Based on the cross- and intra-linguistic distribution of Person Case Constraint (PCC) effects, I argue that there can be no agreement in phi-features (person, number, gender/noun-class) which systematically lacks a morpho-phonological footprint. That is, there is no such thing as "abstract" phi-agreement, null across the entire paradigm. Applying the same diagnostic to instances of clitic doubling, we see that these do involve syntactic agreement. This cannot be because clitic doubling is agreement; it behaves like movement (and unlike agreement) in a variety of respects. Nor can this be because clitic doubling, qua movement, is contingent on prior agreement – since the claim that all movement depends on prior agreement is demonstrably false.

I propose that clitic doubling requires prior agreement because it is an instance of non-local head movement, and movement of X⁰ to Y⁰ always requires a prior syntactic relationship between Y⁰ and XP. In local head movement (à la Travis 1984), this requirement is trivially satisfied by (c-)selection. But in non-local cases, agreement must fill this role.