JingMing Li

【Abstract】In this research paper two empirical methodologies are used for studying the relation between cigarette price and cigarette consumption in America with available statistical information. The purpose of the paper is to investigate whether the price of cigarette is a powerful method for cutting cigarette consumption. The statistical information used in the paper is collected from 48 U.S. states over the period from 1985 to 1995 for examining the effect of cigarette price and others independent variables on cigarette consumption. Ordinary least squares （OLS） regression model and Least square dummy variable model are used to determine effect of cigarette price. Furthermore， other factors such as GDP per capita， population and Consumer price index （CPI）， have been added into the model to attest to their potential nexuses with cigarette consumption. The result of the report shows that any increase in the price of cigarettes will decrease personal consumption of cigarettes. Higher prices increase costs to consumers and discourage cigarette consumption. The percentage decline in consumption caused by a percentage increase in price is measured by price elasticity of demand. Based on the analysis， it could be safely concluded that increasing price is still an effective instrument for cutting down the cigarette consumption.

【Key words】price elasticity； cigarette consumption； tobacco taxes

【摘要】兩種經(jīng)驗方法在這篇研究論文中使用，為了調(diào)查在美國香煙價格跟香煙需求的關(guān)系通過可以找到的數(shù)據(jù)信息。這篇論文的目的為了調(diào)查香煙的價格是否是一個強有力的方式去減少香煙的需求。論文中的的數(shù)據(jù)收集來源于美國的48個州從1985年到1995年，目的是檢測香煙價格跟其他獨立的變量對香煙需求的作用。最小二乘回歸模型跟虛擬變量的最小二乘法模型已經(jīng)使用去決定香煙價格的作用。此外，其他因素像人均GDP，人口，CPI也使用在模型中去證實潛在的關(guān)系對于香煙需求。報告結(jié)果顯示了任何方式的香煙價格上升將會導致個人香煙需求的下降。香煙需求的百分比下降取決于香煙價格的百分比上升，這個現(xiàn)象可以通過需求的價格彈性去估量。基于報告的分析可以放心的作出結(jié)論，香煙價格上升仍然是一種有效的工具去減少香煙的需求。

【關(guān)鍵詞】價格彈性 香煙需求 煙草稅

Introduction

Cigarette use and consumption have been identified as one of the overarching causes of preventable morbidity and dying prematurely in America （American Lung Association， 2012）. This paper portrays the landscape of cigarette consumption between 1985 to 1995 by studying in detail the relation between cigarette prices and cigarette consumption with the available statistical information （see figure 1 & 2）. In the past decades， authors and public decision-makers have been increasingly aware of the pluralistic determinants of cigarette consumption in the United States.

In the United States， however， Sheu， Hu， Keeler， Ong and Sung （2004） investigate the effect of a major cigarette price change in California， brought about by Proposition 10 and the Tobacco Settlement on consumption， using a zero-inflated negative binomial （ZINB） regression model. However， their statistical analysis shows that this change in price does not cut down cigarette consumption. They make an investigation about the smokers. The results show that some other causalities and policies should be factored into effective anti-smoking strategies.

Contrary to the empirical findings above， the effect of increasing cigarette prices brought about by the United States Tobacco Settlement in 1998 is more progressive. Keeler， Hu， Ong and Sung （2004） estimate the demand function of tobacco before and after the Settlement and identify an 8.3% decrease in cigarette consumption over the period 1990-2000. Nevertheless， due to the widespread tobacco advertisement on media after the Tobacco Settlement， cigarette consumption increases by 2.7% to 4.7%， and the cigarette price increases by about 33% to 57%.

Furthermore， a couple of other control variables， such as GDP per capita， population and Consumer price index （CPI）， have been added into the model to attest to their potential nexuses with cigarette consumption. There is no doubt that GDP per capita is highly correlated to individual income， so that they are likely to change in the same direction. The relationship between population and cigarette consumption is unclear. For one thing， people might feel lonely and depressed in sparsely populated communities where cigarette products might be a desired consolation. For another， due to the Anti-Smoking media campaign and laws restricting smoking in public areas， people in densely populated communities are less disposed to smoke， at least in public. Finally， there is no evidence implicates that CPI directly affects the cigarette consumption.

