To cope with higher demands, future radio access networks will consist of densely deployed base stations (BSs). As the BS ranges in such deployments will often overlap, areas exists that could be served individually by one of several BSs or cooperatively by all members of a group of BSs (using techniques like CoMP). One BS typically serves several such areas. Each BS has to decide, which area to serve when. This problem is compounded by the need to coordinate cooperative transmissions across several BSs. We model this as a queuing problem where each queue represents one particular area that is served by a specific combination of BSs; each BS has to decide which queue(s) to serve when. For three decision strategies, we derive a Markov model and solve it numerically to obtain steady-state solutions. These solutions let us quantify average latency and trade-offs between latency and number of BSs per area for both direct and cooperative scenarios. We show that cooperation reduces the number of BSs needed to cover an area but that a latency price has to be payed.