BD TaxReport Final Regular Format

8/8/2019 BD TaxReport Final Regular Format

1/21

Taxation of Tobacco Productsin Bangladesh

Fidis fm th 2009 ITC Baadsh Suvy

AprIl 15, 2010

The International Tobacco Control Policy Evaluation Project

ITC projeCT WorkIng pAper SerIeS

nia nais ad Ummu Hasaath ruthbah

University of Dhaka, Bangladesh

gffy T. F

University of Waterloo and

Ontario Institute for Cancer Research, Canada

Suggested Citation: Nargis, N., Ruthbah, U. H., and Fong, G. T. (April 2010). Taxation of Tobacco Products in Bangladesh: Findings from th

2009 ITC Bangladesh Survey. ITC Project Working Paper Series. University of Waterloo, Waterloo, Ontario, Canada.


2/21

1

TaxationofTobaccoProductsin

Bangladesh:Findingsfromthe

2009ITCBangladeshSurveyNigarNargis,UmmulH.Ruthbah,andGeoffreyT.Fong

April2010

Tobaccouse isa leading causeofdeath anddisability inBangladesh, asit is inthe vast

majority of countries in the world. Currently, there are 41.1 million people who usetobacco, including 20.9 million people who smoke either cigarettes or bidis or both.1

Although an estimated 57,000 people die a year of tobacco use,2 this will climbconsiderablyinthenearfuture.Infact,epidemiologicalestimatesfromIndiasuggestthat

aboutonethirdtoonehalfof the20.9millionpeoplewhosmokewilldieof asmoking

related disease, at an average loss of life between 610 years.3 The level of tobaccoconsumption is moved even higher in Bangladesh by a bottom heavy demographic

structure(57.7%ofthetotalpopulationisagedbelow25yearsandonethirdisbelow15years),widespreadilliteracy(halfoftheadultpopulationisilliterate),andpoverty(about

halfofthetotalpopulationlivesbelowthepovertyline).4Byanystandard,then,tobacco

use representsa criticalthreat tothe health andwelfareof theBangladeshi peopleandstrongactionmustbetakentoavertthepresentandeverdeepeningthreat.

Bangladeshhasahistoryofacommitmenttotobaccocontrol.ItwasthefirstcountrytosigntheWHOFrameworkConventiononTobaccoControl(FCTC)andamongthefirst40

countries to become a Party to the FCTC. The FCTC obligates the Parties to engage intobaccocontrolactionsacrossabroadrangeofpolicydomains,includingmoreprominent

warning labels, bans/restrictions on advertising, smokefree laws, reducing illicit trade,

andincreasesintobaccotaxationandharmonizationoftaxationacrosstobaccoproducts.

In 2005, Bangladesh enacted the Tobacco Control Act (TCA), with the correspondingregulations being implemented in 2006. However, recent evidence from two nationally

representativesurveysconductedin2009theGlobalAdultTobaccoSurvey(GATS) 5and

theInternationalTobaccoControl(ITC)BangladeshSurvey 6havefoundthatdespitetheenactment of the TCA, Bangladesh has experienced an alarming increase in tobacco

consumptionoverthepastfiveyears.

1 ITC Project. ITC Bangladesh National Report. University of Waterloo & University of Dhaka.2

World Health Organization. Impact of Tobacco-Related Illnesses in Bangladesh. World Health Organization,

South East Asia Region, New Delhi, India, 2006.3

Jha, P., et al. A nationally representative case-control study of smoking and death in India. New England Journal of

Medicine, 2008; 358: 1137-47. Personal communication with Prabhat Jha, March 26, 2010.4

Bangladesh National Census, Bangladesh Bureau of Statistics, 2001.5

World Health Organization, Global Adult Tobacco Survey: Bangladesh Report, 2009.6 ITC Project. ITC Bangladesh National Report. University of Waterloo, Waterloo, Ontario, Canada; University of

Dhaka, Bangladesh, April 2010.


3/21

2

Inpart,thelowimpactoftheTCAinfailingtoreducetobaccoconsumptionandprevalenceinBangladeshmaybeduetolowlevelsofenforcementofnontaxmeasuresoftheTCA,

such as the advertising ban and smokefree public places, and relatively low levels ofimplementationofwarninglabels(which,inaccordancewiththemorerecentArticle11

Guidelines,shouldincludegraphicimagesratherthanthecurrenttextonlywarnings).Butweknowthatthemostimportantandpotentiallyeffectivepolicyforreducingtobacco

consumption and prevalence is to set tobacco taxes at a sufficiently high rate and to

harmonizethetaxratesacrosstobaccoproducts.Inthisreport,wedescribefindingsfromourrecentInternationalTobaccoControlPolicyEvaluation(ITC)SurveyinBangladeshthat

demonstratetheurgentneedtotakeactioninthedomainoftaxation.

Bangladeshhasampleroomtoraisetobaccotax

ThecurrentcigarettetaxinBangladeshiscomposedoftwocomponents:avalueaddedtax(VAT) at the rate of15%,and a supplementary duty (SD) that varies atdifferent price

rangesofcigarettesper10sticks,withtherategoingupfrom32%atthelowestpricetierof 7.258.75 Taka, to 52% for 16.2517.25 Taka, to 55% for 23.2529.25 Taka, to a

maximumof57%forpricesabove46.25Taka.7AccordingtotheITCBangladeshSurvey,

conducted in 2009, 13.4% of our nationally representative sample of 2,788 cigarettesmokers purchases cigarettes in the lowest price tier, a majority (71.8%) purchase

cigarettesinthesecondlowesttier,11.0%inthethirdlowesttier,and3.8%inthehighesttier. Based on this distribution of cigarette smokers within each of the price tiers, the

weightedaverageofcigarettetaxratesincludingSDis49.8%.Theweightedaverageofthe

totaltaxrateincludingSD(49.8%)andVAT(15%)is72.3%. 8Weightedaveragepriceofa

packof10sticksofcigarettesis18.15Taka,accordingtotheITCdata.Thetotaltaxthusaccountsfor42%ofthesaleprice.9Thisshareoftaxinsalepricefallsonthelowersideoftheglobalscalerangingfrom0to80%andwellbelowtheWorldBankrecommendationto

settherateattwothirdsofthesalepriceataminimum.10

Table 1 presents the calculations based on ITC data on the impact of increases in

supplementarydutyonthetotaltaxrate,averagesalepriceperpack,increaseinsaleprice,

and the tax percentageof the saleprice, According toour calculations, bringing the taxpercentageuptotheWorldBankrecommendedlevelof67%forcigaretteswouldrequire

anincreaseoftheSDof225%fromitscurrent49.8%to161.9%.

