180 likes | 271 Views
An Innovative Measurement Method of Basic Needs Mixing Objective and Subjective Information Work in Progress Christophe Muller DEFI, AMSE, Aix-Marseille University July 2011. 1. Introduction. ▪ A proper notion of poverty in society Corresponds to the well-thought opinions of citizens
E N D
An Innovative Measurement Method of Basic Needs Mixing Objective and Subjective InformationWork in Progress Christophe MullerDEFI, AMSE, Aix-Marseille University July 2011
1. Introduction • ▪ A proper notion of poverty in society • Corresponds to the well-thought opinions of citizens • Conflicts with current approaches of poverty lines and poverty statistics • Expert opinions • Biological benchmarks • Arbitrary statistics (1 $ a day, half median...) • What people think poverty means ▪ Uses of Self-Reported Basic Needs • Poverty and income distribution analyses • Individual and household decision models
Potential issues with self-evaluated basic needs • -Comparability across respondents • Non independence from outcomes to explain • Application to assistance system • Financial Incentives to lie • Less reliable than objective measures • Comparing self-assessed needs with consumption for each household (i.e. distribution matches) yields too noisy results to be usable • -hard to observe accurately • insincere answers • unclear to respondent • no clues • erratic individual effects
Potential advantages of self-evaluated basic needs • There is reliability (~ 0.5) • how to best extract the relevant core information • No consensus on the poverty line method anyway • Unreasonable methods are not rare • Using nutritional minima is unrealistic for many countries • - not subject to the ignorance of individual situations by external observers • utility-consistent if individuals know what is best for them • do not always require equivalence scales
2. Context and Data • Republic of Mauritius in 2006/7 • 2006 Household Budget Survey • Nutritional Poverty Profile • Request for adaption of poverty statistics to an advanced development stage • A Special Survey for Measuring Subjective Basic Needs • 2008 Living Condition Survey • Collaboration CSO-UNDP • Aim: getting better poverty lines anchored on realistic basic needs • Sub-sample of 2006 Household Budget Survey • Uses of the new poverty lines • Official poverty statistics • Targeting statistics • Design and improvement of social policies in Mauritius
Our Strategy for Basic Needs Indicators • Selecting logically consistent answers • An observed household is deemed consistent when: • either (1) its consumption is in excess of its self-stated basic needs AND it declares itself as non-destitute in a considered qualitative question; • or (2) its consumption is below its self-stated basic needs AND it declares itself as destitute in a considered qualitative question. • For different categories of goods • Controlling for individual erratic effects • - Concentrating on food basic needs: the better observed needs and consumption • Aggregating to use a central tendency as anchor for the poverty line estimation • Excluding outliers and mistakes • Controlling for individual effects: • * A new econometric method for cross-section regressions • * Extracting individual effects from other basic needs equations
A New Method for Individual Effect Control • Taking advantage of similar phenomena simultaneously measured for the same individuals • Self-Assessment of basic needs for several consumption categories: food, housing, clothing, health, education • SMij, j= 1,...5 are the goods, i is the individual index • The model: SMij = gj(Xi) fi uij • Xi are typical independent variables, • fi is the unobserved individual effect variable • uij are error terms
Simple estimators of individual effects fi can be generated from each secondary good equation • Empirical analogs of: • ln(SMij ) – Mean(ln(SMij)) - gj(Xi) + Mean(gj(Xi) ) • For j different from 1 • To include in the ln(SM1) equation for food.
