It might be a factor, certainly cellular studies only show a partial long-term memory factor in cells from the hippocampus. Noted cell changes only cover a very short period in long-term memory suggesting that there is a write-back effect that writes the memory back to the cortex during the first 2-3 years of storage.
To my thinking, attributing accountability for infantile amnesia to the immaturity of the PFC is analogous to blaming the loss of a battle upon the incompetence of a Commander-in-chief who hadn't had any army to fight it - or (perhaps more fittingly), who lacked the means by which to interpret any intelligence from the front line describing its course of events. The competence of any executive role - that of deciding what tactics to deploy in any given situation - is squarely based in receiving intelligence of actual situations, in experientially derived estimates of their negative and positive implications, and the feasible goals that are to be prioritized in relation to each situation.
The development of the PFC's executive "competencies" - its functional maturation - is one thing. The development of the competencies required to define integrated feature-rich autobiographical memories is quite another, and there is no logically necessary dependency of one upon the other. The competencies required for either are provided for by cognitive assemblies that encode experientially derived (or evolved) knowledge of the world and the significance of it to the individual's moment-by-moment standing in the world.
In infancy, the activity of cognitive assemblies, which obviously can initially have no knowledge of the extra-uterine world, is necessarily taken up with learning to selectively attend to particular sensory stimuli from the jumbled totality that presents itself at every given moment, specifically those kinds which enable the infant to develop control over its motor actions - and the significance and import of the knowledge consequently encoded by cognitive processes, I would conjecture, amounts to no more than that required for achieving such control. From such classic experiments as those of Held and Hein (1963) and Thorndike (1898) we know that the ability to self-initiate orientated, exploratory movements is prerequesite for acquiring understanding the meaningful implications of environmentally presented features, while the development of successful visual search abilities would seem to be initially dependent upon recalling motor-patterns associated with achieving the goal of particular searches that cognitive assemblies could eventually co-represent in a generic form, prior to PFC's ability to retain short-term imagery of a sought object's form and likely location.
Cognitive development in infancy, then, involves the progressive encoding of increasingly composite and structured context-associated representations of environmental data. The sequence of developing the component abilities that eventually permit rapid encoding of events by cognitive assemblies - the necessary competencies for encoding the richly-detailed, thoroughly differentiated and yet integrated representations that constitute long-term autobiographical memories - spans the first 2 1/2 years of infancy, the amnesiac epoch. It is only then that the PFC's potential for conscious re-representing and mental manipulating of the output of cognitive representations kicks in. Neither then nor subsequently, however, does PFC take over the task of encoding autobiographic memories; rather, it decodes the content of whatever is fed to it and decides which downline modules should receive it.
Memory, for a child younger than three years old is vastly subconscious as we know. In my opinion, this is where we should look for an answer. The right hemisphere controls this memory up to those three years old. In a sense, the whole body of a child memorizes. I touch that point in my work to be published: Vers le Cerveau du Bonheur de l'Homo Totus, as well as in my two published books. Memory is such an amazing subject, thanks for posting the question.