The long run government debt target

In a recent post I had an imaginary interviewer asking “But surely no government can keep on borrowing more forever.” To which my suggested reply was “Of course not. But the right time to cut government borrowing is when the economy is strong, and the cost of borrowing is high.” This prompted a little discussion in comments about the long run desirability or otherwise of government borrowing. What I have to say here is only about the long run, and has no immediate relevance while we are still in a recession.

I have to stress here that by long run, I really do mean very long run. It is the period to which Keynes dictum applies. Why? One of the most robust ideas when it comes to government debt is that it should adjust very slowly, and absorb any shocks coming from the economy along the way. The reason, which is just tax smoothing, I have discussed at greater length here. Which prompts an obvious question: if any long run debt target is meant to be achieved in centuries rather than years or decades (I did say very long run), do we need to worry too much about it? This turns out to be a rather good question.

What little literature there is on this issue contains the ‘steady state random walk debt’ result. What this means in ordinary English is that it can be optimal to have no target for government debt. Perhaps a better way of putting it is that the costs of adjusting towards any target outweigh the long run benefits of achieving it. Imagine a recession raises debt. To get debt back down we need to increase taxes in the short run by a lot. Alternatively we could make no attempt to reduce debt, but instead just raise taxes by enough to pay the interest on the extra debt. That will mean raising taxes by less, but having to do so forever. So in one case (debt targeting) taxes rise by a lot in the short run, but not at all in the long run, while in the other (debt accommodation) taxes rise by a little forever.

You might think the second (debt accommodation) alternative must be worse, because the pain is ever lasting, but you would be forgetting about discounting. Which is better depends on the size of the discount rate relative to the size of the real interest rate. It turns out that if the two rates are equal, the second alternative (debt accommodation) is optimal. And it just so happens that in our benchmark macroeconomic model, where agents care about their children and so effectively live forever, the two rates are indeed equal.

Obviously this ignores default, which might put an upper limit on debt. There are other, potentially important, caveats, which I have discussed elsewhere. However the basic result relies on the exact equality of the discount rate and real interest rate. If the real interest rate is even slightly greater than the discount rate, and there are good reasons for thinking in the long run that it might be, then it makes sense to adjust to a debt target, although the adjustment should be very slow. So what should this target be?

The simplest answer is also the most extraordinary – the government should aim to hold assets, not debt. The long run debt target should be negative. The reason is that with lots of assets, the government could pay for all its spending out of the interest on those assets, and as a result taxes could be abolished. Well, maybe not all taxes: some are designed to influence incentives. But most taxes are designed to raise revenue, and these distort incentives, so if we could get the revenue another way that would be great. (Even if you think the preoccupation of many economists with the negative incentive effects of taxes is overdone, it must be worrying that in the US the costs of complying with individual and income tax requirements for 2010 has been estimated to be 1% of GDP.)  

Now many of you will think that I’ve entered that imaginary world that is the one economists like to dream about, but which is a million miles from reality. If so, have a look at this chart, which is of government net financial liabilities in 2007, before the recession. (Source OECD Economic Outlook)

Norway is a special case, of course, but note Australia, Denmark, Finland.... Now I do not know much about the reliability and comparability of these net debt figures, but Australia’s gross debt was less than 15% of GDP in 2007, and both they and New Zealand had a clear policy to reduce debt towards zero, although for different reasons than the one suggested above.

Turning government debt into assets may seem an impossible goal now, and it is a goal we should be ignoring in a recession. However once the recession is over, and given that adjustment should be very very slow, it is not so obviously a ridiculous target.  If we combine incredibly slow adjustment with an incredibly ambitious target, we might end up with something reasonable – which is to aim for a gradually falling debt to GDP ratio once the recession is over.

There are lots of qualifications I would want to throw at this result, but this post is already long. (Those who are interested can read this working paper.) The only one that I have come across which completely overturns this idea is the literature on safe assets, but that definitely requires another post. So let me end instead with an amusing (at least to me) little story. I was recently giving this working paper in a UK economics department. One of the department’s members is a well known and for me inspirational macroeconomist, but notoriously right wing. At first he liked the message of my paper, which was that government debt should come down. Until, that is, he saw where it was leading – to an economy where the government owned a large proportion of assets, and therefore inevitably a large proportion of the capital stock. I think this had unfortunate resonances for him!