This paper provides evidence in support of two important puzzles regarding the relationship of financial development and economic growth documented in a number of recent papers. The first puzzle relates to the finding that banks have a positive effect upon economic growth, when data are averaged, but a negative one when the highest annual frequency is used. The second is about the positive effect of stock markets upon economic growth irrespective of the averaging. On the first puzzle, although our results provide further empirical support to a negative effect of banks’ development upon economic growth in the short run (annual data), they do not provide the expected evidence of a strong positive relationship for the long-run. On the second puzzle related to the impact of stock markets upon economic growth, we find that the sign of the relationship strongly depends on the variables chosen, the method of estimation and the possible role of self-selection bias. This paper uses recent developments in panel data analysis, including panel unit root tests for a sample of 120 countries for 37 years.