Forecast future production of municipal waste on the basis of a panel data model in Algeria

Transcription

1 MPRA Munich Personal RePEc Archive Forecas fuure producion of municipal wase on he basis of a panel daa model in Algeria Brahim Djemaci Universiy of Boumerdes (Algeria) 16 January 2016 Online a hps://mpra.ub.uni-muenchen.de/68879/ MPRA Paper No , posed 18 January :27 UTC

2 Forecas fuure producion of municipal wase on he basis of a panel daa model in Algeria Brahim Djemaci 1 Universiy of Boumerdès (Algeria) Absrac: This sudy analyses he facors ha influenced he producion of municipal wase in Algeria. I carries an esimae of fuure quaniies of wase on he basis of daa from 48 deparmens from 1997 o We use economeric projecion of he wase o deermine he facors ha influence he producion of wase. The analysis shows ha he producion of municipal wase in Algeria is relaed o several facors: populaion densiy, he reail rade. The projecion of fuure municipal wase amoun achieves he 28 million ons in JEL : Q53, C33, C53, F18 Keywords: municipal wase, forecas wase, araciveness of he erriory, panel daa, Algeria. address:brahim.djemaci@gmail.com

3 Inroducion In recen years, Algeria has experienced srong economic growh ha exceeds 5%, he GDP per capia is increasing. Imporaion from he EU is progressing very remarkable. Following he various programs for economic recovery, several regions have recorded projecs including in urbanizaion wih he consrucion of over one million unis over he all erriory, he developmen of Small and medium firms in he conex of devices assisance o micro projecs, which sill wihin he naional agency o suppor youh employmen... In parallel wih his growh, he generaion of municipal solid wase has also experienced significan developmens. Evaluaed in 2005 o abou 8.5 million ons per year or on per day, and his producion is growing significanly. According o he MATE, he hreshold of 12 million ons of municipal solid wase is cerainly reached in In he sudy of he MATE on he sae of he environmen in Algeria indicaed ha he producion of wase per capia in urban areas increased from 0.76 kg / day in 1980 o 0.9 kg / day in 2002 (Meap 2004), o reach 1.2 kg / day on average in 2005 (MATE, 2005). On he oher side, here is a gap of almos 30% beween he rae of wase collecion in smaller owns and ciies. I is emphasized ha he coasal ciies of Algeria, denser populaion generae amoun of wase significanly higher han hose of Highlands and Deep Souh. As for he capial, she has produced over 0.87 million ons in A dozen ciies produce beween 200 and ones of MSW per year: his is he case of major ciies such as Oran in he wes, Consanine in he eas and Tizi Ouzou in he cener. Tweny medium owns produce beween 100 and ons per year. Finally, a some ciies produce lower amouns of MSW o ons per year, hey are generally concenraed in he far souh (Sahara) and are characerized by low populaion densiies by RGHP 2008.(ONS, 2008) Urban solid wase was resuling from he household consumpion, he public insiuions, he commercial premises and he firms. The amoun of wase produced varies from one ciy o 2

4 anoher in developing counries, according o several facors; he mos imporan is populaion growh. To illusrae, during 2007, esablishing Ne-Com has colleced of ons of wase (household garbage) in he 28 municipaliies in he Wilaya 2 of Algiers where he provides delivery. The principal quesion of his paper is o esimae he fuure producion of urban wase in Algeria if he same developmen programs recorded during he period 1994 o 2005 coninuing ino 2025? For his, we engaged an economeric analysis of facors ha can explain he producion of urban wase in Algeria? As a resul we make a projecion based on hese facors. We firs presen a summary of he economic lieraure devoed o quesions of facors ha may explain he wase flow. The objecive of his lieraure review is o deermine hese facors, and developed economeric models o produce a projecion of wase. Some of hese sudies are based on he assumpion of a proporional change beween wase generaion and he level of economic aciviy, he oher on he relaionship beween municipal wase generaion and privae consumpion. Models devoed o developing counries use he income and populaion as he mos imporan deerminans of municipal wase generaion. Nex, esablish he second poin by an empirical sudy devoed o develop an economeric model based on panel daa, which includes he producion of wase a he 48 Wilayas, and a se of daa on he araciveness of such erriory, he esablishmen of reail rade, wholesale, firms and populaion densiy. And finally we will conclude his sudy wih some remarks and suggesions. 1. Lieraure Review of Economerics projecion of wase The aricle "Fuure wase disposal in The Neherlands" of D. Nagelhou e al published in 1990 is considered a reference in he lieraure review of he forecas wase. The auhors have made an esimae of fuure amouns of wase in he Neherlands, and he impac of prevenion 2 Wilaya is a se of municipaliy (equivalen of deparmen) 3

