In fourth generation (4G) wireless networks, users will be able to roam freely from one type of wireless access network to another while preserving the main characteristics of their connections. While various aspects of this vision have been discussed since mid 1990s, there remain fundamental challenges. Of these challenges is performing resource management across different access networks while considering various irregularities. This work advocates that any proposal for resource management in such networks should act reactively toward operational dynamicity while proactively allocating resources in a manner that sustains demand uncertainty. It shows how proactive allocations can be made using a formulation based on Stochastic Programming. The objective is to maximize the allocations while minimizing underutilization and rejection. It also discusses how a comprehensive proactive module can be realized. To the best of our knowledge, this is the first attempt to directly address joint resource management in wireless heterogeneous networks.