Demand Analysis of Non-Alcoholic Beverages in Japan

Transcription

1 Journal of Agrcultural Scence; Vol. 7, No. 5; 2015 ISSN E-ISSN Publshed by Canadan Center of Scence and Educaton Demand Analyss of Non-Alcoholc Beverages n Japan Mchael Fesseha Yohannes 1 & Toshnobu Matsuda 2 1 Unted Graduate School of Agrcultural Scences, Tottor Unversty, Japan 2 Faculty of Agrculture, Tottor Unversty, Japan Correspondence: Toshnobu Matsuda, Faculty of Agrculture, Tottor Unversty, Japan. E-mal: matsudat@muses.tottor-u.ac.p Receved: February 12, 2015 Accepted: March 15, 2015 Onlne Publshed: Aprl 15, 2015 do: /as.v7n5p143 URL: Abstract Ths paper estmates the demand of non-alcoholc s n Japanese household usng the lnear approxmaton quadratc almost deal demand system model (LA/QUAIDS). Eght expendture shares and prces demand equatons for non-alcoholc group: green tea, black tea, tea, coffee, coffee, frut and vegetable uce, carbonated and mlk are estmated for two or more households n forty-nne ctes for the perod January 2000 through March The expendture elastcty results ndcate that green tea (2.320), black tea (1.357), coffee (1.090) and frut and vegetable uce (1.019) are luxury goods whle tea (0.836), coffee (0.896), carbonated (0.844) and mlk (0.677) are necesstes n the Japanese household. The demographc effects reveal that people under the age of 18 prefer mlk (5.928) than any other s whereas elderly people tend to drnk more green tea (24.427). Moreover, temperature effects shows t has mostly postve effect on demand for tea, coffee, frut and vegetable uce, and carbonated and negatve effect on green tea, black tea, and coffee n most of the months. Keywords: elastcty, Japanese household, LA/QUAIDS, non-alcoholc, temperature effects 1. Introducton The global non-alcoholc ndustry s one of the bggest n the world wth market sze of bllon dollars n 2013 (Euromontor, 2014). Japan s among the man market n the non-alcoholc market wth sales revenue of 31.3 bllon dollars n 2013 (Japan Soft Drnk Assocaton, 2014). In the Japanese non-alcoholc ndustry, tea holds the largest market segment (wth 5.4 bllon lters produced n 2013) and s followed by carbonated (3.6 bllon lters n 2013), flud mlk (3.0 bllon lters n 2013), coffee (2.9 bllon lters n 2013), mneral water (2.8 bllon lters n 2013), frut and vegetable uce (1.7 bllon lters n 2013), and sports (1.6 bllon lters n 2013). Other smaller categores s such as lactc (496.0 mllon lters n 2013) and soy mlk (164.9 mllon lters n 2013) comprse the remanng share of the group (Japan Soft Drnk Assocaton, 2014). In terms of largest market sales globally, Japan has the thrd largest demand for frut and vegetable uce after the Unted States and Chna wth net sales of 10.4 bllon dollars n However, ts per capta consumpton s lower than many natons at 18 lters (Euromontor, 2012). Smlarly, Japan s consumpton per capta for carbonated s also relatvely smaller at 28.6 lters n 2013 (Japan Soft Drnk Assocaton, 2014). Durng the same year, the Unted States per capta s fve tmes more than Japan s at lters (Statsta, 2014). Even though, demand for both tea and coffee s sgnfcant n the Japanese non-alcoholc ndustry, ts per capta consumpton has remaned relatvely the same snce the year 2000 at 1.0 lter and 3.4 lters per person respectvely (Helg Lbrary, 2011, ICO). As for the demand of flud mlk, ts per capta consumpton shows t has been on a declne for the past decade from 39.0 lters n 2000 to 31.8 lters by 2010 (Mnstry of Agrculture, Forestry and Fsheres of Japan). Watanabe and Suzuk (2006) confrmed ths valdty n ther analyss of perceptons and atttudes concernng mlk consumpton stressng lttle can be done to ncrease the demand of mlk. Moreover, Schluep Campo and Beghn (2005) ponted out that mlk consumpton s regular n 80 percent of Japanese household and by 90 percent of the total populaton, as 4 out 5 Japanese are aware of ts hgh nutrtonal qualty. There haves been qute few studes that analyzes the demand for non-alcoholc s n dfferent countres over the years: Knnucan, Mao, Xao, and Kaser (2001), Yen, Ln, Smallwood, and Andrews (2004), Pofahl, Capps Jr, and Clauson (2005), Zheng and Kaser (2008), and Dharmasena, Capps Jr, and Clauson (2009) to name 143

