INCOME AND ECONOMIES OF SCALE EFFECT ON HOUSEHOLD FOOD DEMAND PATTERN IN PAKISTAN USING PSLM DATA

Academc Research Internatonal INCOME AND ECONOMIES OF SCALE EFFECT ON HOUSEHOLD FOOD DEMAND PATTERN IN PAKISTAN USING PSLM DATA Falak Sher Department of Economcs, Unversty of Sargodha, PAKISTAN. falak.sher@uos.edu.pk Dr. Nsar Ahmad Department of Economcs, Unversty of Sargodha, PAKISTAN. nsarahmad_25@hotmal.com Shreen Safdar Department of Economcs, Unversty of Sargodha, PAKISTAN. shreenjafr786.14@gmal.com ABSTRACT The study evaluates the household food demand patters for varous ncome groups n Pakstan namely: The Lower Income Group( 5000), Lower-mddle Income Group(5001-10000), Mddle Income Group(10001-15000), Upper-mddle Income Group (15001-20000) and Upper Income Group (20001+), respectvely. The study uses Pakstan Socal and Lvng Standard Measurement Survey (PSLM) 2007-08, data conducted by the Federal Bureau of Statstcs (FBS), Government of Pakstan, Islamabad. The food expendtures are the sum of expendtures by each household on the followng commodty groups: cereals, pulses, fruts, vegetables, dary products, chcken, meat, fsh, condments, sugar, edble ols, drnks and mscellaneous food products. The household sze and ncome elastctes are estmated to explan the food consumpton trends n Pakstan. The results of the study ndcate that all the ncome and household sze elastctes are postve and sgnfcant at one percent level of sgnfcance. Results of the study ndcate that food ncome elastcty s the hghest (0.896) for the Lower- Mddle Income Group and the same s the lowest (0.706) for Upper Income Group. Smlarly, household sze elastcty s the hghest (0.171) for the Upper Income Group and the same s the lowest (0.057) for Lower Income Group. Key Words: Households, Income Groups, PSLM, Income Elastctes, Sze Elastctes. INTRODUCTION Household demand s a combnaton of two mportant components household and demand. The household s defned as the person or a group of persons lvng n the same dwellng (Sheffrn, 2003). Ths s bascally the fundamental resdental unt n whch economc producton; consumpton, nhertance, chld rearng, and shelter are organzed and carred out by all members. Accordng to household surveys, household conssts of persons who share ktchen. A household may be ether a soltary beng or a mult-person household. On the other hand the demand refers to the ablty and wllngness of ndvduals to buy a partcular quantty of a good or a servce at a specfc prce durng a gven tme perod (Chaudhary, 2005). Ths s bascally the consumpton of varous types of goods and servces by the members of a famly n the socety whch s helpful for them to provde physcal and psychologcal well-beng and satsfacton. Consumpton of varous goods and servces s the reflecton of household expendture pattern whch are largely nfluenced by household composton, needs, taste and fnancal means. Everybody drects hs hghest nterest to the contnuous satsfacton of those wants that stem drectly from hs human nature, to the expanson of these wants and also to the attanment of the necessary ways to satsfy the hgher expanded wants. Economsts use the concept of utlty to defne the level of satsfacton or welfare that comes from a specfc allocaton of ncome among dfferent products. The progress and development of any economy s related wth the consumpton n a way that t measures the welfare of the people who are makng expendture on the purchase of varous consumpton heads. Then agan, t s helpful n the extenson of busness actvtes because the entre nvestment set up s dependent upon the consumpton pattern n the country. Due to the mportance of consumpton n the economc theory, a number of researcher have carred out research on household deeds n Pakstan and n other countres rangng from ts smplest form to very complcated by usng dfferent types of data and varety of econometrcs technques. Copyrght 2012 SAVAP Internatonal 50

