The Food-Away-from-Home Consumption Expenditure Pattern in Egypt

Transcription

1 The Food-Away-from-Home Consumpton Expendture Pattern n Egypt Jacnto F. Fabosa Workng Paper 08-WP 474 July 2008 Center for Agrcultural and Rural Development Iowa State Unversty Ames, Iowa Jacnto Fabosa s a scentst n the Center for Agrcultural and Rural Development and co-drector of the Food and Agrcultural Polcy Research Insttute at Iowa State Unversty. Ths project was conducted durng the author s 2007 Fulbrght research grant n Egypt, from March to August of 2008, Caro, Egypt. Dr. Gamal Sam of Caro Unversty provded the data and suggested model specfcaton mprovements to an earler draft of the paper. Any remanng errors are the sole responsblty of the author. Ths paper s avalable onlne on the CARD Web ste: Permsson s granted to excerpt or quote ths nformaton wth approprate attrbuton to the authors. Questons or comments about the contents of ths paper should be drected to Jacnto Fabosa, 579 Heady Hall, Iowa State Unversty, Ames, Iowa ; Ph: (515) ; Fax: (515) ; E-mal: jfabosa@astate.edu. Iowa State Unversty does not dscrmnate on the bass of race, color, age, relgon, natonal orgn, sexual orentaton, gender dentty, sex, martal status, dsablty, or status as a U.S. veteran. Inqures can be drected to the Drector of Equal Opportunty and Dversty, 3680 Beardshear Hall, (515)

2 Abstract Ths study characterzes the household food-away-from-home (FAFH) expendture pattern n Egypt. Specfcally, a standard Tobt model was estmated to quantfy the responsveness of Egyptan household FAFH expendtures to changes n ther ncome and selected household demographc characterstcs. We found that the proporton of households wth a postve FAFH expendture s small, at 36% to 38% of the total number of households. These households spent 5% to 8% of ther total expendture on FAFH. Households that are located n urban areas, wth more famly members, and whose household head s young and male had generally hgher levels of FAFH expendture. The estmated condtonal ncome elastcty s only 0.02, and the uncondtonal ncome elastcty s 0.52, suggestng that most of the growth n ths sector wll be drven by new households partcpatng for the frst tme n FAFH expendtures. These elastcty estmates are relatvely low when compared to those of other countres. However, prelmnary estmates from more recent data seem to suggest a hgher ncome elastcty, whch s consstent wth the expanson of the sector of hotels, restaurants, and other nsttutons n Egypt. Keywords: condtonal and uncondtonal elastcty, demand, Egypt, food away from home, HRI (hotels, restaurants, and other nsttutons).

3 The Food-Away-from-Home Consumpton Expendture Pattern n Egypt 1. Introducton One common consumpton pattern that s shared by many countres as ther economes develop s the ncreasng sgnfcance of expendture on food away from home (FAFH) n consumers food basket. Ths emergng consumpton pattern can be of sgnfcant nterest to polcymakers for a number of reasons. Among others, although FAFH s generally taken as a result of economc growth, t can also serve as a dynamc stmulant n the development of ndustres such as those related to hotels, restaurants, and other nsttutons (HRI) n the food sector (Ma et al., 2006). Also, as the FAFH expendture share becomes more mportant n household budgets, possble dfferental nutrtonal content n food at home and food away from home may have mportant mplcatons for the detary status of populaton groups (Ln, Guthre, and Frazao, 1999). The emergence and growth of FAFH expendture can be drven by many factors, ncludng the greater proporton of women partcpatng n the labor market, whch rases the opportunty cost of home producton of food (Nayga and Capps, 1992). Ths s partcularly true n socetes n whch food preparaton s performed by the female members of the famly. Another contrbutng factor s the demand for greater varety by consumers. In both cases, the growth n demand for FAFH s drven by ncreases n household ncome. It s for ths reason that FAFH demand s usually assocated wth hgh responsveness to ncome and, as a result, the rapd rse n the proporton of ncome that s spent on FAFH reported n many countres such as the Unted States (Ln, Guthre, and 1