2 a few have all analyzed the demand pattern of non-alcoholc s n the Unted States usng annual tme seres data. The numbers of prevous studes on demand of non-alcoholc s n Japan however are very few. Nonetheless, there are a number of studes that focuses manly on mlk demand. Some of the relevant studes: Watanabe, Suzuk, and Kaser (1997) examned consumer s preferences of mlk and other s usng Japanese consumer data and Quantfcaton theory type III. Ther results ndcate that mddle-aged people prefer soda and alcoholc s to mlk s whle younger people prefer mlk. Watanabe, Suzuk, and Kaser (1998) usng a logt model analyzed the consumpton of mlk for nne s n Japan. Ther results concluded that many of the consumers attrbute had a maor relaton wth the consumpton of other s. Stroppana, Rethmuller, and Kobayash (1998) analyzed the consumpton of mlk n nne Japanese regons. Ther fndng ndcates that structural factors such as consumer age and sze of the household have an effect on the consumpton of mlk at regonal and natonal level n Japan. Watanabe et al. (2006) estmated demand for mlk and other detary products usng Heckman s two-step estmators of the type II Tobt model. Ther result shows that hgher estmaton on mlk utlty and hgher health concerns are sgnfcant determnant for mlk consumpton. Most of the studes mentoned are related manly on mlk and do not concentrate on other s n the Japanese household. Moreover, they do not ncorporate a complete demand model n addton to examnng a wde range of factors and varetes. Our study apples a comprehensve demand model for eght expendture shares and prces n the Japanese non-alcoholc s household. Snce Japan s one of the bggest producng and consumng naton of non-alcoholc s, ths study could be a useful tool for manufacturers as well as for polcy makers. Regardng consumpton of non-alcoholc s, there are several factors that could nfluence ts demand pattern. Some of these factors nclude expendture, prce, soco-demographc varables, and temperature. Temperature s an mportant feature of consumer behavor. There have been qute few studes that analyzed temperature effects on consumer demand, to menton some: Harrson (1992), Agnew and Palutkof (1999), Parker and Tavasoll (2000), Roslow, L, and Ncholls (2000), Murray, D Muro, Fnn, and Popkowsk Leszczyc (2010), and Bahng and Kncade (2012). However, these studes do not examne the demand for non-alcoholc s wth regards to temperature effects. Our study, usng the LA/QUAIDS model, attempts to analyze the demand for non-alcoholc s n Japanese household. To attan the analyss of ths study, we ontly estmate eght demand equatons takng nto account the effects of demographc and temperature by evaluatng the monthly cty data for green tea, black tea, tea, coffee, coffee, frut and vegetable uce, carbonated, and mlk. In addton, to a general concern about demand of non-alcoholc s n Japanese household, t s a great nterest to determne whether the s are necessty, luxury, or an nferor good. Accordng to our knowledge, the novelty of the study s t s the frst study to analyze temperature effect on non-alcoholc s demand n Japanese household. The rest of the paper s drawn as the followng, a dscusson of the data set and model, followed by a dscusson of the results and lastly the concluson. 2. Data and Model Monthly aggregate pseudo panel data, whch s repeated cross sectonal data, s used n ths study for two or more person households from January 2000 to March 2013 for 49 ctes, attanted from the Famly Income and Expendture Survey (FIES) of Japan, conducted by the Mnstry of Internal Affars. Pseudo panel data, suggested by Deaton (1985), s an alternatve econometrc method for estmatng demand models of ndvdual behavor. One of the advantages of pseudo panel data over panel data s that t allows us to estmate models over a longer perod of tme. In addton t elmnates ndvdual-level measurement error. The use of aggregate pseudo panel data enables us to capture the mpact of the mportant demographc and temperature varable affectng non-alcoholc s patterns n Japan. Consderng the changes n prces, all expendture data for green tea, black tea, tea, coffee, coffee, frut and vegetable, carbonated and mlk were deflated usng the consumer prce ndex. To conduct the estmaton process, we used the SHAZAM (verson 10.2) econometrcs software. A sample sze of 7789 s ncluded n ths study. The teratve seemngly unrelated regresson (ISUR) was used to estmate the lnear system of eght equatons. To ensure non-sngularty of the error covarance matrx, we deleted the 8th equaton for mlk. We apply two types of dummy varables nto the demand system: monthly dummy varables to adust the monthly varaton and cty dummy varables to capture the cty varaton. Several demographc varables from the FIES were ncorporated n the study to understand the demand patterns of non-alcoholc s n the Japanese households. These 144

