Ol Dscovery, Real Excange Rate Apprecaton and Poverty n Gana Carles Acka and Dens Medvedev ABSTRACT Te dscovery of commercal quanttes of ol and gas off te coast of Gana n June 2008 as gven rse to muc optmsm for mproved prospects for accelerated growt and poverty reducton. However, altoug proper management of ncreased foregn excange nflows may go a long way towards elpng te government aceve ts growt and poverty reducton objectves, te assocated onset of Dutc dsease may also reduce compettveness n key export and mport-competng sectors, and ave an adverse mpact on te lveloods of major segments of te populaton. In order to assess te extent of vulnerablty of Ganaan ouseolds to te lkely apprecaton n te real excange rate, we estmate te potental sort-term mpact on poverty usng a mcro-accountng metodology drawng on te most recent round of te Ganaan Lvng Standards Survey. Te man fndng of te paper s tat wtn a plausble range of real excange rate apprecaton, aggregate poverty s lkely to ncrease only margnally. However, a far amount of dstrbutonal varaton underles ts small aggregate cange. Export farmers n te better-off soutern regons of te country are lkely to experence welfare losses, n many cases, large enoug to pus tem below te poverty lne. Insttute of Statstcal, Socal and Economc Researc (ISSER), Unversty of Gana. E-mal: cacka@ug.edu.g. Economc Polcy Unt, Latn Amerca and te Carbbean, Te World Bank Group.
2
1. Introducton Wen Gana dscovered commercal quanttes of ol and gas off ts coast n June 2008, many of ts ctzens became optmstc of mproved prospects for accelerated growt and poverty reducton. However, after two years of ol producton and exports, tere are not many wo stll beleve tat te dscovery of ol n Gana wll nure to te benefts of te poor n Gana. Tere ave been publc concerns for te proper management of te ncreased foregn excange nflows and te transparent use and far dstrbuton of te fnancal gans. Tere are also fears of a possble Dutc dsease, wc could reduce compettveness n key export and mport-competng sectors and adversely mpact te lveloods of major parts of te populaton. Altoug proper management of ncreased foregn excange nflows may go a long way towards elpng te government aceve te growt and poverty reducton objectves, te assocated onset of Dutc dsease may also reduce compettveness n key export and mport-competng sectors and adversely mpact te lveloods of major parts of te populaton. In order to assess te extent of vulnerablty of Ganaan ouseolds partcularly tose at or below te natonal poverty lne to te lkely apprecaton n te real excange rate (RER), ts paper estmates te potental sort-term mpacts on poverty usng a mcro-accountng metodology drawng on te most recent round of te Ganaan ouseold survey. Te rest of te paper s organzed as follows: Secton 2 provdes a bref lterature revew wle Secton 3 descrbes te data, metodology, and provdes some descrptve statstcs. Secton 4 presents te poverty and welfare estmates, and Secton 5 concludes. 3
2. Lterature revew Tere s a growng uneasness over a trend tat seems to suggest tat a country wc s abundantly endowed wt natural resources must defntely suffer low economc performance. Some scolars attrbute ts penomenon partly to overdependence on natural resources wle te resource-rc country smultaneously experences less trade and foregn nvestment, more corrupton, less equalty, less poltcal lberty, less educaton, less domestc nvestment, and less fnancal dept (Bresnger, 2009). Moreover, governments of resource rc countres, apart from beng tempted to pursue polces nmcal to growt, also tend to relax n ter tax revenue efforts. Gylfason (2001) explans tat abundant natural resources may mbue people wt a false sense of securty and lead governments to lose sgt of te need for good and growtfrendly economc management ncludng free trade, bureaucratc effcency, and nsttutonal qualty. Typcal examples of countres tat ave suffered ts penomenon nclude Sudan and Ngera wc over te past ffty years ave made several efforts to leverage resource wealt for development but ad woefully faled n te process (Moss and Young; 2009). Indeed, a sgnfcant body of researc over te last two decades as sown tat te natural resource endowment of many developng countres adversely affected te effcency wt wc tey ave managed ter resources. Ts often resulted n worse socal and economc condtons a penomenon known as te resource curse. Among te factors tat ave been used to explan te resource curse s te Dutc Dsease were te ncreased revenue from te mneral resources exported as te effect of rasng te prce of domestc goods relatve to foregn goods. 4
