Price Elasticity Of Alcohol Demand In India

Download Price Elasticity Of Alcohol Demand In India

Preview text

Department of Economics and International Business Working Paper No. 16-10 November 2016
Santosh Kumar Department of Economics and International Business
Sam Houston State University 1821 Ave I, Box 2118 Huntsville, TX, 77341 [email protected]

Abstract Using a household survey conducted in 2014, this study estimates price elasticity of demand for beer, country liquor, and spirits in India. Ordinary least square models were used to estimate the responsiveness in alcohol demand due to price change. We include a large number of control variables to adjust for potential confounding in the model. Inter-district variation in alcohol consumption is adjusted for by including district fixed-effects. Alcohol prices are negatively associated with demand for alcoholic beverages. The price elasticity of demand ranged from -0.14 for spirits to -0.46 for country liquor. Low level of education was positively associated with spirits consumption. The magnitude of elasticity varied by rural-urban, education, and gender. Results indicate a policy mix of price controls and awareness campaigns would be most effective in tackling the adverse effects of harmful drinking in India. Keywords: Price elasticity of demand, Alcohol demand, Public policy, India JEL Codes: I18, D12
Acknowledgments: The survey was financially supported by International Alliance for Responsible Drinking (IARD). This publication presents the work product, findings, viewpoints, and conclusions solely of the named author. The views expressed are not necessarily those of any of IARDs sponsoring companies.

Price Elasticity of Alcohol Demand in India
INTRODUCTION Alcohol consumption in India has been rising rapidly in the last decade. On average
30% of Indians consume alcohol, out of which 4-13% are daily consumers and more than half of those who consume alcohol are hazardous drinkers (WHO, 2012; Ray et al., 2004). The per capita consumption of alcoholic beverages in India increased by 38 percent, from 1.6 litres in 2003-05 to 2.2 litres in 2010-12 (WHO, 2012). Against the global average of 16 percent, about 11 percent of Indians were binge drinkers.1 Excessive consumption of alcoholic beverages has been found to have a detrimental effect on health. There is overwhelming evidence to suggest that alcohol consumption is associated with a variety of disease and disability (Whiteford et al., 2010; Lim et al., 2012). Liver cirrhosis, cancers, tuberculosis, HIV, and injuries are some of the adverse health effects caused by drinking alcohol (Baan et al., 2007; Shield, Parry & Rehm, 2013). The WHO reports that excessive use of alcohol accounts for 5.9% of all deaths worldwide (WHO, 2014). In India alone, 350,000 deaths were attributed to alcohol consumption in 2010 (Lim et al., 2012). The recent data indicates that 15 people die every day or one every 96 minutes from the harmful effects of alcohol consumption in India (NCRB, 2013). In addition to adverse health impacts, alcohol use also contributes to poverty and impoverishment either due to diversion of resources away from productive use or increasing healthcare cost associated with alcohol-related problems ((Benegal, 2005; Bonu et al., 2005; Gajalakshmi and Peto, 2009; Rathod et al, 2015).
1 Binge drinking or heavy episodic drinking is described as heavy consumption of alcohol over a short period of time. 1

Given the increasing evidence on the harmful effects of alcohol consumption, policymakers have resorted to either increasing alcohol prices through taxation or have put blanket bans on alcohol consumption. Alcohol consumption is prohibited in the Indian states of Gujarat, Kerala, Bihar, and Nagaland. However, either the prohibition or the price increase can be an effective policy for reducing alcohol consumption if the demand for alcoholic beverages is price sensitive and price-elastic.2 If the consumers have inelastic demand for alcoholic products, then price control through taxation or prohibition may not be an effective policy instrument to curb the adverse effects of alcohol consumption. Prohibition is less desirable because it severely restricts freedom of individual choice and may have undesirable and unintended effects as was the case in the failed alcohol ban in the USA from 1920 to 1933 (Thornton, 1931, Mahal, 2000).3
Therefore, having reliable information on price elasticity of demand (PED), the percentage change in demand for alcohol resulting from a one percent increase in alcohol price) by different characteristics of drinkers (such as gender and caste) are important for formulating appropriate tax policies to decrease alcohol consumption. There is lack of credible estimates of price elasticity for alcohol beverages in India, which is important for implementing effective interventions. There have not been many estimates of price elasticities for different alcoholic beverages in India to date except Mahal (2000) and using a representative cross-section of households from five states, this paper provides additional and more reliable estimates of price elasticity of demand for beer, spirits and country liquor India.4
2 Several studies have shown that alcohol price is a key determinant of consumption (Anderson et al., 2009; Wagenaar et al., 2009). 3 The period was marked by rampant smuggling, corruption and black market. 4 These five states account for one-third of India’s population.