3 varables nclude sze (household sze), under (number of persons per household under the aged of 18), elders (number of persons per household over the age of 65), age (age of the head), and rent (rate of those payng rent). The study doesn t nclude varables such as gender and race snce those data are not avalable at the FIES. To assess the temperature effects, monthly data from January 2000 to March 2013 were outsourced from Japan Meteorologcal Agency. The quadratc almost deal demand system (QUAIDS) model, whch was developed from utlty maxmzaton by Banks, Blundell, and Lewbel (1997), not only holds the desrable propertes of Deaton and Muellbauer s (1980) almost deal demand system (AIDS) model but also s more versatle n modelng consumer expendture patterns. The QUAIDS model gves rse to quadratc logarthmc Engel curves, that s, allowng such crcumstances where ncremental n expendture would change a luxury to a necessty, whereas for the AIDS model, t gves rse to Engel curves that are lnear n logarthm of total expendture. Moreover, for QUAIDS model, the expendture elastcty depends on the level for expendture whle for the AIDS model the elastctes are not dependent of expendture level (Banks et al., 1997). The recent appled studes on QUAIDS are presented n Matsuda (2006). For our study purpose, we employ the lnear verson of QUAIDS (LA/QUAIDS) model by Matsuda (2006). We selected ths specfc model, LA/QUAIDS, because t features the characterstcs of Closure Under Unt Scalng (CUUS) even wth demand shfters such as demographc varables and monthly and cty dummes (Alston, Chalfant, & Pggott, 2001). Wth demand shfters, the orgnal QUAIDS of Banks et al. (1997) does not fulfll CUUS. Accordng to Pollak and Wales (1992), CUUS s a property that ensures that estmated economc effects are constant to the scalng of the data. Moreover, Pollak and Wales (1980) stated that only demand systems consstent wth CUUS should be used for emprcal demand analyss. Followng Matsuda (2006), the LA/QUAIDS model s derved as: n 2 y y w logp log log 1,2,..., n C Z C (1) 1 P P P where w s the expendture share of good, y s total expendture wthn the system, p s the prce of good, α, β, γ, and λ are parameters to be estmated. C log P w logp (2) The loglnear analogue of the Laspeyres prce ndex s shown n Equaton 2. P C s nvarant to changes n unts. stands for the sample mean. We apply ndex P Z as proposed by Matsuda (2006) n Equaton 3: n Z p log P w w log (3) 1 p where P Z could be seen as a zero degree homogenous analogue of the Tӧrnqvst prce ndex analogue and s lkewse nvarant to changes unts. We defne the estmated LA/QUAIDS model usng expendtures shares and prces for the demand of green tea, black tea, tea, coffee, coffee, frut and vegetable uce, carbonated and mlk wth regards to demographc, temperature, and other dummy varables as follow: y y w 0 1t, 1k Zk, 7mDm 19h, 19rM r logp log log 1, 2,...,8 C Z C (4) k1 m1 r1 1 P P P where t ndexes the tme n months, Z k are demographc varables, D m are monthly dummy varables, h s temperature, M r are cty dummy varables. The parameters consequently are expected to meet the followng restrctons: k 1,2,...,67 (5) 0 k n (6) 0 (7) 1,2,..., 8 0 1,2,..., 8 (9) The ensung demand system ontly ensures that t fulflls addng up and homogenety. In addton, symmetry s 2 (8) 145

