This is not a direct answer to the original question. But, in the comments I invited the OP to take a broader perspective on the issue, which eventually led to a reconsideration by him on the most adequate approach to take. I'm including this as an answer at the OP's request.
In general, when you are trying to customize a publisher provided style, such as
elsarticle-harv, you should think carefully about it. If the publisher supplied the style it is likely that "customizations" might break submission guidelines. I'm not saying there aren't cases in which such adjustments are considered legitimate and are accepted. But you should consider carefully what you are doing, why you are doing it and if such changes are indeed acceptable by the publisher.
But, if you are trying to adjust one of these styles just because it seems to you a "nice starting point" and the style itself is not required from you for reasons of submission, then you might well consider changing to one of the standard styles supplied by
natbib. There is plenty of styles for you to choose from, most likely one fits your bill. You will probably get better support on them and also have more flexibility in their use.
authoryear style as an example. It indeed is close to the format you specified and because, as I am an humanities guy,
biblatex is usually the best choice for me. But that might not be your case.
In choosing between
bibtex) you should weight their advantages and disadvantages for your case. Take a look at: bibtex vs. biber and biblatex vs. natbib
Summing up the main issues there, I think it is fair to say that
biber) has a number of technical advantages, is in active development and is more easily customizable. On the other hand, a number of publishers do require that you stick to
bibtex). Anyway, more detailed info at the link.