Food prices, taxes, and obesity in Canada and its implications for food taxation

22 2019, XXII, 1 Economics DOI: 10.15240/ ul/001/2019-1-002 In oduc ion An impo an heal h i nancing issue acing Canada and o he OECD coun ies (OECD, 2017) a e he heal h consequences o obesi y. S a is ics Canada (2014) epo s ha 51.6% o adul Canadians we e o e weigh o obese in 2009 compa ed o 53.6% in 2013. The p opo ion o Canadians who a e o e weigh di e s by sex, wi h 59.2% o Canadian males o e weigh in 2009 compa ed o 62% in 2013 and 43.9% o emales o e weigh in 2009 compa ed o 45.1% in 2013. These inc eases ha e led o calls o policies o con ol obesi y (see Cla k e al., 2014). These a es o obesi y a e based on he body mass index (BMI) which is he a io o weigh (in kilog ams) o he squa e o heigh (in me e s). C an i eld (2007) uses he Canadian Communi y Hea h Su ey (CCHS) o examine he de e minan s o he body mass index (BMI) o Canadians. Un o una ely, he e is insu i cien in o ma ion in he CCHS da a on he die a y beha iou o esponden s so ha he way in which die s in l uence BMI canno be di ec ly s udied using ha da ase . This lack o in o ma ion on die a y choices is a common ea u e o s udies ha measu e he BMI o indi iduals. Fo example, he BMI da a used by Goldman, Lakdawalla, and Zheng (2009) also does no include obse a ions o he die s o indi iduals. One solu ion o his p oblem is o include he p ices o die a y elemen s in place o quan i ies since, while quan i ies may no ha e been obse ed, p ices ypically a e. Nu ien p ices a e included in he BMI s udies o Goldman, Lakdawalla and Zheng (2009) and Cla k e al. (2014) whe eas ood g oup p ices a e included in s udies by Chou, G ossman and Sa e (2004), Boizo -Szan ai and Fab ice (2005) and S um and Da a (2005) and (2008). In his s udy we discuss some conside a ions o he de e mina ion o he app op ia e se o p ices o use wi hin a s udy o obesi y. We compa e he use o ood-g oup-p ices o nu ien -p ices. By a he mos popula app oach in his a ea o esea ch is he use o ood g oup p ices (e.g., Chou, G ossman, & Sa e , 2004; Boizo - Szan ai & Fab ice, 2005; S um & Da a , 2005; 2008; along wi h Bea y & LaF ance, 2005; Chouina d e al., 2005; Smed, Jensen, & Den e , 2007; Allais, Be ail, & Nichele, 2010). Fewe s udies use nu ien p ices (Goldman, Lakdawala, & Zheng, 2009; Cla k e al., 2014). A heo e ical model de eloped by Chen e al. (2004) explo es he ela ionship be ween nu ien s and hype ension. Thei model clea ly demons a es ha i nu ien s a e a de e minan o a speci i c heal h ou come (hype ension, in hei case) hen nu ien p ices a e he mos app op ia e se o p ices o use. In hei empi ical model hey do no use nu ien p ices bu ood g oup p ices, p oposing ha hei es ima ed model is a “ educed o m” model. We asse ha Chen e al. (2004) a e co ec in hei in e p e a ion o hei model and ha ood g oup p ices included in obesi y s udies lead o es ima es o a educed o m model and nu ien p ices lead o es ima es o a “s uc u al” model. Gi en his in e p e a ion, ood g oup p ices can be used o es ima e he model i one is in e es ed in a educed o m model o he de e minan s o obesi y and nu ien p ices should be used i one is in e es ed in es ima ing a s uc u al model o obesi y ha is mo e amenable o in e p e a ion o policy pu poses. The dis inc ion is c i ical since educed o m models allow o he es ima ion o o he non- p ice a iables in he sys em o make su e he e is no an “omi ed” a iable p oblem bu hey a e less use ul o policy pu poses unless some addi ional app oach is g a ed on o he s udy o he pu pose o unco e ing he s uc u al pa ame e s. Pa ame e es ima es om he use FOOD PRICES, TAXES, AND OBESITY IN CANADA AND ITS IMPLICATIONS FOR FOOD TAXATION S ephen J. Cla k, Ludwig O. Di ich, S ephen M. Law, Dana S a á, Mi osla Ba ák EM_1_2019.indd 22EM_1_2019.indd 22 8.3.2019 9:15:498.3.2019 9:15:49 23 1, XXII, 2019 Economics o ood g oup p ices should no be used o make judgemen s conce ning he heal h impac s o axes on consump ion o oods om many o he ood g oups because oo many ac o s a e con ounded in he educed o m model ela ing hose ood g oup p ices o he unde lying nu ien s. We p o ide some examples o his sho coming o he ood g oup app oach below. The use o nu ien p ices is a mo e p omising line o esea ch in his a ea since his app oach cap u es he p ices o he speci i c nu ien s hough o be causing undesi ed heal h isks by nu i ionis s. The di i cul y wi h nu