Economists have long argued that developing countries have the potential for high productivity growth if they adopt existing technologies and apply them to the local context. This report brings to bear a battery of new data sources to explore the innovation "paradox": despite the potential for very high returns, developing countries invest far less in adopting and inventing new processes and products than advanced countries. The report posits three broad factors underlying this paradox. The first is that firms in developing countries lack the managerial and technological capabilities to undertake meaningful innovation projects. This implies that conventional innovation policies are unlikely to be effective, and moving firms up the "capabilities escalator" becomes central. A second factor is that firm capability is only one of many critical ingredients - for instance, access to financial markets, macroeconomic stability, and imported machinery - that are complements to the innovation process, and whose absence lowers the return to innovation in developing countries. This implies that cultivating an effective innovation system will be a greater policy challenge, and that standard measures of innovation performance, such as research and development or GDP, are misleading. Finally, government capabilities required to redress these two points are also correspondingly weaker in developing countries, so building these capabilities needs to be explicitly integrated in formulating innovation policy