Decomposition of Poverty Reduction Between 1993/94 and 2009/10

We examine some of the factors which contributed to the decline in poverty between 1993/94 and 2009/10. The potential impact of growth, had there been uniform growth rate of MPCE across states and between rural and urban areas are analysed using the parameter estimates of log-normal distribution fitted to the expenditure data of the states. We have assumed that per person total expenditure of a household (y) is distributed log-normally with mean 9 and standard deviation k. The mean p of y, Gini-coefficient (G) and head count ratio (HCR) are given by

Table 3.3 Annual growth rates of Atkinson inequality (A_e)

Note Same as in Table 3.1

***Significant at 1% level, **significant at 5% level, *significant at 10% level

Fig. 3.2 Trends in Atkinson inequality (A₂₀) and rural urban disparity

where, ф is cumulative distribution function of the standard normal deviate with mean 0 and variance 1; and z is poverty line.^[1]

The parameters of the log-normal distribution are estimated for states with a break up on rural and urban from the NSS unit level data.^[2] The fitted log-normal distributions are found to give a good fit for most of the states. The HCR of All India is estimated as weighted average with states population as weights. The

Table 3.4 Simulation results for poverty reduction (HCR %) between 1993/94 and 2009/10: all India and selected states

decline in HCR between 1993/94 and 2009/10 in the states as well as All India, estimated from the fitted distributions are found to be close to those directly computed from the unit level data.

Simulations are conducted by treating log-normal distribution estimated for 1993/94 and states population for 1993/94 as base. HCRs under alternative simulations are computed by replacing the MPCE of states with new MPCE corresponding to alternative scenarios. Tables 3.4 and 3.5 give the estimated poverty reduction under alternative growth scenarios and the decomposition of observed poverty reduction between 1993/94 and 2009/10. Table 3.4 also shows the inequality effect on poverty reduction.

• 4^
• 00

Table 3.5 Decomposition of poverty reduction between 1993/94 and 2009/10: all India and selected states

^aFor All India, growth rate of MPCE is assumed to be uniform across the states and between rural and urban, and for the states it is assumed to be same between rural and urban at the observed (rural + urban) poverty of the state

R. Radhakrishna et al.

We find the reduction in head count ratio would have been 18.9% points had the MPCE of all states as well as rural and urban increased at the same rate as that of the MPCE of All India (rural + urban) instead of the realized rate of 14.3% points. Hence the gain in poverty reduction under uniform growth would be 4.6% points. Since the MPCE (rural + urban) of all India is fixed in the simulations, some states whose MPCE growth was higher than that of All India have lower poverty reduction under uniform growth scenario than their observed poverty reduction between 1993/94 and 2009/10. The All India poverty reduction under uniform growth scenario is found to be higher than the observed one, even though its MPCE remains the same. This may be due to the fact that the states with low growth are found to have low MPCE and low inequality in 1993/94 (see Table 3.6). States with higher incidence of poverty such as Bihar, Chhattisgarh, Madhya Pradesh, Rajasthan and Uttar Pradesh are found to have higher reduction in poverty under uniform growth compared to their actual reduction between 1993/94 and 2009/10. States such as Haryana, Punjab and Kerala are the losers in poverty reduction under uniform growth. The loss is due to both reductions in their MPCE as well as their high inequality in 1993/94.

The last row of Table 3.4 shows the effect of changes in inequality between 1993/94 and 2009/10 on poverty reduction. Poverty reduction would have been higher had the inequalities remained at the 1993/94 levels. Poverty reduction in All India would have been higher by about 2% points had the inequalities of the states been at 1993/94 levels. The Gini-coefficient as well as Atkinson measure of inequality show substantial worsening of inequality between 1993/94 and 2009/10 in all states except Jammu and Kashmir and North Eastern states; inequality worsened, more significantly in Kerala, Punjab and Haryana^[3] (Table 3.6). Consequently, the effect of the increase in inequality on poverty reduction is found to be about 7% points for Kerala and Punjab and 5% points for Haryana.

Table 3.5 gives the decomposition of poverty reduction between 1993/94 and 2009/10 derived from Table 3.4. Among the selected states, inequality aggravated poverty in all states and rural-urban disparity also aggravated poverty in all states except Kerala and Haryana. The inequality effect is more pronounced than that of rural-urban disparity. In Kerala and Haryana, the change in rural-urban disparity between 1993/94 and 2009/10 has positive effect on poverty reduction. This can be attributed to the decline in rural-urban disparity in these states during 1993/94 and 2009/10 (see Table 3.6).

U

о

Table 3.6 State wise inequality measures: all India and major states

Source Computed by the authors

R. Radhakrishna et al.

• [1] In this paper we used the poverty lines of Expert Group (2009)/Planning Commission.
• [2] The unit level households MPCE have been expressed at 1993/94 urban prices by adjusting forinter temporal, rural-urban and interstate price variations. The price adjustments have been madeusing Planning Commission’s rural and urban poverty lines at current prices for all India and states.The procedure assumes uniform price index for rural/urban expenditure groups within a state.
• [3] It may be pointed out that the interstate correlation between Atkinson inequality and Ginicoefficient is found to be close to one. Moreover, both show similar changes in inequality between1993/94-2004/05-2009/10 across states.