Asian Economic and Financial Review ESTIMATE OF POVERTY LINE AND ANALYZE OF POVERTY INDICES IN IRAN ( ) Morteza Afghah.

Transcription

1 Asa Ecoomc ad Facal Revew, 2014, 4(11): Asa Ecoomc ad Facal Revew joural homepage: ESTIMATE OF POVERTY LINE AND ANALYZE OF POVERTY INDICES IN IRAN ( ) Morteza Afghah Faculty of Ecoomcs, Shahd Chamra Uversty of Ahvaz, Ira Am Masour Faculty of Ecoomcs, Shahd Chamra Uversty of Ahvaz, Ira ABSTRACT The purpose of ths paper s to aswer ths uesto that f poverty status regardg varous dces follow smlar treds durg the war betwee Ira ad Ira ad developmetal plas for urba ad rural areas Ira perod. Usg the Grager Causalty test ad correlato aalyss, the ter-relato of dces ad correlatos amog areas s examed. Results dcate that whle the mothly poverty le urba areas from 1982 to 2007 creased 80 tmes, mothly poverty le rural areas creased from 1982 to 2007 creased 77 tmes. Grager causalty test shows that rural poverty type dd ot affect other types of poverty; but some kds of poverty dces urba areas are very effectve o each other. Fally, t reveals that the poltcs mpacts o dces are dfferet varous regos AESS Publcatos. All Rghts Reserved. Keywords: Poverty le, Lear expedture system, Poverty dces, Correlato aalyss, Grager causalty, Ira. JEL Classfcato: I32, C87, C22. Cotrbuto/ Orgalty Ths study s oe of very few studes whch have computed ad compared all avalable dces related to poverty Ira. Furthermore, ths study cotrbutes to the exstg lterature wth examg the correlatos betwee all estmated poverty dces. The Grager causalty of smlar dces betwee rural ad urba areas s also computed. 1. INTRODUCTION Poverty s oe of the oldest forms of socal pathology all areas. Sce last decades of the tweteth cetury, eradcatg poverty ad supportg deprved people were at the core of UN plas. I Ira, although the efforts to protect ad provde socal securty of poor ad vulerable 1502

2 Asa Ecoomc ad Facal Revew, 2014, 4(11): people have at least 40 years hstory, the achevemets of developmet plas Ira terms of reducg poverty ad come vulerablty s ot remarkable. To measure poverty status ad detfy the umber of poor ad o-poor people, the poverty threshold or boudary as "Poverty Le" s ecessary. However, the ma ssue s the fact that how to recogze the poor people. Here, the crtero s called Relatve Poverty (RP) Le that s defed as the amout of reveue (or cost) whch s reured to provde the least subsstece wth regard to cultural, socal ad ecoomcal backgroud of the commuty. It seems that, a estmatg method usg moetary value or cost of cosumer goods ad servces (due to the capablty of gatherg t) through a system of Lear Expedture System (LES) s the best way to calculate the relatve poverty le. Ths method, based o the theory of cosumer behavor ad demad aalyss for major groups of goods, estmates the poverty le usg exstg data. LES s a demad euato system that ca provde demad theory restrctos through a defed utlty fucto. Although there are varous deftos for poverty, but (Sto, 1954), expermetally put LES as base for studyg demad euatos system, usg utlty fucto of the Kle ad Rub for the frst tme. Afterward, gorg assumpto of stablty of mmum cosumpto over tme, Pollak ad Walles (1969), preseted varous models of spedg system patters wth cosumer habts, whch the mmum cosumg expedtures (least subsstece) presets as a radom varable ths model. I addto, Luch (1973) provded publc lear expedture system ad preseted Exteded Lear Expedture System (ELES) usg CES, whch utlty fucto of the Kle ad Rub s a specfc type of ths system. Berges ad Casellas (2002) coducted a research o a demad system aalyss for food Argeta s poor ad o-poor households. A complete system of demad euatos, the Lear Expedture System (LES), has bee used due to ts relatve emprcal advsablty. It cludes the moey value, the uattes ad types of food purchased by the households over a perod of oeweek (March 96-Aprl 97). The results showed that the estmated parameter chage at dfferet levels of come. Lower-come famles choose to cosume relatvely more meat, chcke ad bread. The margal budget shares dffer sgfcatly betwee both groups. Poor households exped more of ther superumerary come o meat, bread ad vegetables tha o-poor households. Skoufas (2003) vestgate the prce ad come effect o food ad calore demad. Usg SUSENAS data of 1996 ad 1999, ma purpose of ths study s examed the chage of cosumer behavor of cosumpto after the ecoomc crss. Noparametrc methods was appled to observe the dffereces elastcty estmates of poor ad o-poor households. The emprcal fdgs revealed that the come elastcty of calore demad s mldly hgher 1999 (post ecoomc crss perod) compared to 1996 (pre-crss perod). Ths result dcates the calorecome elastcty s ot sestve to prce chages eve whe the prce s very volatle the crss tme. The households smooth ther cosumpto the tme of crss, as show ths study, through the crease of cereals calore-come elastcty whle the calore-come elastcty for other food decreased. Wdodo (2006) estmated household demad fucto ad welfare measuremet usg the Lear Expedture System (LES) the case of Japa ad Idoesa. These 1503