Dutc Dsease s derved from an economc penomenon observed n Holland were te dscovery of natural gas reserves n te 1960s led to a slump n oter sectors lke manufacturng. Fears of a Dutc Dsease customarly assume tat a szeable nflow of ODA wll exacerbate macroeconomc nstablty, namely, by rasng nflaton and apprecatng te real excange rate. Moreover, t s also assumed tat growt wll be mpared because excange-rate apprecaton wll amper te compettveness of a country s export sector (Rajan and Subramanam, 2005). Accordng to Wunder (1997), te Dutc Dsease fundamentally tells a story of sectoral reallocaton n an open economy facng a foregn excange transfer. Wunder used te tree sector model developed by Corden (1982) and Corden and Neary (1982) to explan te Dutc Dsease : Te frst sector s te boomng sector tat generates a wndfall foregn excange proft, probably as a result of rsng ol prces, generatng ger natonal ncome and spendng. Ts rse n aggregate demand causes a ger prce for non-tradable goods (NT) - prvate and publc servces, constructon goods, etc. - wc cannot be mported from abroad. However, rsng demand for goods from te non-boomng tradable sector (T goods - agrculture and/or ndustry) s satsfed by an ncrease n mports - te prce of tese T goods s strctly world-market determned (te "law of one prce") - and tus fxed. An overvalued currency was te frst dentfed symptom assocated wt te Dutc Dsease (Gylfason, 2001b). Te natural resource boom and te assocated surge n rawmateral exports tend to drve up te value of te domestc currency n real term. Snce natural resource abundance s often accompaned by booms and busts, excange rate volatlty s gly mmnent (bd). However, te unstable excange rates create uncertanty tat tends to urt exports and oter trade, ncludng foregn nvestment. Apart from te excange rate volatlty, te ncreasng currency value also means tat te non- boomng trade sector s no longer attractve to te workforce. As for tat matter, labour 5
turn to te non-trade sector (suc as servces) and troug an ncrease n te demand for non-traded goods, te prce of te non-trade sector ncreases. Ts leads to a steep boost n te real excange rate (Gordon, 2006). Implcaton of natural resource boom to te ncdence of poverty remans unambguous n te lterature. Wle t s clear resource boom sometmes propels economc growt, suc growt mgt not be pro-poor. Accordng to UNDP (2006) growt s sad to be pro-poor wen.t s labour-absorbng and accompaned by polces and programmes tat mtgate nequaltes and facltate ncome and employment generaton for te poor, partcularly women and oter tradtonally excluded groups. Te ol sector as been found not to be too labour-absorbng (Gordon, 2006); t absorbs only few gly sklled personnel, wle sometmes contrbutng to te contracton n te nonboomng sector s exports, and even slacks n aggregate supply response. However, suc contractons n te performance of te non-ol exports can worsen te plgt of te poor wo are manly found n tese sectors (bd). Gylfason (2001b) fears tat uge natural resource rents and market mperfectons can encourage rampant rent-seekng beavour on te part of producers, especally n te developng economes often caracterzed by ll-defned property rgts, mssng markets, and lax legal structures. He explans tat rent-seekng beavours can dvert resources away from more socally frutful economc actvtes. Wen ts appens, te poor suffer as te few powerful and corrupt poltcans and rc busnessmen would contnue to make more money. Addtonally, an abundance of natural resources may attract foregn governments to nvade wt destructve consequences tereby promptng te domestc autortes to 6
