Analysis of Poverty and Vulnerability and Its Determinants in Rural Hubei

Transcription

1 6th PEP Research Network General Meetng Sheraton Lma Hotel, Paseo de la Republca 170 Lma, Peru June 14-16, 2007 Analyss of Poverty and Vulnerablty and Its Determnants n Rural Hube Shjun Dng Mnstero de Economía y Fnanzas For more nformaton about the 6th PEP Research Network General Meetng, Please vst:

2 Analyss of Poverty and Vulnerablty and Its Determnants n Rural Hube RESEARCH PROPOSAL Presented to PEP Network By Shjun Dng & Yupng Chen Hatao Wu Zhe L L Feng CHINA May 14, 2007 PDF created wth pdffactory tral verson

3 Abstract Ths research s proposed to analyze poverty and vulnerablty, and ts determnants n rural Hube. The man objectves nclude: 1) measurng poverty and vulnerablty to poverty by household groups n rural Hube; 2) dstngushng the determnants of poverty (transent and chronc) and vulnerablty to poverty n rural Hube; 3) suggestng polcy optons for government to adopt for reducng poverty and vulnerablty. Data employed for analyss wll be a unque panel data set of 1900 households surveyed between 2000 and 2004, from Rural Household Survey Team of the Hube provncal Statstc Bureau. We wll decompose measured household poverty nto chronc and transent components, and classfy the household accordng to ther characterstcs and measure each group s chronc poverty, transent poverty, and vulnerablty whch we wll adopt an emprcal method to quantfy. We wll then use Censored Quantle Regresson to nvestgate the determnants of poverty and vulnerablty usng panel data for rural Hube. It s expected that the research team members, consstng of two males and three females, two senors and three young PhD students aged 20s, wll gan much experence and knowledge n analyzng and understandng poverty and vulnerablty, and are able to do further and broader analyss by employng approaches establshed n ths research project. Ths wll certanly mprove the research capacty of the ndvdual team members and the research nsttuton. 1. Man research questons and core research objectves Economc growth n the past two decades n Chna has no doubt mproved lvng standard across the country, and analyss of poverty trends, by whatever measures, has showed a dramatc declne, especally n rural areas. However, recent years studes suggest that the speed of poverty reducton n rural Chna become slower. By the end of 2004, PDF created wth pdffactory tral verson

4 there are 26.1 mllons absolute poverty populatons n rural Chna, and a large number of households hover around poverty lne. Although a seres of polces of ex-post poverty allevaton nterventons have been mplemented, poverty detecton shows that there s hgh rato of households fallng back nto poverty. Addressng the objectve of poverty reducton, polces should hghlght not only ex-post poverty allevaton nterventons but also ex-ante poverty preventon nterventons, whch had been emphaszed by the World Bank s Socal Rsk Management framework whch hghlghts three types of rsk management strateges: preventon, mtgaton and copng (Holzmann et al, 2000). It s clearly necessary to go beyond measurng poverty to catalogng of who s currently poor, how poor they are, and why they are poor. An assessment of household s vulnerablty to poverty - to fgure out who s lkely to be poor, how poor are they lkely to be, and why they are vulnerable to poverty, seems to be more mportant. Poverty and vulnerablty (to poverty) are two sdes of the same con. Poverty s an ex-post measure of a household s well-beng. It reflects a current state of deprvaton. Vulnerablty, on the other hand, may be broadly construed as an ex-ante measure of well-beng, reflectng not so much how well off a household currently s, but what ts future prospects are (Chaudhur, 2003). Current vulnerablty lkely causes future poverty. As decomposng poverty nto transent and chronc poverty, a vulnerable household wll run nto transent poverty when t s exposed to adverse shocks, on the other hand t wll get nto chronc poverty for beng exposed to adverse shocks and havng lmted long-term ncome generatng capacty. Poverty reducton has become an mplct objectve of development polcy n Hube, a PDF created wth pdffactory tral verson