3 Asa Ecoomc ad Facal Revew, 2014, 4(11): studes obta some coclusos. Frst, for food cosumptos, Idoesa households have maxmum margal budget shares o meat ad the mmum margal budget shares o fruts. Meawhle, Japaese households have ther maxmum margal budget shares o fsh ad shellfsh, ad ther maxmum budget shares o dary products ad eggs. Secod, Idoesa households have a smaller gap betwee mmum food cosumpto (subsstece level) ad average food cosumpto tha Japaese households have. Thrd, wth the same level of prces crease o foods, the smulato shows that omal exchage (Ye) Japaese households get greater welfare decrease tha Idoesa households. However, the percetage of total expedture, Idoesa households get greater welfare decrease tha Japaese households. Pagarbowo ad Tsega (2011) aalyzed the demad resposes of Idoesa households to food prces, come chages ad other socoecoomc factors. They use the Idoesa Famly Lfe Survey data ad employed a exteded form of the Quadratc Almost Ideal Demad System model whch cludes demographc ad regoal factors. The results revealed that the well-kow patter that food demad behavor vares sgfcatly betwee urba ad rural households smlar to come groups. The poorest households cosume relatvely more staple food as well as alcohol ad tobacco goods, whle the rchest households cosume relatvely more meat, sack ad dred food. It s show the poorest households expedture elastcty of alcohol ad tobacco s hgh, mplyg that the poorest households trasfer ther extra resources o alcohol ad tobacco stead of other utrtous food tems. Results also showed that prce ad expedture elastcty have dfferet value across tme ( ). Prce elastcty creased for most food tems mplyg that people become much resposve to the chages food prces. Otherwse, the expedture elastcty s low for most food tems (except for alcohol ad tobacco goods ). Thus, we are tryg to estmate the poverty le usg the Seemgly Urelated Regresso (SUR) ad Lear Expedture System for the perod of ths paper. The we wll cocetrate o the results dffereces urba ad rural areas of Ira, calculated by varous poverty estmato tools, followed by revewg the relatoshps betwee dces usg Grager Causalty Aalyss. Employg correlato aalyss, the effects ad the terdepedece of the poverty dces urba ad rural areas are also examed Lear Expedture System (LES) The Kle-Rub utlty fucto (1947) s show as follows: (1) Ut ( t t ) t 0 1 Where, t s the amout of good cosumpto the perod of t, t s the mmum cosumpto expedture of each commodty group. If ths fucto cosdered as a lear combato ad by applyg logarthm (Sto-Geary utlty fucto), the we wll have: (2) U log( ) Therefore, ( / ), thus 1 t 1 t t