4 guaranteed by the addtonal restrcton:, 1,2,...,8 (10) The expendture, uncompensated, and compensated prce elastctes can be calculated as follows: 2 y 1 log 1,2,...,8 (11) Z C w wp P w y y 2w w w log log, 1,2,..., 8 (12) Z C C w w w P P P c w (Slutsky equaton), 1, 2,...,8 (13) Where δ s the Kronecker delta: 1for = ; for Results Table 1 reports the descrptve statstcs of varables. The expendture share of mlk (0.371), frut and vegetable uce (0.184) and coffee (0.107) show they have the hghest mean expendture shares whereas black tea (0.018) and carbonated (0.057) have the lowest mean expendture shares n the household. Table 1. Descrptve statstcs of varables Varables Mean Std. devaton Mnmum Maxmum Number of persons per household (z 1 ) Number of persons per household under the age of 18 (z 2 ) Number of persons per household over the age of 65 (z 3 ) Age of the head (z 4 ) Rate of those payng rent (%)(z 5 ) Expendture share of green tean (w 1 ) Expendture share of black tea (w 2 ) Expendture share of tea (w 3 ) Expendture share of coffee (w 4 ) Expendture share of coffee (w 5 ) Expendture share of frut & vegetable uce (w 6 ) Expendture share of carbonated (w 7 ) Expendture share of mlk (w 8 ) Prce of green tea (p 1 ) Prce of blacke tea (p 2 ) Prce of tea (p 3 ) Prce of coffee (p 4 ) Prce of coffee (p 5 ) Prce of frut & vegetable uce (p 6 ) Prce of carbonated (p 7 ) Prce of mlk (p 8 ) Temperature (h) Table 2 reports the estmates of expendture and prce coeffcents at the mean shares. The prce coeffcents for green tea (-0.127), tea (-0.103), coffee (-0.019), frut and vegetable uce (0.042), and carbonated (-0.133) are statcally sgnfcant at the 1 percent level, whereas black tea, coffee, and mlk are not sgnfcant. In the case of total expendture, wth the excepton of frut and vegetable uce, all other s are sgnfcant at the 1 percent level. As for the quadratc log-lnear expendture, green tea (0.174), tea (-0.015), coffee (-0.027), carbonated (-0.020) and mlk (0.174) are sgnfcant at the 1 percent level. Black tea, coffee, and frut and vegetable uce are not sgnfcant. The estmated R 2 results were satsfactory. They ranged from for black tea equaton to for tea equaton ndcatng the model explans well the expendture shares and prces for the non-alcoholc n the household budget. 146

5 Table 2. Estmates of expendture and prce coeffcents Left-hand varable w Regressor y log p 1 log p 2 log p 3 log p 4 log p 5 log p 6 log p 7 log p 8 log P C 1 y log Z C P P 2 R 2 Green tea *** *** *** 0.173*** (-3.890) (-0.812) (0.325) (-0.372) (0.965) (3.271) (1.011) (0.227) (32.280) (15.094) Black tea * * 0.006*** (0.709) (-0.604) (0.618) (-1.659) (-1.252) (1.103) (-0.193) (1.847) (5.061) Tea *** *** *** 0.034*** *** ** (-7.990) (-1.421) (-3.092) (-3.007) (2.882) (0.041) (-7.383) (-2.426) Coffee ** * *** (1.150) (2.394) (0.860) (1.797) (0.385) (3.235) (-0.652) Coffee * *** 0.060*** *** *** (-1.936) (-0.146) (-2.658) (4.624) (-3.393) (-5.158) Frut & vegetable uce 0.042* *** (0.448) (1.897) (1.260) (-7.183) (1.241) Carbonated *** 0.084*** *** *** ( ) (7.249) (-4.964) (-4.294) Mlk *** 0.173*** Note. The degrees of freedom of the demand system are 53,595. The correspondng crtcal values of the t-dstrbuton for 1%, 5%, 10% sgnfcance levels are 2.576, 1.960, and 1.645, respectvely. ***, **, and * mean that the estmate s dfferent from zero at the 1%, 5%, and 10% sgnfcance levels, respectvely. Defned as the squared correlaton between the observed and predcted shares, R 2 s computed for each sngle equaton. t-values are n parentheses. (0.933) ( ) (15.090) The estmates of expendture and uncompensated prce elastctes are shown n Table 3. The expendture elastcty for green tea (2.320), black tea (1.357), tea (0.836), coffee (1.090), coffee (0.896), frut and vegetable uce (1.019), carbonated (0.844) and mlk (0.677) are all sgnfcant at the 1 percent level. More precsely, the expendture elastcty for green tea, black tea, coffee and frut and vegetable uce are elastc, whle tea, coffee, carbonated and mlk are expendture nelastc. In other words, green tea, black tea, coffee and frut and vegetable uce are luxures whle tea, coffee, carbonated and mlk are necesstes. Ths means that for nstance a 1 percent ncrease n the non-alcoholc expendtures ncreases demand for green tea by percent and black tea by percent respectvely. 147