ien p ices is ha hey a e no ypically collec ed by s a is ical epo ing ins i u ions and a e no di ec ly obse able. Nu ien p ices need o be cons uc ed om exis ing da a bases and linked o he obse a ions on expendi u es and BMI. The goal o he s udy is o in oduce new nu ien p ice model and shows i s supe io i y o ood g oup model based on Canadian da a. This s udy is o ganized as ollows. The nex sec ion e iews ood policy in Canada, speci i cally supply managemen egimes and he axa ion o ood, and he ways in which he s uc u e o hese policies e l ec s conside a ion o heal h ou comes, i a all. The ollowing sec ion e iews he li e a u e on p ices o ood and nu ien s in s udies o obesi y and o he heal h ou comes. This is ollowed by a discussion o he da a and me hods. The penul ima e sec ion discusses he esul s and he i nal sec ion p o ides a discussion o policy implica ions, ele ance o nu ien axa ion model o o he coun ies and conclusions. 1. Food P ice and Tax Policies in Canada and Thei Rela ionship o Obesi y While he e a e no ood policies in Canada ha explici ly a ge obesi y in Canada, he e a e se e al policies in place ha in l uence he ela i e p ices o a ious oods in ela ion o o he oods. T adi ionally, P o incial and Fede al HST and GST axes in Canada ha e exemp ed many oods. Howe e , he e a e h ee majo a eas o ood axes ha a e common o many p o incial and ede al axes in Canada ha may in l uence obesi y: (1) Taxes on es au an ood; (2) Taxes on con enience ood; and (3) “Taxes” on ood ela ed o supply- managed ood indus ies. 1.1 Taxes on Res au an Food Res au an ood has been subjec o ede al axes (GST) and p o incial axes (PST o he p o incial componen o HST) in Canada. In he obesi y li e a u e, hese axes a e oughly compa able o axes on “ as ood” in he li e a u e. The e o e, in as much as “ es au an ” ood lines up o a ood g oup called “ as ood”, Canada has had an obesi y ax o a long ime. 1.2 Taxes on Con enience Foods Ano he obesi y ax ha is o en ad oca ed in ela ed li e a u e is a ax on “junk” ood, high a o high calo ic ood ha is hough o con ibu e o obesi y. While he e a e no ood axes in Canada ha explici ly a ge “junk” ood o axing, HST o PST and GST a e applied o u he -p ocessed oods, o “con enience” oods. These axes a e une enly and inconsis en ly applied i hey a e iewed om he pe spec i e o axes whose pu pose is o con ol obesi y. While his ca ego y does co espond o se e al oods ha a e commonly conside ed o be “junk” oods (e.g., po a o chips), o he high a , high suga oods a e un axed (e.g., ice c eam) and oods which a e in ended o be obesi y- educing (such as Weigh Wa che s® ozen ood dinne s) a e also axed. Fu he mo e, he con enience ood axes a e inconsis en ly applied o speci i c oods. Fo example, indi idual cookies a e axed, whe eas cookies pu chased in bags in quan i ies g ea e han six cookies a e no (as long as hey a e no indi idually w apped). 1.3 Supply Managemen Indus y Taxes The i nal ca ego y o ood p ice inc eases (o “ axes”) in Canada a e hose imposed on consume s by he supply managed indus ies. These include dai y, poul y, and egg p oduc s. Fo a discussion o he ela i e size o hese implici axes, see Ca dwell, Lawley and Xiang (2015). In he case o poul y and egg p oduc s, i is deba able whe he inc easing hei p ice will dec ease obesi y, since hese p oduc s a e o en singled ou as p oduc s ha con ibu e o a heal hie die by nu i ionis s, especially poul y p oduc s. To he ex en ha poul y and eggs a e “good” oods, he implici axes om supply managemen in hese indus ies migh be conside ed obesogenic a he han lep ogenic. P ice inc eases om supply managemen in he dai y sec o a e a unique case om he poin o iew o obesi y axes. Dai y supply managemen EM_1_2019.indd 23EM_1_2019.indd 23 8.3.2019 9:15:498.3.2019 9:15:49 24 2019, XXII, 1 Economics has wo impac s on consume s. The i s is he usual ax on consume s om aising he e ail p ice o dai y p oduc s o consume s. Fo hese axes, since many dai y p oduc s a e high a and he e o e lead o an obese ou come, dai y axes p omo e a less obese socie y. The second is he peculia way dai y quo a is alloca ed on he basis o he a con en o milk. This alloca ion sys em lowe s he amoun o dai y a ha en e s he sys em. None o hese axes a e explici ly imposed o p omo e heal hy ea ing on he pa o consume s. The e o e, hey do no necessa ily a ge he nu ien s esponsible o causing obesi y in Canada. Cla k and Di ich (2010) a gue