4 Asa Ecoomc ad Facal Revew, 2014, 4(11): Now, f the mmum cosumpto expedture o each commodty group ay perod s cosdered as a radom varable, the cosumers' utlty fucto wll be the maxmum, regardg to budget costrats, so the euato wll be as: (3) L log[ t ] ( I t pt ) 1 Where s cosumer come, f we maxmze utlty fucto, thus the demad fucto s: 1 (4) Et ptt pt t ( I t pt ) Wth ths codto, 1 the cosumed spedg euatos are lear wth cost ad 1 come varables ad are olear wth parameters. However, based o defto, the relatve poverty le s the expedtures oe speds o subsstece level of each commodty group. (Sto, 1954): (5) Z t t pt. 2. POVERTY INDICES 2.1. Head-Cout Rato Idex To calculate ths dex, the umber of the poor dvdes to total populato. (6) H 0 H 1 Where, s populato of poor people ad s the total populato of coutry. The amout of ths dex vares from zero (o poor people the socety) to oe (all socety people the socety are poor). The most mportat problem wth Head-Cout Rato dex s that t s sestve to trasferrg come amog the poor ad eve betwee the poor ad others. Further, t s ot sesble to the reducto of the poor come Icome-Gap Rato Idex (Poverty Gap) Ths dex s the average come gap of the poor to the poverty le: 1 (7) I z y / Z z y p 1 / z 1 y z p Where dcate Icome-Gap Rato, y 1 p y 1 s the average come of the poor, z shows the poverty le, ad z-y reflects the poverty gap. Ulke Icome-Gap Rato, ths dex s sestve to trasfer of come from oe to aother. It must be oted that the poverty gap would be zero for the poor (Se Amartya, 1973). 1505

5 Asa Ecoomc ad Facal Revew, 2014, 4(11): Se Poverty Idex The Se Poverty dex, whch s based o ordal welfare of people, dcates that f G coeffcet of poor people come y 1, y2,..., y s show as: (8) G 1 1 y 2y 3y... y I addto, f s hgh eough, the: y p (9) P HI 1 I Where H shows Head-Cout Rato dex, I s Icome-Gap Rato Idex ad G s G coeffcet of poor people. The amout of ths dex vares betwee zero (dcate o oe s poor) ad oe (all or the people are poor) (Se Amartya, 1973). G 2.4. Poverty Idex of Foster, Greer ad Thorbeack (FGT) Foster, Greer ad Thorbeack beleved that a good poverty dex should be expermetal, so they troduced a ew poverty dex. Ths dex s based o the assumpto that the poverty rate could be a aggregato of varous populato subgroups poverty rates. Thus, oe ca compute a ut rate of poverty of whole populato. Therefore, they troduced the dex as follows: (10) 1 FGT 2 z 2 g 1 Ths dex s based o dfferece betwee poor dvdual (household) come ad the poverty le. Further, those who have the most dstace to poverty le have the greatest weght ths dex. Therefore ths dex cocetrates o the poorest people codtos. I other words, ths dex calculates the poverty gap ad thus, the lower rate reflects a lower poverty gap (Foster et al., 1984) Se, Shorrocks ad Tho (SST) Idex Shorrocks used SST dex The SST dex s show as: 1 g (11) SST (2 2 1) 2 z Where, s the populato, g s the poverty gap ad z reflects poverty le. Shorrocks the developed the dex as follows: 1 (12) SST HI 1 G ) ( s I ths dex, H: Head-Cout Rato dex, I: the Icome Gap Idex, G s 1 g 1 G coeffcet of gap betwee the poor. Note that the z (2 2 1) 2 HI 1 dex value the worst codto s two ad the best value s zero (Xu ad Osberg, 2001). 1506