7NationalBoardofRevenue,GovernmentofBangladesh,ValueAddedTax,GeneralOrderNo.10/VAT/2009,11June2009.8(1+0.0498)(1+0.15)1=0.723.9Pretaxprice(exfactoryprice+markupatwholesaleandretaillevel)=18.15/(1+0.723)=10.53Taka.

Amountoftaxperpack=18.1510.53=7.62Taka.Percentageoftaxinsaleprice=(7.62/18.15)x100=42%.10CurbingtheEpidemic:GovernmentsandtheEconomicsofTobaccoControl,TheWorldBank,Washington

D.C.,1999,p.45.


4/21

3

Table1:Projectionoftaxandpriceincreasesforcigarettesandbidis

IncreaseinSD

over2009

(%)

SD

(%)

VAT

(%)

Totaltax

rate

(%)

Averagesale

priceper

pack(Taka)

Increasein

saleprice

(%)

Taxas

percentageof

saleprice(%)

CIGARETTES0 49.8 15 72.3 18.15 0 42.0

50 74.8 15 101.0 21.16 16.6 50.2

100 99.7 15 129.6 24.18 33.3 56.5

150 124.6 15 158.3 27.20 49.9 61.3

200 149.5 15 187.0 30.22 66.5 65.2

225 161.9 15 201.3 31.73 70.8 66.7(WorldBank

recommendedlevel)

250 174.4 15 215.6 33.24 83.2 68.3

300 199.4 15 244.3 36.26 99.8 71.0

350 224.3 15 272.9 39.27 116.4 73.2

400 249.2 15 301.6 42.29 133.1 75.1

BIDIS

0 20 15 38.0 6.00 0 27.5

50 30 15 49.5 6.50 8.33 33.1

100 40 15 61.0 7.00 16.67 37.9

150 50 15 72.5 7.50 25.00 42.0

200 60 15 84.0 8.00 33.33 45.7

250 70 15 95.5 8.50 41.67 48.8

300 80 15 107.0 9.00 50.00 51.7

350 90 15 118.5 9.50 58.33 54.2

400 100 15 130.0 10.00 66.67 56.5450 110 15 141.5 10.50 75.00 58.6

500 120 15 153.0 11.00 83.33 60.5

550 130 15 164.5 11.50 91.67 62.2

600 140 15 176.0 12.00 100.00 63.8

650 150 15 187.5 12.50 108.33 65.2

700 160 15 199.0 13.00 116.67 66.6(WorldBank

recommendedlevel)

750 170 15 210.5 13.50 125.00 68.0

Notes: 1.Totaltaxrate=[(1+SD/100)(1+VAT/100)1]x100

2.Percentageoftaxinsaleprice=(Taxperpack/Averagesalepriceperpack)x100,where:Taxperpack=AveragesalepriceperpackPretaxprice

Pretaxprice(exfactoryprice+markupatwholesaleandretaillevel)=

Averagesalepriceperpack/(1+Totaltaxrate/100).


5/21

4

In case of bidis, currently, an SD at the rate of 20%is added to theVATof 15%.11 The

currenteffective tax rateofbidis isthus38%.12 According toITCdata, the averagesalepriceofapackof25bidisticksis6Taka,ofwhich27.5%istax.13Becausethecurrenttax

rateforbidisissolow,itwouldtakeamuchgreaterproportionalincreaseinSDtoachievetheWorldBankrecommendedtaxpercentageof67%.Specifically,asshownatthebottom

ofTable1,theSDwouldneedtoberaisedby700%fromitscurrent20%to160%.ItfollowsfromthisanalysisthatBangladeshhasampleroomtoraisethetaxrateforboth

cigarettes and bidis. It also shows that to bring the tax rate for bidis up to the same

recommended percentage of sales pricewill require a substantially greater proportionincreaseintaxes.

RaisingtobaccotaxescaneffectivelyreducetobaccouseinBangladesh

Intheprevioussection,wedemonstratedhowthetaxratesoncigarettesandbidismustbe

increasedinordertoconformtotheWorldBankrecommendedtaxlevelsoftwothirdsofthe sale price. In this section, we analyze the implications of taxes in decreasing both

consumption and prevalence of tobacco use in Bangladesh. How would consumers inBangladeshrespondtochangesinthepriceoftobacco?

Inorder to estimate the responsiveness of tobacco consumption to changes inpriceof

tobacco it is necessary to calculate three parameters: (1) elasticity of smoking

participation,(2)conditionalpriceelasticity,and(3)totalpriceelasticity,wherethethirdis a sum of the first two parameters. Specifically, the price elasticity of smoking

participation represents the impact of a price change on the prevalence of smokingresultingfromcessation,lesserrateofinitiationanduptakeofsmoking,andlesserrateof

relapseofquitters.Theconditionalpriceelasticitymeasurestheresponsivenessoftobaccoconsumptiontopricechangeforthosewhosmoke.Totalpriceelasticityisasumofthesetwoandindicatestheoverallimpactofapricechangeontobaccoconsumptionresulting

fromlowersmokingprevalenceaswellaslowersmokingintensityofexistingsmokers.

Pricesensitivityofcigaretteconsumption

Mostestimatesofpriceelasticityofcigarettedemandinlowandmiddleincomecountries

rangefrom0.5to1.0whilethoseforhighincomecountriestendtofallintherangefrom

0.25to0.5.14UsingITCBangladeshenumerationandwave1surveydata,weestimatetheelasticityofparticipationincigarettesmokingat0.29,theconditionalpriceelasticityof

11

NationalBoardofRevenue,GovernmentofBangladesh,ValueAddedTax,GeneralOrderNo.11/VAT/2009,11June2009.12(1+0.20)(1+0.15)1=0.38.13Pretaxprice(exfactoryprice+markupatwholesaleandretaillevel)=6/(1+0.275)=4.35Taka.Amount

oftaxperpack=64.35=1.65Taka.Percentageoftaxinsaleprice=(1.65/6)x100=27.5%.14

Chaloupka, F. J., Hu, T., Warner, K.E., Jacobs, R., and Yurekli, A. (2000). The taxation of tobacco products. In

Tobacco Control in Developing Countries (ed. P. Jha and F.J. Chaloupka), pp. 237-72, Oxford: Oxford University

Press.


