'POVTIME': module to compute aggregate intertemporal poverty measures Carlos Gradín Universidade de Vigo 1 Description • ‘povtime’ computes aggregate intertemporal poverty measures (poverty accounting for time) in a balanced panel of individuals. • The program computes the family of FGT-type intertemporal poverty measures proposed in: – Gradin, Del Rio, and Canto ("Measuring Poverty Accounting for Time", Review of Income and Wealth, 58(2): 330-354, 2012). • Other measures that can be interpreted as particular cases of this general family: – Foster (“A Class of Chronic Poverty Measures” in Poverty Dynamics: Interdisciplinary Perspectives, OUP, 2009) and – Bossert, D'Ambrosio and Chakravarty ("Poverty and Time", Journal of Economic Inequality, 2012. 2 Measuring poverty • Poverty in a cross-section of individuals • y=(y1, y2, ... , yq , yq+1, ... , yN) Poor z Non poor • Poverty index: P(y; z) 1 FGT ( ) N z yi z , i 1 q 0 FGT(0) = Headcount rate (H=q/N) FGT(1) = Poverty gap ratio (HI) FGT(2) = Poverty severity • Stata modules: povdeco, apoverty, sepov 3 Measuring longitudinal poverty • Poverty in a (balanced) panel • N Individuals observed T times Y y11 y12 ... y21 y22 ... . . ... . . ... . . ... yN 1 yN 2 ... y1T y2 T . . . y NT y1 y2 Y y1 , y2 ,..., yN ' yN • Poverty index?: P(y; z) • Stata modules: povtime z y t it git zt 0 if yit zt otherwise 4 i) summarize the complete individual information in time individual intertemporal poverty index 1 T pi yi ; z g it wit T t 1 sit wit T 0 0 ii) then construct an aggregate poverty index that takes into account a social preference for equality among individuals p1 p2 ... pN pi if pi 0 pi 0 if pi 0 1 P Y ; z N p i N i 1 5 Gradín, Cantó and del Río (RIW, 2012) 1 N 1 T sit git N i 1 T t 1 T P Y ; z 1 N 0 q pi N i 1 N 𝛽 ≥ 0; 𝛼 ≥ 0; 𝛾 ≥ 0 if 0 if 0 P satisfies all desirable properties for 0 , 1 Foster (OUP 2009) 𝛽 = 0; 𝛼 = 1; 𝛾 ≥ 0 Bossert, D’Ambrosio and Chakravarty (JOEI 2012) 𝛽 = 1; 𝛼 = 1; 𝛾 ≥ 0 6 Advantages • A code that allows for measuring agggregate poverty in a panel – complementing existing codes for measuring poverty in a cross-section – following various measures recently proposed in the literature, – in a way consistent with how poverty is measured in a cross-section. • Easy to undertake in-depth analysis – robustness (dominance analysis), – decomposition into components (incidence, intensity, inequality), – analysis of the distribution of individual poverty indices, etc. • Easy to obtain inference using bootstrapping 7