S1). But construction of proposed accounts for this SAM with NSSO data does not ensure that this SAM is balanced. This implies the row total is not equal to corresponding column total, because the NSSO data are obtained through primary sample survey and are available in market price, whereas data presented in IO table are based on factor price. Therefore, to make this SAM balanced we have followed following two steps. First, estimate the distribution patterns for households factor income from different types of labour and itemwise consumption expenditure and sector wise labour payment using the NSSO data. Secondly, a pro-rata adjustment method has been applied to distribute the aggregated data of existing 2007–2008 SAM according the distribution pattern. Thus, we have adjusted the NSSO data with the aggregate-level data on household income, consumption expenditure and sectorwise labour payment as obtained from the existing SAM of the year 2007–2008 to complete the construction procedure.

Figure 1 describes the contribution of various social groups of households in India’s NDP of the year 2007–2008. If we add the percentage contribution to NDP by the various social groups of households (including rural and urban), we find almost 87% of NDP is contributed by the households and rest is contributed by the government, private corporate and public non-departmental enterprises. In addition to estimating contribution to NDP by various households groups, we have also estimated the composition of various social groups in India’s total population and it is superimposed in Fig. 1. Figure 1 shows that urban area in India is residence of almost 30% of her total population and contributing almost 43% of her NDP. By contrast, the 70% population lives in rural area and contributing only 44% of total NNP which is almost 30% lower than their population contribution. This scenario is quite obvious as rural economy is dependent on agriculture sector which is lowest contributor to GDP in India as compared to manufacturing and service sector. Again, more than half of the Indian population is based on agriculture sector as their primary source of livelihood. Yet again, it is interesting to observe that the gap between percentage contribution towards population and NDP is lower for the households belong to other social group in rural. According to Fig. 1, the other social group of households occupies only 17% of total population and contributes almost 13% India’s NDP in the year 2007–2008, whereas the households belong to OBC social group occupy one-third of the India’s population and contribute only 21% of India’s NDP. Thus, it can be argued from this figure that the average productivities of the rural population that belong to the social groups schedule tribe (ST), schedule caste (SC) and other backward class (OBC) are lower than the rural population belong of other social group. On the other hand, in urban area the percentage share in population is lower than their contribution to NDP across every social group. Hence, it can also be argued that the urban households irrespective of their social belonging are more productive than their rural counterpart.

Further, in Table 7 we have shown the distribution of worker across their skill. The number of workers is estimated using NSSO 68th round survey and their categorization according the level of skill is based on their level of education (see Table 7). Table 7 shows that in rural area 48% of total workers supplied by the ST category households are unskilled. In case of SC category households, the share of unskilled worker is 43% and it is 36% for the OBC category households. The share of unskilled worker in total worker for other social group households is lowest (i.e. 23%) in rural areas. However, in urban areas the share of unskilled worker in each category of social group is lower as compared to their rural counterpart. Therefore, it can be argued from the above discussion that the source of income of the socially marginalized group in India is largely dominated by the income from unskilled worker, and hence, the average productivity is lower for these categories of households in India as shown in Fig. 1.

Again our value-added disaggregated SAM categorized households according to the expenditure deciles class. Therefore, the households income data presented in our value-added disaggregated SAM can be used to estimate the level of inequality across rural and urban households. In this study, we have estimated Gini coefficient as a measure of inequality and the results are shown in Fig. 2.