5 provnce n central Chna. Based on the above dscusson, to explore vulnerablty and poverty n rural Hube, there are a few questons that need to be further explored and answered: 1) How to measure poverty and vulnerablty to poverty for households n rural Hube? 2) Who s more lkely to fall nto transent poverty or chronc poverty? What are the major characterstcs of households n transent and chronc poverty, respectvely, n rural Hube? 3) Who s vulnerable to poverty? What are the major characterstcs of households wth vulnerablty n rural Hube? Do characterstcs featurng households wth poverty and households wth vulnerablty dffer? 4) What are the determnants of poverty (transent and chronc) and vulnerablty, respectvely, for households n rural Hube? 5) Can polces that reduce poverty also reduce the vulnerablty to poverty? Whch strateges shall be chosen to reduce poverty and vulnerablty, respectvely? Core research objectves of ths research nclude: 1) Measurng poverty and vulnerablty to poverty n rural Hube; 2) Dstngushng the determnants of poverty and vulnerablty to poverty n rural Hube; 3) Suggestng polcy optons for government to adopt for reducng poverty and vulnerablty? 2. Scentfc contrbuton of the research ncludng a short lst of key references n the lterature and knowledge gaps The prevous studes on poverty show a transton from statc poverty to dynamc poverty. Pror to the emergence of the dynamc poverty lterature, economsts were already concerned about the poverty decomposng and vulnerablty to poverty. PDF created wth pdffactory tral verson

6 Researchers have developed approaches to measure transent and chronc poverty (Ravallon, 1988; Morduch, 1994; Thorbecke, 1998; Hulme, 2003; Duclos et al., 2006), and to explore ts determnants (Jalan and Ravallon, 1998, 2000; Yue et al., 2005). Ravallon (1988) attempted to decompose total poverty nto a chronc and a transent component. Jalan et al. (1998) defned transent poverty as the poverty that can be attrbuted to nter-temporal varablty n consumpton, and dstngushed transent from chronc poverty usng data from rural Chna. Usng robust sem-parametrc methods, they found that households average wealth holdng s an mportant determnant for both transent and chronc poverty, however, household demographcs, educaton levels and health status of the household members whle mportant for chronc poverty were not sgnfcant determnants for transent poverty. Yue et al. (2005) nvestgated chronc and transent poverty and the determnants of each component. They found transent poverty account for large proporton of total poverty observed n the poor rural area of Chna, and demographc characterstcs of households are very mportant to the poverty status of households. Duclos et al. (2006), usng data from 82 vllages n 9 provnces of Chna, developed the measurement of transent and chronc by provdng correctons for the statstcal bases ntroduced. Both of transent poverty and chronc poverty reflects the ex-post poverty dynamcs. Scholars have ncreasngly recognzed that explorng vulnerablty s very necessary for understandng ex-ante poverty dynamcs and polcy nterventons. Carlo et al. (2006) announced that rsk and vulnerablty consderatons n poverty analyss would be future drectons. Researchers have developed dfferent approaches to nvestgate vulnerablty based PDF created wth pdffactory tral verson

7 on dfferent conceptual frameworks. These ncluded approaches that use change or varablty of consumpton; that use the dfference between expected consumpton and the poverty lne; that use the expected poverty; and that use the probablty to fall nto poverty. Coudouel and Hentschel (2000) used average ncome and ts standard devaton to measure vulnerablty. Kamanou and Morduch (2002) ponted out that f the hstory of shocks s a predctor of future shocks, a smple startng pont for measurng vulnerablty would be to compare standard devatons of consumpton and ncome changes, and households or groups are judged to be more vulnerable f standard devaton of past consumpton changes are hgher. An mmedate practcal problem s that few data sets have a long enough tme dmenson to yeld a relable standard devaton for each household over tme. The technques dscussed above are statc n nature as no efforts were made to predct future poverty. Kamnous and Morduch (2002) further proposed a dynamc framework, and defned vulnerablty n a populaton as the dfference between the expected value of a poverty measure n the future and ts current value. By defnng vulnerablty to poverty as expected poverty, Chaudhur (2003) measured the vulnerablty of rural Chna (Guangdong, Guangx, Yunnan and Guzhou). Researchers ncreasngly defne vulnerablty as the probablty to fall nto poverty. The World Development Report (2000/2001) defned vulnerablty as the rsk that a household or ndvdual wll experence an epsode of ncome or health poverty over tme. Prtchett et al. (2000) calculated the Vulnerablty to Poverty Lne as the level of expendtures below whch a household s vulnerable to poverty and mplement the approach usng two panel data sets from Indonesa. Lgon et al. (2003) took a utltaran PDF created wth pdffactory tral verson