Academc Research Internatonal Engel's law (1857) states that as ncome ncreases the consumer s spendng upon food s decreased. It also mples that the ncome elastcty of food demand les between zero and one. It means that ncrease n expendtures on food tem by the consumers s less than the ncrease n ncome of the consumers (Tmmer et al., 1983). A number of studes are accomplshed both for developed and developng countres that confrm the Engel s law regardng the household s behavor evaluaton. The valdty of the Engel s law regardng the consumpton patterns of the household have been tested by research studes. Western households n the Unted States consume less food energy, proten, ron, thamn and nacn but more calcum and vtamn A as compare to Southern households and household consumpton of vtamn A, vtamn C, and calcum s largely affected by ncome (Nayga, 1994). Functonal descrpton of Engel s Law s known as Engel curve whch descrbes that how household expendture on a partcular good or a servce vares wth change n total ncome or expendture. Budget share Engel curves depct how the share of household expendtures on a specfc good or servce changes wth varaton n ncome (Cha & Moneta, 2010a). Engel curve of a commodty reflects ts ncome elastcty and ndcates whether a partcular good s an nferor, normal or a luxury good (Cha & Moneta, 2010b). No establshed theory subssts that could explan the pragmatc shape of Engel curves and ther assocated ncome elastcty values. Ernst Engel hmself argues that households have a herarchy of wants that determnes the shape of Engel curves. As household ncome rses, some ncentves become more promnent as far as household expendtures are concerned that domnates consumpton patterns such as starvaton, ultmately become satsfed at hgher ncome levels (Wtt, 2001). The specfc objectves of the present study are numerous. Frst s to determne the household food demand pattern among dfferent ncome groups. Second s to measure the economes of scale effect n household consumpton by ncludng the household sze as an explanatory varable n Engel curve equaton. Thrd s to estmate Engel elastctes and fnally to suggest some polces. The remanng part of paper s organzed as follows: The revew of lterature s presented n secton 2. The descrpton of data s gven n secton 3. Secton 4 descrbes methodology. Results and dscusson are presented n secton 5. Concludng remarks are gven n secton 6. LITERATURE REVIEW The revew of lterature s helpful to obtan the base and n depth evaluaton of profcent knowledge and understandng of prevous research studes related to household consumpton trends that have been conducted so far both natonally as well as at nternatonally. Ths secton revews the theoretcal as well as emprcal lterature on the analyss of household demand system. Nayga (1994) nvestgates the mpact of household socoeconomc and demographc characterstcs upon the consumpton of 10 food nutrents n the Unted States. Ordnary Least Square (OLS) regresson s employed to make the consumpton analyss of vtamn A whle the demand pattern for remanng nutrents s evaluated by employng Weghted Least Square (WLS) technque. The mpact of soco-demographc factors on the household demand pattern s statstcally sgnfcant. It s further observed that Western households consume less food energy, proten, ron, thamn and nacn but more calcum and vtamn A as compared to Southern households. Household consumpton of vtamn A, vtamn C, and calcum s largely affected by ncome and ncome elastctes are relatvely small for poor as compared to rch households. Holcomb et al. (1995) study the household expendture pattern of total food consumpton at home (FAH) and food away from home (FAFH) n the Unted States. The Engel s law s verfed by employng four estmaton technques namely: the Workng-Leser Model, Sem-Logarthmc Model, Double-Logarthmc Model and Quadratc Expendture Model. The parameter estmates of weekly ncome for total food, food at home and food away from home are negatve and statstcally sgnfcant n Workng-Leser form whle are postve and statstcally sgnfcant n sem logarthmc, double logarthmc and quadratc functonal practces, respectvely. In all expendture categores, the ncome elastctes are less than one for each emprcal form that confrms the accuracy of Engel s law. The consumpton of food away from home s largely affected by ncome than the consumpton of food at home. All the household sze elastctes are postve and are comparatvely hgher for FAH rather than FAFH. Copyrght 2012 SAVAP Internatonal 51