Data

Accordingly， this paper adopts the cigarette consumption panel data set from 48 states over the period from 1985 to 1995 to empirically examine the effects of price， income， advertising and a couple of other variables on cigarette consumption. This cigarette consumption panel data set is made up by annual data from the 48 continental U.S. states. The quantity of cigarette consumption is identified by annual per capita cigarette sales in packs per fiscal year， arising from state tax collection data. The price refers to the retail cigarette price per pack on average during the fiscal year， including taxes. Income is marked by per capita income. The general sales tax is the average tax due to the state sales tax applied to all consumption goods， while the cigarette-specific tax is the one merely applied to tobacco.

Methods

The empirical methodologies used in this paper are ordinary least squares （OLS） regression model and Least square dummy variable model （LSDV）. The ordinary least squares （OLS） regression model is used to determine the effect of average cigarette price. However， other factors such as personal financial status， economics condition and unobserved variable， which can influence the cigarette consumption variously over time.

lnpackpc=β0 + β1*lnavgprs +a + u eq（1）

Number of packs per capita is represented by lnpackpc， β1 is the coefficient on the average price of cigarette variable， a is the additional control variables and u is the error term.

The least square dummy variable model （LSDV） provides a good way to understand fixed effects. The effect of cigarette price （lnavgprs） is mediated by the difference across states. By adding the dummy for each state it can estimate the pure effect of cigarette consumption （lnavgprs）. A fixed effect regression is suitable to eliminate any observable and/or unobservable differences between states.

yit = βi*Priceit + αi + uit eq（2）

where i represents each state and t denotes time （from 1985 to 1995）. Number of packs per capita is represented by yit， βi is the coefficient on the average price of cigarette variable， αi is the intercept term for each state and uit is the error term. A fixed effect regression， used on panel data， is ideal to control the variation between states.

In this paper， the instrumental variable used on average cigarette prices is the average state， federal， and average local excise taxes and the average excise taxes for each fiscal year， including tobacco taxes.Overall， a total four models will be estimated in this paper：

Model 1： average cigarette price on number of packs per capita.

Model 2： average cigarette price， personal income and GDP per capita on number of packs per capita.

Model 3： average cigarette price， personal income， GDP per capita， population and CPI on number of packs per capita.

Model 4： average cigarette price， personal Income， GDP per capita， population， CPI and advertisement expense for cigarette on number of packs per capita.

Model 2 aims to capture income and economic factors. Additional explanatory variables that concern personal financial status and economic dynamics are used into this regression. Model 3 attempts to capture other social and economic differences. Model 4 includes all explanatory variables. The hypothesis test for this study is to test：

Whether there is a relationship between cigarette price and cigarette consumption.

In Table 1， the OLS regression with the instrument variable， coefficients for cigarette price in model 1 is strongly related to cigarette consumption， with an expected negative relationship. The coefficient of average cigarette price is significant at the 1% level. No matter if the instrumental variable is used， the coefficient is significant. Also， the instrumental variable provides a larger coefficient of average cigarette price， which implies the realistic relationship between cigarette price and consumption estimated by the instrumental variable.

In model 2 to 4 on the table， control variables of personal income， GDP per capita， Population， CPI and advertisement expense are consecutively added. As a result， the coefficient of average cigarette prices remains statistically significant and it is negatively correlated with cigarette consumption. On the contrary， most of the control variables have no statistical significance and are small in size. Although the effects of personal income and GDP per capita are large in size， the potential effect of multicollinearity between them demonstrates that GDP per capita is not that important compared to personal income and it would be better to exclude this variable from the model. Nevertheless， after adding other control variables， the income elasticity remains roughly the same with our expectation. Surprisingly， population has a significantly negative relationship with cigarette consumption. The result shows that under the condition of controlling other variables， people will smoke less in densely populated areas and a result we could not properly interpret now. Finally， CPI and advertisement expense are two variables that are not significantly related to cigarette consumption.

In contrast， Table 2 repeats the regressions from Table 1 but with state-specific fixed effects， which is documented in Table 3. Such effects allow the researcher to consider any unobservable time-invariant factors such as culture and history that might make a contribution to exceptionally high or low cigarette consumption. Results of the impact on price increased in cigarette consumption in Table 1 still hold for Table 2. Most coefficients are statistically significant at the 1% level； the multiple R-squared value suggests that variation of cigarette prices explains approximately 95.59% of the total variation of cigarette consumptions through fixed effect model that shows in table 3. Exceptions are the regression in Model 4， estimated by instrumental variable， where the coefficient does not reach statistically significance.