6/21

5

cigaretteconsumptionat 0.14,andthe totalpriceelasticityof cigaretteconsumptionat

0.43 (Table 2). This estimateofprice elasticityofcigarette demand is low compared tootherlowincomecountriesintheworld.15Onecanreasonablyarguethatitisthelowsale

priceofcigarettesthatmakesitsoaffordableforcigarettesmokerstobehighlysensitivetopricechangeswithinacloserangeoftheirpurchaseprice.Thevariationofpricewithina

crosssectionofpopulation,therefore,maynotbelargeenoughtoresultinalargedegreeof price sensitivity. We can expect to obtain a larger estimate of price elasticity withlongitudinalobservationsandlargervariationofpriceovertime.

Table2:PriceelasticityofdemandforcigarettesandbidisinBangladesh,2009

Cigarette Bidi

Smokingprevalence(%)(exclusiveanddualusers) 18.7 12.4

Smokingparticipation:Logitcoefficientonprice 0.45***(0.06)

0.53*(0.31)

Elasticityofsmokingparticipation 0.29 0.46

Numberofobservations 43,075 83,356

Priceelasticityoftobaccodemandconditionalonsmoking

participation

0.14***

(0.04)

0.18

(0.12)

Numberofobservations 1,850 597

Totalpriceelasticity 0.43 0.64

Source:ITCBangladeshEnumerationandSurvey,2009.

1. Cigarettedemandfunctionisestimatedusingtheadultmalepopulationofthemainsample.2. Bididemandfunctionisestimatedusingtheadultmaleandfemalepopulationofthemain,tribaland

floatingsample.

3. ***,**,and*implysignificantlydifferentfromzeroatthe1%,5%and10%levelsrespectively.4. Standarderrorsareprovidedintheparenthesesbelowtheestimates.Theseareadjustedfor

clusteredsamplingdesignofITCBangladeshenumerationandsurvey.

15One reason for such a low value of elasticity could be the endogeneity of self-reported cigarette pricethat can bias the price coefficient in the demand equation downward. It can happen when smokers choose

the particular brand of cigarette for its price so that they can smoke some chosen number of sticks per

day. In our survey, 55% of cigarette smokers mentioned that they chose the brand they regularly smoke

for its price, among other things such as taste, quality, perceived health effect, and popularity. However,

the Hausman specification test between ordinary least squares estimate using self-reported price and two-

stage least squares estimate using predicted price from first stage estimation does not support the

contention that self-reported price is endogenous.


7/21

6

Thefindings ofprice elasticityofdemand forcigarettes have important implicationsfor

publichealth.Forasmoker,positivehealthbenefitsarerealizedtoa(much)greaterextentfor quitting than for reducing consumption. And thus, the proportion of the totalprice

elasticitythatisaccountedforbyelasticityofparticipationisrelevanttoanunderstandingoftheimpactofincreasingpriceonquitting.TheWorldBanktreatisefindsthatoverallfor

highincomecountries,theproportionofthetotalpriceelasticitythatisaccountedforbyelasticityofparticipationisabout50%.FromouranalysisofITCBangladeshSurveydata,theproportionis.29/.43=67%.Thus,iftaxeswereincreasedoncigarettesinBangladesh,

therewouldbegreaterimpactonreducingtheprevalencerate(leadingtogreatergainsin

enhancinghealthatthepopulationlevel)thanwouldbethecaseinmostothercountries.

Pricesensitivityofbidiconsumption

Using ITC data, we estimate the elasticity of participation in bidi smoking at 0.46,

conditionalpriceelasticityofbidi consumptionat 0.18 and totalprice elasticityofbidiconsumptionat0.64(Table1).Theconditionalelasticityisnotstatisticallysignificant.The

lowconditionalelasticityimpliesthatbidismokerswhocontinuetosmokeareexpectedtobe relatively unresponsive to price increases in terms of reduction of their currentconsumption. The estimate of participation elasticity for bidis is, however, statistically

significantandhigherthanthatofcigarettedemand.Thisisbecausebidisareconsumedmostlybylowerincomesmokerswhoalsotendtobemorepricesensitive(showninalater

sectioninthisreport).Thisresultindicatesthatanincreaseinbiditaxeswouldbemore

effective in reducing bidi consumption by reducing prevalence. The proportion of totalpriceelasticityduetoelasticityofparticipationis .46 / .64=72%,higherthanthe67%

foundearlierforcigarettes.

Thus,increasingtaxesonbidiswouldhaveagreaterimpactonreducingconsumptionand

agreaterproportionalimpactonreducingprevalence,relativeto thesameproportionalincreaseintaxesoncigarettes.

Effectofpriceincreaseoncigaretteandbidiconsumption

To further explicate the beneficial impact of tax increases on cigarettes and bidis, weprovideanextended exampleof the impact of increasing specific taxesoncigarettesby

50%and increasing specific taxesonbidisby50%.Based on the existing cigarette taxstructure, weestimate that a 50% increase in the supplementaryduty on domestically

manufactured cigarettes at each price tier would raise cigarette prices by 16.6% on

average.Totaltaxwouldaccountfor50.2%ofaveragesalepriceperpackofcigarette.Itis

assumedthattheincidenceofthistaxincreaseisfullypassedontothesmokersandthemarkupovertheexfactorypriceatthewholesaleandretaillevelsisunaltered.Giventheabove estimates of price elasticity of cigarette demand it can be shown that this 50%

increaseinsupplementarydutyonfactorymadecigaretteswouldresultina4.8%(0.29x

16.6%)decrease incigarettesmokingprevalence and 2.3% (0.14x 16.6%)decrease indaily cigarette consumption of those who continue to smoke. The total effect is thus

estimatedat7.1%decreaseinannualcigaretteconsumption.


8/21

7

Thisanalysismeansthatincreasingthesupplementarydutyofcigarettesby50%ateach

pricetierwouldleadtoadecreaseof856,000cigarettesmokersfromthecurrentpoolof20.9millionsmokers, ofwhom17.9millionare cigarette smokers(including thosewho

smoke both cigarette and bidi). Furthermore, the current volume of annual cigaretteconsumptionof6,534millionpacksof10cigaretteswoulddecreaseby466millionpacks,

from the drop incigarette smoking prevalence aswell as the decrease incurrent dailyconsumptionofcigarettesbycontinuedsmokers.