ha nu ien axes (e.g., axes on he a con en o ood) be e a ge he p oblem han do ood g oup axes (i.e., axes on ood composi es) and he e o e nu ien axes a e mo e e ec i e as a means o p omo e less obesi y. All o he axes in Canada, wi h he excep ion o he alloca ion o milk quo as by a con en , a e ood g oup axes. The p oblem o wha p ices a e app op ia e o obesi y s udies is discussed in he nex sec ion. Tab. 1 summa izes ood axes on con enience oods and es au an oods by p o ince in Canada oge he wi h he a e age BMI by p o ince. The able sugges s ha he e is no consis en s o y ha can be old ha ela es he le el o he sales axes on con enience oods and es au an oods in Canada o obesi y. Fo example, Albe a has he lowes a e o axes on con enience and es au an oods in Canada and Albe ans ha e he lowes a e age BMI in he coun y, while A lan ic Canada has he highes axes on es au an and con enience oods and A lan ic Canadians a e among he highes a e age BMI in he coun y. This indica es ha he e could be conside able con ounding om simul aneous posi i e and nega i e e ec s appea ing in he aw co ela ions o es au an and con enience ood axes wi h he BMI o Canadians. In On a io es au an s meals a e exemp om p o incial ax i expendi u e a e less han 4 CND. 2. P ices in Obesi y S udies The empi ical li e a u e in his a ea is p ima ily composed o s udies which ha e concen a ed on es ima ing he educed o m ela ionship be ween ood g oup p ices and an obesogenic ( ha is, “obesi y-inc easing”) nu ien o die a y elemen , such as a , ca bohyd a es, o calo ies. The mos common app oach (Huang, 1996; Bea y & LaF ance, 2005; Chouina d e al., 2005; Smed, Jensen, & Den e , 2007; Allais, Be ail, & Nichele, 2010) is o ela e he educed o he obesogenic nu ien o o e all demand o he ood g oup p oduc by combining he a e age nu ien con en o ood in i xed p opo ions wi h he o e all quan i y demanded o he ood g oup o p oceed om a educed o m model o ood demand o a s uc u al model o nu ien demand. P o ince P o incial Tax Fede al Tax Combined Tax A e age BMI in 2010 Albe a n.a. 5% 5% 25.987 Saska chewan n.a. 5% 5% 26.862 B i ish Columbia 7% 5% 12% 27.411 Mani oba 8% 5% 13% 27.361 New B unswick 8% 5% 13% 27.657 New oundland &Lab ado 8% 5% 13% 27.839 On a io 8% 5% 13% 26.746 P ince Edwa d Island 9% 5% 14% 26.984 Quebec 9.975% 5% 14.975% 26.001 No a Sco ia 10% 5% 15% 27.238 Sou ce: own calcula ion based on S a is ics Canada Tab. 1: A e age BMI (2010) and ax a es o es au an meals and con enience oods in Canada (as o May 2015) EM_1_2019.indd 24EM_1_2019.indd 24 8.3.2019 9:15:498.3.2019 9:15:49 25 1, XXII, 2019 Economics This app oach has h ee ad an ages. Fi s , demand elas ici ies om ood demand sys ems can be used o de i e he esponse o consump ion o he obesogenic nu ien o changes in ood g oup p ices. Second, gi en he i xed p opo ions assump ion, a demand elas ici y o he obesogenic (o combina ions o obesogenic) nu ien p ices can also be es ablished. The i xed p opo ions assump ion allows an analysis o ei he he e ec o changes o a ood g oup p ice o nu ien p ice on consump ion, while empi ically using only ood g oup p ices o es ima e he ood demand sys em. Thi d, es ima es o ela ionships can be es ablished using exis ing demand sys em es ima es which ha e been in ensi ely s udied and so excellen es ima es o demand elas ici ies a e a ailable. The e a e h ee p oblems wi h his app oach. Fi s , he assump ion o i xed p opo ions be ween he o e all elas ici y and he obesogenic nu ien ha allows o he cons uc ion o a s uc u al elas ici y om a educed o m o a s uc u al elas ici y is a main ained hypo hesis ha may no be alid. This app oach allows o ac oss- g oup subs i u ions o obesogenic nu ien s bu does no allow o wi hin-g oup subs i u ions o nu ien s. Cla k and Di ich (2010) obse e ha ood g oups ha e a la ge numbe o elemen a y goods and ha wi h changes in p ice subs i u ions o nu ien s wi hin g oups commonly ake place. Second, he ood demand sys ems ypically es ima ed a e based on a -home ood expendi u e da a. This means a la ge ca ego y o ood expendi u es ha may a ec obesi y is no cap u ed in he analysis. This ca ego y comp ises 50% o ood expendi u es in he US in 2014 (Uni ed S a es Depa men o Ag icul u e). The assump ion ha an a -home dai y elas ici y adequa ely cap u es he away- om-home dai y elas ici y seems dubious a bes since he die a y habi s o consume s wi h espec o a -home and away- om-home ood a