Article Text

Download PDFPDF

Economics of tobacco control in Pakistan: estimating elasticities of cigarette demand
  1. Nasir Mushtaq1,
  2. Saghir Mushtaq2,
  3. Laura A Beebe1
  1. 1Department of Biostatistics and Epidemiology, University of Oklahoma Health Sciences Center, Oklahoma City, Oklahoma, USA
  2. 2Pakistan Institute of Development Economics (PIDE), Islamabad, Pakistan
  1. Correspondence to Nasir Mushtaq, Department of Biostatistics and Epidemiology, University of Oklahoma Health Sciences Center, 801 NE 13th Street, CHB-309, Oklahoma City, Oklahoma 73104, USA; nasir-mushtaq{at}


Background Despite ongoing global efforts for tobacco control, low-income countries with struggling economies have challenges to effectively implement tobacco policies and programs. Due to the complexity of the tobacco control issue and lack of comprehensive policies, tobacco use is increasing in Pakistan.

Objective The aim of this study was to assess the effect of taxes on tobacco demand in Pakistan.

Methods Various surveillance indicators of tobacco use were assessed from 2001 to 2009. Price elasticities of cigarette demand in Pakistan were investigated.

Results During 2003–2009, annual per capita cigarette consumption increased by 30%. Analysis of economic data indicated that a 10% increase in cigarette prices would lead to 4.8% decrease in cigarette consumption while controlling for per capita income in the short term. The long-term price elasticities of cigarette demand were estimated at −1.17. The estimations provided support for myopic addiction model for cigarette consumption in Pakistan.

Conclusions Increasing tobacco taxes would have a significant impact on tobacco consumption in Pakistan. Cigarette consumption could decrease by 11.7% in the long term if there was a 10% increase in its price. The results of this study should benefit policymakers as it provides information on the characteristics of the cigarette consumption and cigarette demand function that may help in planning tobacco control strategies in low-income and middle-income countries.

  • Economics

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Tobacco consumption is the leading cause of preventable death and disability worldwide. Currently 5.4 million deaths are annually attributed to tobacco use worldwide, with 50% of them occurring in Asia. If current smoking patterns persist, the number of tobacco-related deaths will rise to 8.3 million by 2030; 80% of these will be in Asia, surpassing the combined number of deaths from AIDS, tuberculosis, automobile accidents, maternal mortality, homicide and suicide.1 As multinational cigarette companies target the huge and still underdeveloped market of low-income countries, tobacco use among the low-income and middle-income countries has become 80% of the global consumption.2 While in high-income countries smoking prevalence is decreasing, an upward trend is projected in low-income and middle-income countries. Pakistan is a developing country that produces tobacco and has well established cigarette manufacturers. According to 2007 estimates, Pakistan with 25 million smokers (36% of the adult population) and annual tobacco consumption of more than 90 000 tons was ranked the 16th largest consumer of tobacco.2 3

Low-income countries with struggling economies, such as Pakistan, hesitate to take steps that might affect one or more of the significant revenue streams of their economy. Over the years the tobacco industry has created a formidable barrier for the policymakers of these countries to overcome by highlighting contributions of the tobacco economy in generating revenues through agriculture, exports, manufacturing, taxes and creating more jobs. Pakistan has a domestically driven agroindustrial economy, which is currently in a state of a balance of payments crisis and threatened by inflation. Unlike other low-income agrarian countries such as Zimbabwe and Malawi, Pakistan is closer to the countries with balanced tobacco economies, consuming about the same as they produce.4 Tobacco is one of the six leading contributors to the economy, providing 1.16 billion Pakistani Rupees (PKR) through taxes during current fiscal year.5 Cigarettes are subject to two different taxes; sales tax and excise tax. Sales tax is charged as 13.8% of the retail price, whereas for excise tax, domestic cigarettes are categorised into three tiers on the basis of the retail price. Tier I, the lowest-priced cigarettes (maximum price PKR 14.86/20 cigarettes) have specific excise (PKR 6.34). Cigarettes in tier II are the mid-price range (PKR 14.86–32/20 cigarettes) and are subject to specific excise (PKR 6.34) and incremental excise (69% per over PKR 14.86). Excise for the higher-priced cigarettes (more than PKR 32/20 cigarettes), tier III, is 63% of the retail price.6

An increase in prices lowers the demand of consumer goods; the same is true for tobacco consumption. In developed countries an increase in cigarette taxes has consistently led to significant reduction in cigarette smoking. For high-income countries, estimated price elasticity of cigarette demand ranged from −0.25 to −0.50, whereas for low-income and middle-income countries it ranged from −0.5 to −1.0.7 8 There is a lack of any studies estimating price elasticity for Pakistan. Such estimates would provide vital information for the creation of public health and economic policies, and are needed in order to formulate a tax structure to control the tobacco epidemic in the country. Evidence on tobacco consumption behaviour is also an important factor when interpreting the results of price elasticity. To date, no studies have been conducted to estimate the dynamic specification of cigarette demand with complete information on price elasticities, including addictive behaviours.

