CIRPÉE Centre interuniversitaire sur le risque, les politiques économiques et l emploi - PDF

CIRPÉE Cntr intrunivrsitair sur l risqu, ls politiqus économiqus t l mploi Cair d rcrc/working Papr On t Impact of Bttr Targtd Transfrs on Povrty in Tunisia Sami Bibi 2 Octobr/Octobr 2002 I am gratful

Please download to get full document.

View again

of 26
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.


Publish on:

Views: 21 | Pages: 26

Extension: PDF | Download: 0

CIRPÉE Cntr intrunivrsitair sur l risqu, ls politiqus économiqus t l mploi Cair d rcrc/working Papr On t Impact of Bttr Targtd Transfrs on Povrty in Tunisia Sami Bibi 2 Octobr/Octobr 2002 I am gratful to t ERF, CIRPEE, URECA, and FSEGT for tir financial support. An arly vrsion of tis papr was prsntd at t worksop on t analysis of povrty and its dtrminants in t MENA rgion, organizd by t Economic Rsarc Forum (ERF) and t World Bank in Sana a July 3-August, 200. I addrss a spcial tank to Jan-Yvs DUCLOS for is vry usful commnts. I also tank Moamd GOAIED, Ragui ASSAAD, Cristop MULLER, Brnard FORTIN, and Véroniqu NICHELE for tir commnts. Of cours, all rmaining rrors ar min. 2 Faculté ds Scincs Economiqus t d Gstion d Tunis (FSEGT) t Unité d Rcrc n Econométri Appliqué (URECA). Addrss: FSEGT, Campus Univrsitair, Bd. 7 Nov. El Manar 2092 Tunis. Républiqu Tunisinn. Fax : (26) Abstract : Tis papr dscribs t ffcts of gnral food subsidis on povrty in Tunisia, as rvald by ousold survy data for 990. T analysis indicats tat t poorst crtainly tak advantag of tis systm, but at t pric of considrabl lakags to non-poor popl and at a sizabl conomic fficincy loss rsulting from rlativ pric distortions. Furtr, non-paramtric stimations suggst tat tr ar no commoditis prdominantly consumd by t poor. Tis implis tat targting by commoditis is not an ffctiv way to figt against povrty and so, it is unlikly tat rstructuring t currnt scm would improv significantly t living standards of t lss wll-off mmbrs of socity. W tn invstigat t impact on povrty of a mor targtd transfr scm, basd on proxy mans-tsts, using an appropriat conomtric tcniqu to modl it. Simulations sow tat tis dsign would b mor ffctiv in rducing povrty tan t us of gnral food subsidis. Finally, dominanc tsts sow tat tis dsign would first-ordr-dominat food subsidis scm witin a rang of povrty lins including all tos stimatd and gnrally usd for Tunisia. Ky Words: Povrty; Targting; Subsidis; Transfrs. JEL classification: D2; D63; H53; I32; I38. 2. Introduction T living standards of t lss wll-off mmbrs of a socity affct public policy and ar also of tical intrst. Bcaus govrnmnts av limitd rsourcs, and it is important to us tm fficintly, t qustion of ow to dsign policis to lssn povrty is likly on of t most important concrns of policymakrs. Among availabl tools to nanc t lot of t poor, targting by commoditis, tat is, by subsidizing food stapls mainly consumd by t poorst, as bn vry popular in dvloping countris. Yt targting by commoditis fails to promot fficincy, by crating conomic distortions. Furtr, t lakag of food subsidis to t non-poor popl is gnrally considrabl wil failur to improv substantially t wll-bing of t poorst is wll known. T objctiv of tis work is tn twofold. First, w aim to analyz t ffcts of on of t most important tools for allviating povrty in Tunisia, tat is t food subsidis program. Tis analysis rquirs tat w av conomtric stimats of t rlvant dmand systm. Givn tis information, it is possibl to assss t impact on povrty of tis scm; using a wid rang of povrty lins and povrty masurs for robustnss analysis. It will b also instructiv to tst wtr rforming tis systm to incras t poorst sar of food subsidis bnfits is fasibl. If t prsnc of som commoditis tat ar largly consumd by t poorst is provd, incrasing food subsidis to ts commoditis sould b an ffctiv way to lssn povrty. Otrwis, dsigning an altrnativ mans to aciv tis goal bcoms appaling. Unfortunatly, stimation rsults suggst tat tr ar no commoditis prdominantly consumd by t poor. Hnc, t scond part of tis work illustrats ow proxy mans tsts, using an appropriat conomtric tcniqu to