5 on he various mehods of disposal and recovery. The sudy period exend from 1986, base year, from 2000, projecion year. The objec of he sudy is o presen some resuls of research conduced by RIVM 3 has developed scenarios for wase disposal in he fuure. Ten of 17 caegories of wase were idenified by he sudy concerned wih garbage, indusrial wase, consrucion wase, used car's ires, car end of life... ec.. To achieve he forecas amoun of wase, he auhors close he generaion of each caegory of wase o facors including consumpion and producion. Forecasing scenarios of economic growh have been made by he Cenral Planning Board for he period Srong growh wih a rae of 4%, he average rae (3%) and low growh a 1.4%. The model of D. Nagelhou e al (1990) is based on a simple regression. The explained variable is indicaed on he basis of 100, and each ype of wase is explained by well-based scenarios of growh. For example, he garbage were explained by he consumpion of food / drink, and oal consumpion. The households wase and assimilaes are linked o he consumpion of durable goods, commercial and indusrial wase are explained by he differen producion secors. The sudy resuls show ha in he case of low and high growh scenario, increased amous of wase from 1.2 o 2.2% per year, an increase of 15% in 2000 compared o 1986, and his in he absence of any improvemen in prevenion. However, applying o 13 prevenion approaches esablished by he deparmen, a small reducion in he volume of wase in 2000 compared o he siuaion in 1986 was recorded. However, a major change is regisered in he amouns of wase recycled or reused in he even of a change in environmenal policy. The second sudy is he Beede and Bloom (1995). The auhors sudied he producion and managemen of municipal solid wase (MSW) hrough economic objecives. The objec was o make projecions ino he fuure by involving he curren rends in he amoun of MSW and o evaluae he elasiciy of income and populaion wih he producion of MSW. They focused 3 Naional Insiue of Public Healh and Environmenal Proecion 4

6 heir analysis on developing counries, aking ino accoun he example of he Unied Saes. The analysis is based on economic reasoning. The MSW sudied were divided ino wo ypes, which can be reused wase (boles, newspapers...) and wase ha canno be reused (for wase healh risk). The explanaory variables used by he auhors are gross domesic produc (GDP / capia) and populaion. The dependen variable is he oal producion of MSW is esimaed a abou 1.3 billion ons in 1990 o 149 counries wih an average of 660 kg per capia per day. Several hypoheses have been assumed by he auhors on differen daa ses. The growh of naional GDP is assumed o be sable during he eighies, and populaion growh coninues o increase as projeced by he World Bank. The overall producion of MSW is expeced o double during 1990 o 2019 (ha is o say an average annual growh rae of 2.4%), he proporion of MSW generaion per capia will no double by The auhors esimaed he model by OLS, using he following regression: log(d) = α + β 0 log(pib hab ) + β 1 log(pop) + ε (1) Where (D) is he oal amoun of MSW ons, β 0 is he elasiciy of income generaion DSM β 1 is he elasiciy of populaion generaion MSW. Beede and Bloom have he example of Chinese ciies and U.S. Saes, which hey proceed o calculae he elasiciy s wih a model of cross-generaion rae of MSW wih he explanaory variables in 36 counries and in 45 Chinese ciies and 33 U.S. saes. The producion was realized by hree models, in he firs model (M1) Chinese ciies have been compared o U.S. saes, he second (M2) ranks counries by heir levels of wealh ino four groups (poor: GDP less han 600 $., below average income: 630 $ o 2490 $, above average income: 2490 $ 7050$, and higher incomes: more han 9550 $) by calculaing he proporion of each group in he overall producion MSW (million ons and percenage), and he size of heir populaions in he world populaion and he share of heir GDP in oal GDP. The hird model (M3) compares Taiwan and he Unied Saes. In each of hese models, he same ype of variables was used 5

7 (GDP and populaion). The resuls ha emerge from heir sudy show he exisence of a significan discrepancy beween he proporions of DSM indusrial counries in he world and deposi heir share of populaion in he world populaion. While in developing counries, he dispariy is refleced in he share of he oal MSW by MSW and he share of heir income o worldwide income. The counry analysis shows ha MSW generaion is posiively correlaed wih he change in per capia income, and ha oupu per capia does no vary wih he size of he populaion in counries wih comparable per capia incomes. The auhors esimae ha he producion of MSW is increasing a an annual rae of 2.7% in developing counries unil In he model (M1), a good relaionship exiss beween he producion of MSW o one side, and income and populaion on he oher side. The resuls show an increase of 1% of income per capia will generae an increase of 0.34% of oal DSM, and an increase of 1% of he populaion will increase overall producion from DSM of 1.04%. Model (M2) esimaes he wase in counries wih low income o 0.53 kg/inhab/day o 1.2 kg/inhab/day. On he basis of populaion, counries wih high income are disproporionaely represened in MSW (hese counries accoun for less han a sixh of he world's populaion, bu produce more han a quarer of oal MSW). On he oher side, on he basis of income, developing counries accoun for a disproporionae share of DSM (wih less han half of global GDP, hey produce hree quarers of oal MSW). Model (M3), presens he resuls for Taiwan and he Unied Saes. Regarding Taiwan, he oupu elasiciy of DSM o changes in income is However, i is 1.63 per populaion variaion. Where MSW generaion per capia does no vary wih he populaion in counries wih comparable incomes, he elasiciy of DSM wih income per capia rises, i goes o 0.72 (i is less han 1) bu i decreases wih he populaion, i is 1. For U.S elasiciy of DSM wih income are 0.86 and 0.63 for he populaion. On he assumpion ha MSW generaion per capia does no vary wih he populaion (aking he per capia income consan he elasiciy of DSM overall income fell o I is near ha 6