Turningnowtobidis,a50%increaseinthesupplementarydutyondomesticallyproduced

bidisfromthecurrentrateof20%perpackof25sticksofbidiswouldincreasethesalepriceofbidis(whichsellsfor6Takaperpackonaverage)by8.3%,providedthatthistax

increasewouldbefullypassedontothesmokersandthemarkupatthewholesaleandretaillevelsovertheexfactorypriceremainsunchanged.Itwouldincreasetheshareoftax

inthesalespriceofbidisfrom27.5%to33.1%.Giventheaboveestimatesofpriceelasticity

ofbididemand,this50%increaseinsupplementarydutyonbidiswouldresultina3.8%(0.46x8.3%)decreaseinbidismokingprevalence.Althoughstatisticallyinsignificant,we

consider the effect of price increase on bidi consumption conditional on smokingparticipationtoestimatea1.5%(0.18x8.3%)decreaseindailyconsumptionofthosewhocontinuetosmokebidis.Thetotaleffectisthusestimatedat5.3%decreaseinannualbidi

consumption.

This analysis means that a 50% increase in the supplementary duty on domestically

producedbidiswouldleadtoadecreaseof456,000bidismokersfromthecurrentpoolof20.9millionsmokersofwhom11.9millionarebidismokers(includingthosewhosmoke

bothcigaretteandbidis).Furthermore,thecurrentvolumeofannualbidiconsumptionof2,379millionpacksof25bidiswouldbelessenedby127millionpacks,fromthedropin

bidi prevalence as well as the cut in current daily consumption of bidis by continued

smokers.

We set this extended example at a relatively low ratewhich would increase the tax

percentage of totalpriceupto50.1% for cigarettes, and 33.2% for bidis, far below theWorldBankrecommendationof67%.Still,suchanincreaseintaxes,thoughmodest,would

represent an important first step inanoverall longterm strategyof increasing taxationthatwouldeventuallybringthetaxationratesuptothatrecommendedlevel.Weoutline

suchastrategylaterinthispaper.

The effects on cigarette and bidi consumption corresponding to different percentage

increasesinspecifictaxes(supplementaryduty)arepresentedinTable3.


9/21

8

Table3:Projectedeffectoftaxandpriceincreaseoncigaretteandbidiconsumption

inBangladesh

IncreaseinSD

over2009 SD(%)

Increasein

saleprice(%)

Decreasein

numberof

smokers

Decreaseintotal

consumption

(Millionpacks)

CIGARETTES

0 49.8 0

50 74.8 16.6 856,096 466

100 99.7 33.3 1,712,191 932

150 124.6 49.9 2,568,287 1,399

200 149.5 66.5 3,424,382 1,865

250 174.4 83.2 4,280,478 2,331

300 199.4 99.8 5,136,574 2,797

350 224.3 116.4 5,992,669 3,263400 249.2 133.1 6,848,765 3,729

450 274.1 149.7 7,704,860 4,196

500 299.1 166.3 8,560,956 4,662

550 324.0 182.9 9,417,052 5,128

BIDIS

0 20 0

50 30 8.33 456,250 127

100 40 16.67 912,500 254

150 50 25.00 1,368,750 381

200 60 33.33 1,825,000 508

250 70 41.67 2,281,250 634

300 80 50.00 2,737,500 761

350 90 58.33 3,193,750 888

400 100 66.67 3,650,000 1,015

450 110 75.00 4,106,250 1,142

500 120 83.33 4,562,500 1,269

550 130 91.67 5,018,750 1,396

600 140 100.00 5,475,000 1,523

650 150 108.33 5,931,250 1,650

700 160 116.67 6,387,500 1,776

Notes: 1. Decreaseinnumberofsmokers=IncreaseinsalepricexPriceelasticityofcigarettesmoking

participationxEstimatednumberofsmokersin2009.

2. Decreaseintotalconsumption=Increaseinsalepricex(Priceelasticityofsmokingparticipation+

Priceelasticityofconsumptionconditionalonparticipation)xEstimatedtotalconsumptionin

2009


10/21

9

Dualusageofcigarettesandbidis

Theanalysisaboveconsidersthepriceresponsivenessofcigarettesandbidisindependentof each other. However, a large portion of the smokers are observed to smoke both

cigarettesandbidis.IntheITCdata,thesedualusersaccountfor9.3%oftheenumeratedpopulationrepresenting8.8millionsmokersofthenationalpopulation.Moreover,halfofthe cigarette smokersand threefourthsof the bidi smokers are dual smokers, and that

about half of the smokers in Bangladesh frequently switch between cigarette and bidi

consumption.Hence,demandforcigarettesisnotindependentofdemandforbidisandtheseparate equations estimated for cigarette and bidi consumptionmay bias the results.

Considering this interdependence, we estimate a single demand function for bothcigarettesandbidis.

Inthisfunction,weconsiderdualsmokersasbidismokersastheirdailysmokingintensityisgreaterinbidiconsumption.Theaveragedailybidiconsumptionofdualsmokersis13.9

sticks,while their averagedaily cigarette consumption is5.8sticks.Hence,weusetheirreportednumberofbidismokedperdayalongwiththenumberofbidissmokedperdayfor exclusive bidi smokers and the number of cigarettes smoked per day for exclusive

cigarettesmokersasthedependentvariable.Similarly,theirreportedpriceofbidisisusedasanindependentvariablealongwiththepriceofbidisforexclusivebidismokersandthe

priceofcigarettesforexclusivecigarettesmokers.

Theelasticityofparticipationinsmokingcigarettesorbidisisestimatedtobe0.34,and

the elasticity of conditional demand for cigarette or bidi is 0.12. Thus the total priceelasticityofdemandforcigarettesandbidisisestimatedat0.46.Thegreaterelasticityof

overall smoking participation compared to cigarette use is driven by the much greater

elasticityofparticipationofbidismokers,asshownintheprevioussectionsoncigaretteandbididemandequationsestimatedseparately.

Our finding about the large scale usage of both cigarettes and bidis suggests that thepotentialofswitchingbetweenthesetwoproductsishighintheaftermathofchangesin

theirrelativeprices.Smokersderivingsatisfactionfromeithercigaretteorbidismoking,whichever iseconomicalatthe pointofpurchase, are expected tobepersistent intheir

smoking habit and hence less likely to quit or to reduce their consumption. This

compensatingbehaviorofsmokersislikelytodampentheintendedeffectofataxincreaseandthusunderminethepublichealthgoalsoftobaccocontrol.

Inviewof thepossibilityofproductswitching inresponseto relative pricechanges,theprojectionoftheeffectofataxincreaseonsmokingprevalenceandtobaccoconsumptionis

notasstraightforwardasitwasincaseofindependentestimationforcigarettesandbidis.Itisnecessarytoaccountforthepotentialofproductswitchinginestimatingtheeffecton

priceincreasesontheprobabilityofsmokingbyanindividualaswellashis/hersmoking

intensity.Moreover,itislikelythatsmokersswitchtocheapercigarettebrandsinthefaceof rising expenditureand thus are able tomaintain their current smoking participation

withequalintensity.