In high-income countries, the literature on estimation of price elasticity of demand for alcohol products is quite extensive but diverges markedly in the magnitude of elasticity estimates. Some studies indicate that alcohol demand is elastic (price elasticity is greater than one), while other studies suggest the demand to be price inelastic (price elasticity is less than one). Three recent meta analyses comparing cross-beverage elasticity have found that beer, wine, and spirits have different own-price elasticities, with beer appearing to be less elastic than wine and spirits (Fogarty, 2010; Gallet, 2007; Wagenaar et al., 2009). Gallet (2007) and Wagenaar et al., (2009) reported an average PED for alcohol of -0.5, meaning that a 20% increase in alcohol price would reduce the demand for alcohol by 10%. In UK, Meng et al. (2014) found the price elasticity estimates to range from -0.08 to -1.27 and beer was most elastic beverage.
The dearth of research on estimation of PED for alcoholic beverages in low and lower middle-income countries, including India, calls for additional research. To the best of our knowledge, we are aware of the following two studies that deals with the estimation of PED for alcohol in India. In a simulated study, Mahal (2000) found that the own price elasticity of participation in moderate to heavy alcohol consumption is 1.00 for individuals aged between 15 and 25 years and 0.50 for individuals aged 25 years and above. The estimates in Mahal (2000) are smaller than estimates for one state (Andhra Pradesh) by Reddy, Reddy, and Dheeraja (1999). Reddy, Reddy, and Dheeraja (1999) found an arc elasticity of demand for arrack (local liquor) in the range of -1.23 to -1.36, but this analysis was carried out on a very small sample of 86 moderate to heavy alcohol consumers of arrack in Andhra Pradesh.

In India, the prevalence of alcohol consumption has been on the rise and policy makers are struggling to design an appropriate tax system to reduce alcohol consumption. In several instances, higher alcohol prices has led to consumption of spurious alcohol by poor households thereby resulting in premature loss of lives. Given the complex socioeconomic conditions of households and lack of credible estimates of PED for alcohol in India, findings of this study will be important to design alcohol price strategy so that harmful effects of alcohol consumption can be minimized.
The data used in this study are from the Survey of Unrecorded Alcohol in India (SURA) collected in 2014. Data collection for this cross-sectional survey was funded by the International Alliance for Responsible Drinking (IARD) in order to assess the prevalence of unrecorded alcohol drinking in India. The survey sampled approximately 1200 respondents in each of the following five states- Andhra Pradesh, Kerala, Madhya Pradesh, Maharashtra, and West Bengal. The sample was selected under a semi-purposive, multi-stage probability design, and oversampled respondents in rural areas. In the first stage, two districts were randomly selected based on the socio-economic profile of the districts in each state.5 In the second stage, 10 urban wards/towns and 20 rural villages were selected from each district using the probability proportional to size (PPS) sampling method. Urban wards/towns and rural villages formed the primary sampling units. Finally, in stage three, 20 respondents were selected from each primary sampling unit in each
5 Districts were stratified based on proportion of schedule caste and tribe population, female literacy rate, and percentage of households belonging to lowest wealth quintile.

district. In addition, 50 respondents were purposively sampled from two randomly selected slums in each of the sampled urban wards/towns. The overall response rate was about 85%, and there was no significant difference in response rates between the urban and rural samples.
The survey covered individuals aged 15 years or older. Among the eligible individuals in the households, the member with the most recent birth date were selected for the interview. Our initial sample included 6088 individuals. Of these respondents, 3988 (65%) respondents resided in rural areas while 2100 (35%) respondents resided in urban areas. The survey included questions about past and current drinking and about the frequency and quantity of alcohol use in the past year. Of the total sample, 38.6% were current drinkers, 53.6% were lifetime abstainers, and 7.8% were former drinkers.
Detailed questions about the drinking habits, patterns, and beverage type were asked to current drinkers only. The survey collected information on the socio-economic and demographic characteristics of the respondents, such as age, gender, caste, marital status, income, and family size. Price information was collected for the “most consumed drink (MCD).” Using the information in beverage-specific alcohol consumption module, the most consumed drink is identified as the beverage with highest consumption by volume (quantity x frequency). Price and quantity data on the MCD were used to estimate price elasticity of demand for different types of alcoholic beverages. We restrict the analyses to the sample of respondents who reported beer, spirit, and country liquor as their most consumed drink. Price information about homemade alcohol drinks was missing for a large number of homemade alcohol respondents, therefore, homemade drinkers were excluded from the analysis.

Estimation The standard approach to estimate price elasticity of demand is to quantify the
empirical relationship between price and alcohol demand, after adjusting for socioeconomic characteristics of the respondents including income. Socio-economic characteristics are able to capture differences in tastes and preferences across individuals. The linear relationship between price and demand is transformed into logarithmic (log) form, and the estimated model can be represented by the following equation for each beverage:
𝐿𝑜𝑔 (𝑄𝑢𝑎𝑛𝑡𝑖𝑡𝑦𝑖𝑑𝑠) = 𝛼1 + 𝛽1log(𝑃𝑟𝑖𝑐𝑒𝑖𝑑𝑠 ) + 𝛾1𝑋𝑖𝑑𝑠 + 𝜂𝑑 + 𝜖𝑖𝑑𝑠 (1) where 𝑄𝑢𝑎𝑛𝑡𝑖𝑡𝑦𝑖𝑑𝑠 is the dependent variable measuring quantity consumed of each beverage (beer, spirit, country liquor) by respondent i in district d and state s; the main independent variable is log of price of each beverage; 𝑋𝑖𝑑𝑠 is a vector of socio-economic and demographic characteristics of the respondents that can potentially affect alcohol demand (for example, age and gender of the respondent, education level of the respondent, monthly income of the respondent, whether respondent lives in the rural area); 𝜂𝑑 represents district fixed effects so that time invariant characteristics of district that may affect alcohol demand can be adjusted for; and finally 𝜖𝑖𝑑𝑠 is the idiosyncratic error terms in individual-level alcohol consumption, which are uncorrelated with other covariates included in the model. Standard errors are clustered by district to adjust for the possibility that residuals are not independent and identically distributed.

In Eq. (1), 𝛽1 is price elasticity of demand for beer, spirit, and country liquor. In econometric models, where both the dependent and the independent variables are logtransformed, the regression parameter (𝛽1) is interpreted PED. The magnitude of 𝛽1 shows the percentage change in alcohol demand for a specific beverage by respondent i, in response to a percentage change in price of that specific beverage:
𝛽1 = 𝛿[𝛿lo[lgo(g𝑄(𝑢𝑃𝑎𝑟𝑛𝑖𝑐𝑡𝑒𝑖𝑡𝑖𝑑𝑦𝑠𝑖𝑑)𝑠] )] = %%𝛥(𝛥𝑄(𝑢𝑃𝑎𝑟𝑛𝑖𝑐𝑡𝑒𝑖𝑡𝑖𝑑𝑦𝑠𝑖𝑑)𝑠) (2) Some prior studies have used average alcohol price or community-level price instead of beverage-specific actual price paid by individuals (Aayagari et al., 2013; Goryakin, Roberts, & McKee, 2016). Alcohol prices are aggregated due to unavailability of individual-level data on actual price paid by the respondents and to reduce measurement error in individual prices. For comparison, we also estimate average price elasticity by estimating a pooled model that combines the sample of beer, spirit, and country liquor drinkers. For normal goods, the negative relationship between price and demand means that the value of 𝛽1 will be a negative number, meaning that individuals may reduce the demand or shift their consumption to a substitute drink as a result of increase in price.
RESULTS Sample characteristics
Table 1 shows the summary statistics of the variables used in the analysis for current drinkers only. We define current drinking status in terms of whether an individual has consumed alcohol in the past 12 months. Using drinking frequency, number of drinks, and size of the drink, we estimate annual consumption of each beverage in litres. The annual consumption is transformed in natural log. The average log price of alcoholic

beverages ranged from 5.32 to 6.44. Distilled spirits are the most expensive drink type. The majority of current drinkers are male (91%) and the average age of current drinkers is 41 years. About two-fifths of current drinkers are illiterate, and about 60% of the respondents who are current drinkers live in rural areas. Close to two-fifths of the analytical sample earns less than 4000 rupees (equivalent to $65) per month. The average daily alcohol consumption is 25 grams of pure ethanol in rural areas and 30 grams of pure ethanol in urban areas. Price elasticity of demand
In Table 2, we report the results on price elasticities of demand for beer, country liquor, and spirits from the ordinary least square method for current drinkers. Each column reports results from separate regression models. In general, the results in Table 2 indicate that an increase in price has a small negative effect on alcohol demand. The estimated PED for beer and country liquor are -0.33 and -0.46, respectively. The elasticity estimates for beer and country liquor are statistically significant and are consistent with estimates reported in the USA and other developed countries and are well within the range of previous estimates (Wagenaar et al., 2009). The magnitude of -0.33 means that a 1% increase in the price of beer is associated with 0.33% reduction in beer consumption. The PED for spirits is 0.139. However, it is not significantly different from zero. The absolute value of all elasticities is less than one, indicating that alcohol demand is not very sensitive to price change. Male and age are positively associated with alcohol demand, but the coefficients are statistically insignificant except for spirits drinkers. Education is positively associated with spirits demand: illiterate individuals consume more spirits than literate

Preparing to load PDF file. please wait...

0 of 0
Price Elasticity Of Alcohol Demand In India