The real lesson of the debt burden debate
It was such a simple argument.
Every ‘IOU’ has a corresponding ‘UOMe’. So, if you consider the set of all people alive at any point in time, the IOUs can’t be a burden on those people because (at the aggregate level) any repayment is being made to other people who are alive. The set of people alive cannot be burdened (if there is no transfer between time periods). Here’s Nobel laureate Paul Krugman on the topic a few days ago (emphasis mine):
That’s not to say that high debt can’t cause problems — it certainly can. But these are problems of distribution and incentives, not the burden of debt as is commonly understood. And as Dean says, talking about leaving a burden to our children is especially nonsensical; what we are leaving behind is promises that some of our children will pay money to other children, which is a very different kettle of fish.
I used to buy this line, until I grokked Nick Rowe’s argument that if different cohorts of people overlap, then Krugman’s argument (that it is nonsensical to say that debt could be a burden on our children) falls apart. There are still empirical questions as to how much of a burden it is*, but it is not a question of identity or tautology.
This is an illustration of a profound epistemological problem with using grossly simplified models of reality. Before this argument began bouncing around the blogosphere, I had a kind of model in my head that did not factor in overlapping generations of people. It produced the result that debt cannot be a burden on future people. In the context of that model, that is the correct result. But by leaving out overlapping generations, it so happened I left out a crucial detail.
There are good reasons that we simplify in order to model and understand. Sometimes (arguably always) it is just not possible to factor in everything. Whether or not this is a problem basically boils down to whether the stuff you omitted makes a difference to the answer to the question you’re asking. But how do you know you haven’t omitted a crucial detail? In the physical sciences, you can create testable predictions on the basis of your model and run experiments. Most of the time, you can’t do that in economics (or at least every result is controversial due to the problems of trying to isolate particular variables in complex systems). We can’t run an experiment in the real world and say “hmm, the future people in this experiment have been burdened by the debt, so I guess my model is missing out something important”.
Luckily, in the case of the debt burden you can show the model is quite possibly missing out something important just through a simple thought experiment. That’s the exception rather than the rule. Most of the time, we have a lot less independent confirmation of our models than we think we do.
To paraphrase CS Lewis – not only do we believe that our models are right, but it is often by our models that we decide what is right. This is, to put it mildly, a problem.
*Essentially, it depends on the degree to which either the debt is sustainable because NGDP growth is higher than the interest rate, or older cohorts leave bequests to their children