Do alliances work?

The signing of the ANZUS Treaty.

With ANZUS a core pillar of our own strategic policy, it should come as no surprise that Australians frequently turn (and return) to the subject of just how reliable that alliance is. Most of the debate tends to be remarkably impressionistic. For some, history is the best guide—and Britain’s inability to come to Australia’s aid after the fall of Singapore in 1942 a salutary warning about the dangers of a smaller power becoming too reliant on a great power to protect it. For others, reliability is simply assumed—sometimes on the basis that if the US refused to honour its ANZUS commitments all of its other alliances would come under increased pressure.

But we should look at some data to take the impressionism out of the debate. We should be interested not just in the big question—is ANZUS reliable or isn’t it—but in the specifics: how reliable is it? There are several ways of judging the utility of alliances—including whether they deliver strategic gains during peacetime through training, technology, intelligence exchange and the like. Still, the real test of an alliance’s reliability is whether alliance partners end up honouring their commitments to each other on the battlefield.

It’s instructive, then, to turn to the academic literature for a set of insights on just how reliable alliances actually are.

Here, there are relatively few major studies, and I’ll discuss only two of them. An American academic, Alan Ned Sabrosky, back in the 1980s, put a sizeable dent in the reputation of alliances when he concluded that a study of allies’ behaviour over a 150-year period showed that they fought alongside each other in wars only 27% of the time. On 61% of occasions, allies sat by while their partner was fighting. Worse, on 12% of occasions they fought on the other side.

Those unsettling conclusions—enough to make any country doubt the value of its alliance—have, in more recent years, been revisited by other researchers. Brett Ashley Leeds and two of her colleagues at Florida State University revisited the Sabrosky data-set in 2000, and argued that his test for ‘alliance reliability’ left much to be desired. Sabrosky had tested only whether one ally fought alongside another in any conflict, not whether it did so in the circumstances in which it had a specific alliance commitment to do so. Moreover, Sabrosky had counted as ‘alliances’ agreements that might more properly be described as ‘ententes’ (agreements merely to consult) or non-aggression pacts. Leeds recoded the data to reflect the specific obligations laid down in alliance commitments.

The results were substantially different. Alliance ‘reliability’, redefined as the honouring of specific agreements, rose from 27% to 74.5%—or, near enough, from one quarter to three quarters of cases. Leeds’ research lies at the basis of an important truth in alliances: specifics matter. Alliance reliability increases when we take the specific commitments made by nations into account. Most alliances are not general purpose pledges to fight alongside another country in all circumstances, and shouldn’t be judged as such.

Now, what does all that mean for ANZUS? Well, if we genuinely believe that ANZUS might well be a more substantive alliance in the 21st century than it was in the 20th, then we might want to go back and re-read the specifics of the agreement. Further we might want to discuss with Washington how we both interpret the specifics. We can read Sabrosky’s research as a sign of just how weak international commitments are if they are regarded as general-purpose undertakings. But we can take Leeds’s research as an affirmation of just how strong international commitments are in relation to specific undertakings. They certainly aren’t absolute guarantees—but they seem to work three quarters of the time. I suppose there’s a caveat needed somewhere here similar to the sort used by fund managers selling their products: that past performance should not be taken as an assurance of future outcomes. Still, we should draw comfort from the broader pattern.

Rod Lyon is a non-residential fellow at the Australian Strategic Policy Institute. Image courtesy of Flickr user kohane.

 

Here, there are relatively few major studies, and I’ll discuss only two of them. An American academic, Alan Ned Sabrosky, back in the 1980s, put a sizeable dent in the reputation of alliances when he concluded that a study of allies’ behaviour over a 150-year period showed that they fought alongside each other in wars only 27% of the time. On 61% of occasions, allies sat by while their partner was fighting. Worse, on 12% of occasions they fought on the other side.

Those unsettling conclusions—enough to make any country doubt the value of its alliance—have, in more recent years, been revisited by other researchers. Brett Ashley Leeds and two of her colleagues at Florida State University revisited the Sabrosky data-set in 2000, and argued that his test for ‘alliance reliability’ left much to be desired. Sabrosky had tested only whether one ally fought alongside another in any conflict, not whether it did so in the circumstances in which it had a specific alliance commitment to do so. Moreover, Sabrosky had counted as ‘alliances’ agreements that might more properly be described as ‘ententes’ (agreements merely to consult) or non-aggression pacts. Leeds recoded the data to reflect the specific obligations laid down in alliance commitments.

The results were substantially different. Alliance ‘reliability’, redefined as the honouring of specific agreements, rose from 27% to 74.5%—or, near enough, from one quarter to three quarters of cases. Leeds’ research lies at the basis of an important truth in alliances: specifics matter. Alliance reliability increases when we take the specific commitments made by nations into account. Most alliances are not general purpose pledges to fight alongside another country in all circumstances, and shouldn’t be judged as such.

Now, what does all that mean for ANZUS? Well, if we genuinely believe that ANZUS might well be a more substantive alliance in the 21st century than it was in the 20th, then we might want to go back and re-read the specifics of the agreement. Further we might want to discuss with Washington how we both interpret the specifics. We can read Sabrosky’s research as a sign of just how weak international commitments are if they are regarded as general-purpose undertakings. But we can take Leeds’s research as an affirmation of just how strong international commitments are in relation to specific undertakings. They certainly aren’t absolute guarantees—but they seem to work three quarters of the time. I suppose there’s a caveat needed somewhere here similar to the sort used by fund managers selling their products: that past performance should not be taken as an assurance of future outcomes. Still, we should draw comfort from the broader pattern.

Rod Lyon is a non-residential fellow at the Australian Strategic Policy Institute. Image courtesy of Flickr user kohane.

 

Comments are closed.