Selecting a Targeting Method to Identify BPL Households in India

Abstract

This paper proposes how to select a methodology to target multidimensionally poor households, and how to update that targeting exercise periodically. We present this methodology in the context of discussions regarding the selection of a targeting methodology in India. In 1992, 1997, and 2002 the Indian government identified households that are below the poverty line (BPL) and in updating the 2002 methodology, alternative methods have been proposed and vigorously debated. A fourth BPL method was published and a corresponding Socio Economic Caste Census (SECC), implemented. Using the third National Family Health Survey (NFHS-3), this paper illustrates how a BPL targeting method using SECC variables might be calibrated to a multidimensional poverty measure. This paper compares the fit between a benchmark measure of multidimensional poverty and several plausible targeting methods to determine which method(s) approximate it—as well as related measures—most closely. We find a ten-item binary scoring method, which uses variables already available in the SECC questionnaire, provides a strong proxy. The emphasis of this paper is to illustrate how a particular targeting method can be justified, rather than to advocate any particular solution.

This is a preview of subscription content, log in to check access.

Fig. 1

Notes

  1. 1.

    For example, for Cote D’Ivoire, Glewwe and Kannan (1989) finds it harder to predict accurately the per capita expenditure of rural residents. Using cross-country simulations, for 30 % eligibility threshold, Grosh and Baker (1995) finds that the under-coverage rate and leakage rate in urban Jamaica to be 43 and 26.1 %, respectively. The corresponding rates are 39.3 and 24.1 % for urban Bolivia, and 53.8 and 35.1 % for urban Peru. Narayan and Yoshida (2005), in case of Sri Lanka, find that the under-coverage rate and the leakage rate for the model with best predicting power to be 28 and 31 % for the 40th cutoff percentile.

  2. 2.

    The questionnaire appears in “Appendix 1”.

  3. 3.

    For an example on how corruption could cause loss of welfare during a redistribution of subsidized rice to poor households in the Indonesian context, see Olken (2006).

  4. 4.

    For example, one method was proposed by Mehrotra and Mander (2009) and we have used an adaptation of that simply as a matter of convenience. The second author was a member of the Saxena Committee expert group and the first author was a former member (See Datta 2009).

  5. 5.

    For example, the share of female-headed households increased from 5.2 % in NFHS-2 for 1998/99 to 9.2 % in NFHS-3 for 2005/06. It is important to monitor and understand the levels of deprivation among these different identity-based groups in order to understand whether to interpret this increase as an increase in poverty—or not. To do so requires a benchmark measure.

  6. 6.

    For an application of multidimensional poverty to Mexico, see Foster (2007) and CONEVAL (2011); for an application to Colombia, see Angulo et al. (2011); for Bhutan, see RGoB (2011). The counting approach has also been used to identify the BPL card recipients in Kerala, see Thomas et al. (2009).

  7. 7.

    The multidimensional nature of poverty has also been acknowledged by GoI (2009), while devising an improved method for estimating consumption poverty.

  8. 8.

    The number of interviewed women and men reported above are usual residents. Overall, the survey interviews 124,385 women from 87,016 households and 74,369 men from 51,443 households.

  9. 9.

    10,539 sample households with at least one usual resident, do not have all information available. The sample drop normally varies between 11.9 and 23.8 % across states. The two states with extreme values are 9 % in Manipur and 29 % in Goa. This does not invalidate our results: if we show that the disagreement in identification exists for this subsample, then the extent of disagreement in absolute number cannot be lower if the whole sample were used.

  10. 10.

    Our sample does not cover the rural population of the union territories including Delhi, but these regions cover merely 0.2 % of the entire rural population of India.

  11. 11.

    Our sample does not represent those households headed by elderly members with no member in the respondent’s age group (which tend to be smaller in size: only 5.2 % of the total rural households). As a result, households in certain criteria may be non-representative. For example, for the missing sample, the percentage of households with no adult member between the ages of 16 and 59 is 29.9 %, which is nearly 5.1 % of the rural population of 28 states. Our results thus are likely to under-report elder poverty.

  12. 12.

    The exclusion criteria and the criteria that have been used to construct the pseudo-BPL-scores for Saxena (2009), the alternative method, and SECC 2011 are listed in “Appendix 2”.

