Estimating the Tobacco Kuznets relationship in Spain (1957–2023): income curvature, projected turning points, and policy implications

Abstract

This paper estimates a Tobacco Kuznets Curve (TKC) for Spain on the full national series, 1957–2023, to estimate the income–consumption curvature underlying the Tobacco Kuznets hypothesis and derive the associated model-implied turning points. Understanding these model-implied turning points may contribute to discussions of tobacco control, taxation, and regulation, and to translating long-run demand patterns into operational policy benchmarks. Using annual adult cigarette consumption, real GDP per capita, and real pack prices, we estimate three nested models: (i) a simple bivariate benchmark specification included for illustrative comparison purposes only; (ii) an augmented model adding ln(real price) and post-2005/2010 indicators; and (iii) a segmented-trend specification interacting each indicator with time. Baseline estimation relies on OLS with Newey–West HAC inference, while robustness checks, such as GLS(AR1), Huber RLM, DOLS with leads/lags where available, and ARDL long-run coefficients, ensure that the identified turning points are not artifacts of persistence, outliers and dynamic adjustment. Across specifications, ln(GDP pc) is positive and [ln (GDP pc)]2 is negative, confirming an inverted-U conditional on price and policy timing. The estimated curvature parameters imply turning points between approximately €39,000 and €58,000 and the implied peak consumption (≈ 118–158 packs per adult/year). Short-run price elasticities are negative (about − 0.17), and post-reform dynamics are consistent with a persistent decline. These thresholds should be interpreted as model-implied projections rather than empirically observed policy thresholds. While they help characterize the long-run relationship between economic development and cigarette consumption, they do not by themselves imply specific excise-tax recommendations.