Academc Research Internatonal Byrne and Capps (1996) conduct the consumpton analyss for food away from home for the Unted States households. The mpact of varous socoeconomc and demographc varables s also determned and margnal effects and elastctes are estmated by adoptng the nverse Mll s rato and Probt model technques. It s found that ncome s postvely related wth the household food consumpton away from the home. The trend of food consumpton away from home s most common n the Mdwest and South as compared to the Northeast. Households belongng to Whte group are more lkely to consume food away from home as compared to the Black households. Most of the expendture elastctes are postve and less than unty. Salvanes and Devoretz (1997) determne the household demand pattern for fsh and meat products n Canada. The mpact of household demographc characterstcs s also determned and the separablty test of meat and fsh commodtes are conducted, usng three demand models namely Model-1, Model- 2 and Model-3. Four commodty groups are ncluded n Model-I. Model-2 and Model-3 have sx commodty groups and eght commodty groups, respectvely. These demand models are estmated by usng the Lnear Approxmaton of An Almost Ideal Demand System (LA/AIDS). All the own prce elastctes are found to be negatve whle the cross prce elastctes are postve. Smlarly, all the expendture and cross prce effects are sgnfcant at one percent level of sgnfcance. Farooq et al. (1999) dscuss the farm household s response n terms of ther consumpton of sx food tems to changes n prces of food, ncome and the age composton of the household n Punjab Pakstan. Lnear Expendture System, An Almost Ideal Demand System (AIDS) and Complete Demand System are used for the estmaton purpose. It s observed that paddy and wheat are gross complements; pulses and meat are gross substtutes, dary products and meat are luxures and all own prce elastctes are negatve and sgnfcant. Ozer (2003) estmates the household demand pattern of sx major food groups n Turkey. The Lnear Expendture System (LES) s used to estmate the prce and expendture elastctes. The demand pattern s found to be expendture elastc for meat, fsh, poultry, tobacco, lquors and beverages whle expendture nelastc for the remanng food groups and all the expendture elastctes are found to be postve. The uncompensated cross-prce elastctes show that bread and cereals are gross complement whle the compensated cross-prce elastctes show that all the food groups are net substtutes. The ncome-compensated own-prce elastctes are lower n absolute terms as compared to uncompensated prce elastctes. Cross prce elastctes are low as compared to own prce elastctes ndcatng that consumers are more senstve to changes n own-prces of food tems. Babar and Shahnawaz (2010a) explan the changes n consumer s preference for meat group and ther per capta consumpton by usng the Lnear Approxmaton of An Almost Ideal Demand System (LA/AIDS) n Pakstan. All the own prce, cross prce and expendture elastctes are sgnfcant and reasonable n magntude. The uncompensated own prce elastctes are negatve for beef, chcken and fsh and are postve for mutton. The expendture elastctes of meat components are postve and less than unty and are comparatvely hgher for fsh as compared to other meat tems. It s also observed that beef & mutton; beef & chcken and beef & fsh are substtutes whle the mutton & fsh and chcken & fsh are complements due to the own and cross prce substtuton effects between them, respectvely. Manzoor et al. (2011) analyze the mpact of nflaton on the household consumpton pattern of 11 food and nonfood tems n Pakstan. It s concluded that household expendtures on food tems ncrease wth the ncrease n prce level and the same lowers ther access to basc socal