Conventional wisdom has it that trade en- hances economic efficiency and thus promotes growth. At least since Robert M. Solow's (1957) pioneering work, however, technologi- cal progress has been recognized as the domi- nant factor in determining the rate of growth. This is presumably even more true for develop- ing countries, for which the possibilities of clos- ing the knowledge gap with advanced industrial countries offers especially large growth poten- tial. We examine the impact of trade restrictions in economies in which technological spillovers within countries and across industries are fun- damental to the process of growth (see Kenneth J. Arrow, 1962a, 1962b; Paul M. Romer, 1986; Stiglitz, 1987). Since that work, it has been clear that markets, by themselves, do not nec- essarily, or in general, lead to overall dynamic efficiency; and that there are often trade-offs between static inefficiencies (e.g., associated with patent protection) and long-term growth. We find, here in particular, that the dynamic benefits of broad trade restrictions may out- weigh their static costs. Our analysis provides the basis of an infant economy (as opposed to an infant industry) argument for protection. This paper develops a simple two-sector model with an industrial (modern) and a tradi- tional (craft or agricultural) sector. There are four key features to the model: (a) there are spillovers from the industrial sector to the craft sector, for which firms in the industrial sector are not compensated; (b) such spillovers are geographically based, that is, it is only produc- tivity increases in the industrial sector in the developing countries that affect productivity in- creases in the traditional sector; (c) innovations are concentrated in the industrial sector; and (d) size is among the important determinants of the pace of innovation in the industrial sector. Earlier critiques of trade policies encouraging the development of the industrial sector in de- veloping countries ignored these spillovers. They argued, in effect, that Korea would always have a comparative advantage in growing rice; therefore, it was foolish for it to try to restrict imports of industrial goods, even if by so doing productivity in the industrial goods sector was increased. It could never catch up, so the pro- tection would have to be permanent. Year after year, the country would have been better off if it simply specialized in its own comparative advantage, growing rice. Korea could, and did, catch up, however, at least in certain areas. If catch-up is possible, then dynamic comparative advantage differs from static comparative ad- vantage. But even if Korea's comparative ad- vantage remained in agriculture, industrial protection might be desirable, because by sup- porting it, one might have a more dynamic agricultural (traditional) sector. Trade restric- tions enhance the size of the industrial sector; the benefits spill over to the rural sector; and national income grows at a possibly far faster pace. After presenting the model, we explain why the underlying hypotheses are plausible and argue that the model is broadly consistent with historical experience and empirical evidence.