8 found in he model (M1). The overall conclusion of he model Beede and Bloom is ha MSW generaion is relaed o income as he populaion, bu is more linked o he populaion o income, ha is o say, is near a β 1 and β 0 is less han 1. Coopers and Lybrand (1996) conduced a sudy o make a projecion of municipal wase by he year 1997 o 2000 in he Neherlands, based on wo scenarios. The firs scenario forecass growh of municipal wase generaion. The second scenario includes prevenive measures ha will reduce he wase sream. The saring poin of he auhors is o assume a rend scenario, where amoun of wase are used as reference o evaluae prevenive measures o reduce wase. The model is based on an approach developed by RIVM o forecas growh in producion of household wase and bulky in he Neherlands. The RIVM model is based on he following assumpions, which are derived from a regression analysis of hisorical daa: - Increasing he amoun of household wase is proporional o he increase in privae consumpion of food and luxury goods. - The evoluion of bulky wase is relaed o changes in consumpion of durable goods. The auhors have aken a significan even, where he oal producion of municipal wase is relaed o he growh of oal privae consumpion. Therefore, he funcion of municipal wase is a linear funcion of household consumpion. The funcion is wrien: Q mw = f(cp) (2) The alernaive scenario includes wo addiional scenarios abou he effeciveness of prevenive policies due o various uncerainies in he daa. This uncerainy makes i difficul o predic he impac of hese policies on wase producion. Faced wih his siuaion, Coopers and Lybrand suggesed wo siuaions, reducion of wase wih high raes, 5% in 1997 and 10% in 2000 rend of he siuaion. The oher siuaion, he reducion is low, 2.5% in 1997 and 5% in 2000 compared o he baseline. This sudy highlighs ha he producion of municipal wase 7

9 will increase, bu ha he evaluaions are subjec o significan margins of error due o variaions in he qualiy and availabiliy of daa (he problem of heerogeneous daa). In 1997 and Bruvoll and Ibenhol (1997) develop a macroeconomic model of general equilibrium o esimae he fuure producion of indusrial wase in Norway. The auhors' objecive is o show he usefulness of an economic model based on he use of producion inpus (raw maerials...) and echnological progress o explain he fuure producion of indusrial wase indusry. The auhors used a general equilibrium economic model o address he mulidimensional problems of wase and which akes ino accoun echnological change, relaive prices of inpus and ineracion of differen secors. The empirical daa concern he producion of indusrial wases in 1993 ( housand ons), and includes paper, plasic, glass, exiles... ec.. Bruvoll and Ibenhol and assume ha he producion of wase is proporional o he oupu and inpu (raw maerial...) in each secor. These inpus may have wo purposes, a commodiy, or a residue (wase, polluion). So here is a relaionship beween he amoun of goods and he amoun of wase, increased producion of he produc involves an increase in wase. This relaionship can be changed if a change in echnology is inegraed. This increases he number of unis produced by keeping he wase producion consan. In oher words, i will reduce he proporion of wase and producion of goods. In general, he model is based on he assumpion ha he acual amouns of wase are proporional o he explanaory facors, and ha he proporionaliy coefficien is consan or exogenous in a given period. The model developed by he auhors is as follows: D ij () is he amoun of wase ype (j) produced in a secor (i) during he period () D ij () is given by he following funcion : D ij () = U ij () D ij ( 0 ) d ij () (3) Where U ij () represens he growh rae of he explanaory variables (value of oupu or inpu in he (i)), his variable depends on he ype of wase (j). d ij () is he parameer of evoluion 8

10 exogenous explanaory level wase (e.g, he effecs of policy measures ha influence he producion of wase). D ij ( 0 ) he amoun of wase ype (j) generaed in he (i) during a reference year, in our case ( 0 ) is 1993, he amoun supplied by he saisics. The oal amoun of wase ype (j) is he amoun of wase from all secors, given by he funcion: D j () = D ij () i (4) The reference year (0)=1988 ; he simulaion period is from 1988 o The model conains 33 secors and 48 producs. The auhors idenified he share of each secor in he producion of wase in he base year, and he growh of inensiy of he maerials. The resuls presened by Bruvoll and Ibenhol and show ha he reference rajecory used o perform a projecion of indusrial wase akes an average of 1% of echnological change in all secors of producion. This means ha he inpu demand per uni of oupu fell by 1% per year, all hings being equal. The effec of relaive prices of inpus and echnological change varies across secors. The increase in indusrial wase from 1994 o 2010 ranged from 45% o 110% depending on he ype of wase, while he oal increase is esimaed a 64% and indusrial hazardous wase o 58%. This increase exceeds he growh of commodiy producion, and is higher in he case of a domesic maer even wih echnological advances. In general, increasing wase sream would be included, depending on he ype of wase beween 35% o 60% by The projecion shows an increase in wase inensiy of 2.3% on average. For some wase ypes, his inensiy reaches 18% over he simulaion period, while i fell for hazardous wase. Increased inpus mainly explain he increase in wase and increase oal producion explains he increase of inpus o 52% over he simulaion period. Subsiuion beween producion facors conribuing o increased inpus. Technological change affecs he price of inpus. Conrary o he model and Bruvoll and Ibenhol, Anderson and his co-auhors (1998) have developed a simple model ha allows for fuure projecions of producion of household and indusrial wase. This model links he producion of various ypes of wase o various economic 9

