Sunday, 14 October 2012

The Burden of Government Debt

The recent ‘exchange’ on this topic (see Nick Rowe here, here, here and here, and apparently on the other side Dean BakerBrad DeLongPaul Krugman and Noah Smith) may just have confused many, so here is my attempt to unconfuse. I’m doing this because (a) the issue is tricky (as I know from my own experience) (b) I’ve written about this before (c) I happen to be teaching this stuff right now (d) Nick Rowe needs some support (although no help). Bottom line: government debt can be a burden on future generations (the current generation can use it to take resources from future generations) even if there is no impact on future output, but it is also likely to reduce future output, so we really should worry about the size of government debt in the longer term. But none of these worries applies when the economy is demand constrained, as it is right now.

There are a number of ways that the current generation can exploit (take resources away from) future generations, and paying themselves using the device of government issuing debt is one. This can happen because generations overlap, and even if there is no impact on future output. Take the most basic example. At any point in time there are two generations: old and young. Only the young work and they produce 2000. They plan to save 1000 for their retirement, so consuming 1000 in each period of their two period lives. (Ignore interest payments for simplicity.) Now the old at time T get a gift of 100 from the government, paid for by issuing debt to the young. Suppose the young cut their consumption to 900 to purchase this debt. The old at T are better off by 100, and consume 1100 at T. The gift effectively comes from the young at T, but as the young bought an asset, they may think they will be OK at T+1, when they sell that asset and have consumption that period of 1100. The government then pays back its borrowing at T+1 by taxing the now (at T+1) old by 100. The T+1 old do indeed get their money back by selling the asset, but they are also taxed, so their consumption is 1000 in T+1, not 1100 as they had hoped. The economy at T+1 produces and consumes as much as it would have without this temporary creation of government debt, but the sum of consumption in both periods of the old in T+1 is 1900, while the sum of consumption in both periods of the old at T in 2100. It is as if the young just paid the old 100 at T, which is what would happen in a one-off unfunded social security scheme.

Now have the debt paid back at T+2 rather than T+1. The young at T, who are the old at T+1, are indifferent: they consume 900 at T and 1100 at T+1. The young at T+1 buy the debt from the old at T+1, and consume 900. It is when they become old at T+2 that they are worse off, when taxes rise to pay back the debt. If we call ‘society’ at some date the combination of the two generations living at that date, then we can say society at T is better off and society at T+2 worse off. But what if the debt is never repaid? I have misgivings about the relevance of this thought experiment (see here), but for what it is worth the standard result is that future generations get exploited if the real interest rate is greater than the rate of growth of the economy. However, no exploitation need occur if the debt is used not to increase the consumption of the old, but to invest in assets that future generations can benefit from (see here).

All this assumes that there is no impact on what society can produce. But that is not a realistic assumption. In the example above I just assumed that the young would be happy to postpone 100 of consumption to buy some government debt. But suppose instead they were not, and substituted that debt for 100 of their 1000 saving for their retirement. If that saving was in the form of productive capital, then the government debt ‘crowds out’ that capital. Now one for one crowding out like that is probably too extreme, but I think there are good reasons to believe that some crowding out occurs when investment in capital is governed by the availability of savings. This effect occurs not because government debt is in any sense necessarily bad – you can get exactly the same effect from investment in housing depending on who inherits houses. Indeed, if society has too much capital (which is theoretically possible), having government debt crowd out capital would be a good thing. However the balance of evidence is that society probably has too little rather than too much productive capital, which means we should be concerned about the long run impact of government debt.

Need this have any relevance to the debate on stimulus versus austerity? Absolutely not: indeed quite the reverse. No sensible person is arguing that current increases in debt should be permanent – instead they should be reversed, gradually, once the economy recovers. Until the economy recovers, investment is being held back by lack of demand, not a lack of savings, so the crowding out issue does not arise. I think you could make a good case that, by stifling the economy today through austerity, you are damaging productive capacity in the future. So in the current circumstances it is austerity, not increasing debt, which is harming future generations.


  1. Simon: I very much appreciate the support. (And I think I do need some help too).

    I agree that the case where there is deficient demand is tricky. It's going to depend on what the government buys (e.g. really good schools?) and the amount of crowding in of private investment.

  2. I must say, this latest blogosphere debate has been very mature and civil compared to previous spats, and has been very informative too!

  3. What I can't get my head round is that: at time T, we are not consuming money, we are consuming resources and producing waste. If you want more money then we can print it anytime but the resources are not being replaced, some because they take million of years, other by because politics doesn't seem to allow a long term view these days. Its hard not to get pulled in by the doom mongers.

  4. “But total consumption of the old in T+1 is 1900, while total consumption of the old at T in 2100”

    ( Typo? / (I’m finding that part difficult to understand))

    I would interpret it that the (T + 1 old) have consumed 1900 over their lifetime instead of an expected 2000.

    I’m struggling to figure out the derivation of 2100.

