Why the Cost of the Bailout Doesn’t Matter, part 1

Let’s suppose that, as some people seem to insist, the government will not recover a cent from the distressed assets it buys in the big bailout. What will the bailout cost the taxpayers?

As I argued in my previous post, it will not cost today’s taxpayers anything. But what will it cost future taxpayers? I suggested in my last post that perhaps the government could roll over its debt indefinitely, and no taxpayers would ever have to pay it off. That might sound ridiculous at first blush. You will say, perhaps, that investors will be unwilling to refinance the debt once they catch on to the government’s strategy of rolling it over forever. I would agree, if the result of the rolling over the debt were a debt-to-revenue ratio that went up over time. In that case, investors would see that the rollover strategy was unsustainable, and they would refuse to lend. But, I will argue, we should not expect that to be the case. In all likelihood, provided that the primary deficit (i.e. deficit before interest payments) can be brought under control in subsequent years, the debt-to-revenue ratio will fall over time.

The key to my argument is that the long-term growth rate of the economy should be higher than the (expected long-term average) interest rate paid by the government. In that case, revenues will keep rising at a growth rate that is higher than the interest rate, so if you discount those revenues at the interest rate, the present value of the stream will be infinite. Thus the government effectively has infinite assets, so whatever debt it takes on, it can keep rolling over that debt and financing the interest payments with new debt, and the debt will still be a smaller and smaller multiple of revenues every year.* (This works as long as the new debt is a one-time thing. If the government keeps running large primary deficits year after year, then its debt may grow more quickly than its revenues, and that would be unsustainable.)

But will the growth rate be higher than the interest rate? Yes, I will argue, it normally will be. My argument relies on the premise that most capital income is reinvested. It’s reasonable to expect that most capital income will be reinvested, since people saving for retirement will typically reinvest their capital income and then add some new savings, whereas retired people will consume their capital income to an extent likely to roughly offset the extra saving by younger people. (OK, I'm working on a better argument, and I'm sure there is one that somebody has written a paper about -- probably a paper I read years ago and then forgot. Actually several arguments, each of which has produced several papers, all of which I either have not read or have forgotten. But let this argument do for now. In any case, to me, it seems intuitively likely that most capital income will be reinvested.)

If capital income is reinvested, then the growth rate of capital (due to that reinvestment) will be equal to the return on capital (which provides that income). In a steady state, by definition, the growth rate of capital has to equal the growth rate of labor. The economy will always tend toward that steady state because, if capital is growing faster than labor, then the amount of capital per worker is growing, which means that the return on capital must be falling (since over time, every unit of capital has less and less labor to work with and therefore becomes less productive), which means that the growth rate of capital must be falling, so it will eventually fall to the growth rate of labor. (That argument applies in reverse if capital is growing more slowly than labor.)

So here’s why the growth rate will be higher than the interest rate. The overall return on capital, as discussed above, should roughly equal the growth rate. At the level of the whole economy, there is a capital structure, where some capital is risky and has a high return, while some is safe and has a low return. Since government bonds are the safest type of capital asset, they have the lowest return. That means that the average return on capital across the economy must be higher than the return on government bonds. Since the average return on capital roughly equals the growth rate, that growth rate must be higher than the return on government bonds. Thus the growth rate is higher than the interest rate paid by the government.

Consequently, the government, in effect, has, in the capitalized stream of future tax revenues, an infinitely valuable asset, which it can put up as collateral for anything it wishes to borrow, no matter how much. It will only be a problem if the borrowing is perceived as part of an ongoing pattern, in which case the government’s asset will no longer be infinitely valuable (since future revenues have to be sufficient to pay off not only the current debt and its interest, but also all future primary borrowing, which, in the “ongoing pattern” case, is expected to be too much for the expected revenue stream to ever pay off).

Maybe you think that we are in the “ongoing pattern” case, since the government has been running fairly large deficits for several years now. Personally, however, I believe that these primary deficits will, at least in the short run, come under control. We will either have a Democratic president who promises policies to raise tax revenues, or we will have a Republican president who will be working with a Democratic congress and will thus have great difficulty extending the expiring tax cuts or instituting major new expenditures. In the longer run, we do have to worry about the effect that an aging population and rising health care costs will have on entitlement programs. I’m worried about that, but I’m not much more worried than I was a week ago.

Now, perhaps you will try a reductio ad absurdum by saying, “If any amount of one-time borrowing can be financed by rolling over the debt indefinitely, why stop at $700 billion? Why not borrow $700 trillion? Or $700 quadrillion?” It does get absurd when the numbers get ridiculously high, because it gets to the point where the debt will be huge relative to revenues and it will take a ridiculously long time before revenues catch up. In other words, mathematically speaking, the present value formula (which gives the value of the government’s revenue stream) is not very useful if it takes many centuries for the cumulative sum to come close to its theoretical limit. But with the national debt already well into the trillions, and with investors obviously willing to refinance the current debt at very low interest rates, I seriously doubt that another $700 billion – or even another $2 trillion, should it come to that – will make the difference between a reasonable level of debt and an unreasonable one.

Yes, Virginia, when it comes to financing government bailouts of financial assets, there is a Santa Claus. I should make clear, though, that Santa Claus is not omnipotent. He can come down the chimney and deliver a financial bailout on Christmas Eve, but he can’t make something out of nothing. If the government were to attempt some type of bailout that involved the production of large quantities of real goods and services, Santa Claus could only help to the extent that the economy had slack resources (as it does now and likely still will in the immediate future, but probably only a few hundred billion dollars worth, at most). When the economy runs out of resources, no bailout can make more of them.



*See Michael R. Darby, “Some Pleasant Monetarist Arithmetic,” for a more precise version of this argument in a different context.