11/21

10

Inordertoidentifyindividualswhoswitchoverproductsorbrandsofthesameproducttomaintaintheirsmokinghabit,weneedtoobservetheirtobaccoproductpurchasepattern

overtime.Thecrosssectionalstudiesontobaccodemandarelimitedbythelackof theselongitudinal data for a particular society. The ITC Bangladesh Survey, however, is a

longitudinalcohort survey,anduponcompletionofsecondwave (2010)and thirdwave(2011) surveys, we will be able to investigate this important aspect of tobaccoconsumptionbehavioramongthecohortofsmokersinBangladeshandwillthusbeableto

estimatetheimpactofincreasedtobaccotaxesontobaccouseinamoreprecisemanner.

Harmonizationoftaxacrosscigarettesandbidis

Comparing the independent estimates of price sensitivity for cigarette and bidi

consumption,wefindthattheabsolutedecreaseinbidiconsumptionaftera50%increase

in supplementary duty on bidis would be much less than in the case of cigaretteconsumption,despitelargerestimatesofpriceelasticityforbidiconsumption.Thisisdue

tothefactthattheproportionoftaxinthesalespriceofbidisis27.5%comparedto42%for cigarettes. It indicates that tobring forthcomparabledecrease incigarette and bidiconsumption, the supplementary duty on bidi should be brought at par with that on

cigarettesintermsofthepercentofsaleprice.Itisonlythenthattheuniformpercentageincreasein the ad valorem tax rates, suchassupplementaryduty,across cigarettes and

bidiswouldnotincreasetherelativepriceofcigarettesinfavorofbidismokingandwould

bringdownbothcigaretteandbidiconsumptionsimultaneously.

Theharmonizationoftaxisofgreatimportancefortheeffectiveimplementationoftobaccotaxation. The dual usage of cigarettes and bidis and the habit of product switching by

smokers,asmentionedintheprevioussection,isenhancedbythelargetaxdifferentialthat

existsbetweencigaretteandbidiinBangladesh.Itappearstobeamajorweaknessinthepresenttobaccotaxstructurein Bangladesh.Asmentionedearlierin thisreport, thetax

rateappliedtobidismustbeincreasedatafargreaterratethanthatofcigaretteforthe

convergenceofpercentageshareoftaxinsalepriceforthesetwoproducts650%forbidiand200%forcigarettesforthissharetobeabouttwothirdsofsalepriceofeitherproduct.

Atthatpoint,theSDratewouldbeabout150%.

Itshouldalsobepointedout thattheaveragepriceofbidis(6Taka)isveryclose tothe

lowestpricetierofcigarettes(7.25to8.75Taka)specifiedintheprogressivetaxstructurefor cigarettes. The continuity of price level from bidis to cheaper brands of cigarettes

potentiallycontributes to thefrequent switchingofdual smokersbetweencigarette and

bidi.Inthesecircumstances,increasingthecigarettetaxrateatthelowesttierwouldmakethe corresponding brands relatively expensive and make the smokers of these brands

switch to bidis in greater proportions, undercutting their probability of quitting orloweringtheirsmokingintensity.Thustherewouldbenopublichealthgain.Itmighteven


12/21

11

result in further loss of welfare as bidis are found to be more harmful to health than

cigarettes.16

TheGovernmentofBangladeshcansubstantiallyincreasetaxrevenueby

raisingtobaccotaxes

Theinelasticdemandforcigarettesandbidis(elasticityislessthan1)suggeststhatagivenincreaseintobaccopriceswouldleadtoalessthanproportionatedecreaseinconsumption

leadingtoanincreaseintobaccoexpenditureofsmokersandtherebygeneratinggreatertaxrevenue.HencetheGovernmentwouldgainsignificantlyfromataxincreasethatraises

theprice.Accordingtoourestimation,aninitial50%increaseinSDineachoftheprice

tierswouldresultin16.8%increaseincigarettetaxrevenuefromdomesticallyproducedcigarettes.This increase corresponds toan increaseof845croreTaka to the estimated

currentlevelofcigarettetaxrevenueof5,021croreTaka.17

Figure 1 shows the annual level of cigarette tax revenue corresponding to percentageincreases in SD at successively higher rate beginning from the current state to 550%increase.Theprojectionrevealsthatthereisa limittotheexpansionoftaxrevenuefrom

increasing tobacco taxes. The tax revenuefromraising cigarette taxmay increase upto

350% increaseintax rate.For further increaseintax ratebeyond 350%of the currentlevel,revenuestartstofall,althoughitwillcontinuetobehigherthanthecurrentlevelof

collectionforsomerangeoftaxincrease.Ourcalculationshowsthatcigarettetaxrevenue

ismaximizedwhentherateofSDis224%.IncludingtheVAT,thetotaltaxratewouldbe273%and the totaltaxwould represent73.2%ofthesalesprice.Thiswouldresult ina

116%increaseinthesalepriceofcigarettesthatis,amorethandoublingofthesalepricefromthecurrentlevel.Themaximumtaxrevenueisprojectedtobealmosttwiceashighas

thecurrentlevel.Thelimittotheincreaseintaxrevenuefromincreasedtaxratesisposedbythecontraction

of the size of the market from falling smoking prevalence that causes tobaccomanufacturerstocutdownproduction.Thefallintaxrevenuefromlowerproductionlevel

in turn outweighs the increased tax collection from the continued smokers. Thus the

demand reduction strategy of raising tobacco tax has proved to be very effective inreducingtobaccoproductioncomparedtosupplysidedisincentives.

Atthispointofmaximizationoftaxrevenue,thenumberofsmokerswillalsodecreasebyabout6million,andcigaretteconsumptionwillbelowerby3,263millionpacks(Table3).

16

Prakash C. Gupta, Mangesh S. Pednekar, D.M. Parkin, and R. Sankaranarayanan, Tobacco associated mortalityin Mumbai (Bombay) India. Results of the Bombay Cohort Study, International Journal of Epidemiology, 2005;

34: 13951402.

17 1 crore = 10 million.


13/21

12

Figure1:Annualcigarettetaxrevenue(CroreTaka)fromgivenpercentageincrease

insupplementarydutyfrom2009rate


Inotherwords,thecalculationsusingdatafromtheITCBangladeshSurveydemonstratethat increasing the total tax rate to the pointwhere cigarette tax revenueismaximized

wouldalsoleadtoadecreaseof6millionsmokers.Thisisacompletewinwinsituation.