6 Asa Ecoomc ad Facal Revew, 2014, 4(11): METHODOLOGY Data used ths research are: - Expedture ad come of urba ad rural households the perod of are crude ad curret Iraa Rals (IRR); - Cosumer Prce Idex s used for eght groups of goods ad servces durg as costat prces 1997; - The umber of sampled households s based o come groups for the perod of ; - The average come IRR for the perod of ; (Statstcal Ceter of Ira, ). However, ths paper, eght goods ad servces tegrated to four groups as: Food=Food ad Tobacco, Socal affars=clothg, Health, Educato ad Etertamet, Housg= Housg ad Servces, Other=other goods ad servces ad trasportato Further, as others, socal affars ad housg groups clude several other groups, we have used weghted average of dces. For ths reaso, the share of each group etre group s computed at frst the the weghted average of dex s calculated;.e. to calculate other dex, have used the followg formula: (13) E other E 2 2 E w pother 1 Eother 1 w p Where, dcate other group ad trasportato sector mscellaeous group, E, P ad W are expedtures dexes ad the share of spedg a group respectvely. 4. ESTIMATING SUBSISTENCE AMOUNT USING LES - A) Estmatg : For calculatg the share of commodty groups from come, the Egel coeffcet,.e. f (I) E, s used the system. I ths approach, 1 euato s estmated ad the share of the last group s approxmated through Egel coeffcet euato could be ether lear or logarthmc. However, ths research the lear model used to estmate. Thus, the Egel coeffcet for three groups of Food, Socal affars ad others was estmated ad the Egel coeffcet for housg s computable by deductg the sum of other groups from oe. The estmated results usg E I are show table 1. The fgures are estmated through Ordary Least Suares ad we have used AR varable to remove ay probable autocorrelato our models. 1507

7 Asa Ecoomc ad Facal Revew, 2014, 4(11): Table-1. Results of commodty group shares urba ad rural areas Commodty Group Food Socal affares Housg Other Urba (t-rato) 0.21 (5.88) 0.18 (111.5) 0.36 (51.7) 0.25 (18.7) Rural (t-rato) 0.31 (2.75) 0.2 (53.9) 0.26 (26.5) 0.23 (18.18) B) Estmatg : To estmate, havg groups share of come, we have employed Stoe-Gary utlty fucto based o lear expedtures systems. To estmate euatos for commodty groups for the perod of , we have employed Seemgly Urelated Regressos. I the case of autocorrelato, AR varable s used. Table-2. Subsstece amout of commodty groups urba ad rural areas (IRR) commodty group Food Socal affares Housg Other Urba Rural As show table 2, the food group has a very hgh share subsstece of urba ad rural areas. 5. POVERTY LINE ESTIMATION Usg estmated subsstece of commodty groups, we have calculated the commodty poverty le ad total poverty le for both groups, based o p t t ad Z t t p 1 t euatos respectvely. Calculatg poverty le based o commodty groups has the advatage that t shows the mmum come ecessary for a famly to provde a group of commodty. Furthermore, ths approach eables us to coclude absolute poverty from poverty le of food group. I other words, the mmum come to provde food group s a dex for poverty le. To summarze the results, we have show oly the commodty poverty le of food groups table 3. Mothly poverty le s obtaed va dvdg to 12 aual poverty les. Table-3. Commodty poverty le of food group ad aual ad mothly total poverty le for urba ad rural areas as curret IRR Urba Rural Food group Poverty le Mothly poverty le Food group Poverty le Mothly poverty Year (mothly) (mothly) le Cotue 1508