spend vast resources on natonal defense. Unfortunately, mltary expendtures tend to nbt growt troug ter adverse effects on captal formaton and resource allocaton Gylfason (2001b). Also, natural resource abundance may reduce prvate and publc ncentves to accumulate uman captal due to a g level of non-wage ncome suc as dvdends, socal spendng and low taxes. Apart from te above, resource abundance may also affect poverty troug ncreases n te prces of goods and servces (Ose and Domfe, 2010). Resource boom often leads to nflatonary pressures bot from te pont of vew of demand pull or cost pus factors of nflaton. As a result of te ol boom, te capacty of te manufacturng sector tat produces agrcultural nputs may be adversely affected and n te process cause ncreases n te prces of agrcultural produce. Moreover, te laggng manufacturng sector may be affected n producton of oter consumable goods wc cannot be mported. Te scarcty of suc goods n te local market wll cause prces ncreases, worsenng te lvng condton of te people (bd). In te nutsell, te value of te local currency rses; manufacturng sector becomes uncompettve, leadng to an excessve slump. Ts eventually leads to job losses n te manufacturng sector snce most of te manufacturng companes would eter stay on and cut producton or would completely move away to oter countres. Poverty creeps mmedately to te ouseolds wose breadwnners may suffer job losses. 3. Data, metodology, and descrptve statstcs Te data for te exercse s te 2005/2006 round of te Gana Lvng Standards ouseold Survey (GLSS5), descrbed n detal n GSS (2007a, 2007b). Consstent wt 7
most analytcal work usng te GLSS5 data (e.g., Coulombe and Wodon, 2007), ts paper uses consumpton per adult equvalent as te welfare metrc, wt te poverty lne set at 3,708,900 ceds per adult per year (2006 prces), wc rougly corresponds to te $1.25 per day nternatonal extreme poverty lne (GSS, 2007a). Despte te mpressve pace of poverty reducton over te past 15 years (documented n World Bank, 2007), poverty ncdence remans relatvely g at more tan 28 percent of te populaton. Furtermore, as sown n te frst ( ntal ) column of Table 1, te ncdence of poverty s gly based towards rural areas, partcularly tose n te Nort of te country. Te same pattern s evdent n te dstrbuton of poverty across soco-economc groups, as food crop farmers and ter famles wo account for more tan 72 percent of te populaton n te Nortern, Upper West, and Upper East regons experence poverty ncdence tat s nearly twce as g as next poorest group, export farmers. Altoug te poverty rate among te latter s relatvely g, te dstrbuton of welfare n ts group s muc wder tan among food crop farmers, wt nearly 45 percent of export farmers earnng ncomes tat place tem n te top alf of te natonal dstrbuton. 8
Table 1: Poverty effects of real excange rate apprecaton n Gana (natonal poverty lne) Intal 10% RER apprecaton 25% RER apprecaton Natonal Headcount rato 0.2855 0.2874 0.2882 Poverty gap 0.0959 0.0957 0.0964 Squared poverty gap 0.0459 0.0458 0.0460 Headcount rato by: Localty Urban 0.1074 0.1066 0.1072 Rural 0.3932 0.3967 0.3976 Regon Western 0.1857 0.1925 0.2096 Central 0.1994 0.2012 0.1841 Greater Accra 0.1179 0.1176 0.1231 Volta 0.3171 0.3254 0.3134 Eastern 0.1473 0.1481 0.1537 Asant 0.2051 0.2097 0.2067 Brong Aafo 0.2968 0.2934 0.2941 Nortern 0.5215 0.5226 0.5246 Upper East 0.7050 0.7025 0.7034 Upper West 0.8789 0.8776 0.8797 Sector of ouseold ead Publc sector employment 0.0782 0.0782 0.0741 Prvate formal employment 0.1015 0.1059 0.1075 Prvate nformal empl. 0.1707 0.1693 0.1763 Export Farmers 0.2412 0.2754 0.3013 Food Crop Farmers 0.4573 0.4555 0.4530 Non-Farm Self-Employed 0.1668 0.1666 0.1647 Non-workng 0.1304 0.1303 0.1359 Gender of ouseold ead Male 0.3136 0.3165 0.3194 Female 0.2350 0.2352 0.2323 Note: ead of ouseold s te economc ead. RER s te real excange rate. Te frst step n assessng te potental mpacts of real excange rate apprecaton on poverty and ncome dstrbuton s to dentfy te extent to wc te domestc markets are lnked to nternatonal markets. Ts s captured by te ratos of exports to total output and mports to total demand (absorpton), calculated usng data from te most recent (2005) socal accountng matrx (SAM) for Gana and sown n Annex 9
Annex Table 2. Tese export and mport sares, calculated for 66 broad product categores dentfed n te SAM, are ten appled to te full set of product categores avalable n te ncome generaton and consumpton modules of te GLSS5. Due to te lack of data on excange rate pass-troug, perfect pass-troug s assumed. Ts means tat for eac good, te sare of te good tat s tradable does not vary across ouseolds or regons. 1 Te ncome and expendture flows assocated wt eac product are ten summed over te full range of products produced and consumed by eac ouseold, and te resultng dstrbuton s sown n Fgure 1 by populaton percentle (ranked from poorest to te rcest). Te fgure sows tat, despte relatvely g macro openness numbers exports-to-gdp rato of 43 percent and mports-to-gdp rato of 71 percent te average ouseold consumes and earns most of ts ncome from non-tradable sources. 