This is a very interesting question. Your question seems to mix up two components of a research paper---(1) the theory or hypothesis, which is what the authors set out to validate through the study, and (2) the literature review, which is where the authors use existing literature to build a case for their theory.
To quote your researcher friend: "The theory I wish to develop does not exist out there; I develop and support it." This would not constitute the literature review but the actual study itself or the research question.
Coming to the literature review, it cannot be entirely based on speculation, although it's okay for the authors to take a guess at some specific points and then look for references to support them. Ideally much of the literature review should be undertaken before beginning the study. Authors should read through whole papers and related papers cited within them to form a view of the established trends of thought in the field. It would also be useful at this stage itself to highlight or mark out statements that they think might be useful references for their manuscript. They should then formulate their research question and study design so as to fill a gap in the existing literature.
At the manuscript writing stage, the authors should already have a good grasp of the related literature that exists. They then need to put this literature into perspective to support their research question and findings. At this stage, they may make specific statements and then look for references to support them, in the manner you have mentioned. However, since by this time, they should already be familiar with the literature, the statements they make in their manuscript are not likely to be purely speculative. It is also important that the authors cite statements from previous studies with due consideration to the context in which these statements were made in the original studies.
I hope this answers your question!