This is hard. Which is why you identify obvious outliers first. It's much easier to identify obvious slackers and top engineers who sell their soul for your company (again, not a good thing, but for the sake of the argument -- and realism -- we're forgoing ethics here) than it is to identify lower-than-average or higher-than-average employees who are much closer to the median.

Next, a company looking for saving costs would be looking at top-earning employees with a long tenure. There is most likely not a lot of fat to trim here, but the profits could outweigh the risks.

If you're looking to increase productivity and profits only, you'd also have a look at non-value-producing employees, or those who introduce friction. These employees may add to a company's culture by providing equality strategies or acceptance groups in strong economic times, which in turn could also attract more employees with a similar culture fit, but money is scarce now and shareholders want profits, so they must go. I've seen this happen first hand.