After reading advanced papers on MSF (notably the Project Management Discipline), I put some thought on the relationship between MSF and COCOMO. MSF has reference to COCOMO, but does not particularly address the relationship between the two.
Starting from COCOMO, they have the basic equation of Effort = a * (KSLOC ^ b). The closest thing to MSF is the KSLOC variable. I would need to find the parameters to complete the following equation: KSLOC = ? (MSF attributes).
Looking at MSF track and cycles, the earliest item that MSF has would be Vision, Persona, Scenario, and Quality of Service Requirement. Judging from quantitative relationship, the above equation would be something that looks like: KSLOC = (c * number of Persona) + (d * number of Scenario) + (e * number of QoS Requirement).
My first impression, c, d, and e constants are to be generated emprically. But this approach would require a lot of data and the data itself has to come from the same source (that means, all Scenario has to be of the same scope, etc.) The problem lies in researching for the data, and normalizing the data itself. We'll see how this will evolve in the future.