8 approach to defne vulnerablty n a rsky envronment and construct a measure of vulnerablty. Applyng ther measure to a panel dataset from Bulgara n 1994, they found that poverty and rsk play roughly equal roles n reducng welfare. McCulloch et al. (2003) estmated the determnants of chronc poverty and vulnerablty usng the data of rural Schuan and found the determnants of chronc poverty and vulnerablty appears to be smlar, suggestng that polces to reduce chronc poverty wll also reduce vulnerablty. Zhang et al. (2006) followed Kuch (2003) for measurng vulnerablty and explored whether dversfcaton and educaton affect vulnerablty n southeast coastal rural Chna. Chaudhur (2003) took vulnerablty level of a household h at tme t as the probablty that the household wll fnd tself consumpton poor at tme t+1: V ht = Pr( ch, t+ 1 z ) Where c h, t+ 1 s the household s per capta consumpton level at tme t+1 and z s the approprate consumpton poverty lne. Researchers attempt to explore the lnk between dfferent methods of decomposng poverty nto transent and chronc components and measurng vulnerablty. Barrentos et al. (2004) summarzed three man approaches that are used to decompose poverty, ncludng duraton approach, shortfall approach and predcted consumpton approach. The duraton approach defnes transent and chronc poverty focused on the duraton of poverty, and those whose ncome or consumpton s persstently below the poverty lne are dentfed as chronc poor. The shortfall approach suggests ncome and consumpton have constant and fluctuatng components, and decomposes chronc poverty from transent poverty accordng to permanent component of ncome or consumpton beng at or below the poverty lne. The predcted consumpton approach defnes a household as chronc poor PDF created wth pdffactory tral verson

9 f ts current consumpton s at or below the poverty lne and, on the bass of current nformaton, has a hgh probablty that future consumpton wll also be at or below the poverty lne. Both duraton approach and shortfall approach exclude rsk and vulnerablty as a relevant factor n dentfyng chronc poverty. However, the predcted consumpton approach explctly ncorporates rsk and vulnerablty nto chronc poverty. Practcally, researchers try to combne poverty ndces wth measurement of vulnerablty. Prevalent poverty ndex s Foster, Greer and Thorbecke (FGT) measure: z y p ( y ) t = t z f y z t < = 0 otherwse α By defnng vulnerablty as expected poverty, Chaudhur (2003) adopted the FGT measures as the base for constructng measures of vulnerablty: V α, t α z z y, t+ 1 f ( y, t+ 1) = E[ p,, + 1(, + 1) (, + 1)] = ( ) ( ) α t y t F y t F z dy, t+ 1 y z F( z) Where F ) and f ) respectvely denote the cumulatve dstrbuton and ( y, t+ 1 ( y, t+ 1 densty functon of y, t + 1. The approach assocated vulnerablty measure wth poverty ndex. The dstncton between poverty measure and vulnerablty measure s that vulnerablty to poverty s the expected poverty. The nature of the measure changes when varyng the value of α. The smple headcount rato s found by α=0, whle ncreasng values of α ncrease the sensblty of the measure to ncome or consumpton gap. When α=1, the ndex becomes the poverty gap rato, and wth α=2, the ndex means square poverty gap (SPG). The three ndces reflect dfferent characterstcs of poverty, and can explan poverty from dfferent profles. Unlke the head-count ndex (proporton of people below the poverty lne) or poverty gap ndex (mean proportonate dstance below the lne), PDF created wth pdffactory tral verson

10 the squared poverty gap penalzes nequalty among the poor (Foster, et al., 1984). As summarzed above, whle prevous studes focus on poverty measurement and poverty decomposng, recent studes have ncreasngly amed to measure vulnerablty and to explore the methods lnk between decomposng poverty and vulnerablty. In understandng poverty and vulnerablty and ts determnants n rural Chna, a few studes recently have emprcally nvestgated household transent poverty (Yue et al., 2005) and vulnerablty (Zhang and Wan, 2006). However, there are needs n further nvestgatng poverty and vulnerablty n rural Chna, especally focus on central Chna. The followng gaps need to be flled n. 1) Although there are studes decomposng poverty nto transent and chronc ones, there s less detaled nvestgaton on ts determnants usng current nformaton n rural Chna. 2) Study on vulnerablty n rural Chna s few. No such study focuses on rural Hube. Relevant study focuses on more developed coastal area (for example, Zhang and Wan (2006) used 600 households data n Shangha, the most developed area n Chna) and less developed southwest areas. They may have less relevance for central parts of Chna gven that there s tremendous economc and geographc dversty n rural Chna. 3) Few studes compared poverty and vulnerablty n terms of dfferent household groups and ndvduals characterstcs, such as household sze, educaton, age, gender, asset and geography. There s no comparson between the determnants of poverty and determnants of vulnerablty n rural Chna. 4) Research methods and approaches on studes of vulnerablty and poverty need to be appled to rural Chna. More understandng on poverty and vulnerablty s requred, so that government ant-poverty polces can more effcently target dfferent groups. PDF created wth pdffactory tral verson