6 Table 3. Estmates of expendture and uncompensated prce elastctes at the mean shares Demand q Expendture y Green tea Black tea Tea Prce p Coffee Coffee Frut & vegetable uce Carbonated Mlk Green tea 2.320*** *** ** 0.835*** *** 0.338*** 0.601*** *** (56.276) (-9.882) (-2.448) (6.061) (-5.146) (2.758) (3.381) (-0.173) (-3.444) Black tea 1.357*** ** *** 0.812*** *** ** (18.783) (-2.240) (-4.483) (2.542) (-2.735) (-1.329) (2.233) (0.405) (0.246) Tea 0.836*** 1.000*** 0.154*** *** *** *** 0.673*** (35.837) (7.115) (2.705) ( ) (0.301) (-5.428) (-4.207) (9.330) (1.523) Coffee 1.090*** *** ** *** *** ** *** (39.440) (-3.545) (-2.574) (-0.208) ( ) (-0.186) (9.193) (-2.093) (-3.540) Coffee 0.896*** 0.665*** *** *** *** ** 1.108*** (29.867) (3.509) (-1.224) (-5.476) (0.116) (-8.568) (-3.068) (-2.440) (7.016) Frut & vegetable uce 1.019*** 0.459*** 0.091** *** 0.387*** *** *** *** (55.933) (4.710) (2.387) (-4.490) (9.399) (-3.213) (-5.771) (-0.263) (-9.345) Carbonated 0.844*** *** * ** *** 1.534*** (27.867) (0.610) (0.514) (9.317) (-1.756) (-2.407) (-0.060) ( ) (9.627) Mlk 0.677*** ** *** *** 0.247*** *** (61.564) (-0.217) (0.855) (1.963) (-1.592) (7.512) (-7.936) (10.000) ( ) Note. The same as Table 2. As for the uncompensated own- prce elastctes, all eght s are sgnfcant at the 1 percent level. Wth the excepton of frut and vegetable uce and mlk, whch are own-prce nelastc, all other s are own-prce elastc. More specfcally, carbonated has the most own-prce elastc and frut and vegetable uce has the most own-prce nelastc. Ths mples that f prces for green tea, black tea, tea, coffee, coffee, and carbonated are lowered, total expendture wll ncrease snce to the quantty sold wll augment by large percentage than the decrease n prce (see Table 3). Whereas for frut and vegetable uce and mlk, the ndcaton s that a fall n prce of the s wll affect a smaller percentage change n quantty demanded. Usng the LA/QUAIDS model, Dharmasena et al. (2009) also found the own-prce elastcty of coffee to be elastc at Moreover, Zheng and Kaser (2008) usng the LA/AIDS model found the uncompensated prce elastcty for soft drnks (carbonated ) and mlk to be prce nelastc at and respectvely, whch s consstent wth our results. Usng Rotterdam and Translog demand system model, both Knnucan et al. (2001) and Yen et al. (2004) also found carbonated and mlk to be prce nelastc. The uncompensated cross prce elastctes shows that nneteen pars are gross substtutes. Among the s group, tea have the most number of substtutes whereas coffee has the least number of substtute pars. The top par of substtute s carbonated and mlk. Ths means that, for nstance, a 1 percent ncrease n the prce of carbonated ncreases demand for mlk by percent, whle a 1 percent ncrease n the prce of mlk ncreases demand for carbonated by percent. In addton, ths result s consstent wth Watanabe et al. (1997), who also found mlk to be substtute wth carbonated n the Japanese household. As for complementary s, twenty pars are found to be gross complement n the s group. Coffee and frut and vegetable uce have the most number of pars among the group whle black tea has the least number of complementary par. Table 4 reports the estmates of compensated prce elastctes at the mean shares. All eght compensated own-prce elastctes are sgnfcantly negatve, whch are consstent from theoretcal perspectve. As for the compensated cross prce elastctes, twenty-four pars are found to be sgnfcantly gross substtutes. Among the s group, tea has the most number of par and seems to be substtute wth every other s wth the excepton of coffee. Carbonated has the least number of par among the group. Moreover, sxteen pars are found to be gross complements. Coffee and frut and vegetable uce have the most pars among the group whle carbonated and mlk have the least number of pars. One of the complementary pars s mlk and frut and vegetable uce. Ths result s consstent wth the studes of 148

7 Yen et al. (2004), Zheng and Kaser (2008), and Dharmasena et al. (2009). In ther study, Yen et al. (2004) found mlk as a complement to uce. Table 3 shows coffee and mlk as gross complement however Table 4 shows they are gross substtute. Table 4. Estmates of compensated prce elastctes at the mean shares Demand q Green tea Black tea Tea Coffee Prce p Coffee Frut & Vegetable uce Carbonated Mlk Green tea *** * 1.063*** ** 0.489*** 1.028*** (-9.075) (-1.956) (7.722) (-2.151) (3.988) (5.769) (1.059) (1.108) Black tea * *** 0.945*** * *** (-1.956) (-4.403) (2.963) (-1.922) (-1.017) (2.855) (0.695) (1.461) Tea 1.084*** 0.169*** *** 0.105** *** *** 0.721*** 0.512*** (7.722) (2.963) ( ) (1.980) (-4.733) (-2.866) (9.996) (3.855) Coffee ** * 0.097** *** *** * (-2.151) (-1.922) (1.980) ( ) (1.515) (12.035) (-0.579) (1.646) Coffee 0.754*** *** *** ** * 1.441*** (3.988) (-1.017) (-4.733) (1.515) (-8.171) (-1.970) (-1.908) (9.107) Frut & vegetable uce 0.561*** 