When I consider the operation what is "grouping the number of employees whose income is over 2000 by the depertment from employees", we have to write following because there is no way to filter for Collector:
(Note: In this case, we need the entry which the value is 0)

Map<Department, Long> map = emps.stream()
    .collect(groupingBy(Employee::getDepartment,
        collectingAndThen(toList(),
            es -> es.stream().filter(e -> e.getSalary() > 2000).count())));

When I add filtering like following to Collectors, we can write it easy, and it would be more efficient.
    public static <T, A, R>
    Collector<T, ?, R> filtering(Predicate<? super T> filter,
                               Collector<? super T, A, R> downstream) {
        BiConsumer<A, ? super T> downstreamAccumulator = downstream.accumulator();
        return new CollectorImpl<>(downstream.supplier(),
                                   (r, t) -> {
                                       if (filter.test(t)) {
                                           downstreamAccumulator.accept(r, t);
                                       }
                                   },
                                   downstream.combiner(), downstream.finisher(),
                                   downstream.characteristics());
    }

Map<Department, Long> map = emps.stream()
    .collect(groupingBy(Employee::getDepartment,
        filtering(e -> e.getSalary() > 2000, counting())));