11 3. Polcy relevance Poverty reducton has become an mplct objectve of development polcy n Chna. Although the poverty allevaton funds have rsen year after year, poverty has not been totally elmnated, and rato of fallng back nto poverty s hgh. To reduce poverty more effectvely, ant-poverty nterventon should carry out from two essental aspects. One s ex-post poverty allevaton nterventon such as provdng subsdes, relef, and relevng taxes, the other s ex-ante poverty prevent nterventon such as buldng capacty, educaton, offerng opportuntes of work for the poor, so as to reduce vulnerablty to poverty. To mplement the two nterventons effectvely, separately targeted mechansms to both poor and vulnerable households are essental, so that ant-poverty nterventons can reach the target groups. It s crucal that ant-poverty nterventons target the conformable groups, for example, free tranng for the less capable poor, socal relef for households who suffered temporarly adverse shocks. Clearly there are dfferent needs among households wth poverty and wth vulnerablty. Can the same ant-poverty measure be appled to dfferent groups wth poverty (transent and chronc) and wth vulnerablty? Should poverty reducton and vulnerablty reducton programs adopt dfferent strateges? These questons are obvously polcy-relevant. Ths research project ams to provde evdence for settng dfferent polces targets, and to suggest alternatve polcy nterventons. 4. Methodology Presentaton of the specfc technques that wll be used to answer the research questons and how exactly they wll be used to do so 1) Decomposng poverty Although the predcted consumpton approach (as dscussed before) explctly PDF created wth pdffactory tral verson

12 ncorporates rsk and vulnerablty nto chronc poverty, the approach s complex to decompose poverty, and does not separate vulnerablty from poverty. The shortfall approach s employed n ths study. Accordng to Jalan and Ravallon (1998), the transent component (T ) of aggregate poverty (P(.)) can be defned as T C = P, y ) C ( y 1, y2, L = P( Ey, Ey, L, Ey t ) Where ( y, y, 2, y ) 1 L t s household s consumpton stream over t perods, and P y, y, L, y ) s an aggregate nter-temporal poverty measure for household, whch ( 1 2 t s the sum of the chronc components ( C ) and transent components ( T ). The chronc poverty component ( C ) s defned as ts value f consumpton dd not vary around the tme mean, and Ey s the expected value of consumpton over tme for household. Poverty wll be emprcally measured as follow: z y P ( y ) t = t z f y z t < = 0 otherwse α Here z s poverty lne. In ths study, two poverty lnes wll be used for analyss: One wll be assessed by usng Extended Lnear Expendture System (Shaukat, 1995); the other wll be natonal poverty lne of Chna followed the World Bank s defnton. When α =0, the poverty ndex s smply a bnary ndcator of whether a household s poor. By ths the poverty headcount s calculated. When α =1, the ndex becomes the poverty gap rato, and wth α =2, the squared poverty gap, reflectng nequalty amongst the poor. The three ndces reflect dfferent characterstcs of poverty, and can explan poverty from dfferent profles. The emprcal poverty measure employed n ths study wll be the squared poverty gap (SPG) of Foster, Greer and Thorbecke (1984) ndex. PDF created wth pdffactory tral verson

13 2) Measurng vulnerablty to poverty Several approaches can be used to measure vulnerablty. Researchers, however, are ncreasngly usng probablty approach to measure t (as proposed by, for example, Prtchett et al. (2000) and Chaudhur (2003)). To make ths study comparable wth prevous studes, probablty approach wll be employed n ths study. In addton, for comparng determnants of poverty and vulnerablty, the expected poverty approach wll then be used to measure vulnerablty. Followng Chaudhur (2003) for measurng vulnerablty, for a gven households, the vulnerablty at tme t s defned as the probablty of ts consumpton beng below poverty lne at tme t+1: V ( ln y ln z) t = Pr, t + 1 < Where V s vulnerablty of household at tme t, and ln y, t + 1 denote the t consumpton of household at tme t+1. Conceptually, the household s consumpton conssts of three parts: tme-varyng consumpton (e.g. medcal expendture and agrcultural nputs), tme-nvarant consumpton (e.g. educatonal level and dependency rato), and a zero mean measurement error term. The consumpton functon s estmated as: ln y = P γ + R λ + e t t Where P t and R respectvely represents a vector of tme-varyng and tme-nvarant characterstcs for household, and et s a mean-zero dsturbance term that captures dosyncratc shocks that contrbute to dfferent per capta consumpton levels. It s assumed that the structure of the economy s relatvely stable over tme and hence, future consumpton stems solely from the uncertanty about the dosyncratc PDF created wth pdffactory tral verson