4 Frazao, 1999), Chna (Ma et al., 2006, Ba et al., 2008), Malaysa (Lee and Tan, 2006), and Span (Angulo, Gl, and Mur, 2002). The objectves of ths study are to (a) characterze the FAFH consumpton expendture pattern n Egypt; (b) dentfy factors that affect FAFH; and (c) estmate the responsveness of FAFH to changes n consumer ncome and selected household demographc characterstcs. 2. Model Household producton theory (Becker, 1965; Lancaster, 1971) has been used n many studes as the underlyng theoretcal bass for analyzng the demand for FAFH (Stewart et al., 2004). In ths case, households are maxmzng utlty n the consumpton of homeproduced goods subject to a household producton functon, tme constrant, and ncome constrant. The soluton to ths optmzaton problem gves the standard demand functons for market goods that are used as nputs n household producton. When multpled by ther respectve prces, these demand functons can be expressed n terms of expendtures. We follow ths approach n dervng the demand for FAFH n ths study. Snce n the survey data used n the study, many respondents may report zero FAFH expendture, we use a standard unvarate Tobt model to specfy the estmatng equaton for the demand model for the FAFH consumpton expendture. That s, [1] y x, * where y* s a latent varable representng household FAFH consumpton expendture; and x s a set of explanatory varables, whch n ths partcular case ncludes total expendture as a proxy for household ncome and demographc characterstcs ncludng household sze and household locaton (.e., urban or rural).βs a vector of parameters to be 2

5 estmated, andεs the ndependently dentcally dstrbuted error wth mean of zero and varance of sgma. Observatons of the FAFH consumpton s treated as censored snce many respondents n consumpton surveys report zero FAFH. Ths can be represented n the Tobt model as, [2] y * y f y 0. * 0 f y 0 The log-lkelhood functon (LLF) of ths model specfcaton s [3] ' ( y x ) ' x ln 1 ln { y 0} { y 0}, where the frst term of the LLF represents the lmt observatons, that s, a zero dependent varable (FAFH), whle the second term represents the postve FAFH observatons. Optmzaton of the LLF wth respect to the parameters wll gve the set of equatons that can be used to estmate the unknown parameter values. The condtonal value of y gven that the latent varable y*>0 s [4] * ' E( y y 0) x, ' ( x / ) ' ( x / ) whereφandφare the standard normal probablty densty and cumulatve densty functons, respectvely. It s the presence of the second term n equaton [4] that wll result n based estmates f the households wth zero FAFH observatons n the sample are not properly accounted for. The condtonal elastcty based on [4] s [5] * c E( y y 0) x * x E y y ( 0), whch s equal to 3

6 [6] 2 ' 2 c ( x / ) x 2. * E ( y y 0) The uncondtonal mean s [7] E y P y E y y P y E y y, * * * * ( ) ( 0)( ( 0) ( 0)( ( 0) whch when smplfed s equal to [8] E y x x x. ' ' ' ( ) ( / ) ( / ) The uncondtonal elastcty s computed from [8] and s gven n [9], [9] E ( y ) x x x. ( ) ( ) u ' x E y E y The dfference between the uncondtonal [9] and condtonal [6] elastcty can be nterpreted as an elastcty of partcpaton (Fabosa, 2005 and 2006), that s, the responsveness of households who currently do not have any spendng on FAFH to begn partcpatng n ths sector, or [10] p u c. 3. Data and Results Ths study uses the Egypt Integrated Household Survey (EIHS). The EIHS s a multtopc, natonally representatve household survey carred out by the Internatonal Food Polcy Research Insttute (IFPRI) n coordnaton wth the Mnstry of Agrculture and Land Reclamaton and the Mnstry of Trade and Supply of the Government of Egypt. The sample conssts of 2,500 households n 20 governorates, and the survey data collecton was conducted n Sample households were asked n the survey f anyone from the household consumed away from home durng the past seven days of the survey perod. If the answer was yes, 4

7 they were asked further how much the household members spent n total for regular meals ncludng breakfast, lunch, dnner, school meal, and other lght meal (e.g., bscuts, sandwches, etc.). The household FAFH expendture s defned as the sum of all expendtures assocated wth all these events. For the purpose of ths study, we used total expendture to proxy household ncome. Ths s the sum of food expendtures and nonfood expendtures. Food expendtures covered all food tems ncluded n the survey such as grans and cereals, pulses, eggs and mlk products, cookng ols, green leafy vegetables, fruts and nuts, fsh and meat, spces and condments, sweet and confectonary, and beverages. Non-food expendture ncluded housng, utltes, fuel, clothng, servces, and health. We also ncluded other demographc varables such as household sze and locaton of households, whether urban or rural. Of the 2,500 households ncluded n the survey, only 2,492 were used n ths study. Estmaton was done usng SAS verson 9.1 for Wndows. Of the total number of respondents, 46% of the household observatons used were located n urban areas and the other 54% were located n rural areas. Food expendture of households n rural areas represented 58% of total expendture and 46% for households n urban areas. A sgnfcant proporton of households n both urban and rural areas dd not report any FAFH expendture: 64%, of rural households and 62% of urban households (see table 1). Households that dd not have any FAFH expendture had ncomes that were only 84% to 85% of the ncome of households that reported postve FAFH expendture. Urban households that reported postve FAFH consumpton spent 1.7 tmes on FAFH compared to rural households. Ths expendture accounted for 7.9% of urban household total expendture, whch s much hgher compared to the 5.2% for rural 5

