Disclaimer: This assignment was written by a student and is not an example of our proffesional work. You can view professional samples here.

Any opinions, findings, conclusions, or recommendations expressed in this material are those of the authors and do not reflect the views of BusinessTeacher.org. Information in this assignment should not be used to form the basis for any kind of financial or investment advice, as the content may contain inaccuracies or be out of date..

How Much Do Price Changes in Cigarettes Influence Demand?

4249 words (17 pages) Business Assignment

17th Nov 2020 Business Assignment Reference this

Tags:

Abstract

This paper analyses the relationship between price elasticity of demand for cigarettes whilst controlling for other relevant variables. An Ordinary Least Squares regression model was the estimator used, making use of cross-sectional microdata. The data was also tested for heterscedasticity, multicollinearity and omitted variable bias in order to improve the models validity. Supported by relavent literature, it was found that cigarette demand can be influenced by educational attainment and an individuals age, the relationship between age and smoking prevalence is however concave.  Coefficients representing the P.E.D and Y.E.D are in line with other relavent literature, however it was difficult to obtain high enough significance levels.

Introduction

The consumption of cigarettes is one of the main causes of preventable death throughout the world, with various estimates showing that smoking leads to the premature death of approximately 8-million people per year. [1] Further to this, cigarette smoking costs the world economy approximately $1.2 trillion[2] per year, equating to 1.8% of the worlds GDP. The purpose of this paper is to add to the existing literature in assessing the ‘Price Elasticity of Demand’.

There are a number of costs to cigarette smoking in the form of personal costs but also through subsequent negative externalities produced, which is why increasing smoking cessation remains one of the main targets of Governing bodies. This can be seen through Ekpu & Brown (2015) who find a major cost to the third part of smoking comes through the decrease in productivity, within the USA, they estimate a total loss of $151 billion as a consequence of smoking. Further to this, it has been estimated that passive smoking costs the US economy $1.9 billion (Yao et al 2018).

This paper aims to provide insights into the effect that price changes have on decreasing the demand for cigarettes, whilst expanding this analysis to include other relevant variables that can influence demand such as age, education level and an individuals income.

Literature Review

There are numerous studies providing insights into the P.E.D for cigarettes. As a result of this, there are differing viewpoints on how the consumption of cigarettes is directly affected by price changes. In spite of this, a common theme seems to emerge by which an increase in the price of cigarettes will lead to a fall in demand, often importantly showing how the demand can change from being price elastic to price inelastic.

Much of the existing literature makes use of cross-sectional survey data, including a range of control variables. Work produced by Tauras (2005) makes use of cross-sectional data from surveys covering the civilian non-institutionalised populations of the USA from 1993/1994 and finds that the P.E.D for cigarettes varies from -0.541 and -1.092 moving from price inelastic to price elastic.

More recent work by Yeh et al (2017) looks into the effects of a rise in cigarette price on consumption throughout 28 European Countries. Making use of panel data from 2005 to 2014, collected by Euromonitor International, the World Bank and the World Health Organisation. Their regression model includes cigarettes prices, cigarette price in Eastern European Countries, GNI and the rural population levels. They yield similar results of those mentioned above with the P.E.D estimates ranging from -0.503 to -1.227, again moving from inelastic to elastic demand.

A report produced in the Republic of Ireland by Kennedy et al (2015) provides estimations that the P.E.D within Ireland ranges from -1.6 to -2.0, with an average of -1.8. This indicates that if there was an imposed price increase of 10%, there would be a fall in demand by 18%. The above highlights how research can provide a wide range of P.E.D estimates.

This paper will be another intervention into the continuing debate surrounding the ‘Price Elasticity of Demand’ for cigarettes. Similar to previous literature making use of Cross-sectional data.

Theoretical Analysis

In order to gain an insight into the P.E.D for cigarettes, I will make use of a mainly log-log model, with the dependent variable being cigs and the independent variable’s cigpric and income, this will allow for me to assess the impact of income on cigarette demand. The remaining variables will be in linear form, educ and age, and the final variable being age-squared to gain an idea if there is a concave or convex relationship.

Data Analysis

This paper makes use of data (Wooldridge 2000) collected from the ‘Child Health Supplements’ to the 1988 National Health Interview Survey, it features 807 observations throughout different states within the USA. The data presents itself in cross-sectional microdata form.