  13. 13.

    Note that our conclusions are constrained by our ability to match the inclusion, scoring, and exclusion criteria. In each case there are some inaccuracies introduced by differences between NFHS variables and the proposed criteria. As we cannot know the overall direction or magnitude of the errors, our conclusions are illustrative only.

  14. 14.

    Ferreira and Lugo (2012) argue that it is crucial for policy analysis to understand the joint distribution of deprivations in multidimensional analysis besides when constructing indices. They suggest three different approaches for understanding joint distribution: by stochastic dominance analysis, by simple cross tabulation and Venn diagrams, and by using the copula function. However, the use of Venn diagram becomes complicated when there are more than three dimensions or indicators. In this paper, we try to understand the number of dimensions in which households are jointly deprived in order to explore the extent of joint deprivations.

  15. 15.

    Note that all comparisons in this section refer to the proportion of poor households rather than poor people, because the BPL targeting exercise identifies the household as a unit of analysis.

  16. 16.

    What we call a bunching problem occurs when the fraction of households identified as poor changes drastically due to a change in the poverty threshold—for example because too few variables are being considered.

  17. 17.

    We have also compared the percentage of households identified as MD-poor to the percentage of households identified as BPL by the alternative scoring method for different poverty caps. When 35–36 % of households are identified as BPL and MD-poor, only 15.9 % of them are identified as poor by both methods. Similarly, when 45–46 % of households are identified as BPL and MD-poor, only 25.5 % of the MD-poor are identified as BPL. Thus the corresponding match indices are lower, and under-coverage rates and leakage rates are higher, for the alternative method than for Saxena (2009) with similar poverty cutoffs.

  18. 18.

    Among those 8.4 % of all rural households who are deprived in one-third of all weighted indicators but would have been automatically ‘excluded’ by the SECC exclusion criteria, 79.4 % do not have an improved sanitation facility, 75.9 % have at least one woman or child under-nourished, 92.8 % use unimproved cooking fuel, and 60.3 % do not live in houses with improved floor material. Thus it seems that the exclusion criteria may be mis-identifying as non-poor households that should qualify as BPL.

  19. 19.

    Among those 6.4 % of all rural households who are deprived in one-third of all weighted indicators but would have been ‘excluded’ by the first six SECC exclusion criteria, 78.9 % do not have an improved sanitation facility, 76.9 % have at least one woman or child under-nourished, 91.9 % use unimproved cooking fuel, and 56 % do not live in houses with improved floor material.

  20. 20.

    The questionnaire was accessed at http://rural.nic.in/sites/downloads/circular/questionnaire09082011.pdf on March 8, 2012.

  21. 21.

    We have also calibrated the ten-item binary scoring method using other criteria, such as the source of clean drinking water or the type of cooking fuel used, instead of the disability indicator and find that match between the multidimensionally poor and BS-poor (Table 5) are robust to these alternatives. However, we do not choose to use these criteria as they are similar to the sanitary latrine criteria used in BPL-2002, which the Ministry of Rural Development refers as disincentive criteria.

  22. 22.

    We have conducted the standard t test for the difference in means and find the differences to be statistically significant at the 99 % level.

References

  1. Ahluwalia, M. S. (2011). Prospects and policy challenges in the twelfth plan. Economic and Political Weekly, 46, 88–105.

    Google Scholar 

  2. Alkire, S., & Foster, J. E. (2011). Counting and multidimensional poverty measurement. Journal of Public Economics, 95, 476–487.

    Article  Google Scholar 

  3. Alkire, S., & Santos, M. E. (2010). Acute multidimensional poverty: A new index for developing countries. Working paper no. 38, Oxford Poverty & Human Development Initiative, Oxford University.

  4. Alkire, S., & Seth, S. (2008). Determining BPL status: Some methodological improvements. Indian Journal of Human Development, 2, 407–424.

    Google Scholar 

  5. Alkire, S., & Seth, S. (2009). Measuring multidimensional poverty in India: A new proposal. Working paper no. 15, Oxford Poverty & Human Development Initiative, Oxford University.