8 households. 1 The average household ncome of urban households that reported postve FAFH expendture was 1.14 tmes hgher than the ncome of rural households wth postve FAFH expendture. Table 2 shows that lunch s the most reported FAFH consumpton event, at 13%, n terms of the proporton of households that reported postve FAFH expendture. Ths s followed by breakfast, at 12%, and then dnner, at only 4%. In terms of the amount spent per week, lunch FAFH consumpton also reported the hghest amount spent, followed by dnner and then breakfast. Households also reported school meal and others as events for FAFH expendture. The urban and rural household FAFH consumpton expendture pattern s shown n table 2. Rural households had 8 to 9 percentage ponts more reportng zero FAFH expendture compared to urban households. For those reportng a postve FAFH expendture, urban households spent LE 31 per week more for dnner and lunch FAFH expendture compared to rural households, and LE19 per week more for school meals and other FAFH expendtures. Table 3 reports the standard Tobt model estmates for the FAFH expendture model. Wth the excepton of household sze and martal status of the head of household, the rest of the explanatory varables are sgnfcant at the 1% level. The sgns are also consstent wth expectatons. That s, ncome has a sgnfcant postve nfluence on FAFH expendture. The sze of the household also shows a postve nfluence on FAFH expendture, although not sgnfcant. That s, households wth more famly members 1 The proporton of FAFH to total household expendture s smaller n more recent household expendture survey data (3 to 4 percent for rural and urban households, respectvely), whch may suggest that the FAFH pattern has not changed sgnfcantly wth what s reflected n the EIHS dataset. 6

9 spent more on FAFH compared to households wth fewer members. 2 Moreover, urban locaton of households has a postve nfluence on FAFH expendture. Ths household locaton effect can be explaned from both the demand and supply sdes. That s, urban locaton s hghly assocated wth hgher ncome, whch has a postve nfluence on FAFH expendture on the demand sde. On the supply sde, there are more nsttutonal food supplers n urban locatons, so ths sector s larger n urban compared to rural areas. When households are headed by males who are younger, ths shows a postve nfluence on FAFH expendture. The martal status of the head of household s not sgnfcant. The estmated condtonal elastcty of FAFH expendture to ncome for households that are already spendng on FAFH s only 0.02, whch suggests that for every 1% ncrease n household ncome, FAFH expendture of households who are already spendng on FAFH ncreases by only 0.02%. The estmated uncondtonal elastcty s That s, for every 1% ncrease n household ncome, FAFH expendture of households ncreases by 0.52%. The dfference between the uncondtonal elastcty and condtonal elastcty can be nterpreted as the elastcty of partcpaton. In ths case, t s equal to That s, for every 1% ncrease n household ncome, FAFH expendture of households ncreases by 0.52%, 0.50% of whch accounts for households who are spendng on FAFH for the frst tme. Ths represents households who prevously had no spendng n the FAFH category who are now partcpatng n ths sector. These estmates show that the growth n ths sector s drven more by new households that are partcpatng for the frst tme than by the ncrease n FAFH expendture for households that are already spendng on FAFH. The share of the new partcpatng households n the 2 Other studes (Stewart et al., 2004) found an nverse relatonshp wth household sze because of some economes of scale ganed n home producton wth larger household sze. 7

10 total response s 96%, whle the share of the households that are already partcpatng s only 4%. The elastcty of FAFH expendture n Egypt s relatvely small compared to estmates n other countres (see table 4). 3 For example, Ma et al. (2006) reported an FAFH consumpton expendture elastcty wth respect to ncome for urban Chna of 0.21 to 0.77 for the condtonal elastcty and 1.02 to 2.54 for the uncondtonal elastcty. Ba et al. (2008) reported a condtonal elastcty of 1.38 and an uncondtonal elastcty of Angulo, Gl, and Mur (2002) reported an uncondtonal elastcty of 0.80 for Span. Stewart et al. (2004) reported a condtonal elastcty of 0.64 for the Unted States. 4. Summary and Conclusons Ths study characterzed the household FAFH expendture pattern n Egypt. We found that based on the 1997 IFPRI survey data, only 36% to 38% of households ntervewed reported postve FAFH expendtures. On average, household FAFH expendture accounted for 5% to 8% of total food expendture. In general, larger households and those located n urban areas spent more on FAFH. Lunch s the most mportant event assocated wth FAFH both n terms of the proporton of households reportng postve FAFH consumpton expendture as well as the amount spent per month. Ths s followed by breakfast n terms of the proporton of households wth postve FAFH expendture, and then dnner n terms of the amount spent. 3 A more recent household and expendture survey for the year 2004/05 has been conducted, but only hghly aggregated numbers are avalable to the publc. Offcal publcatons report an average HRI expendture and total household expendture for 20 ncome classes for urban and rural households. Usng these numbers n a double-log functon gves an FAFH ncome elastcty of The estmated functon has hghly sgnfcant explanatory varables and very good ft. Although valdty of the recent data s not guaranteed, new estmates suggest that the responsveness of household FAFH to changes n ncome may have ncreased. Ths s consstent wth the obvous expanson of the HRI sector n Egypt over tme. 8