2 1 In oter words, ts calculaton assumes a natonal Armngton agent. 2 Altoug te exact source bend ts dscrepancy s dffcult to establs, te lkely reasons nclude a lower mport content of prvate consumpton relatve to nvestment or publc spendng, te tendency of ouseold surveys to capture small transactons more accurately tan larger ones, and te under-samplng of wealter ndvduals wo are more lkely to consume mported goods. For a more detaled dscusson of potental reasons for substantal dfferences between averages computed usng ouseold survey data and aggregates reported n natonal accounts, see Deaton (1997). 10
Fgure 1: Dstrbuton of non-tradable consumpton and ncome n Gana 100% Sare of non-tradables n total consumpton, by percentle 100% Sare of non-tradables n total ncome, by percentle 95% 95% 90% 90% 85% 85% 80% 80% 75% 0 10 20 30 40 50 60 70 80 90 100 75% 0 10 20 30 40 50 60 70 80 90 100 As glgted n te recent lterature on te food prce crss (e.g., Wodon and Zaman, 2008; De Hoyos and Medvedev, 2008), polcymakers are sometmes presented wt a dlemma were te poorest parts of te populaton are not always te most vulnerable to a partcular sock. In te case of real excange rate apprecaton, te most vulnerable groups are tose wo earn a large sare of ter ncome from tradable actvtes and spend a large sare of ter consumpton budget on non-tradable goods. As llustrated by Fgure 1, te poorer parts of te populaton devote a larger sare of ter expendtures to non-tradable goods tan te rcer parts, wc speaks to te vulnerabltes of te poor on te consumpton sde. At te same tme, owever, te poorer groups also earn a larger sare of ter ncome from non-tradable sources. Moreover, te sare of ncome earned from non-tradable actvtes s n many cases above te sare of expendtures devoted to non-tradables for ouseolds below te poverty lne, wc suggests tat te adverse mpacts of real excange rate apprecaton on te poor are lkely to be lmted. 11
Te next step n estmatng te lkely mpacts of a cange n relatve prces on poverty s to map prce canges to welfare outcomes. Ts paper reles on a mcroaccountng metodology to calculate te frst-order approxmaton to te lkely cange n ouseold welfare by focusng solely on prce canges. 3 Ts means tat no allowances are made for ouseolds re-optmzng te structure of ter consumpton bundle or te varety of goods tey produce n response to a prce sock. In ts framework, canges n ouseold welfare are determned only by te magntude of RER apprecaton and te sares of ncome and consumpton derved from non-tradable sources. If a gven ouseold earns more from non-tradable actvtes tan t spends on non-tradable goods, ts ouseold benefts from an ncrease n te relatve prce of non-tradables. Te converse s true for a ouseold for wc te sum of ncome from non-tradable sources s below te total consumpton of non-tradable goods. Terefore, te approac essentally segments te populaton nto net consumers and net producers of non-tradables wt te former lkely to lose and te latter to gan from RER apprecaton wle retanng te flexblty to undertake more detaled analyss on bot te consumpton and producton sdes. Te approac detaled below dentfes separately te mpact of te sock on ouseold ncomes and te cost of ouseold consumpton basket, and estmates te fnal cange n poverty as a combnaton of ncome (employment) and consumpton socks. Allow te welfare ndcator W of ouseold to be a functon of ts consumpton and ncome, wc are n turn functons of prces and quanttes: W f ( C, Y ) f ( c( p, x ), y( w, q )) 3 See te Annex for te metodologcal detals and te matematcal statement. 12
13 Note tat ouseold ncome Y s also a functon of prces and quanttes suppled: ), ( ), ( s x p g w q y Y Usng te envelope teorem, we can approxmate te near-term cange n ouseold welfare by keepng track of canges n prces: d s dp p x p c dp p x p g dw ), ( ), (,, Assumng perfect pass-troug of te RER sock, defne tradablty sares for eac product usng detaled balance of payments data: g g g c c c ) (1 ) (1 We can ten wrte te canges n ouseold ncome and consumpton as a functon of tradablty sares and te cange n te real excange rate r, defned as te prce of nontradable goods relatve to te prce of tradable goods: dr c dp p x p c dr g dp p x p g d s,,,, ), ( ), ( Terefore, te total mpact on ouseold welfare s just a functon of te sare of ncome derved from non-tradable goods relatve to te sare of consumpton spent on non-tradables and te extent of te RER apprecaton: c g dr dw,, A complcaton tat often arses wen workng wt bot ncome and expendture aggregates n ouseold surveys s te serous underestmaton of ncome flows. Many tmes, ts s a problem not only of averages (.e., average ncome s sgnfcantly below