14 shocks, e. For non-based estmate of the parameters, we need to allow the varance of the dsturbance term (heteroskedastcty). We assume that the varance of the dsturbance term depends on: ln δ 2 e = Ptθ + R t η + u t Where 2 δ e t s the varance of the dsturbance term ( e t ).The estmates of γ, λ,θ and η could be obtaned usng a three-step regresson. Then we can compute the expected log consumpton and the varance of log consumpton for each household as follows. E [ln y P, R ]= P + R λ t t t γ V [ln y θ η t Pt, R ] = Pt + R Assumng that consumpton s log-normally dstrbuted, Φ(.) denote the cumulatve densty of the standard normal dstrbuton, the estmates of vulnerablty can be straght-forwardly generated usng the propertes of the normal dstrbuton. V t = Pr ( ) = Φ ln z µ 2 ln y, t + 1 ln z Ptγ Rλ ln y, t+ 1 < ln z µ ln y, δ = Φ, t + 1 ln y, t + 1 δ ln y P θ + η + t R, t 1 As mentoned above, to compare determnants of poverty and vulnerablty, measures of poverty and vulnerablty should be based on same ndex, for example the FGT ndex, whch can be taken to calculate the expected poverty. Vulnerablty can then be expressed as follow: V * t = z Ey z t α f Ey t < z = 0 otherwse Where * V t s vulnerablty (also known as expected poverty). SPG ndces of poverty and expected poverty (vulnerablty) can then be calculated for nvestgatng determnants of poverty and vulnerablty. PDF created wth pdffactory tral verson

15 3) Investgatng the determnants of poverty and vulnerablty To nvestgate the determnants of poverty and vulnerablty, the measures of poverty and vulnerablty are, respectvely, regressed on the same set of household characterstcs. The model of transent poverty s: T = f T 0 Where * T * > * T T T = x β + µ = 0 otherwse Where * T s a latent varable, T s the observed transent poverty, T β s a kx 1 vector of unknown parameters, x s a kx 1 vector of explanatory varables, and T µ are the model resduals. The analogous model for chronc poverty and vulnerablty are gven by: C = f C * > 0 Where * C * C C C = x β + µ = 0 otherwse * v v V = x β + µ Accordng to Jalan and Ravallon (1998), Powell s Censored Quantle Regresson (CQR) model s used to estmate the determnants of poverty and vulnerablty (the model can be estmated usng the QREG command n STATA software). 5. Data requrements and sources For measurng poverty and vulnerablty and comparng the determnants of poverty and vulnerablty, ths research requres detaled nformaton on characterstcs of household such as household sze, demographcs and resources endowment, and ts ncome and consumpton expendture. Net ncome refers to the household s ncome n cash and n-knd after the deducton of all costs and taxes. Consumpton expendture refers PDF created wth pdffactory tral verson

16 to expendture on food, clothng, housng, commodtes, health, educaton and recreaton, transportaton and communcaton. Gven that the probablty to be poor s assgned to a gven year or successve years, the mpact of the number of retaned perods and ther duraton on the measurement of poverty and vulnerablty should be noted. Researchers have used dfferent number of retaned duraton on the measurement of poverty and vulnerablty. Jalan and Ravallon (1998) measure transent and chronc poverty usng sx years household data of western Chna. Chaudhur (2003) adopt sx years data of Southern Chna. Kozel et al. (2005) assess poverty and vulnerablty of Zamba usng four years data between 2002 and McCulloch and Calandrno (2003) measure vulnerablty and chronc poverty usng fve years data of Schuan provnce n Chna. Hulme and Shepherd (2003) support crude fve year crteron n poverty analyss based on three reasons. Frstly, n most cultures, fve years s perceved as a sgnfcant perod of tme n an ndvdual s lfe-course. Secondly, there s often a fve-year gap between data collecton ponts when panel data are created. Fnally, some emprcal materals ndcate that people who stay poor for fve years or more have a hgh probablty of remanng poor for the rest of ther lves. Data requred for ths analyss has been systematcally collected by the Rural Household Survey (RHS) team of the Natonal Bureau of Statstcs. By long tme collaboratve research work wth the RHS offce n Hube provnce, where the team and the host unversty locate, the team has obtaned such data set for Hube from 2000 to A fve-year 1900-household panel data set had been constructed based on the data coverng more than 3000 households, contanng over thousands varables. Ths provdes ample source for the purposes of ths research project. PDF created wth pdffactory tral verson