Figure 2 plots the Gini coefficients for different social groups of households across rural and urban area in India for the year 2007–2008. Higher value of Gini coefficients explains higher level of income inequality. Here, the highest level of inequality is observed for the urban households (0.44) than their rural (0.29) counterpart. In rural area, highest level of income inequality is observed among the other category households (0.63) which is 2.5 times higher than the OBC category households (0.24), 3 times higher than SC category households and 4 times than ST category of households. Therefore, the income inequality among the ST households (0.16) category is lowest in rural area. By contrast, the same is higher for the same social group (ST category) of households in the urban area, i.e. 0.34. The lowest income inequality in urban area is observed for the SC category households (0.22). Further, it is interesting to note that the Gini coefficients for urban area are higher for all social groups of households that the rural areas except the households belongs to other social group. The Gini coefficient for other category of households is 0.54 for urban area as compared to 0.63 in rural areas. Now if we relate these income inequality estimates to their corresponding per capita income as shown in Table 8, we see a direct relationship between the two indicators for each group of households. Therefore, it can be argued that with the increase in per capita income the level of income inequality increases irrespective of region and social status of the households. This issue corroborates the rising part of the Kuznets curve. But at what point this level of inequality will starts falling is a crucial research question. Using the SAM multiplier model, one can further simulate with various policy options to understand their impact on income inequality in India. Below we have described an illustrative example of SAM multiplier analysis on the impact of growth in various crop and food processing sectors on rural income and inequality.

In this section, we have presented the SAM multiplier analysis to see the impact of agricultural growth on rural and income and inequality. The following table describes the cropwise impact on rural income for different types of households. The impact on income is obtained from the SAM multiplier matrix corresponding to the interaction between rows of the households and column of the economic activities. Thus, the value 0.085 for ST category households corresponding to paddy crops explains the impact on income of that particular group of households due to growth in paddy crop sector by one unit. Similarly we can explain the rural income values for each household category corresponding to various agricultural activities. Using this multiplier value, we have estimated the Gini coefficient to observe how the increase in income will be distributed across various decile classes of households within each households class. Further we have also estimated the overall impact on rural income and rural income inequality. Again though we have ten various crop sectors in our SAM, we have selected few major crops for illustration purpose.

As it is observed from Table 8, if there is an increase in output of paddy crop sector by one unit, income of rural ST households will increase by 0.085 unit which is highest ever the increase as compared to other listed sectors in this table. Moreover, this increase in income will be distributed more equally among the ST households as compared to growth in income due to growth in other crop sectors. Similar fact is also observed for the SC households. However, the OBC category and other types households gain income more due to growth in livestock sector and this income growth is more equally distributed (lowest GINI coefficient, i.e. 0.17 and 0.37, respectively, for OBS and others) among the households of those classes than the growth in other sectors. In totality, the rural income inequality will be lower, i.e. 0.17, due to growth in paddy crops than the growth in other activities reacted to agriculture and listed in this above table.

It is also interesting to observe that the growth in pulses crops will lead to overall growth in income by 0.807 unit which is lower than other farm-related activities but higher than the growth in food processing industry, i.e. 0.723. Yet again the growth in income due to growth in pulses crops leads to rise in rural income inequality (i.e. 0.38) than the growth in paddy (0.17), livestock (0.23) and even vegetables crop (0.32). Moreover, growth in pulses crop leads to rise in income inequality among various social groups of households.

The SAM multiplier is a static observation for the year 2007 and is highly influenced by the social and economic structure of the Indian economy. Cereals being a principal crop in Indian agriculture (especially paddy) and paddy being the major staple food for most of the India, a large share of agricultural land is employed for this crop cultivation. As a result, the growth in paddy crop leads less income inequality among the rural households in India. Further, the pulses crops are dominated in arid and dry agro-climatic zones in India, and due to such in equal distribution of pulses growers, the income effects also lead to rise in income inequality in Indian economy. Therefore, unless the area under pulses increases in other parts of the country the inequality in rural income cannot be reduced. In this context, Government of India has integrated pulses crop under national food security mission to promote pulses crops in India and rice fallow region of eastern India is also the prime focus for pulses production. Once achieved, we may expect reduction in income inequality within the rural area. It is also to be noted that the Indian economy has been moving through the structural transformation process since 1991, but to reduce the income inequality, a structural transformation is required within the agriculture sector with more focus towards livestock and pulses crops.