Pharmaceutical expenditure and gross domestic product: Evidence of simultaneous effects using a two-step instrumental variables strategy.

Mujaheed Shaikh,Afschin Gandjour

doi:10.1002/hec.3832

Abstract

This paper estimates the income elasticity of government pharmaceutical spending and assesses the simultaneous effect of such spending on gross domestic product (GDP). Using a panel dataset for 136 countries from 1995 to 2006, we employ a two‐step instrumental variable procedure where we first estimate the effect of GDP on public pharmaceutical expenditure using tourist receipts as an instrument for GDP. In the second step, we construct an adjusted pharmaceutical expenditure series where the response of public pharmaceutical expenditure to GDP is partialled out and use this endogeneity adjusted series as an instrument for pharmaceutical expenditure. Our estimations show that GDP has a strong positive impact on pharmaceutical spending with elasticity in excess of unity in countries with low spending on pharmaceuticals and countries with large economic freedom. In the second step, we find that when the quantitatively large reverse effect of GDP is accounted for, public pharmaceutical spending has a negative effect on GDP per capita particularly in countries with limited economic freedom.

Highlights

This paper confronts two questions related to medical spending
We find that the income elasticity of public pharmaceutical expenditure is greater than unity in the full sample, and that there is a negative effect of public pharmaceutical expenditure on gross domestic product (GDP) per capita
The estimate is significant at a 5% level and quantitatively means that a 1% increase in public pharmaceutical expenditure is related to a 0.09% decrease in GDP per capita

Summary

| INTRODUCTION

This paper confronts two questions related to medical spending. First, the paper estimates the income elasticity of government pharmaceutical spending, and second, it assesses the simultaneous effect of such spending on income itself. Our contribution to the literature in health economics is twofold by—(a) assessing the income elasticity of public pharmaceutical spending across a multitude of countries using IVs, something which past studies have not considered, and (b) assessing the effect of pharmaceutical spending on a nonhealth outcome such as economic growth. It is important to mention here that there will be correlation between the error terms of the two equations only if there are omitted variables that will affect the within‐country changes in both GDP and pharmaceutical expenditure simultaneously For this reason, we include several time varying controls in Wit that may be important and may determine both GDP and pharmaceutical expenditure along with country and time fixed effects that take into account other unobservable heterogeneity related to country specific characteristics and common shocks, respectively. To interpret the estimates as elasticities, we take the log form of both GDP and public pharmaceutical expenditure

| MAIN RESULTS

Findings

| CONCLUSION