War in Taiwan: Could China Win?

“I don’t know if China can take Taiwan — and I don’t think anybody else does, either. I think it’s unknowable.”

If China did invade Taiwan, who would prevail? Mike O’Hanlon, senior fellow at Brookings, takes on that impossibly difficult question at multiple levels of analysis. We discuss:

• The limits of scenario analysis for predicting the outcome of a China-Taiwan conflict;

• Why governments should bet on war being far more expensive than they think at first glance.

Are War Models Useful?

Jordan Schneider: Can China take Taiwan? Mike O’Hanlon is a senior fellow and director of research at Brookings. He was also the first person to give me a job in DC way back in 2010 when I was his intern.

Let’s start off with some scenarios. Which were the ones you picked to model, and why?

Mike O’Hanlon: The overall answer to the big question is, “I don’t know if China can take Taiwan” — and I don’t think anybody else does, either. I think it’s unknowable; and that may seem like an analytical punt — and maybe it is a little bit, because I admit there are various scenarios that I don’t completely understand at a technological or an operational level.

But I think that my agnosticism or uncertainty about who would win is a fundamental reality of the situation. That shouldn’t be surprising: these are two countries that have outstanding militaries with technologies that have never really been tested in combat, especially on the Chinese side. But even on our side, we haven’t had to make our integrated command and control networks operate against an adversary that has the ability to disrupt them. When our mutual friend General Stan McChrystal commanded special operations and NATO forces in Afghanistan, he didn’t have to worry about the Taliban conducting cyberattacks that could bring down his networks or jam his radars or ground his aircraft.

I like to make an analogy with sports, because sports is a fundamentally competitive human endeavor within very artificial and circumscribed parameters: a certain field, a certain length of time, a certain number of players on the field — the same on both sides — referees that are making sure people adhere to the same rules. And yet, we could have two teams that know each other, that play each other, even in the same year — and one time they could have a lopsided victory for team A, and the other time could be a lopsided victory for team B.

Even though people would have placed bets about who will win before each game, basically they didn’t know — because there is a human element of unpredictability that has to do with courage, creativity, tactics, cleverness, or just getting out of the right side of the bed that day. All these factors are enough to mean that even a sporting event confined within certain very specific parameters is unpredictable in many cases. How much more must that be the situation in war between two countries of roughly comparable capability?

Now yes, the United States still has a much better military overall than China. But we’re not talking about fighting China in a neutral location like Chile or Antarctica. We’re talking about fighting them 100 miles off their coast, and 8,000 miles off ours. You bring that together — as well as the uncertainty about how Taiwan would itself perform, how Japan or anybody else that would get involved would do — and it’s just unknowable.

So that’s the philosophical level. I know your question was more specific, but I would just say that, starting at the philosophical level, we really shouldn’t think that we can predict the outcome of wars between comparably capable countries, especially when they haven’t fought each other and haven’t tested their technologies in high-end combat in decades.

Jordan Schneider: So given that, what’s the point of doing modeling exercises, and how much can they tell you?

Mike O’Hanlon: I began with that philosophical point that I just laid out — but what I try to do is put my instincts to the test and see how well I think I can take simple mathematics, simple models of combat, and apply them to this problem.

Even if I can’t get all the best data — and maybe the best data doesn’t even exist, because none of the tests that we’ve done have been completely realistic, and even the classified tests can’t really be fully insightful — I try to establish an optimistic and a pessimistic case to bound a problem from an American point of view.

Let’s say that most of the questions about missile defense, the survivability of command and control, and other major ingredients or raw materials all break the American-Taiwan way — based on what we know about these weapons and systems, let’s go ahead and do the math, follow that through, and see who we think wins. And then let’s make a similarly pessimistic set of assumptions from our point of view and see who wins.

What I try to argue in this long paper I did over the summer is that, with plausibly optimistic assumptions about American weapons performance and military performance overall, we do in fact defeat China in its effort to squeeze Taiwan into submission through a blockade.

If the assumptions are plausibly pessimistic, however — and again, using what real-world data we do have from unclassified tests and previous battles — then I think China wins.

And these findings are fairly robust; in other words, it’s not a close call in either case. Even if China spends five, ten, twenty percent more on its military, or even if we do the same, these are not close calls. So, at the risk of speaking in oxymoronic language, I feel that I have demonstrated that the result is unknowable, and therefore both sides should be chastened about any optimism heading into such a fight — and hopefully, therefore, never have the fight at all.

Jordan Schneider: I’m with you through most of that, but I think the key piece at the very end is that these models are only policy-relevant if you’re odds are really at 95-5 — because the downside risks of running “experiment” are so enormous for Beijing (which is something that I discussed in a podcast with Jude Blanchette and Gerard Dipipo). Let’s hope it never happens.

Another point you made in your paper about “fancy modeling” versus “back-of-the-envelope” modeling: I love this example from Desert Storm where RAND had dozens of researchers do some fancy model — and it turned out that the back-of-the-envelope calculations that took previous, relatively comparable wars ended up being far more accurate in how the correlation of forces ended up playing out.

Next, we get into:

• The lower and upper bounds of war costs;

• The closest analogies in recent military history to a fight for Taiwan;

• O’Hanlon’s experience war-planning for the Congressional Budget Office in 1990.