8 Asa Ecoomc ad Facal Revew, 2014, 4(11): As see the above table, all poverty le dces durg the study perod have creasg tred. Ths may be due to flato rate ad creasg tred of prce dex durg the study perod. As results show, the mothly come of a urba famly should have bee creased from 53,000 Rls 1982 to 4,250,000 Rls 2007 (79 tmes greater) f they wat to keep the same urba lvg stadard. Further, f a urba household just wats to survve, ts come should have bee creased from 16,000 Rls 1982 to 1,170,000 Rls However, for a rural famly, mothly come 2007 was 76 tmes greater tha 1982;.e. from 32,000 Rls 1982 to 2,490,000 Rls 2007, to keep the same rural lvg stadard. Further, a household mmum come ecessary to survve should have bee creased from 16,000 Rls 1982 to 1,140,000 Rls 2007, the amouts smlar to those urba areas. 6. CALCULATING POVERTY INDICES I ths secto, descrbed dces such as Head-Cout poverty (H), Icome Gap (I), Se Idex (P), FGT dex ad SST dex are calculated. To compute Se Idex, however, we eed to estmate the G coeffcet of the poor (G) ad to calculate SST dex, we eed to estmate the G coeffcet of poverty gap (G s ). Table-4. Poverty dces urba areas (percet) Year (SST) (Gs) (FGT) (P) (G) (I) (H) eutoc 1509

9 Asa Ecoomc ad Facal Revew, 2014, 4(11): Table-5. Poverty dces rural areas (Percet) Year (SST) (Gs) (FGT) (P) (G) (I) (H) REVIEW AND ANALYSIS OF POVERTY INDICES DURING THE FIVE YEAR DEVELOPMENT PLANS 7.1. Head-Cout Idex Results show that Head-Cout Idex urba ad rural areas has a costat crease durg Ira-Ira war perod ( ). Durg frst year of frst developmet pla, the percetage of 1510

10 Asa Ecoomc ad Facal Revew, 2014, 4(11): poor creased both urba ad rural areas but ecoomc growth 1990 led to a postve mpact o poverty reducto. I early years of Secod Developmet Pla, whle Ira was experecg the hghest flatos rate, amout of poor people urba ad rural areas were at the hghest level, though t was hgher urba areas. Ths may be due the fact that flato has had more effect o urba areas. However, the coutry expereced the lowest level of poor people durg the Thrd Developmet Pla, wth the begg of the Fourth Developmet Pla; the poverty status expereced a reverse tred the amout of poor people both urba ad rural areas, whle may govermet polces cocetrated o poverty reducto ad mprovemet come dstrbuto Icome Gap Idex Results tables 4 ad 5 show that, whle come gap has decled urba areas durg the war t was almost costat rural areas. Ths meas that poor people had expereced a better codto durg the war perod. Durg the Frst, Secod ad Thrd Developmet Pla, ths dex was almost fxed for both urba ad rural areas, though t was lower urba areas compared wth rural areas. Ths tred, however, reversed the fal years of the Thrd Developmet Pla ad durg the Fourth;.e. the come gap bega to crease both urba ad rural areas, though lower rural areas Se Idex As metoed, Se Idex s smlar to both prevous dces, but t volves the G Coeffcet of poor people the formula. Revewg the results, Se Idex show that poverty was creasg urba areas but decreasg rural areas durg the war perod. I geeral, the results from Se Idex are smlar to Head-Cout Idex durg the four developmet plas. However, 1992 ad 1999 the poverty dex show reductos urba areas but crease rural areas FGT Idex Results of ths dex show deterorato poverty stuato both rural ad urba areas durg the war perod. The poverty dex was at the hghest level the fal years of the war. However, the codto of the poorest people rural areas were better tha the codto of the poorest people urba areas. Whle experecg the hghest ecoomc growth at the begg of Frst Developmet Pla 1990, the poorest codto mproved dramatcally. I 1994, the dex was at the maxmum level of the Secod Developmet Pla. Ths dex, however, show a stable codto durg the Thrd ad Forth Developmet Pla SST Idex Results of ths dex represet poverty testy smlar to those of Se results durg the war perod. I other words, the poverty testy has had a creasg slope urba areas but decreasg rural areas. Ths dex, however, expereced results relatvely smlar to Se dex durg the years after the ed of war wth two hghest levels 1990 ad