11 aciviies in Denmark. The auhors defined he dependen variable (he amoun of wase divided by ype and source), and he explanaory variables which are he economic aciviies generaing he wase. Hisorical daa covering he period 1994 o In erms of household wase in 1996, ISAG has idenified million ons, accouning for much of he wase produced. These hree ypes include household wase, household wase, green wase and similar wase biodegradable. The basic assumpion of his model is ha wase generaion is relaed o he level of economic aciviy. I assumes ha he variaion is proporional beween he wase producs and he level of economic aciviy. The screening was performed by Eq.(4) which shows he relaionship beween he amouns of wase and he explanaory variables. I reads as follows: D f,s = D 0 f,s [1 + α f,s ( X f,s Xf,s 0 )]. β 0 f,s X f,s + φ f,s Ds f,s (4) D f,s where:, D 0 f,s is he amoun of wase of ype f, he source s in year, and he reference year 0. X f,s, X 0 f,s is he explanaory variables for he ype f, and he source s, year, and he reference year 0 Ds f,s is he amoun of addiional wase in addiion o he caegory in year, can be posiive or negaive. α f,s is he coefficien of proporionaliy beween changes in he amoun of wase and he explanaory variable. β f,s is he ime-dependen facor o explain changes in relaionships. For example, if D f,s households (source s), he explanaory variable X f,s nondurable goods., D 0 f,s are he amouns of wase paper / cardboard (f ype) of, X 0 f,s is he privae consumpion of As α f,s equal o 1, changes in he amouns of wase are proporional o changes in he explanaory variable. By cons when α f,s equal o 0.5, a 1% increase in he explanaory variable implies an increase of 0.5% of he amouns of wase. 10

12 β f,s is a series of ime coefficiens normalized o 1 in he reference year. If β f,s changed in some ime, hen he raio of wase is also changing (he raio beween he amoun of wase and he explanaory variable). The assumpion is α f,s = 1, he coefficien of Wase period, is he produc β f,s and he coefficien of he year 0 is o say: ( D f,s = β X f,s f,s D 0 f,s ). If β 0 f,s varies from 5% of he reference year 0 o year, he wase coefficien also varies from 5%. This variaion may be influenced by changes in policy on wase managemen, cusoms agens, packaging of goods. The amoun Ds f,s may be posiive or negaive; ofen he wase is ransferred from one caegory o anoher. The assumpions are: α f,s = 1.0; β f,s D f,s = 1 and φ f,s = 0, Eq.(4) will be reduced o: D f,s = c f,s X f,s (5) Where c f,s is a consan coefficien of wase on he base year. I is calculaed as: [ D 0 f,s ]. This model can be used o analyze how he economy affecs he fuure producion of wase. Connec he wase generaion in economic aciviies should be a a more deailed level. The assumpion of a consan coefficien of wase implies ha he amoun of wase follows economic developmen. Changing he composiion of producion and consumpion may change he proporionaliy a he aggregae level. In general, in he case of an increase of 10% of all economic aciviies, household wase are also increasing by 8% o 10% (ha is, on he assumpion ha he index of 1996 is 1 i is 1064 in 2005). The amoun of household wase increased from 2,741,200 ons in 1996 o 3,300,577 ons in A he same ime, a oal increase in wase generaion by 6.6%, 1.5% is aribued o he household. In addiion, if he privae household consumpion increases by 10%, he index of household wase increased from 1 in 1996 o 1,063 in A mehodology for projecing fuure wase has been developed by he EEA in 1999 following he repor on Europe's environmen published in In his repor he issue of wase D 0 f,s 11

13 producion in he EU Member Saes (15) was examined. I focuses on municipal wase (paper, glass, vehicle end of life). The saring poin for developing a model projecion of wase generaion in Europe was o creae an invenory of he siuaion regarding wase managemen (amoun, composiion, mehods of reamen, ype of wase, wase-producing indusries... ). The saemen was as follows: - The lack of comprehensive and reliable daa on wase, - No common paern has been developed for he screening of wase a he European level, - The exisence of a common feeling ha he producion of wase is relaed o he level of economic aciviy, - The effecs of naional policies on wase are no appreciaed, - The assessmens are subjec o a margin of error due o very significan variaions in he qualiy and availabiliy of daa. The model developed by he EEA is dispue models Bruvall and Ibenhol (1997); and Anderson e al (1998), which is based on he relaionship beween he level of economic aciviy and he amoun of wase produced. These models assume proporionaliy beween wo variables. Tha is, when economic aciviy increases by 10%, producion of wase increases by 10%, which is proporionaliy consan. The EEA aims o link he producion of wase a a level of economic aciviy in more deail. The saring poin is ha economic aciviy in some measures, may explain he producion of wase. Bu linking his producion o GDP is no a fair approach, given he origin of municipal wase, and he fac ha flucuaions in naional income do no necessarily affec consumpion. I is based on he explanaion of he producion of municipal wase and household goods by he share of naional income spen on privae consumpion. This creaes a lo of misakes, following he increase in oher expenses of consumpion (leisure, housing, energy), which will limi he share of consumer spending on producs ha generae wase. 12

14 The EEA model assumes ha he amoun of a ype of wase depends on a specific economic aciviy and ime. The funcion ha defines his relaionship is: D i = f(y i, T ) (6) Which, (D i ) is he amoun of a ype of wase (i), he period (), (Y i ) is he producion of a specific economic aciviy (subjec) expressed in moneary erms ha produces he ype of wase in he period (), and (T ) is ime. The funcion f is assumed o be a log-linear funcion, and is wrien: log(d i ) = a 0 + a 1 log(y i ) + a 2 T (7) I assumes ha he amoun of municipal wase and household changes proporionally wih he consumpion of goods. Three caegories of goods were deermined by he agency, food and drink, clohing, furniure and household equipmen. The amoun of wase have been esimaed by he following Eq.(8) based on Eq.(7) log-linear. log(d mm ) = a log(c alim + C habi + C equi ) + a 2 T (8) Where D mm is he amoun of municipal wase, household variables C alim, C habi, C equi represen he consumpion of food / drink, clohing and equipmen securiy. Given he lack of wase daa, he esimae used he following Eq.(9) based on Eq.(11) wih consan coefficien. = [ D mm D 0 mm (C alim +C habi +C equi ] (C alim ) + C habi + C equi ) (9) Where he big hug is he coefficien of Wase in he base year 0, and he sum of he caegories of privae consumpion is he explanaory variable. Assumpions relaed o he developmen of model are as follows: a 0 is a consan, a 1 is a coefficien of proporionaliy beween he amoun of wase generaed and he value of economic aciviy, a 2 is a endency o equal he annual% change in coefficien wase. The esimaion of hese coefficiens, based on hisorical observaions, is difficul because of insufficien daa and mulicollineariy. Therefore, 13

