Many theoretical and empirical studies have analyzed the relationship between the economic cycle and tobacco consumption using the GDP and unemployment rates as the key variables for measuring economic phases. However, few studies focus on the pathways that cause tobacco consumption to be linked with the economic cycle, and there are no studies analyzing the heterogeneous effects underlying this relationship across nations and regions. This article explores the relationship and its pathways in 16 Spanish regions for the period 1989-2018. To this end, we apply a Granger causality analysis based on the augmented vector autoregressive (VAR) model in levels and extra lags. This method provides more efficient and robust results than the standard VAR model, which can lead to biased results with limited samples, especially in a region-by-region analysis. The empirical results suggest that the impact of the business cycle on tobacco consumption is heterogeneous and specific to each region. In addition, although recession phases cause a decline in tobacco consumption in Spain, in line with the literature, this procyclical relationship does not occur for expansion phases in all regions. One of the main findings of this article is that in expansion phases, tobacco consumption is sensitive to GDP, while in recession phases, tobacco consumption is affected by unemployment. National and regional governments should consider these results when they develop smoking control policies because homogeneous strategies can lead to heterogeneous results. Thus, the results can be useful for policymakers dealing with tobacco control strategies.
Keywords: Asymmetric causality; Bootstrap; Economic growth; Granger causality; Panel causality; Tobacco.