Assuming that you are looking at the project as a whole, you need to identify items that can be measured and that can be collected on all projects.
Also as this is for senior management you need to be careful about the level of information provided, they usually prefer a higher level of detail.
Some things that could be measured for projects, at least we could in my environment are:
<ul type="square"> [*]Project Completed on Time[*]Project Completed within budget[*]SDLC phases completed as planned[*]Help desk Rickets opened after project goes into production[/list]
I would add that these metrics do not measure efficiency - efficiency in general is rather subjective and relative and would only work if you defined efficiency as delivering a project on time and within budget.
I have not failed. I've just found 10,000 ways that won't work" --Thomas Edison
One way to measure efficiency is by comparison.
What are you going to measure? Things like: Percentage of cycles that completed within the time frame? – Maybe the time frame was too short, or too long – so it is an error in the process design. Budget is the same thing.
You have to compare how the team is doing compare to other teams working on similar projects (with taking into account the budget differences, number and professionalism of members, and so on) or measure your team throughout history of projects.
One thing that you can check without comparison is how are you using your resources? For instance, if you have a developer in your QA team, what is the percentage of time that she spends on tasks that are not development related?
I am not sure if I understand correctly what you mean by "development efficiency" but one way to evaluate projects could be to compare the results of the project (expressed in lines of code, function or use case points for instance) and the ressources needed to provide them (man/month or $ for instance). Please note that if you start to debate on how to measure outputs and inputs of software project you will get more answers that particpants ;o)
Are you not talking about a metric which would display how many defects were found in development, how many in testing and then how many in the field? You'd then have a figure, a weighted figure, which would indicate how efficient dev was? Or maybe you'd have to link that in to timesheet data to work out how efficient it was. And anyway, what does how efficient the development of a product was matter?