15 he model was simplified by assuming a 1 = 1, meaning ha he raio beween he amoun of wase produced and he producion of economic aciviy follows an exponenial rend. The coefficien a 2 was esimaed from hisorical observaions, which give he following equaion: log(d i ) log(y i ) = a 0 + a 2 T log [ D i Y ] = a 0 + a 2 T (10) i In case of unavailabiliy of daa, i is assumed a 2 = 0in Eqs. (7) and (8), he model is consan coefficien, hence e a 2 T = 1 and a 0 = 0, we obain he following equaion: [ D i Y i ] = a 0. e a 2 T (11) In pracice he coefficien a 0 is esimaed by calculaing he average excess amouns of wase over he period. In his sudy, only he las observed coefficiens were esimaed. I implies ha oher facors are consan over he forecas period. Two approaches have been developed o make projecions: he firs approach considers he daa of wase producion in recen years using he Eq.(8), and compare hisorical daa wih acual daa for hose years. If here is a good correlaion beween hisorical daa and real daa, he model esimaed by his equaion is reasonable and can be used o achieve he projecion. A good correlaion is achieved if a 2 is beween (-0.02) and (0.02), -suden is significan and he R 2 greaer han 0.6. The second approach is based on he consan coefficien model if he correlaion is weak. This involves using he values of he mos reliable hisorical daa o produce fuure projecions. Boh he above approaches have been developed for he ype of wase paper, cardboard and glass. In raison o lack of daa for Ausria and he Neherlands he projecion of wase was esimaed by Eq.(8). The resuls show ha equal and a 2 and ha increased amouns of wase reaches a oal of 55% and 74% respecively for he wo counries during he simulaion period from 1995 o 2010 (or 4110 housand ons and 12,480 housand ons), i should be noed ha his rae was 11% during he period 1990 o For he remaining counries (excep 14

16 Luxembourg) he consan coefficien model was used (Eq.9), he esimaed resuls for he same period from 1995 o 2010 showed a oal increase of 22% based on 11% from 1990 o 1995 (more han 191,454 housand ons in 2010). Oher models, he model of V. Karavezyris (2000) which saes ha he esimaed amoun of wase produced is done a wo levels, naional / inernaional or local / regional. The auhor analyzes he differen forms of model associaed wih he esimaion procedure. A he inernaional level, he projecion of wase is based on economic aciviies and ime. The explanaory variables are complee consumpion, inpus or oupus of producion. The model applied can be described as follows: log D i = α + βlogy i + γ (12) where D i is he amoun of wase produced by he caegory (i) he period (), Y i is he amoun of a specific economic aciviy expressed in moneary or physical, β is he consan coefficien raio wase producion and appropriae economic aciviy, and γ is he coefficien of a dynamic raio of wase generaion and producion of appropriae economic aciviy. Following he lack of reliable daa and significan margins of error in esimaes, he model is reduced o simples form: Y i = κy i (13) Where κ is he coefficien of wase from he base year. A he municipal / regional esimaes of fuure amouns and composiion of household wase are derived primarily as a funcion of demographics. The model can be represened as follows: D i = λ P i (14) Where λ is a consan erm and P is he populaion. 2. Empirical sudy 15

17 Algeria, an area of 2,381,741 km 2, consiss of 48 wilayas (deparmens), wih a 2008 populaion of abou 34.8 million inhabians spread over 1541 municipaliies. Algeria has a populaion densiy of inhabians per km 2. Abou 40% of Algerians are concenraed along he norhern coass of 2% of he oal area wih an average densiy of 300 inhabians per km 2. Thus, much of he counry, paricularly he Sahara is sparsely populaed. The populaion densiy varies widely beween coasal ciies and owns in he Souh. According o he census of 2008, Algeria has 5.76 million households or 71% in urban owns, 15.7% in secondary owns and 13% in sparse areas. The average size of households is 5.9 persons agains 6.6 people in 1998 and is 7.7 persons in nomadic households (ONS, 2008). In his approach we esimae he oal amoun of wase in Algeria aking ino accoun characerisics of araciveness of he erriories a he 48 wilaya of Algeria. The objecive is o deermine he facors ha influence he producion of wase in an area relaive o anoher o make a projecion for The daa was presened below Daa Presenaion Our endogenous variable is always he oal amoun of wase produced a each wilaya. Our daa come from esimaes by he naional agency of he wase, hey cover he period from 1997 o On exogenous variables, we incorporae ino he model he following variables: Densiy (DENS): The explanaory variable used in mos empirical sudies is he populaion. Here we use populaion densiy insead of populaion o measure he araciveness of he erriories a 48 wilayas. The daa on densiy were calculaed using he populaion census published by he Office for Naional Saisics and hey cover he period from 1997 o Five general censuses of populaion and housing (RGPH) 4 have been made since 1966, 1977, 1987, 1998 and Saisics beween he wo periods are esimaes made by NSO on he 4 hp://rgph2008.ons.dz/ 16