facltes and luxures. The overall household consumpton behavor s sgnfcantly affected by the current nflaton stuaton. DATA The data for ths study s drawn from the Pakstan Socal and Lvng Standards Measurement (PSLM) Survey 2007-08, conducted by the Federal Bureau of Statstcs (FBS), Government of Pakstan, Islamabad. It s based upon two-stage stratfed samplng desgn. Ths survey, based on a natonal sample, covers the unverse consstng of all urban and rural areas of the four provnces of Pakstan apart from forces restrcted areas. Ths study uses a sample of 15495 households out of total 15512 households covered by the PSLM 2007-08 due to unreported and mssng values for 17 households. Copyrght 2012 SAVAP Internatonal 52

Academc Research Internatonal The expendtures n the form of only pad consumpton are used for computaton because the goal of study s to compare consumpton patterns wthn varous earnng sets. The food consumpton groups are as: cereals, pulses, fruts, vegetables, dary products, chcken, meat, fsh, condments, sugar, edble ols, drnks and mscellaneous food products. Frequency of food tems data n PSLM 2007-08 s of two types.e. fortnghtly and monthly. The 14 days data s frst converted nto monthly nformaton and then both of these groups are joned to make the household total food consumpton durng the month. Thus the dependent varable s the natural log of the monthly expendtures on food tems. The total household expendtures are used as a proxy for ncome as an explanatory varable because of the fact that ncome data generally suffers from measurement errors and may also nclude a transtory component of ncome (Burney & Khan, 1991). The use of total expendtures nstead of ncome s a common practce n Engel curves estmaton because the expendtures mostly reflect the permanent ncome of the households. Household total expendtures and household sze are computed n the form of natural log and then are used as an explanatory varable n ths regresson analyss. Household sze ndcates the numbers of persons lvng n the sngle house. Ths varable s computed n the natural log form of total famly sze of the household. Havng certan advantages, the famly sze s used as a separate ndependent varable. The same s valuable to drectly determne the economes of scale effect, avods the loss of nformaton problem and gves more effcent results regardng the household members (Malk & Sarwar, 1993). Table 1: Households Dstrbuton among Dfferent Income Groups Income Groups (Monthly Income) Number of Households Percentage of Households 5000 3577 23.08 (Lower IG) 5001-10000 6095 39.34 (Lower-Mddle IG) 10001-15000 2683 17.32 (Mddle IG) 15001-20000 1217 7.85 (Upper-Mddle IG) 20001+ 1923 12.41 (Upper IG) Total 15495 100 IG = Income group Source: Author s Calculatons usng PSLM 2007-08 data. In order to determne the food demand pattern and to make the consumpton comparson; households are dvded nto fve ncome groups. Table 1 shows the household dstrbuton wth respect to dfferent ncome groups at the natonal level, measured n Pakstan Rupee. Income groups nclude: The Lower Income Group ( 5000), Lower-mddle Income Group(5001-10000), Mddle Income Group(10001-15000), Upper-mddle Income Group (15001-20000) and Upper Income Group (20001+), respectvely. METHODOLOGY For computng ncome elastctes, the method of Ordnary Least Squares (OLS) s used on the bass of both common practce as well as convenence. The cross secton data s based on the assumpton that all the households face the same prces for every commodty so t s dffcult to capture the mpact of prce varaton on consumpton pattern. The prce effects can be determned by usng the unt values, calculated as the expendtures of consumpton tems dvded by the quanttes of tems used as a proxy for prces (Tokoyama et al., 2002). The economc theory of consumer behavor s based upon the assumpton that consumer attempts to maxmze the utlty of goods and servces subject to the gven budget constrant. Consumpton of varous commodtes are usually taken n terms of expendture rather than quanttes n the Engel s curve approxmaton because the expendture takes much care of Copyrght 2012 SAVAP Internatonal 53