The dependent variable in our study ‘cigs’ reports the average number of cigarettes smoked per day by each individual, for the purpose of this research, I have transposed this variable into logarithmic form in order to draw conclusions relating to the price elasticity of demand.

As mentioned above, the estimations for the P.E.D vary significantly within the existing literature, to draw our own conclusions I have introduced a ‘price’ variable, this measures the price per packet of cigarette throughout the United States. This is measured by ‘cents per pack’, again this has been transposed into a logarithmic form which will allow us to draw conclusions relating to the percentage change in price and how this impacts the demand for cigarettes in percentage terms.

In order to bring analysis further than a simple price and demand relationship, income has been added as its often considered an important factor. As a result, I have included income which reflects the average yearly income of respondents. This has been transposed into a logarithmic form which will provide an insight into the income elasticity of demand. Existing literature also varies on results, work produced by Sackrin (1957) making use of cross-sectional data finds that Y.E.D stands at 0.12, this would imply that a 10% increase in income, would lead to a 1.2% increase in the demand for cigarettes.

Additionally, educational attainment is often considered an important variable when assessing the demand for cigarettes. Thus an education variable has been included which gives an insight into the surveyed individuals years of schooling –  A possible disadvantage is the data does not state the level of education that was attained, just the number of years spent in school. Nonetheless, existing work (Silles 2015) highlights that those with a higher level of education face a lower probability of smoking, an increased probability of stopping smoking through an increase in the probability of quitting. It was also important to include a variable to control for the impact of age on smoking habits, further an age-squared variable was added which can allow us to determine if there is a concave or convex relationship between age and smoking.

Table 1: Descriptive Statistics

Variable

Mean

Std. Dev.

Min.

Max.

cigs

8.686493

13.72152

0

80

cigpric

60.30041

4.738469

44.004

70.129

income

19304.83

9142.958

500

30000

educ

12.47088

3.057161

6

18

age

41.23792

17.02729

17

88

agesq

1990.135

1577.166

289

7744

log_cigs

-3.143642

4.794245

-6.907755

4.382039

log_cigpric

4.096048

0.082918

3.784303

4.25035

log_income

9.687316

0.7126956

6.21461

10.30895

n=807

       

Table 1 represents the descriptive statistics for each variable within the study, with 807 observations. It can be seen that on average, individuals smoke 9 cigarettes per day (rounded up), the mi5nimum is zero including those who do not smoke, with a maximum of 80. The age variable provides some intuition of the composition of the survey, with the average respondent being 41, youngest and oldest are 17 and 80 respectively, providing a wide range of observations.   

The income variable shows the average yearly earnings of respondents, with an average of $19,304 per year, the minimum and maximum being $500 and $30,000 respectively showing a significant difference between the lowest and highest earners. Educational controls are measured through the years of schooling, the average observation has 12 years of schooling with a minimum and maximum of 6 and 18 years respectively. Our dependent variable ‘cigpric’ measures the cigarette price, with an average of $0.60 per packet, the minimum and maximum vary quite a degree from $0.44 to $0.70.

There is a potential issue with endogeneity relating to the price variable within our study, the failure to account for this may lead to bias of the price elasticity estimates. A potential way to get around this would be through an ‘instrumental variable’ that is correlated with price but does not have a direct influence on demand, for example, a common instrumental variable would be an excise tax on cigarettes. This would allow us to provide more accurate estimates which could allow us to provide more insight into government policy, especially relating to taxation policy which depends critically on the price elasticity of demand. The ‘instrumental variable’ must be correlated with the endogenous variable, as it is used to pick up the part of the variable which is uncorrelated.

Empirical Analysis

Before beginning the econometric analysis, a test for heteroscedasticity was completed which produced a p-value of 0.00, which, therefore, lead to the rejection of the null of constant variance. With this in mind, I have controlled for this in my regression using robust standard errors. It was also relevant to test for multicollinearity, which produced satisfactory results in which I concluded no further adjustments were necessary.

In each of my four regressions, the dependent variable is the log of cigarettes denoted ‘log_cigs’. In the first instance, I compute a simple regression which highlights the relationship between both the price and demand for cigarettes, the coefficient reads -1.48, being classified as elastic this would indicate that an increase in the price of cigarettes by 10% would lead to a fall in the demand for cigarettes by 14.8%. This coefficient complies with the work of Nguyen (2012) who finds within their model that, the long run P.E.D stands at -1.48 within the UK. Our result is however statistically insignificant at all levels, this could be due to the lack of variation amongst the data.