  6. Alkire, S., & Seth, S. (2013). Identifying BPL households in India: A comparison of methods. Economic and Political Weekly, 48, 49–57.

    Google Scholar 

  7. Angulo, R. C., Diaz, B. Y., & Pardo, R. (2011). Multidimensional poverty index (MPI-Colombia) 19972010. Department of Planning (Departamento Nacional de Planeación), Colombia.

  8. Atkinson, A. B., Cantillon, B., Marlier, E., & Nolan, B. (2002). Social indicators: The EU and social inclusion. Oxford: Oxford University Press.

    Google Scholar 

  9. Atkinson, A. B., & Micklewright, J. (1983). On the reliability of income data in the family expenditure survey 1970–1977. Journal of the Royal Statistical Society, 146, 33–61.

    Google Scholar 

  10. Consejo Nacional de Evaluacion de la Politica de Desarrollo Social (CONEVAL). (2011). Coneval presents 2010 poverty levels for each Mexican municipality. Mexico D. F., Press release no. 15. http://web.coneval.gob.mx/Informes/COMUNICADOS_DE_PRENSA/Press%20release%20municipal%20poverty%t20levels%202010.pdf. Accessed on 14 February, 2012.

  11. Datta, K. L. (2009). Breach of faith. Economic and Political Weekly, 44, 632–633.

    Google Scholar 

  12. Drèze, J., & Khera, R. (2010). The BPL census and a possible alternative. Economic and Political Weekly, 45, 54–63.

    Google Scholar 

  13. Ferreira, F. H. G., & Lugo, M. A. (2012). Multidimensional poverty analysis: Looking for a middle ground. Policy research working paper no. 5964. Washington, DC: World Bank.

  14. Foster, J. E. (2007). A report on Mexican multidimensional poverty measurement. Working paper no. 40, Oxford Poverty & Human Development Initiative, Oxford University.

  15. Glewwe, P., & Kanaan, O. (1989). Targeting assistance to the poor: A multivariate approach using household survey data. Policy, planning and research working paper no. 225. Washington, DC: World Bank.

  16. Government of India. (2006a). Report of the XI plan working group on poverty elimination programmes. New Delhi: Planning Commission.

    Google Scholar 

  17. Government of India (2006b). Social, economic and educational status of the Muslim community of India. New Delhi: Prime Minister’s High Level Committee, Cabinet Secretariat.

  18. Government of India (2007). Report on conditions of work and promotion of livelihoods in the unorganized sector. New Delhi: National Commission for Enterprises in the Unorganized Sector.

  19. Government of India. (2008). Eleventh plan (2007–2012): inclusive growth (Vol. 1). New Delhi: Oxford University Press.

    Google Scholar 

  20. Government of India (2009). Report of the expert group to review the methodology for estimation of poverty. New Delhi: Planning Commission.

  21. Government of India. (2011). Socio economic & caste census 2011 in rural India. Ministry of Rural Development, July 25. http://rural.nic.in/sites/BPL-census-2011.asp Accessed on October, 29, 2011.

  22. Grosh, M., & Baker, J. (1995). Proxy means tests for targeting social programs-simulation and speculation. Living standards measurement study working paper no. 118. Washington, DC: World Bank.

  23. Grosh, M., & Glewwe, P. (Eds.). (2000). Designing household survey questionnaires for developing countries: Lessons from 15 years of the living standards measurement study. Washington, DC: The World Bank.

    Google Scholar 

  24. Hasan, Z. (2009). Politics of inclusion: Castes, minorities, and affirmative action . New Delhi: Oxford University Press.

  25. Hirway, I. (2003). Identification of BPL households for poverty alleviation programmes. Economic and Political Weekly, 38, 4803–4808.

    Google Scholar 

  26. Jain, S. K. (2004). Identification of the poor: Flaws in government surveys. Economic and Political Weekly, 39, 4981–4984.

    Google Scholar 

  27. Jalan, J., & Murgai, R. (2007). An effective “targeting shortcut”? An assessment of the 2002 below-poverty line census method. Mimeo. New Delhi: World Bank.

  28. Khera, R. (2008). Access to the targeted public distribution system in Rajathan. Economic and Political Weekly, 43, 51–56.

