Number of things at play.

Most companies can't take advantage even in theory of saving costs if they have an office today. If they own it, who is going to buy in this climate? (keeping in mind that if it is office space, then it pretty much has to stay office space, without exorbitant effort and money to change it to something else) My company has 10 years left on their lease, with penalties of vacating early so bad that they would be just as well letting the lease run. If there's one thing a company hates it's being forced to spend money/have assets that are not seeing use.

Contributing to the above, a lot of these folks have a big part of their portfolio invested in real estate, so collapse of the office space segment of real estate would be bad news for them.

A lot of management looks awfully superfluous in a remote worker scenario. Without the visual aid of dealing with people in person, it seems like maybe you could double the sizes of departments, maybe erase a layer of middle management. So management needs people in person to maintain the appearance of relevance.

Companies also like to give tours to clients and show a busy looking office space of people working toward their customers goals. You need people in person in order for those tours to look adequately impressive.

Some of their levers to get longer work out of people work better in the office. For example, my office had way less parking than they had people coming in, and an overflow lot dangling a literal mile away from the buildings. In response to complaints that there's no reasonable parking, that there's no shuttle, that folks have to cross a fairly busy street without a signal light, an executive said "if you cared about your work, you'd come in early enough to get a good spot". They considered people who came in at 8AM to be lazy slackers, because the real dedicated people came in at 7AM, even though office work technically started at 9AM for most of them.

Frankly, remote work isn't objectively more productive across the board. You can find/create metrics to "prove" either side of the argument (measuring "productivity" is really subjective, and many of the studies are self-reporting where employees decide for themselves their productivity, or even outright state how productive the workers feel. In my experience, individual productivity may see a boost with improved focus, removal of commute, fewer work social distractions. However, the relevance of the work may suffer (for example, in my group one guy spent three weeks doing something no one needed done because he didn't have the presence of others to remind him about what really mattered) and others that depend on collaboration may falter (for example, new college hire is left adrift because it's really hard for an early career person to get traction in a pure remote scenario). We tend to care less about folks who are little more than icons next to text most of the day, or a disembodied voice for select meetings. Ambient collaboration takes a hit, as the barrier to talk to someone is a bit higher when you have to explicitly go to the trouble of typing a message or calling. It seems more intrusive.

As to why the message tends to be softball, well a number of things.

They don't want to get into the "data" game because the employees can find studies with data saying the exact opposite. Employees have a vested interest in believing their favored data.

Other statements are too aggressive, and they want to try to maintain some semblance of morale by being the "good guys". At least at the company level. From what I can tell, the corporate level at my job gets to send the happy, gentle prods to come into office, but the managers are expected to go as asshole as needed to "fix" the attendance problem.