11 Asa Ecoomc ad Facal Revew, 2014, 4(11): ESTIMATING GRANGER CAUSALITY BETWEEN POVERTY INDICES I ths secto, we have examed the probable causal relatoshp amog all estmated poverty dces ths research. For ths reaso, the Grager Causalty Test was employed usg Vector Auto Regressos (VAR) te two-sded euatos (twety euatos) for each rego. No causalty were show betwee dces of poverty rural areas. Results show that the poverty dces rural areas are depedet. However, there were causalty relatoshps amog some of the poverty dces urba areas. we have preseted oly the dces that showed causal relatoshp table 6 bref. Table-6. Grager o-causalty of poverty dces (urba) Null hypothess F-Statstc Prob Null hypothess F-Statstc Prob I: grager ocausalty of H causalty of I H: grager o SST: grager ocausalty of P causalty of H SST: grager o FGT: grager ocausalty of H causalty of I SST: grager o FGT: grager ocausalty of P causalty of SST FGT: grager o The results show that all ull hypotheses were rejected, so the causalty relatoshp exsts. Ths meas that whe the come gap creases the percetage of poor people has also creased, ad vce versa. I addto, the testy of poverty s affected by come gap, Head-Cout ad Se Idex. I fact, wth a crease percetage of poor people ad come gap or a decrease relatve welfare, the testy of poverty wll crease; cosderg that the causalty of other sde has ot bee cofrmed. The depth of poverty s also affected by three dces: Head-Cout, Se ad SST of poverty; that meas a crease the percetage of poor people or relatve welfare or a decrease poverty testy, depth of poverty wll deterorated. However, the causalty of other sde has ot bee cofrmed. 9. ANALYSIS OF THE CORRELATION BETWEEN INDICES OF POVERTY I geeral, may macro-level polces socety affect come dstrbuto ad, thus, poverty codto. Sce ths paper, the poverty codto both urba ad rural regos are studed separately, the preset secto t s tred to uderstad how poverty codto urba ad rural areas s affected by polces macro level. The correlato aalyss help us to exame the correlato betwee two varables regardless of beg exogeous or edogeous. Thus, the correlatos betwee poverty dces urba ad rural areas were computed. Results are show table

12 Asa Ecoomc ad Facal Revew, 2014, 4(11): Table-7. Correlato coeffcets of dces of poverty urba ad rural areas Varable SST FGT P I H r Results show that ay polcy to reduce poverty testy (SST Idex) wll have the same effect o urba ad rural areas by 86 percet. Furthermore, polces to mprove poverty gap socety has by 43 percet smlar effect o urba ad rural areas. However, polces affectg FGT, I ad H socety have smlar effect o urba ad rural areas by 56%, 79% ad 75% respectvely. 10. CONCLUSION I ths paper, usg budget data for rural ad urba households for the perod of , total poverty le ad commodty poverty le Ira was estmated. To estmate poverty le, based o socety subsstece computed by the Lear Expedture System usg utlty fucto of Stoe-Gray, we have employed Seemgly Urelated Regressos. The poverty le data, the, used to compute some related poverty dces such as Head-Cout, come-gap, Se, FGT ad SST, resulted data are aalyzed to compare poverty codto Ira-Ira war ( ), as well as frst three developmet plas. Fally, we have examed ad aalyzed the Grager causalty relatoshp betwee poverty dces ad correlato aalyss of poverty dces urba ad rural areas. The results showed that the mothly come of a urba household must have bee creased from 53,000 Rls 1982 to 4,250,000 Rls 2007 (creasg 79 tmes) to keep ts codto above poverty le. Further, for a household to survve, the come must have bee creased from 16,000 Rls 1982 to 1,170,000 Rls Smlarly, the rural households' mothly come should have bee creased from 32,000 Rls 1982 to 2,490,000 Rls 2007 (79 tmes crease) to keep ts codto above poverty le. Furthermore, for a rural household to survve, ts mothly come should have bee creased from 16,000 Rls 1982 to 1,140,000 Rls Results also show that the subsstece come s almost eual both rural ad urba areas. The calculated poverty dces show that the poor percetage the socety s creased ad relatve welfare of the poor ad the poverty testy deterorate durg war. All poverty dces show relatve mprovemet durg the frst three developmet plas. However, from the begg of the Fourth Developmet Pla, the codto chaged ad the dces show deterorato. The poor come gap has almost smlar codto durg the war perod ad frst ad secod plas, but durg the thrd ad the fourth plas, a reducto was expereced. The depth of poverty was deterorated durg war perod but after a relatve mprovemet, over the frst ad secod plas, t showed a sgfcat mprovemet durg thrd ad fourth developmet plas. Causalty test results showed o sgfcat relatoshp for rural areas but for urba areas. It s cocluded that a crease the percetage of urba poor wll cause a crease poverty gap, ad vce versa. Further, the poverty testy s affected by the poor percetage, come gap ad relatve welfare reducto Lkewse, the depth of poverty s flueced by the poor percetage, reducto tesve poverty ad relatve welfare. 1513