18 basis of he number of birhs and deahs each year in each wilaya. We expec ha his variable has a posiive influence in all regions. Small and medium firms (SMFs) and crafs (ART): The daa relaing o SMFs and crafs come from he Minisry of SMFs and crafs 5. They include wo ypes of SMEs; public and privae. The census number of SMEs is based on businesses regisered wih he Naional Social Insurance Fund (CNAS). The scope of he aciviy of privae SMEs is very wide, weny-wo secors are concerned wih he consrucion indusry and rade are he mos aracive. The number of SMFs per year akes ino accoun birhs, reacivaions and radiaion hereof o recognize ha he SMFs acive in he same year. In conras, public SMEs operae in 33 public secor (ourism, ranspor,...). As for he arisans, heir enumeraion shall be made wih he 31 Chambers of Crafs and Trades (CAM) and ake ino accoun he number of regisrans and he number of delised on 31 December each year. Three ypes of craf are concerned, craf goods producion, craf producion services and radiional crafs. These arisans are individual arisans, cooperaives or arisan enerprises. In our sudy, we focus on SMFs and Arisans daa disribued over he 48 wilayas counries. The inclusion of his variable in our model is jusified by doing ha hey are concerned by he TEOM and by heir producion of MSW including packaging of he raw maerial needed o produce finished producs. We expec ha his variable posiively influences he producion of MSW in he big ciies. Trader: Oher variables have been inroduced ino our model, hey relae o daa on numbers of regisered raders wihin he Cenre Naional du Regisre du Commerce (NRC). We disinguish wo caegories of business: wholesale rade (COM_GT) and rade in deail (COM_D) or legal persons or naural persons. The number of raders per year per wilaya is 5 hp:// 17

19 composed of he number regisered by deducing he radiaion o obain he acive raders. These wo caegories of raders end o influence he producion of MSW Méhodologie e modèle économérique The model used here is unexpeced by he model of Beede and Bloom (1995). We seek o esimae he impac of he economic araciveness of Wilaya on he producion of household solid wase (MSW). The general model is wrien as follows: Log(wase w ) = α + β 1 log (DENS w ) + β 2 log (PME w ) + β 3 log (COM_D w ) + β 1 log (COM_GT w ) + β 1 log (ART w ) + ε w Which, (wase w ) represens he amoun of wase produced by all residens, merchans, crafsmen and businesses in he wilaya (w) in year, (DENS w ) is he densiy of he populaion in each wilaya, (PME w ) is he number of small and medium companies operaing in he wilaya (w) in year, (COM_D w ), (COM_GT w ), are he number rader for deails, and large respecively, (ART w ) is he number of arisans. ε w is he error erm. The coefficiensβ i represen he elasiciy s of DENS, SMEs, COM_GT, COM_D, and ART producion of solid wase. Table 1 : The differen esimaion models LDENS LPME LCom_g LCom_d LAr ρ σ μ σ ν OLS (0.000)*** (0.726) (0.000)*** (0.000)*** (0.000)*** Wihin (0.000)*** (0.020)** (0.277) (0.000)*** (0.006)*** Beween (0.016)** (0.538) (0.000)*** (0.002)*** (0.009)*** Wallace e Hussain (0.002)*** (0.000)*** (0.012)** (0.000)*** (0.115) Wansbeek e Kapeyn (0.000)*** (0.003)*** (0.206) (0.000)*** (0.009)*** Swamy and Arora (0.000)*** (0.000)*** (0.000)*** (0.000)*** (0.499) MLE (0.005)*** (0.000)*** (0.012)** (0.000)*** (0.837) Table 1 summarizes he resuls of he coefficiens and heir p-value esimaed by he OLS models, Wihin, Beween. And hree oher evaluaions of GLS. The esimaor of Wallace and Hussain esimaor Wansbeek and Kapeyn and Swamy and Arora esimaor gives more of he 18

20 coefficiens of he variables values, σ μ, σ ν. Esimaes made for he esimaor under Eviews Swamy and Arora (1972), gives σ μ = 0,24, σ ν=0,08, ρ = σ μ2 (σ μ2 + σ ν2 ) = 0,89. The resuls of various models show ha he variable densiy (DENS) is a posiive sign significan a a hreshold of more han 99%. Variable (SMFs) is significan and posiive in Model Wihin a hreshold of 95% and in he hree random effecs models. However his variable is no significan in he OLS model and Beween. The reail rade (Com_d) o a posiive and significan in all models excep for he model and he esimaor Wihin Wansbeek and Kapeyn. Wholesale rade (Com_g) is significan and posiive in all models. Arisanal aciviy (Ar) is significan and posiive in models and in he Wihin esimaor Wansbeek and Kapeyn, and a negaive effec in he OLS and Beween. However, his variable is no significan in he remaining models. We presen in deail he esimae by OLS, hen he wo fixed effecs models and random effecs. Table 2: The OLS esimae Variable Coefficien Sd. Error -Saisic Prob. C LDENS LPME LCOM_GT LCOM_D LART R-squared Adjused R-squared Reading he resuls of he OLS model shows ha he probabiliy of Suden's maching variables are significan a 1% excep for he variable ha EPCA is no significan. Noe ha he variable densiy in a posiive coefficien, supporing he hypohesis fairly obvious ha populaion growh is posiively correlaed wih increasing amouns of wase. The assumpion of he commercial appeal is also suppored by he posiive coefficien. The resuls also show ha he araciveness scale impac negaively on he amoun of wase. The indusrial araciveness is no significan wih a negaive sign. The R2ajusée of he esimae is equal o