The aim of this study is to assess the impact of taxes on cigarette consumption in Pakistan. We conducted an empirical analysis of cigarette demand in Pakistan to investigate price elasticity. The long-term and short-term elasticities are estimated and addiction behaviours of smokers are assessed.


Data related to tobacco consumption and economic indicators were obtained from various departments of the government of Pakistan and international institutions. These included the Ministry of Finance, statistics division, Ministry of Commerce, Pakistan Tobacco Board, International Monetary Fund (IMF), United Nations Commodity Trade Statistics Database (UN Comtrade), Food and Agriculture Organization (FAO) and the US Department of Food and Agriculture.

Tobacco consumption

Since direct data about annual cigarette consumption is not available, cigarette consumption was calculated from production data and foreign trade data using the following equation: cigarette consumption=no. of cigarettes produced−no. of cigarettes exported+no. of cigarettes imported.

Per capita cigarette consumption was defined as number of cigarettes consumed per adult (15 years or older) per year. Annual cigarette production was obtained from Ministry of Finance and from US Department of Agriculture. Tobacco export data was obtained from Pakistan Tobacco Board, UN Comtrade; however, import data was taken from UN Comtrade.9 For certain years cigarette imports were reported in units of weight. In order to estimate number of cigarettes imported, the previously specified conversion of 1 cigarette=1 g was used.10 Cigarette import data for 2009 was not available; therefore, 2008 import data was used for estimating cigarette consumption.


Price data used for estimation was the average price of a pack containing 10 cigarettes for each year. These prices were for the first tier brands, which account for about 85% of the market. The price was adjusted for inflation. Price data was obtained from Ministry of Finance and PKR was used as currency benchmark.5


Per capita income was calculated by dividing gross domestic product (GDP) by the adult population (15 years or older). Income was also adjusted for inflation by using the Consumer Price Index (CPI) as a deflator. GDP and CPI data were obtained from the Pakistan Ministry of Finance and IMF.5

Statistical analysis

Cigarette demand was analysed using annual data from 1981 to 2009 to estimate price and income. Based on the economic theory of demand, quantity of cigarettes consumed is a function of real price and per capita income. In order to accommodate time trends, the following model was estimated:lnQt=β0+β1lnPt+β2lnYt+β3trend+et

Where Qt is number of cigarettes consumed per adult in time (year) t, Pt is average price of cigarette during year t, Yt is per capita income in year t, trend is the time trend variable and e is the error term.

The autoregressive distributive lag (ARDL) cointegration technique was used to generate short-term and long-term elasticities. The following form of ARDL specification for the cigarette demand function was used to analyse the long-term relationship and to estimate the long-term demand elasticities:Δlog(ct)=β0+i=1pβ1iΔlog(cti)+i=0pβ2iΔlog(pti)+i=0pβ3iΔlog(iti)+β4log(ct1)+β5log(pit1)+β6log(it1)+ut

Before estimating these elasticities, the unit root test was used to check the stationarity of the data, and the augmented Dickey–Fuller (ADF) test was used to check the order of integration. For estimating the long-term parameters of the model, the order of the ARDL (Q,P,Y) model was selected on the basis of Schwartz–Bayesian criterion (SBC) and Akaike information criteria (AIC). The short-term dynamics of the model were checked to verify convergence towards steady/stable relationship. Therefore short-term elasticities were also obtained. Cumulative sum control chart (CUSUM) and CUSUMQ tests were used to check the stability of the estimated model.

The statistical software packages SAS V. 9.2 (SAS, Cary, North Carolina, USA), Eviews V. 5.1 (HIS Eviews, Irvine, California, USA) and MicroFit V. 4.0 (Oxford University Press, Oxford, UK) were used for statistical analyses.


Analysis of the economic data indicated that there was a steady increase in cigarette production and consumption in Pakistan. Cigarette production has been increasing for the last 20 years, with 75.6 billion cigarettes produced during 2009 (11.35% more than 2008 and 36.46% more than the production in 2004). Estimates indicated an upward trend in cigarette consumption. During the last 10 years, the number of cigarettes consumed declined from 2001 to 2003, but more recently an average annual increase of 7.5% has been observed. Per capita cigarette consumption also increased from 2003 to 2009. Cigarette consumption per adult was 685 cigarettes in 2009, which was 30% more than the per capita consumption in 2003. The results of cigarette consumption are summarised in figure 1.

