To take this debate further we need to get a bit clearer about what we are debating. Are we discussing whether we want END to survive, or whether we expect it to survive? Many of the posts we have seen so far argue that END is a good thing that we should want to preserve if we can. Some comments on my earlier post seem to imply I think the opposite.
Let me be clear: I think END has been an extremely successful and beneficial policy, and I wish it could last forever. No debate there.
But can it survive? This is a very different and to me much more pressing question, because, as I argued in my earlier post, the conditions under which END has worked until now are changing. I think END is unlikely to survive these changes, because I see it as a complex, delicate thing, and the conditions required for its survival are specific and demanding. This is the underlying point on which I differ from many of the excellent posts in this debate so far. In what follows I will respond primarily to Bruno Tetrais' second post, and to my colleagues Stephan Fruehling and Ben Schreer.
For them, END is very simple. America promises it will come to an ally's aid if it is attacked, and the ally is happy to have that assurance. The ally need not be sure that the US would honour its promise in a crisis, because what matters is the effect on the supposed adversary, who cannot be sure the Americans will not honour their promise, and will be deterred by the possibility that it might. This account of END makes it seem as if it delivers big benefits for little cost. Why shouldn't it last?
But I do not think END is that simple, and especially not in the cases where it matters most. It is complicated because of the substantial costs that both the US and its ally must incur in sustaining END. These costs mean that END will only survive if it delivers benefits to both sides that justify them. That is what I doubt.
What are these costs? Let's start with the ally. Being a US ally and accepting END carries several kinds of costs. The most important of these is the need to forgo other ways to protect oneself — what one might call strategic opportunity cost. These costs are quite low for a relatively weak yet secure country such as Australia. But for a strong ally in a risky position — in other words, for the most important allies — the strategic opportunity cost can be substantial. Strong allies have other options, and for allies at risk, the imperative to find the most effective defence is strong, so the costs of sticking with a sub-optimal policy is high.
Now consider America's most important ally today. END is not a free good for Japan. Preserving END costs Japan the option of developing its own nuclear forces. The future of END for Japan depends not simply on whether Japan believes it benefits from America's promise, but whether it believes accepting END provides Japan with more security than the alternative. This is a pressing question for Japan today, not just because its fears of China are growing, but because Tokyo's confidence that America's promises will stick are fading.
The reason is that END carries costs to the US as well. As I said in my earlier post, these costs are low when the adversary cannot retaliate against the US. But against an adversary capable of nuclear attack on American soil, the costs are considerable. It is easy to underestimate the costs of END imposed by the risk of retaliation if we discount the risk that a crisis involving a US ally would escalate to a nuclear exchange. But should we do that?
To explore this we need to look quickly at the relationship between END and its close cousin, Extended Conventional Deterrence (ECD). Several posts in this debate have touched on the connection between END and ECD, but I'm not sure we are clear about it. At first glance it seems simple enough: in both cases the US promises to protect its allies, and the only difference is the kind of forces it is prepared to use. Much is made of the effectiveness of conventional US forces to protect allies, which carries the optimistic implication that the costs to the US of this kind of extended deterrence are low. Indeed optimists about the durability of END tend to blur the distinction between END and ECD, viewing them as two sides of the same coin and suggesting that END is really not much more risky than ECD for the US.
I do not agree with this. First, how distinct are ECD and END in practice? Much depends on the context. It is easy to exclude the possibility of nuclear use by the US against a non-nuclear adversary that does not have clear conventional superiority. But against an adversary which has either the capacity to prevail against US and allied conventional forces, or has nuclear weapons, or both, then excluding a nuclear response would drastically weaken the value of US promises of support. So against these adversaries — the ones that matter, in other words — any extended deterrence must be nuclear to be effective.
Second, the threshold at which the nuclear element of extended deterrence comes into play in a crisis would be much lower than people may think. Several posts in this debate — Stephan and Ben's talking about the Baltics, for example — seem to suggest that the US could respond to conventional threats with conventional forces, and hence fulfill its promises to allies, without running the risks of nuclear war. This is because the threat of US nuclear retaliation would deter the Russians from threatening nuclear action in response to US conventional operations.
My hunch is that, on the contrary, there is a real risk of escalation to a nuclear exchange in any conventional clash between two nuclear powers, as long as there is asymmetry in their calculations of cost and risk. In particular, geography imposes structural asymmetries of interest between the US and its Eurasian competitors which would profoundly affect calculations in a crisis.
For example, in a crisis between the US and Russia over a Baltic state, the risks might be symmetrical but interests are not. The Baltics matter a lot more to Russia than they do to America. This does not mean Russia would be prepared to accept a nuclear attack to get hold of Estonia (for example). But it does mean Moscow would probably calculate that the US would not be willing to risk nuclear attack to save Estonia. So, assuming the Russians believed they could prevail in a conventional attack on Estonia, they might well calculate that the US would not be willing to escalate from conventional to nuclear warfare to stop them. And the Estonians might well think so too.
Now back to Japan. It is entirely credible that the US would retaliate with nuclear forces to any direct nuclear attack on Japan, so this element of END remains intact. But that is not enough for Japan to feel secure, especially from China, because it leaves open the credibility of US undertakings in the event of a conventional crisis.
Consider a Chinese conventional attack on Japan's Ryuku Islands. America's formidable sea-denial capability would be able to interdict Chinese sea-borne forces, but China could threaten substantial conventional retaliation against US interests, so the costs to the US of supporting Japan in this situation would be very high. The question would arise whether the Ryukus were worth it. But say firmness prevailed in Washington, and the Air-Sea Battle swung into action. China might then threaten nuclear escalation if the US persisted — initially perhaps against Guam.
America would of course threaten nuclear retaliation against China. But while the risks are broadly symmetrical, the interests are not: China cares more about the Ryukus that the US does. It is the Cuban Missile Crisis in reverse. So American decision-makers might well decide that China's threats must be taken seriously. They would then ask themselves whether the Ryukus are worth LA, just as General Xiong suggested. The answer is not a forgone conclusion.
This is very important for Japanese decision-makers weighing the costs and risks of different options for Japan over the next decade or two. Going nuclear would carry huge costs for Japan, but relying on END carries big risks.
END evolved during the specific circumstances of the Cold War when, for reasons I sketched in my earlier post, American interests in Europe and Asia were as great as those of the Europeans and Asians themselves. Today this is no longer true, because neither China nor Russia today poses the kind of threat to America that the Soviets once posed. It is this widening asymmetry of interest between America and its Eurasian allies that undermines END.
The Nuclear Reactions column is supported by the Nuclear Security Project of the Nuclear Threat Initiative, as part of a wider partnership between the NSP and the Lowy Institute
Photo by Flickr user discoodoni