21 Regarding he fixed effecs model, he mos relevan is R2 R2 Wihin because i gives an idea on he par of inra-individual variabiliy of he dependen variable explained by hose variables. I is of In his model he variable reail which is no significan and his conrary o our predicion. Table 3 : The esimaed fixed effec model (Wihin) Coefficien Sd. Error -Saisic Prob. C LDENS LPME LCOM_D LCOM_GT LART The random effecs model gives he variable craf insignifican. Table4 : Esimaing a random effecs model Coefficien Sd. Error -Saisic Prob. C LDENS LPME LCOM_D LCOM_GT LART The economerics of panel daa requires wo ess, he firs es used o verify he exisence of significan specific effecs hrough he Fisher es. The second is used o make he choice of esimaor beween he fixed effecs model and random effecs model across he Hausman es. We es he firs hypohesis of he exisence of specific effecs: Y i = α + f i + βx i + μ i H0 : α 1 = α 2 = α n = α H1 : α 1 α 2 The Fischer es rejecs H0. 20

22 In he fixed effecs model, he probabiliy of he Fisher es is 0.000, so i validaes he presence of specific effecs in he model. The Hausman 6 es is o es he hypohesis ha αi is correlaed wih Xi: H0: αi is uncorrelaed wih Xi and herefore he GLS esimaor is consisen and effecive; H1: αi are correlaed wih Xi and herefore Wihin esimaor is preferable. Table 5 : Hausman Tes Chi-Sq. Tes Summary aisic Chi-Sq. d.f. Prob. Cross-secion random Cross-secion random effecs es comparisons: Variable Fixed Random Var(Diff.) Prob. LDENS LPME LCOM_D LCOM_GT LART Table 5 informs us ha he probabiliy of he Hausman es is less han 10%, so we rejec he null hypohesis. I indicaes ha he specific effecs are correlaed wih he explanaory variables, so he fixed effecs model is preferable o random effecs model. Table 6 : Tes of fixed effecs Effecs Tes Saisic d.f. Prob. Cross-secion F (47,465) Cross-secion Chi-square Period F (10,465) Period Chi-square Cross-Secion/Period F (57,465) Cross-Secion/Period Chi-square We evaluae he effecs of specifying fixed effecs model. Eviews esimaed hree specificaions: he effec period is fixed, he individual effec is hen fixed wih a common consan. The resuls show hree es secions. Each secion conains wo ess o es he significance of effecs. The firs wo ess measure he significance of individual effecs (F 6 To execue he Hausman es in Eviews, we mus firs esimae a random model, hen View / Fixed / Random Effecs Tesing / Correlaed Random Effecs - Hausman Tes. 21

23 Fisher and Chi-square). The wo saisical values are and and p-values respecively. We rejec he null hypohesis ha he effecs are redundan. The oher four values assess he significance of common period effecs and global effecs, respecively. All resuls sugges ha he effecs are saisically significan. Table 7 : Correlaion es errors Coefficien Sd. Error -Saisic Prob. RESID01(-1) R-squared Mean dependen var Adjused R-squared S.D. dependen var S.E. of regression Akaike info crierion Sum squared resid Schwarz crierion Log likelihood Hannan-Quinn crier Durbin-Wason sa Under he null hypohesis ha errors specific (idiosyncraic) are uncorrelaed, he residues of he funcion mus have an auocorrelaion coefficien of Here, we obain a ρ -1 = 0388 appears o be far from zero. The Wald es rejecs he null hypohesis of no correlaion of residuals. Table 8 : Wald Tes Tes Saisic Value df Probabiliy F-saisic (1, 479) Chi-square Null Hypohesis Summary: Normalized Resricion (= 0) Value Sd. Err C(1) Resricions are linear in coefficiens. If we rejec his hypohesis, i.e he errors are auo correlaed individuals. So we mus adjus our model o accoun for auocorrelaion errors. We run he fixed effecs model a second ime by including he mehod of Whie diagonal sandard errors & covariance under Eviews o consider he robusness of our model. The resuls are presened in he following able: 22

24 Tableau 9 : Whie correcion procedure model Coefficien Sd. Error -Saisic Prob. C LDENS LPME LCOM_D LCOM_GT LART Effecs Specificaion Cross-secion fixed (dummy variables) R-squared Adjused R-squared Resul Afer he esimaion of fixed effecs model, he resuls led o he following conclusions: firsly ha i has variable densiy (DENS) is significan. The elasiciy of he densiy of wase producion, however, is The amouns of wase produced are highly correlaed wih densiy. The densely populaed ciies end o produce more wase han ciies wih low densiy. This confirms he finding by he deparmen in which he raio per capia per day in ciies was 1.5 kg in medium ciies and 0.9 kg and less han 0.7 kg in Saharan ciies (MATE, 2005). I appears from he sudy ha he aciviy of reail rade (Com_d) is significan a 100% wih a posiive effec on wase producion and an elasiciy of The sale of producs in reail requires removing packages conaining hese producs (including packaging wase such as cardboard). These packages ha faciliae he ranspor of such goods or merchandise from he wholesaler or reailer are colleced direcly in he reail merchans. In conras, he packaging of he produc iself (boxes, canned,...) are generaed a he household level is ha can be explained by he variable populaion densiy. We also noe ha he craf (Ar) is significan a 95% wih a small posiive elasiciy of This low yield is caused largely by he craf service producion ha does no require he raw maerial ha may be likely o generae wase. The only caegory of wase is produced handicrafs producion and ha of many radiional handicrafs wih very small amouns for he number of crafsmen in each wilaya, which remains very low. 23

