Productivity measured in LOC?

LOC can be used as a measure to count productivity.
But when one does that, it makes the following assumptions.

1. LOC counting is standardized. There is a uniform method for counting lines and this method is applied across projects.
2. Different technologies, domains, languages and methods used for development are segragated and productivity is computed separately.

Overall productivity measure of the project can be obtained taking into account LOC productivity. Assumptions are:

1. For a given domain and development approach, the final LOC is a normalized size of measure.
2. Therefore the effort involved in requirements, design, developing test cases, and conducting testing will be proportional to the size in LOC. The relation between size and effort is not linear but exponential
3. That is, if a 100KLOC project took 100person months - including requirements, design, coding, testing etc., then we can estimate that a 50KLOC project will take proportional effort to go through the entire cycle, provided, the domain, language and development approach are same.
4. Factors like complexity affect such productivity estimates, but once you clearly segragate data based on domain, technology, language, development approach etc., its effects will be negligible.