0.109*** *** 0.496*** ** *** *** (5.769) (2.855) (-2.866) (12.035) (-1.970) (-4.120) (0.920) (-5.144) Carbonated *** * *** 1.847*** (1.059) (0.695) (9.996) (-0.579) (-1.908) (0.920) ( ) (11.577) Mlk *** 0.037* 0.252*** *** 0.286*** *** (1.108) (1.461) (3.855) (1.646) (9.107) (-5.144) (11.577) (-8.579) Note. The same as Table 2. Table 5 reports the estmates of monthly rates of shft and demographc effects at the mean shares. The effect of the lnear tme trend (t) on demand s the monthly rate of shft. Fve of the eght s are shown to have a sgnfcant effect on quantty demanded. Wth the excepton of black tea, coffee, and frut and vegetable uce, all other s are sgnfcant at the 1 percent level. As tme goes by, demand for green tea (0.182), tea (0.166), coffee (0.175), and carbonated (0.146) ncreases whle demand for mlk (-0.179) decreases n the household. When the household sze ncreases (Z 1 ), demand s postve for coffee (10.563), carbonated (22.601), and mlk (4.271), whle t s negatve for green tea ( ), and black tea ( ). When the household sze s younger (Z 2 ), demand for frut and vegetable uce (12.752) and mlk (5.928) s postve whle demand for green tea ( ), tea (-6.584), coffee (-9.630), and coffee ( ) s negatve. Intutvely, ths s true; as n Japan chldren tend to drnk mlk due ts many nutrtonal values and health beneft. Ths result s consstent wth Watanabe et al. (1997) who found younger people, larger famles, and people wth calcum dstress consume more mlk than any other segment of the household. In addton, they mentoned that ncrease of consumpton of mlk among age group has a lot to do wth breakfast habts, as people are more lkely to consume mlk durng breakfast than any other meal. When there are more elders n the household (Z 3 ), demand for green tea (24.427) and mlk (10.438) s postve whle black tea ( ), tea (-8.452), coffee ( ), coffee ( ), and carbonated ( ) s negatve. Ths s true as far as Japan s concerned snce elderly people tend to be more conscous on ther health and det and s such as green tea and mlk contrbute to that aspect. Ths fndng s also consstent wth Watanabe et al. (1997) fndngs of postve relatonshp between health concerns and preferences for mlk demand for elderly people. When the household head s older (Z 4 ), demand for green tea (0.807), coffee (0.840), carbonated (0.582) and mlk (0.407) are postve whle demand for tea (-0.864), coffee (-1.781), and frut and vegetable uce (-0.926) are negatve. When the percentage of those payng rents ncreases (Z 5 ), demand for tea (0.093), coffee (0.231), frut and vegetable uce (0.166), and carbonated (0.339) s postve whle demand for coffee (-0.091), and mlk (-0.205) s negatve. 149

8 Table 5. Estmates of monthly rates of shft and demographc effects at the mean shares Demographc varable z k Demand q Monthly rate of shft (%/month) Number of persons per household (%/person) Number of persons under the age of 18 (%/person) Number of persons per household over the age of 65 (%/person) Age of the head (%/year old) Rate of those payng rents (%) Green tea 0.182*** (2.852) *** (-5.495) * (-1.941) *** (3.463) 0.807* (1.756) (1.080) Black tea (1.273) ** (-2.420) (0.022) * (-0.893) (1.105) (1.242) Tea 0.166*** (3.420) (0.099) * (-1.719) ** (-2.114) *** (-3.317) 0.093** (2.076) Coffee 0.175*** (7.763) (-0.634) ** (-2.116) *** (-5.052) 0.840*** (2.716) * (-1.708) Coffee (-0.281) *** (2.990) *** (-3.417) *** (-3.392) *** (-5.320) 0.231*** (4.022) Frut & vegetable uce (0.499) (-0.860) *** (4.266) (-0.726) *** (-4.558) 0.166*** (4.763) Carbonated 0.146** (2.548) *** (6.337) (-1.608) *** (-3.953) 0.582* (1.720) 0.339*** (5.837) Mlk *** ( ) 4.271*** (3.296) 5.928*** (3.282) *** (5.536) 0.407*** (3.317) *** (-9.714) Note. The same as Table 2. Table 6 shows the estmates of temperature effects at the mean shares. Temperature shows negatve effects on demand of green tea and black tea n most of the months whereas coffee s negatvely sgnfcant all year round. Moreover, temperature for tea, coffee, fruts and vegetable uce and carbonated mostly show postve effects. Ths s true wth regards to Japan as s such as tea, coffee, frut and vegetable uce and carbonated are usually served cold whereas green tea, black tea and coffee are usually served hot through vendng machne. For nstance, carbonated s such as coca-cola and sprte are usually