The second variation within the model controls for the income levels of those surveyed. This variable is included as it allows us to assess the changes in the demand in cigarettes as a result of changes in an individual’s income. The corresponding coefficient for income reads at 0.11, implying that in an increase in an individual’s income will lead to an increase in the overall demand for cigarettes, however, the coefficient implies that a change in cigarette demand will not be very responsive given a change in the level of income. The coefficient at 0.11, would imply that an increase in an individual’s income by 10% would lead to an increase in the demand for cigarettes by 1.1% - our result however is statistically insignificant, which again could be due to the lack of variation within our data. Although our output is statistically insignificant, the corresponding coefficient is in line with the work of Sackrin (1957) who utilizes cross-sectional data and finds the income elasticity of demand to be 0.12.

The third variation within our model controls for the age and agesq of our respondents. The inclusion of this variable allows us to assess changes in the demand for cigarettes given someone’s age. The introduction of the agesq variable makes it possible to assess the impact of an age change as an observation gets older. The coefficient from our regression model shows 0.184, this would imply that for every unit increase in age, that the demand for cigarettes increases by 18.4% statistically significant, further the agesq coefficient shows -0.002, this would imply that when age increase by a year, the slope of the curve decreases by 2% which would imply a concave relationship, this result is also statistically significant. Work produced by Jordan et al (2017) finds that older smokers are less likely to attempt to quit smoking – further that there are potential opportunities missed in facilitating smoking cessation among older smokers.

The fourth and final specification within our model introduces a human capital aspect through an education metric, allowing us to see how an additional increase in a year of schooling will impact the demand for cigarettes. The coefficient reads at -0.281, this would imply that for every additional year of education an individual completes, there is a fall in the demand for cigarettes by 28% statistically significant at all levels, and this is line with the work of Silles (2015) who concludes that a higher level of educational attainment implies individuals are less likely to smoke. Additionally, we can see a diminishing impact of the price of cigarettes on the quantity demanded, as the price elasticity of demand falls from -1.489 to -1.286, this would imply that now an increase in the price of cigarettes by 10% would only lead to a fall in demand of 12.86% still however statistically insignificant at all levels.

Table 2: Regression Results

 

(1)

(2)

(3)

(4)

VARIABLES

log_cigs

log_cigs

log_cigs

log_cigs

         

log_cigpric

-1.489

-1.568

-1.428

-1.286

 

(2.085)

(2.091)

(2.069)

(2.048)

log_income

 

0.120

-0.208

0.105

   

(0.229)

(0.243)

(0.251)

age

   

0.185***

0.214***

     

(0.053)

(0.053)

agesq

   

-0.002***

-0.003***

     

(0.001)

(0.001)

educ

     

-0.281***

       

(0.055)

Constant

2.954

2.118

1.722

1.217

 

(8.544)

(8.686)

(8.584)

(8.500)

         

Observations

807

807

807

807

R-squared

0.001

0.001

0.027

0.054

Robust standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

The R-squared value has increased with the additional control variables in our model, from 0.001 to 0.054, showing some form of improvement, however, the value is far from desired at 1. Due to this, it was relevant to test for omitted variable bias. The result from this gave a p-value of 0.045, allowing a rejection of the null hypothesis at both the 5% and 10% significance level that there were no omitted variables.

Conclusion

This paper supports the claim that those who are more educated within society are less likely to smoke, perhaps governing bodies could find ways in which to reduce the prevalence of smoking through campaigns that target those individuals with a lower level of educational attainment. Work by Zhuang (2015) states that smokers with a lower level of education have consistently lagged behind higher-educated counterparts in quitting smoking, and thus smoking control groups should focus more on the lower educational groups. Similarly, more should be done to take advantage of smoking prevalence within older groups in society. Our income and price variable did not produce the required significance levels to draw any credible assumptions. 

Within this study, the data is in cross-sectional form. It is plausible that panel data would be better suited to research the relationship between price and the demand for cigarettes as it allows us to follow individual observations through a time-variant position. The use of panel data would allow us to use an IV-regression model with the scope of making more accurate assumptions.