13 Asa Ecoomc ad Facal Revew, 2014, 4(11): Fally, the correlato aalyss suggests that polces affectg poverty dces such as Head- Cout dex, come gap, relatve prosperty, ad depth of poverty have smlar effect o urba ad rural areas by 75%, 43%, 79%, 56% ad 86% respectvely. As results of ths research, followg polces ca be suggested to polcy makers: 1- It s clear that the polces should lead to reducto of flato rate as the ma factor affectg the crease relatve poverty. 2- As was metoed above, each kd of poverty rural areas s ot affected by other kds of poverty; o the cotrary, urba areas all kds of poverty have close relatoshp to each other. Thus, t s recommeded that depedg poverty target polcy makers should be aware of dfferet results of poverty reducto polces urba ad rural areas. REFERENCES Berges, M.E. ad K.S. Casellas, A demad system aalyss of food for poor ad o poor Households: The case of Argeta, the Xth EAAE cogress: Explorg Dversty the Europea Agr-Food System, Zaragoza (Spa): Foster, J., G. Joel ad E. Thorbecke, A class of decomposable poverty measures. Ecoometrca, 281: Luch, C., The exteded lear expadture system. Europea Ecoomc Revew, 4(1): Pagarbowo, E.H. ad D.W. Tsega, Food demad aalyss of doesa households wth partcular atteto to the poorest. ZEF- Dscusso Papers o Developmet Polcy, (151). Avalable from [Accessed August 1, 2011]. Pollak, R.A. ad T.J. Walles, Estmato of lear expedture system. Ecoometrca, 37(4): Se Amartya, K., Levels of poverty: Polcyad chage, world Bak staff workg paper. U.S.A: 7. Skoufas, E., Is the calore-come elastcty sestve to prce chages? Evdece from doesa. World Developmet, 31(7): Statstcal Ceter of Ira, The statstcal yearbook. Sto, J.R.N., Lear expedture system ad demad aalyss: A applcato to the patter of Brtsh demad. Ecoomc Joural, 64(255): Wdodo, T., Demad estmato ad Household's welfare measuremet: Case studes o Japa ad Idoesa. Avalable from [Accessed December 1, 2011]. Xu, K. ad L. Osberg, How to decompose the se- shorrocks-poverty dex: A practtoer's gude. Joural of Icome Dstrbuto, 10(1-2). BIBLIOGRAPHY Rowetree, B.S., Poverty: A study of tow lfe. Lodo: Macmla. Seddegh, H.R., K.A. Lawler ad A.V. Katos, Ecoometrcs: A practcal approach. U.K: Suderlad UveRlsty. 1514