preferred durng a hot temperature whereas drnks such as green tea or black tea are preferable durng a cold temperature. As for the demand of mlk, t s only sgnfcant n the month of July. One possble explanaton for could be mlk s not usually served n the vendng machne. Table 6. Estmates of temperature ( o C) effects at the mean shares Demand q January February March Aprl May June Green tea (-0.788) (-0.439) (0.200) (-0.101) 4.025*** (5.196) ** (-2.000) Black tea (-1.082) (-0.695) (0.504) ** (-2.590) (-0.584) *** (-3.397) Tea 0.755*** (2.692) 0.704*** (2.772) 1.071*** (3.976) 2.369*** (7.800) (1.643) 1.322** (2.556) Coffee *** (-4.493) *** (-4.168) *** (-6.071) *** (-6.510) *** (-4.789) *** (-5.194) Coffee (0.684) 0.634* (1.945) 1.132*** (3.273) 1.285*** (3.293) (0.674) 1.355** (2.037) Frut & vegetable uce 0.433** (1.981) (1.512) (1.265) (1.315) (-0.468) 1.008** (2.495) Carbonated (-0.659) (-1.169) (0.483) (0.226) (-1.112) * (1.910) Mlk (1.211) (0.456) (-0.998) (-1.507) (-0.258) (1.184) Demand q July August September October November December Green tea *** (-6.130) ** (-2.086) *** (-4.507) *** (-7.252) * (-1.707) 2.043*** (4.367) Black tea * (-1.838) (-0.447) (-1.586) * (-1.757) (-0.145) (0.691) Tea 2.095*** (6.699) 2.214*** (5.916) 0.610* (1.889) 1.495*** (5.057) 1.293*** (4.749) 1.304*** (4.913) Coffee *** (-4.784) *** (-7.232) *** (-5.808) *** (-7.302) *** (-6.745) *** (-9.583) Coffee 1.326*** (3.297) 1.456*** (3.023) (0.990) 1.220*** (3.209) (-0.361) (0.033) Frut & vegetable uce 2.106*** (8.631) 0.577** (1.974) 1.319*** (5.236) 1.644*** (7.129) 0.923*** (4.351) (0.384) Carbonated 1.511*** (3.721) (0.740) 1.991*** (4.748) 2.128*** (5.543) 1.152*** (3.258) *** (-4.665) Mlk *** (-3.769) (0.795) (1.423) (0.591) (-0.810) (1.504) Note. The same as Table

9 The hghest negatve temperature effects are partcularly seen n the month of July, September and October on green tea, Aprl, June, and July on black tea, June, August, and December on coffee. To gve an example, demand for green tea decreases by percent n September as temperature rses by 1 degree Celsus. As for postve effects s concerned, t s mostly hgh n July and August both on tea and coffee, from July to October for frut and vegetable uce, and from September to October for carbonated. For nstance, as temperature rses by 1 degree Celsus, demand for tea ncreases by percent n the month of July. Wth the excepton of mlk, temperature effect s mostly sgnfcant for all other s. Moreover, these effects postve and negatve are hgher durng the summer seasons. 4. Concluson Ths study usng the LA/QUAIDS model analyzed the demand for non-alcoholc s n Japan for eght s group usng monthly data obtaned from FIES. The estmated results reveal that expendture elastctes for green tea, black tea, coffee, and frut and vegetable uce are elastc whle tea, coffee, carbonated, and mlk are nelastc. Ths mples that green tea, black tea, coffee, and frut and vegetable uce are luxures whereas tea, coffee, carbonated, and mlk are necesstes n the household. The emprcal results for uncompensated own-prce elastctes show that demand for green tea, black tea, tea, coffee, coffee, and carbonated are own-prce elastc whle demand for frut and vegetable uce and mlk are nelastc. The cross prce elastctes for both uncompensated and compensated show that most of the s are complement to each other. As for soco-demographc effect s concerned, t s found to play a key role n the determnant of the non-alcoholc s consumpton. The results reveal that people under the age of 18 prefer mlk and frut and vegetable uce than any other s, whle elderly people prefer more green tea n addton to mlk. Ths ndcates that n the Japanese household demand, there s a strong assocaton between consumer behavor and ts health and detary mplcaton. Wth regards to seasonal effects, the study reveals that when temperature rses, people consume more tea, coffee, frut and vegetable uce, and carbonated whereas when temperature drops consumers prefer more green tea, black tea and coffee. In realty, ths s true snce green tea, black tea, and coffee are usually served hot and consumed at home whereas s such as tea, coffee, frut and vegetable uce and carbonated are served cold and are frequently consumed n convenence shops and through vendng machnes. The fndngs of our study have an mplcaton for