More interestingly， in Table 2 the coefficient on population is statistically significant in most regressions， but it is negatively correlated with cigarettes consumption. This indicates that while more populous states tend to consume more cigarettes， conversely there is a decline in the percentage of people who are the cigarette consumers. However， the result is not true for GDP per capita， CPI， or advertising expense. Additionally， the estimated income elasticity remains to be a large and statistically significant coefficient from Table 1 and instrumental variable gives a more conspicuous result. Finally， since the t-statistic for cigarette prices is larger enough through all 4 models， it demonstrates that cigarette prices have significant impacts on cigarette consumptions over times.

Discussion

Overall， the results in Table 1 to Table 3 clearly speak in defense of the argument that the U.S. price elasticity of cigarettes is significantly negative over the period from 1985 to 1995. Besides， cigarette consumption is also strongly linked to income and population. However， GDP per capita， CPI and advertising expense do not demonstrate a statistical significance with respect to cigarette consumptions.

The results of the analysis in the report are in accordance with previous studies， which show that any increase in the price of cigarettes will decrease personal consumption of cigarettes. Higher prices increase costs to consumers and discourage cigarette consumption. As the price of an item increases by a certain percentage， consumption of the item falls. The percentage decline in consumption caused by a percentage increase in price is measured by price elasticity of demand. Also， price elasticity of demand can help to show that some of the decline in cigarette consumption can be explained by higher prices.

The empirical finding also demonstrates that advertising expense in this period is not as important as I considered before. The positive relationship between cigarette advertising and cigarette consumption is unverified. This result suggests that advertising elasticity is in decline trend as the cigarette market in the United States matures （Assmus， Farley and Lehmann， 1984）. Advertising attracts repeated purchasers rather than new smokers. Moreover， the restrictions imposed on the cigarette industry about advertising also contribute to the indifference to cigarette consumption and make the customers less responsive to cigarette advertising （Andrews and Franke， 1991）. Due to the shortage of statistical information involved in the cigarette consumption panel data set， it is beyond reach to investigate the potential demographic and behavioral determinants of cigarette consumption.

Policy Implication

Take the pronounced negative price elasticity and positive income elasticity of cigarette consumption into consideration， it shows that the increase in price is still the most effective instrument for cutting down the popularity of cigarette products. On the other hand， the low-income groups in some areas are not sensitive to price changes. However， to deal with cigarette consumption， smokers consider quitting or decreasing tobacco consumption in response to rise in cigarette price. （Farrelly， Bray， Pechacek and Woollery， 2001）. Moreover， the increased price from taxation could strongly influence cigarette consumptions. According to the study conducted by Guindon， Tobin and Yach in 2002， there is ample room to increase tobacco prices through taxation. In too many countries， cigarette prices have failed to keep up with increases in the general price level of goods and services， rendering them more affordable in 2000 than they were at the beginning of the decade. As a consequence， in contrast with other ineffective cigarette price support programs， such as minimum cigarette price laws in the United States （Feighery， Ribisl， Schleicher， Zellers and Wellington， 2005）.

Limitation and Future Motivation

There are essentially three aspects of limitations over this research and future study should lay stress on them. In the first place， as explained above， this paper concludes that rise in cigarette price， especially cigarette excise tax， is an effective tool to cut down cigarette consumptions. However， some authors oppose that the increased price will unjustly hurt specific groups， notably the minorities and the poor （Farrelly， Bray， Pechacek and Woollery， 2001）. Even though this paper uses the price elasticity and income elasticity to deal with the problem， a better understanding of this issue should be deepen about how the price increase will affect the welfare of a variety of demographic groups.

Secondly， by using cigarette consumption panel data set and IV method， this paper estimates the basic elasticity for a couple of variables over the period from 1985 to 1995. Nevertheless， it fails to investigate the trends in this relationship over time， indicating that a more pragmatic policy with respect to reducing cigarette consumption is necessary for a solid foundation.

Lastly， the methods used in this paper can only offer a general trend and a rough approximation of the influence of price increase. But the issue of how a specific policy affects personal behavior patterns remains in question.

Conclusion

Based on the cigarette consumption panel data set， this paper estimates various elasticities with or without instrumental variable. The estimated results reveal that a sale of cigarettes is highly related to price change， income variation and population. The inelastic nature of demand for cigarettes indicates that only some of the decline in cigarette consumption can be explained by higher prices during the last three decades. Other variables， such as GDP per capita， CPI and advertisement expense， do not significantly have a strong effect on cigarette consumption. Based on the analysis above， considering the pronounced negative price elasticity and positive income elasticity of cigarette consumption， it could be safely concluded that increasing price is still an effective instrument for cutting down the popularity of cigarette products， especially among lower income adults， in consideration of the pronounced negative price elasticity and positive income elasticity of cigarette consumption. For future researches， socio-demographic information， including gender， income， race， age and other related message among different groups， should be included into the panel data set. Besides， it fails to investigate the trends in this relationship over time and the behavioral patterns. As a result， concerning the pronounced income elasticity and price elasticity， more effective measures should be taken to cope with the issue of cigarette consumption， especially smoking among young people.

References：

[1]American Lung Association.（2012）.Trends in tobacco use.2011.Accessed at.

[2]Guindon，G.E.，Tobin，S.，& Yach，D.（2002）.Trends and affordability of cigarette prices：ample room for tax increases and related health gains.Tobacco control，11（1），35-43.

[3]Hu，T.W.，& Mao，Z.（2002）.Effects of cigarette tax on cigarette consumption and the Chinese economy.Tobacco control，11（2），105-108.

[4]Gallus，S.，Schiaffino，A.，La Vecchia，C.，Townsend，J.，& Fernandez，E.（2006）.Price and cigarette consumption in Europe.Tobacco control，15（2），114-119.

[5]Sheu，M.L.，Hu，T.W.，Keeler，T.E.，Ong，M.，& Sung，H.Y.（2004）.The effect of a major cigarette price change on smoking behavior in California：a zero‐inflated negative binomial model.Health Economics，13（8），781-791.

[6]Zhang，P.，Husten，C.，& Giovino，G.（2000）.Effect of the tobacco price support program on cigarette consumption in the United States：an updated model.American Journal of Public Health，90（5），746.

[7]Keeler*，T.E.，Hu，T.W.，Ong，M.，& Sung，H.Y.（2004）.The US national tobacco settlement：the effects of advertising and price changes on cigarette consumption.Applied Economics，36（15），1623-1629.

[8]Townsend，J.（1996）.Price and consumption of tobacco.British Medical Bulletin，52（1），132-142.

[9]Peretti-Watel，P.，& Constance，J.（2009）.“Its all we got left”.Why poor smokers are less sensitive to cigarette price increases.International journal of environmental research and public health，6（2），608-621.

[10]Andrews，R.L.，& Franke，G.R.（1991）.The determinants of cigarette consumption：a meta-analysis.Journal of Public Policy & Marketing，81-100.

[11]Leigh，J.P.，& Schembri，M.（2004）.Instrumental variables technique：cigarette price provided better estimate of effects of smoking on SF-12.Journal of clinical epidemiology，57（3），284-293.

[12]Assmus，G.，F(xiàn)arley，J.U.，& Lehmann，D.R.（1984）.How Advertising Affects Sales：Meta-Analysis of Econometric Results.Journal of Marketing Research（JMR），21（1）.

[13]Farrelly，M.C.，Bray，J.W.，Pechacek，T.，& Woollery，T.（2001）.Response by adults to increases in cigarette prices by sociodemographic characteristics.Southern Economic Journal，156-165.

[14]Guindon，G.E.，Tobin，S.，& Yach，D.（2002）.Trends and affordability of cigarette prices：ample room for tax increases and related health gains.Tobacco control，11（1），35-43.

[15]Feighery，E.C.，Ribisl，K.M.，Schleicher，N.C.，Zellers，L.，& Wellington，N.（2005）.How do minimum cigarette price laws affect cigarette prices at the retail level？.Tobacco control，14（2），80-85.

[16]Farrelly，M.C.，Bray，J.W.，Pechacek，T.，& Woollery，T.（2001）.Response by adults to increases in cigarette prices by sociodemographic characteristics.Southern Economic Journal，156-165.

[17]Hu，T.W.，Sung，H.Y.，& Keeler，T.E.（1995）.Reducing cigarette consumption in California：tobacco taxes vs an anti-smoking media campaign.American journal of public health，85（9），1218-1222.