*This preview shows
page 1. Sign up
to
view the full content.*

**Unformatted text preview: **count Poverty Rate 97.0 77.0 1600 96.0
95.0
94.0 1300 93.0 constant 2005 international $ 98.0 78.0 Headcount Poverty Rate 1982 4000
1980 92.0 72.0
91.0 71.0 90.0 1000
1980 70.0 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 1000
1980 1982 1984 1986 1988 1990 2 USD a day 1992 1994 1996 1998 2000 2002 GDP per capita, PPP (constant 2005 international $) 4 USD a day 2004 Source: World Bank, World Development Indicators and POVCALNET 75 GDP per capita, PPP (constant 2005 international $) 2004 Annex 3. Alternative Specifications of Poverty Elasticities Table A3.1 presents the coefficients of the Poverty‐GDP pc elasticity using a linear spline transformation. This exercise allows the estimation the relationship between y and x as a piecewise linear function, which is a function composed of linear segments. In this particular case, the first linear segment represents the Poverty‐per capita GDP elasticity for periods of economics crises (negative per capita GDP change), while the second linear segment represents the Poverty‐per capita GDP elasticity for periods of economic growth (positive per capita GDP change). Two models were estimated, with two different specifications. The first model looked at the extreme poverty elasticity, and the second looked at the moderate poverty elasticity. The two alternative specifications considered the full dataset and all observations but Argentina, given the very particular magnitudes of the changes in this country. All models were estimated using ordinary least squares (OLS) algorithm, with standard errors clustered at the country level. These models allow us to test the equality of the elasticities during periods of economics crisis and growth. As it can be seen in the last two lines of Table A5.1 none of the specifications allowed the rejection of the hypothesis that the coefficients are identical. Table A3.1: Poverty Elasticities using Splines (OLS) Extreme Poverty Moderate Poverty Full Without Argentina Full Without Argentina d_lngdppc: (.,0) ‐2.481 ‐1.494 ‐1.332 ‐0.732 (1.221) (0.822) (0.684) (0.392) d_lngdppc: (0,.) ‐3.767** ‐3.965* ‐2.423* ‐2.547* (1.178) (1.315) (0.929) (1.037) d_lngini3 ‐0.015 ‐0.011 ‐0.010 ‐0.008 (0.008) (0.008) (0.006) (0.006) Year ‐0.001 ‐0.002* ‐0.000 ‐0.001 (0.001) (0.001) (0.001) (0.001) Constant 2.534 4.426* 0.698 1.806 (1.862) (1.646) (1.148) (1.053) Adj.R‐squared 0.335 0.301 0.348 0.310 Obs. (unweighted) 107 102 96 91 Test: d_lngdppc_a1 ‐ d_lngdppc_a2 = 0 F‐stat 0.323 1.346 0.529 1.632 P‐val 0.580 0.271 0.480 0.226 Note: clustered standard errors; population weighted. Inference: * p<0.05, ** p<0.01, *** p<0.001 76 Table A3.2 presents the same models presented in Table A3.1, estimated as median regression. The advantage of this model over the OLS is the fact that it is much more robust to the presence of outliers. As it can be seen the moderate poverty elasticities for periods of economics crisis [d_lngdppc: (.,0)] are not very differen...

View
Full
Document