Efficient planning for disasters can significantly reduce their adverse impact on the public; however, assigning people to shelters becomes problematic especially while planning for those at-risk populations such as older adults who may need special assistance. In addition, hurricane registries may not reflect the actual number of people with access and functional needs. This can lead to a shelter capacity deficit in a county, which may be mitigated by collaborations among neighboring counties. The overall effectiveness of a cross-county collaboration (compared to the no-collaboration case) in shelter assignment was previously shown using a p-Median model. However, it is also necessary to investigate the cost of such a collaboration at the block group level, which is the main scope of this study. This collaboration cost includes both the cost of being assigned to a farther shelter, and the cost of not being assigned under the collaboration cases. To monitor the total performance of shelter assignment under collaboration, performance measures are developed to evaluate the benefits and cost of this collaboration (i.e., the total number of assigned/ unassigned people, the number of people assigned to farther shelters, the average cost of assignment, the number of people assigned/not assigned under the collaboration). A GIS-based numerical example is provided for 5-county Panhandle region of Florida with a focus on the 85 + population and special needs shelters (SpNS). Results indicate that it is possible to increase the total number of people assigned in the 5-county region by 2-7% using the cross-county collaboration.