Academc Research Internatonal the quantty and the qualty of the goods consumed to confscate the problem of aggregaton of heterogeneous tems (Burney & Khan, 1991). It s further assumed for cross sectonal data that all the households face dentcal prces for all commodtes. So the Engel curve equaton can be wrtten as: E Y (1) Where E s quantty demanded or expendture on th consumpton head by th household, Y s total ncome of the household, and β are the unknown parameters to be estmated and s stochastc error term. The choce of an approprate functonal form n estmatng the Engel s curve has been a matter of great nterest. There are so many functonal forms that are used to estmate the Engel curves but yet no sngle form found ts general acceptance (Islam and Swar, 2005). Thus by ncorporatng total expendture and household sze as explanatory varables n Engel curve equaton and takng the natural log the functonal form gves as follows: Where ln F ln E ln HS (2) F s food expendture by th household, sze and s the stochastc error term. RESULTS AND DISCUSSION E s total expendture by th household, The computed food expendture elastctes at natonal level are gven n Table 2. Table 2: Food Income Elastctes by Dfferent Income Groups Income Groups Total Expendtures Ln (TE) Overall 0.778* Lower IG 0.876* Lower-Mddle IG 0.896* Mddle IG 0.840* Upper-Mddle IG 0.814* Upper IG 0.706* IG = Income Group. Dependent varable: Natural log of monthly food consumpton expendtures (LnF). * Imply that the coeffcents are sgnfcant at 1% Level. HS s household All the expendture elastctes are less than one and sgnfcant at one percent level of sgnfcance mplyng that all food tems nclude n the analyss are necesstes n the whole computaton. Table 2 shows that food ncome elastcty s the hghest (0.896) for the Lower-Mddle ncome group and the same s the lowest (0.706) for Upper ncome group. Overall analyss shows that the food consumpton ntally ncreases wth ncrease n ncome but gradually declnes for hgher ncome groups. The expendture elastctes show a lttle cyclcal trend of elastctes n dfferent ncome groups. The reasonng for ths pattern may be explaned n terms of quanttatve as well as qualtatve changes n terms of consumpton basket. For a gven qualty of a commodty, the mmedate concern of a household s to consume the commodty up to a certan mnmum desred level. If the household are not consumng the sad commodty n the desred mnmum amount then the expendture on that commodty ncrease wth the ncreased level of ncome. Once the desred quantty s acheved the expendture on that commodty declne wth ncrease n ncome. As ncome contnue to ncrease the Copyrght 2012 SAVAP Internatonal 54

Academc Research Internatonal household swtch to a better qualty of products thus expendtures starts ncreasng agan. Ths pattern s repeated as ncome contnues to ncrease. The household sze s used also as an explanatory varable n order to nvestgate the economes of scale effect n the food demand pattern for varous ncome groups. The computed household sze elastctes at natonal level are gven n Table 3. Table 3 shows that household sze elastcty s the hghest (0.171) for the upper ncome group and the same s the lowest (0.057) for lower ncome group. Economes of scale effect may occur because the publc goods wthn the household can be shared and the same serve ther functon wthout needng to be replcated n relaton to the number of ndvduals wthn the household. Larger households may receve quantty dscounts because they buy larger quanttes. The sze elastctes are also sgnfcant at one percent level and depct the specfc effect n the food demand patterns analyss. Table 3: Household Sze Elastctes of Food by Dfferent Income Groups Income Groups Household sze Ln (HS) Overall 0.127* Lower IG 0.057* Lower-Mddle IG 0.075* Mddle IG 0.144* Upper-Mddle IG 0.147* Upper IG 0.171* IG = Income Group. Dependent varable: Natural log of monthly food consumpton expendtures (LnF). * Imply that the coeffcents are sgnfcant at 1% Level. CONCLUDING