17 6. Dssemnaton strategy: How and where research results wll be dssemnated to academcs, polcy-makers and the publc: publcatons, polcy brefs, semnars, conferences, etc. (see PEP's consultaton and dssemnaton strategy for deas) It s expected that ths study produce research reports and dscusson papers at the frst stage, and polcy brefs and journal artcles at the later stage. As the team members, Drs Shjun Dng and Yupng Chen, have good record of publcatons n scentfc journals. Much of the lnks have been establshed between the team and outsde nsttutons and researchers. The fndngs and polcy mplcatons from the study wll be conveyed to polcy makers through nternal channels, by submttng research reports whch could be papers, lectures, compact dscs and polcy brefs. As a consultant, the team leader, Dr Shjun Dng, has been montorng governmental poverty reducton programs from the State Councl Leadng Group Offce of Poverty Allevaton and Development (LGPAD) and the Offce of Hube Provncal Poverty Allevaton and Development. The team has many occasons each year to dscuss poverty-related ssues wth the offcals from above mentoned government agences. Dr Shjun Dng s currently supervsng two PhD canddates, who are from the State Councl Leadng Group Offce of Western Regon Development (LGWRD) and are responsble for mplementng poverty reducton and human resources development programs n the regon, thus, the fndngs of the proposed research project can be channeled through LGWRD to varous relevant government departments. The research fndngs wll also be presented and dscussed n varous polcy dssemnaton workshops, whch nvolve key stakeholders from academcs, non-governmental organzatons and members of publc organzaton, such as the World Bank, the World Insttute for Development Economcs Research (WIDER), Center for PDF created wth pdffactory tral verson

18 Chnese Agrcultural polcy (CCAP) and the IDRC/CIGI Young Chna Scholars Poverty Research Network. Dr Shjun Dng had been consultants of World Bank s project (for example, Energy, Poverty and Gender: Chna Case Study n 2001), and has been ntensvely collaborated wth researchers from the Insttute of Development Studes, at the Unversty of Sussex, UK. He s also havng contacts wth the Chronc Poverty Research Center at the Unversty of Manchester, UK. As a resource person, Dr Guanghua Wan s a senor research fellow and the drector of vstng fellowshp and nternshp programs at WIDER. Together wth Hatao Wu, Dr Shjun Dng has close lnk wth the IDRC/CIGI Young Chna Scholars Poverty Research Network. Furthermore, the team members have closely contact wth PEP members n Chna (for example, Dr Fenglan Du and Dr Xuqng Wang) through meetngs, emals and telephone contacts. Through dalogue wth nternatonal and natonal agences and members, the research results wll be more nfluental and nformatve to ant-poverty polcy-makers, academes and the publc. 7. Short lst of key references 1) Barrentos, A., D. Hulme and A. Shepherd, 2004, Can Socal Protecton Tackle Chronc Poverty, ProtB arrentos.pdf 2) Chaudhur, S., 2003, Assessng Vulnerablty to Poverty: Concepts, Emprcal Methods and Illustratve Examples, keny_0304/ke_0304/vulnerablty-assessment.pdf 3) Chrstaensen, L. and K. Subbarao, 2004, Towards an Understandng of Household Vulnerablty n Rural Kenya, World Bank Polcy Research Workng Paper No. 3326, Washngton DC. PDF created wth pdffactory tral verson

