In this paper we add new results to the emerging field of investigating productivity levels rather than productivity change, as initiated by Hall and Jones (1996, 1997). To obtain measures of relative productivity levels we depart from traditional growth accounting and calculate the Malmquist index of total factor productivity change using the nonparametric Data Envelopment Analysis (DEA) for a broad cross country sample. This index can be decomposed in measures of technological progress and efficiency change that are cumulated to level measures. The so obtained heterogeneity in productivity levels is next related to several determinants of technology driven growth in an econometric exercise. Doing this (1) we are able to provide confimation of the validity of the decomposition of the Malmquist index and (2) we find innovation-related explanations for international technological frontier shifts and imitative catching up and falling behind.