Sunday, November 19, 2017

On the thorniness of historical counterfactuals

Both the economist and the historian are students of human organisations and behaviour.

The economist (at least in his empiricist manifestation) is usually interested in understanding causation - what causes drive what outcomes. This is at least one way of testing our models. The aim is to understand the structure of the world we live in, with the ultimate aims being prediction and policy improvements.

The historian has a choice to be either an economist or a journalist. The economist version wants to understand why things happened the way they did, and what can be learned about the world as a result. The journalist version scorns the grandiose conceit of trying to pin down causality, and instead sticks to the smaller question of "what happened", telling the stories of the past.

Of course, economists are also historians - they take a series of events that have happened, and assert that they're comparable along the dimensions that matter and thus worthy of being used to understand general principles.

For a genuine historian, especially of world events, the tools needed are different. One must treat historical events as case studies, to be explored in detail, rather than line items in a spreadsheet. If you're understanding World War 2, for instance, for many questions there's pretty much only one line in the spreadsheet. So running a regression isn't exactly on offer, and you probably would do better to crack open a book instead.

Even though it's generally out of fashion among historians, understanding causality and counterfactuals is very important if you want to draw actual lessons from history. The counterfactual says what could have been, if some other choice had been made. It aims to make, in a literary sense, the claim to causally identifying the effect of a single decision.

It goes without saying that this is incredibly hard to do. But for those of us not bound by the standards of academic publishing, it's among the more interesting ways of thinking about history.

Getting the counterfactual right, of course, is largely a matter of judgment and opinion, since we can't actually re-run the past and find out.

But there is one aspect of thinking about counterfactuals that is beyond dispute. If you don't have a clear sense of what the counterfactual is, then you don't have any idea what the actual impact of the decision was.

You might think that nobody could possibly be this stupid, but you'd be wrong. The surest sign is when people complain about some decision that was followed by bad consequences, but never explain exactly what the alternative was and how it was meant to work.

One case where I've come across this is in the role of the British in partitioning India in 1947 into two countries (India and Pakistan) when they left. This is part of the long list of standard recitations given as to how beastly the British were in all matters of administering the British Empire.

As everybody knows, what happened after partition was a total disaster, with widespread violence and population transfers under extreme duress. Estimates of the deaths involved range from several hundred thousand to two million, according to La Wik. The problem is the classic one in the colonial critic's playbook - that the lines drawn on the map didn't correspond to messy demographic reality. When the two came into conflict, the result was maybe a million deaths.

Which sounds pretty bad, no?

But again, bad compared to what? What other choices did they have, and what would the consequences have been?

Here, the dilemma is not so much the "what" as the "consequences". The main alternative was leaving the two places as a single country. The issue, of course, is how that would have actually played out. But for some reason, the people who denounce Britain never seem to spend much time discussing this aspect.

There are, to be fair, arguments that partition was a mistake, because a single country would have done a better job of calming ethnic tensions. To my mind, the strongest of these is the relative levels of antagonism between Indian Hindus and Pakistani Muslims, versus the antagonism between Indian Hindus and Indian Muslims. 

People have a tendency to forget that the last-mentioned category of Indian Muslims not only exists, but is enormous, around 172 million (almost as many, in fact, as the 193 million in Pakistan). And while there is violence and conflict between Hindus and Muslims in India, it's not on anything like the scale or seriousness as the tension between India and Pakistan. The latter could actually devolve into a nuclear missile exchange. 

By contrast, I don't even hear about serious pushes for Muslim separatism in India. I'm sure it exists in some form. But it's less prominent than, say Quebec or Belgium. Indeed, India is perhaps the only other exception (along with Switzerland) to the Holmes rule that there are no stable multilingual countries. Somehow, they mostly manage to rub along. The surest sign that the conflict with India's Muslims is probably not an existential threat to the country is that it's possible to go months without reading any newspaper stories on "Indian Muslims". Certainly, you go a lot longer than you do without reading something about the India/Pakistan tensions.

So this reads like a pretty bad indictment of the British, at least ex post. 

But there's another counterfactual which gets discussed even less.

What if Britain left them as a single country, but they had a bloody separatist war afterwards anyway?

The amazing thing about this example is that it's not even a hypothetical. It literally happened.


Bangladesh. It used to be called East Pakistan, and was part of the same country. They were divided by geography, but shared a common religion. Surely, we would reason, they should work well in a single polity!

Except they didn't. Bangladesh fought a bloody war of independence against what was (and I quote La Wik):
An academic consensus prevails that the atrocities committed by the Pakistani military were a genocide.
As coincidence would turn out (it seems in poor taste to call it "luck" in this context), the estimated number of civilian deaths ranges from 300,000 to 3 million, rather close to the deaths during partition. 

Huh! Doesn't seem quite so cut and dried now, does it?

It leaves me actually quite agnostic on the whole question, to be honest. It seems very unclear whether things would have actually turned out better or worse if the British hadn't partitioned India when they left. It may just be that when they left the place, there was a high chance there was going to be widespread ethnic conflict no matter what they did.

And yet, to slightly paraphrase the great Mr Bastiat, people have a tendency to judge actions primarily by what is seen, not what is unseen.

The people that might have died under an alternative political arrangement are not salient at all. The deaths during partition are highly salient. 

Sometimes there are just no good options, at least based on what you can reasonably know at the time. Sometimes, you have to call things as you see them, suspecting that a lot of people might die either way.

It seems depressing, but likely, that the only actual lesson of partition is the one that the Bard wrote centuries ago - uneasy lies the head that wears a crown.


  1. Or... you know... the British could have simply not left. They were far better at running India than the Indians have ever been.

    -- An Indian.

    1. I'm quite sympathetic to that view, actually. The big question is how many other things would have to have been different before staying on would have been a real option. All counterfactuals involve some degree of unreality, but the question is how many deviations are involved and what they are. It's relatively easy for me to imagine the British rulers making a last minute decision to not implement partition. But as I've written about before, the process of decolonialisation started quite a lot earlier than most people realise, and I don't fully understand it. (See Chris B's objections to my post above, for instance, though I don't think he enunciates a clear answer either). In other words, I don't know how I exactly I would go about hypothetically altering Britain's history to make "staying on" happen.