Figure 1

Cigarette consumption in Pakistan 1981–2009. Data source: economic survey, UN Comtrade, US Department of Agriculture (USDA).5 10

Demand and elasticity analysis

Results of the long-term cigarette demand functions were estimated with the ARDL cointegration technique. As a first step to evaluate the existence of the long-term relationship, the maximum lag ρ were selected and an F test was applied to the joint significance of coefficients of the lagged levels of the variables that is, in ARDL specification, (1) Ho456=0 against the alternative H1 β4≠β5≠β6≠0.

The calculated F value (16.2621) at lag length of order 1 was compared to the asymptotic critical value bound (lower bound=6.31 and upper bound=7.52 at the 1% significance level), which indicated an existence of the long-term relationship. On the basis of SBC and AIC, ARDL of order (1,0,0) was selected for estimating the long-term parameters of the model. The ARDL of order (1,0,0) model showed that the given cigarette demand model followed the partial adjustment or myopic addiction model. Table 1 summarises the results of the ARDL model for the long-term cigarette demand functions.

Table 1

Results of long-term elasticities of cigarette demand. The long-run demand model based on ARDL (1,0,0)

The long-term estimates showed that there would be a significant negative effect of price and income on the demand for cigarettes; −1.17 (p=0.014) and −0.84 (p=0.028), respectively. The demand for cigarettes was elastic, as the estimated long-term price elasticity indicated an 11.7% decrease in cigarette demand for a 10% increase in cigarette price. Similarly, a 10% increase in income would result in an 8.3% decrease in cigarette consumption in the long term.

The short-term dynamics of the cigarette demand model are reported in table 2. The estimate for error correction term ECT (−1) was significant; the negative estimate showed the existence of cointegration relationship in the model. The results also indicated that, once deviated from the steady state, the model adjusts at a fast rate of approximately 41% towards a long-term stable relationship. As observed in the long-term estimates, price and income elasticities were negatively associated with the cigarette demand. The short-term income and price elasticities are smaller than their long-term counterparts. In the short term, a 10% increase in price would result in a 4.8% decrease in cigarette consumption. Likewise, a 3.4% decrease in cigarette consumption would be observed with a 10% increase in income. Model diagnostic tests for the ARDL model are summarized in table 3.

Table 2

The short-run dynamics: error correction representation of the selected ARDL Model

Table 3

ARDL model diagnostic tests

The results of CUSUM and CUSUMQ tests indicated that the estimated model was stable. The plots were within the critical bounds at the 5% significance level; thus, the parameters of the model were stable.

Based on the estimated price elasticity of cigarette consumption, the possible impact of increase in price of cigarettes on cigarette consumption is illustrated in table 4. Average prices and taxes were based on 2009 data. The average price of a pack of cigarettes (10 sticks) was PKR 9.11, which included 52% (PKR 4.74) taxes. The government collected total revenue of PKR 35.82 billion in the form of these taxes from the sale of 75 620 million cigarettes. If cigarette taxes are increased to the optimal tax rate of 58.5% with a new retail price of PKR 10.50, consumption will decrease by 18% and in the long term there will be a 30% decrease in cigarette consumption. With these increased taxes, cigarette consumption will decrease to 61 810 million cigarettes. However, tax revenues will increase by 6%, resulting in a total contribution of PKR 38.09 billion to the national economy.

Table 4

Impact of tax changes on consumption and tax revenues in Pakistani Rupees (PKR)


Despite global efforts, the per capita cigarette consumption is increasing in Pakistan. The various surveillance measures of tobacco control do not provide evidence of progress in tobacco control, placing Pakistan in stage 2 of the tobacco epidemic model as proposed by Lopez et al11 Furthermore, tobacco control policies are not well developed and effectively implemented due to lack of public and political support. Many developing countries hesitate to take concrete measures to reduce tobacco use because of the perceived economic benefits that the country derives from tobacco agriculture, manufacture, trade and taxes. However, the economic contributions of tobacco are offset by the harm caused by its use, as the rising social and healthcare costs related to tobacco result in a significant burden to the public sector.

This paper approaches the problem of the tobacco epidemic from an economic point of view. It is timely with the new world trends of tobacco use, tobacco industry and those of policy development. It is the first study to conduct an empirical analysis of cigarette demand in Pakistan. Article 20 of the WHO Framework Convention on Tobacco Control, ratified by Pakistan in 2004, requires member countries to ‘establish a national system for epidemiological surveillance of tobacco consumption, and related social, economic and health indicators’.12 Pakistan lacks any such surveillance system; therefore various indicators of tobacco consumption were obtained from multiple national and international institutions and departments. Per capita tobacco consumption provides a surveillance measure in the absence of prevalence data and provides historical trends. Although it does not indicate who is using tobacco and is prone to distortion because of smuggling or unreported cigarette production or sale, it provides more conservative estimates due to underestimation of the number of cigarettes consumed. Data regarding cigarette imports was not available in physical units for certain years, therefore weight units were converted to physical units (1 cigarette=1 g). This conversion might not be valid, thus biasing the consumption estimates. It is likely this would underestimate total consumption.10