Bibliography

  • Ekpu, V. and Brown, A. (2015) 'The Economic Impact of Smoking and of Reducing Smoking Prevalence: Review of Evidence', Tobacco Use Insights, 8, pp. 1-35.
  • Jeffrey M. Wooldridge, 2000. "Smoke," Instructional Stata datasets for econometrics smoke, Boston College Department of Economics.
  • Jordan, H., Hidajat, M., Payne, N., Adams, J., White, M. and Ben-Shlomo, Y. (2017) 'What are older smokers’ attitudes to quitting and how are they managed in primary care? An analysis of the cross-sectional English Smoking Toolkit Study', BMJ Open, pp. 1-9.
  • Kennedy, S., Pigott, V. and Walsh, K. (2015) Economics of Tobacco: An analysis of cigarette demand in Ireland, Ireland: Statistics and Economic Research Branch of the Office of the Revenue Commissioners.
  • Nguyen, L., Rosenqvist, G. and Pekurinen, M. (2012) Demand for Tobacco in Europe An Econometric Analysis of 11 Countries for the PPACTE Project, Finland: National Institute for Health and Welfare.
  • Sackrin, S. (1957) 'Income elasticity of demand for cigarettes: A cross-section analysis', Agricultural Economics Research, 9(1), pp. 7.
  • Silles, M. (2015) 'The causal effect of schooling on smoking behaviour', Economics Of Education Review, 48, pp. 102-116.
  • Tauras, J. (2005). An Empirical Analysis of Adult Cigarette Demand. Eastern Economic Journal, 31(3), 361-375.
  • Yao, T., Sung, H., Wang, Y., Lightwood, J. and Max, W. (2018) 'Healthcare costs attributable to secondhand smoke exposure at home for U.S. adults', Preventative Medicine, 108, pp. 41-46.
  • Yeh, C. (2017), Schafferer, C., Lee, J. et al. The effects of a rise in cigarette price on cigarette consumption, tobacco taxation revenues, and of smoking-related deaths in 28 EU countries- applying threshold regression modelling. BMC Public Health 17, 676
  • Zhuang, Y., Gamst, A., Cummins, S., Wolfson, T. and Zhu, S. (2015) 'Comparison of Smoking Cessation Between Education Groups: Findings From 2 US National Surveys Over 2 Decades', Am J Public Health, 105(2), pp. 373-379.

Data Appendix

Omitted Variable bias

Abbreviations Used

Price Elasticity of Demand – P.E.D

Income Elasticity of Demand – Y.E.D

Do File

Installing Data Set

SSC Instal Bcuse

Bcuse Smoke

 

Summarising Data

Summarize

Histogram age, normal

Histogram cigpric, normal

Histogram cigs, normal

Histogram cigs, normal

Histogram income, normal

Histogram educ, normal

 

Generating Variables

Gen log_cigs = log(0.001+cigs)

Gen log_income = log(0.001+income)

Gen log_cigpric = log(0.001+cigpric)

 

Dropping un-used variables

Drop restaurn

Drop lcigpric

Drop lincome

Drop white

 

Creating descriptive statistics table

Sum cigs cigpric income educ age agesq log_cigs log_cigpric log_income

 

Completing Regressions

Reg log_cigs log_cigpric

Reg log_cigs log_cigpric log_income

Reg log_cigs log_cigpric log_income age agesq

Reg log_cigs log_cigpric log_income age agesq educ

 

Test for Heteroscedasticity

Estat hettest

Test for multicollinearity

Estat vif

Installing Outreg

Ssc instal outreg2

 

Regressions controlling for heterscedasticity and outreg

Reg log_cigs log_cigpric, robust

Outreg2 using regression_smoke, replace word dec (2)

Reg log_cigs log_cigpric log_income, robust

Outreg2 using regression_smoke, append word dec (2)

Reg log_cigs log_cigpric log_income age agesq, robust

Outreg2 using regression_smoke, append word dec (2)

Reg log_cigs log_cigpric log_income age agesq educ, robust

Outreg2 using regression_smoke, append word dec (2)

Testing for omitted variable bias

Estat ovtest


[1] According to the World Health Organisation.

[2] World Bank Statistics

Cite This Work

To export a reference to this article please select a referencing stye below:

Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.

Related Services

View all

DMCA / Removal Request

If you are the original writer of this assignment and no longer wish to have your work published on the UKDiss.com website then please: