We study an international climate agreement that assigns emission quotas to each participating country. Unlike the simplest models in the literature, we assume that abatement costs are affected by R&D activities undertaken in all firms in all countries, i.e. abatement technologies are endogenous. In line with the Kyoto agreement we assume that the international climate agreement does not include R&D policies. We show that for a second-best agreement, marginal costs of abatement should exceed the Pigovian level. Moreover, marginal costs of abatement differ across countries in the second-best quota agreement with heterogeneous countries. In other words, the second-best outcome cannot be achieved if emission quotas are tradable.