Abstract

SDMX and DDI represent two of the most widely utilized implementation standards for statistical data processing. The initial versions of both standards were published approximately 20 years ago. Since that time, each has developed independently, with different objectives and minimal communication between the individuals and organizations responsible for their advancement.

It is therefore not surprising that both tend to address similar issues in different ways, each with its own approach. It is also natural in this context that there is a considerable overlap in the coverage of the statistical process cycle.

This duality has led to the perception of DDI and SDMX as "competitors", creating a false opposition that has forced users to choose one or the other, or has "assigned" each to "areas" of the GSBPM where it performs better than the other.

In fact, there is a considerable degree of overlap between the two standards, even greater than has been recognized in the past, but far from this being a problem leading to a forced choice between one or the other, recent work by the Supporting Standards Group and the COSMOS meeting on interoperability has demonstrated that there are significant opportunities for complementarity between the two.

This paper explains some of the conceptual differences between SDMX and the "variants" of DDI, and how both standards could mutually adopt concepts from each other and enhance their capabilities to facilitate such interoperability.