How responsive is cigarette demand to price and

How responsive is cigarette demand to price and income (expenditure ) changes in Uganda? Grieve Chelwa Emerging Research Programme CAPE TOWN 22 – 26 June 2015

Introduction & motivation • Uganda’s smoking prevalence relatively high for SSA – – 17% male (vs 14% for SSA) 2% female (vs 3% for SSA) 19% young males (vs 9% SSA) 11% for young females (vs 3% for SSA) • GDP growth has averaged 7% per year – People getting better-off • Inflation at 10% over same period – Cigarettes getting cheaper in real terms Price and income elasticities 22– 26 June 2015

Introduction & motivation • Increase in the incidences of heart disease, chronic obstructive pulmonary disease, strokes and cancer (WHO, 2012) • We know that tax & price measures are single-most effective tool (Chaloupka & Warner, 2000; IARC, 2011) – Excise tax burden very low in Uganda • But still policymakers demand local evidence – Few studies on demand elasticities for SSA outside South Africa – Small but growing literature using expenditure surveys • Nothing yet for Africa Price and income elasticities 22– 26 June 2015

Preview of results • I find estimates of the price elasticity & income elasticity of demand in-line with international evidence – Price elasticity: -0. 30 to -0. 40 – Income elasticity: ~ 0. 10 Price and income elasticities 22– 26 June 2015

Literature review • 2 approaches to estimating price & income elasticities of demand for cigarettes – Time series – Cross sectional data • Time series studies – Reviewed in Chaloupka & Warner (2000); IARC (2011); Guindon et al. (2015) – Price elasticity: -0. 2 to -0. 6 – Income elasticity: 0 to 0. 6 Price and income elasticities 22– 26 June 2015

Literature review • Cross sectional studies – Shortcomings with time series analyses – Most African countries rarely keep time series records • New literature uses expenditure data & method by Deaton (AER, 1988) – John (2005); John (2008); Eozenou & Fishburn (2009); Guindon et al. (2011); Chen & Xing (2011) – Price elasticity: -0. 10 to -0. 60 – Income/expenditure elasticities: 0 to 2 Price and income elasticities 22– 26 June 2015

Data • I use the 2005, 2009 editions of the Uganda National Panel Survey – Nationally Representative – Conducted by UBOS – Sample of ~ 3, 000 households • I also run specifications where I pool the two samples Price and income elasticities 22– 26 June 2015

Method Motivating assumption: cigarette prices vary across geographical space in Uganda Price and income elasticities 22– 26 June 2015

Method • Price and income elasticities 22– 26 June 2015

Method • Unit values are however not “prices” – Measurement error – Quality effects • We can check for quality effects by running: Price and income elasticities 22– 26 June 2015

Quality effects Price and income elasticities 22– 26 June 2015

Method • Standard demand equation: Price and income elasticities 22– 26 June 2015

Demand equation Price and income elasticities 22– 26 June 2015

Method • Now work at cluster level: • Cluster-level regression: Price and income elasticities 22– 26 June 2015

Elasticity formulae • Price elasticity of demand: • Expenditure/income elasticity: Price and income elasticities 22– 26 June 2015

Price elasticities Price and income elasticities 22– 26 June 2015

Expenditure elasticities Price and income elasticities 22– 26 June 2015

Robustness: Price elasticities Price and income elasticities 22– 26 June 2015

Robustness: income elasticities Price and income elasticities 22– 26 June 2015

Conclusion • Cigarettes in Uganda are price inelastic – Demand likely to decline by 3% to 4% for every 10% increase in prices – Within estimates from elsewhere • Cigarettes are likely normal goods in Uganda • Government can decrease consumption of cigarettes without sacrificing tax revenues Price and income elasticities 22– 26 June 2015