modl tm, could b usd to rac a bttr outcom on povrty, rvnu-nutral. Housold caractristics ar usd as xplanatory variabls to comput t incom transfr to b awardd to ac ousold; as dducd from a modl dsignd to minimiz t svrity of povrty givn an anti-povrty budgt. T approac dvlopd in tis papr as bn applid using a Tunisian data st. Expnding t sam anti-povrty budgt currntly dvotd to food subsidis, simulations rval larg potntialitis in allviating povrty, if targting by commoditis is rplacd by t transfr scm rsulting from our proxy mans-tsts mtodology. For instanc, using robustnss analysis to avoid critical coics of povrty lins and/or povrty masurs, t rsults sow tat Wit rvnu-nutral rforms, t problm of t optimal siz of t govrnmnt is ignord. An altrnativ assumption rquirs information on t ousolds willingnss to pay for public goods. S King (983). 3 t simulatd dsign would dominat targting by commoditis scm witin a wid rang of povrty lins, wic includs all tos stimatd and gnrally usd for Tunisia. Tis papr is structurd as follows. Sction 2 provids a broad caractrization of t food subsidis ffcts on povrty, as wll as an valuation of t lakags and dadwigt loss rsulting from tis scm. Sction 3 illustrats ow targting by indicators, wn it is drivd from an adquat framwork, can b usd to aciv a bttr outcom on povrty. Sction 4 offrs som concluding obsrvations. 2. Targting by commoditis: ffcts on t wlfar of t poor population Sarcing for a povrty-allviating rform prsupposs agrmnt on a masur of t standard of living. Tis is an important and controvrsial subjct and would rquir a sparat study on its own. In lin wit most studis, w us t total xpnditur pr capita as a basis for masuring t (prmannt) incom. 2 W assum tat bfor implmnting t food subsidis scm, ac ousold as an xognous incom y and facs t pric systm p o. Aftr implmnting t targting by commoditis program, by xpanding t compnsation fund B availabl to financ tis policy, ac ousold as t sam nominal incom, y, but facs a nw pric systm p p. W aim to compar t lvls of a ousold's wlfar wn it facs diffrnt pric systms. To aciv tis goal, w coos a bncmark pric systm, dnotd by p r, and w dfin as in King (983) t concpt of quivalnt incom: for a givn budgt constraint (p, y), quivalnt incom is dfind as tat incom lvl wic allows, at t bncmark pric systm, t sam utility lvl as can b racd undr t givn budgt constraint. Formally, w av: r v( p, y ) = v( p, y), () wr v(.) is t indirct utility function, p is a vctor of pric systm, and y is a vctor of a ousold s incom pr capita. Notic tat sinc p r is fixd across all ousolds, y is an xact montary mtric of actual utility v(p, y) bcaus y is an incrasing monotonic transformation of v(.). Tus, invrting t indirct utility function, w obtain t quivalnt incom in trms of t xpnditur function: y = r ( p ; v( p, y) ) r = y ( p, p, y), (2) 2 Obviously, tis is not t bst option, altoug tr ar good rasons to us tis indicator in practic. Daton (997) provids an xaustiv discussion about tis subjct. 4 wr (.) is t xpnditur function, tat is t minimal amount of incom ncssary to rac utility v at prics p, and y (.) is t quivalnt incom function. 3 If p r is st to b qual to t non food subsidis pric systm, tat is p r = p o, t mov from t original situation to t currnt on, wn som food itms ar subsidizd, could b considrd as a first mans to figt against povrty by nancing t purcasing powr of ousolds. T maximum gain pr capita for ac ousold, Μ, rsulting from tis policy could b computd using t nxt formula: p o Μ = ( p p ) x, (3) wr x is t consumption baskt pr capita of ousold purcasd following tis policy. Bcaus it dos not rquir any ypotsis on t consumption bavior of ousolds, Μ is asy to comput. Yt considring t dadwigt loss (DL) rsulting from distortionary subsidis, tis masur ovrstimats t wlfar improvmnt. A satisfactory masur of t ousolds valu of tis program is t cang in tir quivalnt incom. Tis masur is known as t quivalnt gain pr capita, Ε, and it is givn by: Ε o p = y ( p, p, y = y y. T DL rsulting from distortionary subsidis lt t quivalnt gain always lss important tan t maximum gain. Hnc, a natural dfinition of t xcss