19 4) Coudouel, A. and J. Hentschel (2000), Poverty Data and Measurement, Dscusson Paper, Washngton, DC: World Bank. 5) Duclos, J., A. Araar and S. Gles, 2006, Chronc and Transent Poverty: Measurement and Estmaton, wth Evdence from Chna, the Insttute for the Study of Labor (IZA) n Bonn, Dscusson Paper No ) Foster, J., J. Greer and E. Thorbecke (1984), A Class of Decomposable Poverty Measures, Econometrca, Vol. 52, pp ) Holzmann, R and S. Jorgensen, 2000, A New Conceptual Framework for Socal Protecton, and beyond, Socal Protecton Dscusson Paper No.0006, the World Bank. 8) Hulme, D. and A. Shepherd, 2003, Conceptualzng Chronc Poverty, World Development, Vol. 31, No. 3, pp ) Jalan, J. and M. Ravallon, 1998, Transent Poverty n Post-reform Rural Chna, Journal of Comparatve Economcs, Vol. 26, No. 2: pp ) Jalan, J. and M. Ravallon, 2000, Is Transent Poverty Dfferent? Evdence for Rural Chna, Journal of Development Studes, 36(6), ) Kamanou, G. and J. Morduch (2002) Measurng Vulnerablty to Poverty, WIDER Dscusson Paper No. 2002/58, Helsnk: UNU-WIDER. 12) McCulloch, N. and M. Calandrno, 2003, Vulnerablty and Chronc Poverty n Rural Schuan, World Development, Vol. 31, No. 3, pp ) Prtchett, L., A. Suryahad and S. Sumarto, 2000, Quantfyng Vulnerablty to Poverty: A proposed Measure, Appled to Indonesa, Polcy Research Workng Paper No. 2437, the World Bank, Washngton, DC. 14) Ravallon, M., 1988, Expected Poverty under Rsk-nduced Welfare Varablty, PDF created wth pdffactory tral verson

20 Economc Journal, 98(393): ) Ravallon, M. and J. Jalan, 1998, Determnants of Transent and Chronc Poverty: Evdence from Rural Chna, Polcy Research Workng Paper No.1936, the World Bank. 16) Shaukat, M., 1995, Poverty assessment: Pakstan s case, The Pakstan Development Revew, Vol.34, No.1, pp ) World Bank, 2000, World Development Report 2000/1, Attackng Poverty, New York, Oxford Unversty Press. 18) Yue, X., 2005, The Causes of Transent Poverty and Its Implcaton to Poverty Reducton Polcy n Rural Chna, 19) Zhang, Y. and G.,Wan, 2006, An Emprcal Analyss of Household Vulnerablty n Rural Chna, Journal of the Asa Pacfc Economy, Vol. 11, No. 2, pp Lst of team members pror tranng and experence n the ssues and technques nvolved. Team members: 1) Shjun DING, male, 43, PhD, professor n agrcultural economcs, team leader, nvolved n (and n charge of) several research projects on rural development ssues, as can be found n hs CV. He has been manly focusng on poverty reducton n rural Chna. He has been carryng out several nternatonally collaborated poverty reducton programs wth researchers from the World Bank, the Ford Foundaton, the Rockefeller Foundaton, the Insttute of Development Studes, and others. Hs areas of expertse nclude quanttatve analyss, partcpatory poverty assessment and ssues of poverty reducton regardng rural Chna. 2) Yupng CHEN, female, 42, PhD, assocated professor, gender analyss expert and agrcultural technologcal extenson practtoner, traned n the UNDP Bejng offce and PDF created wth pdffactory tral verson

21 once had bee employed as UNDP volunteer (CPR/V/332) served as poverty and gender analyss and agrcultural extenson expert. 3) Hatao WU, male, 27, PhD canddate, wth ntensve tranng on statstcs and econometrcs, nvolved n several agrcultural economc research projects on drought and rce farmers copng strateges, households energy use and poverty allevaton, and so on. 4) Zhe LI, female, 26, PhD canddate, wth experence n poverty analyss, gender analyss, rural household survey (quanttatve and qualtatve) and data management. 5) L FENG, female, 26, PhD canddate, wth experence n rural household survey and data analyss Resource person Guanghua WAN, male, PhD, senor research fellow and project drector n the World Insttute for Development Economcs Research, nvolved n (and n charge of) several research projects on poverty and nequalty. As a resource person, he s not part of the team for ths proposed research project and entrely voluntary wthout any fnancal mplcaton on the project. 9. Expected capacty buldng Descrpton of the research capactes that team members and ther nsttutons are expected to buld through ther partcpaton n ths project. Ths s an mportant aspect n the evaluaton of proposals and should be presented n some detal. What technques, lterature, theores, tools, etc. wll the team and ther nsttutons learn or deepen ther knowledge of? Please also ndcate what tasks each team member would carry out n executng the project. Zhongnan Unversty of Economcs and Law, to whch the research team belongs, s among the top unverstes specfed n economcs, management and law n Chna. The team members have been nvolved n several research projects regardng poverty allevaton n rural Chna. Professor Shjun Dng, the team leader, has conducted research projects wth fnancal support from the Rockefeller Foundaton, the Ford Foundaton, and PDF created wth pdffactory tral verson