In the sameway,weproject that for price elasticity ofdemand for bidis inBangladeshestimatedinthisstudy,a50%increaseintheexistingSDonbidiswouldleadtoa23.4%

increaseinbiditaxrevenuefromtheexistingestimatedlevelof395croreTaka.Usingthesamemethodasincaseofcigarettetaxrevenue,wemeasurethattaxrevenuefrombidican

bemaximizedatalevelof834croreTakaifSDisincreasedby500%withincreaseinsalepriceofbidiby83.3%.Itwouldresultin4.6millionlessbidismokersand1,269million

lessofbidipacksconsumed(Table3).

As shown inTable 1, the shareof tax corresponding to themaximumbidi tax revenue

(500%increaseinSD)is60.5%,whichislowerthantherecommendedproportionof67%.Incontrast, for cigarettes, the shareof tax correspondingto themaximumcigarette tax

revenue (350% increase in SD) is 73.2%, which is higher than the recommendedproportion.Thiscontrastpointstothefactthatthelimittotheexpansionofrevenuefromtaxing bidis is reachedearlier than that from taxing cigarettes.This occurs because the

smokingprevalenceofbidismokersismoresensitivetopriceincreasessothatthenumber

ofcontinuingbidismokersistoolowtobeabletosustaintherisingtrendintotalrevenuecollection by paying larger amount of tax. This point is further elaborated in the next

section.

5021

5866

6972

7850

85118970

92399330925890358673

8186

0% 50% 100% 150% 200% 250% 300% 350% 400% 450% 500% 550%

Cigare'etaxrevenue(croreT

aka)

%increaseinsupplementaryduty


14/21

13

Tobaccotaxationandthepoor

Itiswidelyrecognizedthatthepoorerpeoplearemoresensitivetopricechangesthanthe

richer,becausetheopportunitycostofagivenincreaseinexpenditureisgreaterforthepoor.ThispredictionisconfirmedintheITCBangladeshSurvey.Usinghousingindexfor

stratificationofrespondentsintolow,medium,andhighsocioeconomicstatus(SES),weestimategreatersensitivityofthelowerSESpeoplewithrespecttoincreasesinpricesas

reflectedinthetotalpriceelasticityfiguresinthebottomrowofTable4forcigarettesand

bidis. The inverserelationshipofpriceresponsivenesswithSES isdrivenmostly bythegreaterlikelihoodthatpoorpeoplewillquitorabstainfromsmoking,atthesametimeas

they aremore likely tobesmokers inthe firstplace. It isreflected intheir larger price

elasticityof smokingparticipationvisvistheirgreatersmokingprevalenceasshowninTable4.

The price elasticity of cigarette consumption conditional on participation is, however,

insignificant for the low SES and greater among themedium SES than the high SES. It

indicatesthatthoselowSESsmokerswhocontinuetosmokearethemostaddictedandareinsensitive toprice increases.A similarpatternisobserved forbidi smokingprevalence

and daily consumption. Between the moderate and high SES smokers, however, the

conditionalelasticityofbidiconsumptionisgreateramongthehighSESsmokers.

Thecomparisonofpriceelasticitiesbetweencigarettesandbidisfurtherreflectsthatbidismokersaremorepricesensitivethancigarettesmokers.Thisdifferenceisexplainedby

thelargepricedifferentialbetweencigarettesandbidisthatmakebidismoreaffordableto

smokersatlowerSESwhoarealsomorepricesensitive.Moreover,bidismokersatthelowSESdemonstrateexceptionalpricesensitivitytheirdemandforbidisispriceelastic(total

priceelasticityisgreaterthan1).Theirelasticdemandismostlyexplainedbytheirhigh

probabilityofquittingorabstentionfromsmokinginthefaceofapriceincrease(elasticityofsmokingparticipationis0.90).

Asaresultofhavingthehighestpriceresponsivenessofsmokingprevalence,peoplefrom

thelowSESasapopulationgroupwouldbenefitmostfromincreasedtaxandpricesintwo

ways. First, increasedtax/pricewould lead toarelativelygreaterproportionof lowSESsmokers who would quit smoking altogether, thereby reducing the likelihood of the

adversehealthoutcomeofsmoking.Second,agreaterproportionoftheresourcesthatlowSESsmokersspendontobaccowouldbefreedupforpossibleuseinproductiveandmore

beneficialpurposes.


15/21

14

Table4:PriceelasticityofdemandforcigarettesandbidisinBangladeshbysocio-economicstatus,2009

Cigarette Bidi

Low Moderate High Low Moderate HighSmokingprevalence

(%)(exclusive&dualusers)

45.6 38.1 30.1 19.1 11.9 4.1

Smokingparticipation:

Logitcoefficientonprice0.63***(0.14)

0.43**(0.16)

0.21***(0.05)

1.11***(0.40)

0.88*(0.47)

0.07(0.37)

Elasticityofsmoking

participation0.33 0.27 0.14 0.90 0.77 0.07

Numberofobservations 13,930 15,303 13,842 29,574 28,936 24,846Priceelasticityoftobacco

demandconditionalon

smokingparticipation

0.10(0.08)

0.12**(0.05)

0.10*(0.05)

0.28(0.14)

0.14**(0.309)

0.37***(0.08)

Numberofobservations 573 563 714 309 209 79Totalpriceelasticity 0.43 0.39 0.24 1.18 0.91 0.44


1. Cigarettedemandfunctionisestimatedusingtheadultmalepopulationofthemainsample.2. Bididemandfunctionisestimatedusingtheadultmaleandfemalepopulationofthemain,tribaland

floatingsample.

3. Low,mediumandhighsocioeconomicstatusareassignedbytertilesofCASHPORhousingindex.4. ***,**,and*implysignificantlydifferentfromzeroatthe1%,5%and10%levelsrespectively.5. Standarderrorsareprovidedintheparenthesesbelowtheestimates.Theseareadjustedfor

clusteredsamplingdesignofITCBangladeshsurvey.

DiscussionFromthe analysisofWave1 ofthe ITCBangladeshSurvey,weconclude that increasing

tobaccotaxescansignificantlyreduceconsumptionofcigarettesandbidisthroughreducedsmoking prevalence aswell as lower smoking intensity of continued smokers.We also

showthatraisingtaxratecanincreasetobaccotaxrevenuetoalargeextent.Inorderto

make maximum revenue gain from tobacco taxation, the government can increasesupplementarydutyoncigarettesby350% andonbidiby500% thatwouldmorethan

doublethecigarettepriceandnearlydoublebidiprice.

Doublingthepricelevelsofcigaretteandbidiatatimemayseemadrasticsteponthepart

ofthegovernment.Inabsoluteterms,itmeansthatapackofcigarettethatcosts18Takanowwillsellfor36Takaandapackofbidisthatcosts6Takanowwillcost12Takaafter

increasingSDtotherevenuemaximizinglevel.Thisincreaseincigaretteandbidipricescan

barelyoffsetthemorethantwofoldincreaseintheiraffordabilityoverthepastoneandahalfdecades.IndeedtheITCBangladeshSurveyfoundthatonly4.7%ofsmokersagreed


16/21

15

that in the last 6months there was a time when the money they spent on cigarettes

resultedinnothavingenoughmoneyforhouseholdessentialslikefood.Althougheconomicallyviable,therearelimitsontheextenttowhichtobaccopricescanbe

increased;asuddenlargeincreaseintobaccopricemaynotbepoliticallyfeasible.Becauseofthispoliticalchallenge,itisadvisabletoimplementthisincreaseintaxrateinphases,

whichwouldallowareasonabletimeperiodforconcernedconsumersandproducerstoadjusttotheoverallprice/taxincrease.Onepossiblestrategywouldbefortobaccotaxestobeindexedtoinflationandeconomicgrowthinamannerthattherealpricewouldincrease

by at least 10% every year. This would and keeps pace with growth in income and

purchasingpowerofpeople.Byimplementingthisstrategy,itwouldtakeabout10yearstoachievetherealincreaseinpriceadoublingofthecurrentrealprice.

TheimportanceofswitchingfromAdValoremtaxestoSpecifictaxes

Thebestmethodforreducingtobaccoconsumptionandprevalencethroughtaxationisforthattaxtobespecifictounitsofconsumption(e.g.,perunit(stick,pack,weightoftobacco)

ofcigarettesorbidis).Specifictaxescanbeusedtoservethedualpurposeofexpandingrevenue and discouraging consumption of the taxed commodity. Because specific taxesfocusonthequantityof theproductratherthanonthevalueofconsumption,specifictax

acts as a disincentive to consume tobacco in general across all price tiers and socioeconomicstatusofusers.Specifictaxesareuniversallyacknowledgedbyeconomiststobe

more effective than an ad valorem tax, such as VAT and SD, which is imposed as a

percentageofprice.

ThecurrenttobaccotaxesinBangladeshareadvaloremtype,whichisknowntoencouragesmokers to switch to lowerpriced products rather than toquit.Bangladesh isone ofa

handfulofcountriesintheAsianregion(includingCambodia,Thailand,andVietnam)that

levyadvaloremtaxesontobacco.Incontrast,18countriesintheregionlevyspecifictaxeson cigarettes: Australia, Brunei, Hong Kong, India, Korea, Japan, Laos,Macau, Maldives,

Mongolia,Nepal,NewZealand,PapuaNewGuinea,Philippines,Singapore,SouthKorea,Sri

Lanka,andTaiwan.Fourcountriesuseamixedsystemofspecificandadvaloremtaxes:China,Indonesia,Malaysia,andPakistan.Indonesiaintroducedspecifictaxinadditionto

existingadvaloremtaxfromJuly2007. 18ItistimeforthegovernmentofBangladeshtotailoritstaxadministrationforthetransitionfromadvaloremtospecificexcisetaxationof

tobacco.

For generating the amount of revenue from cigarette and bidi taxes in Bangladesh

equivalent to the maximum revenue calculated for increased SD, as discussed in the

sectionsabove,thespecifictaxshouldbeabout20Takaonapackof10cigarettesticksandabout6Takaonapackof25bidisticks.Toimplementthesetaxesovera10yearperiod

18FordetailsonthetobaccotaxstructureinAsianregion,seeEmilM.Sunley,TobaccoExciseTaxationinAsia:RecentTrendsandDevelopments,apaperpreparedfortheFourthMeetingoftheAsiaTaxForum,

April1820,2007,Hanoi,Vietnam.


17/21

16

wouldimplyaddingtaxof2Taka(in2009prices)perpackofcigaretteand0.60Taka(in

2009prices)perpackofbidieveryyear.

The immediate need isto forcetobaccoproductsoutofthepurchasingpowerofpeoplewithintheexistingrubricoftobaccotaxation.Thelongtermtargetmustbesetatmoving

from ad valorem to specific taxation and collapsing the differential tax rates betweencigaretteandbidi,aswellasacrosscigarettesatdifferentpricetiers,intoauniformrate.

Thispolicyprescriptionmayseemunfairtopeopleatthelowerstrataofsocioeconomic

status because the tax burden would be disproportionately larger for those who usecheapertobaccoproducts.Thusthespecifictaxisregressiveinnature.However,itcanbe

assertedthatraisingtobaccotaxesisoneofthoserarepolicymeasuresthatcanserveboththepurposesofequityandefficiency.Ourfindingssuggestthatthepoorerpeoplearemore

responsive to price increases and hence tax increases will reduce their tobacco

consumptiontoagreaterextent.Consequently,largerhealthgainswillbeexperiencedbythis segment of the population, which will contribute to a reduction in the health and

economicdisparitythatcurrentlyexistsinBangladesh.Overandabovethepublichealthgain,thelimitedresourcesofthepoorthatusedtobespentontobaccoconsumptionwillbedivertedfromaddictivetoproductiveusestothebettermentofsocioeconomicwelfare

ofthepoor.Thereisnoalternativetoincreasingtaxesontobaccoproductsinachievingthesegoals.

The rationale of increasing tobacco taxes is generally challenged by the followingrepercussionsanticipatedintheaftermath:

Lossofgovernmentrevenuefromtobaccoexcisetaxation Anunfairlyhightaxburdenonthepoor Increasedsmugglingoftobaccoproductsintothecountry Joblossesintobaccocultivationsector,tobaccoproductmanufacturingsector,and

theadvertisingsectorandmedia

Lossoftobaccoindustrysponsorshipofsportingandculturalevents

Thefirsttwoissueshavebeenaddressedinthepresentanalysis.Itshowsthatgovernment

does not need to fear the loss of revenue from tobacco taxation until the tax rate isincreasedbyasignificantamountandtobaccoconsumptionfallstoasufficientlylowlevel.

Furthermore,thisanalysisshowsthatthepooraremorepriceresponsivethantherichso

thatagivenpriceincreaseislikelytoshifttheburdenofincreasedtaxesfromthepoortotherich.Thedemandanalysispresentedherecannotaddresstheotherthreeissuesonits

own.Itneedstobesupplementedwiththeanalysisofdataonsmuggling,employmentand

the activities of tobacco industries,whichare subject to further explorationof ITC dataalongwithgovernmentandtobaccoindustrydocuments.


18/21

17

PolicyRecommendationsforTobaccoTaxationinBangladesh

BasedonaneconomicanalysisoftheITCBangladeshSurveydata,thefollowingtobacco

taxationstrategyisrecommended:

1. Inordertomaximizerevenuegainfromtobaccotaxation,thegovernmentshouldincreasesupplementarydutyoncigarettesby350%andonbidiby500%inthelongrun.Theseincreaseswouldmorethandoublethecigarettepriceandnearlydouble

thebidiprice.Whilethisincreaseseemsdrastic,itwouldbarelyoffsetthemorethantwofold increase in the affordability of cigarettes and bidis over the past two

decades.

2. Itisadvisabletoimplementthisincreaseintaxrateinphasestoallowreasonabletimeforconcernedconsumersandproducerstoadjustforthepriceincrease.For

example,tobaccotaxshouldbeindexedtotheinflationandeconomicgrowthsothatthe real price increases by at least 10% everyyear above the inflation rate and

keeps pace with growth in income and purchasing power of people. This

incrementalstrategywouldtakeabout10yearsforthesalespricetodoubleabovethebenchmarklevel.

3. ItistimeforthegovernmentofBangladeshtotailoritstaxadministrationforthetransition fromad valoremtospecificexcise taxationoftobacco.The specifictax

shouldbeabout20Takaonapackof10cigarettesticksandabout6Takaonapack

of 25 bidi sticks. To implement these taxes over a 10year period would implyaddingtaxof2Taka(in2009prices)perpackofcigaretteand0.60Taka(in2009

prices)perpackofbidiseveryyear.

The longterm target must be set at moving from ad valorem to specific taxation and

collapsingthedifferentialtaxratesbetweencigarettesandbidis,aswellasacrossbrandsof

cigarettesatdifferentpricetiers,intoauniformrate.

Conclusion

Opponentsoftobaccotaxincreasesarguethatsuchincreaseswillresultinasignificantlossof government revenue and impose anunfairly high tax burden on the poor.Economic

analysesoftheITCBangladeshSurveyWave1resultsshowthatthegovernmentdoesnot

needtofearthelossofrevenueuntilthetaxrateisincreasedbyasignificantamountandtobacco consumption falls to a sufficiently low level. The ITC Bangladesh findings are

consistentwithauniversalfindingineconomicstudiesoftobaccouse:thatthepoorare

morepriceresponsivethantherich.Ataxincreaseismorelikelytoreducetheirtobaccoconsumption, resulting in larger health gains and therefore lowerhealth and economic

disparityinsociety.Inthismanner,thelimitedresourcesofthepoorwillbedivertedfromaddictivetoproductiveuses,improvingthesocioeconomicwelfareofthepoor.

In other words: significant increases in tobaccospecific taxes will both increasegovernment revenue and reduce the consumption of tobacco aswell as the number of

peoplewhousetobacco.


19/21

18

APPENDIX

EconometricModelofPriceElasticityofDemandforCigaretteandBidi

In the first stage we calculate the elasticity of tobacco use conditional on smoking from the

following equation estimated by Ordinary Least Squares (OLS) method using wave -1 surveydata:

C = a0 + a1 P + a2 I.Edu + a3 Married + a4 I.Village/Ward + a4 Addict + e2

where a1 = Conditional price elasticity for smokers.C = Log of the number of whole cigarettes smoked daily

P = Log of reported price of tobacco products per stickEdu = An index of education

Married = Dummy variable for marital status = 1 if married and 0 otherwiseVillage/Ward = An index of area of residence (village/ward)

Addict = Addiction variable corresponding to the number of years of smoking.

Next we estimate the following equation for smoking participation (prevalence), where the

probability of smoking is a function of price P. We calculate these probabilities using the logitmodel:

P(Smoking) = [1+ebo+b1P]

1

The elasticity of participation is then calculated as follows:b = b1(1 Proportion of smokers) = price elasticity of smoking participation.

We estimate this equation using the census data. Since the census data contains no price

information we first estimate the following price equation for both smokers and nonsmokers

from wave-1 survey data:

P = c0 + c1Age + c2I.Upazila + c3Urban + c5I.SES + e1

where Age = Age of the respondent

Upazila = An index of Area (upazila)Urban = Dummy variable for place of residence = 1 if urban and 0 if rural

SES = Housing index for socio-economic status

The non-smokers did not report any price for tobacco products and it is assumed that they facethe same product prices as smokers who are otherwise similar with respect to observable

characteristics such as, age, socio-economic status and area of residence. We estimate thecoefficients for the price equation from the survey data and then use the estimated coefficients

into the census data to calculate predicted price P for both smokers and nonsmokers. We use

this price P to estimate the elasticity of participation. The total price elasticity is calculated as:

= b + a1


20/21

19

We then repeat the procedure for each of the three levels of socio-economic statuslow,medium, and highto see how the responsiveness to a price change varies across different

income groups.

One criticism that is often raised in the context of the estimation of demand equation is that price

is an endogenous variable, and hence the coefficients estimated in the consumption equation arebiased. We conducted a Hausman test in order to detect whether the reported price deviated fromwhat would be expected if it was exogenous. The Hausman test compares the OLS estimate of

the demand equation under the assumption that price is exogenous and the two-stage least square(2SLS) estimate of the demand equation with price predicted from instrumental variable

estimation. The Hausman test failed to demonstrate that the price variable was not exogenous inall the equations.


21/21

Australia

Baladesh

Bhuta

Brazil

Caada

Chia (Mailad)

Frace

germay

Idia

Irelad

Malaysia

Mauritius

Mexico

netherlads

new Zealad

South korea

Thailad

Uited kidom

Uruuay

Uited States of America

20 countries 50% of the worlds population60% of the worlds smokers 70% of the worlds tobacco users

The International Tobacco Control Policy Evaluation Project

The ITC ProjectEvaluating the Impact of FCTC Policies in...

Geoffrey T. Fong, Ph.D.

Department of Psychology

University of Waterloo

200 University Avenue West

Waterloo, Ontario N2L 3G1 Canada

Email: [email protected]

Tel: +1 519-888-4567 ext. 33597

www.itcproject.org

BD TaxReport Final Regular Format

Documents

Transcript of BD TaxReport Final Regular Format