average consumpton) but also of dstrbuton, wt te ncome-consumpton gap growng as one moves from te poor to te rc parts of te ncome dstrbuton. Ts s also te case n Gana, as sown n Fgure 2. Te man reason for a growng gap between ncome and consumpton s te well-known problem of ms-measurement and under-reportng, combned wt te ncluson of auto-consumpton n bot te ncome and consumpton aggregates. As a result, te gap between ncome and consumpton s small wen autoconsumpton represents a large sare of ouseold budget, wc s te case for te poorest parts of te populaton. As welfare rses, te mportance of auto-consumpton declnes and te underestmaton of ncome (from sources oter tan auto-consumpton) becomes more obvous. Fgure 2: Declnng sare of ncome to consumpton across te dstrbuton Rato of ouseold ncome to consumpton, by percentle 200% 180% 160% 140% 120% 100% 80% 60% 40% 20% 0% 0 10 20 30 40 50 60 70 80 90 100 Te pattern sown n Fgure 2 presents an obvous problem n calculatng welfare canges usng bot ncome and consumpton aggregates. Followng te above 14
metodology wll ntroduce a progressve dstrbutonal bas n te welfare calculaton, wt rcer parts of te populaton losng more tan te poorer ones. In order to avod ntroducng ts bas n te calculatons, ts paper does not use te ncome aggregate drectly. Instead, we calculate te sare of eac ouseold s ncome tat s due to nontradable sources usng te ncome aggregate, and ten apply tat sare to te consumpton aggregate (wc s te man welfare metrc) to calculate adjusted ncome by ouseold. 4. Poverty and dstrbutonal mpacts of a 10 (25) percent apprecaton n te real excange rate Usng te data and metodology descrbed n te prevous secton, te followng paragraps dscuss te potental canges n poverty and te dstrbuton of ncome wt a ypotetcal 10 and 25 percent apprecaton of te real excange rate. Te poverty effects are summarzed n te second and trd columns of Table 1, wle te mpacts on ouseold welfare (per adult equvalent) are sown n Annex Table 3. Te results ndcate tat poverty s lkely to ncrease only margnally, wt a 0.2 and 0.3 percentage pont ncrease n te eadcount rato, respectvely. Te dept of poverty may actually decrease slgtly wt a 10 percent apprecaton, as sown by te declne n te poverty gap and squared poverty gap statstcs. Ts ndcates tat te poorest parts of te populaton may actually beneft somewat from RER apprecaton, even toug te ouseolds n te vcnty of te poverty lne may be adversely affected. 15
In order to understand te dstrbutonal mpacts of RER apprecaton, t s useful to consder te growt ncdence of a 10 percent apprecaton sown n Fgure 3. 4 Te verson of te growt ncdence curve drawn n te fgure dffers slgtly from te orgnal growt ncdence curve concept of Ravallon and Cen (2003) by sowng te percentage gans n welfare by percentles of te orgnal dstrbuton of welfare. For te bottom alf of te dstrbuton, te ncdence of RER apprecaton s welfare-enancng and qute progressve, even toug ouseolds very close to te poverty lne (wc s drawn near te 28 t percentle of te dstrbuton) are lkely to experence negatve welfare canges. Tese outcomes are determned by te ncome and consumpton patterns n Fgure 1, wc sowed tat te sare of ncome earned from non-tradable actvtes by te poorest members of te Ganaan socety s above te sare of total expendture devoted to te consumpton of tese goods. 4 Te growt ncdence curve for a 25 percent RER apprecaton s not sown because welfare canges (summarzed by te curve) are monotonc n te RER. Ts s evdent by comparng te 10 and 25 percent apprecaton columns n Annex Table 3. 16
