Single ordered_non_unique index vs 2 hashed_non_unique indices

Question

Let's say I need to query\delete employees by

• their department
• their department & name

The first option would be to use a single ordered composite index and do partial key look ups:

boost::multi_index::ordered_non_unique<
              boost::multi_index::tag<ByDepartmentAndName>,
              boost::multi_index::composite_key<
                  boost::multi_index::member<
                      Employee, int, &Employee::department>,
                  boost::multi_index::member<
                      Employee, std::string, &Employee::name>>>

Or I can define 2 separate hash indices:

boost::multi_index::hashed_non_unique<
              boost::multi_index::tag<ByDepartmentAndName>,
              boost::multi_index::composite_key<
                  boost::multi_index::member<
                      Employee, int, &Employee::department>,
                  boost::multi_index::member<
                      Employee, std::string, &Employee::name>>>>,
boost::multi_index::hashed_non_unique<
              boost::multi_index::tag<ByDepartment>, 
              boost::multi_index::member<
                      Employee, int, &Employee::department>>

I would assume the first option is more space efficient since it is a single index. But it is probably slower due to O(log^N) search vs O(1) search of hash index.

What would be the right option here?