REMARKS The study evaluates the double logarthmc analyss to determne the household food demand pattern n Pakstan. The data for ths purpose s drawn from Pakstan Socal and Lvng Standards Measurement Survey (2007-08). The households are dvded nto fve ncome groups. All the coeffcents of ncome and household sze elastctes are postve and sgnfcant at one percent level of sgnfcance. The legtmacy of Engel s law s verfed because the proporton of food consumpton s lower as compared to ncome. The household sze analyss confrms the exstence of economes of scale for food consumpton among numerous ncome classfcatons. The plannng for future ncrease n food supply s mostly dependent upon accurate predcton of future demand. As the estmates presented n ths study are based on recent household level data, so they are much benefcal to predct the true consumpton pattern accurately. In partcular, ncome specfc estmates are helpful to provde a better understandng of changes n food consumpton pattern by enablng polcy makers to focus on households n dfferent ncome groups. Copyrght 2012 SAVAP Internatonal 55

Academc Research Internatonal REFERENCES Babar, A., & Shahnawaz, M. (2010a). Surmsng Consumer Demand System & Structural Changes Usng Tme Seres Data. Munch Personal RePEc Archve, 1-25. Burney, N. A., & Khan, A. H. (1991). Household Consumpton Patterns n Pakstan: An Urban Rural Comparson usng Mcro Data. The Pakstan Development Revew, 30, 145-171. Byrne, P. J., & Capps, O. J. (1996). Does Engel's Law Extend to Food Away from Home? Journal of Food Dstrbuton Research, 22-32. Cha, A., & Moneta, A. (2010a). Retrospectves Engel Curves. Journal of Economc Perspectves, 24(1), 225 240. Cha, A., & Moneta, A. (2010b). The evoluton of Engel curves and ts mplcatons for structural change. Dscusson Paper, Grffth Busness School. Chaudhry, M. H. (2005). Economc Theory: Mathematcal and Mcroeconomcs. The Caravan Press. Darbar Market, Lahore. Farooq, U., Young, T., & Iqbal, M. (1999). An Investgaton nto the Farm Households Consumpton Patterns n Punjab, Pakstan. The Pakstan Development Revew, 38(3), 293-305. Government of Pakstan (2008). Pakstan Socal and Lvng Standard Measurement Survey (PSLM) 2007-08. Federal Bureau of Statstcs. Holcomb, R. B., Park, J. l., & Capps, O. J. (1995). Revstng Engel's Law: Examnng Expendture Patterns for Food at Home and Away From Home. Journal of Food Dstrbuton Research, 1-8. Islam, M., & Swar C. (2005). Impact of the Fnancal Crss on Expendture Patterns n Malaysa: Specal Reference on Low-Income Households. Journal of Economc Research, 10, 145 173. Malk, S. J., & Sarwar, N. (1993). Some Tests for Dfference n Consumpton patterns: The Impact of Remttances usng Household Income and Expendture Survey Data of Pakstan 1987-88. The Pakstan Development Revew, 32(4), 699-711. Manzoor, M. M., Sddque, A. A., Sattar, A., Fahm, M., & Rasheed, S. (2011). Impact of Inflaton on Household Consumpton-A Case of Pakstan. Internatonal Research Journal of Fnance and Economcs, 68, 162-167. Nayga, M. R. J. (1994). Effects of Socoeconomc and Demographc Factors on Consumpton of Selected Food Nutrents. Agrculture and Resource Economcs Revew, 171-182. Ozer, H. (2003). Consumpton Patterns of Major Food Items n Turkey. The Pakstan Development Revew, 42(1), 29-40. Salvanes, K, G., & Devoretz, D. J. (1997). Household Demand for Fsh and Meat Products: Separablty & Demographc Effects. Marne Resource Economcs, 12, 37-55. Sheffrn, S. M. (2003). Economcs: Prncples n acton. Upper Saddle Rver, New Jersey 07458: Pearson Prentce Hall, 29. Tmmer, C. P., Falcon, W. P., & Pearson, S. R. (1983). Food polcy Analyss. Johns Hopkns Unversty Press. Baltmore. Tokoyama, Y., Takag, S., Ishbash, K., & Chern, W. S. (2002). Recent Food Consumpton Pattern of Japanese Households Drvng Forces behnd Westernzaton. Selected paper, Annual Meetng of the Amercan Agrculture Economcs Assocaton. Wtt, U. (2001). Learnng to consume. A theory of wants and the growth of demand. Journal of Evolutonary Economcs, 11(1), 23-36. Copyrght 2012 SAVAP Internatonal 56