    Google Scholar 

  29. Mehrotra, S., & Mander, H. (2009). How to identify the poor? A proposal. Economic and Political Weekly, 44, 37–44.

    Google Scholar 

  30. Mukherjee, N. (2005). Political corruption in India’s below the poverty line (BPL) exercise: Grassroots’ perspectives on BPL in perpetuating poverty and social exclusion. New Delhi: Development Tracks in Research, Training & Consultancy.

    Google Scholar 

  31. Narayan, A., & Yoshida, N. (2005). Proxy means test for targeting welfare benefits in Sri Lanka. Report no. SASPR-7. South Asia Region. Prem working paper series. The World Bank.

  32. Olken, B. A. (2006). Corruption and the costs of redistribution: Micro evidence from Indonesia. Journal of Public Economics, 90, 853–870.

    Article  Google Scholar 

  33. Roy, I. (2011). “New” list for “old”: (Re-) constructing the poor in the BPL census. Economic and Political Weekly, 46(22), 82–91.

    Google Scholar 

  34. Royal Government of Bhutan, Gross National Happiness Commission (2011). Eleventh roundtable meeting - turning vision into reality: The development challenges facing Bhutan. Thimphu, Bhutan.

  35. Saxena, N. C. (2009). Report of the expert group to advise the Ministry of Rural Development on the methodology for conducting the below poverty line (BPL) census for 11th five-year plan. New Delhi: Ministry of Rural Development, Government of India.

    Google Scholar 

  36. Sen, A. K. (1976). Poverty: An ordinal approach to measurement. Econometrica, 44, 219–231.

    Article  Google Scholar 

  37. Sen, A. K. (1992). Inequality reexamined. Cambridge: Harvard University Press.

    Google Scholar 

  38. Sen, A. K. (1996). On the foundations of welfare economics: Utility, capability and practical reason. In F. Farina, F. Hahn, & S. Vannucci (Eds.), Ethics, rationality and economic behaviour (pp. 50–65). Oxford: Clarendon Press.

    Google Scholar 

  39. Sen, A. K. (2004). Capabilities, lists and public reason: Continuing the conversation. Feminist Economics, 10, 77–80.

    Article  Google Scholar 

  40. Sharan, M. R. (2011). Identifying BPL households: A comparison of competing approaches. Economic and Political Weekly, 46(26–27), 256–262.

    Google Scholar 

  41. Sundaram, K. (2003). On identification of households below poverty line in BPL census 2002: Some comments on proposed methodology. Economic and Political Weekly, 38, 896–901.

    Google Scholar 

  42. Thomas, B. K., Muradian, R., de Groot, G., & de Ruijter, A. (2009). Multidimensional poverty and identification of poor households: A case from Kerala, India. Journal of Human Development and Capabilities, 10, 237–257.

    Article  Google Scholar 

Download references

Acknowledgments

This paper has gone through many versions. We are grateful to participants in the June 2008 OPHI research meeting in Oxford, the 2008 HDCA conference in New Delhi, the WIDER conference on Frontiers of Poverty Research in Helsinki, the 2011 International Economic Association Sixteenth World Congress in Beijing, the 2010 Michaelmas OPHI Lunchtime Seminar Series in Oxford, the 2011 South Asia in transition conference in Oxford, and to Jean Drèze, Reetika Khera, Rinku Murgai, Abhijit Sen and anonymous referees for comments on previous versions of this draft. All errors remain our own.

Author information

Affiliations

Authors

Corresponding author

Correspondence to Sabina Alkire.

Appendices

Appendix 1

See Table 9.

Table 9 2002 BPL census questions

Appendix 2

See Table 10.

Table 10 Pseudo-criteria for the Saxena method and the Socio-Economic Caste Census (SECC) 2011 method, using NFHS-3 data

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Alkire, S., Seth, S. Selecting a Targeting Method to Identify BPL Households in India. Soc Indic Res 112, 417–446 (2013). https://doi.org/10.1007/s11205-013-0254-6

Download citation

Keywords

  • Multidimensional poverty
  • Below the poverty line (BPL)
  • Socio Economic Caste Census
  • Targeting methods
  • Binary scoring
  • Poverty in India