25 Unlike wha we expeced, he variable (SMFs) and (Com_G) are no significan. The explanaion we can give o he relaionship beween SME aciviy and producion of wase a each wilaya is ha his aciviy is characerized by he ype of aciviy performed by hese SMFs is generally no generae wase reaed as household wase so i has a sysem of collecion or disposal raher special (indusrial wase for indusrial, iner wase for consrucion aciviy, green wase from agriculural aciviies). On he wholesale (Com_g) has no relaion wih he amouns of solid wase produced. Among he explanaions ha can be given for his is ha he main role of he merchan and provide a link beween producers or imporers and raders in deail, he producs are sold in heir original condiion wihou any amendmen, and reail merchans and consumers bear he wase. 3. Projecions of wase amoun for 2025 Our projecions of he amoun of wase a each wilaya are based on assumpions ha he growh of all explanaory variables follows he same rend recorded during he period 1997 o 2007 o he year Projeced amouns of wase were performed using he command in Eviews Forecas 6. In a general model, Eviews calculaed for each case he fied value of Y using he parameers evaluaed and he corresponding values of X and Y: y = c (1) + c (2)x + c (3)z So he forecas made by he saic mehod, calculaes he resul of a sequence sep forward, using acual raher han prediced values for he dependen variables (delayed). Here we ignore he uncerainy coefficien of error ypes (sandar error). The model is wrien: y S+k = c (1) + c (2)x s+k + c (3)z s+k 24

26 The able shows he average annual prediced amouns of wase each year during he period 1997 o 2007 and The esimaed oal amoun of wase in 2007 was million onne, i will increase o million ons in To ge a beer reading of our resuls, he prediced amouns were reorganized ino hree groups, each represening a region (Norh Highlands, and Souh). Fig 1 : Projeced amouns of wase from norhern Wilayas The projeced amouns of wase in he norhern region o he year 2025 shows ha he wilaya of Algiers is he firs ciy in Algeria in he producion of solid wase wih a doubling compared o he amouns of 2007 or 3, 39 million ons in The second is afer Algiers Oran Wilaya wih nearly 2M ons. The norhern region will produce more han 45% of oal amoun ha can be domesically produced, is million ons. The high concenraion of populaion and commercial araciveness are he wo mos producive areas of wase in hese ciies. The increase in populaion causes a sharp increase in food consumpion. The developmen of new 25

27 echnology has pushed households o change heir old elecronic equipmen (TV,...). These producs are producs ha require packaging including cardboard. Figure 2: Projeced amouns of wase in he wilaya of Highlands The Highlands region by 2025 will produce more han 41% of he oal amoun of wase. The wilaya of Seif is he larges producer of wase wih 2.36 million on. Figure 3: Projeced amouns of wase in he Souh and large Souh 26

28 Wilayas locaed in he Souh or he deep souh of Algeria are he ciies generaing less solid wase. The share of his region is 13.36% compared o he oal amouns. The wilaya of Adrar on by iself produce more han 1.4 million ons in Conclusion Urban developmen is a very imporan elemen in economic developmen. This developmen requires wo essenial seps: firs, he developmen of land hrough urban developmen projecs (residenial, governmen, business, rade), secondly, provide he infrasrucure necessary o remove solid wase from he all acors in everyday life. We can expec a growing awareness of wase amouns a he level of araciveness. Wilayas will cerainly pay more aenion in he years o come o heir environmenal policy o address he increased amouns of wase. This increase will creae anoher problem is ha he capaciy of landfills ha an average life of beween 7 and 15 years wih a onnage of 100,000 ons per year. Among he policies adoped o address his phenomenon, i is he policy of source reducion 27

29 hrough soring, eco-design of packaging, reuse, ec... Anoher policy is he recycling of wase including plasic, paper, glass. In he absence of a policy of recycling wase from a household Algerian consis of all of hese maerials in addiion o organic wase. Inroduce he principle of exended producer responsibiliy is also needed in environmenal policy. Reference Beede, D.N., and D.E. Bloom, The economics of municipal solid wase. The World Bank research observer 10(2), Augus, Bruvoll, K. Ibnehol Fuure wase generaion. Forecass on he basis of a macroeconomic model. Resources, conservaion and recycling 19: Coopers & Lybrand, 1996: Cos-Benefi Analysis of he Differen Municipal Solid Wase Managemen Sysems: Objecives and Insrumens for he Year 2000, final repor for European Commission, DGXI. March European Environmen Agency Baseline projecions of seleced wase sreams. Developmen of a mehodology. Technical repo N 28. F.M. Andersen e al (1998) «A Scenario Model for he Generaion of Wase», Riso Naional Laboraory, DK Roskilde. Karavezyris, V Theoreical Approaches o Forecasing of Solid Wase, Paper presened a he 4S/EASST conference, Vienne/ MATE Analysis and recommendaions in covering of he coss of he managemen of he municipal wase in Algeria, repor realized by Erns and Young agency. Meap Regional projec of Managemen of he Solid Wase in he Counries of Mashreq and he Maghreb: repor of he counry Algeria. January, Nagelhou, D., Joosen, M., and K. Wieringa Fuure wase disposal in he Neherlands. Resources, Conservaion and Recycling, 4 : ONS. 2008, Recensemen général de la populaion e de l habia, en ligne hp://rgph2008.ons.dz/ Snigdha Chakrabari and Prasenji Sarkhel (2003), Economics of Solid Wase Managemen: A Survey of Exising Lieraure, Economic Research Uni, Indian Saisical Insiue. 28