Correlates of Consistent Log Food Basic Needs • Number of obs = 920 R-squared = 0.4793 • Coef. Std. Err. t P>|t| • ei_cloth | .0959786 .0171521 5.60 0.000 • ei_housing | .0965442 .0248727 3.88 0.000 • ei_health | .0208055 .0127216 1.64 0.102 • age | .0016652 .0013742 1.21 0.226 • room | .0016167 .0074483 0.22 0.828 • sex | -.1685794 .0472513 -3.57 0.000 • n13 | .0923126 .0240365 3.84 0.000 • n410 | .0687213 .0181334 3.79 0.000 • n1116 | .1238146 .0183668 6.74 0.000 • n1721 | .1175631 .0185373 6.34 0.000 • n2259 | .152178 .0125792 12.10 0.000 • n60 | .1980666 .0254644 7.78 0.000 • district_d~2 | .0792172 .0809322 0.98 0.328 • district_d~3 | .1448836 .0819476 1.77 0.077 • district_d~4 | .2223407 .0802721 2.77 0.006 • district_d~5 | .2521609 .083598 3.02 0.003 • district_d~6 | .1573245 .0841895 1.87 0.062 • district_d~7 | .0275595 .0435975 0.63 0.527 • district_d~8 | .2928428 .0860717 3.40 0.001 • district_d~9 | .2066021 .0919369 2.25 0.025 • district_~10 | .0630789 .0845483 0.75 0.456
building_d~2 | .0173235 .0265064 0.65 0.514 • tenure_dum~1 | .060169 .0454859 1.32 0.186 • tenure_dum~3 | .0426515 .0536443 0.80 0.427 • educ_no~y | .0172653 .0292946 0.59 0.556 • educ_high | .0195854 .0291216 0.67 0.501 • educ_co~e | .0151804 .0567269 0.27 0.789 • activit~1 | .0595621 .0360397 1.65 0.099 • activit~2 | .0626847 .0416989 1.50 0.133 • activit~4 | .1103974 .0553114 2.00 0.046 • cooklpg_du~y | .030092 .042303 0.71 0.477 • lcsmarital~1 | .0199681 .041463 0.48 0.630 • car_dummy | .0156389 .0375727 0.42 0.677 • van_dummy | .0630376 .0651886 0.97 0.334 • dcab_dummy | .0133242 .0798519 0.17 0.868 • mcycle_dummy | .0440583 .0260842 1.69 0.092 • lsp_sq | -2.92e-09 3.79e-10 -7.69 0.000 • lsp | .0000894 9.43e-06 9.48 0.000 • savings | .0087989 .0238832 0.37 0.713 • priority_~p1 | .0503988 .0308288 1.63 0.102 • priority_e~4 | .0343247 .0323022 1.06 0.288 • priority_e~5 | .0330141 .0434224 0.76 0.447 • priority_e~8 | .0320459 .0384065 0.83 0.404 • reqsocialaid | -.0343265 .0314591 -1.09 0.276 • check1 | -.0820029 .0311165 -2.64 0.009 • telephone | -1.44e-08 6.21e-09 -2.32 0.021 • urbanrural | -.0862697 .037277 -2.31 0.021 • _cons | 7.536526 .1631736 46.19 0.000
Poverty Line Estimation • - Accounting for consumer substitutions • Based on an estimated food Engel curve • Linearized QAIDS • Mean self-assessment of their food basic needs by consistent households → defining food poverty thresholds: ZF • si = α + β ln(xi) + γ [ln(xi)]2 + Ni’ δ + εi, • Food budget share of household i = si • Total expenditure of household i = xi • Household and environment characteristics = Ni
Solving for the poverty line • Once the parameters are estimated the poverty line, Zj is obtained by solving in Zthe following equation: • ZF/Z = Concentrated intercept + β ln(Z) + γ [ln(Z)]2 • For example with a Newton method • Poverty line for Mauritius: 2217 Rupees a month. • For Rodrigues: 1556 Rupees a month. • 7.06 percent of households are under the poverty line. • The poverty rates: • 7.79 percent in the whole Republic • 7.54 percent in Mauritius Island • 15.2 percent in Rodrigues.
Table 1: Estimated Poverty Rates (%) Comparison of general poverty profile and nutritional poverty profile : Table 2 – Headcount poverty rates by urban & rural regions
Higher general poverty for households led by : • Unemployed heads • Separated heads or widows • Female heads • Elderly heads • Little educated heads • Other categories of households especially affected by general poverty are: • Large size households • Households dwelling in disadvantaged areas in terms of the Relative Development Index used in Mauritius to characterized disadvantaged area. • Higher levels of poverty measures than with nutritional profile, while still realistic.
5. Conclusion • A new method mixing subjective and objective information to estimate basic needs and poverty lines • More realistic than current methods, except for extremely poor countries: eliciting the well-thought opinion of the population • Matching a special survey with typical household budget survey • Selection of consistent answers using destitution information • A new method for controlling for individual effects in cross sections • Large number of independent variables in basic needs equations, including individual effects, living standard, demographics, human capital, environment, collection checks, relative income… • Central tendency food to anchor poverty line instead of matching distributions of needs and consumptions • Yields an ‘objective’ core from subjective data • Application to poverty analysis and social policy in mauritius