burdn arising from tis distortionary transfr is: ) y ( p o, p o, y ) (4) DL = Μ Ε, (5) wr Μ and Ε ar rspctivly t avrag of t maximum and quivalnt gain. Along wit information about t distribution of wlfar gains among ousolds, it is worty to assss t social impact of t scm undr considration. Sinc t main objctiv of t currnt program is to improv t wlfar of t poor, a natural masur of its social impact could b givn by t dclin of a pr-spcifid povrty indx. Many povrty masurs can b xprssd in trms of povrty gaps g for incom y and povrty lin z as: g = max{ z y,0}. Following Jnkins and Lambrt (997), povrty masurs wic ar dfind in trms of g and wic ar in lin wit t aformntiond axioms blong to t class of Gnralizd 3 It is obvious tat y (p r, p r, y) = y. 5 Povrty Gap (GPG) indics. An important subst of GPG is t FGT st of additivly dcomposabl indics suggstd by Fostr t al. (984), wic can b writtn as: 4 = α H n Ρα ( z, y) = ( g ), (6) H n wr H is t total numbr of ousolds in t survy, n is t siz of ousold, n is t avrag siz of ousolds, and α may b dmd as a masur of povrty avrsion: a largr α givs gratr mpasis to t poorst of t poor. Wn α bcoms vry larg, Ρ α (.) approacs a Rawlsian masur, wic considrs only t poorst ousolds wlfar. T FGT class involvs many commonly usd povrty masurs as spcial cass. For instanc, wn α = 0, Ρ 0 (.) is t adcount ratio wic is insnsitiv to variations of t poor s incom; wil for α =, Ρ (.) is t (non-normalizd) povrty dficit. 5 Finally, for α , Ρ α (.) is snsitiv to inquality witin t poor. Considring w wis to assss t ffcts of t food subsidy program, povrty masurs sould b snsitiv to pric systm variations. For tis purpos, w prsnt tm in trms of valus of t quivalnt incom function. Hnc, t social impact of tis antipovrty program could b capturd using t nxt formula: Ρα = H H = n n H o o α n o p α ( z y (, p, y )) ( z (,, )), + y p p y + p (7) H n wr z is t quivalnt povrty lin, tat is t minimum xpnditur lvl rquird at p o to rac t indiffrnc curv corrsponding to t minimum standard of living on. Furtrmor, sinc t povrty masurs ar stimatd on t basis of sampl obsrvations, w nd to tst wtr t prdictd magnitud and dirction of cang in povrty following tis anti-povrty scm is statistically significant, wic is possibl using t tst of Kakwani (993): wr σ(.) is t standard rror of Ρ α : 6 = Ρα κ =, (8) σ Ρ ) ( α σ ( Ρ α ) = Ρ 2α ( z, y) [ Ρ nh α ( z, y)] 2 Ρ + 2α ( z, y + Ε) [ Ρ nh α ( z, y + Ε)] 2. 4 In rality, tis is a non-normalizd vrsion of t FGT povrty masurs class, adoptd also by Cakravarty and Mukrj (998). T normalizd vrsion is givn by ϕ, wr: ϕ = max{ y /z, 0}. 5 Tis is a good masur only if all t poor av t sam living standards. S t Axiom N of Sn (976). 6 It is vidnt from quation (4) tat y (p o, p p, y ) = y + Ε. 6 T mtodology prsntd abov is applid to data st from t 990 Tunisian survy. Tis is a multipurpos ousold survy wic provids information on xpnditurs and quantitis for food itms and xpnditurs for non-food itms, as wll as on many otr dimnsions of 7734 ousolds bavior; ducation, ousing, rgion of rsidnc, dmograpic information, and conomic activitis. T application of tis mtodology rquirs to comput formost t quivalnt incom function, y (.), t construction of wic calls for t stimation of an appropriat dmand systm. 7 Arguably, a gnral quilibrium modl is rquird to licit t saring out of food subsidy bnfits btwn firms and ousolds. Most computabl gnral quilibrium modls broadly assum tat all production functions ar omognous of dgr on and tat tr is prfct comptition. Undr ts assumptions, t supply curv of ac commodity is orizontal so tat consumrs rapd t ntir bnfits of t indirct transfrs. For simplicity, w assum suc framwork, altoug tr is noting in t followd approac tat prvnts t introduction of altrnativ ypotss. Hnc, troug tis scm, consumr pric is lowrd blow marginal cost by 37 prcnt for ard wat, 35 prcnt for tndr wat, 9 prcnt for otr wat, 4 prcnt for poultry and ggs, 8 prcnt for milk, 24 prcnt for sugar, and 34 prcnt for grain oil. T outcoms of tis program, for diffrnt povrty lins and povrty masurs, ar summarizd in t following tabl. 8 7 Bibi (998) as stimatd t paramtrs of IQAIDS dmand systm according Daton s (988) mtodology, wo assums spatial pric variation in t survy and maks us of unit valus as indicators of markt prics. W us t stimation rsults of tis dmand systm to comput t quivalnt incom. 8 T idntification problm is byond t aim of tis study. So w us a wid rang of povrty lins to avoid an arbitrary coic. Not tat t official povrty lin stimatd (undr t vctor p p ) by t National Statistic Institut (s t World Bank (995)) is btwn 96 and 252 DT. Ayadi and Matoussi (999) found tat z is ratr btwn 23 and 262 DT. 7 Tabl : Impact of targting by commoditis program on povrty 9 α z Ρ α (z, y) Ρ α (z, y + Ε) Ρ α (%) κ T prsnc of targting by commoditis scm is a maningful sourc of wlfar improvmnt for t poor, as t dclin of all povrty masurs provd. In addition, tabl sows tat t subsidis on foodstuffs bnfitd mor t poorst of t poor tan t ricst in rlativ trms, tat is targting by commoditis program is progrssiv in rlativ trms. For instanc, w not tat povrty rduction is lss important as t quivalnt povrty lin ris for a givn povrty indx. Tis rsult would b strngtnd if it could b sown tat t contribution of quivalnt gain to total xpnditur, (Ε /y ), dclind monotonically wit incrass in incom. In figur, w display t rsults of t link btwn ts two variabls obtaind using a non-paramtric stimation, wit Gaussian krnl and bandwidt slctd to minimiz t man intgratd squar rror. 0 T rsults ar rvaling, sinc ty confirm t progrssiv fatur of t currnt program. Also, as t slop of figur is always 9 For convninc, w rport [Ρ α (.)] (/α) instad of Ρ α (.) for α. So, if α = 2, w av a quadratic avrag of povrty dficit. Anotr attraction of tis incrasing monotonic transformation is tat, sinc t inquality among t poor lt [Ρ α (.)] (/α) Ρ (.), t diffrnc btwn [Ρ α (.)] (/α) and Ρ (.) could b intrprtd as an ovrall cost of inquality. 0 S Silvrman (986). 8 ngativ, rgardlss of t (quivalnt) povrty lin cosn, w can argu tat t postsubsidy distribution of xpnditurs dominat t (ypottical) pr-subsidy on witin all t rang of incom variations. Figur : Sar of quivalnt gain Sar of quivalnt gain 0,2 0,0 0,08 0,06 0,04 0,02 0, Total xpnditurs pr capita Prformancs of targting by commoditis in rducing povrty do not indicat, owvr, tat it is an optimal transfr dsign. Indd, t magnitud of t incom transfr to t nonpoor, tat is t lakags of t program bnfits, is vry important. T distribution of t absolut bnfits btwn t diffrnt quintil groups, arrangd in ascnding ordr from t poorst quintil to t ricst, is rportd in t following tabl. Tabl 2: Distribution of quivalnt and maximum gain from food subsidis program 2 Q s ys Εs Μ s n Ε Q s / B n Μ Q s / B Q (0.78) 3.0 (3.0) Q (25.35) 7.57 (30.67) Q (4.6) 9.84 (50.5) Q (59.55) (72.78) Q (83.27) (00.0) In t stocastic dominanc litratur, tis rsult is known as first-ordr dominanc. 2 Valus btwn parntss indicat t cumulativ distribution of t variabl undr considration. 9 Tabl 2 sows tat t anti-povrty program in and bnfitd t ric mor tan t poor in absolut trms. T ricst quintil group of t population rcivd 2.2 tims mor of t quivalnt gains from food subsidis tan t poorst, wit an avrag quivalnt gain pr capita ( Εs ) of 4.25 DT and 8.75 DT rspctivly. 3 Tis mistakn awarding of transfrs to t non-targtd group rducs t vrtical fficincy of tis scm and lads to t lakag of program bnfits. It is obtaind by adding t transfrs tat ar givn to tos wo ar inligibl in prfct targting and t total cost of t dadwigt loss. Indd, t xcss burdn cost rprsnts anotr sourc of lakag, wic is spcific to t distortionary transfrs, and sould also b addd to t traditional lakag cost. T lakag ratio, wic is obtaind by dividing t lakags by t availabl budgt, approximats at last 75 prcnt of t anti-povrty funds vn if w sould admit tat 40 prcnt of t population is poor. 4 Furtr, 6.7 points of prcntag of t lakag ratio ar rlatd to t xcss burdn cost. In t absnc of tis distortionary cost, t quivalnt gains would qualiz t maximum gain and t impact on povrty would b mor important. 5 Rgarding t wigt of t dadwigt loss and, spcially, t wak bnfits targtd to t poorst group, t rstructuring of tis scm must b a prssing priority. Tus, it is worty to invstigat first, rform possibilitis witin t xisting framwork of targting by commoditis ovr tos rquiring a nw institutional structur to avoid t transition costs of suc a moving. 6 T prvious analysis raiss t qustion of ow to improv targting by commoditis so as to rduc as muc as possibl t lakag of its bnfits to t non-poor. For tis purpos, w nd an stimat of t xpctd xpnditur on ac commodity conditional upon individuals incom, wic could b obtaind consistntly by non-paramtric rgrssions. If t stimation rsults rval t prsnc of som commoditis prdominantly consumd by t poorst, incrasing food subsidy to ts commoditis sould b an ffctiv way to rais t poorst sar of food subsidis bnfits; and so to lssn mor povrty. Estimation rsults of t incom-xpnditur rlationsip for diffrnt commoditis (t Engl curvs) ar displayd grapically in figur 2. 3 In 990, DT is clos to US dollar. 4 Crdy (996) as distinguisd btwn t vrtical xpnditur infficincy, tat is qual to t lakag ratio as dfind r, and t povrty rduction fficincy wic includs also t total transfrs wic, altoug rcivd by t pr-transfr poor, ar in a sns unncssary sinc ts individuals ar raisd abov t povrty lin. Bcaus t cost of unncssary transfrs is rally insignificant wn anti-povrty dsign is basd on targting by commoditis, w ignor its computation in tis sction. 5 Simulations sow tat t opportunity cost of t xcss burdn in trms of povrty rduction is btwn 2 and 3 prcnt according to t povrty lin and povrty masur slctd. S tabl A-3 in annx for mor dtails. 6 Ts costs ar rlatd to t political, conomic, and administrativ constraints of implmnting a mor targtd program xcluding powrful groups of t population. On tis issu, s Bsly and Kanbur (993). 0 Figur 2: Non-paramtric rgrssions of Engl's curvs 20,00 Expnditur on ac food commodity 00,00 80,00 60,00 40,00 20,00 0, Total xpnditur pr capita Hard wat Cooking oil Tndr wat Otr subsidizd commoditis It is wortwil to giv anotr intrprtation of tis figur. Considr a policymakr wo aims to dcras t incidnc of povrty but dos not know wo is rally poor. To b on t saf sid, it is bst to lssn t adcount ratio rgardlss of t povrty lin cosn. 7 Tis rquirs t prsnc of som commoditis wos xpnditurs ar downward sloping across diffrnt incom groups (at last from a trsold lvl of incom) and vn fall down at igr incom groups. Unfortunatly, t main fatur of t rgrssions sown in figur 2 is tat tr is no commodity fulfilling tis rquirmnt. Rstructuring t currnt scm, by rducing subsidis on otr subsidizd itms and raising tm on wat and cooking oil, could probably dcras to som xtnt lakags. Givn tat t non-poor purcas all commoditis, tis is not an ffctiv mans to incras targting accuracy and so, it is unlikly to improv significantly t living standard of t poorst witout looking for an altrnativ targting procdur. 7 Tis is ncssary to av a post-rform incom distribution wic first-ordr-dominats t pr-rform incom distribution, rgardlss of t povrty lins and povrty masurs cosn. 3. An altrnativ mans to figt against povrty: proxy mans tsts T prvious findings cifly rval tat lakags from food subsidis to non-poor popl ar vry larg wil a failur to improv substantially t living standards of t poorst is manifst. Bcaus govrnmnts av limitd rsourcs, and givn tat it is important to us tm fficintly, looking for an altrnativ targting tool to aciv mor povrty rduction is an appaling goal. Howvr, it is not usually asy to idntify t poor dirctly. Wilst suc idntification is rquird to curb povrty givn an availabl budgt, it is unlikly to b administrativly fasibl caply [Bsly and Kanbur (993)]. It is for tis rason tat among targting options, awarding bnfits to t poorst basd on targting by indicators could b an attractiv altrnativ to targting by commoditis. 8 Tr is plnty of tortical modling and mpirical stimation concrnd wit t qustion of ow to provid assistanc to poor popl wn it is possibl to obsrv som individuals caractristics, but not tir incom. For instanc, Ravallion and Cao (989) av modld t targting problm as o
Related Search
Similar documents
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks