The industry is experiencing a resurgence in interest in Cobol, both to support existing applications and to take advantage of the continuing evolution of the zSeries platform. Commercial unit testing tools for Cobol are able to exercise code at the level of a whole load module, but cannot exercise individual Cobol paragraphs in isolation. That limitation means we cannot achieve the same degree of granularity in microtests as we can when working in other languages, such as Java, Kotlin, C#, Python, or Ruby.
Given that capability, we might anticipate some of the same effects on the design of Cobol code as we see in the design of code in other languages, when fine-grained microtesting is used. These include factors like modularity, smaller source units, closer alignment with basic software design guidelines such as separation of concerns and single responsibility, and so forth.
As much of the work in this space will involve support for existing code bases, we might also anticipate that incremental refactoring will become easier once we have a safety net of fine-grained examples in place. We can apply the same techniques we use to improve the design of existing code in other languages.