11 The estmated condtonal ncome elastcty of FAFH expendture n Egypt s only 0.02 and the uncondtonal ncome elastcty of FAFH s 0.52, suggestng an ncome elastcty of partcpaton of Ths mples that the growth n the FAFH sector wll largely be drven by new households enterng nto the sector, spendng on FAFH for the frst tme. These elastcty estmates are relatvely low when compared to elastctes reported for other countres. However, estmates from more recent data seem to suggest that these elastcty estmates may have ncreased. 9

12 5. References Angulo, A.M., J.M. Gl, and J. Mur Spansh Demand for Food Away From Home: A Panel Data Approach. Presented at the European Agr-Food System, Zaragoza, Span. Ba, J., T.I. Wahl, B.T. Lohmar, and J. Huang The Determnants of Consumng Food Away from Home n Chna: The Role of Hosted Meals. eals.pdf Becker, G.S Theory of the Alocaton of Tme. Economc Journal 75: Fabosa, J.F Growng Demand for Anmal-Proten-Source Products n Indonesa: Trade Implcatons. CARD Workng Paper 05-WP 400, Center for Agrcultural and Rural Development, Iowa State Unversty. Fabosa, J.F Westernzaton of the Asan Det: The Case of Rsng Wheat Consumpton n Indonesa. CARD Workng Paper 06-WP 422, Center for Agrcultural and Rural Development, Iowa State Unversty. Lancaster, K Consumer Demand. New York: Columba Unversty Press. Lee, H., and A. Tan Determnants of Malaysan Household Expendtures of Food-Away-From-Home. Presented at the Internatonal Assocaton of Agrcultural Economsts Conference, Gold Coast, Australa. Ln, B-H., J. Guthre, and E. Frazao Nutrent Contrbuton of Food Away from Home. Chapter 12 namerca s Eatng Habts: Changes and Consequences, pp AIB-750, E. Frazao, edtor. Washngton, DC: U.S. Department of Agrculture-Economc Research Servce. Ma, H., J. Huang, F. Fuler, and S. Rozele, Getng Rch and Eatng Out: Consumpton of Food Away from Home n Urban Chna. Canadan Journal of Agrcultural Economcs 54: Nayga, R., and O. Capps, Jr Determnants of Food Away from Home Consumpton:Update. Agrbusness 8: Stewart, H., N. Blsard, S. Bhuyan, and R. M. Nayga, Jr., The Demand for Food Away from Home: Full-Servce or Fast Food? Agrcultural Economc Report Number 829 (electronc report), U.S. Department of Agrculture-Economc Research Servce. 10

13 Table 1. Data statstcs Household Locaton Data Urban Rural Number of observatons Lmt (zero) Observatons Number Percent Total expendture (LE/month) Means (for FAFH>0) Food away from home expendture (LE/month) Total expendture (LE/month) Percent Table 2. Food-away-from-home consumpton expendture by type of meal Lmt Observatons Postve Observatons Number Percent Number Expendture Breakfast Urban Rural 1, Lunch Urban Rural 1, Dnner Urban 1, Rural 1, School meal Urban 1, Rural 1, Others Urban Rural 1,

14 Table 3. Tobt estmaton results for food-away-from-home consumpton expendture Explanatory Varables Coeffcent Std Error t-rato Probablty Intercept <.0001 Total Expendture <.0001 Household Sze Urban Rural Locaton <.0001 Head of Household Gender (Male) Martal Status (Marred) Age Standard Error <.0001 Table 4. Comparson of ncome elastcty of food away from home Country / Study Partcpaton Condtonal Uncondtonal Egypt - Fabosa Chna (Urban) Ma et al Chna Ba et al Span Angulo, Gl, and Mur 0.80 U.S. (Full servce) Stewart et al