e p obably much di e en . Fo example, ood away- om-home may be conside ed a “ ea ” whe e ypical a -home ules o die a y discipline do no apply. Also, consume s ha e less con ol o nu ien in ake om away- om-home ood han a -home ood. Fu he mo e, e en i a ca ego y in ood demand sys ems es ima ed an away- om-home ood demand elas ici y, his could no be combined wi h nu ien da a, since nu ien da a a e de eloped using a - home ood g oups and no away- om-home ood g oups. Fo example, while he e a e da a on agg ega e a le els in he ood g oup “dai y” o a -home consump ion, he e a e no da a on a p opo ions in he co esponding away- om- home ood g oup. Thi d, ood demand sys ems in o m us abou he ood pu chases by consume s, bu hey may no ell us e y much abou wha is ac ually ea en and his is wha ul ima ely in e es s us o a s udy o obesi y. In o he s wo ds, es ima ing elas ici ies o nu ien s pu chased by consume s, which is wha hese s udies ac ually do, may ell us li le abou obesi y. Kan o e al. (1997) es ima e ha 26% o ood pu chased by Ame icans is no consumed bu en e s he economy as ood was e. Encou aging consume s o ea mo e esh ui s and ege ables as a way o educe obesi y may no be as e ec i e as hypo hesized i 32% o pu chased esh ui s and ege ables a e no consumed bu was ed (Kan o e al., 1997). Fu he mo e, ood was e o esh ui s and ege ables may be much highe a he ma gin. The e o e, combining an a -home ui s and ege ables elas ici y o demand in i xed p opo ions wi h he a e age nu ien con en in ui s and ege ables p obably o e es ima es he impac on obesi y o a pa icula nu ien . Cla k and Le edahl (2006) and Xu (2010) show how he p oblem o i xed p opo ions can be o e come empi ically by eg essing nu ien s on ood g oup p ices. These es ima ions gene a e a se o elas ici ies o nu ien esponse o ood g oup p ices. Va iable p opo ions allows o wi hin-g oup subs i u ions o nu ien s and he model does no use ood expendi u e da a o gene a e elas ici ies o nu ien s o ood g oup p ices and so nu ien quan i ies include bo h ood a -home and away- om-home ood g oups. While he a iable p opo ions app oach seems p omising, i has no pe o med pa icula ly well empi ically in ea lie s udies. Fo example, Xu (2010) does no i nd subs an ial s a is ical signi i cance om es ima ing a , ca bohyd a e, and calo ie elas ici y es ima es om ood g oup p ices using US da a. While Cla k and Le edahl (2006) do i nd a s a is ically signi i can ela ionship be ween a and mea p ices, he mea g oup is no pa icula ly s iking as a majo sou ce o a causing obesi y. Nei he o hese s udies includes nu ien p ices in he analysis. Also, hese es ima es only include elas ici ies o ood disappea ance o nu ien s EM_1_2019.indd 25EM_1_2019.indd 25 8.3.2019 9:15:498.3.2019 9:15:49 26 2019, XXII, 1 Economics o p ices, and no es ima es o nu ien s ea en, and he e o e was e is no conside ed. Bea y (2007) uses an en opy app oach o gene a e shadow p ices o nu ien s bu he es ima es a e s ill based on only ood a -home demand elas ici ies. Ano he popula app oach is o measu e he heal h ou come di ec ly a he han indi ec ly h ough changes in some obesogenic nu ien , such as a om ood pu chased o a -home consump ion. Fo example, he e a e se e al s udies ha ela e he BMI o popula ions o a ious socio-economic a iables including p ices o ele an goods (e.g., Chou e al., 2004; Boizo -Szan ai, & Fab ice, 2005; S um & Da a , 2005; 2008; Auld & Powell, 2009; Goldman, Lakdawalla, & Zheng, 2009; Cla k, e al., 2014). Chen e al. (2002) use simila me hods o examine blood p essu e. A common ea u e o he BMI da a used in hese s udies is ha in o ma ion on nu ien in ake is una ailable, and so p ices a e included in place o quan i ies. A basic ques ion o hese s udies, and seldom discussed, is wha p ices ough o be included in BMI s udies? Two ypes o p ices a e examined in his s udy: nu ien p ices and ood g oup p ices. Chen e al. (2002) demons a e ha he use o ood g oup p ices esul s in es ima ion o he educed o m model and nu ien p ices es ima e he s uc u al model. Which p ices bes cap u e nu ien s in die s as hey ela e o obesi y is discussed in Cla k and Di ich (2010) who a gue ha nu ien p ices be e cap u e he e ec s o axes o con ol obesi y han ood g oup p ices. Bea y (2007) implies ha he app op ia e p ice is a nu ien p ice in his s udy o de elop shadow alues o nu ien s. Chen e al. (2002) de elop a heo e ical model in which blood p essu e