polcy. Mlk s set to be good for health, and as our fndngs show coffee and carbonated are substtutes for mlk. Consequently, mlk consumpton wll ncrease f the government taxes coffee and carbonated. Acknowledgements Ths research was supported by Grant-n-Ad for Scentfc Research (C), no , Japanese Mnstry of Educaton, Culture, Sports, Scence and Technology. References Agnew, M. D., & Palutkof, J. P. (1999). The mpacts of clmate on retalng n the UK wth partcular reference to the anomalously hot summer and mld wnter of Internatonal Journal of Clmatology, 19, Alston, J. M., Chalfant, J. A., & Pggott, N. E. (2001). Incorporatng demand shfters n the almost deal demand system. Economcs Letters, 70, Bahng, Y., & Kncade, D. H. (2012). The relatonshp between temperature and sales: Sales data analyss of a retaler of branded women s busness wear. Internatonal Journal of Retal & Dstrbuton Management, 40(6), Banks, J., Blundell, R. W., & Lewbel, A. (1997). Quadratc Engel curves and consumer demand. Revew of Economcs and Statstcs, 79, Deaton, A. (1985). Panel data from tme seres of cross-sectons. Journal of Econometrcs, 30, Deaton, A. S., & Muellbauer, J. (1980). An almost deal demand system. The Amercan Economc Revew, 70, Dharmasena, S., Capps Jr, O., & Clauson, A. (2009). Nutrtonal contrbutons of non-alcoholc s to the U.S. Det: In Southern Agrcultural Economcs Assocaton Annual Meetng, Atlanta, GA. 151

10 Harrson, K. (1992). Whether the weather be good. Super Marketng, Helg Lbrary. (2011). Data on Tea Consumpton Per Capta. Retreved from Japan Soft Drnk Assocaton. (2014). Statstcal Beverage Data Retreved from Knnucan, H. W., Mao, Y., Xao, H., & Kaser, H. M. (2001). Effects of advertsng on US non-alcoholc demand: Evdence from a two-stage Rotterdam model. Advances n Appled Mcroeconomcs, 10, Matsuda, T. (2006). Lnear approxmatons to the quadratc almost deal demand system. Emprcal Economcs, 33, Mnstry of Agrculture, Forestry and Fsheres. (MAFF). Current stuaton of (mlk) consumpton. Japan. Retreved from Murray, K. B., D Muro, F., Fnn, A., & Popkowsk Leszczyc, P. (2010). The effect of weather on consumer spendng. Journal of Retalng and Consumer Servces, 17(6), Parker, P. M., & Tavassol, N. T. (2000). Homeostass and consumer behavor across cultures. Internatonal Journal of Research n Marketng, 17, Pofahl, G. M., Capps Jr, O., & Clauson, A. (2005). Demand for non-alcoholc s: Evdence of from the ACNelsen home scan panel. Amercan Agrcultural Economcs Assocaton annual meetng, Provdence, RI. Pollak, R. A., & Wales, T. J. (1980). Comparson of the quadratc expendture system and translog demand systems wth alternatve specfcatons of demographc effects. Econometrca, Pollak, R. A., & Wales, T. J. (1992). Demand System Specfcaton and Estmaton. Oxford Unversty Press Catalogue. Roslow, S., L, T., & Ncholls, J. A. F. (2000). Impact of stuatonal varables and demographc attrbutes n two seasons on purchase behavor. European Journal of Marketng, 34, Schluep Campo, I., & Beghn, J. C. (2005). Dary food and consumpton, producton, and polcy n Japan. Food and Agrcultural Polcy Research Instttute (FAPRI) at Iowa State Unversty. Statsta. (2014). Per capta consumpton of soft drnks n Unted States from 2000 to Retreved from soft-drnks/ Stroppana, R., Rethmuller, P., & Kobayash, K. (1998). Regonal Dfferences n the Japanese Det: The Case of Drnkng Mlk. Economc Analyss and Polcy, 28, Watanabe, Y., Suzuk, N., & Kaser, H. M. (1997). Identfyng consumer characterstcs assocated wth Japanese preferences toward mlk products. Agrbusness, 13(4), (199707/08)13:4<357::AID-AGR1>3.0.CO;2-Y Watanabe, Y., Suzuk, N., & Kaser, H. M. (1998). Factors affectng consumers choce of s n Japan. Agrbusness, 14(2), >3.0.CO;2-2 Watanabe, Y., Suzuk, N. (2006). Is Japan s mlk consumpton saturated? Journal of the Faculty of Agrculture Kyushu Unversty, 51(1), Yen, S. T., Ln, B. H., Smallwood, D. M., & Andrews, M. (2004). Demand for nonalcoholc s: The case of low-ncome households. Agrbusness, 20, Zheng, Y., & Kaser, H. M. (2008). Advertsng and U.S. nonalcoholc demand. Agrcultural and Resource Economcs Revew, 37,

11 Copyrghts Copyrght for ths artcle s retaned by the author(s), wth frst publcaton rghts granted to the ournal. Ths s an open-access artcle dstrbuted under the terms and condtons of the Creatve Commons Attrbuton lcense ( 153