Tourism-led growth and granger causality : issues on publication bias

Abstract

We present the main findings of a research project concerned with the study of the presence of publication bias and significance bias in the literature on the relationships between tourism and economic growth by means of tests of Granger causality. We aim at giving answers to three related research questions: (!) Is there evidence of a preference for the publication of studies tending to the validation of any of the Granger causal hypotheses between tourism and economic activity? (2) Is there evidence of a preference for the publication of statistically significant results, independently of the Granger causal direction dectected? (3) Do the results of this line of research represent agenuine empirical association between tourism and income? The answers found are, respectively, "yes", "yes" and "no". Together with the additional finding that some methodological choices are correlated with the size of the emmpirical effects, we conclude that there are reasons to mistrust the typical methodological approaches followed by the literature reviewed, which cast doutt on its usual implications for economic policy. Nevertheless, purged from bias, we confirm previous assertions that the variability of the emperical effects can be explained by the degree of tourism specialization, by the level of economic development and by the size of the countries analyzed.