Poverty levels in the country have fallen from 26.1 per cent in 1999-2000 to 21.8 per cent in 2004-05, based on the mixed recall period consumption method. Those for rural areas fell from 27.1 to 21.8 while those for urban areas fell from 23.6 to 21.7 per cent.

While the poverty estimates for 1999-00 and 2004-05 are not strictly comparable with earlier poverty estimates, some broad conclusions can be drawn on the rate of decline of poverty vis-�-vis the rate of growth of the economy.

Both the rate of decline in poverty and income growth between 1993-94 and 1999-00 have been the highest for the six periods given in Table 1. Yet, the elasticity of poverty reduction (0.78) is exactly the same as that between 1983 and 1987-88.

Does this mean that the trickle-down hypothesis is working for poverty reduction? The correlation between aggregate income growth and rate of decline of poverty is not very strong (0.47).

However, it also provides sufficient evidence against rejecting the trickle-down hypothesis. The trickle-down hypothesis is stronger for urban areas -- the correlation between aggregate income growth and rate of decline of poverty for urban areas is 0.73, while the same for rural areas is 0.36.

While at an aggregate level there has been some evidence of the trickle-down hypothesis, the cross-sectional evidence of trickle down between 1999-00 and 2004-05 is rather weak. The correlation between the rate of decline in poverty (RDP) and gross state domestic product (GSDP) growth between 1999-00 and 2004-05 is 0.20, suggesting a weakness in or the absence of trickle-down.

All major states witnessed a positive GSDP growth between 1999-00 and 2004-05, while the incidence of poverty increased in Delhi, Goa, Haryana, Jammu and Kashmir, Maharashtra, Rajasthan, Dadar and Nagar Haveli and Daman and Diu.

Even after excluding states that experienced an increase in poverty, the evidence of trickle-down (correlation coefficient = 0.30) is weaker than that observed at the national level. Among the states which experienced a decline in the incidence of poverty between 1999-00 and 2004-05, the magnitude of the RDP varies from 2.22 per cent in Kerala to 16.77 per cent in Arunachal Pradesh.

Between 1999-00, the average annual growth rates of Andhra Pradesh (6.54 per cent), Gujarat (6.85 per cent), Haryana (6.86 per cent) and Kerala (6.88 per cent) were of a similar order while the RDP varied drastically. It was 6.77 per cent for Andhra Pradesh, 2.41 per cent for Gujarat and 2.22 per cent for Kerala and the incidence of poverty increased in Haryana by 2.44 per cent per annum.

Similarly, Assam and Arunachal Pradesh were able to achieve an RDP of 16.77 per cent and 16.11 per cent respectively between 1999-00 and 2004-05 with a corresponding average annual growth in GSDP at 4.86 per cent and 5.23 per cent. Undivided Bihar with a 5.01 per cent average annual growth of GSDP between 1999-00 and 2004-05 was able to achieve an RDP of 4.95 per cent.

However, poverty in Rajasthan and Maharashtra increased by 2.79 per cent and 0.12 per cent per annum respectively, despite their GSDPs rising by 5.39 per cent and 5.13 per cent respectively. Clearly state level policies with regard to poverty alleviation, safety nets and developmental activities are mainly responsible for divergent trends of poverty reduction vis-�-vis economic growth.

This is not the first time that rural poverty has declined faster than urban poverty has. This was observed in earlier periods also --1977-78 to 1983, 1983 to 1987-88 and 1993-94 to 1999-00. The ratio of rural to urban RDP at 2.54 observed between 1999-00 and 2004-05 is slightly higher than the 2.34 observed between 1983 and 1987-88. Both these time periods coincide with a low average agricultural GDP growth (0.06 per cent in 1983 to 1987-88 and 1.76 per cent in 1999-00 to 2004-05).

What are the reasons for the faster decline in poverty in rural areas compared to urban areas? Rural employment estimates suggest that rural employment alone is not growing fast enough to reduce rural poverty faster -- the shift of the labour force from the less profitable and stagnant agricultural sector to the more profitable and growing sectors could, however, be the reason for the faster decline in rural poverty levels.

Provisional results of the fifth economic census suggest that between 1998 and 2005, the average annual growth of enterprises in rural areas-- at 5.53 per cent-- has been nearly 50 per cent higher than the growth of enterprises in urban areas.

At the same time the growth in employment offered by these enterprises in rural areas at 3.33 per cent has been nearly twice the employment growth in urban areas. Between 1999-00 and 2004-05, five states, namely, Assam, undivided Bihar, Tamil Nadu, undivided Uttar Pradesh and West Bengal accounted for 72 per cent of the decline in the number of rural poor across the country.

A combination of higher agricultural growth rate and higher employment growth in non- agricultural/farm (excluding crop production and plantation) activities led to a decline in the absolute numbers of poor in these states.

While in undivided Bihar it was higher agricultural growth that led to the decline in the number of rural poor, in the case of Tamil Nadu and undivided Uttar Pradesh it was higher non-agricultural growth that was responsible. In short, the politics and economics of Indian states are so divergent that it is difficult to pinpoint a single factor responsible for the change in poverty or any other developmental indicator.

The author is Associate Director, Fitch Ratings India Private Ltd. The views are personal.