is he heal h ou come and conclude ha nu ien p ices a e app op ia e o es ima e he s uc u al model which in u n iden i i es pa ame e s ha can be used om a policy pe spec i e. The ood p ice model is a educed o m model om which he coe i cien es ima es a e subjec o con ounding o nu ien e ec s wi hin and ac oss ood g oups. Chen e al. (2002) call nu ien p ices “implici ” p ices in hei model, and his e m highligh s he ac ha nu ien p ices a e no eadily a ailable. This could be one eason why hey es ima e a model including ood g oup p ices a he han a model including nu ien p ices. Goldman, Lakdawalla and Zheng (2009) show how da a on ood expendi u es, p ices o ood g oups, and ood disappea ance o nu ien s can be used o gene a e nu ien p ices ( o a -home ood consump ion). In his s udy we es ima e wo models, one including nu ien p ices and he o he including ood g oup p ices, o compa e he empi ical pe o mance o he wo app oaches. 3. Me hods and Da a The me hodology is based on he p e ious wo k o Cla k e al. (2014). Da a on BMI o Canadian adul s (eigh een yea s o age o olde ), socio-economic da a and o he da a ela ed o indi idual heal h s a us o indi iduals is a ailable om he pe sonal cha ac e is ics da a ha comes om Canadian Communi y Heal h Su ey (CCHS) (S a is ics Canada, 2012b) and he pe sonal heal h indica o s o Canadians (S a is ics Canada, 2012c). The CCHS is an annual su ey con aining in o ma ion on heal h s a us, heal h ca e u iliza ion and heal h de e minan s o he Canadian popula ion g ea e han wel e yea s old. The a iables included in his s udy om he CCHS su ey included BMI, income, educa ion, age, gende , household size and a a ie y o heal h s a us indica o s (see Cla k e al., 2014, pp. 252-253). Cla k e al. (2014) a e ci ing Goldman, Lakdawalla and Zheng (2009) who de elop a o mula ha shows how ood disappea ance da a on nu ien in ake can be combined wi h ood expendi u e da a o de elop a p oxy o a nu ien p ice. Speci i cally, hei o mula is (1) whe e NPli is nu ien p ice o he l h nu ien ( a , ca bohyd a es, p o ein) o he i h p o ince, (New oundland, P ince Edwa d Island, No a Sco ia, New B unswick, Quebec, On a io, Mani oba, Saska chewan, Albe a, B i ish Columbia and he e i o ies – Nuna u , Yukon and he No hwes Te i o y – included in a “p o ince” called No h Canada), Sij is he expendi u e sha e o he j h ood g oup (milk; cheese; eggs; ui ; ce eals; ege ables; bee ; po k; poul y; i sh; suga and swee ene s; a s and oils; and co ee and ea) in he i h P o ince, expendi u e sha e is expendi u e sha e o ood g oup j in P o ince i pe capi a pe day (S a is ics Canada, 2012d; 2012e), Nl EM_1_2019.indd 26EM_1_2019.indd 26 8.3.2019 9:15:508.3.2019 9:15:50 27 1, XXII, 2019 Economics is nu ien ood disappea ance o he l h nu ien pe capi a pe day (S a is ics Canada, 2012c) and CPIij is consume p ice index o ood g oup j in P o ince i (S a is ics Canada, 2012 ). In he ood g oup p ice model, he CPIij o P o ince j o he ollowing ood g oups a e used: dai y, eggs, ui s and ege ables, eggs, ce eals, ed mea s, poul y, i sh, suga and swee ene s, a s and oils and es au an ood (S a is ics Canada, 2012 ). (see Cla k e al., 2014, p. 252). No e ha a -home ood g oup expendi u e sha e da a a e used o gene a e nu ien p ices. This means ha p ices a e gene a ed o he a -home ca ego y o nu ien p ices and do no include he away- om-home sha e o he nu ien in ood. This means ha measu es o nu ien p ices used in his s udy a e dependen upon he ood py amid agg ega ion sys em. All o he da a a e o he yea o 2010, excep he da a on nu ien ood disappea ance (S a is ics Canada, 2012c), which is no a ailable o 2010. We use he 2009 nu ien ood disappea ance da a in Canada as a p oxy o 2010. While i is no possible o know how closely he 2009 da a p oxies o 2010 da a, da a om The Cen e o Nu i ion and Public Policy in he US indica e ha nu ien ood disappea ance pe capi a changes e y slowly. A e dele ing he obse a ions ha we e eco ded as unknown, no applicable, e used o answe o no s a ed, he sample consis s o esponses om 39,805 Canadians eigh een yea s o olde (see Cla k e al., 2014). Food expendi u e da a and CPI da a we e no a ailable o he No h egion, and so we use a popula ion-weigh ed p ice o de e mine he CPI in No h Canada and ega d he a e age Canada expendi u e sha e as he expendi u e sha e in No h Canada. Since so ew o he esponde s o he CCHS we e om he No h egion, he esul s did no change app eciably when No he n egion’s esponses we e excluded om conside a ion. These esul s a e no shown, bu a ailable on eques . The da a on p ices gene a ed om equa ion (1) a e combined wi h BMI and o he socio-economic and heal h s a us da a o es ima e he BMI equa ions (see Cla k e al., 2014, p. 252). The empi ical model es ima ed is: (2) whe e BMIki is body mass index o indi idual k in P o ince i (S a is ics Canada, 2012b), PCki is a ec o o pe sonal cha ac e is ics da a o indi idual k in P o ince i, Pi is a ec o o nu ien p ices o a ec o o ood g oup p ices in P o ince i and β and γ a e ec o s o pa ame e s o be es ima ed (Cla k e al., 2014). 