Fgure 3: Growt ncdence curve of RER apprecaton n Gana 0.80 Percent cange n welfare from 10% RER ncrease, by percentle 0.60 0.40 0.20 0.00-0.20-0.40-0.60 10 20 30 40 50 60 70 80 90 100 Ts may also be seen n te rows of Table 1 and Annex Table 3 tat sow te cange n poverty and welfare by te sector of employment of te ouseold ead. For example, food crop farmers and ter famles, wo account for 69 percent of all poor n Gana, spend 87 of ter consumpton budget on non-tradable goods but derve 91 percent of ter ncome from non-tradables. Terefore, ts poorest segment of Ganaan socety could actually beneft from an ncrease n te relatve prce of non-tradable goods. On te oter and, export farmers accountng for just 6 percent of all poor n Gana spend 86 percent of ter consumpton bundle on non-tradables but earn just 56 percent of ter ncome from non-tradable sources. Not surprsngly, ter welfare declnes by nearly 3 percent wt a 10 percent RER sock, and poverty ncdence also rses by more tan 3 percentage ponts. 17
Te decomposton of canges n te poverty eadcount by localty sows tat urban poverty s lkely to declne wle rural poverty may rse. Almost 92 percent of te total ncrease n rural poverty s explaned by an ncrease n poverty among export farmers, more tan 90 percent of wom resde n rural areas. Altoug poverty ncdence also rses among formal workers n te prvate sector, ter densty around te poverty lne s low and terefore te small ncrease n urban poverty due to welfare losses n te formal sector s outweged by gans among self-employed and nformal sector workers. Te canges n poverty at te regonal level are somewat dffcult to analyze because tey are dependent on te magntude of te excange rate sock. For example, poverty ncdence could ncrease n te Central and Volta regons wt a 10 percent RER apprecaton, but declne wt an apprecaton of 20 percent or more. Te reason s tat a 10 percent RER apprecaton s suffcent to pus some export farmer ouseolds below te poverty lne, but te ncrease n welfare of food crop farmers s not enoug to pull tem above te poverty lne. Wt a larger sock, poverty reducton among te latter, more numerous group, outwegs te poverty ncrease among te former. Smlarly, poverty ncreases n te Nortern regon wt bot a 10 and a 25 percent sock because te ncrease n te relatve prce of non-tradables of ts magntude, altoug welfareenancng, s not enoug to lft most of te food crop farmers out of poverty, but enoug to pus some self-employed workers below te poverty lne. Houseolds were a female s te most mportant earner (economc ead) are lkely to experence almost no cange n poverty ncdence wt a 10 percent RER apprecaton and some reducton n poverty wt an apprecaton of 25 percent. On te oter and, male-eaded ouseolds wc represent two-trds of te Ganaan populaton could 18
face an ncreasng ncdence of poverty. Ts s due to te fact tat te proporton of expendture spent on non-tradable tems for male-eaded ouseolds s ger tan te sare of ncome from non-tradable sources, wle te opposte s true for ouseolds eaded by females. 5 For example, male-eaded ouseolds are 50 percent more lkely to earn most of ter ncome from export crops tan female-eaded ouseolds. Fgure 4: Decomposton of consumpton and ncome effects -8.2 Percent cange n welfare from 10% RER sock appled to consumpton only, by percentle 10.0 Percent cange n welfare from 10% RER sock appled to ncome only, by percentle -8.3 9.5-8.4-8.5 9.0-8.6 8.5-8.7 8.0-8.8-8.9 7.5-9.0 10 20 30 40 50 60 70 80 90 100 7.0 10 20 30 40 50 60 70 80 90 100 As te prevous dscusson placed a lot of empass on te relatve strengt of ncome and consumpton effects n explanng te poverty and dstrbutonal outcomes of RER apprecaton, t s useful to consder eac of te effects separately, as sown n Fgure 4. Te left panel of te fgure sows te growt ncdence curve of a 10 percent RER apprecaton appled to te consumpton aggregate only, wtout cangng ouseold ncomes. Gven te dstrbutonal pattern of consumpton of non-tradables sown n Fgure 1, t s not surprsng tat te ncdence of te sock s qute regressve wt te poorest ouseolds losng as muc as 8.9 percent of ter orgnal welfare wle te 5 Wle te average sare of non-tradables n consumpton s rougly te same for bot groups (86.2 percent for female-eaded ouseolds and 86.6 percent for male-eaded ouseolds), te sares of ncome derved from non-tradable sources are very dfferent: 85 percent for males and 91 percent for females. 19