22 the Asan Scholarshp Foundaton on rural development and poverty allevaton, as well as provded consultances to the World Bank and the Shell Foundaton on energy and rural poverty studes, as can be seen n hs CV. Other team members have been nvolved n most of Professor Shjun Dng s research projects. Undertakng ths study wll certanly enrch ther understandng of poverty n rural Chna, and mprove ther ablty n dynamc analyss of poverty. A number of young PhD students (two of three n the team are females) wll be drawn nto ths project and beneft from an exposure to dynamc poverty analyss, and ths research wll help to bult a team consstng of young PhD students who wll be able to undertake such studes for other regons n the country, and to strengthen the capacty of young researchers n Chna n analyzng poverty dynamcs. Most of the lterature mentoned above s collected from the team leader s several scentfc vsts to the Insttute of Development Studes (IDS), at the Unversty of Sussex, UK., and from the World Insttute for Development Economcs Research (WIDER). Dr Guanghua Wan, who s a senor research fellow at WIDER, wll act as resource person for ths proposed project. Major technques to learn from the project nclude practces of poverty measures n monetary term and n dynamc term, measurement of vulnerablty n rural Chna, use of large panel dataset of rural household survey, as well as, for the young PhD students, the statstc software such as STATA and SPSS. Shjun Dng wll act as team leader, be responsble for desgn and mplementaton of the study. Together wth Hatao Wu, Shjun Dng wll be responsble for mathematc and econometrc specfcaton of varous knds of models employed n the study. Shjun Dng and Yupng Chen wll be responsble for data analyss and wrtng up of research reports PDF created wth pdffactory tral verson

23 and papers for journal publcaton. Zhe L and L Feng wll be manly nvolvng n data management and statstc descrpton of data, whle also engagng ntensvely n learnng n methodologes of poverty and vulnerablty analyss. The team and the nsttuton are expected to ganed knowledge and technques n vulnerablty measurement by nvolvng n ths study. 10. Any ethcal, socal, gender or envronmental ssues or rsks whch should be noted The project wll follow natonal ethcal standards for research. In partcular, all survey and ntervews wll be undertaken wth the consent of the ntervewee and the nformaton wll be kept strctly confdental. Informaton on natonal ethcal revew wll be carred out. 11. Lst of past, current or pendng projects n related areas nvolvng team members (name of fundng nsttuton, ttle of project, lst of team members nvolved) 1) The Rockefeller Foundaton, Improvng Food Securty of Upland Communtes n southern Yunnan, , Shjun Dng, project manager, Yupng Chen and Hatao Wu were nvolved 2) The World Bank Bejng Offce, Baselne Assessment of Current Status of Household Bogas Systems n Western Provnces of Chna, , Shjun Dng, project consultant, Hatao Wu, L Feng and Zhe L were nvolved. 3) The Natonal Natural Scence Foundaton of Chna, Drought Rsk and Farmers Copng Mechansms n Rce Producton Systems of Southern Chna, , Shjun Dng, project manager, Yupng Chen and Hatao Wu were nvolved. PDF created wth pdffactory tral verson

24 4) The Asan Scholarshp Foundaton, A Comparatve study of the Rural Elderly Support n Chna and Thaland, , Shjun Dng, project manager. 5) The Rockefeller Foundaton, Drought Rsk and Farmers Copng Mechansms n Rce Producton Systems of Southern Chna, , Shjun Dng, project manager, Yupng Chen and Hatao Wu were nvolved. 6) The Ford Foundaton n Bejng, Support for the Elderly n Rural Chna durng Economc Transton, , Shjun Dng, project manager. 7) The World Bank, Energy, Poverty and Gender: Chna Case Study, , Shjun Dng, project consultant, Yupng Chen was nvolved. 8) The Department for Internatonal Development, UK, Socal Securty n Rural Chna, , Shjun Dng, project consultant, Yupng Chen was nvolved. PDF created wth pdffactory tral verson