The properties of the productive morphological causative traditionally understood to be suggestive of biclausality have in more recent accounts been attributed to multiple layers within the verbal domain. This paper evaluates two approaches to the causative verbal domain: one requiring multiple stacked vPs as in Harley 2008 and Folli & Harley 2007, and a second in which causation instantiates a separate causative head as in Pylkkännen 1999, 2008. Features particular to the Indic language Kashmiri, such as verbal agreement with the internal argument in ergative clauses, nominative causees, and iterated causation, provide new empirical ground for testing distinct predictions made by these two approaches. This paper ultimately argues that the CauseP approach better accounts for the facts in Kashmiri, and may also be better suited to capture parametric variation of productive causation crosslinguistically.