average welfare reducton s around 8.5 percent. In terms of te poverty eadcount, ts consumpton-only sock leads to a 4.4 percentage pont ncrease n te poverty eadcount, rasng t to 32.9 percent of te populaton. On te oter and, te ncomeonly sock s rater progressve, wt te poorest ouseolds ganng as muc as 9.5 percent wle te average gan s closer to 8.6 percent. If real consumpton were to reman uncanged, ts knd of a sock would lead to 3.8 percentage pont declne n poverty, brngng te eadcount rato down to 24.7 percent. 5. Concludng remarks Wle te exact magntude of real excange rate apprecaton n Gana wll reman uncertan for several years, ts paper as sown tat wtn a plausble range of canges n te real excange rate suggested by general equlbrum models of te Ganaan economy canges n welfare and poverty at te natonal level are lkely to be mld. Even wt a 25 percent apprecaton n te real excange rate, real consumpton per adult equvalent would ncrease by just 0.08 percent, wle te ncdence of poverty could rse by 0.9 percentage ponts. Bend te small aggregate numbers, owever, a far amount of dstrbutonal cange s lkely to take place. Te poorest segments and regons of te Ganaan socety food crop farmers n te Nortern, Upper East, and Upper West regons are lkely to beneft from an ncrease n te relatve prce of non-tradables, altoug often tmes not enoug to rase ter ncomes above te poverty lne. On te oter and, export farmers and formal prvate sector employees n te Asant and Western regons are lkely to experence substantal welfare losses, wt a number of tese ouseolds slppng below te poverty tresold. 20
Tese results come wt a number of mportant caveats, and sould terefore be nterpreted as llustratve scenaros rater tan predctons of wat may actually appen. For nstance, te paper assumes no adaptaton and no beavor on te part of te ouseolds: wle ts may be a reasonable assumpton n te sort and medum term, over te long run ouseolds are lkely to re-optmze ter consumpton bundles towards less expensve goods and fnd new jobs n better-payng sectors. Ts suggests tat te results presented n ts paper may overestmate te fnal poverty effects. Smlarly, te analyss presented ere gnores te ncome effect tat s te source of Dutc dsease: te necessary condton for te ol revenues to result n real excange rate apprecaton s tat tey are spent, at least partally, on domestcally produced goods, and ts spendng s lkely to beneft certan groups of te populaton more tan oters. One can envson dfferent spendng strateges, some of wc may beneft ouseolds tat are most vulnerable to te real excange rate sock, wle oters may focus on te poorest ouseolds, wo may gan somewat from RER apprecaton but stll reman te least well-off members of te Ganaan socety. Fnally, te analyss presented ere does not allow for nternal or external mgraton of ouseolds or ter members n response to cangng economc opportuntes, wc as tradtonally been an mportant response of ouseolds to varous socks. 21
6. References Bresnger, C., Turlow, J. and Duncan, M. 2007. A 2005 Socal Accountng Matrx (SAM) for Gana. Gana; Wasngton, D.C.: Gana Statstcal Servces (GSS); Internatonal Food Polcy Researc Insttute (IFPRI)(datasets). ttp://www.fpr.org/data/gana03.asp Bresnger C., Dao X., Scweckert R., Webelt M., (2009): Managng Future Ol Revenues n Gana: An Assessment of Alternatve Allocaton Optons ; Internatonal Food Polcy Researc Insttute (IFPRI) Dscusson Paper 00893. Coulombe, Harold and QuentnWodon. 2007. Poverty, Lveloods, and Access to Basc Servces n Gana. In World Bank, Gana Country Economc Memorandum: Meetng te Callenge of Accelerated and Sared Growt, Report No. 40934-GH. De Hoyos, Rafael and Dens Medvedev. 2008. Poverty Effects of Hger Food Prces: A Global Perspectve. Paper presented at te Segundo Encuentro Regonal sobre Modelos de Equlbro General Computable: Sus Aportes en la Formulacón de la Polítca Económca en Amérca Latna y el Carbe conference, San Jose, Costa Rca. Gordon, A. 2006. Dutc Dsease: How One Industry Causes Natonal Economc Downturn. Retreved February 19, 2010, from ttp://ezneartcles.com Gylfason, T. 2001: Lessons from te Dutc Dsease: Causes, Treatment, and Cures ; Insttute of Economc Studes, Workng Paper Seres, W01:06 ISSN 1011-8888, Unversty of Iceland Odd at Sturlugata, 101 Reykjavk Iceland. Gylfason, T., (2001b): Nature, Power, and Growt, Scotts Journal of Poltcal Economy, Scotts Economc Socety, vol. 48(5), pages 558-88. Ravallon, M. and Cen. S. 2003. Measurng Pro-Poor Growt, Economcs Letters, Vol. 78(1): 93-99. Moss T., and Young, L. 2009: Savng Gana from Its Ol: Te Case for Drect Cas Dstrbuton. Center for Global Development, Workng Paper 186, 1800 Massacusetts Ave., NW Wasngton, DC. Ose R.D. and Domfe, G. 2010: Macroeconomc Implcatons of Ol Boom: Lessons For Gana, unpublsed paper submtted to Fredrc Ebert Stftung, Gana, as a Capter n a book enttled, Managng Gana's Ol Boom. UNDP (2006): Macroeconomc polces for poverty reducton: Te case of Sudan, Publsed for te Unted Natons Development Programme n Sudan, Kartoum, Sudan. 22
