An employee resigns from a department and isn't replaced (not through any affirmative decision to reduce headcount - no one would take the job). Productivity in the department goes up. Another employee goes on maternity leave. Their direct supervisor decides to earmark her salary to hire outside contractors if necessary to backfill the workload rather than hire a maternity leave replacement. None were necessary, and productivity goes up once more. The department manager resigns and those remaining operate relatively autonomously, reporting in to the VP to whom the former manager reported. Once again, productivity, effectiveness, quality, interactions among both internal and external constituencies, all increase. By any accounting, these three particular individuals contributed negative productivity to the department.
I wouldn't be surprised to hear that you're not surprised. We all have run into people among our organizations who are negative contributors. I'm not talking about those who "contribute" their negative personalities - each of these individuals was quite likable, affable, and pleasant to be around. They merely took far more resource to manage, to brief, to keep in the loop, and to work around than the value they contributed to the organization and its various constituencies. What's perhaps surprising about this situation is, the company has an extensive, rigorous, Taleo-based performance management system to track contribution against plan. All goals entered into the system are necessarily "SMART" goals. All performance reporting involves a self-assessment, a next-level, supervisory assessment, an objective achievement-against-metrics measurement, and tracking individuals' demonstrated competencies against standards for competencies required by job function according to the values pillars of the organization. (Yes, I realize that last one is a bit headache inducing, but it's real.) And each of these negatively contributing employees were rated as "meets expectations" in their most recent annual performance review.
For all its industrial-model soundness, a performance management system that cannot detect and correct negative productivity isn't worth very much. This becomes especially galling when one considers that administering such a system in an organization of any significant size - that is, one large enough to consider using such a system - is itself a productivity sink-hole with an annual cost of countless person-years of effort, not to mention several full-time employees to administer it. This is not (entirely) the fault of the software system, although technology will often intensify a problematic system so that it becomes even more problematic, faster. The fault lies in the underlying industrial assumptions: that decomposing organizational missions into so-called SMART goals, distributed among departments and from there to individuals will sum back up to accomplishing the organization's mission. That a job-holder whose resume best satisfies all the requirements of the abstraction that is a job description is the best person to fulfill a particular role in an organization. That an individual working against a checklist of tasks - even if they nominally accomplish every single one of those tasks - will necessarily contribute to the overall success of the organization.
If those assumptions were ever true, even in the prior century in which businesses operated on an explicitly industrial-age model, they certainly aren't true now. The sum of individual goals will not necessarily fulfill the department's goal. Neither will the sum of departments' objectives yield the organization's mission. The best person-on-paper for the job who is also the best performer in an interview situation is rarely, if ever, the best person for the actual embodied role. Pre-determined tasks set in annual planning cycles - whether top-down or bottom-up - cannot accurately predict the location and direction of the organization's trajectory a year out, let alone several years from now, in a complex, volatile business environment.
Supposedly, the job of a performance management system is to ensure that the right tasks are being done by the right people at least to an extent that "meets expectations." The intention is for the organization to be ever more efficient and effective at accomplishing its annual objectives towards an overarching mission, in service of some long-term vision. In instances where people leaving the organization actually increase productivity, the performance management system clearly operates "below expectations." It therefore needs to be put onto a "Performance Improvement Plan," or simply managed out of contemporary organizations.
Perhaps outplacement services can find it an appropriate new home. Maybe there's a role for it in managing athletes' training. Or in a weight-loss clinic. Because those are the types of environments for which the traditional, industrial models of performance management systems are best suited. If your role in your real organization does not share attributes, characteristics, and interactions akin to losing weight or training for athletics, one size does not fit all. What might a better, more effective approach be? Something more like this.