Composite indicators (CIs) have increasingly been accepted as a useful tool for benchmarking, performance comparisons, policy analysis and public communication in many different fields. Several recent studies show that as a data aggregation technique in CI construction the weighted product (WP) method has some desirable properties. However, a problem in the application of the WP method is the difficulty and subjectivity in determining the weights for sub-indicators. In this paper, we extend the WP method and propose a multiplicative optimization approach to constructing CIs. This approach requires no prior knowledge of the weights for sub-indicators. Instead, the weights are generated by solving a series of multiplicative data envelopment analysis type models that can be transformed into equivalent linear programs. Additional relevant information on the weights, if available, can be incorporated into the proposed models. We apply the proposed approach to the 2005 data of 27 economies in the Asia and the Pacific region in the United Nations' "Human Development Index" study and present the results.