    1. I have changed the wording slightly to make it clearer. Your interpretation is correct. The old at T consumed 1000 in T-1 and 1100 at T, so lifetime consumption is 2100

  5. Nick is going soft here. He convinced me, in the comments section of his original post last year, that, in the "base case" with overlapping generations, no liquidity constraints, no bequest motive, and no stable government Ponzi scheme, debt is a burden on future generations even if the economy is demand-constrained (provided that it isn't expected to remain so forever). The reason is that rational people in an OLG model smooth their consumption over the life cycle. So if the debt is used to finance consumption or something that can substitute for it (or if there is a multiplier-accelerator kind of effect that causes current-period consumption possibilities to increase even if they aren't financed by the debt), then young people will choose to save more and then consume more when they are old. If the economy has returned to full employment by the time they are old, this means they will be taking resources away from future generations.

    1. Andy: I'm getting old, and my brain is going. I vaguely remember arguing something like that. It does sound like it might be right.

  6. I think the strongest case for the "no debt burden" argument (aside from the possibility of a stable Ponzi scheme) is where the economy is demand constrained and individuals are liquidity constrained. For a marginal increase in government debt, people who were liquidity constrained will still be liquidity constrained on the margin and will therefore still have no savings, so they won't be able to increase their consumption when they are old. Combine this with some non-liquidity-constrained people (since some obviously exist) who behave according to Ricardian equivalence (since it's not too implausible that most non-liquidity-constrained people have bequest motives) and you end up with a plausible setup in which debt is not a burden but Ricardian equivalence still fails in aggregate. I don't think Dean Baker mentioned liquidity constraints, but I think they're kind of critical to the case he was making.

  7. There is something very confused about this whole debate. Three issues:

    1) Labour is time constrained, use it or lose it.
    2) Iron ore or coal (for example) are of limited stock, any time we use then, we steal from the possible wealth of future generations.
    3) Symbolic tokens are not of limited stock, they can be created and rearranged - and notably have been multiple times through history.

    It seems you are most interested in proving that (3) is, in particular circumstances (stable economic systems and zealous adherence to particular economic models) in fact of limited stock. That's fair enough - I'm not wholly convinced by your argument, partly because there is evidence that capital has been piling up in storage pools for about 10 years now, partly because there is little accurate historical evidence connecting investment to savings. Still, I'm happy to agree to differ. But this leaves a huge concern:

    You implicitly agree that (1) is more serious (although Nick Rowe is notably much less clear about this - and I think that is damaging.)

    However, it's not at all clear why (3) is anywhere more serious than (2).

  8. The debt in this case is no different from a tax transfer. (You do say this, but I think it's misleading to talk about a debt burden in this case.)

    You're example is zero sum. Add a tax for all income over 1000 and pay it to all income below 1000.

    So, it's the interest burden (again, you say this), that's the problem. (Perhaps, though, we're "hiding" transfers in debt because of the unified budget.)

    I'd like to see a discussion about what we should do in the case interest becomes a problem.

  9. but I think there are good reasons to believe that some crowding out occurs when investment in capital is governed by the availability of savings.

    Is this an argument assuming credit constraints?

    I.e. if there are no credit constraints, then every investment that earns a return in excess of the risk free rate will get funded. Households will borrow to fund it, and banks will be able to create enough funds to lend to households, holding the purchased investment as collateral. Therefore the pool of savings never governs the amount of investment unless you assume some form of credit constraint. Only the pool of savings *demanded* governs the amount of investment, and here only if the CB is making real time adjustments in response to these demands.

    But this is the part I don't understand. Even if the older generation has no regard for the younger generation, so that they *do* view the government bonds as net wealth, the younger generation will understand this and view the government bonds as a net liability. The savings demands of the older generation will fall, but the savings demands of the younger generation will increase by the same amount.

    Therefore all real investments will still get funded and you have no crowding out.

    I don't see how you can argue that, in the absence of credit constraints, you get a crowding out result.

    But with credit constraints, the whole analysis really needs to shift to whether the deficit spending is harmful or beneficial based on who is the recipient of the spending, when the spending occurs, and how constrained they are.

  10. Nick Rowe and Prof Simon’s arguments are valid, given their assumptions: in particular that when the private sector sacrifices cash in exchange for government bonds, oldies react by burning up more of their net assets in retirement than had they kept their cash.

    But why would they do that? If I’m retired and have decided to burn up 50% of my assets in retirement and pass on 50% to my children, the fact that some of my cash is turned into government bonds won’t persuade me to change my 50:50 plan, will it? And if I was a youngster, doubtless I’d buy SOME BONDS, but I don’t see why youngsters would buy vastly more than oldies.

    1. Then there are people like me who don't have children and who hope to leave as little bequest as possible (though the situation is complicated by an uncertain lifespan). To the extent that people intend to leave assets for their children, Ricardian equivalence ought to apply (although maybe it doesn't for other reasons). Nick's original post last year was pointing out the irony that the people who don't believe public debt is burdensome to future generations tend to be the same people who most strongly reject Ricardian equivalence.

  11. Surely this example confuses the real and nominal economy.

    The old are not taking future output from the young (they can't it hasn't been produced), they are taking current output.