Fred Brooks said that a "programming product" takes about 3 times as much work as a plain program (so it becomes a module reusable by anybody). The extra work is in things like generalizing, testing and documenting. He claimed that a "programming systems component" takes about three times as much work again (that is, a module reusable somewhere else), with the work going into things like precisely defining interfaces, and integration testing. p.6-7, MM-M. He said all that long before object oriented was coined (even before ADTs), so he wasn't talking about objects, but about code reuse in general.
What's the evidence? We see notoriously few FP libraries, but 10,000's of Java libraries (as an OO eg). The latter is a major reason touted by JVM languages for targeting it.
While FP individuals do seem to reuse their own code quite a bit, the code reuse of systems programming products remains mostly theoretical for FP - but standard practice for OO.
I think a more relevant distinction is between component consumers and component producers. The former learns libraries; and latter creates them. The former is closer to the customer and their application domain (where value becomes tangible); latter is closer to the raw materials, the essence of programming and nature's wild bounty (where value originates).