Trade and openness are Pareto optimal, which is the main argument for globalization. Empirically this has been a difficult question to answer. In the past 15 years or so empirical analysis focused on micro-level plant-level, data instead of aggregate time series. There have been two competing hypotheses: “self selection” and “learning-by-exporting”. In the former, causality runs from productivity to exporting; the opposite in the latter. Clearly there are serious policy implications. We test those hypotheses for four Arab countries (Morocco, Egypt, Oman and Qatar) using a variety of methods. There is evidence for both self-select and learning-by-exporting in Morocco. Exporting industries’ productivity levels dominate those of non-exporting industries in Morocco and Egypt. But there is no evidence in favor of exporting industries in the oil-producing countries, Oman and Qatar.