4. Resul s and Obesi y Tax Assessmen The model is es ima ed in linea na u al loga i hm o m, meaning ha pa ame e es ima es a e elas ici ies. Tab. 2 p esen s O dina y Leas Squa e (OLS) es ima es o he educed o m ood g oup p ice model (elas ici y es ima es in he 2nd column and - alues in he 3 d column) and he s uc u al nu ien p ice model (elas ici y es ima es in he ou h column and - alues in he i h column) esul ing om es ima ing he OLS es ima es o he BMI o Canadians in 2010. Obesogenic nu ien p ices and ood g oup p ices ha e nega i e coe i cien s and indica e ha a ax will lowe he BMI o Canadians and lep ogenic nu ien p ices and ood g oup p ices ha e posi i e coe i cien s and indica e ha a subsidy will lowe he BMI o Canadians. As app op ia e de ices o ensu ing o he pa ame e es ima es in he model a e no biased, bo h models gi e simila esul s. The elas ici y es ima es o all o he a iables included in each model excep o p ice a iables a e qui e simila , excep pe haps he household size coe i cien , which is insigni i can in bo h models. The a ge ing p oblem discussed in he p e ious sec ion is bo ne ou in he esul s when we compa e he es ima ed elas ici ies o he p ice coe i cien s be ween he wo models: nu ien p ices a ge he sou ce o he nu ien impac on BMI be e han do ood g oup p ices. Based on s a is ical signi i cance, he nu ien p ice model is supe io o he ood g oup p ice model, since he p ice o ca bohyd a es, he p ice o p o ein and he p ice o a a e highly s a is ically signi i can . This is in con as o he ood g oup p ice model whe e, o he nine ood g oup p ices included in he model, only h ee, he p ice o eggs, he p ice o a s and oils and he p ice o es au an ood a e s a is ically signi i can . The esul s poin ou he dange s o using g oup p ice elas ici ies in he place o nu ien elas ici ies when making judgemen s conce ning he e ec o public policies on ood ou comes. Fo example, G ay e al. (2005) use EM_1_2019.indd 27EM_1_2019.indd 27 8.3.2019 9:15:508.3.2019 9:15:50 28 2019, XXII, 1 Economics Food G oup P ice Model Nu ien P ice Model Reg esso Es ima e - alue Es ima e - alue Reg esso * In e cep 2.736 33.902 2.782 79.792 Age 0.039 10.831 0.038 10.475 Gende (Male = 1) 0.042 23.002 0.041 22.834 Ma ied (Yes = 1) 0.018 6.048 0.020 6.481 Common Law (Yes = 1) 0.015 4.387 0.014 4.064 Di o ced (Yes = 1) 0.003 0.816 0.003 0.918 Household Size 0.000 0.177 0.001 0.271 Educa ion -0.021 -3.846 -0.023 -4.363 Has Job 0.027 11.201 0.027 11.414 Income 0.006 3.550 0.006 3.459 Sel pe . heal h (5 = e y unheal hy) 0.064 30.078 0.064 30.160 Has As hma (Yes = 1) 0.030 9.500 0.030 9.433 Has A h i is (Yes = 1) 0.027 10.696 0.028 11.104 Has High Blood P essu e (Yes = 1) 0.070 28.357 0.071 28.549 Has Diabe es (Yes = 1) 0.087 24.094 0.087 24.190 Has Hea Disease (Yes = 1) -0.015 -3.598 -0.015 -3.541 Has Cance (Yes = 1) -0.028 -4.597 -0.028 -4.626 Dai y P ice -2.207 -1.004 -0.369 -11.923 Fa P ice F ui s & Vege ables P ice -0.136 -0.337 -0.214 -10.359 Ca b. P ice Red Mea s P ice 0.412 0.634 0.608 12.339 P o ein P ice Egg P ice -2.262 -4.320 Ce eals P ice -0.464 -1.290 Poul y P ice -2.567 -1.782 Fish P ice 0.243 0.483 Suga & Swee ene s P ice -0.521 -1.775 Fa s & Oils P ice 1.629 2.359 Res au an Food P ice -0.647 -2.845 R20.143 0.140 F-S a is ic 255.100 339.800 Sou ce: own calcula ion based on S a is ics Canada No e: *I di e en han ood g oup p ice model. Tab. 2: OLS Es ima es o double log speci i ca ion o he BMI o Canadians, 2010 EM_1_2019.indd 28EM_1_2019.indd 28 8.3.2019 9:15:508.3.2019 9:15:50 29 1, XXII, 2019 Economics an a -home ood expendi u e elas ici y o bu e , which is a de eloped om a sub-u ili y analysis o he a s and oils ca ego y aken om Godda d and Amuah (1989), in place o a a elas ici y o dai y a o analyze he impac he Canadian Milk Ma ke ing Boa d on he incidence o hea disease on Canadians. The Godda d and Amuah (1989) s udy did no conside he kinds o c oss ood g oup subs i u ions he consume s can make when hey buy ood om supe ma ke s, anspo i o hei homes and combine i wi h o he oods o p oduce meals and ood was e. I he heal h ou come we e obesi y a he han hea disease, he closes elas ici y p esen ed in Tab. 2 o ha used by G ay e al. (2005) would be o use ha elas ici y on a s and oils p ice (1.629) in place o he a p ice elas ici y (-0.369) o public policy pu poses. One can see ha he wo bea no esemblance o one ano he . In addi ion, hese elas ici ies only conside a -home consump ion o a s and oils, and no away- om-home consump ion o a s and oils. While he a s and oils elas ici y es ima e seems nonsensical (since i is posi i e and would imply a hin subsidy be placed on his ood g oup), i may no be so absu d when one conside s wha his elas ici y ac ually measu es. The “ a s and oils” ca ego y includes p oduc s like ege able oils used in salads. Lowe ing he p ice o a s and oils could po en ially lead o an inc ease in he pu chases o ege ables as consume s ea mo e salads, leading o a educ ion in BMI. This would mean ha calo ic in ake, which inc eases wi h he educ ion in he a s and oils p ice, is mo e han o se by a educ ion in calo ic in ake wi h he inc ease he consump ion o salads, especially i he inc ease in he consump ion o salads dec eases he consump ion o , say, F ench ies. The ambigui y and con ounded na u e o nu ien s associa ed wi h he use o ood g oup p ices in he educed o m model is demons a ed in he 2nd and 3 d columns o Tab. 2. Nega i e coe i cien s a e indica ed o p ices o dai y, ui s and ege ables, ce eals, eggs, suga and swee ene s, and es au an ood. Howe e , only he eggs and es au an p ice coe i cien s a e s a is ically signi i can . A nega i e coe i cien indica es ha a ax on he ood g oup should be e ec i e in educing BMI and his esul seems coun e in ui i e o ui s and ege ables, ce eals and poul y. The BMI-dai y p ice ela ionship is elas ic wi h an es ima e o -2.20, bu no s a is ically signi i can . The nega i e coe i cien on poul y p ice is unexpec ed, gi en ha many nu i ionis s ha e sugges ed ha whi e mea s a e supe io o ed mea s in p omo ing lowe BMI. Taxing eggs would appea o be an e ec i e way o educe he BMI o Canadians, wi h an elas ici y es ima e o -2.26 and his esul also seems coun e in ui i e gi en ha eggs ha e been iden i i ed as a good sou ce o p o ein. The coe i cien on suga & swee ene s sugges s inelas ici y and a s a is ically insigni i can impac on he BMI o Canadians, indica ing a ax on his ood g oup would ha e a small o negligible impac on educing he BMI o Canadians. The es au an p ice coe i cien is inelas ic and s a is ically signi i can indica ing a ax on away- om-home ood would educe he BMI o Canadians. Food g oup p ice coe i cien s wi h a posi i e coe i cien include ed mea s, i sh and a s and oils, wi h a s and oils ha e he only s a is ically signi i can coe i cien o 1.629. This conclusion seems un easonable bu is a good example o why es ima es coming om ood g oup p ice as he basis o imposing axes o subsidies may no be a good guide o public policy. Elas ici y es ima es and - alues o nu ien p ices a e p esen ed in columns 3 and 4 o Tab. 2 espec i ely. Two obesogenic ( a and ca bohyd a es) and one lep ogenic (p o ein) nu ien a e included in he model. Bo h a and ca bohyd a es elas ici ies a e nega i e and s a is ically signi i can , indica ing hey a e obesogenic nu ien s. P o ein is posi i e and s a is ically signi i can , indica ing a lep ogenic nu ien . These esul s a e consis en wi h a p io i expec a ions. These esul s a e also consis en wi h he model o Chen e al. (2002) who ound ha nu ien p ices mo e accu a ely e l ec he p ice o nu ien s in he die s o consume s as hey ela e o obesi y han do ood g oup p ices. 5. Discussion and Policy Implica ions o Ac ual and Po en ial Food Taxes The esul s highligh he p oblem wi h using ood g oup p ice elas ici y es ima es coming om a educed o m model o obesi y. Many o he ood g oup p ice coe i cien s a e s a is ically insigni i can . The esul s a e consis en wi h a small bu g owing empi ical li e a u e ha EM_1_2019.indd 29EM_1_2019.indd 29 8.3.2019 9:15:508.3.2019 9:15:50 30 2019, XXII, 1 Economics s udies he e ec o p ices on obesi y: nu ien p ices a ge he unde lying nu i ional ac o s he a ec he BMI o Canadians mo e accu a ely han do ood g oup p ices. The esul s o he nu ien p ice model a e a mo e plausible o assess public policy on obesi y axes. Ei he a ax on he a o ca bohyd a e con en o ood o a subsidy on he p o ein con en o ood would educe he obesi y o Canadians; howe e , he elas ici y es ima es a e small. The nu ien p ice in o ma ion could also be used o impose a nu ien index ax on ood, wi h he weigh s o his ax being 0.607 on a , 0.352 on ca bohyd a es and -1 on p o ein (no malizing by he p o ein elas ici y). The esul s indica e ha axes on es au an ood would educe he BMI o Canadians. Gi en ha es au an ood is subjec o sales axes, he esul s indica e ha exis ing axes a e an e ec i e means o educe he BMI o Canadians. Inasmuch as he p ice o es au an ood is a good p oxy o “ as ood” hese esul s suppo hose who ad oca e a “ a ” ax on as ood as an e ec i e me hod o educe obesi y in Canadians. Howe e , he es ima e is inelas ic, indica ing ha his e ec is small and he e could be subs an ial nu i ional con ounding unde lying he es ima e ha make i an un eliable as a guide o policy wi hou some u he elabo a ion conce ning he nu i ional link be ween which oods ha a e “ as ” a e obesogenic and no lep ogenic. The coe i cien s on he egg p ice, poul y p ice, and dai y p ice se e as a guide o he impac o implici ood axes om supply- managed indus ies. The able indica es ha axes on eggs a e an e ec i e means o educing he BMI o Canadians and he e ec is bo h s a is ically signi i can and elas ic. This esul is also ue o he implici ax on poul y om supply managemen al hough his esul is s a is ically insigni i can . Bo h o hese esul s a e con a y o popula pe cep ion, whe e egg consump ion is o en encou aged due o i s high p o ein con en and poul y consump ion is encou aged due o i s low a con en ela i e o o he mea s. This is a case o con ounding which a ises in he use o measu es o elemen a y ood consump ion ins ead o meal consump ion (o , e en be e , nu ien consump ion). Fo example, i could be ha eggs a e almos always cooked using a high a me hod and consumed wi h o he high calo ic oods such as b ead, po a oes, and ed mea s. One could hink o simila con ounding o poul y-mea based meals e l ec ed in he es ima ed poul y p ice elas ici y. The peculia na u e in which dai y supply managemen ope a es implies ha he esul s o bo h he ood g oup p ice and he nu ien p ice model can be used o assess he implici axes o he dai y supply managemen sys em. Dai y supply managemen alloca es quo a (and hus es ic s supply) on he basis o he a con en o milk. This aspec o dai y supply managemen es ic s he supply o a in he sys em and inc eases he p ice o a in ood. The BMI-elas ici y o a p ice o -0.369 is he ele an elas ici y o his policy: a a ax is BMI- educing. The dai y supply managemen sys em also inc eases he p ice o dai y p oduc s o consume s. This would lead o a educ ion o a -home dai y p oduc s pu chased and he es ima ed elas ici y o -2.207 o dai y p ice would sugges ha his, oo, would lead o a lowe BMI o Canadians. Howe e , he es ima ed dai y ood g oup p ice elas ici y is no s a is ically di e en om ze o. Recommenda ions ela ed o heal hy ea ing, o example, o ea mo e esh ui s and ege ables and less a s and oils, a e e l ec ed in changing ela i e p ices in he ood g oup p ice. While his may be e ec i e in he a -home ca ego y o ood o in l uence die a y choices, i may no be e ec i e o he away- om-home ood ca ego y and his ca ego y is a ising p opo ion o consume expendi u es on ood. The shi in consume p e e ences owa ds ood away- om-home means ha he ade-o s in ood ca ego ies ep esen ed by ood py amid p ices may no help in he unde s anding o he ise in obesi y which may be d i en mo e by inc eased consump ion in he away- om- home ca ego y compa ed o he a -home ood ca ego y. On he o he hand, he use o nu ien axes and subsidies would p esumably a ec bo h he a -home and away- om-home ca ego ies o ood pu chase beha io o consume s and he beha io o p oduce s as well, when hey adjus p opo ions o nu ien s in ood when ela i e nu ien p ices change. The p e e ence o he use o nu ien p ices in s udies and nu ien axes in p ac ice e l ec s a desi e o be able o mo e ully unde s and and in l uence he choice o ood a -home e sus ood away- om- home as well as he choice be ween aw and u he -p ocessed ood a supe ma ke s which EM_1_2019.indd 30EM_1_2019.indd 30 8.3.2019 9:15:508.3.2019 9:15:50