Wodon, Q. and Zaman, H. 2008. Rsng food prces n Sub-Saaran Afrca : poverty mpact and polcy responses, World Bank Polcy Researc Workng Paper 4738. World Bank. 2007. Gana Country Economc Memorandum: Meetng te Callenge of Accelerated and Sared Growt, Report No. 40934-GH. Wunder S., (1997), From Dutc Dsease to Deforestaton - A Macroeconomc Lnk? A case study from Equador : Dans Insttute for Internatonal Studes Publcatons from te former Centre for Development Researc CDR Workng Paper 976. 23
Annex Annex Table 2: Trade exposure n Gana, by sector Imports-to-absorpton Exports-to-output Maze 15.3 0.0 Rce 64.3 0.0 Oter cereals 12.0 0.0 Cassava 0.0 0.5 Yams 0.0 2.0 Cocoyams 0.0 0.0 Cowpea 0.0 0.0 Soyabean 0.0 0.0 Palm ol 0.0 30.5 Groundnuts 0.0 8.3 Tree nuts 0.0 35.7 Frut (domestc) 0.0 0.0 Frut (export) 0.0 82.2 Vegetables (domestc) 0.0 0.0 Vegetables (export) 0.0 79.9 Plantans 0.0 0.0 Cocoa beans 0.0 86.6 Oter crops 31.2 0.0 Export ndustral crops 0.0 77.5 Ccken broler (mostly mported) 94.3 0.0 Eggs and layers (domestc) 15.2 0.0 Beef 33.7 0.0 Seep and goat meat 12.8 0.0 Oter meats 18.5 0.0 Forestry 0.0 73.1 Fsng 0.0 27.5 Gold 0.0 96.7 Oter mnng 0.0 77.3 Oter formal food processng 81.1 0.0 Informal food processng 0.0 0.0 Cocoa processng 60.9 80.1 Dary products 12.9 0.0 Meat and fs processng 42.8 22.6 Textles 60.7 11.2 Clotng 71.6 1.4 Leater and footwear 38.3 0.9 Wood products 0.0 68.0 Paper products, publsng and prntng 32.7 0.3 Petroleum 58.8 0.0 Desel 0.0 0.0 Oter fuels 97.6 0.0 Oter Cemcals 66.4 1.6 Rubber products 62.9 50.7 Non-metallc mneral products 50.5 0.0 Metal products 48.1 3.8 24
Non-electrcal macnes 97.8 0.0 Electrcal macnes 97.8 0.0 Rado and televson equpment 99.3 0.0 Medcal and optcal equpment 99.7 0.0 Motor vecles and parts 99.8 0.0 Oter transport equpment 99.9 0.0 Oter manufactured products 45.3 0.0 Constructon 0.0 0.0 Water 0.0 0.0 Electrcty and gas 0.9 0.0 Trade servces 0.0 0.0 Reparng, otel, and restaurant 29.2 0.0 Transport servces 0.0 51.0 Communcaton 0.0 0.0 Bankng and busness servces 0.0 0.0 Real estate 0.0 0.0 Communty and oter servces 0.0 0.0 Publc admnstraton 0.0 0.0 Educaton 0.0 0.0 Healt 0.0 0.0 Source: Bresnger et al (2007) 25
Annex Table 3: Cange n ouseold per capta welfare ndex Level Percent cange Intal 10% appr. 25% appr. 10% appr. 25% appr. Natonal 7,367,964 7,370,213 7,373,587 0.03% 0.08% Welfare decle 1 1,531,052 1,535,510 1,542,198 0.29% 0.73% 2 2,619,460 2,625,529 2,634,633 0.23% 0.58% 3 3,433,787 3,435,813 3,438,853 0.06% 0.15% 4 4,228,252 4,226,839 4,224,720-0.03% -0.08% 5 5,103,160 5,099,677 5,094,453-0.07% -0.17% 6 6,118,064 6,116,302 6,113,659-0.03% -0.07% 7 7,285,038 7,278,469 7,268,616-0.09% -0.23% 8 9,034,603 9,043,963 9,058,002 0.10% 0.26% 9 11,676,753 11,686,854 11,702,005 0.09% 0.22% 10 22,665,560 22,669,242 22,674,764 0.02% 0.04% Localty Urban 10,506,222 10,531,344 10,569,026 0.24% 0.60% Rural 5,470,986 5,459,410 5,442,046-0.21% -0.53% Regon Western 7,795,184 7,724,746 7,619,089-0.90% -2.26% Central 8,325,059 8,341,824 8,366,972 0.20% 0.50% Greater Accra 10,891,068 10,918,339 10,959,245 0.25% 0.63% Volta 6,322,330 6,328,224 6,337,064 0.09% 0.23% Eastern 7,791,521 7,791,058 7,790,364-0.01% -0.01% Asant 8,278,984 8,272,862 8,263,678-0.07% -0.18% Brong Aafo 6,657,145 6,657,116 6,657,071 0.00% 0.00% Nortern 4,746,855 4,776,444 4,820,826 0.62% 1.56% Upper East 3,393,715 3,406,143 3,424,786 0.37% 0.92% Upper West 2,352,809 2,369,635 2,394,873 0.72% 1.79% Sector of ouseold ead Publc sector empl. 11,634,901 11,708,542 11,819,003 0.63% 1.58% Prvate formal empl. 11,583,756 11,576,377 11,565,309-0.06% -0.16% Prvate nformal empl. 7,912,996 7,946,122 7,995,810 0.42% 1.05% Export Farmers 6,511,220 6,317,359 6,026,567-2.98% -7.44% Food Crop Farmers 4,909,646 4,923,707 4,944,796 0.29% 0.72% Non-Farm Self-Employed 8,935,009 8,943,806 8,957,000 0.10% 0.25% Non-workng 10,172,853 10,222,492 10,296,950 0.49% 1.22% Gender of ouseold ead Male 7,149,412 7,129,475 7,099,568-0.28% -0.70% Female 7,759,610 7,801,619 7,864,632 0.54% 1.35% Note: ead of ouseold s te economc ead. 26