Tobacco taxes are an important tobacco control measure. Ideally there should be a decrease in the revenue generated through tobacco taxes because of effective tobacco control. However, increasing tobacco taxes is an exception. In Pakistan, tobacco is one of the leading contributors to tax revenues; sales tax and excise duty from cigarettes and tobacco contribute a 5% share of indirect taxes. During 2005–2008 there was a 22.6% (US$360 million–441.3 million) increase in these indirect tobacco taxes. An increase in tobacco taxes is an important tobacco control measure. Some policymakers argue against such an increase on the basis that increased taxes would lead to a decrease in sales, resulting in lower revenues. However, the opposite has been observed in developed countries. The amount of tobacco taxes has decreased in Pakistan since 2003. In 2003, total taxes on cigarettes were 78% of the retail price (63% excise duty+15% sales tax) whereas in 2009 these were 52% of the retail price (38.2% total excise+13.8% sales tax).13

To estimate the long-term and short-term cigarette demand functions we applied the ARDL cointegration technique developed by Pesaran et al.14 The merits of using this technique are that it does not require the underlying variables to be integrated of the same order and it gives robust results for small data size. The cointegration technique is based on the idea that some non-stationary variables may deviate from each other in the short term but they tend to drift at roughly the same rate. The ARDL approach selected the optimal lag lengths of the variables, suggesting that the myopic addiction model best explains the smoking behaviour in Pakistan.

Unlike other low-income countries where cigarette demand is more price sensitive (price elasticity ranges between −0.5 and −1), price elasticity of cigarette demand in Pakistan was similar to middle-income countries, such as Turkey and China.8 15 The estimated price elasticity suggests that a 10% increase in cigarette price would result in a 4.8% decrease in short-term demand for cigarettes. However, the same increase of 10% in price will lead to a 11.7% decrease in long-term cigarette demand. These findings for short-term estimates are similar to the studies conducted on pooled data from other countries, but estimates for long-term price elasticity are more elastic as compared to other developing countries.16 17

Contrary to economic theory, income elasticities bear a negative sign. Income elasticity indicated that a 10% increase in income would lead to a 3.4% decrease in cigarette consumption in the short term and an 8.4% decrease in the long term. Although, these estimates of income elasticity are different from the results based on data from other developing countries, the same association is observed in developed countries. The negative income elasticities can also be explained by the fact that there are certain factors peculiar to particular economies for which economic theory does not explain the true depiction of the data. Furthermore, the cigarette price data was for the cheapest brands, and their common use among lower income groups highlights their lack of awareness about harmful effects of the tobacco use. As tobacco use is associated with lower socioeconomic status, a higher prevalence of smoking in the lower income group as compared to the higher income group would result in this negative association. The most plausible reason for then negative impact of income on the demand for cigarettes (pertaining to time series analysis) may be the fact that as per capita income increases, overall awareness in the population increases and cigarette consumption is contained.

The effect of price on cigarette consumption has implications for economic and health policies. An increase in cigarette prices through taxes would result in decreased consumption and increased tax revenues to the government. An increase in the existing tobacco taxes to the optimal tax rate of 58.5% will result in a 16% increase in cigarette prices, and in the long term will lead to an 18% decrease in cigarette consumption. A proportionate increase in price through elevating taxes should exceed the proportionate decrease in cigarette consumption; reduction in tobacco demand will not have an adverse effect on tobacco tax revenues. In fact, there will be a 6% increase in revenue generated through tobacco taxes. In the short term, cigarette demand is not elastic; the percentage increase in prices would be higher than the cigarette demand. However, in the long term demand is more elastic and a relatively lower decrease will result in a greater decrease in cigarette consumption. An increase in tobacco taxes will result in achieving the goals of tobacco control in Pakistan.

What this paper adds

  • While in high-income countries smoking prevalence is decreasing, an upward trend is projected in low-income and middle-income countries. In developed countries an increase in cigarette taxes has consistently led to significant reduction in cigarette smoking. This is the first study to conduct an empirical analysis of cigarette demand in Pakistan.

  • In Pakistan, an increase in cigarette prices through taxes would result in decreased consumption and an increase in tax revenues to the government. This finding can benefit policymakers as it provides information on the characteristics of cigarette consumption and cigarette demand functions that may help in planning tobacco control strategies in low-income and middle-income countries.


Authors would like to acknowledge the support of the Oklahoma Tobacco Research Center.



  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.